Sacmi Vol 2 Inglese - II Edizione

Applied Ceramic Technology Volume II

Copyright 2005 SACMI IMOLA s.c. a r.l. Via Selice Provinciale 17/A - 40026 Imola (BO) Italy Tel. 0542/607111 - Fax 0542/642354 www.sacmi.com e-mail: [email protected] Not for sale All rights reserved. Translation, electronic storage, reproduction or adaptation by any means (microfilm and photocopy included), whole or partial, is prohibited. ISBN 88-88108-55-6 Editrice La Mandragora s.r.l. Via Selice 92 - Cas. Post. 117 - 40026 Imola (Bo) Italy Tel. 0542/642747 - Fax 0542/647314 e-mail: [email protected]

INDICE

Introduction ................................................................................................................................ 13 Evolution of ceramic tile production technology ............................................................. 13 Evolution of ceramic tile plant engineering ....................................................................... 16 Grinding ....................................................................................................................................... 17 Pressing ........................................................................................................................................ 19 Porcelain tiles and production line innovations ................................................................. 22 Firing ............................................................................................................................................ 25 Chapter I - Grinding .............................................................................................................. 29 Definition and purpose of solid material grinding ........................................................... 29 Properties of solids ................................................................................................................... 29 What happens during grinding; choosing the right machine ........................................ 30 Dry grinding and wet grinding ............................................................................................. 32 Machines commonly used in the grinding department ................................................... 33 Wet grinding theory ................................................................................................................. 36 Discontinuous wet grinding theory with Alsing or ball mills ....................................... 38 Note n. 1 ....................................................................................................................................... 45 Note n. 2 ....................................................................................................................................... 45 Practical load calculation for discontinuous wet grinding mills .................................... 51 Calculating the load for a discontinuous mill (example) .................................................. 52 Continuous wet grinding ......................................................................................................... 53 The continuous wet grinding plant ...................................................................................... 53 Continuous grinding technology ........................................................................................... 55 Description of continuous mills; how to choose the right size ...................................... 56 Description of the mill ............................................................................................................. 56 Mill sizing .................................................................................................................................... 60 Mill productivity ........................................................................................................................ 60 Control and production parameters in the continuous and discontinuous grinding department ................................................................................................................. 62 Definitions and units of measure .......................................................................................... 64 Appendices ................................................................................................................................... 66 The continuous wet grinding mill and the porcelain tile industry ............................... 66 Technological and managerial parameters .......................................................................... 67 Chapter II - Spray drying of ceramic slips .................................................................... 71 Classification of spray driers .................................................................................................. 72

General description of nozzle-type spray drier ................................................................. 74 How it works ............................................................................................................................... 74 The spray drying cycle ............................................................................................................. 75 Description of main spray drier devices .............................................................................. 77 Dust separator device ............................................................................................................... 86 The dynamics of dried granulate formulation ................................................................ 87 Characteristics of spray dried powders ............................................................................... 92 Morphology of the grain and particle size distribution of the powders .................... 92 Grain size ..................................................................................................................................... 92 Powder flowability ..................................................................................................................... 93 Variations in physical characteristics of spray dried powders ....................................... 96 Relationships regulating working conditions of spray driers ....................................... 96 Energy consumption ................................................................................................................. 98 The heat transfer diagram ....................................................................................................... 98 Essential parameters and examples for the evaluation of energy consumption and spray drier output capacity ................................................................. 100 Calculating heat energy consumption ............................................................................... 100 Practical method for calculating evaporating power of a spray drier ....................... 101 Calculating the size and output of the spray drier ........................................................ 101 Energy savings ........................................................................................................................ 103 Energy recovery ..................................................................................................................... 105 Chapter III - Pressing ......................................................................................................... 107 Introduction ............................................................................................................................. 107 Pressing systems ..................................................................................................................... 107 Pressing semi-dry powders .................................................................................................. 108 Different press types .............................................................................................................. 108 Toggle press ............................................................................................................................. 108 Friction press ........................................................................................................................... 108 Hydraulic presses .................................................................................................................... 110 Requirements for a modern press ....................................................................................... 110 Different types of press for different tiles ........................................................................ 111 Machine devices ...................................................................................................................... 114 Machine structure ................................................................................................................... 115 Different types of die used to form the tiles .................................................................... 116 The pressing sequence ........................................................................................................... 119 Main parts of the die ............................................................................................................. 120 Different punch types ............................................................................................................ 121 Characteristics of the ceramic powders used in pressing ............................................ 121 Variables in the pressing cycle ............................................................................................. 122 Definition of terms ................................................................................................................ 122 Physical characteristics of the particles ........................................................................... 123 Powder fluidity ........................................................................................................................ 126

Powder particle size distribution ........................................................................................ 126 Bulk density, tapped density and Hausner index as a function of variations in powder parameters ............................................................................................................ 129 Ceramic powder pressing: technological aspects ............................................................ 129 Relationship between bulk density of the pressed material and green bending strength (practical aspects) .................................................................................................. 135 Effect of forming pressure on firing performance of ceramic powders (practical aspects) .................................................................................................................... 138 Future developments in pressing........................................................................................ 142 Appendices ................................................................................................................................ 145 Green physical-mechanical characteristics of tiles obtained from different powders (spray dried/granulated/dry ground) .............................................................. 145 Particle size distribution of powders produced using different systems .................. 145 Compaction and uniformity of pressing ........................................................................... 147 Green and dry bending strength ........................................................................................ 149 Porosity and fired tile bending strength ........................................................................... 149 Chapter IV - Drying ............................................................................................................ 153 Conventional hot air driers for ceramics........................................................................... 154 Technological considerations............................................................................................... 155 Glossary .................................................................................................................................... 159 Machines ................................................................................................................................... 161 Rapid vertical driers ............................................................................................................... 162 Horizontal driers .................................................................................................................... 166 Evaporating conditions ......................................................................................................... 169 Infrared rays............................................................................................................................. 170 Microwaves .............................................................................................................................. 171 Technological aspects ............................................................................................................ 171 Relationship between drying cycle and tile thickness/size .......................................... 171 General observations ............................................................................................................. 172 Chapter V - Glaze and decoration application techniques................................... 175 Glaze preparation ................................................................................................................... 175 Preparing screen printing pastes ........................................................................................ 182 The glazing department ....................................................................................................... 182 Glaze application and glazing lines .................................................................................... 184 Key units on the glazing line ............................................................................................... 188 Other units ............................................................................................................................... 190 The glazing line ...................................................................................................................... 190 Printing machines ................................................................................................................... 191 The scraper blade (squeegee) ............................................................................................... 191 Flat screen printing ................................................................................................................ 192 Rotary printing machines ..................................................................................................... 192

Roller decoration ..................................................................................................................... 195 Screen mesh types................................................................................................................... 196 The frame .................................................................................................................................. 198 Emulsion ................................................................................................................................... 199 Photoengraving ....................................................................................................................... 200 Silk screen printing inks ....................................................................................................... 200 Mixing the paste ..................................................................................................................... 201 Refinement ................................................................................................................................ 201 Screen printing vehicles (medias) ....................................................................................... 202 Screen printing decoration techniques .............................................................................. 203 On-glaze decoration ............................................................................................................... 203 Under-glaze decoration ......................................................................................................... 203 Colour gradation ..................................................................................................................... 204 Plain tiles .................................................................................................................................. 204 Colour overlap ......................................................................................................................... 205 Chapter VI - Firing ............................................................................................................. 207 General ...................................................................................................................................... 207 The transformations that take place during firing ......................................................... 207 The firing cycle ....................................................................................................................... 213 Different types of firing ........................................................................................................ 215 Fuels ........................................................................................................................................... 216 Ceramic firing kilns ................................................................................................................ 220 Heat exchange ......................................................................................................................... 220 Heat transmission by convection ........................................................................................ 220 Kiln construction .................................................................................................................... 221 Combustion system ................................................................................................................ 223 Controls ..................................................................................................................................... 232 Rollers ........................................................................................................................................ 234 Metal rollers ............................................................................................................................ 234 Ceramic rollers ........................................................................................................................ 236 Raw materials and formulations .......................................................................................... 236 The most common causes of roller breakage .................................................................. 237 Chapter VII - Sorting, packaging and palletizing lines ......................................... 243 Introduction ............................................................................................................................. 243 Analysis and classification of tiles ...................................................................................... 243 Automatic tile inspection ...................................................................................................... 252 Introduction ............................................................................................................................. 252 Tile vision systems: characteristics .................................................................................... 253 Installing an automatic sorting system ............................................................................. 258 Defects detected by an automatic sorting system .......................................................... 259 Automatic sorting systems: performance and advantages ........................................... 260

Automatic sorting systems: other uses ............................................................................. 261 Classification and stacking ................................................................................................... 262 Packaging .................................................................................................................................. 264 Printing and labelling zone .................................................................................................. 267 Palletizing zone ....................................................................................................................... 267 Sizing a sorting and palletizing line .................................................................................. 271 Production control software ................................................................................................ 272 Sorting line configurations for porcelain tiles ................................................................. 273 Chapter VIII - Polishing .................................................................................................... 275 Introduction ............................................................................................................................. 275 Past developments and current trends .............................................................................. 275 Lines and machines for the rough flattening - polishing - squaring of porcelain tiles ........................................................................................................................... 276 The work cycle ........................................................................................................................ 276 Glazed wall tile squaring lines ............................................................................................ 282 Polishing - satin finish - semipolishing lines for glazed, unglazed and third firing ceramic ........................................................................................................................... 282 Appendices ................................................................................................................................ 283 Water treatment and recycling ............................................................................................ 283 Recovery/re-utilisation of porcelain tile polishing sludge .......................................... 283 Polishing sludge ...................................................................................................................... 284 Difficulties in using the sludge ............................................................................................ 286 Appendix 1 - The environmental impact of the ceramic industry..................... 289 Pollutants in raw materials for bodies ............................................................................... 291 Pollutants in glazes ................................................................................................................ 292 Pollutants in gaseous emissions .......................................................................................... 292 Atmospheric pollution ........................................................................................................... 294 Prevention and separation .................................................................................................... 295 Bag filters .................................................................................................................................. 296 Fume filtration......................................................................................................................... 297 Abatement of organic substances....................................................................................... 297 Water pollution ....................................................................................................................... 298 Water treatment (separation) ............................................................................................... 303 Porcelain tiles ........................................................................................................................... 305 Solid waste and residues ....................................................................................................... 305 Appendix 2 - Production controls .................................................................................. 309 Controls on raw materials or bodies .................................................................................. 309 Body preparation department controls ............................................................................. 321 Spray dried powder controls ................................................................................................ 322 Press department controls ................................................................................................... 323

Controls on dried tiles ........................................................................................................... 326 Controls on the biscuit (double fire only) ......................................................................... 327 Controls in the glazing department ................................................................................... 328 Controls on the finished product ........................................................................................ 329 Laboratory controls on raw materials for frits and colours ......................................... 330 General criteria for quality control tests on raw materials for glazes ...................... 332 Appendix 3 - Defects ........................................................................................................... 333 Defects associated with raw materials ............................................................................... 335 Problems attributable to the technological characteristics of the body ................... 347 Problems associated with the nature of the raw materials that appear after firing ................................................................................................................................. 349 Other defects associated with raw materials .................................................................... 349 Body preparation defects ....................................................................................................... 350 Pressing defects ....................................................................................................................... 352 Glaze and glazing defects ..................................................................................................... 354 Defects caused by the glaze .................................................................................................. 354 Application defects .................................................................................................................. 356 Poor glaze-body match .......................................................................................................... 357 Glazed surface defects ............................................................................................................ 357 Nature and formation of gaseous emissions .................................................................... 358 The influence of stains .......................................................................................................... 361 Glaze application defects ....................................................................................................... 363 Decoration defects .................................................................................................................. 367 Firing defects with effects on the glaze ............................................................................. 368 Firing defects ........................................................................................................................... 370 Tile bursts in the pre-kiln (explosions) ............................................................................. 370 Breakage during preheating ................................................................................................. 370 Cooling breakage (or cooling cracks) ............................................................................. 375 Identifying the point of tile failure .................................................................................... 375 Crazing ...................................................................................................................................... 376 Uniformity of shrinkage over the kiln cross-section .................................................... 377 Glaze brilliance and shade .................................................................................................... 380 Pin holes - holes - bubbles in the glaze ............................................................................. 381 Degassing ................................................................................................................................. 383 Degassing in double fire processes ..................................................................................... 384 Black core .................................................................................................................................. 384 Planarity defects ...................................................................................................................... 386 Downturned corners .............................................................................................................. 387 Upturned corners ................................................................................................................... 388 Convexity .................................................................................................................................. 389 Concavity .................................................................................................................................. 392 Roller effect .............................................................................................................................. 395

Priest hat ................................................................................................................................... 397 Asymmetric deformation or warping ................................................................................ 399 Monoporosa ............................................................................................................................. 401 Convex planarity (Monoporosa) ......................................................................................... 401 Non-uniform planarity across the kiln load in monoporosa ........................................ 405 Non-uniform planarity over time (Monoporosa) ............................................................ 406 Non-uniformity of size across the load cross-section in monoporosa ...................... 407 Production contamination .................................................................................................... 408

Introduction

INTRODUCTION

Evolution of ceramic tile production technology The developments seen in the ceramic industry since the end of the Second World War, have, quite simply, been extraordinary. The way in which building bricks, sanitaryware, tableware and ceramic tiles are produced has changed radically. This is especially true of tiles, where production plants and markets have been revolutionised; on the one hand, this has come about as a result of the enormous demand for industrial and residential building materials and, on the other, continuous efforts to increase productivity and improve health and safety standards for workers in the ceramic industry itself. Not so long ago the ceramic industry was an almost craft-like affair (fig. 1) that depended on an abundant supply of manual labour. Progressively, yet rapidly, it soon became a mechanised industry, thus eliminating much of the pure physical toil. From here rationalisation of the work process led to gradual replacement of complex control and organisational activities via the application of electronic automation. Generally speaking, such changes have resulted in uniformity of output and the transformation of the human role, the latter now largely being limited to supervision and control of highly automated even robot-controlled processes. Fortunately, more uniform production results have been counterbalanced by extremely keen interest in the development of new tile shapes, new decorations and new products, thus avoiding the blandness of standardisation altogether. CRAFT-LIKE ORGANISATION ORGANIZZAZIONE ARTIGIANALE

MECHANISATION MECCANIZZAZIONE RATIONALISATION RAZIONALIZZAZIONE

AUTOMATION AUTOMAZIONE

UNIFORMITÀ DIOF PRODUZIONE UNIFORMITY OUTPUT MANUAL WORK LAVORO MANUALE

SUPERVISION ATTIVITÀ DI SUPERVISIONE

Fig. 1.

13

Applied Ceramic Technology

In fact, this demand for ever-better aesthetics has driven technological development onwards and upwards. Plant engineers and machine designers have been forced to elaborate more and more complex manufacturing solutions to meet the fast-rising demand for a creative finish. The three most important factors that have led to attainment of high-performance, attractive tiles and consequently high product sales are (fig. 2): the machines, that have succeeded in meeting ever-more demanding tolerance, automation and performance requisites. extremely attentive selection of raw materials as a function of the increasing sophistication of production machinery and rising demand for high performance tiles. the role of human resources, which, because of reductions in the number of workers per unit of output caused by increased automation, has been modified significantly: todays workers, for example, need specialist training, especially in the electronic control field. Forward planning has become vital: continuous efforts need to be made to identify new technological solutions and, today more than ever, to give ones products a qualitative edge through constant refinement of product aesthetics and functionality. The result of this revolutionary expansion of the ceramic tile industry is well known and has allowed some nations, particularly Italy, to become undisputed leaders in terms of overall output, quality and worldwide exportation. GROWTH FACTORS

MACHINES

RAW MATERIALS

HUMAN RESOURCES

Fig. 2.

14

Introduction

Furthermore, in the field of ceramic tile production Italy is advantaged by the worlds highest concentration of ceramic plant and machinery manufacturers. In Italy alone (fig. 3) 190 such companies employ 7,200 workers and generate total sales of around A 1.6 billion/year (2000). For some years now this huge production effort has hinged on an equally huge technological effort, itself largely dependent on mutually rewarding interdependence with the Research and Development activities carried forward in a state of delicate equilibrium by tile producers and suppliers of production plants, machinery and raw materials. The most successful companies have always been those ready to react quickly to technological and aesthetic trends by keeping one step ahead of them and even controlling them. Even better, some companies have managed to impose the results of their constant application of new ideas on the market; these companies are generally aided by a highly evolved, concentrated manufacturing scenario (e.g. Italy, Spain) and a backdrop of flexible, medium size engineering companies ever ready to follow the developments required of them. Simultaneously, such companies have proved adept at responding quickly to the requisites of new technical or standardisation legislation on mechanical and physical tile characteristics (MOR, frost resistance, water absorption etc.). To illustrate just how the tile manufacturing industry has developed, a brief historical aside on recent technological developments is needed. At the start of the 1970s certain negative aspects vis-à-vis the production process and the final product were successfully eliminated, giving a new process and a product that was more remunerative and suited to a much vaster market; this proved to be good news for tiles manufacturers and suppliers of technology (i.e. presses and kilns) alike. PRODUZIONEPLANT DI IMPIANTI E MACCHINE PRODUCTION / MACHINERY OUTPUT

ITALIA ITALY

AZIENDE COMPANIES EMPLOYEES ADDETTI SALES FATTURATO EXPORT EXPORT

Tableware Stoviglieria Sanitari Laterizi Brick Sanitaryware 0,8 Refrattari Refractories 2,5 10,3 0,3

Fig. 3.

15

190190 7,200 7.200 1.6 billion 3.100 miliardi 65%65%

Piastrelle Tiles 85,5


The way in which the industry evolved towards such innovation was typical of the Italian economy; the industry was able to rely upon a high number of suitably sized companies that were willing and able to experiment the new technology internally. Those companies also had the knowledge, skill and drive needed to interpret and guide the experimentation. As already mentioned, there were also a great many mechanical engineering companies with autonomous research capacity and, finally, an intense exchange of ideas and experiences within the industry. The outcome was that Italian tile manufacturers were the first in the world to gain possession of a new product single-firing ceramic that would prove to be extremely competitive in terms of both quality and price with respect to other floor tile types. Compression of energy and labour costs not only made this new innovation profitable, it also made traditional systems, market niches aside, rapidly obsolete. In this scenario, then, the characterising aspect of output was the ceramic body while aesthetics, conditioned by poor familiarity with the new techniques, was characterised by matt or rustic glazes designed to give an impression of strength. Consequently there was soon a levelling out of both the domestic and export markets for these products. Once again, the quest for qualitative advantage gave new impetus to research into aesthetics, giving rise to the use of new decorations, grains and granulates and the creation of products with a glossier, more intricately decorated finish that required more and more applications and silk screen printings. Technology suppliers reacted accordingly, providing ever-more sophisticated tile pressing and automatic handling machines and giving manufacturers the bonus of reduced waste, improved quality and greater production plant flexibility.

Evolution of ceramic tile plant engineering Ceramic tiles, then, began to gain ever-larger world markets not only because their raw materials are found all over the globe and the production process is considered to be a relatively easy one but, above all, because new technologies based on reliable machines were able to provide finished products of excellent quality. The dynamism of the industry also played a key role, acting as a catalyst for continuous evolution of both process and product. Beginning in the 70s and 80s (depending on the country and the degree of industrial development, see fig. 4) a general switch-over from traditional double firing to rapid double and single firing was witnessed. That changeover was accompanied by the increasingly widespread use of wet grinding, larger and larger tile sizes and fast-improving aesthetic/decorative results; such changes clearly involved important technological changes in mills, presses and glazing lines.

16

Introduction

TILE OUTPUT IN EUROPE (year 2000) (Millions of m2/year) ITALY SPAIN TURKEY GERMANY PORTUGAL FRANCE CZECH REPUBLIC POLAND RUSSIA

632 620 150 65 60 50 31 30 20

Fig. 4.

Grinding Just twenty years ago the majority of ceramic manufacturers still used pendulum mills or pin crushers. Nowadays, as a result of increased output requirements and demand for particle size distributions that pivot around the 10 micron mark, with 63 µm screen residues equal to zero (and 45 µm residues of just a few percentage points), there has been a widespread conversion to large, continuous wet grinding mills. For at least ten years now this tried and tested solution has had all the characteristics of a mature technology. While, in the production of low-performance red body tiles, it might still be possible to consider using dry grinding technology, the increasing demand for light coloured bodies of complex formulas, together with the improved technological characteristics of the finished tile, has oriented the market towards more and more extensive use of wet grinding where the excess water is evaporated in spray driers. These changes, of course, are driven by the improved quality of the semi-finished tiles and the improved production flexibility such technology offers. Continuous wet grinding mills (fig. 5) are also used extensively, and have been for many years, in the worlds mining industries. In addition to technical advantages continuous grinding has also yielded technological, energy and management advantages with respect to discontinuous grinding because: it is possible to weigh and formulate the bodies continuously and automatically, thus providing more consistent physical-chemical slip characteristics and easier repeatability as and when necessary. it is possible to reduce the quantity of water in the mill because continuous emptying and the higher temperature (higher by as much as 60 °C) of the usually thixotropic slip make it possible to use suspensions that are denser but just as easy 17


Fig. 5.

CONTINUOUS GRINDING PLANT

VANTAGGI DELLA CONTINUA ADVANTAGES OF MACINAZIONE CONTINUOUS GRINDING +17.4% +17,4%

INCREASED AUMENTO PRODUCTIVITY PRODUTTIVITÀ

–14.8% -14,8%

ENERGY CONSUMO CONSUMPTION ENERGETICO

–14.8% -14,8%

SPECIFIC CONSUMO ELECTRICITY ELETTRICO CONSUMPTION SPECIFICO

DISCONTINUOUS

–80% -80%

LABOUR PERSONALE REQUIREMENTS RICHIESTO

-50% –50%

REQUIRED AREA SPACE RICHIESTA

CONTINUOUS

MACINAZIONE GRINDING DISCONTINUA

MACINAZIONE GRINDING CONTINUA

Fig. 6.

18

Introduction

to unload: this results in evident energy savings at the subsequent spray drying stage (consumption about 15% lower). while system output capacity has been increased, the grinding department is now more compact than in the past and the low number of employees makes for easier management. Furthermore, as only turbo-dissolved clayey raw materials can be used, continuous grinding can be organised to feed pre-formed slips continuously; in this case the productivity of the system, which then focuses its potential on the harder materials, may increase by as much as 40%. Continuous, wet grinding mills are, of course, more suitable for production plants of considerable size as it is here that the described advantages are greatest. They may also be used in medium size plants as base body preparation units, with subsequent batch corrections; the effective load capacity of the machines currently available on the market ranges from 40,000 litres (24 tons of silica grinding media) to 150,000 litres (100 tons of grinding media).

Pressing Recent product innovations have seen so-called porcelain tiles gain ever-greater market shares. Such innovations have also yielded technological improvements at other stages of the production process. For example, enormous progress has been made in pressing, where machines and accessory systems have been completely redesigned to meet demand for larger tiles without reducing specific pressure (which, has, in truth, increased). The 30 × 30 cm and 40 × 40 cm tiles now account for more than half of total output and sizes of 90 × 180 cm and beyond, mere oddities up until just a short time ago, now form part of the standard product range (fig. 7). This trend towards large tiles means that for the same surface area a smaller number of pieces are handled; also, silk-screen printing and dry grain application techniques make decoration much more effective and, thanks to the extensive surface area of the tile, easier. Such decoration techniques now allow manufacturers to imitate a whole range of natural stones almost perfectly. Moreover, the initial investment is amortised quickly on account of the considerable added value of these products. Such production plants are more exploitable in that there has been no slowing down of production cycles, only increased productivity in terms of square metres produced. The market has accepted these larger tiles willingly, not only because of their superb aesthetics, but also because they are more economic to lay even where used in raised flooring or outdoor building facades (fig. 8) and involve fewer intertile joints. Yet none of this would have been possible without the press and die developments that followed in the wake of fast-changing manufacturer needs. 19


Fig. 7.

Fig. 8.

At present, presses used to make large tiles have very high pressing forces (up to 7,000 tons - fig. 9), with die cavities now rapidly nearing 1000 × 2000 mm. The construction parameters for such machinery must thus take into account the need for rigidity, precision of alignment and absence of torsion etc.: those needs have been met through the application of a various construction solutions capable of producing non-welded, suitably pre-loaded structures. 20

Introduction

Fig. 9.

Filling systems have also changed much in recent years: they now involve a large number of multiple operations and provide excellent filling homogeneity, even with specially shaped die cavities. With regard to the actual pressing phase itself, no summary of recent developments would be complete without mentioning the introduction of pressure-compensating or isostatic dies (fig. 10) in which an incompressible fluid occupies a space separating a rigid part of the punch from a flexible part that comes into contact with the powders to be pressed: during pressing, then, this compensates for the density differences caused by the non-uniformity of the powder filling. ISOSTATIC PUNCH

Fig. 10.

21


The advantages offered by this simple solution (derived from the production of technical ceramics), especially in the case of large tiles, include, among other things, a drastic reduction in sizing variations, the disappearance of a host of other dimensional defects (often evident only after firing) and increased press efficiency.

Porcelain tiles and production line innovations Our analysis of recent production technology innovations would not be complete without mentioning the enormous success of porcelain tiles. This argument does not concern an innovation in the production process per se but, rather, a product innovation that has strongly influenced (and is likely to continue doing so) the wall and floor tile market and the configuration of production plants. This type of ceramic material is defined by its superior technological characteristics and generally falls within ISO 13006 class BIa (fig. 11). In Italy it has conquered a significant slice of the market: in 1985 porcelain tiles accounted for just 8 million square metres while forecasts for 2001 anticipated sales in excess of 300 million square metres. In Italy, porcelain tiles now make up over 30% of total floor tile sales. Since some 62% of such Italian-made products are exported (mainly within Europe), it is fairly safe to assume that European and world demand will continue to grow. The irresistible commercial success of these tiles has had knock-on effects in the production process. A different chemical and mineralogical composition (fig. 12)

PRESSED CERAMIC TILE CLASSIFICATION CATEGORY

WATER ABSORPTION (%)

PRODUCT TYPE

B III

> 10

MONOPOROSA OR RAPID DOUBLE FIRING TILE

B II b

6 – 10

SINGLE FIRED TILE (SEMI-POROUS)

B II a

3–6

VITRIFIED SINGLE FIRED TILE

BIb

0.5 – 3

FROST-RESISTANT SINGLE FIRED TILE

BIa

< 0.5

PORCELAIN TILE

Fig. 11.

22

Introduction

PORCELAIN TILE COMPOSITION Feldspar

Plastic Clay Kaolin Quartz Talc

KAOLIN PLAS. CLAY FELDSPAR QUARTZ TALC

Fig. 12.

has, as seen, necessitated better grinding machinery performance; this is because the high degree of compactness and vitrification that characterises these tiles is dependent not only on accurate raw material selection but also optimum particle size distribution: the latter provides the extensive inter-particle surface area needed for proper sintering at the firing stage. Pressing parameters (and machines), via which compaction and thus reactivity at high temperatures is optimised, have also been influenced. Standard porcelain tile pressures now stand at around 350-450 kg/cm2. Hence the growth in demand for the larger tiles has increased the requirement for larger, more powerful presses. Another key development in the production process associated with the success of porcelain tiles concerns decoration. Given its characteristics of surface hardness and abrasion resistance porcelain tiles were, up until about 3 or 4 years ago, produced without any glaze-like vitreous covering as the latter would have been detrimental to performance. This posed the question of how to obtain an aesthetic result that would free the manufacturer from the relative banality of plain colours and so-called granito ceramics, obtained simply by colouring bodies with appropriate oxides and mixing different combinations of coloured powders together (fig. 13). New decoration techniques have thus been perfected which, without compromising the cited technical performance, have allowed the creation of entirely new product types and led to marked differentiation of products available to the consumer. Such techniques have also allowed porcelain tiles to make inroads in areas traditionally reserved for products of very high aesthetic quality. 23


VARIABLY SHADED PORCELAIN TILE

VARIABLY VEINED PORCELAIN TILE

Fig. 13.

Such innovations have also permitted the re-introduction of glazing lines specially designed for porcelain tiles. Given the possibility and need, as is often the case in full-body products like porcelain tiles, of effecting post-firing polishing to increase the tiles aesthetic qualities, there has been intense development of both application-type decoration and decoration with pigmented substances dispersed in appropriate vehicles (water, glycols, PEG, polycarboxylic acids etc.). The latter are absorbed in a controlled manner into the semi-finished product (tiles that have been pressed and dried or pre-fired at 900-1000 °C), and colour it from within the body. This has resulted in a far-reaching reassessment of aesthetic aspects, allowing, for example, the reintroduction of silk-screen printing decoration (fig. 14), without in any way altering the body and, above all, surface features of the material. Furthermore, final polishing can now be performed as a function of just how far the pigments are absorbed into the ceramic body and as a function of the need to obtain the shading effects that give the finished tile a natural look. These new needs, all aimed at enhancing the aesthetic quality of the product, have stimulated enormous innovation within the accessory machines field; these carry out tasks such as granulation, re-granulation, micronization, multiple die cavity filling on the press, application, decoration. Such needs have also driven intense research into precursors, raw materials, pigment and soluble pigmenting salts etc. (fig. 15). Fast-growing demand for a wider range of products and better aesthetics has, in very recent years, resulted in the widespread reappearance of surface glazing techniques for porcelain tiles; this is largely performed with special glazes or glassceramics that are extremely resistant to chemical aggression, staining and exhibit very good bending strength. This form of decoration, together with a seemingly endless flood of imaginative solutions put forward by designers and applicators, has gone hand in hand with the increasingly diffused utilisation of porcelain tiles on 24

Introduction

PORCELAIN TILE DECORATED WITH SALTS

Fig. 14. TYPES OF POWDER THAT CAN BE PRESSED SPRAY-DRIED DRY GROUND MICRONIZED PIGMENTS RE-GRANULATES FLAKES GRAINS PELLETS

Fig. 15.

walls as well as floors. Porcelain tiles now provide a high quality touch in many public buildings (outdoors and indoors) and are sometime used on external facades where their technical and geometric qualities (porcelain tiles can be ground to size and chamfered) are ideal.

Firing The popularity of large tiles and high-density, almost fully sintered materials has inevitably produced changes in that most critical stage of the production process, firing. Firing has undoubtedly been affected by more technical innovations and plant engineering developments than most other stages of the production process, espe25


cially over the last few decades. As mentioned in the introduction, this has also involved the development of specially formulated bodies suitable for the new firing cycles. The advent of single-layer roller kilns, which began to appear in the mid-70s, shortened firing cycles to 25-65 minutes for a whole range of items, from the simplest double fired product to the densest, thickest porcelain tile body. Moreover, flexibility in terms of load and regulation allowed manufacturers to adjust firing curves to suit tile size and thickness and the glaze type. While the reader will certainly be aware of much more recent developments concerning flexibility, electronic control, energy consumption, productivity and adaptability to certain products, it is nevertheless worth taking a brief look at the latest innovations which have provided manufacturers with even better performance. In particular, there has been much progress in the optimisation of temperature distribution across the cross-section and length of the kiln via the use of tubed or boxed burners (fig. 16). In addition to supplying a mix of convective and radiated heat, these also allow the heat to be directed to more sensitive sections of the kiln, such as side walls or at the centre of the chamber as required. Moreover, these devices make it possible to programme temperature regulation during initial preheating of the tiles by using low temperature air. Fine regulation of cooling zones clearly advantageous where large or high mass tiles are produced is also being developed. The ability to adjust temperatures and times allows for the creation of special cycles that yield certain colouring effects (where iron, manganese and other colouring elements are present) in glaze applications or special body compositions. The innovative technological aspects illustrated here represent just a few examples of a continuous upgrading process that involves everything and everyone in INNOVATIVE BURNERS

Fig. 16.

26

Introduction

the ceramic industry from raw materials to rheological and silk screen printing additives, from glaze manufacturers to tile producers. Suppliers of complete plants and machinery are especially committed in this regard as they play an ever-greater role as vehicles of know-how and are a driving force in the industrys R&D sector. Recent efforts aimed at optimising sorting, packaging and storage lines (fig. 17) illustrate this point perfectly: sophisticated electronic control systems using intelligent vision systems that effect on-line product classification are increasingly common as are automatic laser or wire-guided tile transfer systems etc. (fig. 18). So far nothing has been said about key innovations in glazing and screen printing, now available as continuous systems or about the huge efforts that have been made to bring the industry into compliance with environmental standards; this last aspect is

Fig. 17.

Fig. 18.

27


particularly important in highly industrialised areas and is a key factor when it comes to designing a new product or production line. This second volume examines each stage of the ceramic tile production process, from grinding all the way to sorting and packaging. Its aim is to provide clear, concise, comprehensive and updated explanation of production technology and complete the preliminary information on materials, products and reactions provided in the first volume.

28

Grinding

Chapter 1 GRINDING

Definition and purpose of solid material grinding Grinding involves a whole series of tasks intended to reduce the dimensions of solid raw materials. It begins by crushing the raw materials to produce smaller lumps and ends with the creation of a fine powder. Hence grinding is not just a simple, preliminary affair. It is a complex process that aims to produce a material of a desired average particle diameter and particle size distribution appropriate for a specific final product. Broadly speaking, while there are a variety of reasons for reducing the dimensions of a solid, the most important aspect is that the increase in specific surface area gives the mass a high degree of homogeneity and results in more complete, faster chemical reactions during firing (see relevant chapter in Vol. 1). Properties of solids Solid materials have certain properties that significantly influence the efficiency of the grinding process. The most important of these are: a) Linear dimensions of the particles of the materials to be ground. These concern: the diameter, where particles are spherical the side length, where particles are cube-shaped etc. Clays are usually supplied to ceramic manufacturers as lumps no larger than 10-20 cm. b) Dimensions of the external surface of the particles to be ground Surface area can be calculated easily for spherical or cubic shapes, even those of irregular form. Quarried clay lumps are frequently rounded or laminated; the laminated ones can sometimes be very hard (shale-clays). Other materials are spherical, rectangular or cubic such as the harder materials: calcite, dolomite, feldspar and silica. c) Hardness This is one of the most important parameters in grinding. It is essential to have an understanding of: compression resistance: especially important in the dry grinding of hard materials.

29


impact resistance: comes into play in the dry grinding of clays and wet grinding of hard materials. abrasion resistance: important in the wet grinding of hard materials. d) Material structure May be compact or heterogeneous, with casual fracture planes or well determined cleavage fractures. Clays generally have a compact structure but some often have fracture planes. The ceramic industry also uses schistose clays with evident, well defined cleavage fractures. Hard materials are generally compact; these include materials such as quartzites, feldspars, chemical sedimentary limestones and highly fissured metamorphic rocks such as homogeneous-sedimentation calcites and dolomites that have not been excessively modified by diagenesis. e) Specific weight Not a particularly important factor in the grinding of compact solid materials. However, in the case of natural mixes or blends with particles of different mineralogical composition, specific weight takes on more significance regarding the possibility of segregation during transportation within the plant. There follow some specific weights for several ceramic materials: Clays Quartz Feldspars Calcite Dolomite

from 2.5 to 2.8 2.65 from 2.53 to 2.67 2.75 2.95

g/cm3 g/cm3 g/cm3 g/cm3 g/cm3

f) Moisture content and hygroscopic characteristics Both exert considerable influence on grinding in that they reduce the efficiency of the grinding machinery. As moisture content rises grinding output capacity falls (for the same fineness). g) Tendency to agglomerate or flocculate Reduces grinding efficiency. Other factors which influence grinding considerably, especially where ceramic materials are in their natural state, are shown in fig. 1. What happens during grinding; choosing the right machine The grinding process involves the following: a) simple compression (crushing) b) percussion c) impact 30

Grinding

MORPHOLOGICAL AND PHYSICAL CHARACTERISTICS OF CERAMIC RAW MATERIALS INFLUENCING THE GRINDING PROCESS CLAYS • MINERALOGICAL NATURE • PLASTICITY • ELECTROSTATIC INTERACTION WITH WATER

THESE CHARACTERISTICS ARE IDENTIFIED VIA ANALYSIS OF: • PERCENTAGE AND NATURE OF RESIDUAL MATERIALS • CLAYEY PARTICLE DIMENSIONS • SPECIFIC SURFACE AREA • IONIC EXCHANGE CAPACITY • PRESENCE OF SOLUBLE SALTS

NON-CLAYEY MATERIALS • SIZE AND PARTICLE SIZE DISTRIBUTION • BASE STRUCTURE Fig. 1. Morphological and rheological factors influencing grinding.

d) abrasion e) cutting. All the machines used in the grinding process involve the above-listed principles. Comminution can be sub-divided into two main phases: crushing, the range of which extends from the rough blocks extracted in the quarry to pieces just a few millimetres in size. Crushing itself may be divided into primary crushing (or pre-crushing), effected on materials that range from the tout-venant quarry blocks to pieces about 100 mm in size, and secondary crushing that produces grains with a size of about 10 mm. grinding, instead, has a field of application that extends down into the microns. This stage of comminution is divided into primary grinding, which involves pieces as small as 0.5 mm, secondary (or fine) grinding, which yields particles with a size of some tens of micrometers, and final micronization, which reduces the particles to a size of just a few microns (fig. 2). 31


BASIC GRINDING OPERATIONS PEZZATURA SIZE

• CRUSHING FRANTUMAZIONE:

IN 30-20 mm

OUT 5-10 mm

- FRANTUMAZIONE – PRIMARY CRUSHINGPRIMARIA

100 mm

- FRANTUMAZIONE SECONDARIA – SECONDARY CRUSHING

5-10 mm

• GRINDING MACINAZIONE:

IN 5-10 mm

MAYPUÒ BE SUB-DIVIDED INTO:IN: ESSERE SUDDIVISA – PRIMARY GRINDING - MACINAZIONE PRIMARIA

OUT < 50 µmm ~ 0.5 mm

- MACINAZIONE SECONDARIA – SECONDARY GRINDING (FINE)(FINE)

20-100 µm < 50 µm

- MICRONIZZAZIONE – MICRONIZATION

Fig. 2. Grinding: different stages and relative piece/particle sizes.

In the case which is of most interest to us (i.e. grinding of raw materials for ceramic tile bodies), it should be born in mind that highly heterogeneous materials are often used; these may be heterogeneous from both a mineralogical and a physical standpoint. Certain necessities thus need to be taken into account: dispersion of the various components in the clay or mix must be all but perfect. some mineralogical components need to be ground to different degrees. certain impurities contained in the raw materials need to be removed from the mix before grinding it. Dry grinding and wet grinding The raw materials used in ceramic bodies can be ground using either dry or wet grinding technology. As a rule it can be said that wet grinding is preferred as it reduces in-mix particle size greatly and provides better homogenisation. Dry grinding technology may be used where raw materials are homogeneous from both a morphological and hardness standpoint, when the final product is of modest quality or with double-firing products in general. With wet grinding the raw materials are initially ground together as a suspension of solids to reduce the size of the natural particles even further. The substance obtained at the end of this grinding process is called slip. Chemical deflocculants (which can also reduce the quantity of water in the slip, thus providing economic gains) allow manufacturers to produce particles with a diameter of even less than one micron (almost zero residue on a 63 µm screen). 32

Grinding

The choice of grinding technology, however, depends only in part on the degree of particle fineness needed to obtain a certain final product; other factors need to be taken into account. Dry grinding, in fact, is used for mixes made up of a maximum of two or three clays that are similar to each other in terms of both mineralogical constitution and physical characteristics. Any silt, sand or coarse residue must only be present in very limited quantities. Dry grinding can thus be used to produce double fire biscuit (i.e. to be fired a second time) and in the production of wall tiles (majolica). Its application in the production of floor tiles (e.g. porcelain tiles) or monoporosa is somewhat limited as wet grinding is generally preferred. Wet grinding is used with natural mixes of clays and hard materials, that is, with heterogeneous blends sized in such a way as to make their refinement with even the most efficient dry grinding systems impractical; bodies are, above all, prepared using wet grinding technology where ceramic compositions consist of various components, specific weights and particle sizes. Wet grinding is also preferable where clays contain impurities that need to be eliminated from the bodies: the preferred procedure is to disperse them and then sieve the slip through a suitable mesh. Finally, wet grinding is preferred for the manufacture of vitrified products or those utilising very fast firing cycles, in that the wet process makes composition corrections easier and provides powders that, once spray dried, have good fluidity. Fluidity is important as it ensures good die cavity filling performance at the pressing stage (fig. 3). From the above, then, it is evident, from a purely technological viewpoint, that there can be no real competition or dualism between dry and wet grinding: they exist only as separate alternatives. Once the raw materials have been analysed and the technical characteristics of the tile to be produced have been taken into account, only one technology will prove feasible. Machines commonly used in the grinding department Dry grinding plants a) Clay grinding Ceramic tile manufacturers generally use clays with a maximum moisture content of 4-5%; in mills heated with hot air (which performs a drying function), this figure can be as high as 10%. Primary crushing reduces the size of the incoming materials from about 20 cm to about 6 cm max. This is normally done in pin-type or toothed-wheel crushers. Subsequent crumbling (secondary crushing) is performed in impact crushers; here, the material is reduced to a maximum particle size of 5 mm. Grinding itself is carried out in centrifugal pin mills, fixed or mobile hammer mills and, above all, in pendulum mills, depending on the required degree of particle fineness. 33


FACTORS INFLUENCING THE CHOICE OF PROCESS (DRY OR WET GRINDING)

-

TECHNICAL CHARACTERISTICS OF THE FINISHED PRODUCT TYPE OF RAW MATERIAL QUANTITY OF COMPLEMENTARY COMPONENTS MORPHOLOGICAL CHARACTERISTICS OF AVAILABLE RAW MATERIALS - COST OF FINISHED PRODUCT - OVERALL INVESTMENT Fig. 3. Factors determining the grinding method.

b) Grinding chamotte or other hard materials Materials with a maximum moisture content of 1.0-1.5% are used. Primary crushing is done in jaw crushers; starting with a piece size of about 1520 cm, the material is reduced to a tout-venant with a maximum particle size of 4-6 cm. Subsequent granulation is effected in impact mills where maximum particle size drops to 0.5-1.0 cm. Final grinding is generally effected in annular gap mills where medium particle size distribution is required or in pendulum mills should greater refinement be required. Body preparation (dry grinding): machines In defining the machines needed for the grinding process and, more generally, those needed to prepare powders for the pressing of ceramic tiles, three essential aspects need to be taken into consideration: 1) the nature of the material to be ground, that is, the type of component (e.g. clays and, chamotte, etc.) and its relative physical characteristics (size, moisture content, hardness and impact strength). 2) the final particle size distribution to be attained. 3) the technology adopted for the transformation of the ground material into a powder suitable for pressing: wetting and any re-granulation (or, better, agglomeration). From the above it follows that no one machine is able to perform all the required functions: production plants therefore use a range of machines to deal with specific situations as and when they occur.

34

Grinding

The key machines in dry grinding are: Pre-crushers PIN-TYPE LUMP CRUSHERS Crushers JAW CRUSHERS IMPACT CRUSHERS HAMMER MILL Refiner mills PIN CRUSHERS ANNULAR GAP MILL PENDULUM MILL VERTICAL ROLLER PULVERIZER a) b) c) d) e)

As stated previously, grinding involves the following: simple compression percussion impact abrasion cutting.

All the machines used in the grinding process perform one or more of the abovelisted actions. Fig. 4 shows particle size distribution ranges for the dry ground powder and compares it to the granulate obtained from spray drying. In dry technology, regulation of the degree of grinding depends on the classification system (dynamic air flow sizing for pendulum mills and vertical roller pulverizers, mesh for pin mills). Note that the machine efficiency decreases over time because the grinding media is subject to wear. Such wear progresses at a rate proportional to machine productivity, which is, in turn, dependent (except for vertical roller pulverizers) on mill RPM. Wet grinding plants Raw materials generally have a maximum moisture content of 14% for clays and 6% for the harder materials and are reduced in size as follows: for clays: maximum piece size 6-10 cm for hard materials: maximum piece size 0.3 cm. The clays can be dissolved with the aid of high speed or turbo dispersers and continuous or discontinuous drum mills. Hard raw materials, instead, are normally ground in continuous or discontinuous drum mills. 35


Fig. 4. Particle size range of semi-finished products with dry and wet grinding.

The extent of grinding varies enormously depending on the type of product being manufactured. Wet grinding theory There are various grinding theories, all of which tend to explain machine operation in terms of laws expressed as mathematical formulas; the most important of these are Kicks law and Rittingers law: a) Kicks law: states that the energy needed to crush a solid material to a specified fraction of its original size is the same, regardless of the original size of the feed material. This energy is also proportional to the logarithm of the ratio between the initial and final diameters of the material. The law can be expressed mathematically as follows: W = κk lg (D/d) where W = required energy Kk = constant dependent on type of material D = average size of particles before grinding d = average size of particles after grinding. For example, the quantity of energy needed to reduce a given weight of material from 1 cm3 to 0.5 cm3 is equal to the energy needed to reduce the same weight of material from 0.5 cm3 to 0.25 cm3 and so on.

36

Grinding

b) Rittingers law: states that the energy needed to reduce the size of a solid particle is directly proportional to the resultant increase in surface area. The law can be expressed mathematically as follows: W = κr (1/d - 1/D) where W = required energy Kr = a constant dependent on particle shape and energy per unit of surface d = average side length of particles after grinding D = average side length of particles before grinding. In other words, the quantity of energy needed to reduce the size of a given weight of material still depends on the initial and final size of the ground product. For example, it can be estimated that, if it takes 2.5 hours to grind the material from 1 mm to 100 µm (reduction ratio 10:1), in another 2.5 hours the material will pass from 100 µm to just 53 µm (reduction ratio 2:1) because the change in surface area remains the same. It should, however, be observed that, in practice, the amount of energy absorbed by the machines is always greater than the amount of energy calculated via the above formulas because the work also involves: 1) Energy needed to overcome the cohesion linking the particles in the pieces being ground. 2) Deformation energy (plastic and elastic deformation). 3) Energy absorbed in attrition between grinding media. 4) Energy absorbed by vibration. 5) Energy lost as heat. More recently a further theory has been developed by Bond. He observed that the energy required to reduce a material from an initial size of d0 to a final size of d1 is defined by the difference between the total quantities of energy needed to pass from a theoretically infinite size to sizes d0 and d1 respectively. This evaluation has given rise to the identification of a Work Index: an example of values obtained semi-experimentally is given below: Calcined clay 1.43 Glass 3.08 Raw clay 8.16 Pyrite 8.90 Dolomite 11.31 Limestone 11.61 Feldspar 11.67

Quartzite 12.67 Spodumene 13.70 Nepheline syenite 14.90 Silica sand 16.46 Magnesite 16.50 Basalt 20.41 Mica 134.50

37


These figures give a good idea of how the grindability of a material depends not only on its hardness or abrasion resistance. Discontinuous wet grinding theory with Alsing or ball mills The purpose of wet grinding is, as mentioned earlier, not only to reduce the particle size of the raw materials in the bodies to a few microns but also to achieve perfect homogenisation and dispersion of the various components throughout the mix. Wet grinding is almost exclusively performed in Alsing or ball mills (fig. 5). This type of mill has a discontinuous work cycle. The entire work cycle is divided into three phases: 1) Loading of raw materials, water and rheological additives (usually deflocculants). 2) Grinding. 3) Unloading of the obtained slip. Proper grinding and good mill performance depend on the observance of basic rules and the understanding of some basic concepts. The most important of these are: a) Mill speed A mill of diameter D metres, rotating at n revolutions per minute has a peripheral speed of: v = (π/60) D n = metres/sec.

Fig. 5. Discontinuous Alsing or ball mills.

38

Grinding

The speed of the mill exerts a centrifugal force on the pebbles (i.e. the grinding media) lying against the drum wall; this force raises the pebble of mass m by an angle α with respect to the horizontal: this is the point at which the force of gravity is equal to the opposing centrifugal force. cf = m v2 (D/2) = mgsen α This equality indicates that the angle to which the pebble is raised is independent of the mass m of the pebble and depends only on the rotation speed of the mill. Once the lift angle α is exceeded the pebbles up against the drum wall fall away from it because, at this point, gravity overcomes the centrifugal force. They thus fall back into the mill, tracing a parabola as they do so. The angle β that the pebbles achieve beyond the lift angle α is known as the cascade angle. The pebbles that are not against the drum wall but located closer to the axis of rotation have a lower peripheral speed than the outermost ones, are not lifted as high and are enveloped by more external layers as illustrated in fig. 6. The overall motion of the grinding media in a mill thus consists of a cascade drop plus a rolling action: these two movements cause grinding. The angle of lift a can also be expressed as a function of mill RPM: v = (π/60) D n metres/sec and centripetal acceleration expressed by the formula: ca = 0.0055 D n2 m/sec2 therefore sen α = (ca/9) = 0.000561 D n2 Where acceleration ca is considerably lower than g the load is raised only a little above the horizontal and the pebbles tend to slide backwards on the mill lining; grinding action is low and the pebble wear rate high (fig. 7). Tests have shown that grinding is optimised at an angle of 45° - 60° because the pebbles then fall back into the drum in cascade mode, roll against each other and thus maximise grinding; moreover, wear on the pebbles and the mill lining is minimised. Where acceleration ca is approximately equal to g, the pebbles closest to the mill lining stick to it (partial centrifugation) while the inner layers of grinding media trace a parabola and fall back against it. The produced grinding action is still high but wear on both grinding media and mill lining is intense. The pebbles closest to the lining will, of course, continue to orbit the mill without ever falling as they are subject to a centrifugal force equal and opposite to the force of gravity. The peripheral speed at which partial centrifugation begins and pebbles start sticking to the lining is called critical speed (cs).

39


Fig. 6.

Fig. 7.

40

Grinding

When ca is sufficiently stronger than g even the innermost layers of grinding media will adhere to the mill lining as centrifugal force overcomes gravity (total centrifugation). Under these circumstances grinding does not take place at all. The above leads us to conclude that a progressive increase in mill speed (i.e. RPM) initially results in increased grinding efficiency; then, as centrifugation comes into play, further increases in mill speed diminish grinding efficiency. There is thus an optimum mill speed at which grinding time is minimised (fig. 8). Nevertheless, optimum mill performance does not correspond to minimum grinding time but is, rather, achieved at that speed at which the energy lost in friction and impact is minimised and the energy spent on actually grinding the product is maximised. This occurs when the grinding media fall into the mill in cascade mode at an angle of 45° - 60°. As experienced production personnel will be well aware, the noise made by the grinding media can provide important clues as to whether the mill is running at the right speed. Tempo di time Grinding macinazione

VelocitàMill del speed Mulino

Fig. 8.

Fig. 9 shows critical and correct speeds as a function of the internal diameter of the mill. All the mills that run according to the graph in fig. 9 have a constant centripetal acceleration = 50-75% of gravitational acceleration (g). However, to apply the same acceleration values to mills of different diameters is wrong; large-diameter mills are more efficient and acceleration can, and should, be slightly reduced to make the lining and grinding media (subject to more stress than in smaller mills) last longer. Mills are thus run not at constant centripetal acceleration but at constant peripheral speed; at most, such speed will vary within a very tight range. For example: 41


Giri RPM

metres metri

min’

min’

130

180

120

170

110

160

100

150

_ 1 Vc = 133 √D m/min

90

140

2 Vk = 75% Vc

80

130

70

120

3 Vk = 50% Vc _ 4 nc = 42.3/√D

60

110

50

100

40

90

30

80

20

70

10

60 0.5

1.0 1.5 2.0 Diametro utile deldiameter Mulino (m) in metri Effective mill

5 nk = 75% nc 6 nk = 50% nc

2.5 metres metri

Fig. 9.

a) for low and medium density grinding media (2.4 - 2.7 g/cm3) Vk = 95 - 125 m/min, that is: nk = Vk/πD = (30 - 40)/D RPM b) for high density grinding media

(3.4 - 3.5 g/cm3) Vk = 85 - 95 m/min, that is: nk = Vk/πD = (27 - 30)/D RPM.

Fig. 10 shows these values. Of coarse, the above should be taken only as a rough guide; in the final analysis, it is the cascade angle (which should always be around 45°) rather than the mill speed which counts. The cascade angle now depends not only on the peripheral speed but also the type of product (solid or liquid) and, where liquid suspensions are being ground, the density and viscosity of such suspension. b) Grinding media Mill rotation generates not just the overall cascade motion of the pebbles but also reciprocal rotation between them. 42

Grinding

RPM Giri

metres metri

min’

min’

130

130

120

120

110

110

100

100

90

90

80

80

70

70

60

60

50

50

40

40

30

30

20

20

10

10 0.5

1.0

1.5

2.0

Vk = 125 m/min

Vk = 96 m/min Vk = 85 m/min _ 1 nk = 42,3/√D RPM _ 2 nk = 40/√D RPM _ 3 nk = 30/√D RPM _ 4 nk = 27/√D RPM A Porcellana-steatite Porcelain-steatite (densità (density 2.4 2,4+2,75 - 2.75gr/cm gr/cm33)) B Sintered Allumina alumina sinterizzata (density (densità 3.4+ 3,4+ gr/cm3)

2.5

Effective mill diameter (m)

Fig. 10.

Such rotation is particularly grinding-effective where grinding media consists of spherical or cylinder-shaped balls. Much less efficient are silica pebbles on account of their irregular shape, which makes rolling discontinuous and heterogeneous (fig. 11). Rolling and attrition between pebbles of different morphologies The most important aspect of grinding media is its specific weight (normally referred to as density). Classification is as follows: a) low density media (specific weight = 2.4 - 2.7 g/cm3), such as standard porcelain or silica. b) medium density media (specific weight = 2.7 - 3.0 g/cm3): such as steatite, porcelain with a high alumina content etc. c) high density media (specific weight = 3.4 - 3.6 g/cm3), such as sintered alumina, alubit©. The higher the specific weight, the greater the kinetic energy freed during rolling and falling (grinding media volume being equal) and the more effective the grinding action. Moreover, in wet grinding, because of hydrostatics, the force of gravity is proportional to the difference pc-ps (between the specific weight of the media and the 43


Fig. 11.

specific weight of the product). The greater this difference, the greater the grinding action. Fig. 12 shows how the specific weight of the grinding media influences grinding times. c) Grinding media load (pebbles) The amount of grinding media loaded into the mill has a strong influence on grinding times; if the quantity of pebbles in the mill is increased progressively, grinding time initially falls and reaches a minimum when the pebbles fill about half the mill. Beyond this point grinding times start to increase again. Reciprocally, absorbed power initially increases and reaches its peak where the grinding media occupy about half the mill and then starts to drop (fig. 13). According to P.E.I. minimum grinding time corresponds to a pebble load of about 60% of mill volume. In practice, though, the recommended pebble load is 5055% (i.e. apparent volume of pebbles equal to 50-55% of mill volume). % di a 10,000 10.000 mesh/cm 2 of Residuo residue at Maglie/cm3 100 90 1 1 2 2 3 3

80 70 60 50 40

Porcellana normale (densità = 2,4) Residuo porcelain = 0 dopo (density 24 ore = 2.4) Standard Steatite=(densità = 2,6-2,7) Residue 0 after 24 hours Residuo = 0 dopo 20 ore Steatite (density = 2.6-2.7) Porcellana alto Residue = 0 ad after 20contenuto hours di Allumina Porcelain with(densità high = 2,7-2,76) Residuocontent = 0 dopo 16 ore= 2.7-2.76) alumina (density Residue = 0 after 16 hours

30 20 10

2

4

6 8 10 12 Tempo di macinazione in Ore Grinding time (hours)

Fig. 12. Influence of specific weight of grinding media on grinding time.

44

14

16

Grinding

Grinding time

Absorbed power Empty

Full

Level of pebbles as % of mill diameter

Fig. 13. Grinding times and absorbed power as a function of grinding media load.

The corresponding level of pebbles is 50-54% of mill diameter as illustrated in table n. 1 and fig. 14. Note n. 1 The angle between the two tangents at the pebble-particle contact points is called the grip angle; it depends on the diameters S and p (of pebbles and particles respectively, see fig. 15). Calculations show the optimum grip angle to be about 17°: this means that the diameter of the pebbles should be 90 times the average diameter of the particles. For example, where particles have an average initial diameter of 0.7 mm, the pebbles should be approximately about 60 mm in diameter. After a certain time, though, these pebbles will no longer provide efficient grinding. An assortment of pebbles is thus needed: the large pebbles grind the larger particles into smaller ones and the small pebbles refine those smaller particles even further. Pebble diameter should not, in any case, exceed 60-70 mm. The real volume accounted for by the pebbles in a completely full mill is only about 60% as the remaining 40% consists of the gaps between them. Note n. 2 This data (tab. 1) refers to a stack of pebbles of a differing diameter that is intermediate between real volume cubic stacking = (π/6) 100 = 52.4% (stacking by minimum density) and real volume tetrahedral stacking = √(2π /6) 100 = 74% (stacking by maximum density) (for pebbles all having the same diameter). 45


Percentuale volume del Percentage del of mill volume occupied mulino occupato dalla carica di by pebble load biglie

Percentuale delofdiametro Percentage diameterdel of mill taken mulino raggiunto dalla carica up by pebble load delle biglie

5

10

15

21

25

30

35

38

45

46

55

54

65

62

75

70

85

79

95

90

100

100

Tab. 1.

55% of volume 50% of volume

50-54% of mill diameter

a) Apparent volume of pebbles = 55% b) Free volume of mill = 44%

b) a) 54% D

c) Real volume of pebbles = 33% d) Volume of gaps between pebbles = 22%

Fig. 14. Mill filling diagram.

46

Grinding

Fig. 15.

It thus follows that for a pebble load occupying 50% of mill volume, the real volume of the pebbles is approximately 50 × 0.6 = 30% and the volume of the gaps between the pebbles approximately 50 × 0.4 = 20% of mill volume. Summing up: Apparent volume Real volume Volume of gaps Free volume 50% 30% 20% 50% 55% 33% 22% 45%

pe Ap bb pa le re vo n t lu m e le bb pe e l a m Re volu

Real pebble volume

% of mill volume

Volume of gaps

Volume of product above pebbles

Free volume

Apparent pebble volume, real pebble volume and the level of the spheres in the mill are interconnected as illustrated in fig. 16.

% of mill diameter reached by level of the pebbles

Fig. 16.

47


Once the load volume occupied by the pebbles is known it is easy to calculate the weight per unit of volume, i.e. the so-called load ratio. Load ratio = Real vol. % × specific weight (g/cm3) A load occupying from 50 to 55% of mill volume gives, depending on specific weight, the following load ratios: Pebbles in

Specific weight g/cm3

Real volume %

Load ratio kg/m3

Low density porcelain

ð

2.4

ð

30-33

ð

720-800

High density porcelain

ð

2.7

ð

30-33

ð

810-900

High density alumina

ð

3.4

ð

30-33

ð

1080-1120

Fig. 17.

These load ratios are not always observed. It is common for mills to be loaded at much lower ratios, such as 500 kg/m3, according to the well known formula: pebble load in kg/m3 = mill volume in litres/2. This ratio is much too low because even with high density porcelain the real pebble load is just: 500/2.4 = 208 litres/m3 = 21% This means an apparent volume of approximately 35% (as opposed to 50%) and a level of about 37% of mill diameter (13% below the half-full line). This ratio (even less favourable where medium and high density grinding media are used), involves longer grinding times and wears both pebbles and lining faster. Pebble size/assortment must be chosen on the basis of the following factors: a) Mill size: mills of small volume must be loaded with a smaller-sized assortment of pebbles than larger mills. An assortment of three different pebble sizes is generally used, as follows: 45-50% smaller diameter pebbles 25-30% intermediate diameter pebbles 25-30% larger diameter pebbles. b) Particle size distribution of the substance to be ground and the ground substance: where the substance to be ground is made up of coarse particles large-diameter pebbles will be needed. If that substance is to be fine-ground smaller-diameter pebbles will be needed to increase the number of contact points and the total surface area of the pebbles themselves. 48

Grinding

Table 2 gives contact/surface area data for pebbles of different diameters. c) Specific weight of the substance to be ground (wet grinding): in wet grinding, where the specific weight of the product is high (e.g. greater than 2.5 g/cm3), high density pebbles must be used; thus the difference 3.4 - 2.5 = 0.9 is in the same order as that between a pebble of specific weight 2.4 and a product of specific weight 1.5. d) Product load. As with pebble load, there are no hard and fast rules, only general guidelines. The product (wet or dry) must at least cover the pebbles completely. Quantities any lower than this would be ground too quickly and could overheat. Moreover, wear on both pebbles and lining would be intense. The lower product load volume limit is thus given by the volume of the gaps between the pebbles. For example, where pebble load = 50% of mill volume the product must account for at least 20% of mill volume, that is (20/100) 1000 = 200 litres per m3. However, there should really be more product than this so as to cushion the pebbles cascading back into the mill. Bear in mind that the more excess product is added, the longer the grinding time. Fig. 18 shows how grinding time is influenced by product load P, expressed as a percentage of mill volume. The upper product load limit is given by the minimum free space that must be left in the mill; this space represents approximately 25% of drum volume and the upper product load limit is thus about 20 + 25 = 45% of mill volume. Of course, if the mill is this full grinding times will be lengthy and grinding output (kg of product ground per hour) low. In some cases, though, manufacturers are not so much concerned with maximum grinding output as they are with the fact that the mill has to be run for a certain time, perhaps because the product so requires or simply because the factory does not always have personnel at hand to empty and reload the mill; hence it often makes more sense to use the available mill volume completely. Diameter of pebbles in mm Surface area of pebbles in cm

2

Volume of pebbles in cm3 3

20

30

40

50

60

12.56

28.27

50.26

78.54

113.10

4.19

14.10

33.50

65.25

113.00

Number of pebbles per dm (1)

143.10

42.57

19.91

9.19

5.31

Number of contacts per dm3 (2) Surface area of pebbles in m2 per m3

858.70

255.42

107.46

55.15

31.85

179.73

120.34

90.01

72.20

60.05

(1) Calculated according to real volume = 60% of apparent volume, which represents a practical load factor, owing to the average between best and worst possible pebble settling. (2) Calculated according to tetrahedral stacking (with 6 contacts per pebble).

Tab. 2.

49

Grinding time


Product load (% of mill volume)

Fig. 18.

e) Lining and grinding media Both lining and the grinding media can be made of several different materials. The table below shows the most commonly used lining/media combinations for ceramic body and glaze mills. MILLS FOR BODIES Lining Silica Alumina Rubber Rubber

Grinding media Silica Alumina Silica Alumina

MILLS FOR GLAZES Lining Grinding media Porcelain Porcelain Alumina Alumina Such materials generally have the following specific weights SILICA: 2.65 kg/l PORCELAIN: 2.75 kg/l ALUMINA: 3.57 kg/l f) Conclusions The concepts described so far may be summed up as follows. The main factors influencing grinding are: mill speed. 50

Grinding

pebble load (load ratio, assortment, type). product load (load ratio, type, initial and final particle size distribution). Rational use of a mill involves regulation of the above factors so as to minimise the energy absorbed in the wear of the lining and the grinding media and the energy dissipated as heat generated by friction and attrition. These optimal conditions are achieved when: mill speed produces falling-rolling of the pebbles; peripheral speed should approximately be: 95-125 m/min for low density pebbles 75-85 m/min for high density pebbles. The pebble load level should be half the diameter of the mill or a little more (5055% by volume, that is 50-54% by diameter). This involves the following load ratios: 700-800 kg/m3 of low density pebbles (2.4) 800-900 kg/m3 of medium density pebbles (2.7) 1000-1100 kg/m3 of high density pebbles (3.4). The assortment of pebbles must be selected as a function of the initial and final particle size distribution of the product; mill size also needs to be taken into account. As a rough guide, the assortment for silica pebbles should be as follows: 45-50% of smaller diameter pebbles (20-30 mm) 25-30% of intermediate diameter pebbles (40-50 mm) 20-25% of larger diameter pebbles (50-60 mm). There should be at least enough product to fill all the empty gaps between the pebbles (20-22% of mill volume): it is highly advisable to add some excess product (up to a maximum of 25% with respect to mill volume). Practical load calculation for discontinuous wet grinding mills An exact calculation of the quantity of raw material to be loaded in the mill requires certain information; this is usually found in the mill instruction manual or the raw material test lab reports. The required data is: 1. Effective mill capacity This is the space available to the raw materials, water and grinding media. Expressed in litres. 2. Slip density to be prepared in the mill. Expressed in kg/l. 3. Content of solid material in the slip. Expressed as a percentage.

51


The load for body grinding can be calculated with the following formula: Kg of dry material = 0.55 (0.67) × Vu × d × y/100 where Vu = d = y = 0.55 = 0.67 =

effective mill volume slip density kg/l dry material percentage in slip filling coefficient of mills with silica grinding media filling coefficient of mills with alumina grinding media.

Calculating the load for a discontinuous mill (example) The mill to be loaded and the slip to be prepared have the following characteristics: mill capacity = 30,000 l silica lining silica grinding media effective volume 24,000 l slip density 1.6 kg/l dry material in the slip 64% (water = 36%) Kg of dry material = 0.55 × 24,000 × 1.6 × 64 / 100 = 13,500 If the body composition is: Clay A 10% Clay B 23% China Clay 12% Feldspar C 17% Feldspar D 18% Feld. sand 20% the quantity of each dry raw material can be obtained from the proportion Weight of dry material X = (% of dry material X in the body) x (Mill dry load) i.e. where X is china Clay its weight is 12% of 13,500 = 1,620 kg Since raw materials stored in the factory are seldom dry, it will be necessary to determine moisture content and bear this in mind during batching via the proportion Wet batch weight = Dry weight × 100 / (% of dry content in raw material) This calculation provides the sum of the weights of the individual raw materials which, depending on their moisture content, will exceed the theoretical weight cal52

Grinding

culated as mill capacity: loading moist raw materials means loading water. The presence of such water will have to be taken into account when calculating the quantity of grinding water. This is calculated as follows: 36 (% H2O) : 64 (% raw materials) = Z (theoretical litres or kg of water to be loaded) therefore: Z - [(Ytot Total weight moist materials) (Xtot Total weight dry materials)] = kg or litres of H2O to be added Rheological additives (deflocculants) are normally added together with the water in appropriate quantities, calculated in the laboratory as described in Volume 1; in any case such quantities represent only a small percentage (0.2-0.4%) of the total weight of the dry raw materials. Continuous wet grinding Recent years have seen a steady increase in the number of continuous wet grinding plants being installed. These plants use tubular mills. Raw materials (body components and water) are fed continuously from one end of the mill while the slip exits from the opposite end. The continuous wet grinding plant The first section of the plant consists of a continuous computer-controlled weighing and batching system. Each of the different raw materials making up the body composition is stored in a silo. Each of these silos features an extractor which regulates, on the basis of a signal, the quantity of material released onto the weighing conveyor belt. The weighing belt measures the difference between the actual outflow and the set-point and transmits it to the computer. In the event of an anomaly the computer modifies the release rates to ensure that the ratio of components and thus the body formula remains constant. The thus-batched raw materials merge together on a conveyor belt which carries them to the mill pre-loading silo; the latter has a filling cycle regulated by a level control sensor. A mechanical extractor unloads the material from the silo constantly and places it on a weighing conveyor which compares silo outflow rate with the mill infeed setpoint and corrects any fluctuations by adjusting conveyor speed (fig. 19). Just before the mill feed screw there is a device that pre-mixes the raw-material and the deflocculant with the aqueous suspension containing re-circulated screening residue. 53


GRINDING DEPARTMENT: FLOW DIAGRAM

Fig. 19.

At the mill outlet the slip is first passed through a sieve with a mesh size of a few mm to trap any large, solid lumps (including small pebbles); a smaller-mesh sieve set follows. The sieving residue is carried away by the pre-batched water to be recycled into the mill while the slip is sent to a constantly stirred holding tank. From here it is pumped directly into the spray drier or put through a colouring process. Generally, all the raw materials (clay and non clay) making up the body are fed at the pre-mixer stage and subsequently to the mill. However, in some cases a part of the clay (generally the wettest most plastic part) and any colourless unfired scrap is dispersed in a turbo mixer. Possible system configurations associated with this technology (fig. 19) are, essentially: configuration in which a part of the clayey components by-pass the mill; the mill is fed with a mix rich in components that are hard with respect to the formulation of the body, but containing a certain percentage of plastic raw material for suspension purposes. configuration with pre-dispersed clayey materials (and any unfired scrap), fed into the mill after mixing with the other slip components. These solutions are particularly appropriate in cases where the body has a substantial proportion of very plastic, very moist clays (moisture greater than 15-20%), 54

Grinding

as they are difficult to mix with other body components and can lead to aggregation phenomena which reduce grinding efficiency. Continuous grinding technology Constant research into innovative control equipment has resulted in the development of precise, highly reliable weighing and batching systems (raw materials, deflocculants, water). Such microprocessor-controlled systems have led to the introduction of continuous mills in the body preparation plant. Consequently, manufacturers are now able to provide excellent consistency of slip characteristics such as composition, water content, density, viscosity and, last but by no means least, grinding residue. The reasons behind the ceramic tile industrys adoption of the continuous process already used in the mining industry extensively are twofold: technological and technical-managerial. From a technological standpoint the most immediate advantages vis-à-vis product quality are: greater consistency of slip characteristics thanks to automated in-mill batching of the various materials (body components, water and deflocculants) that are much more reliable than manual procedures. greater grinding media efficiency thanks to optimisation of pebble assortment in the different chambers and a grinding media/loaded material weight ratio of at least 2.5:1 (for a 3-chamber mill in silica - silica - alumina) compared to a maximum of 1:1 for discontinuous mills with high density alumina media. as a result of the above, grinding times have been reduced drastically: for porcelain tile bodies, which contain a significant amount of hard, comminutionresistant materials, grinding times vary from 2 to 3-4 hours (depending on whether the pebbles are made of alumina or silica). This is a huge gain on the 10 hours or more (up to 30 hours with silica media) associated with discontinuous mills - improved rheological characteristics of the slip, thus making sieving easier even when density is increased and slip water content drops by 2-3%; the latter can be sieved directly at the mill outlet without having to interrupt agitation, thus preventing the thixotropic phenomena which can disturb sieving; furthermore, slip outlet temperature, on average 20-30 °C higher than in discontinuous mills, keeps viscosity within the required range (about 3 °E) and ensures good fluidity. From a technical-managerial viewpoint the main advantages are: approximately 15% less heat energy is required in the spray-drying process because water content is reduced by about 2% and the slip is warmer (20 °C approx.): the former translates into latent heat savings and the latter in sensible heat savings. 55


thanks to reduced grinding times, machine-specific productivity increases by up to 70% per unit of volume. less personnel are needed to run an automated process. up to a 50% less grinding department space is required per unit of output, favoured by the economies of scale that allow containment of the number of auxiliary machines. Description of continuous mills; how to choose the right size The machine (fig. 20) is essentially made up of: A cylindrical structure with a rigidly bolted base. This has hubs for the support of the entire mill and concentric ports arranged inside the hubs for loading/ emptying. The cylinder is made of thick, special stainless steel that has been precision welded and quality-checked using X-rays and ultrasound. The bases are made of high-strength, precision-made cast iron on the cylinder connection flanges and the bearing drive fits. A series of supports, made of steel profiles and thick sheet metal, are anchored rigidly and partially sunk into the cement foundations. Mill supports are equipped with bearings of the double roller type to sustain rolling load adequately. A drive unit made up of one or two DC (or AC) motors, which, via a series of reducers, moves the mill cylinder: the reduction system may be of the gear or belt type. The drive unit thus provides: a) soft and gradual mill start up with low current absorption. b) a wide range of rotation speeds, thus making it possible to set the most suitable grinding speed for the specific body type (for DC motors). c) gradual mill deceleration with final halt in correct position. An internal cylinder lining made of wear-resistant rubber specially shaped to maximise grinding efficiency. A series of forced oil or grease lubrication systems to lubricate all drive parts. Description of the mill The continuous mill is, then, made up of a cylindrical sheet steel structure with ports for inspection, maintenance and introduction of grinding media; bases featuring ports for continuous infeed of raw materials and unloading of slip, a drive unit, a series of oil/grease lubrication systems and a PLC control panel. The anti-wear rubber lining on the interior of the cylinder usually incorporates 56

Grinding

Fig. 20. Continuous mill.

protuberances known as lifters: these are specially designed to optimise grinding and increase the dynamic effect of the pebbles on the material. The mill is subdivided into two or three chambers by bulkheads featuring adjustable-aperture throughways to control the axial flow of the slip and regulate how long the slip stays in the different zones. Sub-division of the cylinder means that grinding media materials and pebble assortments can be optimised for each different stage of grinding: heavy grinding media of varying size for mixing of the heterogeneous material near the mill infeed point. heavy grinding media of varying size, for the crushing or primary comminution of the harder coarse material: this stage does not necessarily necessitate high-density grinding balls (although they would undoubtedly improve performance). spherical or rounded media (to provide the rolling that optimises refinement of the material through abrasion and cutting): during the body refinement stage these pebbles should be small (to ensure a high number of media-media contact points) and high density (to keep dynamic input high and so maintain a vigorous impact action). In practice, in-mill grinding media separation is not so clear: primary crushing and mixing, already partially begun in the pre-mill feed devices, occur simultaneously in the first chamber and continue, to a lesser extent, in downstream ones. 57


Production of single-firing, porous or vitrified tile bodies generally involves mills with just two chambers, while porcelain tile manufacturers almost always use 3chamber models. The third chamber is generally used for extreme refinement of the body and is loaded with small grinding media. The first two chambers will usually contain medium density media (d≅2.6 g/ cm³) made up of rounded silica pebbles of varying size (from 30 to 100 mm); the third chamber generally contains high density sintered alumina balls (d≅3.5 g/cm³), quite small in size (1-1½), that refine the materials efficiently and ensure high productivity. Costs associated with grinding media wear in a continuous 3-chamber silicasilica-alumina mill are generally less than those incurred with discontinuous mills having alumina grinding (i.e. pebbles+lining). Optimising the grinding process via appropriate distribution of in-mill silica pebble assortment allows the desired particle size distribution to be achieved fast enough to meet output requirements without having to use costly sintered alumina media. Porcelain tile manufacturers generally use one of two different types of continuous mill that meet very different managerial and production needs. These are: very high capacity mills of enormous output potential which generate enormous economies of scale vis-à-vis auxiliary equipment and running costs. Mills of this type may, for example, have an effective internal capacity of 150,000 litres and yield an average output of 14-16 tons/hour of dry porcelain tile body. low capacity mills that are more flexible and easily adaptable to already-operative grinding departments. For example, these include mills with an effective internal capacity of 35-40,000 litres that provide average hourly outputs of 3.8 tons/h of dry body for porcelain tiles. Assuming average unfired scrap to be 5%, overall L.O.I. (loss of ignition) to be 5% and fired tile weight to be 20 kg/m², output for the above mill types may, in practice, be calculated as approximately 15,700 m²/day and 4,100 m²/day respectively. Self-classification of grinding media and grinding type is also possible: this is seen on cylindrical-type continuous mills featuring a helical classifying lining or those that have a range of diameters (i.e. the cylinders are conical). The operating principle of the cylindrical drum with helical classifying lining is based on the fact that large grinding media tend to remain at the periphery of the mass rolling inside the mill owing to the combined effect of gravity, rolling and centrifugal force. However, the lining lifters push the bigger pebbles backward towards the loading material inlet where larger pebbles are needed. Such mills do not have any partitions or separated chambers and the pebbles can be introduced during feeding together with the raw materials and water etc (in other words, on the run) instead of having to stop the mill to re-charge media into the separate chambers. 58

Grinding

Fig. 21. Continuous mill with cylindrical drum and helical classifying lining.

Fig. 22. Continuous conical section mill.

The operating principle of the conical mill is based on the fact that large grinding media, thanks to gravity, rolling and centrifugal force, tend to move towards the large mill diameter zone (feeding side). The main problem with this type is lower efficiency of final refining near the outlet due to decreased diameter (and volume) in this part of the mill. 59


MILL SIZING Mill productivity a) Sizing Continuous mill size and productivity are functions of a coefficient Ks that depends on the type of body to be ground and is defined on a case-by-case basis. This coefficient is assumed to be inversely proportional to the grinding time of the body in question in a standard-compliant discontinuous mill. It follows that the formulas for hourly output (H.O.) and mill size (effective volume) are: H.O. = Vu × d × s × Ks/GT where Vu = H.O. = d = s = Ks = GT =

effective mill volume in litres hourly output in kg/h slip density in kg/l % of dry material in the slip coefficient grinding time in standard-compliant discontinuous mill where it mathematically follows that: H.O. × GT Vu =  d × s × Ks

b) Grinding time GT The time the body takes to reach the desired residue logically coincides with the time it is left in the mill. This time is defined as: Vu × d × s × Kr GT =  H.O. where GT = Vu = d = Kr = H.O. =

time in mill expressed in hours effective mill volume in litres slip density in kg/l filling coefficient (ratio) hourly output in kg/h.

As already observed for discontinuous mills, there are also certain principles and regulations for the efficient running of continuous mills: 60

Grinding

a) Rotation speed To determine the correct speed at which a continuous mill should be run the already-explained rules and formulas for discontinuous mills can be applied. b) Lining shape Previous descriptions of the grinding process refer to mills with an essentially smooth lining in which there is no added lifting force between the lining and the grinding media. The point at which the grinding media detaches from the lining a key determinant of grinding efficiency can be changed by modifying the profile of the lining. Continuous mills are normally fitted with rubber protrusions that aid lifting of the grinding media. The distance between the lifters is constant on all types of lining. c) Different lining types A description of linings should take into account three factors: lining material, lining profile and lining attachment systems. The most commonly used lining materials are: high quality cast steel and rolled steel rubber stone (e.g. microcrystalline quartz) a combination of wood and steel porcelain. Generally speaking, the most widespread materials are rubber and steel, which complement each other.

Fig. 23. Left: illustration of mill interior (note the diaphragm between the two chambers). Right: two different lifter profiles.

61


None of the above materials is ideal for all applications. When coarse material is ground with large pebbles, a steel lining is generally preferred (not for ceramic bodies). Properly designed rubber linings have a particularly wide field of application and are, today, the most common. Rubber is, as a rule, the best material for working temperatures of 80-90 °C. Another important factor in the choice of lining material is the potential for contamination. Rubber is preferred in the grinding of ceramic materials to avoid even very small quantities of iron grains, which can lead to post-firing colour alterations. Rubber coated stone linings and the first rubber-lined mills featured completely smooth profiles. Yet to give the grinding load sufficient kinetic energy, smooth-surfaced mills have to rotate at a higher RPM than those with shaped linings. Moreover, high wear rates are caused by the material rubbing against a smooth lining as its slides backwards. A gradual increase in grinding media lift can, by choosing an appropriate lining profile, be obtained at even medium RPM. d) Grinding media Bear in mind that ceramic bodies can be divided into two main categories: white (or clear) firing and coloured firing. White bodies must not in any way be contaminated by colouring substances while, of course, there is less of a problem with coloured bodies. This distinction influences the choice of grinding media, the specific weight and abrasion resistance of which must be as high as possible. There are two possible solutions: 1) White bodies: grinding media in alumina or silica. 2) Coloured bodies: grinding media in silica (or steel). With coloured bodies, then, it is also possible to use steel grinding media, which provides high grinding efficiency. However, trials have shown that only iron particles smaller than 100 microns do not cause contamination problems; hence use of steel is only possible where the manufacturer has the equipment needed to eliminate the coarser iron grains. As far as linings are concerned a heterogeneous-profile rubber lining is preferred for abrasion resistance and noise dampening purposes. Control and production parameters in the continuous and discontinuous grinding department a) Particle size distribution (residue) This parameter the benchmark indicator of grinding process efficiency is of fundamental importance.

62

Grinding

The particle size distribution of a ceramic body drastically influences its behaviour during firing as it affects: specific surface area and reactivity shrinkage-vitrification curve deformation water absorption black core defects caused by contamination. Particle size distribution is generally determined by sieving a slip sample (on a sieve set of known mesh) corresponding to a known dry quantity and then measuring the cumulative dry residue on the sieves. A single sieve will not on its own be sufficient to determine the particle size distribution of a body. It is often preferable to use the residue percentages from two meshes having apertures of 63 µm (230 mesh, 10000 maglie/cm²) and 45 µm (320 mesh, 16000 maglie/cm²) respectively. However, as long as raw material characteristics or grinding conditions do not change, residue % at 63 µm constitutes a valid process control parameter. Product type Red monoporosa White monoporosa Red monovitrified White monovitrified Fast double firing Porcelain tile

Residue % at 63 µm 58 58 5 10 5 10 5 10 0.6 0.8

Residue % at 45 µm 15 25 15 25 15 25 15 25 15 25 1.8 3.0

Checking frequency: every shift/every mill. b) Density and water content (%) Variations in density and water content affect: mill productivity viscosity particle size distribution settling spray drier energy consumption. Slip water content can be periodically checked by measuring the density of the slip with a container of known volume (e.g. a pycnometer). Water content depends on: the nature of the body the quantity of deflocculant added characteristics of the water. 63


Product type Red monoporosa White monoporosa Red monovitrified White monovitrified Fast double firing Porcelain tile

Discontinuous mills 35 37 34 36 38 40 33 35 35 38 33 35

Continuous mills 33 35 32 34 36 38 31 33 33 36 31 32

Water content (%) Checking frequency: every shift/every mill. c) Viscosity (and thixotropy) If viscosity is too low it can lead to: sedimentation of the slip excessive quantities of water excessively fine spray-dried powders. Excessive viscosity, instead, can cause: longer grinding times mill unloading difficulties sieving difficulties excessively coarse spray-dried powder. The nature of the body, particle size distribution (especially as regards the clayey particles), density, temperature and pH all have significant effects on viscosity. Optimal viscosity is: 2.5-3 °E (250 - 400 cP). Checking frequency: once/twice per shift, every mill. d) Temperature Influences viscosity of the slip and often improves its rheological characteristics. Excessively high temperatures can: cause intense thixotropic phenomena damage the rubber lining. Optimum temperatures are 50-60 °C for discontinuous mills, 70-80 °C for continuous ones. Checking frequency: once/twice, every mill load. Definitions and units of measure While the rheological behaviour of suspensions of solids in liquids has already been described in detail in Volume I, it is worth mentioning some parameters commonly used in grinding departments.

64

Grinding

VISCOSITY = the resistance a fluid opposes to movement F/S = η (Vmax - Vmin /x)

where F = force applied to the liquid (N) S = surface area of application (m2) vmax-vmin = change in speed (m/s) x = thickness of liquid (m) η = absolute viscosity (N s/m2 = Pa s) N.B.: viscosity diminishes as temperature increases. SPECIFIC WEIGHT (DENSITY) = Weight (mass) per unit of volume (grams/litre) Examples: water olive oil slip viscosity mPa s 20 °C 100 300 1 density g/l 10004 °C 918 1700 Practical units of measure of viscosity °Engler Corresponds to the ratio between outflow time for 100 cc of slip and outflow time for 100 cc of water from a standardised funnel. This is the standard check for body slips. Example: outflow time for 100 cc of slip = 12 sec. water outflow time = 5 sec. slip viscosity: 12/5 = 2.4 °E. Ford Cup Measurement corresponds to time taken for 100 cc to flow out of a Ford cup with a ∅ 4 mm nozzle Mainly used with glazes and engobes. Example: slip viscosity = 25 sec Ford cup. °Gallenkamp Measurement corresponds to return rotation degrees (generated by of a twisted metal support wire) travelled by a cylinder that has been immersed in the slip and rotated 360°. Mainly used for casting slips (sanitaryware). Example: slip viscosity = 320 °G. cPoise (Brookfield) A measure of viscosity effected with a fixed-speed rotating viscosimeter. Example: slip viscosity = 300 cP 1 centiPoise (cP) = 1 milliPascal second (mPa s) 65


APPENDICES The continuous wet grinding mill and the porcelain tile industry In the context of porcelain tile manufacturing the continuous grinding system initially appeared to be marred by a certain rigidity as there were difficulties in making the slip colouring process flexible. Today, most companies equipped with a continuous drum mill use it to produce a colourless base slip, which, after sieving, is sent to a holding tank with an agitator. Concentrated colouring syrups are then added to the tank; alternatively, the syrups are batched into the slip by devices installed on the tank feed line or, more rarely, in the pressurised piping that leads to the spray drier. Thanks to the advent of reliable automatic batching systems continuous grinding has now become extremely versatile and has rapidly gained ground in the porcelain tile sector too. The base slip is coloured by using devices (e.g. adding batchers that operate by weight or volume) that add colouring slips to the base body. These are installed on the spray drier service tank feed line and thus allow a microprocessor to generate a signal proportional to density: this signal can be displayed automatically, both locally on the instrument and on the batching control system. The measuring system thus operates independently of temperature, pressure, viscosity, conductivity and fluid rheology. Note: concentrated colorants are, by definition, slips characterised by a very high concentration of colouring pigment. Use of such concentrated products means that the coloured slip preparation department takes up very little space yet provides maximum productivity, optimises the degree of grinding of the pigment and gives optimum slip storage capacity. As there is high pigment concentration in the slip, it will, of course, be very dense and therefore the rheological properties of colouring suspensions must be monitored carefully to prevent sedimentation. Nevertheless, some companies do produce, for some periods, white or superwhite bodies directly in the continuous mill. The rheological characteristics of these bodies differ from those of standard base bodies in that they feature a higher non-clayey complementary content, which is, as a rule, counterbalanced by the use of especially plastic clayey raw materials (e.g. clays with a montmorillonite content); the latter influence the viscosity of the slips and their tendency towards thixotropy. Furthermore, the abundance of grinding-resistant hard components may increase process times, lowering the output potential of the body preparation department considerably. So, while the continuous machine is, of course, ideal for defined, constant operating conditions, it is also capable of providing a certain versatility.

66

Grinding

Technological and managerial parameters The slip can be fluidised with the aid of solid additives such as sodium polyphosphate, introduced in quantities amounting to 0.3-0.4% of the dry body. Alternatively, where body type allows, liquid mixtures based on organic substances and sodium silicates can be used; these are introduced in slightly higher percentages (approximately 0.4-0.5%). Liquid mixtures are advantageous in that they can be adapted to suit a certain body and act faster thanks to the ease with which they are dispersed throughout the aqueous vehicle; batching systems are also more reliable. The water content of the slip generally falls within the 31-34% range. Bear in mind that: as a rule, the amount of deflocculant being equal, a higher in-body clay content increases the amount of water needed to disperse the material. high plasticity of clayey body components (e.g. montmorillonite clays) raises viscosity, which must then be controlled by increasing water or fluidizer content (white bodies, generally featuring a higher percentage of hard complementary components and a clayey fraction of higher plasticity, are subject to the contrasting effects of the plasticity of the clays and their percentage in the body). recycled water that has undergone chemical-physical clarification-flocculation treatment and has consequently been enriched with flocculating ionic substances may work against the effect of the fluidizer and increase the required quantity of water. using greater quantities of deflocculant allows the quantity of water to be reduced; consequently there are heat energy savings at the spray drying stage. attaining a high degree of body refinement means dissipation of large amounts of mechanical energy in the form of heat; the slip thus exits the mill at high temperature (often higher than 70 °C), keeping viscosity low. At room temperature, slip density varies between 1690 and 1740 g/l (at mill outlet temperatures much lower densities are observed): the lowest densities are generally observed in colourless bodies, the highest in white ones. Viscosity generally falls between 2.5 and 3 °E, although slips outside this range that can still be sieved easily are certainly not uncommon. The fluidity of a suspension, in fact, should also be viewed in light of its yield point and thixotropy. In any case, good fluidity means good sieving on small-mesh screens (net apertures of 130-160 µm), indispensable in ensuring body quality and preventing imperfections such as dimples and dots which only show up after firing. Production line checks should, of course, be completed by particle size distribution data from laboratory tests, which, for the most part, consist of sieve residue analyses. The reference figures usually adopted for porcelain tiles are: 0.7-1% of sieve residue with net mesh opening 63 µm 2.5-4% of sieve residue with net mesh opening 45 µm.

67


More accurate, reliable analysis – obtained with the aid of a laser diffraction granulometer – allows the user to evaluate cumulative curves and the average numerical diameter of samples. Results are generally as follows: – average numerical diameter of particles: 17-20 μm. – percentage of particles passing through at 20 μm: 70-75%. Specific electricity consumption on mills in porcelain tile plants usually stands at 50-70 kWh per ton per of dry ground body. The table below compares porcelain tile body production in continuous and discontinuous mills (grinding media quality being equal): Continuous mill

Discontinuous mill

Gross capacity (l)

40,000

38,000

Volumetric ratio

1.053

1

Grinding media

Silica/Silica/Silica

Silica

Rubber

Silica

3,500

-

N° of cycles (cycles/day)

-

1

Slip load (l)

-

17,000

Slip density (g/l)

-

1,690

Water content (%)

-

34

Dry load (kg)

-

18,900

3,500x24=84,000

18,900/1=18,900

84,000/1.053=79,800

18,900/1=18,900

63.7 kW/h

103 kW/h (*)

Lining Dry output (kg/h)

Daily dry output (kg) Output/day per unit volume (kg) Total consumption/ton

(*) This considerable difference is caused by the silica/silica combination (which is quite common). Where alumina is used, grinding times are halved and electricity consumption remains similar. Productivity ratio, gross volume and grinding media quality remaining equal: 79,800/18,900 ≅ 4.2 (in any case this figure will always be greater than 3). Therefore, gross mill volume and grinding media being equal, at least 4.2. discontinuous mills are needed to do the work of a single continuous one. Similarly, overall gross mill volume and grinding media being equal, continuous mill productivity is approximately 4.2 times higher than discontinuous mill productivity. The nature of the grinding media influences efficiency considerably: sintered aluminium media raises machine productivity but also raises costs. However, bea68

Grinding

ring in mind that the benefits of sintered alumina are largely perceived at the refinement stage, such additional costs can be mitigated by using such media in the refinement zone only. Significant economies can thus be obtained simply by changing the grinding media, without excessively penalising productivity.

69


70

Drying

Chapter II SPRAY DRYING OF CERAMIC SLIPS

Spray drying has been used in the foodstuff and pharmaceutical industries for over a century; its introduction into the ceramic industry is a somewhat more recent affair. In ceramics, spray-drying has simplified the overall production process by replacing filter-pressing, drying, grinding, re-moisturising and classification (fig. 1) and has also reduced labour and maintenance requirements. The slip obtained from grinding raw materials in an aqueous suspension is dried by way of a continuous, automatic process that provides a product of controlled moisture content, shape and particle size distribution that is ideal for pressing. Thanks to spray drying, these advantages have been available to the ceramic RAW MATERIALS PRE-TREATMENT WET GRINDING SIEVING

FILTER PRESSING

DRYING SPRAY DRYING

CRUSHING RE-MOISTURISING CLASSIFICATION

SILOS PRESSES

Fig. 1. Comparison of the spray drying and filter-pressing processes.

71


industry for some years now. Spray drying is carried out in machines called spray drier atomisers, or, more simply, spray driers (see fig. 2).

Fig. 2.

Classification of spray driers As a rule, there are two categories of spray drier: a) Direct heat spray driers In these spray driers the heat needed to bring the water in the slip to evaporation point is introduced in the form of combustion gasses or heated air, which comes into contact with the droplets and causes evaporation by convection. This is the type generally used by the ceramic industry. b) Indirect heat spray driers In these machines the heat is transferred to the material to be treated by conduction. The greater the surface area via which evaporation can take place, the faster the evaporation process. The heart of the system consists of a sprayer device that is generally of the rotary (turbine) or fixed (nozzle) type. Atomisation of the suspension is achieved through the use of kinetic energy (in the former) or the pressure exerted on the fluid (in the latter). There are, then, several different types of spray driers, each defined by the suspension atomisation system and the heat flow route (upward or downward). 72

Spray drying of ceramic slips

Fig. 3. Different spray drier configurations.

Fig. 3 illustrates four different types of spray drier configuration: Downward uniflow type, atomisation with nozzles. Downward uniflow type, atomisation with turbine. Counterflow type: atomisation with downflow nozzles, air upflow. Counterflow-uniflow type, upward nozzle atomisation, air downflow.

Ceramic production plants generally utilise the so-called pressure centrifuge nozzle system to atomise the suspension; here, the pressurised flow also imparts, by way of volute-shaped inserts, rotary movement of the particles as they exit the nozzle. Other devices known as lances can also be used to atomise the slip. Within these different nozzle and turbine systems, then, the quality of atomisation depends on the degree of concentration of the solid, viscosity, surface tension of the slip and the equilibrium between several other factors such as pressure, tower volume, nozzle size (in the first case) and the characteristics and speed of the turbine (in the second). A nozzle-type atomisation system has the advantage of providing good particle size distribution control and repeatability. Once the flow rate and the characteristics of the incoming suspension have been determined it is easy to exercise control over all the variables: powder particle size distribution is changed simply by changing the nozzles. 73


A nozzle-type spray drier with compressed air atomiser is rarely used to dry ceramic slips. For products with a very high specific weight, or where very fine particle size distribution is required, the centrifugal disc system is the most suitable. Fig. 4 summarises the complete spray-drying process, from slip infeed to purification of the exhaust gases.

Fig. 4. Illustration of spray drying plant.

General description of nozzle-type spray drier How it works In spray drying, continuous and automatic atomisation and hot air drying of the slip produces powders of controlled particle size distribution and moisture content that can then be pressed in a semi-dry state to make tiles. The spray drier essentially consists of a chamber of shape and volume designed to promote heat exchange between the finely sprayed slip and the hot air/combustion fumes. There are, as illustrated on the previous page, several different types of spray drier: the one most commonly used in the ceramic industry is the mixed-flow type where the air is directed downwards from the top of the chamber and the slip is projected upwards from the base. This solution maximises the time spent by the particles inside the drying chamber: because they first encounter the hot air in counterflow and then in uniflow, heat exchange efficiency is high (fig. 5). 74


Fig. 5. Illustration of the spray drying process. General view of plant.

The finest powders in the spray drier chamber are first separated by cyclones, then a dust collector. Finally, the fumes exit the stack. The spray drying cycle The diagram in Fig. 6 illustrates a spray drier with nozzle-type atomisation and shows the position of all the main devices. 1. 2. 3. 4. 5. 6. 7.

Slip feed pumps Filters Nozzle holder ring Drying tower Powder outlet valve Separator cyclones Pressurisation

8. 9. 10. 11. 12. 13.

Burner Hot air duct Annular hot air distributor Centrifugal ventilator Wet dust separator Stack

The main body of the drier is cylindrical. Hot air is introduced into the chamber from above and distributed tangentially; the initial counterflow stage of heat exchange takes place as this hot air meets the atomised slip being projected upwards from the bottom of the cylinder via nozzles mounted on a ring which is concentric to the chamber perimeter. As the powders lose their momentum and fall back towards the lower, conical section of the chamber the final uniflow stage of drying begins. Kinetic energy is provided by a pair of piston pumps (which operate at 20-30 atm) that impart a nozzle exit speed of about 30 m/sec, sufficient to overcome the low viscosity (about 3° Engler) of the slip, shear it into minute droplets and direct it upwards. 75


The cone-shaped upward spray does not travel in a straight line but, rather, spirals upwards on account of the movement imparted by internal devices, nozzles and spirals of varying shape and size. A simplified description of the overall drying process can be provided with the aid of fig. 6: the slip is fed at constant pressure by the pump (1), passes through the filters (2), and reaches the distributor ring (3) inside the tower (4). The finely atomised slip jet is hit by a vortex of hot air produced by a natural gas or LPG air-flow burner (8) and an intake fan (7) (or by a direct combustion burner using liquid fuel). The air is conveyed to the upper part of the tower via a heat-insulated steel duct (9) and is set in rotation by the annular distributor (10). Dried powders are unloaded via the outlet valve (5) onto a conveyor belt.

13

9

10

8 12 11

4

3 7

6

5

1 2

Fig. 6. Illustration of main parts and devices of a spray drying system.

76


The fine powder residue that remains suspended in the air exhausted by the main fan (11) is, in part, separated out by the cyclones (6) and, in part, by the wet dust separator (12). The exhausted air is then expelled via the stack (13). Description of main spray drier devices Slip feed pump (1) The purpose of the pump is to transfer the slip at a set pressure towards the atomisation devices. It is of the hydraulic piston type, powered by a hydraulic unit. The pistons are made of alumina, a material highly resistant to abrasive fluids. Working pressure (regulated simply by adjusting the valves) is generally in the order of 22-28 bar, depending on the characteristics of the slip and the type of atomisation nozzles. Slip filters (2) Two filters are generally installed on the slip feed line: these trap any impurities or foreign bodies. It is important that these filters operate efficiently because if such impurities/ objects reach the atomisation devices they can clog the nozzles. The filters consist of a cylindrical container containing a perforated cylinder which acts as a support for the filtration mesh. Filters are washed automatically and alternately (where of the automatic type). Distributor ring (3) Made of a stainless steel ring with spray nozzles attached, connected to the slip feed piping via a reinforced rubber flex hose. A simplified illustration of a stainless steel nozzle holder ring, relative feed piping and the ring extraction device is given in fig. 7. A more detailed view of the nozzle ring is shown in fig. 8. As an alternative to the ring system the slip can also be atomised by lances (fig. 9). These are arranged around the tower perimeter and protrude towards its centre. One or more atomising nozzles can be fitted on each lance. Fig. 10 illustrates two different spray driers equipped with ring-mounted and lance-mounted nozzles respectively. a) with ring-mounted nozzles b) with lances.

77


Fig. 7. Diagram of nozzle ring and extraction device.

Fig. 8. View of nozzle ring.

Atomisation nozzles The nozzles are generally of the so-called coil type in which a spiral-shaped insert causes the fluid, driven by the pressure behind it, to rotate in a way that facilitates dispersion of the jet as it impacts with the air. An exploded drawing of a nozzle is given in fig. 11. The spiral is not the only important nozzle part: equally important are the disk inserts with central holes. The (upper) table in fig. 11 shows disc hole diameters and the lower table the spiral heights. 78


Fig. 9. Lance atomiser.

Fig. 10. Cross section and plan of two different spray drier types: a) ring atomising nozzles b) with lances.

In addition to the spirals and discs the other main nozzle parts are: 1. Head 13. Discs without hole 14. Distributor 15. Container with coupling 79


Fig. 11. Nozzle parts.

The perforated discs, made of tungsten or diamond-coated, have calibrated clearance. Most perforated discs have an orifice diameter of 2.5 mm. The coils are located in the interior of the nozzle. These impart a rotary motion and are thus used to narrow or widen the atomised slip cone. The most commonly used types are between 8 and 15 mm high. The number and type of nozzles depend on the type and quantity of spray-dried powder to be produced. 80


Drying tower (4) This constitutes the actual evaporation chamber (fig. 12). Made up of a conical collector, a cylindrical evaporating area and an upper section containing the annular hot air distributor. The internal walls of the tower are made of stainless steel and the outer walls of polished aluminium. A heat insulation layer is sandwiched between the inner and outer walls. A stainless steel duct connects the tower to the separator cyclones. A thermocouple installed on this duct allows the exhaust air temperature to be monitored as it relates to the moisture content of the dust produced. Powder outlet valve (5) This cooler-equipped valve is of the counterweight type; it cools and discharges the spray-dried (or atomised) powder. Ambient temperature air flows through a series of adjustable-aperture slits and comes into contact with the outflowing spray dried material, thus lowering its temperature. There is also another type of cooler known as the enhanced performance type (fig. 13). This allows the external air, drawn in by low pressure, to contact the hot granulate downflow several times and uses special surface geometry to improve cooling performance. Lowering the temperature of the spray dried granulate even further reduces the condensation problems generally encountered later in conveying systems, sieves and silos.

Fig. 12. Cut-away diagram illustrating the tower interior.

81


Fig. 13. Enhanced-performance spray dried powder cooler.

Separator cyclones (6) These are made of stainless steel and equipped with a counterweight powder discharge valve and inspection hatches. These perform initial powder separation from the exhaust air exiting the tower (fig. 14). They work constantly and are always switched on before the main fan so as to prevent the latter being damaged by powder. Before entering the cyclones the exhaust air powder content is in the order of 2-4000 mg/Nm3. When the air leaves the cyclones content is just 400 mg/Nm3. Fuel feed system The type of fuel feed system essentially depends on the type of fuel (gaseous or liquid, light or dense, high or low viscosity). Systems using light liquid fuel (e.g. light oil) do not require heating devices for correct atomisation at the burner nozzle in that their low viscosity allows them to be atomised by way of pressure alone. Systems using, instead, heavy liquid fuels require nozzle pre-heat and special inburner apparatus. Where the fuel is a very thick liquid the pumping duct and the holding tank will also need to be heated. Burner types (7) The different combustion systems adopted in spray driers are as follows: Flu-fire burner (gaseous fuel) with cogeneration system. Air-flow (or in-vein) burner (gaseous fuel). Direct combustion Weishaupt burner (liquid fuel). 82


Fig. 14. Diagram illustrating flow routes within the fine powder separationcyclones.

Air-flow burners are suitable for gaseous fuels and forced-draught burners are suitable for liquid fuels. In both cases the flame is adjusted via a system that modulates the quantity of fuel as a function of required in-tower temperature. Fig. 15 shows an air-flow burner. They provide excess combustion air (about 150% of stoichiometric air). FLU-FIRE burners, instead, do not have their own combustion air fan. Gas combustion thus occurs with process air, which must have a minimum oxygen content. Fig. 16 illustrates the structure of the FLU-FIRE burner. These burners are also suitable in cogeneration plants and, depending on their specific characteristics, may act as after-burners where standard generators are installed (fig. 17). Hot air distributor (8) The hot air distributor, installed in the upper part of the evaporation tower, consists of a peripheral, tangential inlet and a system of spiral ducts that force the air to move towards a corresponding series of inward-facing apertures (fig. 18). 83


Fig. 15. Air-flow burner structure.

Fig. 16. FLU-FIRE burner structure.

The resulting air flow generates a vortex which facilitates heat exchange with the slip (fig. 19). Main fan (9) The main fan, of the centrifugal type, draws in the air needed for the drying process. A partial vacuum is thus present in the section of duct preceding it. 84


Fig. 17. FLU-FIRE burner coupled to cogeneration plant.

Fig. 18. Cut-away diagram showing the hot air distributor.

Fig. 19. The hot air vortexes formed in the distributor (computer generated image).

85


The stack (10) The stack constitutes the final section of the air circulation system and connects the spray drying plant to the exterior. Dust separator device The wet abatement device is located downstream from the main fan. It constitutes the second stage of dust separation and is where the finest dust particles are removed. Spray drier operation (temperature and pressure) Working temperatures and pressures are illustrated in fig. 20. The slip, which exits the mill at 20-70 °C, is pumped at a pressure of 20-30 bar towards the spray drying system. Pressure P1 in the drying tower is approximately -0.5/-1.5 mbar. The hot air intake temperature is in the 550-650 °C range. The temperature of the spray-dried powder is generally in the 40-70 °C range. fume temp. with abatement unit 60-80 °C without abatement unit 80-120 °C

gate

gate

fume outlet

fume outlet temp.

spray dried powder temp.

slip

Fig. 20. Temperatures and pressures in the spray drying system.

86


The air directed towards the cyclones (T1) falls within the 80-120 °C temperature range. The partial vacuum inside the cyclone is in the order of 10/20 mbar. The temperature of the fumes exiting the stack may be 60-80 °C where a dust separator device is installed or 80-120 °C where it is absent. Stack smoke pressure is in the order of +10/+15 mbar. Since the situation varies on a case by case basis all the above figures should be viewed as approximate. The dynamics of “dried granulate” formulation At the precise instant in which the slip is atomised into droplets, the latter are split, projected upwards, captured in the turbulence generated by the downflow of hot air and forced to follow trajectories that involve collisions, aggregations, new splits and impacts against the tower walls. Computer-generated flow models have allowed researchers to study just how a spray drier works by examining the trajectory of the particles and how long they remain in the chamber; researchers have also been able to investigate temperature and humidity ranges both in the process area and the slip droplets themselves (see fig. 21). Like all interfacial phenomena, the relationship between mass (volume) and surface area is crucial. Here, the phenomenon of evaporation is directly proportional to the droplet surface area available for exchange and the temperature, surface tension and other parameters.

Fig. 21. Computer generated image of particle trajectories inside a spray drier.

87


At the same time, the size of the droplets and the direction and speed of the atomised flow dictate the suitability of the type and size of the drying chamber. In the ceramics industry droplets generally range in size from just a few microns to 900-1000 microns with a classic statistical distribution centred around 300-400 microns. During the average life of a droplet (5-6 seconds, the contact time between the two phases) high-energy heat exchange at the liquid interface, with air that has been heated to over 500 °C (maximum air temperatures may be as high as 650 °C), causes evaporation of the water in the droplet and thickening of the particles suspended in it. The size and shape of the dried product obviously depend on the nature of the solid phase (note that the case in hand refers to a ceramic body where the percentage of inert material is very high with respect to the clayey parts). Depending, then, on the size of the droplets, the generated granule types may be classified as follows (see fig. 22): a) For droplet diameters up to 70 microns drying is absolute and the finest particles, in the form of powder, are captured by the outgoing air flow and then recovered by the powder collection device. The smallest particles consist of individual grinding residue particles. b) For measurements in the 70-400 micron range drying conforms to average overall moisture content values and generates a granule, already spherical in shape with a cavity of varying size, that drops into the conical cooling/collection zone of the spray drier. c) Over 400 microns, drying produces a granule with a moisture content higher than the overall average; this is because, on account of its volume, it spends

Fig. 22. Possible particle size distribution of a spray dried body and morphological aspect of grains of different sizes.

88


less time in the drying chamber. This large granule nearly always has the classic shape described above yet also has a net tendency towards agglomeration and inclusion of particles as a result of its high residual moisture content (fig. 23).

Fig. 23. An individual particle (left) and an example of agglomerated particles.

Granule formation mechanism The typical hollow sphere shape generally derives from complex phenomena; these are examined in detail below. Tests have shown that once the drying process has begun the particle maintains it trajectory without rotating: thus one side of the particle always faces forwards as it travels through the air. It is on this side that initial heat exchange takes place. This involves the first, violent superficial evaporation and consequent hardening of the external film of the grain being formed: simultaneously the droplet contracts as the solid particles suspended within it draw closer together (fig. 24). Fig. 25 illustrates, stage-by-stage, the grain formation sequence starting from the moment in which the droplet is expelled from the nozzle. At this point the parameter which determines the final form is the duration of heat exchange between the two phases. If this time is too short the water vapour formed inside the droplet, unable to exit the already hardening film, shatters the grain which, on microscopic examination, takes on the appearance of a split-apart, layered, thickly walled egg shell. If the heat exchange lasts longer, the water vapour forming in the interior can leave the droplet through its rear, which is cooler and has a film still at the formation stage (thinner and less compact). This is why the rear wall collapses inwards and the grain takes on the characteristic hollow sphere appearance, maintaining a residual moisture content. Of course, not all the grains take on the classic spherical shape as there may be small percentages of agglomerates and other, finer particles. 89


Fig. 24. The dynamics of particle-hollow formation.

It is the spherical fraction that provides the best results in terms of uniformity, average composition and good flow performance. It is interesting to note that the particle size distribution of the spray dried grains compared to that of the wet sprayed droplets is smaller by about one third. Starting with a slip having a litre weight of 1.6-1.7 Kg, total evaporation of the water yields a solid with an apparent weight of about 1 Kg/l on account of the empty gaps within the granules and between them (corresponding to the contraction undergone by the droplets as they dry). While the solid parts contained in the initial droplet end their passage through the spray drying system here, the liquid fraction, which has been transformed into water vapour, is exhausted together with the air in the chamber (now cooler after yielding the heat necessary for evaporation). After passing through the cyclones and the wet separators, the so-called fumes exit via the stack at controlled temperature and humidity, taking with them any residual solid particles compatible with environmental protection laws. Transformation of the relevant phases is illustrated in fig. 26. The final powder added to the fine powder recovered by the cyclones is sieved: it is then stored in silos. After an ageing period, which homogenises moisture content and temperature throughout the silo, the powder can then be drawn off and sent to the pressing department.

90


Fig. 25. SLIP liquid phase T 35/60 °C - P 20 - 30 bar solid 65% liquid 35%

HOT FUMES gaseous phase T 500/600 °C dry air CO, CO2

POWDER solid phase T 40/65 °C solid 95% liquid 5%

COLD FUMES water vapour phase T 90/115 °C air, water vapour, fumes, suspended dust

Fig. 26. Characteristics of materials and fumes (phases) entering and exiting the spray drier.

91


Characteristics of spray dried powders Morphology of the grain and particle size distribution of the powders The spray dried grain is, then, more or less spherical in shape with a hollow dimple of varying size; it contains the residual moisture necessary for pressing and this is generally found inside the hollow. Fig. 27 compares the grain morphology of a spray dried powder with that of a dry-ground powder. The spray dried powder (left) appears as a set of spherical grains of varying size while the dry ground powder (right) is an agglomerate of needle-like, acicular particles. This difference clearly implies different mould filling characteristics at the pressing stage: the best results are generally obtained with spray dried powders. Grain size Spray dried powder grains are generally of a diameter in the 100-600 micron range, with much of the statistical distribution falling inside the 180-300 micron band. Maximum particle concentration, though, is generally found in the 250-300 micron range (fig. 22). The three factors grain morphology residual moisture content particle size distribution are extremely important in defining the flow characteristics of the powder, which play a key role in mould filling. Spray dried powders derived from different bodies, such as vitrified single-firing (white and red), porous single-firing, and porcelain tile bodies have, on the whole,

Fig. 27. Grain morphologies of spray-dried (left) and dry-ground powders.

92


similar particle size distribution since it depends more, in fact, on the physical characteristics of the slip (viscosity and density) than it does on the composition of the body. Tab. 1 shows spray dried particle size distribution ranges for different bodies commonly used in the ceramic industry.

GRANULOMETRIA: PARTICLE SIZE DISTRIBUTION: µum mesh/cm m mgl/cm22 above sopra 100 600 “ 196 425 “ 400 300 “ 576 250 “ 1050 180 “ 2500 125 below sotto 2500 125 Lost Disperso Total Totale Approx. average diam.(µm) (µm) Diam. medio approx.

1

2

3

4

2 10 11 43 15 12 7 0 100 283

6 20 13 41 12 7 2 0 100 344

2 13 13 44 14 9 6 0 100 295

1 9 10 42 17 14 7 0 100 272

Particle size distribution of different spray dried powders (different bodies). Note: the differently numbered spray dried powders correspond to the following commonly used industrial products: 1 white vitrified single-firing body, 2 red vitrified single-firing body, 3 red porous single-firing body, 4 porcelain tile body.

Tab. 1. Spray dried powder particle size distribution of different bodies.

Powder flowability The flow properties of spray dried powder are of fundamental importance to the efficient filling of the press moulds and the subsequent tile quality. Conceptually, measurement of powder flowability involves various factors such as particle size distribution, specific weight, shape, mutual adhesions and degree of packing. In practice, however, flowability is measured by quantifying the time taken for a given amount of powder to flow out of a funnel-shaped container. With ceramic powders the tool proposed by AICE consists of a funnel of known volume (fig. 28). This measurement allows the user to obtain and evaluate the following information: fluidity relative to volume (expressed in cm3/s), obtained by dividing volume by emptying time. apparent density (expressed in g/cm3), obtained from the ratio between powder weight and volume occupied (i.e. funnel volume). 93


Fig. 28. Instrument for measuring powder flowability.

angle of fall (expressed in degrees), obtained by measuring the angle of the cone formed by the powder falling from the funnel. Tests carried out on spray dried powders obtained from different bodies made by different producers show very similar fluidity/volume and fluidity/mass values. It has also been observed that the most important factor influencing powder flowability is the residual moisture content of the spray dried powder; note that flowability decreases as residual moisture increases (see tab. 2). Slightly different flowability values are seen in laboratory-prepared mixtures, obtained by mixing together very different granulometric fractions (tab. 3). Observation of the table shows that a mix consisting of a preponderance of medium particles flows better than other mixes classified as coarse or fine.

Powder content Umiditàmoisture delle polveri % %

Fluidità rel. rel. to al VOLUME VOLUME Fluidity 3

/s cm3/s cm Fluidità rel. rel. to alla MASSA g/s Fluidity MASS g/s 3

Densità apparente Apparent density g/cm g/ cm3 Angolo caduta ° Angle ofdi fall

(°)

3.2

5.1

5.9

16.1

16.1

14.8

17

16.9

15.3

1.06

1.05

1.03

30

29

31

Tab. 2. Flowability values for spray dried powders of different residual moisture content.

94

CAMPIONE SAMPLE PARTICLE SIZE DISTRIBUTION: GRANULOMETRIA: mgl/cm2 um above sopra 100 600 “ 196 425 “ 400 300 “ 576 250 “ 1050 180 “ 2500 125 below sotto 2500 125 disperso lost Total Totale Approx. average diam. (mm) Diam. medio approx. (µm) POWDER MOISTURE CONTENT % % UMIDITÀ POLVERI FLUIDITY rel.altoVOLUME VOLUME FLUIDITÀ rel. cm3cm /s 3/s FLUIDITY rel.alla to MASS FLUIDITÀ rel. MASSA g/s g/s APPARENT DENSITY g/ 3cm3 DENSITÀ APPARENTE g/cm ANGLE OF ANGOLO DIFALL CADUTA (°) (°) 0 0 0 100 0 0 0 0 100 275 17.7 18.8 1.06 27

6 44 50 0 0 0 0 0 100 456 15.3 15.8 1.03 30

1 8 9 45 16 12 7 1 100 267 4.2 17.0 18.4 1.08 27

MEDIUM MEDIO

COARSE GROSSO

T.Q.

95

16.5 18.0 1.09 28

0 0 0 0 45 34 21 0 100 162

FINE FINE

16.2 17.2 1.06 29

3 22 25 50 0 0 0 0 100 365

C+M G +M

17.4 19.1 1.10 27

0 0 0 50 23 17 10 0 100 218

M+F M +F

16.4 18.8 1.15 28

3 22 25 0 23 17 10 0 100 308

C+F G +F


Tab. 3. Flowability measurements of a spray dried powder and artificially produced grain blends (different particle size distributions).


Variations in physical characteristics of spray dried powders To obtain satisfactory results in downstream production processes the spray dried powders must, then, have certain residual moisture characteristics. As mentioned above, particle size distributions for spray dried powders produced with different bodies (e.g. vitrified single fired, porous single fired, porcelain tile) are, to a large extent, similar. Residual moisture content may vary from 4 to 7 % depending on the plasticity of the body. It is, to an extent, possible to act on certain variables concerning both the working conditions in the spray drier and some of the slip parameters so as to vary and optimise aspects such as particle size distribution, residual moisture content, spray drier output, quantity of powder reaching the stack, temperature of spray dried product. Fig. 29 shows how these variables (causes) act on the characteristic parameters of the spray drying process (effects). The information in the table is, of course, indicative rather than absolute in that the cause-effect variables are presented individually while, in practice, they interact with each other. In an actual factory several variables are usually altered simultaneously to produce a final result that benefits from such interaction. Relationships regulating working conditions of spray driers There follows a description of the most important theoretical relationships regulating the spray drying process. Although the diagrams in figures 30 and 31 were obtained by atomising water, the results are similar to those that would be obtained by spraying a slip. In spraying water similar spray angles and lower volumetric flow rates are observed. Relations regulating flow rate The volumetric flow rate of the slip is proportional to its density. Volumetric flow ratecaá Portata volumetri

1 density densità

Flow rate, by volume or weight, is proportional to pump pressure. Portata volumetri ca á pressione pompa Volumetric Volumetric flow flowrate rate pump pressure

96


Fig. 29. Variables which control and condition the spray drying process. Output comparison between slip andewater Confronto tra produzione tra barbottina acqua - spacers15-7 15-7spirale coil 13-6 = 20 -Torrette 13-6Pp= 20 bar

Portata Flow ratex 10 x 10

25 20 15

Slip Barbottina

10

Water Acqua

5 0 1

2

3

4

Diametro ugelli Nozzle diam.

Fig. 30. Variation in flow rates as a function of nozzle diameter.

Cone Angoloangle cono

Outputproduzione comparison with slip Confronto con barbottina -- spacer = 20 bar bar Torretta15-7 15 - 7Pp= 70 60 50 40 30 20 10 0

coil 10-4 spirale 10 -4 coil 13-6 spirale 13-6

1

2

3

4

Diametro ugello Nozzle diam.

Fig. 31. Variation in cone angles as a function of nozzle diameter. 97


Flow rates increase as coil clearance increases. Fig. 30 shows flow rates (y coordinate) with coils of different heights in relation to the adoption of nozzles of different diameters (x coordinate). Relationships regulating particle size distribution Using higher pressures results in a smaller particle size distribution. Greater slip pressure => smaller particle size distribution With higher viscosities larger particle size distributions can be obtained. Higher viscosity => larger particle size distribution Energy consumption The optimisation of energy costs during the spray drying process depends on a number of plant engineering and technological factors. The most important of these are: size and type of spray drier continuity of operation conditions on the machine and essential devices such as burners, fans, insulation systems optimisation of overall working conditions technological characteristics of slip (density, viscosity, droplet drying rate). The minimum relative specific consumption (Kcal/kg) and its output capacity expressed in Kg/h - essentially depends on two factors: percentage of solid contained in the slip hence the proportion of water to be evaporated hot air inlet temperature. The heat transfer diagram Fig. 32 illustrates the heat (energy and material) balance in the spray drier. Note that the heat effectively used for evaporation of the water accounts for little more than half of the total heat introduced into the chamber; the rest is, instead, heat loss, mainly through the walls and with the exhaust air and, to a far lesser extent, with the dried product. As tab. 4 illustrates, optimum specific fuel consumption is, on average, about 370 Kcal/kg.

98


Fig. 32. Heat (energy and material) balance in the spray drier.

99


Essential parameters and examples for the evaluation of energy consumption and spray drier output capacity Calculating heat energy consumption The parameters and data which need to be taken into account in evaluating energy consumption are given below (tab. 4). Average specific heat energy consumption is thus as follows: qA =

G e, A .Q Gp

= 372.7 438.5 kcal/kg

ATM Gp

Kg/h

Hourly pressing output Produzione orariapowder di polveri per pressatura

Ub

l/kg

Umidità della barbottina Slip moisture content

Up

l/kg

Pressing Umidità della powder polvere moisture per pressatura content

Q

Kcal/l

Consumospecific specificoheat medio di energia Average energy consumption per litreper of water evaporated termica litro d’acqua evaporata

1000 850

Ge

l/h

Hourly Portata oraria flow rate di acqua of water da evaporare to be evaporated

Ge,A

Q

Kcal/kg

Consumospecific specificoheat medio di energia Average energy consumption per Kg of powder termica perpressing kg di polvere per pressatura

qA

Tab. 4.

100

10000

0.35

---


Practical method for calculating evaporating power of a spray drier Of all the spray drier parameters, evaporation rate is the one that provides the most immediate information as regards its efficiency and the one that will lead to full exploitation of its potential. Table 5 shows a series of formulas with which it is possible to calculate the amount of water evaporated by the spray drier.

H 2O ev. = Pf × Y =

Y

Ui% − Uf% 100 − Ui%

Key to symbols: Ui% Water contained in the slip (%) Uf% Water contained in the spray dried product (%) Pf Final spray dried product, including residual moisture (kg/h) H2O ev. Water evaporated (litres/h) Y Coefficient for calculation of evaporated water (see also table 5) Thus a very simple formula to calculate the maximum output is:

Pf

=

H 2O ev. Y

Calculating the size and output of the spray drier There is often a need to know data as regards the output the spray drier will effectively be expected to provide; such data is used to size the machine in line with effective production plant needs and to run checks on production during routine operation. The following two examples show how some of the sizing questions can be resolved. 1 - If hourly wet powder output is known the quantity of water to be evaporated by the spray drier can be determined. a) Known data Qp quantity of spray dried powder Up powder moisture content H2O slip water content

101

= 6800 kg/h = 6% = 38%


Tab. 5. Table for calculation of evaporative capacity of a spray drier and the quantity of spray dried product.

b) Calculation Determining the quantity of dry powder produced (Qs) Qs = 6800 × 94/100 = 6392 kg/h Where 94 is given by 100% of water in the spray dried powder (i.e. the 94 is the % of solids of the powder derived from the 6% water needed to press it). It is also known that the quantity of water in the powder and thus not evaporated by the spray drier is: 6800 - 6392 = 408 l H2O/h Determining the quantity of water contained in a slip with a 38% water content and 6392 Kg of dry material 38 : 62 = × : 6392 × = 3917 l H2O/h 102


where 38 = % of water in the slip 62 = % dry material in the slip. Determining the quantity of water effectively evaporated by the spray drier 3917 - 408 = 3509 l H2O/h N.B. This calculation is also used to establish the required drying capacity of a spray drier to be installed in a plant of specific capacity. 2 - If the evaporative capacity of the spray drier is known the quantity of moist spray dried powder produced can be determined. a) Known data spray drier evaporative capacity spray dried powder moisture content slip water content

= 3500 l/h = 6% = 38%

b) Calculation Determination of the apparent dry percentage and the water in the slip 62 × 100 / 94 = 65.95% (dry) Where 94 is given by 100% of water in the spray dried powder (94% dry material in the powder and 6% water). Determining hourly spray died powder output at moisture content of 6% 34.043 : 65.95 = 3500 : × × = 6798 kg/h

where 34.043 = % of water to be evaporated (from 100 - % apparent dry material in the slip) 65.95 = % apparent dry material in the slip 3500 = evaporative capacity of spray drier in l/h of water. Energy savings Energy savings at the spray drying stage can be obtained by varying the working conditions of the machine and the technological parameters of the slip. The main variables which have a significant effect on energy savings are: a) incoming hot air temperature b) difference in temperature between incoming air and outgoing air c) air recycling d) increase in the slip solids content. 103


a) Effect of temperature of incoming hot air Increasing the temperature of the incoming air reduces the amount of heat energy needed to dry the slip, for the same air flow rate. Alternatively, where the temperature of the air exiting the spray drier remains constant, the lower the hot air flow rate the lower the energy loss at the stack and the better the thermal efficiency. Fig. 33 illustrates how specific consumption varies as incoming hot air temperature increases. b) Effect of the difference in temperature between inlet air and outlet air The greater the difference between the inlet and outlet air temperatures, the lower the energy consumption per unit of product output. This difference can be widened either by increasing the temperature of the inlet air or decreasing that of the outlet air. c) Effect of recycling air In most cases the air exiting the spray drier is released into the atmosphere. This is undoubtedly a waste of energy, especially where its temperature exceeds 120 °C.

Fig. 33. Specific consumption of a spray drier as a function of hot air inlet temperature.

104


Recycling a part of the outlet air (from 10 to 50%) could provide considerable energy savings. However, recycling more than 50% of outlet air could, instead alter in-chamber hygrometric parameters to the extent that the drying speed of the atomised droplets is reduced. Trials have shown that energy savings of up to 20% can be obtained by recycling outlet air with a temperature of 120 °C. d) Effect of increasing the percentage of solid material A higher proportion of solid material in a slip can yield considerable energy savings; however, this option is not always open to the manufacturer in that an excessive percentage of solid can thicken the slip and cause mill unloading difficulties. However, recovery of waste spray dried material along the production line (e.g. in sieving, pressing etc.) or the fine material separated out by the cyclones is always possible. In a normal production plant this recovered material accounts for some 35% of the total and can be added into the agitated tank, thus preventing mill unloading problems. Nevertheless, it is important that the addition of this extra solid material does not cause excessive variations in slip viscosity and thixotropy and consequent problems at the spray drying stage. Recycling this recovered waste powder and collected dust back into the spray drier slip tank provides energy savings in the order of 3 and 7% respectively. Energy recovery Where the size and power of the spray drying system allow, the plant can be integrated with a cogeneration system. This uses the excess heat produced by an electricity generator (smoke from turbine or combustion engine) as hot air for the spray drier. The electricity can thus be used internally or, if there is a surplus, reintroduced into the mains grid. Recent legislation has provided incentives in this regard. Moreover, it is evident that where a cogeneration system provides a flow of hot air, a recovery system will be able to provide (necessary air temperature and flow rate remaining equal), a considerable reduction in fuel consumption per unit of heat generated.

105

Chapter III PRESSING

Introduction Pressing is that stage of the production cycle in which ceramic products are formed by compacting semi-dry granular powders. Pressing essentially involves: forming of the tile: to give it the required geometry. powder compaction: to confer certain mechanical characteristics. densification of the powders: to limit the empty spaces within the pressed item. Compared to other forming systems (extrusion, casting, etc.) pressing has considerable advantages: high productivity excellent repeatability of size parameters pressed pieces are easy to dry limited drying and firing shrinkage. Pressing systems Pressing powders generally have an evenly distributed moisture content which is important as it confers plasticity on the system and thus aids inter-particle cohesion. Depending on the remaining percentage of water, then, materials may be formed: in a plastic state: the mix has a residual water content of some 20-25%. Pieces are formed, for instance, by extruding the plastic mass through highly polished metal moulds having orifices corresponding to the finished tile dimensions: in a semi-dry state: these powders have a residual moisture content of around 10-15%. in a dry and/or semi-dry state: in this case powder moisture content ranges from 3 to 7%. Loose powders of varying grain size and morphology are normally used, but they must have good flow properties. Flowability largely depends on the shape of the grain and thus on how the powder was prepared, that is:


by dry grinding followed by granulation by spray drying of a ceramic suspension (slip). Compaction and densification of the powders is influenced not only by maximum press power but also by the intrinsic plasticity of the body, moisture content and particle size distribution. There are essentially three different kinds of press, each of which features a different energy application system: mechanical press hydraulic press isostatic press. Pressing semi-dry powders Different press types In industrial tile manufacturing processes that involve the pressing of semi-dry powders, only fully automated presses are used. Historically, it should be noted that, before the advent of hydraulic presses, toggle (or knuckle-joint) and friction presses were used. Toggle press This is a mechanical press featuring a system of two interconnected rods that reproduce a knee-like joint. During pressing the speed of fall of the punch decreases steadily as pressure increases (fig. 1). Friction press Friction-type mechanical presses are characterised by a fast, violent pressing action with a highly dynamic instantaneous impact. The fall and rise of the shaft that transfers motion to the mobile cross-beam is controlled by two rotating vertical discs located to the side of the screw-coupled flywheel (fig. 2). Contact with the flywheel takes place first with one disc then the other, one disc driving the punch down, the other raising it. The friction press is a high-output machine (up to 30 cycles/min) and, given the installed electrical power, features an excellent energy-productivity ratio. Of simple design, it is also easy to maintain. Its limitations lie in the difficulties of maintaining constant, uniform pressing. Hence it has now largely been abandoned in favour of more reliable hydraulic pressing systems. 108

Pressing

Fig. 1. Toggle press.

Fig. 2. Friction press.

109


Hydraulic presses The hydraulic press functions by way of a fluid that exerts a pressure inside a cylinder. The operating principle is based on the pressure increase obtained when the pressure exerted by a narrow column of fluid is connected to a cylinder with a piston of much larger cross sectional area. The greater the differential between the area of the two cylinders the greater the total force on the larger one compared to the narrow one. Modern presses, which use oils with special viscosity characteristics, have a hydraulic control unit that supplies the oil under pressure: once the oil is introduced into the circuit it exerts the compression force of the accumulator and multiplier through the cylinder on the material. This unit consists of two independent circuits, a high pressure one which feeds the actual pressing mechanism and a low pressure one which serves auxiliary units. A single or double-action cylinder unit is installed on the upper structure of the press; this shifts the cross-beam vertically as required and exerts high pressure on the punches at just the right moment. Summing up, the pressing action is effected by transforming hydraulic pressure into a deformation force as per the following formula: where F = Deforming force (in Kg) S = Surface area (in cm2) P = Oil pressure (in Kg/cm2).

F=S×P

It is thus possible to calculate: Po (oil pressure kg/cm2) = St (punch surface cm2) × Ps (specific pressure kg/cm2) / Sp (piston surface cm2) Ps (specific pressure kg/cm2) = Po (oil pressure kg/cm2) × Sp (piston surface cm2) / St (punch surface cm2) The main characteristics of the hydraulic press are: evenly distributed pressing force absolute repeatability of pressing cycles over time. The above characteristics make hydraulic presses especially suitable for use in highly automated production plants and as regards final product technical requirements in the pressing of products with high levels of firing shrinkage. Requirements for a modern press A modern ceramic press must: be reliable provide high output 110

Pressing

be automated be flexible be extremely precise at every stage of the cycle maximise energy savings be versatile.

Versatility refers to the possibility of choosing between a work cycle with maximum energy savings and a cycle with maximum speed. For a detailed description of the main parts of a press system it is best to refer to the specific technical manuals which accompany every machine. Nevertheless, note that a dry-pressing machine essentially consists of the press itself and equally important complementary devices such as: filler box feed devices die cavity loading devices dies powder removal devices.

Fig. 3. First-generation hydraulic press.

Different types of press for different tiles Generally speaking, ceramic tile manufacturers use three different machine ranges for the following purposes: 1. Presses for small tiles and trims. Low tonnage machines (< 500 tons) with sophisticated control devices and precision cross-beam movement, filler box movement etc. Moulds are complex and ejection of the formed piece may involve movement of the die box. Powder filling and pressing force can also be applied horizontally, with the mould closed. Installed power is generally low (20 or 30 kW). 111


Fig. 4. A modern hydraulic press (series 2000).

2. Presses for modern standard size pieces, i.e. single fire floor or wall tiles of up to 40×40 cm. These presses account for the bulk of the market and have maximum pressure in the 600-4000 ton range; characterised by very high productivity (up to 18-20 complete cycles per minute) and flexibility to meet the ever-increasing variability of production lots, often involving several size changeovers per week. For this reason die changeover, filler box or feed unit adaptation and machine adjustment tasks need to be as fast and simple as possible. A powder loading system that uses a filler box with a traditional grating (grid) is already seen as obsolete and can now be replaced by floating gratings supported directly by the die box antifriction inserts. Thus the filler box and load hopper effectively form a single unit and all that needs to be changed during a tile size changeover is the grating (figs. 5 and 6). 3. Presses for medium-large sizes. This range covers larger tiles (e.g. 30×60, 60×60 cm or larger) or tiles decorated on the press itself and not glazed downstream; these items require slower cycles and it is important that energy efficiency is not sacrificed. This is why more suitable, lower productivity presses have been developed, es112

Pressing

Fig. 5. Traditional filling system.

Fig. 6. Filling system with floating grating.

pecially now that medium-large tiles of this type are steadily becoming more popular. As a rule, hydraulic systems use proportional valves, or servovalves, and not the on-off circuiting seen on earlier models. This means that electronic setting systems can be used, thus guaranteeing precision and quality. The load-bearing structure is normally of the column and/or welded type. Presses in the 6001500 ton range have column structures, either of the pre-stressed type (to counteract lengthening of the structure at the moment of maximum load) or arch-like ring frames (made with rings either welded together or linked with threaded bars). 113


4. Presses for large tiles (fig 7). Pressing forces range from 4000 to 7000 tons and beyond [specific loads are generally in the order of 400-650 Kg/cm2 depending on the overall surface area being pressed. Higher pressures are only used on very lean (non-plastic) bodies or very dense granules]. Such presses are characterised by very wide clearance and, of course, sophisticated accessory systems (filler boxes, dies etc.). They are not used to make tiles as such but, rather, very large ceramic slabs (e.g. 120×180 cm); hence reliability takes precedence over speed. Feed systems are specially designed to ensure even distribution of the powder. The dies are made of several parts so that they can be moved around the plant with standard forklift trucks (they would otherwise be too bulky). Production cycles may be lower than 10 cycles per minute. Machine devices Ejector Newly designed piston ejectors have played a key role in making presses more flexible: recent years have, in fact, seen a switch from traditional ejectors, installed underneath the press, to ejectors fitted on the press bed directly beneath the die itself. Because this new design has shorter pistons movement is more precise. Moreover, proportional hydraulics and a greater number of lift points (where there used to be one there are now at least two) give more accurate ejection. To ensure maximum production flexibility, ejectors have even been integrated into the individual dies. This means that the latter can be adjusted outside the press, thus reducing plant downtime.

Fig. 7. A PH 7200 press for large slabs.

114

Pressing

Machine structure

1

Load bearing structure 2 Mobile cross-beam 3 Fixed cylinder support beam 4 Hydraulic piston 5 Mould air filter 6 Traditional ejector (SMU ejector on more modern machines) 7 Multiplier 8 Hydraulic SMU ejector (traditional ejector on older machines) 9 Filler box support 10 Filler box 11 Hopper 14 Filler box drive 15 Mechanical buffer, guards 16 Aspiration hood 17 Spring dampener 18 Hydraulic hoses 19 Hydraulic leak fluid container 20 Logic elements plate 21.1 Valve bank 21.2 Braking unit 22 Ejector and filler box control 27 Hydraulic control unit 28 Cooling pump 29 Guards 30 Electrical wiring 31 Electronic control unit 32 Electrical cabinet 33 Cross-beam position controller 34 Mould/die 35 PLC keyboard

Fig. 8. Main parts of a hydraulic press.

115


Different types of die used to form the tiles There are several traditional types of tile forming die, each of which have welldefined operating principles. Two such types are the entering punch die and the mirror die. Entering punch die The most widespread technological solution: during pressing the upper punches, fixed to the mobile cross-beam, penetrate the corresponding cavities of the die, which is fixed to the lower part of the press. The pressed tile is then removed from the cavity by raising the lower punches. Fig. 9 shows the various parts of an entering punch die (liners, punches etc.).

Fig. 9. Cut-away diagram of an entering punch die: 1) upper punch, 2) lower punch, 3) die box.

Mirror dies In this case, instead, the upper punch contacts the top of the die box, which is, in turn, mobile with respect to the die as it is connected to the lower part by elastic systems (fig. 10). The key characteristic of a mirror die, then, is that the die box is shifted by the force exerted by the upper punches on the die box itself. The resulting advantage is that as the tile is made the right way up the top (or face) side of the tile no longer scrapes along the die box. However, it is no longer possible to produce tiles with edge spacers either. Upper punches and box liners last much longer than they do on entering punch dies. Undoubtedly more complex from an engineering viewpoint, the mirror die is also more costly and relatively more difficult to assemble than an entering punch die. 116

Pressing

Fig. 10. Cut-away diagram of a mirror die: 1) upper punch, 2) lower punch, 3) die box, 4) base plate, 5) pusher.

Upper forming die (SFS) In the wake of increased demand for solutions that allow the tile to exit the press face-up (e.g. because decoration at the press is becoming increasingly popular), other types of die with special mechanical control systems for punches and die walls are becoming more common: such highly technological systems allow the tile to be produced face-up with a spacer, thus avoiding contact of face and die box (fig. 10a). This type of die solves a great many technological and aesthetic problems, allowing the manufacturer to produce high quality products such as coarse grain porcelain stoneware in which the larger particles remain in the upper layers.

Fig. 10a. Cut-away diagram of an upper forming press: 1) upper punch, 2) lower punch, 3) upper die box, 4) lower die box, 5) pusher.

117


Mirror die with pushers Very large tiles cause more marked expansion problems during pressing. Hence solutions have been developed that allow pressing at the same level as the die; this involves the interposition of spacers between the die box and the punch to aid degassing during the compaction phase. The mechanical set-up is optimised so that no alignment work is required (fig. 11). Ever-more sophisticated decoration at the press (especially in the porcelain tile field) has led to the development of ejectors that can be adjusted to different, electronically controlled positions.

mobile cross-beam

clearance adjuster spacers pusher

clearance

mirror top die

lower punch

die box box liners hydraulic support piston press bed

Fig. 11. Cross-section illustrating a mirror die with pusher.

Isostatic dies For some time now special isostatic (or, more properly, isostatic-effect) die sets have been used to optimise filling homogeneity right across the surface of the die cavity; isostatic dies have reduced finished tile defects considerably. Compression is achieved by way of a punch with a rigid back and a hard, yet deformable polymer front that comes into contact with the powders to be pressed; a chamber between the punch and the polymer membrane contains an incompressible fluid. Fig. 12 illustrates the operating principle behind these punches. 118

Pressing

Oil Powder

Fig. 12. Isostatic punch operating principle.

Isostatic punch

Fig. 13. Cross-section of an isostatic punch designed for industrial use.

The pressing sequence A general description of the filling, pressing and ejection sequence for the various kinds of die follows. The pressing cycle begins with ejection of the tile formed during the previous pressing; this ejection consists of the tile being raised by the lower punches from the bottom of the cavity until it is above and level with the surface of the die box. Figures 14, 15 and 16 illustrate the different stages of the pressing cycle (fillingpressing-ejection) for entering punch, mirror and upper forming dies respectively. The cycle is as follows: the grating-equipped filler box moves forwards and shunts the just-formed tiles away from the die box. Then the lower punches drop (first fall) and the cavities are filled with powder; the filler box returns to its home position. The lower punch then drops further (second fall) and the mobile cross-beam begins its descent. Initial first compression of the material then takes place: this is not violent, as its main function is to expel the air from the powder particles (de-airing) The second compression gives the tile its definitive shape and brings the material to the required degree of compaction. The cycle then recommences, with ejection of the pressed tile and a new filling of the die cavities. Electrical impulses emitted by proximity sensors installed at certain points on the press allow precision control and thus automation of all press movements by 119


Traditional mirror die Filling

Pressing

Ejection

Fig. 14. Tile forming sequence: mirror die. Entering punch die Filling

Pressing

Ejection

Fig. 15. Tile forming sequence: entering punch die. S.F.S. upper forming die Filling

Pressing

Ejection

Fig. 16. Tile forming sequence: upper forming die.

way of a feedback system. The impulses are sent to an electronic controller which recognises the process being carried out at that moment, calculates the next movement and relevant execution times and, in turn, transmits the necessary impulses to the hydraulic control unit which governs the pressing sequence. Main parts of the die The main die parts are: upper punch lower punch die box.

120

Pressing

For a more detailed explanation of parts the reader should consult the specific machine handbooks. There follows an overview of the main characteristics that distinguish the various types of pressing punch. Different punch types Hardened, polished steel Generally used where a high quality tile finish and very high resistance to wear are required. The polished surface of the punch gives an excellent tile finish and is therefore particularly suitable for porcelain tiles. Punches with moulded rubber surface The rubber surface (available in varying degrees of hardness) eliminates the need for systematic cleaning of the punches and thus the need for the brushing systems that perform this task. A rubber face is also an effective means of creating surface effects (indentations, rustic effects etc.). Even quite complex surface patterns or embossments (structures) can be achieved economically in this way. Punches with cast resin lining As above, cleaning times are reduced significantly. Unlike rubber surfaces, those in cast resin can be smoothed to give perfectly flat surfaces. Isostatic punch A tile back punch with a membrane on its face which uses the incompressible nature of the oil in its internal channels to compensate for uneven cavity filling. Characteristics of the ceramic powders used in pressing Pressing is intended to: a) provide tiles of a set size without generating defects on either fired or unfired tiles that are the result of pressing itself. One such defect is lamination (the formation of air pockets in the body); this shows up as fissures and surface damage that may appear at any stage of the production process. Also to be avoided are size, geometric and flatness (planarity) defects on the fired tile. All these problems can be traced, together with other causes, to density variations at different points on the same tile or from one tile to another.

121


The main causes of such variations are uneven die cavity filling and changes in press performance. b) contribute to the formation of a microstructure in the unfired tile that gives: the bending strength needed at subsequent stages of tile production: that strength must be available immediately after pressing and after drying. an unfired tile with a mass that is permeable and allows the gases formed during pre-firing and/or firing (combustion of organic substances, dissociation of carbonates etc.) to escape. gives a fired product of standard-compliant technological characteristics (firing shrinkage, porosity, bending strength) with normal firing cycles. Variables in the pressing cycle To meet the requisites in points a) and b) the characteristics of the spray dried powder and the working conditions of the press must be correct and comply with accepted standards. The main factors influencing pressing are thus: a) Characteristics of the powder being pressed. b) Characteristics of the pressing cycle. a) Characteristics of the powder being pressed. Before analysing the powders that are suitable for pressing and the relationship between their characteristics and those of the pressed tile, some clarification is required as regards: definition of terms physical characteristics of the particles technological characteristics of the powders. Definition of terms Particle: may be a primary particle or a cluster of primary particles; the latter is referred to as an agglomerate. Agglomerate: term associated with a small mass of primary particles kept compact by surface forces and/or inter-particle bonds. Grain/Granule: terms used frequently in ceramics where agglomerates are produced deliberately by adding an agent that aids re-granulation (in most cases the agent is water).

122

Pressing

Physical characteristics of the particles Powder characteristics BULK DENSITY Weight per unit of volume of powder; includes solid phase, porosity (internal air gaps) of individual grains and air between individual particles. SINGLE GRAIN BULK DENSITY Parameter that defines the weight per unit of volume: includes solid phase and internal voids in the individual grain. TAPPED DENSITY The bulk density of a known volume of powder after controlled vibration. Characteristics of powders definitions and measurements HAUSNER INDEX This is the ratio between tapped density and bulk density of the powder in question.

I.H. H.I. =

Dv Da

where Dv = tapped density (g/cm3) Da = bulk density of powder (g/cm3). BULK DENSITY Ratio between mass and apparent volume (M and V). M M Bulk dens. = = V tot Vpores + Vsolid REAL DENSITY Ratio between mass (i.e. quantity of material) and actual occupied volume. M Real dens. = V solid

123


TOTAL POROSITY Ratio between pore volume and total volume. Vpor Vtot Vsol 1 Vsol 1 Dbulk Porosity = = = = V total Vtot Vtot Dreal OPEN POROSITY Ratio between the volume occupied by the open pores (i.e. those connecting with the outside) and total volume. CLOSED POROSITY Ratio between the volume occupied by the closed pores (i.e. those not connecting with the outside) and total volume. This value is negligible. COMPRESSION RATIO Ratio between apparent and compacted volume. Mass Dapp. V app. D compacted Compressio n ratio = = = Mass V compacted D apparent D compacted

Characteristics and microstructure of pressed tiles depend on: nature of the particles (shape) particle size characteristics of the agglomerates pressing additives (percentage of water, binders, plasticizers).

In turn, the above factors determine and condition the properties and technological characteristics of the powders, namely: bulk density fluidity compression ratio friction angle of the powders. As mentioned, consistent distribution of the powder in the die cavity is a must for good pressing results. To attain this goal the powder must have good fluidity; this also allows manufacturers to use faster press cycles without compromising pressed tile homogeneity. Yet good fluidity is not, in itself, sufficient: the powder must also be of suitable bulk density. Excessively low bulk density generally means that large volumes of air have to 124

Pressing

be expelled during pressing or a thicker powder filling layer has to be used. Both cause difficulties in the pressing cycle. The bulk density of a powder depends on the volume of the intergranular spaces and the density of the granule or agglomerate. This value is determined by weighing a known volume of powder. Agglomerate density depends on how the powder was prepared (dry grinding, re-granulation, spray-drying) and on the shape and size distribution of the particles in the agglomerate. Bulk density values for different powders (granulates or spray dried powders) of specific grain size are given in fig. 17 and table 1. They illustrate that spray-dried grains have lower densities than agglomerates. Note also that re-granulate density increases slightly as grain size increases while for spray-dried powders the opposite is true because of the existence of the hollows inside the particles, which increase as particle size increases. Such cavities depend on particle shape and size distribution. The agglomerates obtained from dry grinding do not have any such cavities. Table 1 shows bulk density values for: spray dried powder dry ground powder (traditional) re-granulated powder.

Fig. 17. Bulk densities for different (granulates and spray-dried) semi-finished products.

125


MOISTURE UMIDITÀ (%) Regranulated body Granulato Spray dried body Atomizzato Dry ground bodytradizionale Macinato a secco

5.90-6.0 5-7 6-10

DENSITÀ APPARENT APPARENTE DENSITY (g/cm 3) 1.1-1.15 0.33-1.00 0.75-0.8

TAPPED DENSITÀ PER VIBRAZIONE DENSITY (g/cm3) 1.470-1.480 1.240-1.260 1.320-1.350

Tab. 1. Bulk densities for different types of powder (spray dried, dry ground and re-granulated for a given composition).

Powder fluidity Powder fluidity is one of the most important pressing parameters. It is generally expressed as: flow speed a Hausner index value. Flow speed is defined as the time taken for a certain volume of powder to flow out of a funnel-shaped container via a calibrated orifice and is expressed in cm3/sec. The Hausner index is based on the increased density of powder on vibration. The higher the inter-particle fluidity the higher the tapped density and the Hausner index. Although this is the most suitable scale for characterisation of powder fluidity, it is not generally used in routine powder characteristic checks. The effect of residual moisture content and grain size on powder fluidity is illustrated in figs. 18 and 19. In Figure 18 note that fluidity falls as moisture content rises, especially in the 58% range. A similar graph is obtained with agglomerates prepared using dry grinding methods. Fig.19 shows the effect of specific diameter for powders prepared by spray drying and granulation. In both cases smaller grain sizes give lower fluidity. With dry ground granulates particles that are less than 200 microns in diameter reduce flow speed to nearly zero. Those between 200 and 500 microns produce the optimum flow speeds. With spray dried powders higher fluidity levels are seen where particles are in the 125-400 micron range. Both spray dried powders and granulates are affected by the presence of low percentages of small grains. Powder particle size distribution Fig. 20 shows the particle size distribution of a dry ground powder after granulation while fig. 21 gives that of a spray dried powder. 126

Pressing

FLOW SPEED (cm3/s)

Fig. 18. Variations in powder flow speed against changes in the moisture content of a spray dried powder.

GRAIN SIZE (µm)

Fig. 19. Changes in fluidity for spray dried and granulated powders of specific diameter.

127


GRAIN SIZE (µm) Fig. 20. Particle size distribution of a dry ground, re-granulated powder.

GRAIN SIZE (µm) Fig. 21. Particle size distribution of a spray dried powder.

128

Pressing

Note that the granulate (dry ground and re-granulated) has a relatively high content of particles in the low-fluidity band (i.e. less than 200 microns). Compared to the spray dried powder (usually around 5% below 125 microns) the differences produced by the two powder production processes are very significant. Bulk density, tapped density and Hausner index as a function of variations in powder parameters Fig 22 shows, for a spray dried powder, variations in: bulk density tapped density Hausner index. as moisture content changes. Note that an increase in moisture content reduces both the bulk and tapped density: the former is most affected. Hausner index values increase. Fig. 23 shows how spray dried powder grain size influences the Hausner index. Note that the Hausner density curve is similar to the flow speed curve, having similar values in the 200-500 micron particle size band (corresponding to the maximum). Grains outside the 125-400 micron range give the lowest bulk density values. Those inside have the higher density and best packing characteristics. Fig. 24 shows how particle size influences the above-mentioned parameters, this time for a granulate. Again, the Hausner density curve is very similar to that for flow speeds (i.e. for particles larger than 300 microns the Hausner index remains fairly constant). Fig. 25 shows the linear correlation between the Hausner index and flow speed. For powders that do not flow easily and for which flow speeds cannot, therefore, be accurately measured, the best indicator of fluidity is undoubtedly the Hausner index. To see how moisture content influences the fluidity of traditionally prepared powders (wetted dry ground powders), Hausner density values were calculated for industrial powders of different residual moisture content. The results are shown in fig. 26 (with figures for spray dried powders for comparison). Note that moisture content has much more effect on the fluidity of traditionally prepared powders than on spray dried powders. This low grain flowability effect seen with dry ground/re-granulated products is due to the irregular shape of the particles and the high percentage of fine material. Ceramic powder pressing: technological aspects Evaluating how pressing pressure and powder moisture content interact in the compaction of spray dried powders is of great interest to ceramic manufacturers. It is, however, complex and difficult to interpret. 129


Fig. 22. Variations in bulk density, tapped density and Hausner index as a function of residual moisture content (spray dried powder).

Fig. 23. How variations in particle size influence bulk density, tapped density, and fluidity in a spraydried powder.

130

Pressing

Fig. 24. Influence of particle size distribution on bulk density, tapped density and powder fluidity (dry ground re-granulated powder).

Fig. 25. Relationship between Hausner index and flow speed.

131


Fig. 26. How changes in moisture content affect fluidity (Hausner scale) in spray dried, dry ground powders.

During pressing, powders, whether spray dried or dry ground, tend to resist external forces as illustrated in fig. 27 which shows how bulk density changes as forming pressure increases (pressure is expressed on a logarithmic scale). Here, medium-low pressures result in almost uniform densities while higher pressures result in steadily increasing bulk densities. With simple ceramic powders the above relationship is generally a linear one. Whether the product is dry or includes lubricants or binders, during the pressing cycle, at lower pressures, the particles or grains are rearranged, while at higher pressures there may be elastic and plastic deformation, sliding and abrasion or even actual fracturing and grinding of the particles or grains. This evidently involves the actual simple strength of the material itself hence the cited density-strength relationship. Powder compaction curves (e.g. fig. 28) may therefore show changes in gradient against pressure for different products, namely: dry ground powder (V) re-granulated powder (G) spray dried powder (A). Here, the first change in gradient (near-linear tract) for powders G and A occurs 132

Pressing

Fig. 27. Relationship between forming pressure and unfired density (logarithmic scale).

Fig. 28. Compaction diagram for different semi-finished products - (A) spray dried powder, (G) regranulated powder, (V) dry ground powder.

133


where bulk density of the compacted powder Pt, corresponds to a forming pressure of Pf. At this point re-arrangement of the grains is nearly complete (end of phase I) and fracturing and/or deformation of the grains begins (phase II). The flow pressure is considered to be a measure of the simple strength of the agglomerate. The second evident change (where the gradient drops towards the end of the graphs) takes place when the pressed spray dried granules, as a whole, reach densities close to those of the individual granule (Pg, see fig. 28). In this state the pores on the outside of the granule have been all but eliminated (end of phase II) and reduction of intergranular porosity begins (phase III new flattening out of the curves). The compaction diagram is one of the most appropriate tools for checking the behaviour of a granulate material during the compaction phase. Normally, though, to see how the characteristics of the spray dried powder (particle size distribution and moisture content) influence powder behaviour during compaction, simpler checks are run. Results from tests with a spray dried powder for the production of low porosity floor tiles are shown below. Tab. 2 shows its particle size distribution. The spray dried powder was separated into several particle size bands to evaluate the impact of a specific size band on the degree of compaction at the pressing stage. Tab. 3 shows these bands together with the average granule density (Pg) and bulk density compacted by vibration (Pv). The increase in bulk density was determined by plotting the compaction curve using the following variables: forming pressure of up to 600 Kg/cm2 changes in residual moisture content from 3 to 10% semi-finished powders of different particle size distribution. These tests demonstrated that the behaviour of an agglomerate during compaction and its resulting bulk density essentially depend on its moisture content, forming pressure (figs. 29 and 30) and, to a much lesser extent, on particle size distribution (fig. 31 - average grain size). As far as the mechanics of powder compaction are concerned, observe the following: 1. Increased bulk density of spray dried powders during compaction is in line with the models that have been put forward, studied and perfected in the case of nonplastic agglomerates. 2. Flow resistance (or flow pressure) Pf falls linearly as average particle size decreases and moisture content increases (fig. 31). 3. The compaction pressure (Pj) of the agglomerate decreases as moisture content

134

Pressing

Dimensione grano Grain size (µm) >750 750-500 500-400 400-300 300-200 200-125 <125

(%) 0.5 2.5 11 23 31 25 7

Tab. 2. Particle size distribution of the spray dried powder on which tests were carried out. Grain density (Pg) and density of spray dried powdersemicompattato semi-compactedper byvibrazione vibration (Ρv) (Pv) Densità del granulo (Ρg)bulk e densità apparente dellatomizzato

Fraction Frazione

Ρg (g/cm3)

Ρv (g/cm3)

Mix (I)(I) Miscela

-

1.110

125 - 200 µm (F)

2.047

1.130

300 - 400 µm (M)

1.990

1.090

500 - 750 µm (G)

-

1.035

Tab. 3. Grain density (Pg) and bulk density of spray dried products, semi-compacted by vibration, of uniform particle size.

increases (exponentially). The size of the particles is less significant for a given standard moisture content (fig. 32). 4. The behaviour of the spray dried agglomerate at the compaction stage and the bulk density of the pressed tile essentially depend on the residual moisture content of the powders (figs. 29 and 33); grain size is less significant (where mediumlow forming pressures are applied)(fig. 30). Relationship between bulk density of the pressed material and green bending strength (practical aspects) Fig. 27 illustrated how the bulk density of the pressed material is generally linked linearly to the logarithm of forming pressure. These theoretical considerations aside, it is, at this point, important to quote some data obtained from experience at a semi-industrial level; such data concerns the pressing of spray dried powders for vitrified single fire tiles. The graphs in fig. 33 show the pressed tile modulus of rupture at different forming pressures and moisture contents. They illustrate how pressing force and moisture content act synergistically on the bulk density values of the pressed materials, thus al-

135


Fig. 29. Changes in bulk density (d) as a function of pressing force and powder moisture content.

Fig. 30. Bulk density as a function of pressing force and particle size distribution.

136

Pressing

2

4

6

8

Fig. 31. Changes in flow pressure (Pf) of a powder as a function of its particle size distribution and moisture content.

2

4

6

8

10

Fig. 32. Changes in compaction pressure (Pj) as a function of particle size distribution and moisture content.

137


GREEN BENDING STRENGTH (Kg/cm2) PRESSURE PRESSURE PRESSURE PRESSURE

MOISTURE (%)

Fig. 33. Relationship between moisture content and green breaking load.

lowing attainment of uniform (bulk) density values by modifying one variable or the other. The graphs in figs. 34 and 35, instead, show green and post-drying bending strengths for spray dried powders of different moisture contents pressed at different pressures. In this case note how an increase in forming pressure and moisture content (improving powder plasticity and flow) give rise to increased green and dry bending strengths. This behaviour can be explained by the fact that the particles are brought closer together, thus reducing the volume of voids between particles. There is also increased inter-particle contact, forming bridges and links between them. Effect of forming pressure on firing performance of ceramic powders (practical aspects) The graphs on the preceding pages show the relationship between forming pressure and bulk density. However, in order to illustrate the full complexity of the problem, it is, at this stage, important to introduce data on the relationship between semi-finished product characteristics, conditions at the press and the final product that has been obtained in laboratories and test plants. 138

Pressing

Residue 3.7%

Residue 3.7%

Density (g/cm 3)

Pressure (g/cm2)

Density Moisture content

Pressure (Kg/cm2)

Moisture content (%)

Fig. 34. Residue 3.7%

Residue 3.7%

Bending strength (N/mm2)

Bending strength (N/mm2)

Moisture content

Moisture content

350

Pressure (Kg/cm2)

Pressure (Kg/cm2)

Fig. 35.

The relevant variables are: pressing force green pressed tile bulk density powder moisture content working conditions at the press characteristics of the product after firing (shrinkage and porosity).

Observations and data refer to vitreous bodies, where the relationship between the above parameters is more evident. Fig. 36 shows vitrification curves for materials of different bulk density at different firing temperatures. Note that for any given firing temperature higher bulk densities correspond to lower shrinkage and water absorption values. Other observations: 139


Dbulk Dbulk Dbulk

WA %

SHRINKAGE %

Dbulk

TEMPERATURE °C

Fig. 36. Vitrification curves for vitrifiable bodies of different bulk density.

a) The lower shrinkage seen on samples of higher bulk density is attributable to a lower percentage of pores in the unfired material. b) Close to vitrification temperatures, lower shrinkage and lower porosity give greater product stability. Fig. 37 highlights the relationship between unfired tile density and linear shrinkage even more clearly. Yet variations in density are not dependent on pressing force alone. A change in moisture content in the order of 1.5% is equivalent to a change in dry bulk density of 0.03 g/cm3. Another factor that causes variations in pressed tile bulk density is pressing speed (see fig. 38). Different bulk densities in different zones on the same tile a variation attributable to non-uniform filling or other factors discussed previously may cause uneven shrinkage within the same tile and thus dimensional defects. Fig. 39 gives a quantitative example of shrinkage variation where samples of different bulk densities are fired at the same temperature. Note that differences in bulk density of 0.06 gr/cm3 correspond to shrinkage differences in the order of about 0.3%; on a 330×330 mm tile this means up to 1 mm variation on the length of a side. Fig. 40 illustrates the problems that arise where a press cavity is incorrectly filled, resulting in different bulk densities over the tile.

140

SHRINKAGE %

Pressing

TEMPERATURE °C

Fig. 37. Relationship between unfired bulk density and linear shrinkage.

Pressure (Kg/cm2) Fig. 38. Variations in bulk density as a function of pressing speed.

141


Fig. 39. Shrinkage as a function of bulk density.

Future developments in pressing Current trends (2001) indicate that demand for very large, rectangular tiles with aesthetic characteristics ever-more similar to those of natural stone is set to increase further. On the manufacturing side this is sure to mean increased demand for larger pressing areas and higher tonnages so that even larger slabs can be produced: evermore sophisticated on-press decoration is likely to involve widespread use of body/ coloured granule mixes, glaze grains and other dry decoration techniques. In this regard there already exists a new technology that uses a double pressing process. This consists of a relatively simple first press, which forms and, where required, decorates the slab at a preliminary level of compaction; the slab is then conveyed to a small-scale dry decoration line before being compacted in a second press to give a green tile of the required definitive compaction. The advent of larger slabs, which realistically resemble natural stone, has also led to the appearance of new production technologies, which, in a certain sense, have much in common with the technology used in the cutting and preparation of natural materials. For example, there is no longer any need for a spacer gap or 142

Pressing

Fig. 40. Size defects caused by heterogeneous die filling.

chamfered edge as the slabs are laid without an inter-tile gap. In fact, the idea of producing mainly large slabs is beginning to gain consensus: this is because different tile sizes can be produced by cutting the green slab with specially designed conveying and cutting systems, thus allowing the manufacturer to by-pass complicated and costly die changeovers. This technology is also likely to have a significant impact on die manufacturing technology as slab-size dies need to be extremely long lasting. 143


Further improvements are also expected on the units that feed the powder into the die, while hydraulic circuits have already been optimised to confine high pressure to the main circuit only and thus prevent overloading of more delicate components: this reduces unscheduled maintenance, a factor that can seriously undermine production line efficiency.

144

Pressing

APPENDICES Green physical-mechanical characteristics of tiles obtained from different powders (spray dried/granulated/dry ground) Preparation of raw materials Body preparation involves a series of tasks that aim to provide a homogeneous material with an appropriate, relatively fine particle size distribution and a moisture content suitable for the employed pressing system (in the case in question powders with a moisture content of 4-7 % were pressed). The following processes were used: dry grinding process, consists of dry grinding the raw materials and then wetting them with suitable machines. wet grinding process, consists of in-water grinding of the raw materials and subsequent spray drying of the obtained slip. In dry grinding the (very finely) ground powders are treated to aid agglomeration of the particles. This is done with the aid of: standard wetting machines re-granulators. The observations and data below refer to tests that aim to compare the post-pressing characteristics of differently prepared materials (all intended for the manufacture of the same product). The following preparation methods were used: cycle with traditional dry grinding, wetting and re-granulation cycle with dry grinding and grain selection cycle with wet grinding and spray drying. Particle size distribution of powders produced using different systems Figures 41 and 42 compare the particle size distributions of dry ground powders to be wetted and granulated respectively. As mentioned, the fine particle size of dry ground powders has a considerable (and sometimes negative) influence on downstream production processes. While, in fact, fineness generally improves firing conditions because of the increased specific surface area, pressing conditions are generally worsened on account of the excessive quantity of very fine talc-like powders; this can lead to considerable de-airing problems (lamination). These difficulties disappear with granulation as the fine particles are agglomerated. Both granulation and spray drying agglomerate the particles obtained from grinding. The particle size distributions of the respective agglomerates are, nevertheless

145


Fig. 41. Particle size distribution of dry ground powders destined for wetting and granulation.

Fig. 42. Particle size distribution of spray dried powder and grains.

146

Pressing

(with the machinery and under the conditions in question) quite different: fig. 42 clearly shows that the particle size distribution of the spray dried material is largely concentrated around the middle of the spectrum (0.125-0.5 mm) while that of the granulate is, instead, spread more widely across it. Compaction and uniformity of pressing In order to evaluate pressing compaction and the uniformity of such compaction, penetrometer readings were taken on the pressed tiles and permeability measurements were made on the dried tiles. Fig. 43 illustrates typical penetrometer data distributions for the various raw material preparation technologies; pressing pressure is nearly identical, as is the position of the punch in the respective press. Observe that: tiles made from spray dried powders have the lowest penetrometer imprint depths (generally around 0.8 mm, against an overall average of 1.2 mm for the other types of pressed tile). The former are thus more compact. tiles made from spray dried powders are not only more compact, they also feature a compaction that is more uniform; the source of this greater uniformity is not attributable to different press management or control methods but, rather, the morphological and structural characteristics of the spray dried powder itself (more homogeneous cavity filling). This data confirms the superior quality of spray dried powders and their positive influence on compaction performance: that superiority stems from the optimal Penetration (mm)

Dry ground

Granulate

Spray dried

Fig. 43. Distribution of penetrometer readings for different body preparation technologies.

147


die cavity filling offered by the spherical shape and physical characteristics of the granules. The entity and distribution of penetrometer readings on tiles made from dry ground and granulated powders differ only marginally. On the other hand, permeability measurements on dried tile samples give a somewhat different picture (see fig. 44). Forming pressure remaining equal, the permeability readings for dried tiles made from granulated powder are higher (i.e. outflow times are shorter) than those obtained on tiles made with dry ground and spray dried products (N.B. flow time refers to the time taken for air compressed at 6 Atm to flow through the entire thickness of the tile: outflow time is inversely proportional to permeability). Once again, then, granulated powder is shown to produce tiles that are relatively less compact than their spray dried counterparts. Yet another important aspect regards the uniformity of permeability distribution and, therefore, compaction. Readings for granulates are, in this regard, very similar to those for the spray dried product, yet the dry ground tile is characterised by much greater discrepancies between the readings taken on different parts of the same tile; the tile is, in fact, generally more compact at the centre than at the sides. The fact that the granulate tends to be similar to the spray dried product in this regard can be attributed to the shape, particle size distribution and consistency of the granules, which provide good die filling performance. While this does not decisively and definitively show the granulate to be of high quality, it is, nevertheless, an aspect of some importance.

Outflow time (s)

Dry ground

Granulate

Spray dried

Fig. 44. Distribution of permeability readings for different raw material preparation technologies.

148

Pressing

Green and dry bending strength

2 2) BREAKINGLOAD LOADVALUE VALUE(n/mm (N/mm BENDING )

Green bending load is important not only as an indicator of the achieved degree of compaction but also as a performance parameter: it is an important qualitative requisite for tiles that will be moved and handled in their unfired state (as in single firing processes). Fig. 45 shows, for the three materials in question, green and dry bending strengths as a function of forming pressure; the diagram clearly illustrates the superior performance of the spray dried product, especially for dried tiles. At pressures of 230-240 Kg/cm2, the dried tile made with spray dried powder has a breaking load of 4 N/mm2, against approximately 3 N/mm2 for tiles made from granulate and 2 N/mm2 for tiles made from dry ground material. Once again the data seems to indicate that spray dried powders are of higher quality than granulates and that granulates are of much higher quality than dry ground materials.

Fig. 45. Green and dry bending strengths as a function of forming pressure for the three material types.

Porosity and fired tile bending strength The graphs in figs. 46 and 47 give, respectively, the porosity (expressed as WA %) and bending strengths of the analysed materials. 149


BREAKING LOAD VALUE (N/mm2)

Fig. 46. Water absorption (WA%)values for pressed tiles made with spray dried powder, powder granulates and dry ground material.

Fig. 47. Bending loads of fired tiles as a function of forming pressure for the three material types.

150

Pressing

The best porosity values are obtained with spray dried powder, then powder granulates and, finally, dry ground powder. Only at very high forming pressures (over 250 Kg/cm2) do the tiles made with dry ground powder feature a porosity similar to that attainable with granulates. Bending strength tells a similar story: the best results are obtained with spray dried powders, then powder granulates and, finally, dry ground powder. As with porosity, the bending strength of tiles made from dry ground materials only begins to near the performance obtained with granulates where forming pressures are higher than 250 Kg/cm2.

151


152


Chapter IV DRYING

As explained in Volume 1, the term drying refers to removal of the water contained in a solid by way of evaporation (i.e. its conversion from a liquid to a gaseous state). The tendency of a liquid to become gaseous largely depends on its chemicalphysical properties. Kinetic energy causes molecules to leave a liquid as vapour while, simultaneously, vapour molecules condense. The number of molecules abandoning the liquid depends on temperature; the number condensing depends on the number of molecules in a gaseous state that come into contact with the surface of the liquid in a given time and thus the number of molecules in a determined space and their speed (i.e. vapour pressure). To a lesser extent, dying also involves the so-called crystallisation, zeolite or structural water contained in certain minerals (such as montmorillonite, etc.). As it absorbs energy, water, like all liquids, has a tendency to become gaseous. This transformation does not continue indefinitely; it will cease when the vapour reaches saturation point (i.e. when an equilibrium between the number of water molecules evaporating into vapour and the number of vapour molecules condensing into liquid is reached). This equilibrium corresponds to a certain pressure, known as saturation vapour pressure, which increases with temperature. When, by increasing temperature, vapour pressure is raised to a point where it equals atmospheric pressure, boiling occurs (i.e. the vapour escapes not only from the surface of the liquid but also from its interior as bubbles). For all substances, lower ambient air pressures correspond to lower boiling points and every substance has, at a given temperature, its vapour pressure. The saturation vapour pressure of water at 100 °C is 1 Atm (fig. 1). Taking into account the amount of energy absorbed during liquid-vapour transformation, it can be seen that the amount of energy involved during drying is often underestimated. Drying is a process conditioned by three main factors: drying power of the environment: relative humidity, air flow speed etc. forces acting at capillary level. dimensional changes caused by release of moisture. The drying process, then, involves transfer of heat energy from the surrounding environment to the ceramic piece and simultaneous transfer of water in the opposite direction. The heat (energy) needed for the former may be received by the piece by way of convection, radiation or conduction: normally all three are involved. The most common of the above three hot air convection can, as a rough guide, be broken down 153

Vapour tension mm Hg


Temperature °C

Fig. 1. Saturated vapour pressure of water at different temperatures.

into the following stages: transfer of heat contained in air to piece transformation of water from liquid to vapour removal of vapour from surface of piece water transported in liquid state from piece interior to surface. When the evaporation rate at the surface exceeds interior-to-surface capillary transfer capacity, a further drying stage, during which the evaporation front shifts towards the interior of the piece itself, begins. This is generally represented by the classic convection drying curve for porous hygroscopic materials, which features three stages, each corresponding to different drying speed gradients and different depth levels. Conventional hot air driers for ceramics Drying speed is largely determined, then, by the capillary diffusion of water inside the tile. When drying starts (i.e. at the drier inlet) the temperature of the pressed tile is low and diffusion is very slow; evaporation speeds thus need to be contained so as to prevent compaction of the surface layer. This is why, in conventional driers, the incoming material is placed in contact with hot air that is relatively humid. Up to the surface of the pressed tile (i.e. the drying interface), heating occurs by convection; inside the pressed tile heating 154

Drying

occurs by conduction (note that a pressed tile generally has very low thermal conductivity). Heating speed is generally adjusted by modifying air temperature and flow rate. For a given material of defined porosity, the time needed for attainment of uniform heating temperatures that result in proper diffusion kinetics essentially depends on tile thickness. The drying cycle, of course, is influenced not only by thickness but by moisture content. Extruded or cast products, on account of their high moisture content, go through the phases of shrinkage drying, diminishing-shrinkage drying and zeroshrinkage drying; drying of pressed tiles only involves a fraction of the second phase and the entire third phase. Overall shrinkage is therefore much lower and the risk of tension caused by heterogeneous shrinkage thus relatively low. Such low moisture contents and the correlating shrinkage insensitivity, have, in recent years, resulted in the development of machines that employ extremely high heating rates. The purpose of this approach is to counteract natural shrinkage (which can cause fractures) with the opposite tendency of the material to expand when heated intensely: the tile is thus kept in a sort of dynamic equilibrium that allows use of extremely fast drying cycles (lasting just a few minutes). Technological considerations In any case, bear in mind that: diffusion towards the interface requires a moisture gradient. the speed of diffusion increases with the gradient; the pressed tile has an initially homogeneous moisture content which is subsequently reduced at the surface, thus aiding transport kinetics: there exists an upper technological limit to this phenomenon on account of the fissuring that occurs where drying is violent, although this can be avoided by taking the precautions mentioned above. diffusion speed is proportional to green permeability, itself dependent on the chemical-physical properties of the body (composition, particle size distribution, quantity and size distribution of non-plastic hard materials) and on the forming process and work parameters. water flow rates decrease as the migration route lengthens: thus very thick tiles have slower drying speeds. water viscosity inhibits transport kinetics by increasing resistance to flow. To reduce viscosity and aid diffusion the water is heated (heating involves the whole tile). Surface evaporation. This is the kinetics of the process and actually involves two elementary processes:

155


1. Interface evaporation (liquid-vapour equilibrium phenomenon). 2. Removal of the vapour (diffusive transport by way of concentration gradient), which is a function of temperature (or, rather, the vapour pressure of the water) and ambient thermo-hygrometric conditions (at the interface and in the gaseous phase). In conventional drying processes the hot air: carries the heat to the tile body and causes evaporation takes the vapour away. Thus water evaporation and vapour removal rates increase as air temperature and flow speed increase and are aided by low air humidity. Good quality drying is achieved when the two processes (diffusion and evaporation) have kinetics that allow the quantity of water evaporated per unit of time to be replaced by an equal quantity of water from the interior of the pressed tile (evaporation rate = diffusion rate); under standard process conditions the slower of the two is internal diffusion. For any given material (of established geometry, permeability, moisture) the only diffusion parameter that can be adjusted (via heating) is water viscosity. Drying speed = flow of water per unit of pressed tile surface area [Kg/(m2h)]. This last observation illustrates the direct relationship between temperature and pressure (or vapour pressure). At the above equilibrium the vapour is referred to as saturated vapour; if the pressure is below the equilibrium point the vapour is said to be unsaturated. In the case of saturated vapours pressure is merely a function of temperature. To shift the equilibrium a certain quantity of heat energy equal to the latent heat of the substance being evaporated needs to be applied to the system. This heat is peculiar to every substance and is defined as the heat needed to transform a standard unit of weight. Vapour condensation causes the latent heat to show up as calorific energy. The heat of vaporisation is very high if one considers that, theoretically, 605 Kcal are needed to evaporate 1 Kg of water once the water temperature has been raised to 100 °C. Water evaporation obviously diminishes as the degree of saturation of the space overlying the water increases. In ceramics evaporation problems are related to the fact that the water to be evaporated is combined with solids, thus restricting evaporation. The water forms a sort of reticule made up of numerous capillaries that wind their way in between the solid particles. The problem is how to transport this water from the interior of the piece, through the capillaries, to the surface where it can evaporate freely. To resolve this problem two factors and their respective speeds need to be taken into account: diffusion and evaporation. The problem of drying can said to 156

Drying

have been largely resolved when the evaporation speed does not exceed the speed with which the water is diffused from the interior of the piece to the surface and thus the water can evaporate at a speed that is high enough to be economically viable yet not so fast as to cause product damage. Drying is thus piloted by the DIFFUSION SPEED of the water in the interior of the pressed tile and the subsequent EVAPORATION SPEED. The major evaporation conditions are therefore: SURFACE EVAPORATION: occurs under conditions of thermodynamic equilibrium with the surrounding environment, with liquid water being drawn from the interior of the pressed tile in quantities equal to those being evaporated. EVAPORATION UNDER PLASTIC CONDITIONS: occurs in capillaries and interstices at the liquid-vapour interface under equilibrium conditions, with vapour being simultaneously transferred to the surface via the capillaries themselves. The process occurs under dynamic conditions, with transfer of the interface zone deeper and deeper into the piece. INTERNAL EVAPORATION WITHOUT FLOW: similar to the above but there is not enough moisture to generate flow or settling of the clayey particles and thus removal of the water generates interstitial porosity. A study of the graphs found in specialist literature shows that: the volumetric stability that follows drying-induced shrinkage corresponds to a full/empty volume ratio of approximately 75/25. the interstitial water which has reached post-drying stability varies from 10% to 12% by weight of the solid. dimensional variations corresponding to moisture contents below 10-12% are virtually negligible. pressed tile dynamic equilibrium is achieved when the degree of water evaporation corresponds to the degree of interstitial water diffusion in the tile interior; thus in each situation there is a certain evaporation speed, said to be critical in that beyond this speed dimensional variations can create tensions which can, in turn, cause damage. evaporation speeds above the 1st critical Bigot point essentially depend, temperature remaining equal, on ambient conditions; below the 2nd point they depend only on the chemical-physical characteristics of the ceramic material. DIFFUSION SPEED depends on: mineralogical characteristics of the material, which directly influence the orientation of the water molecule and thus its ability to migrate. particle size and shape, which influence width and shape of the capillaries via which the water is diffused. percentage of water content in the piece: as water content drops the solid particles converge (shrinkage only stops when the particles touch each other). This convergence narrows the empty capillaries and seals some of them off, resulting in increased obstruction. 157


pressed tile temperature speed and extent of shrinkage, itself a function of evaporation surface tension of the liquid to be evaporated relationship between vapour pressure of atmosphere, hygrometric state, relative humidity During shrinkage the capillaries narrow and some of them are sealed off, thus increasing resistance to diffusion of the water through the mass of the tile. According to some researchers this resistance is inversely proportional to the fourth power of the diameter of the capillaries. However, it has been found that the shape of a capillary affects diffusion more than its size and that this shape depends on the size and structure of the surrounding particles. Finally, diffusion can only occur if there is evaporation on the surface of the piece. Since evaporation depends on the ratio between the surface tension of the liquid and the atmospheric vapour pressure overlying the system, it is obvious that diffusion speed also depends on this ratio. EVAPORATION SPEED depends on: air temperature: the higher the air temperature the greater the quantity of water it can hold in suspension air speed hygrometric state of air: determines, at a given temperature, the quantity of water that can be removed from the material surface area exposed to air and angle at which piece is tilted with respect to direction of air flow diffusion speed of the liquid in the material surface tension of liquid to be evaporated relationship between vapour pressure of atmosphere Control of air volume, temperature and speed is made easier if its hygrometric state is known.

158

Drying

Glossary ABSOLUTE HUMIDITY the quantity of water per Kg of dry air. vapour pressure of air RELATIVE HUMIDITY saturated vapour pressure quantity of water vapour per Kg of dry air PERCENTAGE HUMIDITY quantity of water vapour that would contain a Kg of dry air if it were completely saturated VAPOUR PRESSURE The vapour pressure in equilibrium with liquid pressure. SURFACE TENSION The force that acts on the surface of a liquid, tending to minimise its area. WET BULB TEMPERATURE Temperature obtained by immerging the bulb in water, the surface of which is exposed to the air. The sensible heat transferred from the air to the water is equal to the latent heat of evaporation of the water. HEAT OF VAPORISATION The quantity of heat energy needed to evaporate a unit of liquid, at constant pressure and temperature. Also known as enthalpy of vaporization, heat of evaporation and latent heat of vaporization. SPECIFIC HEAT The quantity of heat needed to increase the temperature of a material by 1 °C, without any chemical or phase changes. PSYCHROMETRIC (OR HYGROMETRIC) CHART A graph, each point of which represents a specific condition of a gas-vapour system (such as air and water vapour) with regard to temperature (horizontal scale) and absolute humidity (vertical scale); other characteristics of the system, such as relative humidity, wet-bulb temperature, and latent heat of vaporization, are indicated by lines on the chart (fig. 2). From the above it is evident that the drying process is conditioned by the hygrometric state of the air at different temperatures. Rational use of a drier thus requires that air humidity, especially in critical zones, be monitored carefully. Since

159


Fig. 2. Psychrometric chart.

160

Drying

evaporation results in cooling, the temperature of the solid is lowered to an extent that depends on the dryness of the air. The device used features two thermometers: one has a naked bulb and measures ambient air temperature directly and the other has a bulb sheathed in a cotton cover that keeps it constantly wet. The two bulbs give the same temperature reading only when ambient air is saturated (100% relative humidity) because in this situation water evaporation ceases altogether. Vice versa, the difference between the two temperatures is at its greatest when ambient air is completely dry. Wet bulb and dry bulb thermometric measurements allow researchers to trace charts that show the humidity of air-vapour mixes at a constant total pressure of 1 atm (fig. 2). Proper drying requires: careful evaluation of thermo-hygrometric process parameters proper body formulation which translates into reduced shrinkage and limited deformation. the bending strength needed to oppose the stress of drying and allow handling during subsequent stages of production. Thus the main parameters that need to be taken into consideration in body formulation are: percentage of plastic raw materials: these increase the cohesive strength of dried objects, but also increase shrinkage. mineralogical nature of the clays: deflocculated clays in their natural state form a compact, impermeable tissue during drying. They adsorb cations with some difficulty, thus forming more open, permeable structures that allow faster drying. percentage of non-plastic raw materials: these worsen unfired bending strength but help contain shrinkage (especially where particle size distribution is fine). water content: increases shrinkage and plasticity. Particle fineness is, in fact, proportional to the hygroscopic value of the body. forming pressure: influences drying behaviour in that higher pressures reduce permeability and water diffusion. piece geometry: this is very important. Drying is faster for high surface area/volume ratios (thinner tiles dry faster). Machines The driers used in the ceramic tile industry are of both the horizontal and vertical type. Although the technological principles behind them are essentially the same, the dynamics are somewhat different in that variations in the cycles significantly 161


alter the times over which shrinkage, increased bending strength etc. exert their effect. Working conditions on the machine and air parameters (maximum temperature, speed, volume and hygrometric values) are, therefore, necessarily different on the two drier types. Rapid vertical driers High performance rapid vertical driers are ideal for automatic tile production lines. They are designed to provide optimum technological results: low residual moisture content, uniform tile temperatures, good product bending strength and low energy consumption (fig. 3). An automatic vertical drier is essentially made up of a vertical framework in which a chain drive system moves a series of racks (or baskets), which consist of a set of rollers to carry the tiles. Tiles are loaded automatically without touching each other onto the racks by the rollers as they rotate. An air flow generated by burners and fans dries the tiles by convection; this process yields a product that is suitable for the subsequent stages of the production process (glazing, firing etc.), that is, one which has:

Fig. 3. Rapid vertical drier.

162

Drying

low moisture content good bending strength a set temperature. The diagram below (fig. 4) illustrates the main parts of a vertical drier. Fig. 5, instead, shows a plan for a vertical drier with a pre-ventilation unit and a cooling unit at the outlet. Operative conditions in the drier (air temperature, humidity and flow rate) are controlled by three independent units that form three differentiated drier zones. In the first zone (where the material enters the drier and is lifted) the tiles are exposed to an air flow, produced by a fan and burner, which runs parallel to the tile surface. Intake and outlet gratings are used to distribute the air and direct it 1 Load-bearing structure 2 Insulating panels 3 Rack (basket) chain 4 Drive unit 5 Racks (baskets) 6 Rollers 7 Lubrication unit 8 Service platform 9 Upper service platform

10 Internal ducting 11 Burner fan 12 Burner 13 Stabilisation unit 14 Stack 15 Pneumatic system 16 Electrical system 17 Electrical cabinet 18 Control panel

Fig. 4. Cut-away diagram showing main parts of a vertical drier.

163


SECOND ZONE

FIRST ZONE

COOLING ZONE

STABILISATION ZONE

Fig. 5. Air circulation with cooling unit.

to the various rack levels throughout the drier. Once the air has traversed the racks, it is re-aspirated into the internal ducting where most of it is re-circulated by the fan and a small (adjustable) portion is expelled by the exhaust fan into the atmosphere. The circuit in the second zone (where the tiles are lowered) is similar to that in the first zone yet independent from it, in that it has its own fan, burner and external air intake damper. A further fan controls the stabilisation zone, the last area dedicated to the actual drying cycle. Significant resources have been channelled into the development and perfection of this section and so meet the need for uniform tile temperatures. To achieve stabilisation the tiles need to be exposed to a correctly directed flow of constant-temperature air, obtained by mixing cold external air in quantities batched by a modulating valve with the hot air exiting the second zone. 164

Drying

This stabilisation process controls the tile properties required at subsequent stages of the production process (glazing, application of soluble salts, dry polishing etc.). The drying cycle normally exposes tiles to high temperatures immediately (assuming they can withstand initially fast water removal) so that a lower temperature (closer to the final one) can be set in the second zone. This maximises stabilisation efficiency and produces high bending strengths. Where tiles are somewhat fragile the cycle can be modified accordingly: this is achieved by creating (via dampers) an initial tile heating zone with reduced air flow that allows temperature to rise gradually. To obtain very low outgoing tile temperatures (below 65 ºC), the drier can also be equipped with two axial fans that introduce ambient temperature air into the lower drier zone. It is also possible to reach an outgoing tile temperature of 40-50 °C. Smooth tile loading and unloading and versatile tile size accommodation are provided by infeed and outfeed rollers, precise on-rack tile arrangement, a twin roller drive system which prevents slippage and the encoder which controls rotation.

Drying time (minutes)

Rack movement is controlled by an inverter on the chain drive motor; this system provides maximum precision and also protects against overloads. Furthermore, guides placed along the entire rack route ensure proper positioning at all times. The racks are supported by special arms fixed to the chain: these make movement friction-free and thus reduce energy absorption. Thick insulation reduces heat loss to a minimum. The potential of these driers is illustrated in the following graphs and tables (fig. 6 and tab. 1).

6 rows 5 rows 4 rows 3 rows 2 rows 1row Press speed (cycles/min)

Fig. 6. Relationship between press output capacity and drying time.

165


UNFIREDCRUDO SIZE (mm.) FORMATO (mm) width larghezza 161 161 215 269 215 323 355 323 430 538 645

length lunghezza 161 215 215 269 323 323 355 430 430 538 645

thickness Spessore 6 7 7 8 8 8 9 9 10 11 12

TEMPO DI ESSICCAMENTO APPROX.DI DRYING TIME (min.)(min.) INDICATIVO RIFERIMENTO

30 35 35 41 44 44 49 50 53 58 64

Tab. 1. Drying times for tiles of different size and thickness.

Horizontal driers Experience acquired with the first fast driers has led to a greater understanding of some fundamental concepts vis-à-vis drying of pressed tiles with very low residual moisture content (4-7%). 1. Ventilation: air flow must be as parallel to the tile faces as possible to ensure that heating is homogeneous and vapour is efficiently removed from the tiles. 2. Drying air temperature: plays a key role in evaporation speed. 3. Water evaporation speed: less influenced by speed and relative humidity of air than on vertical driers. Horizontal driers can have from just one to as many as five layers of rollers (channels). They are enclosed in an insulated chamber fed with hot air generated by dedicated burners and/or exhaust air recovered from the kiln (see figs 7 and 8). Figures 9, 10 and 11 chart the influence of the main tile drying variables. These are: W = Tile moisture percentage by weight V = Speed of air over the flat tile surface Wr = Relative humidity of the drying air T = Temperature of the drying air. Note: the term drying speed refers to the weight (Kg) of water freed from the surface of one square metre of tile surface in one hour. The graphs show that drying speeds are much higher where samples have a very high water content (e.g. 12%). 166

Drying

Fig. 7. Multi-level horizontal drier.

Fig. 8. Horizontal drier.

Note also that where pressed tile water content is quite low (5-7% or lower) the relative humidity of the drying air is no longer significant. This is an advantage since fast driers sometimes operate discontinuously and relative air humidity can vary considerably, especially during drier start up and stoppages. In (fast cycle) horizontal drying the tile behaves, from a technological viewpoint, more or less as described in the theoretical explanation of drying given in Volume 1. 167


Drying speed (Kg/m2h)

Air speed (m/s)

Fig. 9. Relationship between air speed and drying speed expressed in Kg/m2h (W = tile moisture %). Drying speed (Kg/m2h)

Air humidity (%)

Fig. 10. Relationship between air humidity (%) and drying speed expressed in Kg/m2 h (W = tile moisture %). 168

Drying

Drying speed (Kg/m2h)

Air temperature (°C)

Fig. 11. Relationship between air temperature (°C) and drying speed expressed in Kg/m2 h (W = tile moisture %).

It may be the case, however, if pressed tile moisture content is very low, that evaporation can be accelerated using a non-standard drying cycle that raises tile temperature to over 100 °C, changing the water into steam while it is still in the capillaries. Evaporating conditions Table 2 shows maximum water evaporation speeds for certain materials as a function of treatment temperature. MAXIMUM DRYING SPEEDS (Ve MAX) (Kg/m2h) as a function of treatment temperature Body type (spray dried products) Red monoporosa. 5.3 mm thick, moist. cont. 4% White single fire 8.8 mm thick, moist. cont. 4.7% Red single fire 8.8 mm thick, moist. cont. 5.8%

Treatment temperature 180° 220° 260° 1.2 1.5 2 1.6 2.2 2.9 1.7 2.4 3.1

Tab. 2. Maximum evaporation speeds as a function of treatment temperature (Ve max Kg/m2 h). 169


Fig. 12, instead, indicates in graph form how maximum evaporation speed (Ve max, Kg/m2h) changes as a function of treatment temperature. By taking into account drying speeds attainable for tiles of different size and thickness, it may be affirmed that high evaporation speeds can be obtained without any damage to the tile using hot air alone, without having to use other drying systems (described below).

Fig. 12. Evaporation speed (Kg/m2h) as a function of treatment temperature.

Infrared rays Rather than drying, this is simply heating, because when electromagnetic radiation with a wavelength of 0.8-3 microns encounters a ceramic body it only penetrates a few tenths of a millimetre as even water blocks rays of this wavelength. The surface of the tile thus heats up more quickly than the interior as heat is diffused inside by conduction. While actual drying always requires an air flow to remove the water vapour, it can only take place quickly where the temperature of the tile is high enough to reduce the water viscosity in the capillaries significantly. However, this method has two major drawbacks: firstly, it uses high temperature radiation sources that can cause the material to overheat should the tile stop under the source and, secondly, electricity costs are prohibitive.

170

Drying

Microwaves Ceramic bodies could be heated with microwaves in the 2.5 GHz range. These are directed at the material via specially designed wave deflectors. Water molecules, which behave as resonating dipoles, dissipate energy and it is thus possible to heat the whole of the piece at once (albeit heterogeneously as there is the side effect to take into consideration). The thus-generated vapour is carried away by a hot airflow. Drawbacks: a) Very high plant investment costs. b) Very high drying costs in that the energy, although used efficiently, necessarily requires an electrical source. c) Possible safety and environmental hygiene problems in that radiation leaks can, at such wavelengths, be harmful in many ways and damage the human central nervous system. Technological aspects The technological aspects of fast horizontal driers are, on the whole, similar to those of vertical driers. As before, body formulation needs to take into account the aspects described previously and the main drying parameters (drying cycle times, speed, humidity and temperature of drying air). As with vertical machines, drying requires: attentive evaluation of thermo-hygrometric process parameters proper body formulation which translates into reduced shrinkage and limited deformation. a bending strength that will withstand drying stress and allow handling during subsequent stages of production. The parameters to be taken into account for correct body formulation are the same as those listed for vertical drying. Relationship between drying cycle and tile thickness/size Drier size and drying cycle duration depend, to a large extent, on the dimensions of the tiles. Pilot plant tests and comparisons with results obtained on an industrial scale have led to the establishment of a simple formula which gives appropriate drying cycle times as a function of tile thickness and longest side length.

171


These formulas are given below and refer to two different lay-outs: PRESS

➪

DRIER ➪

GLAZING

Teo = [(S)3/2 + (0.02 × Lm)1.3] × Ko where Teo : drying time in minutes Ko : coefficient of material: 0.40 white and red single fire 0.45 white and red monoporosa 0.55 porcelain tile (soluble salts) S : total thickness mm Lm : longest side in mm PRESS ➪

DRIER ➪ STORAGE

Teo = [(S)3/2 + (0.02 × Lm)1.3] × Ko where Teo : drying time in minutes Ko : coefficient of material: 0.50 porcelain tile (plain, salt and pepper) 0.55 porcelain tile (large grains) S : total thickness mm Lm : longest side in mm. General observations It is possible, then, to use very high evaporation speeds without damaging the product. Hence there is a clear temptation to build driers with very fast cycles. Yet, on modern driers, the most desirable technological parameter is not so much short cycle duration as constant temperature of the outgoing pieces. This is because any variations in this parameter can seriously compromise glazing results. Utilisation of air at temperatures that provide control via special thermoregulation devices over outgoing piece temperature, even where the process has been complicated by lengthy stoppages, is only possible where the drying cycle is of a certain duration. From a thermodynamic viewpoint, the use of very high temperature air would force manufacturers to operate with very high vapour contents so as to prevent excessive loss at the stack but this could result in undesirable condensation on the pieces themselves. The key requirements with this type of drying are, then:

172

Drying

1) Drying cycles must be suitable for tile thickness and temperatures must not be excessively high. 2) Outgoing piece temperature needs to be controlled. 3) Specific energy consumption must be contained. 4) Mechanics must be simple and the machine compact. Generally speaking, all types of press have their own matching drier with an inlet frequently designed to accept twice the number of tiles produced at each pressing cycle; this results in less costly, simpler feed units. Moreover, less space is needed. Installation of machines with wider cross-sections has been made possible thanks to a device that allows roller removal even where very little side space is available. The multi-layer horizontal drier system is hardly an original one as it has been used by the heat treatment industry for decades and, moreover, had already been adopted (in part) on vertical tile driers in split ventilation models (i.e. the drier is designed as a series of ventilations cells as illustrated in fig. 13).

Fig. 13. Multi-layer horizontal drier (the diagram illustrates air flow routes).

Here, each cell is, from a thermo-hygrometric point of view, fully independent as it is equipped with its own circulation fan, hot air generator, air intake and extraction port (which carries exhaust air to the stack). Re-circulation air enters and exits the cell via hopper-like ducts through the side walls opposite the drive units, thus providing more space and a well balanced air feed on the different levels. The system also employs an upside down emission/intake system over the full length of the drying cell. Air flows through openings of variable aperture while specially positioned ducts and deflectors ensure uniform cross-flow. During drying the tiles are exposed to an air flow parallel to their faces; each 173


drier level is bordered by a series of removable metal sheets that define the treatment cell. The exhaust air is extracted at a point designed to yield maximum thermodynamic efficiency (i.e. where the enthalpy or total heat of the re-circulation air is at its minimum). Machines of this kind thus make it possible to construct proper drying curves that include a final stabilisation zone, extremely useful for the control of outgoing piece temperature. In the event of a glazing line stoppage the material may halt inside the drier. This could cause temperature fluctuations on outgoing tiles so regulation logic sets a holding temperature, attained via settable gradients; moreover, internal temperatures can be lowered quickly by servomotor-operated ambient air intake valves. Roller drive units, of the traditional type used on kilns, feature helical gears in closed oil-bath casings. This system is very reliable and is virtually maintenancefree.

174

Drying

Chapter V GLAZE AND DECORATION APPLICATION TECHNIQUES

Glazes can be used in several physical states: as aqueous and non-aqueous suspensions, granulates (sintered or pelletized grains) or powders. Solid, granulate and powdered glaze preparation techniques are generally the domain of glaze manufacturers. Ceramic plants themselves mostly handle the grinding of composite glazes to make them suitable for the delicate application stage. Glaze preparation 1) The glaze preparation department This department should suit the specific plant and logically be sited, where possible, between the storage areas and the glazing lines to minimise transport. In the past glaze preparation departments were often situated in underground areas below the glazing department so that the latter could be supplied directly. Today a system with all the tanks and machines above ground level is preferred: other solutions, generally more expensive and less ergonomic, are used only where space is restricted or there are insurmountable logistic problems. A modern preparation department should be a well organised operation, working to precise specifications so as to supply the application station with a reliable, consistent product. To be economically viable it will necessitate a variety of different sized machines. Taking an average size factory as an example (2 million m2 of single fire tiles per year), the required machine pool must include at least: 2 mills of 5000 l capacity 2 mills of 2000 l capacity 2 mills of 1000 l capacity 1 mill of 500 l capacity 1 mill of 200 l capacity. The department may be producing many different glazes at any one time, requiring at least 2 agitator-equipped tanks (each capable of holding two mill loads) for every ball mill. This allows grinding to be carried out mainly at night, thus reducing energy costs and noise-related problems.

175


Fig. 1. Alsing mill: cross-section, internal view and close-up showing grinding media and lining.

2) The machines Preparation departments usually have the following: Alsing mills, of various capacities. While the 5,000 litre models are the most common, mills can range from the small 300 litre units used for experimental or control tests to the 12,000 litre units for large-scale production runs. Generally equipped with sintered alumina grinding media (which wears at a rate of 5002000 g/ton of ground glaze) and a sintered alumina lining that lasts about 20,000 hours. Rubber and silica are no longer employed in this field. The discharge plug must be made of stainless steel. Energy consumption is in the order of 10 kW/ton of ground glaze per grinding hour at 17 RPM. Agitator tanks. Available in a range of types: octagonal, plastic or tile-lined cement, stainless steel cylinder. Average holding capacity is 5 m3 (approximately equivalent to the contents of two 5,000 litre mills). Tanks must be covered and are equipped with agitators to prevent settling: these consist of a gearmotor mounted directly on the shaft that supports the glaze stirring device. The agitators rotate slowly (about 20 RPM) and sometimes discontinuously (from 10' to 20'/h). Hence the gearmotor system must be designed to withstand the stress caused by repeated restarts. Modern agitators feature inverters that allow RPM to be adjusted as and when desired. A splash pan should be installed between the gearmotor and the shaft to catch any leaking oil and thus prevent contamination of the glaze. The agitator blades should be shaped and sized so that they pass as close to the sides and, most importantly, the bottom of the tank as possible: this prevents sedimentation or, even worse, solidification. Helix and turbine agitators also exist, but these are generally employed as mobile units. 176

Glaze and decoration application techniques

As these run at high RPM they must be used cautiously because the glaze could overheat or air bubbles could get into it, with serious consequences vis-à-vis rheological behaviour. There must be enough tanks to allow proper storage of the glaze. Generally speaking, there should be one tank for current use (from which the glaze is withdrawn and sent to the application stations) and a holding tank, where checking and correcting are done. Emptying operations are best performed with the aid of flex hoses and rapid ball couplings with butterfly valves as these are less likely to jam. Couplings and valves should be made of stainless steel: anything containing copper, bronze or brass must be avoided. Pumps. Several types are used. The most common are: Diaphragm pumps for low pressure glaze emptying and transfer operations. These have the advantage of being very safe, having no electrical system and a wide range of flow rates. Peristaltic pump. Squeezes the slip through a wear-resistant rubber flex hose by way of a rotor with three or more spokes, each tipped with a roller. Has a less marked pulse than the diaphragm pump and is self-priming. Has become very popular over the last decade thanks to good performance even under difficult conditions and low maintenance costs. Centrifugal pump. The most economic type. Rubber lined and stainless steel versions are available to prevent contamination. Not self-priming. High pressure unattainable. Frequent, yet inexpensive maintenance required. Mono pumps of the displacement type. Pressure is provided by a spiral steel screw rotating inside a rubber seat. They must never be run empty and require rather a lot of maintenance. Now used sparingly for both glaze and slip transfer because of high costs. Suitable for long distance transfers on account of high pressures (from 5 to 7 Atm). Vibrosieve. High efficiency circular vibrosieves are now the norm. Efficiency depends on sieve diameter and vibration frequency, which can be optimised through use of an inverter. However, as yet, there are no machines capable of sieving efficiently between 80 and 90 microns at high flow rates and densities. By using the largest diameters available (about 120 cm) it is just possible to reach 110 microns, and even then not for every kind of glaze. Magnetic separators. Removal of magnetic impurities from the glaze is essential. While progress has been made on the technological side, much still depends on the attentiveness and skill of the worker. If, in fact, static magnetic separators are not washed at the right moment they can actually cause the every damage they are intended to prevent: this is because tiny particles, harmless on their own, can agglomerate to form needles, which, once they reach a certain size, could detach and be carried away by the glaze flow, thus damaging output.

177


There are several types of iron remover. The most common are: Bars. Generally consist of cartridges, with three cylindrical, magnetic bars welded to its cover. Limited contact area, but can be installed in groups or series. Easy to clean and inspect. Honeycomb. These provide more surface area. Made up of permanent magnets in the form of overlaid perforated discs. More difficult to clean. Roller. The most recently developed and most costly solution. Consists of a roller (diameter 250 mm, length 800 mm) partially immersed in the glaze. The electromagnetic roller rotates on its own axis and extracts magnetic impurities from the glaze; these are then removed by a scraper. Operation is continuous. On more recent models the 3000 gauss electromagnetic roller, which heats up and loses efficiency to the point where it supplies just 1500 gauss, has been replaced by a 5000 gauss permanent magnet. Magnetic separator efficiency is, of course, inversely proportional to glaze flow rate and viscosity. 3) Scales and weighing systems On the whole ceramic companies would be well advised to have a clear understanding of how their glazes are composed. This would allow them to optimise use of frits and other components, thus avoiding excessive stocks and the corresponding control and management problems. However, this involves more work on the weighing side and the possibility of error. Moreover, a system that allows accurate monitoring of materials purchased becomes indispensable. Intermediate solutions exist: for example glaze manufacturers can supply already made-up sacks or bags that can be modified by adding other components in the ceramic factory. This makes weighing tasks easier yet control of incoming components more difficult. Finally, it is even possible to purchase fully pre-packaged compounds. In many cases these three systems co-exist, with the best mix being chosen on a case-by-case basis. Note that the glaze preparation department is generally equipped with a series of electronic, digital display weighing units with the following weight ranges: 2000 or 3000 Kg floor platform unit 300 Kg floor platform unit. The weighing units can be connected to a printer. PC-connected weighing units that optimise formula weight control and automatically provide information for stock control are increasingly used. These systems can also supply important information and make the switch to new formulas (and their cost calculation) smoother. While weighing systems, then, have evidently evolved, the same cannot be said of mill loading systems. Apart from the cited ready-mixed sacks, all other compositions require several stages of manual handling involving the transfer and lifting of bags and containers. In any case, the extreme variety of the compositions rules out automatic loading. 178


A possible solution is the rationalisation of the glazes used in production, consequently lowering the number of components in their formulation, which could lead to semi-automatic silo storage and weighing systems. Today, in virtually all glaze preparation departments, packs are weighed or even counted on the assumption that they are of constant weight. The mill load is prepared and then placed on pallets which are raised to the mill infeed level with elevators or fork-lift trucks. Alternatively, the contents of the various packs are poured into a ground level container (with a discharge valve) which is then raised to the level of the mill infeed port and emptied into it. Weighing operations must, of course, take into account both moisture content and the weight of packaging. 4) Observations on glaze grinding From an energy viewpoint, grinding is the most costly task in the whole department, with mills accounting for some 10% of plant electricity consumption. This explains why grinding is generally carried out at night using cheaper rates. With glazes, optimisation of grinding is near impossible because compositions vary so much. Nevertheless, much attention can and should be paid to the means that are at the manufacturers disposal: water and additives. These are usually already included in the formulation of the glaze and include CMC-based glues, various sorts of fluidiser (STPP, hexametaphosphate, polyacrylates, etc.) and suspension agents (generally NaCl). Water also plays a key role: if it is drawn from the mains ionic exchange can, where necessary, be performed; if, instead, it is drawn from a well it must be treated according to its inherent impurities: sand or active carbon filtration, hydrocyclones and/or ionic exchange may be required. In any case it is good practice to pass all glaze water through a 63 micron sieve. Using recycled water for glaze preparation or dilution is highly inadvisable: even where this water appears innocuous it will still be characterised by very high conductivity and a high content of rheologically active soluble salts with surfaceactive agent characteristics. Use of additives must be limited to the indispensable, because anything extraneous to the formula itself generally creates problems during downstream stages of the production process. The grinding media load must be kept constant through the regular (weekly or every charge) introduction of balls equal to the medium-to-large sizes of the original charge. Checks are carried out at least once a month by measuring the distance between the surface of the load and the mill hatch. Should the presence of contaminants be suspected (these usually show up on the tiles as coloured specks), the mill must be emptied and the grinding media checked. The latter can conceal a whole range of foreign bodies: coarse frit residue, aluminium (skeets or scoops), electrical wiring, various types of iron, plastic, wood, etc. Checking the grinding media load, by emptying the mill, should be carried out 179


methodically at least once every six-months. It is best not to remove the smaller grinding elements via sieving at this time as this would disturb equilibrium, alter grinding times and modify particle size distribution. Broadly speaking, grinding times are regulated by setting mill running time or revolutions and depend on the type of glaze being ground. One of the most delicate stages of grinding is mill start-up, when the transmission belts are subject to maximum stress. Over the first 30 minutes this stress may actually increase and if the belts are not properly tensioned, they can slip and wear very fast. They can even overheat and catch fire, a scenario to be avoided at all costs as the consequences are extremely costly. Hence mills must be started under the control of qualified personnel. Anti-slip products to be sprayed on the belts at the time of mill start are best avoided as they actually damage the belts. Above all, when grinding coloured glazes, it is good practice to stop the mill and clean any coarse residue from the fissure between hatch and lining as this traps unground particles that can act as a contaminant. 5) Unloading the mill Once grinding is over the mill is unloaded. Because this task is potentially dangerous certain precautions need to be taken: friction between the grinding media and the glaze generates heat and temperatures as high as 35-50 °C may be reached, thus raising in-mill pressure considerably. If the lid is removed without due caution the pressure could eject it with considerable force. The mill operator must, therefore, first vent the pressure via the bleed valve or vent the lid a little at a time by holding it in place with the clamps. The hatch is then replaced with a discharge port having a stainless steel outlet valve and a coarse mesh to stop the grinding media being carried away by the outflow. Once the mill has been rotated so that the discharge port is at the lowermost point and the venting valve on the opposite side has been opened, the mill can be emptied. Following a quick residue check, the glaze is unloaded while still hot (i.e. under optimum sieving and iron removal conditions). It can be passed through the finest sieve possible (only if sieving time does not occupy the mill for too long) and then pumped into holding tanks where it is checked again. These tasks can made easier by using a mobile sieve and pump unit. No attempt should be made to accelerate gravity-assisted drainage of the glaze by pumping air into the mill: use of compressed air is strictly forbidden as mills are not pressure-tested vessels. Once all the glaze has drained the mill is washed. This delicate task has two important purposes: recovery of the glaze captured by the grinding media and preparation of the mill for the next load without any risk of contamination. Achieving the former requires one or two washings with just limited quantities of water. The first washing (about 100 litres of wash water) still contains a high percentage of solids and can usually be added to the emptied glaze directly. The second washing (700-800 litres of wash water), has a solids content of just 1-1.5% 180


(the first washing will already have recovered some 99% of the trapped glaze). More thorough washing is only necessary following the grinding of very strong colours. With monoporosa and double fire glazes the use of only moderate quantities of wash water is essential as these glazes must be applied at high density. The glaze is then pumped into a holding tank and checked. 6) Feeding the glaze to the glazing line The method will depend on the size of the glaze application department and its location with respect to the glaze production unit. Quite often the glaze is transferred from the main tank into smaller ones that are then transported to the glazing department by fork-lift truck and positioned at the point of use. This practice is always used for flashing applications and intensely coloured glazes or where the glaze preparation department is large enough to allow storage of a suitable number of containers. However, it is an inconvenient system in that there is a continual and dangerous movement of trucks through often narrow passageways; bear in mind that the containers are bulky and require manpower. Moreover, it involves the laborious, timeconsuming task of container washing. As an alternative glazes can be pumped from the preparation department into intermediate storage tanks in the application area. When, for logistical reasons, pumping is utilised, the system must be set up with care to prevent leakage, an all too common problem. Pipes should have a diameter of at least 2 or 3 inches (depending on length) and should, at least along their initial tract, be made of stainless steel. Each pipe can feed a number of tanks (usually 3): it should not have any ball or throttle valves, but simply an initial gate valve and a pneumatic-type external valve, fitted on the flex hose section at the end. Piping may cover considerable distances, (100 metres or more), with each 100 m tract containing some 250-280 Kg of dry glaze. Following an initial, vertical tract, pipes must follow as steep a gradient as possible down to the discharge point within the limitations of building height. Washing the pipes is not easy: pumping controlled quantities of water behind the glaze is inadvisable because, given the considerable distances, the water tends to mix and layer with the glaze, thus making it difficult to decide when the flow needs to be diverted into the drains or sedimentation tanks. 7) Glaze storage All glazes especially alkaline glazes are subject to hydrolysis. The migration of soluble salts from the frits may cause serious defects on the finished tile. This sometimes occurs on monoporosa and fast double firing products, yet can also generate various degrees of pin-holing on other glazes. Glues also deteriorate rapidly, losing their characteristics and thus causing serious application difficulties. 181


For many glazes, then, high temperatures (during summer) and lengthy storage times are the ceramists feared enemies. Hence it is best not to rely too much on leftover glazes that have been in stock for a long time just because they were all right before. Overstocking certain glazes is therefore inadvisable and manufacturers must always make sure that glazes are used in the order they were produced. Recovery of altered glazes is, in monoporosa and fast double firing plants, extremely difficult. The altered glaze can usually be mixed with fresh-prepared glaze to reduce soluble salts. Preparing screen printing pastes Recent years have seen the widespread adoption of screen printing techniques. First used with glossy single fire glazes, screen printing soon spread to rustic products, its popularity being aided by the rapid evolution of graphics. Screen printing and, for some years now, flexography, provide excellent repeatability and let manufactures renew their product ranges with relative ease. Over time screen printing has come to be characterised by a constant reduction in applied weights: before such improvements the most commonly used floor tile screens fell in the 30-50 wires/cm range and wall tile screens were generally 50 wires/cm. This meant high glaze consumption and the appearance of black core problems on single fire products caused by the presence of organic vehicles. This in turn, meant slower firing cycles and poor decoration definition (especially with graduated shades). Serigraphic paste preparation techniques were very approximate. The ground base was generally purchased from colour suppliers and then, once mixed in with the vehicle and the necessary additives, passed through a grinder. From here it was, at best, screened in a rotating scraper sieve at 5-6,000 mesh/cm2 (75-80 microns) before being deemed ready for use. More sophisticated serigraphic techniques and materials will be discussed in greater depth further on. The glazing department The glazing department is usually situated between the pressed tile drier outlet and the kiln feeders (or, in double firing, between the two fires, possibly being preceded after the first fire and followed before the second fire by storage systems). Proper spacing apart of the various application stations along the glazing line is essential. The floor, which must have a non-slip surface, must slope towards drainage channels so that glaze spills are collected. Lighting must be adequate. The department must also be equipped with shelving for storage of the seri182


graphic screens, the visual check panels (one per line) on which production samples and standard tiles are displayed, and an area for small spare parts storage and minor maintenance work. In a properly managed glazing department, the fast flow of defect information is essential. Defects do not necessarily show up on selected test samples, hence monitoring at the kiln outlet is also required as is continuous statistical tracking of such defects. Other important elements in a glazing department include: Electrical system The power supply on the glazing line itself should not exceed 50 V (if no differential safety device is used). The lines are supplied from overhead 380 or 60 V cables. Compressed air system Compressed air requirements vary enormously from one glazing line to another, generally ranging from 100 to 300 m3/h per line. This air is used to power spray guns, pneumatic controls (on screen printing systems), fixative application systems (though these can use high pressure airless pumps) and tile surface blowers. The compressed air feed line generally consists of a 1 pipe supplying air at 6-8 Atm. Water supply system There are normally two water circuits in the glazing department, sometimes three. One circuit is generally fed with water from a well or mains pipe and is usually used to dilute the glazes; the second is recycled water for washing purposes. Fume aspiration system Extraction of the fumes that form inside the application units is essential as they contain glaze particles. Proper modulation of shutters and, therefore, extraction performance inside the glazing units (especially disc and spray gun units) not only safeguards worker health, it also shields the tiles themselves from certain application defects: lumps, glaze droplets, condensate. An individual glazing line generally requires a fume aspiration capacity of some 8,000-10,000 m3/h. Data gathering system The glazing line must be equipped with sensors that allow manufacturers to monitor efficiency and other important production control parameters.

183


Speed Conveying speeds are in the order of 20-25 m/min, but may reach 60 m/min along certain tracts (e.g. the gap preceding the box storage system). Glaze application and glazing lines Once prepared, glazes can be applied to dried, single fire tiles (now almost universal) or biscuit (double fire) tiles in a variety of ways. The successful outcome of unfired tile glazing largely depends on dried tile bending strength, as the tile must be strong enough to withstand multiple applications/decorations without breaking. Applying an engobe (a semi-glaze composition) under the glaze is useful in that it limits degassing phenomena and also prevents the glaze coming into contact with contaminants in the body. A uniform film of suitable engobe is sometimes applied to the back of the tiles to stop direct contact with rollers and so reduce the rate at which dirt builds up on them. Generally speaking, a glazing line consists of a belt conveyor: a feeder is installed at one end of the conveyor and a collection unit at the other, with the application stations in between. Tiles are conveyed on trapezoidal cross-section belts, preferred because they are easy to clean, have a limited tile-belt contact area and are versatile. The belts are supported by pulleys mounted on a metal support frame. This conveying system works as a series of interconnected sections each independent from the other in terms of speed (e.g. different speeds may be required during tile preparation, glaze application, decoration operations etc.). In addition to actual application units, conveyor lines may also include auxiliary equipment items such as compenser units (which automatically allow tile collection and release), turntables to rotate the tiles 90°, spacers, side guides, tile cleaning devices (fans, brushes, suction units), scrapers, etc. The main glaze application devices are: a) bell unit b) waterfall unit c) rotating discs d) spray guns e) drop feeders f) dry application units. Bell unit Essentially a convex metal dome. As the glaze runs off its surface it forms a constant, uniform waterfall-like curtain. Glaze is applied to the tiles by passing them through the curtain The glaze, which can be relatively dense and viscous (1950-2000 g/l) tends to deposit in slightly greater quantities on the sides of the tile. 184


The size of the bell on an individual glazing station depends, of course, on the size of the tile being produced (fig. 2). Bell units are the most widely adopted continuous application solution, allowing manufacturers to obtain very smooth surfaces, a prime requisite for certain tile types. The glaze flows out of the apex of the bell, with feed parameters being adjusted to suit the characteristics of the glaze being applied. A circular ring bordering the top of the bell allows the user to apply one or two coats simultaneously. Once set up, the unit generally requires little monitoring but it does involve numerous parameters. One of the most common defects is the formation of bubbles or depressions on the glaze surface. This defect has a number of possible causes: bubbling in glaze transfer and agitation, erroneous fluidification, unsuitable viscosity and density. As the variables in play are so many, there is no easy solution to bell application problems. For example, high glaze density makes sieving difficult. It is now possible to sieve as the glaze returns to its container using large-diameter 4000 mesh vibrosieves. The latter remove the lumps that sometimes detach from the figure-of-eight container or from recycled glaze. Another problem is that bell units are adversely affected even by slight vibration, so the bell support pole must be isolated from the rest of the line. Waterfall unit Consists of a triangular hopper: the glaze flows out through a slot of adjustable aperture, thus creating a waterfall. Tiles are glazed by passing them underneath the waterfall. These units come in different lengths to cover the full range of tile sizes.

Fig. 2. Bell unit (note at right the erroneous flow of glaze highlighted by the hole in the veil).

185


The glaze feed unit is similar to the one used on bell systems: vibrosieve, hopper, feed tube with flow adjuster. Performance is sometimes undermined by waterfall obstruction problems, especially lumps in the glaze and air bubbles. An excessively liquid glaze can also cause problems (irregular stripes): this defect can be corrected by reducing the in-hopper load and narrowing the waterfall slot. Pressurised injection system Glazes, or other slips (even where quite dense), can also be applied by way of a recently developed technique that involves pressurised injection through newly designed waterfall units. This system, called Vela®, makes it possible to apply glaze more homogeneously across the width, yet waterfall-type defects often remain. A special feed system allows several coloured products to be introduced together with the base glaze, thus providing a variety of decorative elements, the shape, position and overlap of which can all be adjusted. Where required, injection can be electronically controlled to ensure good repeatability. Blending the base glaze with the coloured injection products confers considerable creative flexibility. Manufacturers can also apply large quantities of material per unit of surface area per unit of time, allowing them to build up layers of considerable thickness on the tile. Rotating disc system (single or double disc) Applies the glaze by centrifugal force through packs of disks of varying diameter (120-180 mm); they rotate at high speed, about 40-50 cm from the surface of the tile. Characterised by excellent flexibility, this system can apply glazes of greatly differing densities and viscosities as well as those with coarser particles and also provides very good flexibility in terms of applied quantity. The discs are generally arranged as circular packs, fed axially via a holed tube. Drive transmission is provided by belts and pulleys of different diameters that allow adjustment of disc RPM. The latest versions feature inverter control, thus allowing more precise speed and atomisation control. Most packs have a rotation speed of 3000-4000 RPM; while higher speeds are possible, wear rates increase accordingly. Discs, while versatile, are not able to produce a perfectly smooth surface. Moreover, disc application is sometimes marred by defects such as: drops: these form as a result of incorrect glaze rheology, condensation, unsuitable booth configuration, clogged holes etc. lumps: caused by excessively fast drying of the glaze or excessively violent glaze 186


intake; may also derive from a low feed rate, inadequate for the size of the pack, thus allowing the glaze to dry on the discs. stripes: weight too low, irregular disc rotation, erratic feed. Attempts to prevent the formation of drops and lumps have, over the years, led to the construction of booths of every possible shape and size (fig. 3). The discs are fed at a constant flow rate by way of an aperture that is regulated by an overflow control system and an adjuster valve. The glaze is passed through mesh filters on the feed line and through vibrosieves on the return circuit. Drop device Applies rather large drops (approximately 1 cm in diameter) to obtain certain (e.g. rustic) aesthetic effects. Drop applicators may be of the tube or cup type. Tube feeders resemble disc units, but instead of a disc pack there is a 3-4 cm cylindrical holed tube. Cup feeders, instead, feature small, continuously fed crenelated cups, which rotate around a vertical axis, throwing the glaze against the tiles in a way that depends on the characteristics (density and viscosity) of the glaze itself, the shape of the cups, their height, rotation speed and drive speed. Spray gun Allows the application of small quantities of glaze via an atomiser. Flashing effects can be obtained by oscillating the guns (sometimes two differently oriented guns are used in the same booth). Intensity could be varied by way of an inverter on the feed pump.

Fig. 3. Disc application booth and close-up of lamellar disc pack.

187


Where uniform flashing effects are required, constant glaze feed rates are essential. Spray gun glazes, generally used at low densities, require certain rheological properties to prevent the formation of lumps and drops. With spray guns the main inconveniences are deformation of the jet (nozzle dirty or damaged) and droplets which either run off the booth walls onto the tiles or are carried away by the turbulence generated by the injected air. To combat these defects a variety of differently shaped booths have been designed in line with a host of volumetric theories, thus leading to changes in wall angles, positioning of intake ducts etc. Air pressure, which needs to be kept constant, is generally kept in the 2-6 Atm range. High atomisation pressures provide finer droplets but also tend to pressurise the booth, thus creating considerable extraction problems, with the aerosol (harmful to both workers health and the environment in general) exiting the apertures. Dry applicators These are used to apply granular glazes (grains) with a vast range of particle size distributions or various kinds of sands. The grains are re-circulated by way of a pneumatic loop or a sealed-in cup elevator (the latter creates less dust). The grains are loaded into a hopper; this is fitted with an extractor which causes the glaze to fall as a uniform veil onto a roller, belt or distributor sieve (fig. 4). If stripes or steaks defects often stemming from errors at the hopper loading stage are to be avoided, the grain distribution system must work perfectly. Before reaching the grain applicator, specific areas of the tile surface are coated with a glue or wet glaze. As the grains fall onto the tiles they adhere to the glue-coated areas only; excess grains are then blown off the tile surface and returned to the feed circuit. Because they mimic granite almost perfectly and most manufacturers feel that a thick layer of glaze generally provides the product with a certain added value, grain applicators are very popular (1985-2000). Universal grain/powder applicators which provide streaked effects by using rollers or embossed belts are also available. Key units on the glazing line Conceptually, the glazing line consists of a series of individual machines assembled together to form modules which, in turn, make up the line. A description of the main modules follows. Wetting unit: in the traditional version, connected directly to the water mains, or pneumatic version (minimum constant water pressure at least 1 bar). Mainly used on double firing or airless-type single firing lines. A pump feeds up to 3 different appli188


Fig. 4. Dry grain/powder applicator.

cation points. Used on monoporosa and single fire lines where the engobe or glaze is applied with a bell. Glaze scraper module: this module is inserted some 12-15 metres after the last bell application, thus allowing the glaze to dry. The module consists of 2 glaze scraper stations (which scrape parallel sides of the tile), 2 turntables and 1 fan. Manufacturers use different glaze scraper types for single and double firing operations. With the former, any flashing is cleaned off the sides with felt discs so as not to wet the edges, while the latter employs water-bath rubber discs. Screen printing modules: standard screen printing modules are 6-9 metres long and consist of a fixative spray gun (traditional or airless), a compenser and the screen printing unit. The module may incorporate several drive units and usually has two air feed points, one water feed point and a suction intake. Tile back engobe applicator: applies engobe to the back of the tile; fitted on a module about 3 m long. Recommended for use with all production technologies as it is an effective way of protecting the kiln rollers. 189


Other units Rustic effect brushing unit: mostly used on lines producing single fire and porcelain tiles with structured or relief surfaces. The device consists of several abrasive discs or brushes generally positioned at least 6 m downstream from a spray gun unit. Bark effect discs: these may be modified traditional discs (the disc pack is replaced with a holed tube) or special booths designed to confer a granito look. Grain or powder applicators: generally positioned at the end of a line, 3 m after a single or double disc unit. In this case the disc applies the glue that attaches the grains to the surface of the tile. The glazing line Line composition varies as a function of production technology. The tables below give a general idea as to how module compositions change accordingly. SINGLE FIRE LINES approx. length 80 – 100 m

glaze applicators 1 water spray 3 double discs 1 spray gun

optional 1 bell 1 single disc 1 grain applicator

decoration applicators 2 serigraphic units

optional 1 roller unit 1 serigraphic unit

MONOPOROSA LINES approx. length 80 – 100 m

glaze applicators 1 water spray 1 double disc 2 bells 1 spray gun

optional 1 single disc 1 dry applicator



DOUBLE FIRE LINES approx. length 70 – 90 m

glaze applicators

optional

1 airless water spray 1 single disc 1 dry applicator 1 double disc 2 bells 1 spray gun

190




PORCELAIN FIRE LINES approx. length 60 – 80 m

glaze applicators

optional

2 double discs

decoration applicators 2 flat serigraphic units 1 application

optional 1 drier*

* where soluble salts are produced.

Printing machines Screen printing is the most common tile decoration system used by the ceramic industry. Its popularity stems from the fact that it allows manufacturers to obtain a large number of identical pieces at high output rates. Generally involving the use of a fabric screen, the system works by squeezing the ink through the mesh of the open part of the fabric (which forms the pattern) by means of a scraper blade, thus reproducing the same pattern on the tile surface. It is a technique that can be used both on fired tiles and dried pressed tiles (where they have sufficient bending strength), above or below the glaze. A silk screen printing station must have the following characteristics: high piece/minute output to match the output capacity of the latest generation of presses versatility with regards to pattern/decoration types low running costs limited manpower requirements faithful image reproduction. The most common printer types are: Flat (figs. 5, 5a) Rotary (fig. 6) Roller (fig. 6a) The scraper blade (squeegee) This is the blade that pushes the ink through the mesh of the fabric and brings the latter into direct contact with the tile surface (except where the contact technique is used). The scraper blade (or squeegee) is a flexible resin strip attached to a rigid support; the profile of its edge (i.e. the part that touches the fabric) depends on the printing technique.

191


As it is subject to wear, which might ultimately lead to faults in the pattern (stripes, colour changes), the blade edge must be kept sharp at all times. The main printing parameters are: imprinting speed scraper pressure angle of imprint a (angle between scraper and screen). Imprint angles, which depend on scraper flexibility, the pressure exerted and the profile of the scraper edge, have an important influence on the quantity of applied ink. In particular, the lower the angle of imprint (soft, round-edged squeegees) the more ink is deposited. Flat screen printing On flat screen machines the scraper can move both parallel and at right angles to the glazing line during printing and pushes down hard on the screen as it not only has to squeeze the colour through the fabric but also bring the latter into contact with the tile surface (no-contact screen printing, see fig. 5). Rotary printing machines Rotary printing has provided an effective response to the limitations posed by flat screen printing. The principle uses patterns on a screen wrapped around to form a cylinder which rotates over the tile passing beneath (fig. 6). This system provides constant print speeds; thanks to an electronic sensor, which signals the arrival of the tile and its position, screen speeds are varied to make sure the start of the pattern coincides precisely with the leading edge of the tile. Compared to flat screen printing, the main advantage of the rotary method is increased output capacity, mainly because the tiles do not stop but move through continuously. Another key feature of this machine is that the decorative pattern is spread over the entire circumference of the cylinder/has more extensive distribution along the cylinder circumference. Increasing the number of designs on the same cylinder increases the possibilities for product diversification. Thats why, in so-called random printing, rotary printers provide a greater number of graphic combinations compared to flat screen machines. Working conditions being equal, rotary screens also last longer.

192


Fig. 5. Flat screen printing technique: 1) scraper blade, 2) ink, 3) screen, 4) frame, 5) tile.

Fig. 5a. Flat screen printer.

193


T) conveyor belts L) line belts FT) photocell.

Fig. 6. Rotary screen printing: A) feeder, D) imprint roller, P) tile.

Fig. 6a. Roller printing machine.

Rotary screen manufacture can be carried out in two different ways: 1) The traditional way is to stick the film and impress it on the cylinder using lasers, which align and sub-divide the pattern and define the exposure times; this allows the screens to be impressed as they rotate, allowing for greater precision. 2) A more innovative solution impresses the fabric as it lays flat; the fabric is then 194


attached to the cylinder. This technique can be used with all rotary printers and combines the excellent print definition typical of flat screens with the versatility and durability of rotary systems. A specially designed wrap-around unit allows screens to be fitted in the ceramic production plant, thus saving on storage space, transport and disposal costs. Since its introduction, rotary screen printing has evolved steadily and now allows manufacturers to make prints that are as true as possible to the originals. Roller decoration Here, the images to be reproduced are engraved into a polymer or silicon cylinder with a laser; faithful reproduction of the original image is provided by digital processing of the laser-engraved image, giving fine resolution, excellent uniformity of colour shade, high quality definition and constant repeatability over time. Repetitive patterns, shaded effects and infinite reproduction of the original design can all be obtained. Development of standard laser engraving processes has led to the creation of the High Definition (HD) system, a new technique that allows an image to be reproduced on a tile without a base screen or tilts as it is reconstructed point by point. The absence of a screen eliminates all colour overlap problems and allows the original image to be reproduced in the finest detail. In fact, the main advantage of the HD system is that detail, instead of being lost, is enhanced; moreover, softness of colour gradation makes this type of incision especially suitable for marble-like effects, the detailed nuances of natural stone and geometric patterns. With rotary roller systems, the tile is decorated as it comes into direct contact with the surface of the roller; they both move at exactly the same speed so there is no need to stop the tile, thus speeding up the decoration process and reducing the corner damage, cracking and other defects often associated with start-stop systems. It is, then, a decoration system in which movement of all components cylinder, colour, pressed tile is synchronised. Silicon roller engraving creates cavities that contain the colour to be applied to the tile; at every turn of the roller, even before it touches the tile, a scraper removes any excess printing paste. Two mechanisms release the colour from the cavity: the first is assisted gravity whereby the low viscosity colour is transferred to the surface of the tile quickly by a combination of gravity and the centrifugal action of the rotating roller; the second, much more important, mechanism depends on the fact that the colour, when it comes into contact with the tile, is absorbed by it and thus sticks. Another key advantage of this system is roller durability, itself influenced by a number of factors such as scraper pressure (silicon abrasion) and the nature of the glaze (e.g. coarse glazes increase friction between cylinder and tile), especially its 195


moisture content. A still-moist glaze can provide efficient lubrication for the silicon, thus extending the working life of the cylinder. Screen mesh types Whether flat screen or rotary systems are used, it is the mesh that provides the support for the image to be printed and the mesh which ultimately determines the level of definition. To be suitable for high quality printing, a mesh must have the following characteristics: a) uniform weave. b) high abrasion/wear resistance to withstand the mechanical stresses of the printing process. c) high tensile strength to withstand the stresses of print frame construction. d) a fabric appropriate to the type of pattern/decoration and the quality of the tile being produced (this involves such factors as mesh density, mesh aperture and mesh thickness). e) resistance to the chemical products used in printing (e.g. screen printing pastes, water, degreasing agents). f) elasticity so that the screen reverts to its original position after being stretched by the scraper blade. g) resistance to pressure and impact. h) the mesh threads must swell as little as possible in order to prevent changes in aperture and thus paste flow-through parameters. i) it should adhere well to all photosensitive emulsions. The range of meshes the instruments by which passage and deposition quantity of colour are controlled is extensive, giving the user a vast choice in terms of fabric, fibre and mesh density. The most important factor characterising fabric type is mesh density (the number of threads per unit of linear measure, usually per centimetre). It should always be born in mind that colour deposit and consumption are influenced by mesh density directly as the colour has to pass through the mesh. As the mesh is the only support for the image, it must be of a density able to reproduce even the finest details. Thread diameter is also an important factor in this regard. Another factor to be taken into consideration when choosing the fabric is mesh aperture: this becomes crucial when using pastes with very coarse particles. Fabric type is determined by thickness, a value that is a combination of mesh density, thread diameter and on-frame fabric assembly. Theoretical parameters regarding the maximum amount of colour (expressed in cubic centimetres per square metre) left on the tile after the passage of the scraper blade, are referred to as theoretical volume of colour. This varies from fabric to fabric and depends on mesh aperture and fabric thickness. 196


Colour consumption/deposit is also influenced by other factors: on-frame fabric assembly screen manufacture and design technology hardness, profile, pressure and angle of scraper blade imprint speed fabric tension printing pressure in relation to machine type (flat or rotary) number of emulsion applications.

Fabric types The following section takes a look at the fabrics most commonly used by screen manufacturers, products with synthetic fibres capable of withstanding the mechanical, physical and chemical strains of printing (fig. 7). a) Polyester fabric This fibre has good tensile strength and, consequently, has limited thread stretch, thus making it ideal for high definition printing, where alignment accuracy is crucial. It has excellent heat resistance and limited sensitivity to sharp changes in temperature and humidity. Polyester also has very good resistance to abrasion, solvents, mineral/organic acids and is insensitive to light. Monofilament (single wire) polyesters provide excellent colour passage performance thanks to the smoothness of the threads, hence screens last a long time. It should also be pointed out that specially developed technical polyesters also exist; these innovative products offer: homogeneous results that remain constant over time a longer screen life thanks to improved emulsion adhesion excellent uniformity and smoothness of the applied emulsion optimum image resolution and definition reduced work times as the fabric is self-lubricating, reducing screen cleaning requirements considerably. b) Nylon fabric (polyamide) This highly elastic fibre provides excellent mechanical resistance, although this does depend on the degree of extension and polymerisation. Another feature of

Fig. 7. Various fabric types (polyester, nylon and metallic wire).

197


nylon is the fact that its tensile strength and stretch diminish as temperature increases. Highly resistant to abrasion, it is nevertheless, a material that absorbs moisture easily, causing it to swell. On account of their excellent elasticity, monofilament nylon meshes are especially suitable for printing on irregular surfaces. c) Metallic or metallized fabric Metallized high precision fibre is made by coating monofilament polyester with metal (usually nickel). Dimensional stability makes it especially useful where high registration accuracy is required; it differs from stainless steel meshes on account of its elasticity. Surface conductivity prevents any electrostatic build-ups and the metallized surface is also highly resistant to abrasion. Consequently, it is ideal for use with the abrasive colours used in ceramics. Since the colour passes through the mesh smoothly, it is possible to obtain prints with extremely well defined edges. The steel fabric is now used rarely because it has been replaced by new nylon types. The frame The frame provides the necessary support for both flat and rotary printing screens and has a direct influence on screen quality (i.e. uniformity and constancy of tension). As it is subject to continuously changing chemical and mechanical forces, it is best made of stainless alloy so rust does not cause catalytic damage to the mesh. Two types are available: a) steel, which offers the advantages of high specific weight, durability (where properly protected), relatively low costs and good mechanical strength; on the downside steel frames are difficult to handle where large and, unless coated with paint or varnish, may rust. b) aluminium, which has the advantage of being light, easy to handle, rust-resistant and highly suitable for small and medium size frames; however, aluminium offers only limited resistance to acids and alkalis, and is somewhat less appropriate for large sizes in that it is not always rigid enough to withstand the tension of the mesh itself. Aluminium frames are also relatively expensive. Tensioning This is one of the most important tasks in the preparation of a printing screen frame, as it has a direct influence on final results. Only a correctly tensioned screen will provide the desired outcome, namely: good mechanical strength optimum adhesion of the emulsion good image colour. 198


Tensioning may involve one of the following systems: pneumatic: clamps stretch the mesh to the required tension, control being effected with an instrument that measures pressure (in Atm). mechanical: simple and manual, or more sophisticated, with an electric motor coupled to an automatic programming system. Uses different lock systems such as pins, bars and pincers, which hold the screen in place and cover its perimeter. Good printing results depend on uniform tensioning of the screen right across the entire area enclosed by the frame. Workers can measure screen tension during frame preparation via the pressure gauge on the pneumatic pincer tensioner: the pincers are worked by a pressure regulator and the gauge itself. Note that mesh producers supply fabrics complete with tensioning data tables: this information is used to calibrate the tensioning machines accordingly. Tensioning measurements are expressed in atmospheres (bar). More commonly, a tensometer is used. This is a small electronic or mechanical instrument equipped with a blade that comes into contact with the mesh and measures its tension (in Newton/cm). Improper tensioning can lead to defects such as lack of multi-colour printing alignment precision, shifting about of the screen, a stretched out image, premature mesh wear and heterogeneous colours etc. Emulsion The emulsion (or gelatine) is a key component in the screen printing process. It is spread over the tensioned screen with the aid of a scraper blade and the design is impressed on it using a photoengraving procedure. Note that until the mid-80s meshes for the ceramic industry were made with a chrome emulsion; this cost less but offered little resistance to the strains of the printing process and was a source of serious water pollution. Since then, as part of an effort to reduce environmental impact and meet the growing demand for innovation, a diazo gelatine (a photosensitising agent) has been introduced into the production cycle and adapted to the special printing needs of the ceramic industry. Still used for the production of printing screens, this article offers good photosensitised product duration, high definition and is much more environmentally friendly. Technological and graphic developments in the ceramic sector, greater openness to new applications and the need to respond to ever-greater demand for high quality have all led to the development of new emulsions with good printing resistance and high definition. These highly light-sensitive photopolymers provide fast incision, excellent definition and good resistance to the glazes. 199


Photoengraving Once the emulsion has been applied the frame-mesh assembly is left to dry before being photoengraved. Exposure has a direct influence on definition, resolution of the finest details, thickness and the chemical/mechanical resistance of the pattern produced in the emulsion. Correct exposure depends on two important factors: exposure time, essential in conferring quality and resistance intensity and quality of the light. Keeping the light at the right distance during the photoengraving process results in homogeneous distribution over the entire surface area. However, if the light source is too close the light tends to be concentrated in the middle, with diminished intensity at the sides. Likewise, a light source that is too far away causes underexposure and areas with poor mechanical/solvent resistance. Results also depend on photograph image density (which must maintain the finest details), the thread density of the screen fabric and the characteristics of the photosensitive emulsion (on the basis of which the exact UV exposure time needed for complete emulsion penetration is calculated). The most appropriate exposure time can, however, be found by effecting a series of positive test exposures, increasing or decreasing exposure times until results are optimum. Another factor influencing exact reproduction of the design on the screen printing frame is the subsequent development, which influences mesh aperture and thus the passage of colour onto the tile. A catalysing stage follows to reinforce the fabric. Silk screen printing inks These are pastes or liquids of varying viscosity, made up of solid particles dispersed in a fluid vehicle and any complementary substances that may be required. Easily transferred onto the tile, they are deposited in layers about one tenth of a millimetre thick. The solid part consists of colorants that change into glass during firing. Essentially crystalline or fluxing blends with oxidants and pigments, they are prepared via a wet grinding/drying process. The liquid part, which acts as a vehicle for the colorants, generally consists of products such as glycol and polyethylene glycol, characterised by extensive tile covering capacity. Vehicles must have good binding, plastifying and stabilisation qualities; those vehicles with better adhesive performance, especially suitable for use with single fire products, can be prepared with carboxymethyl cellulose in solution, to which polyvinyl alcohol, glycols or polyglycols are added. 200


The solid base is generally prepared by way of a wet grinding method similar to that used with glazes; it is then dried gradually (using an optimised thermal cycle), powdered and stored. Thus obtained, the powder is then mixed and homogenised with the liquid vehicle. Mixing usually takes place in an agitator, with a certain powder weight being added to a pre-determined quantity of vehicle. The paste is then put through final refinement, a procedure that may involve a series of machines (sieving units, cylindrical refiners or roll mills, grinding refiners or micro-ball mills). Mixing the paste Once each of its components has been accurately weighed, the paste must be homogenised before it can be put through the refinement procedure. Mixing takes place in simple cylindrical containers equipped with slow-stir paddles and outflow valves on the bottom. Examples of various screen printing paste compositions used with different production technologies are given in the table below.

Base frit Frit % % Kaolin Colouring pigments Vehicle: base (media)

Double fire Monoporosa Single fire Porcelain tile paste paste paste paste Transparent Transparent Transparent Transparent ~ 1030 °C ~ 1120 °C ~ 1180 °C ~ 1200 °C 95 95 95 95 5 5 5 5 0–50 out of 0–50 out of 0–50 out of 0–50 out of 100 100 100 100 20-100 out of 20-100 out of 20-100 out of 20-100 out of 100 100 100 100

Refinement Performed with the aid of several different machines. Refined pastes must have the following characteristics:

Residue % at 45 mm Density g/l Ford Viscosity at Ø 4 mm Ford

Flat 0 - 0.2 1500 - 1700 > 150”

201

Rotary 0 - 0.2 1500 - 1700 > 150”

Roller 0 - 0.2 1500 - 1600 25 - 30”


Refinement of the screen printing paste (or ink), once carried out in simple grinding units which did not always provide efficient dispersion, is now carried out in more sophisticated machines such as micro-ball mills. The latter are made up of: a cylindrical tank with a sealed lid. a grinding chamber lined with hard alumina containing 5 mm grinding balls. a sealed motor, the shaft of which has two rotors positioned at different heights in the tank; the bottom of the grinding chamber acts as a centrifuge. The roughly mixed screen paste is introduced into the tank and from here is aspirated into the grinding chamber via a circular aperture. Most grinding takes place in the peripheral zone, where grinding energy is maximised by the centrifugal forces acting on the balls and the paste. Driven by the centrifugal force, the paste is circulated through helical fissures on the bottom of the tank. The ceramic quality of the ink is checked by way of sampling procedures which involve both individual raw materials (frits, colorants) and the paste: the latter generally requires adjustment to maintain desired chromatic effects. Screen printing vehicles (medias) The vehicle acts as an ink solvent, transforming the powdered colours into a fluid suspension. From a ceramic viewpoint, such substances are of extremely limited importance in that they do not actually influence the final product. Yet from a printing standpoint they play a key role in establishing the characteristics of the ink and ensuring good image reproduction. The most commonly used vehicles are glycols and polyglycols with an ethylene oxide, propylene and polymer base. These commercially available chemical composites have high powder wetting capacity, are optimum binders and disperse the colour perfectly, creating a mass of good flowability that sticks to tile surfaces well. Vehicles are mixed with dry powders in ratios that depend on the specific application technology. Optimum print conditions must be tested in a production context. The dry base/vehicle ratio can be varied within the following intervals: flat or rotary screen printing: base 20-100, vehicle 100. roller screen printing: base 80-120, vehicle 100.

202


SCREEN PRINTING DECORATION TECHNIQUES On-glaze decoration The most important factor influencing this type of decoration is the moisture content of the glazed surface on which the image is to be printed; if it is excessive it may cause stickiness between glaze and screen, resulting in clogging of the mesh. If, instead, the glaze is too dry, it diminishes cohesion between the glaze and the underlying layers, causing the former to dust the screen (i.e. stick to it); fixatives normally provide an easy solution to this problem. Glazed surface moisture content largely depends on the time that elapses between glaze application and printing (line length and conveying speed), the nature of the glaze (composition and particle size distribution), the porosity of the tile and the presence of under-glaze decorations. Drying times are best modified by adjusting line speed (maximum speed is defined as the speed at which tile breakages begin to occur) and glaze composition: to adjust the latter the percentage of plastic substances (kaolin, clay) is usually altered, or additives are used. Drying times can also be adjusted by making slight changes to glaze grinding parameters, taking care not to alter post-firing properties. In any case, this approach is much less effective than the first two. In addition to the moisture content of the glazed surface, another factor which influences on-glaze decoration is scraper blade pressure; if it is too low decoration may be imprecise or incomplete; if excessive the first colour may stick to the screen of the second application; if inconsistent it may lead to deposits of varying thickness that result in non-uniform colour shades. Identical defects may have a number of different causes; for example, a colour may stick to the underside of the next screen if application stations are too close together or if the first ink has an excessively long drying time. Imprecise/incomplete decoration may be caused by faulty screens, an excessively wide screen-tile gap or improperly levelled tile guides. Variations in decoration thickness are caused by differences in squeegee speed, tiles of non-uniform thickness, worn screens and variations in the composition and/ or viscosity of the ink. Other possible defects include the differences associated with use of colour reagents (caused by absence of parallelism between the screen planes and the scraper blade) and streaks or stripes (blunt scraper blade or worn screen). Under-glaze decoration This type of decoration is applied directly to the tile, which must be cleaned and sprayed with a thin film of water beforehand so as to even up wettability and water absorption. 203


It is also good practice to apply a preparative undercoat to provide a base for the screen printing ink. In addition to the above-described defects (encountered with on-glaze decoration) small dimples, holes or cracks may appear at decoration points, especially where decoration layers are very thin. These problems may be caused by excessively large grains (inefficient powder grinding) or very slow drying which causes the ink to be repelled. Colour gradation Graduated colouring is reproduced on the screen by creating a set of points of various shape, size and arrangement. Graduated decorations, whether printed on top of the glaze or directly on the tile, require particular attentiveness in the choice and preparation of materials (especially the screen printing glaze), the type of scraper blade and above all, the type of mesh. The glaze should be fine-ground to residues of about 0% at 16,000 mesh/cm2, while the scraper blade should be sharp edged so as to obtain maximum image resolution with minimum wear: the blade should also be fairly hard. The mesh which must have a highly regular weave should be selected on the basis of tests aimed at optimising the results required in final production. Generally speaking, the best fabric for this sort of application is polyester with a density of 100-120 threads/cm. Note that this type of decoration, when printed directly on the unglazed tile surface, highlights defects such as a scratched surface; hence it is good practice to equip the press with mirror dies. Plain tiles When a large number of tiles are to be printed all the same colour, the surface on which the colour is to be applied must be perfectly uniform, clean and free from even the slightest defects. If printing on top of the glaze it is essential that the latter does not transfer dust onto the screen. A dust-contaminated tile surface can cause colour variations, shadows and/or pin-holing. Once again, polyester screens are to be preferred on account of their dimensional stability over time (density should not exceed 77 threads/cm). Silk screen printing pastes should have good flow properties so that the spaces corresponding to thread cross-overs are filled in.

204


Colour overlap Overlapping two colours will evidently produce a third one. Screen printing allows manufacturers to take advantage of this simple concept and thus generate a number of colours far greater than the number of actual application stations. In general, decorations of diminishing thickness in any case very thin (60-70 threads/cm) are applied so as to aid drying and allow the use of soft scraper blades; fluxing colours are sometimes used on top of refractory colours. When printing directly on the surface of the tile it is possible to overlap two colours and insert a layer of transparent glaze (crystalline) between them.

205


206


Chapter VI FIRING

General Firing is, perhaps, the most important stage of the whole production process as it actually creates the ceramic material by transforming the raw materials in the body into new crystalline and vitreous compositions that confer key properties: insolubility and solidity to maintain shape, bending strength, porosity or impermeability, resistance to chemicals etc. Firing also causes the vitreous coating to melt and form a continuous layer that is solidly anchored to and interlocked with the tile body, thus providing the required chemical, physical and aesthetic characteristics. Firing consists of heating, and thus transferring energy, to the pressed, dried tile, up to a fixed temperature for a set time, so that chemical and physical transformations cause the body and glaze to acquire the properties required of the final product. In the past the target temperature, the firing time and the firing methods were decided on the basis of experience; nowadays the same parameters are established on the basis of chemical knowledge, an understanding of the technical behaviour of the relevant raw materials and preliminary analyses of the latter, especially thermal analyses (see Volume 1). The equipment available to the modern producer allows him to control manufacturing conditions ever-more precisely and effect the finely-tuned firing cycles that are indispensable for standardised mass production. The transformations that take place during firing Firing takes place by introducing heat into the kiln interior and thus into the ceramic products themselves. Heating increases the vibration amplitude of the atoms in the material, thus causing it to expand. The extent of this expansion depends on the chemical nature of the material, its crystalline or vitreous structure, their relative quantities and the transformations that take place during heating; it has been demonstrated that crystalline compositions expand more than vitreous ones, just as compact structures expand more than porous ones. As firing progresses, certain compounds in the materials disappear and new ones are created; these changes show up as expansion or shrinkage. The exact nature of such phenomena depends on the initial composition of the body, the transformations that occur, and in-kiln temperature; this explains why 207


Expansion (‰)

Expansion (‰)

bodies can behave so differently during the first and second firings even where the achieved final temperature is the same: the thermal expansion generated during a first firing is irreversible while that of a body which has already been fired and heated is reversible (fig. 1). Another physical transformation that takes place during firing is the melting of the fluxes in the body and glaze compositions. In the ceramic industry eutectic mixes are melted to allow the creation of a liquid phase at temperatures lower than those actually needed to melt the individual materials; it has been shown that the more numerous and complex the eutectics among the oxides supplied by the raw materials, the more easily the composition melts. When melting takes place, it counteracts the previously described expansion, and causes a series of closely correlated transformations: reduced porosity à increased density à solid-solid and solid-liquid reactions à improved technological properties. As temperature rises so the quantity of molten material increases steadily, simultaneously resulting in a reduction of system viscosity. Gradual softening of the tiles and a series of events associated with the formation of the liquid phase follow: infiltration of the liquid into the cavities of the mass, solution of the granules, diffusion of dissolved material throughout the rest of the liquid phase, crystallization of the solute in equilibrium with the solvent once saturation point has been reached. Simultaneously, there is an increase in reactions caused by the reduced viscosity of the piece, making control of the material vitrification curve (shrinkage and absorption of water as a function of temperature) more difficult and possibly leading

Fig. 1. Unfired and fired tile expansion curves.

208

Firing

to curvature defects (e.g. pyroplastic deformations) caused by the imperfectly leveled support surface of the tile as it advances on the moving rollers. With glazes the composition is richer in fluxing agents and nearly all the material is transformed into a liquid of a certain viscosity: some substances, added for the purpose of creating certain effects such as glaze opacity, or pigments intended to give colour, may remain undissolved. Infiltration, dissolution, diffusion and crystallization also occur in the glaze-body contact zone and are responsible for the creation of an intermediate layer which anchors the glaze to the body. The depth of this layer depends on the refractoriness of the body and its permeability (fig. 2). As firing progresses and kiln temperatures increase, the material passes through a number of critical thermal zones, generally defined by the chemical reactions that take place within them: up to just over 100 °C: hygroscopic water is eliminated (this is the moisture content left behind after imperfect drying or water reabsorbed during glazing or from the environment). up to 200 °C: the zeolite or crystallization water, the molecules of which are bound by absorption into the crystalline structures, are eliminated. between 350 °C and 850 °C: combustion of organic substances in the clays and oxidative dissociation of the mineral sulphides (e.g. Pyrite FeS2) with freeing of sulphur dioxide. between 450 °C and 650 °C: elimination of structural water and consequent destruction of the clayey crystalline lattice. at 573 °C: allotrope transformation of quartz α to quartz β, causing a sharp increase in volume. between 800 °C and 950 °C: decarbonation of limestone and dolomite with freeing of CO2. above 900 °C: formation of new crystalline phases made up of SiO2, silicates and silica-aluminate complexes.

Fig. 2. Intermediate layer anchoring the glaze to the tile body.

209


above approximately 900 °C: thermal dissociation of other salts such as sulphates and fluorides. if temperatures exceed 1000 °C some body and glaze components such as alkaline oxides, lead oxide, zinc oxide and boric oxide may evaporate. During cooling the molten materials provide both body and glaze with cohesion and solidity. This consolidation may depending on the components in the molten material and the cooling parameters lead to the formation of vitreous and/or crystalline structures. In ceramics, the two circumstances usually co-exist in that the molten material is made up of several components. The ions of certain molten materials, owing to the loss of solubility caused by the drop in temperature, settle according to the geometry of their crystalline structure, while the rest of the liquid solidifies to form a vitreous phase in that cooling times, while lengthy, are not long enough to allow its complete crystallisation. With glazes the molten glass becomes progressively more viscous and takes on a paste-like aspect; simultaneously, it contracts. Yet because it has attacked and penetrated the surface porosities of the tile it remains securely anchored to it. Initially the glaze shrinks but, as it is highly viscous it adapts to the tensions that arise between it and the tile; once the vitreous transition point has been passed, the glass locks and if its shrinkage subsequently fails to match that of the tile, tensions may arise and generate defects such as crazing or flaking. Such tensions sometimes remain hidden and only show up later (latent crazing). Another physical phenomenon caused by increased temperature is the reversible polymorphous transformation that changes quartz α into quartz β (at 575 °C) and then, at even higher temperatures, into tridymite and cristobalite. These transformations entail structural rearrangements, expansive during heating, contractive during cooling. While the overall structure of the tile is, during heating, still elastic and loose enough to absorb expansion without too much difficulty, the contraction associated with the quartz β à quartz α transformation during cooling takes place in a tile that is already rigid and rich in new fragile phases: cooling cycles therefore need to be planned carefully, especially around these temperatures. The same argument applies to materials that have already been fired to a high degree of vitrification and are then fired a second time: in this case similar precautions need to be taken during pre-heating. Applying heat energy to the raw materials causes them to decompose; the compounds that arise from these reactions then become the reagents responsible for the formation of the minerals that constitute the final ceramic article: these changes occur at different temperatures as a function of the energy needed to trigger them. Such reactions can lead to the formation of compounds in a gaseous state, which have a tendency to escape via intergranular interstices: a suitable degree of tile compaction is, then, essential, as it not only optimises sintering but also provides 210

Firing

the permeability needed for expulsion of these pyrolysis and oxidation gases while allowing gaseous exchange with the in-kiln atmosphere (especially the air and the oxygen it contains). In fact, the gases cointained in the in-kiln atmosphere can react with materials in the body, the glaze and even the products of those same reactions. The air is usually directed so as to steer reactions in the desired direction; for those transformations that generate gas or require oxygen air flows are designed to aid gaseous turnover and create an oxidising atmosphere. Vice versa, when an oxygen-poor atmosphere is required, or when the level of oxides needs to be reduced, air circulation can be decreased and/or substances that produce reducing gases such as CO and H2O are introduced into the firing atmosphere. This can be done by varying air inflow via regulation of the burners secondary air ducts or by modifying the relative gas pressure settings in the various kiln sections (keeping settings stable by increasing or decreasing pressure in different parts of the kiln). For example, as seen, organic substances burn between approximately 300 and 850 °C, producing carbon dioxide and water vapour; if heating is gradual, air circulation is good, surface vitrification does not take place and combustion is completed at around 800 °C, resulting in minimum residues and fine porosity. Vice versa, if heating is too fast the organic substances could undergo distillation accompanied by the production of carbon and other reduction products, which blacken the tile (black core defect). Only at higher temperatures does the carbon react with the surrounding oxides which reduce as follows (Me = generic metal): Me2O3 + C à 2MeO + COá or MeO + C à Me + COá These reactions involve gaseous formation of carbon monoxide and cause a colour change towards grey-black, local increase of volume in the mass and local over-firing. Within the same temperature interval (350-500 °C) oxidation of the pyrite (possibly present as an impurity) begins, being completed at higher temperatures: FeS2 + O2 à FeS + SO2 á 4FeS + 7 O2 à 2Fe2O3+ 4 SO2 á If the atmosphere is oxidising the sulphur dioxide changes into SO3 and, in reacting with the basic oxides (CaO) of the body and the glaze, sulphates may be formed: CaO + SO3 à CaSO4 The presence of calcium sulphate in the fired tile is damaging as it could dissolve in the water contained inside the body and appear as saline efflorescence: this emerges on the surface of the pre-fired biscuit and can also cause glaze detachment. 211


Decomposition of alkaline earth carbonates involving the release of carbon dioxide and the formation of the metal oxide takes place between 650-950 °C: the first to decompose are the magnesium carbonates, followed by calcium ones (for a same type, the imperfect or non-crystallised structures decompose first and then the well crystallized ones): CaCO3 Æ CaO + CO2Ç As this is a reversible reaction, decomposition is aided by elimination of the products from the kiln atmosphere: good air circulation and the presence of clayey minerals that react with the calcium oxide by forming silicates and calcium silicaaluminates aid gradual, total dissociation of the carbonate with completion of its fluxing action. Fine particle size distribution also aids decomposition: coarse limestone particles, in fact, decompose only superficially or, in any case do not allow the calcium oxide to fully react with the other oxides. The constitution, at high temperature, of neo-formed crystalline composites such as silicates and calcium silica-aluminates can cause the body to expand, thus counteracting the shrinkage that occurred previously during the destruction of the crystal lattice of the clayey minerals. This increase in volume remains after cooling and explains why products with these body compositions, when fired at the above temperatures, do not shrink in size but, rather, expand. Thanks to the presence of feldspars rich in alkaline elements, the liquid phase formed at high temperatures is initially very viscous and this allows the products to maintain their shape well. Viscosity then decreases as temperatures rise further (this change in viscosity occurs faster in sodium bodies than potassium bodies). Feldspars are thus the vehicle of that vitrification which characterises products with low porosity: the liquid phase, in fact, fills the pores and, as temperature increases, absorbs the clayey mineral oxides more and more, causing considerable shrinkage and increase of density within the mass. Once the liquid saturation point has been reached, the needle-like mullite crystals and aluminum silicate separate, intertwining with each other in the vitreous body to form a structure of high bending strength. This type of mullite (known as secondary mullite) is typical of bodies with a high alumina content, such as porcelain (50% kaolin, 25% quartz, 25% feldspar), fired at very high temperatures (1350 °C) with the formation of an abundant liquid phase. Crystallisation persists as long as the liquid has sufficient fluidity. As regards glazes, compositions consisting mostly of frits do not involve any particular chemical reactions in that these occur during fritting (melting temperature in frit kiln higher of 250-400 °C). Glaze compositions with a higher raw material content have a reactivity that depends on the chemical nature of the materials themselves: sometimes the formation of eutectic mixes causes fusion, while in 212

Firing

The mullite formed in a porcelain body - S primary mullite, A secondary mullite formed from the feldspathic vitreous phase, V vitreous phase, Q dissolving quartz granule.

Viscosity of feldspars as a function of T.

Fig. 3.

other cases dissolution of the materials takes place without them undergoing chemical transformations (e.g. frit to which compounds such as zirconium silicate and quartz have been added). The most significant chemical reactions concern the formation of new compounds which crystallise in the vitreous mass (making the glass opaque or the surface matt) or those that give rise to formation of colorants or certain decorative effects. The firing cycle As firing progresses the product is exposed to different temperatures of varying duration: this temperature-time sequence constitutes the firing cycle. In ceramics a firing cycle is divided into at least three stages: a) heating, up to a maximum temperature that optimises the required ceramic properties (determined via trials): the rate of temperature increase is adjusted according to parameters intrinsic to the material and kiln performance. b) holding (or soaking), at maximum temperature: duration depends on the size of both product and kiln; the larger the tile, the greater the need to homogenise the temperature in order to ensure that the required physical and chemical transformations take place. c) cooling to ambient temperature: occurs at a rate that takes into account the sensitivity of the tile to thermal gradients; other needs must also be considered (e.g. 213


at this stage it may be necessary to aid crystallization phenomena by slowing down the cooling rate within certain temperature intervals). Depending on whether they are endothermic or exothermic, the chemical and physical reactions triggered by firing cause further temperature changes within the body, changes that must be allowed for by firing cycle planners. Generally speaking, decomposition and dehydration reactions and phenomena leading to a more or less condensed state (e.g. fusion or evaporation) are endothermic; oxidations, combustions and the switch from a disordered to a more ordered state (as in crystallization) are exothermic transformations. Within certain temperature intervals tensions can arise within the piece and it is here that the temperature increase/decrease rate (∆T/time = thermal gradient) should be slowed down accordingly, while other intervals may permit much faster temperature change rates. Attainment of appropriate firing cycles (i.e. temperature-time curves) is dependent on a thorough understanding of firing phenomena and the temperatures at which they are triggered. Yet heating gradients and soaking times depend not only on factors concerning the material; of equal importance are the size of the tile (i.e. different cycles for differently sized tiles), load parameters (the heat spreads more efficiently if load density is lower) and the heat diffusion rate of the material given its surface area, thermal conductivity/specific heat and density. Planning a firing curve is thus the difficult art of reconciling kiln productivity with product quality. For centuries kilns with huge chambers and lengthy firing times were used to homogenise heat penetration into the product and so eliminate any internal thermal gradients. Towards the mid-70s, with the advent of single-layer roller kilns, industrial firing times were drastically shortened (to just a few tens of minutes). Nevertheless, studies on rapid firing have shown that even where heating rates are accelerated the chemical-physical transformations are delayed because of the gradient that forms between the temperature of the kiln and that of the products. This phenomena is explained by the fact that it takes time for the kiln heat to spread uniformly through the product mass. It has also been demonstrated that such temperature differences can be diminished by increasing heat transmission coefficients in the material. Thus the practice of holding maximum firing temperatures once they have been achieved is as valid in fast firing as it is in slow firing because it homogenises temperatures and allows reactions to progress and complete. Rapid firing has led to radical changes in kiln loading, kiln structure and the formulation of both bodies and glazes. Tiles subject to rapid firing are introduced into the kiln in a way that maximises the surface area exposed to the heat; tiles are arranged in a single layer so that they can receive heat from every direction as uniformly as possible. 214

Firing

Heating/cooling rates are slowed down only within critical temperature intervals (as explained above) while for other intervals steep gradients can be used. Kilns are now made of refractory materials of low thermal inertia which accumulate little heat, the latter being well distributed throughout the kiln by a series of burners. Thanks to these modifications manufacturers are able to keep output levels high. Different types of firing Firing can be performed in a variety of ways that depend on the sort of product being manufactured. Firing may involve the tile body, the glaze and decorations or glaze and body together. The first involves firing of the dried tile, the second the firing of the glaze as part of a double firing process and the third the single fire process (fig. 4). Body firing involves attainment of temperatures that determine the final qualities of the product (i.e. bending strength, porosity or impermeability, colour). For certain items such as terracotta, refractory, red gres, unglazed klinker and unglazed porcelain tile this fire is the first and final one. In the double firing process the first fire represents only the initial stage of heat treatment, conferring the solidity, bending strength and chemical-thermal inertia needed at the glaze application stage. Once the glaze has been applied the tile can be re-fired at even lower temperatures (usually 30-40 °C less, to prevent the vitreous glaze being affected by any reactions originating from the body). This process concerns faience, earthenware and soft porcelains (i.e. wall tiles and low compaction/cohesion tableware). Single firing allows manufacturers to obtain the desired body and glaze properties simultaneously. This technique offers significant advantages, both economic (faster production process, lower capital investment, energy and labour requirements) and technical (during firing an intermediate layer forms between glaze and body, thus improving body-glaze adhesion). However, the advent of this new firing technique, back in the 70s, necessitated changes to both plant hardware and body/glaze formulation. The body necessitates a composition that is, already in its unfired dried state, solid and strong enough to withstand the stresses of moulding/pressing, glazing and decorating; during firing it must release a minimum quantity of gas before the glaze becomes completely molten so as to prevent the formation of bubbles and pin-hole defects. The body must also maintain open porosity up to about 800-850 °C to allow gaseous exchange with the in-kiln atmosphere and thus prevent black core problems. 215


Temperature (°C)

Curve in presence of black core

Lower zone

Upper zone

Firing time (min.)

Fig. 4. Vitrified material firing curve (single firing).

Finally, the body must be formulated so as to minimise the shrinkage that can cause distortion of decorations. The glaze, which also includes decoration, must be of a composition that matures at the same temperature at which the body attains its final characteristics, must melt when the body has already ceased to generate gases and must spread and level quickly so as to coat the tile evenly; moreover, it should not be affected by the gases/steam in the kiln and must be aggressive enough to react with the body and form the intermediate layer. The hardware changes brought about by single firing largely involve the kiln and automation both upstream and downstream from it. For instance, storage and handling systems for glazed-decorated materials (waiting to be fired) and kiln feeding/unloading units have been extensively automated. Fuels The heat needed to fire the tiles is normally produced by burning solid, liquid or gaseous fuels or even by using electricity. Combustion is a fast oxidation reaction and thus exothermic enough to be used for the production of heat. Because of this reaction the fuel, made up of compounds such as carbon, hydrogen, sulphur, carbon monoxide, hydrocarbons, combines (following ignition) with oxygen in the air to produce reaction products and heat.

216

Firing

C + O2 CH4 + 2O2 H2 + ½O2

+ 94.2 94,2 kcal/mole → CO2↑ → CO2↑+ 2H2O↑ + 192 kcal/mole 58.6 kcal/mole + 58,6 → H2O↑

The heat triggers and continues the combustion. In terms of commercial worth, the main thing that differentiates fuels is their calorific value; this depends on the composition of the fuel and is defined as the quantity of heat (kcal) liberated on complete combustion of 1 Kg of solid or liquid fuel or 1 m3 of gaseous fuel (see tab. 1). During combustion some substances may subtract heat by activating endothermal processes such as changes of state (e.g. liquid-gas) or simply by heating up (e.g. the nitrogen in the combustion air). For the fuel to express its full calorific value it needs to be mixed as well as possible with the combustion air; the latter must also be present in quantities at least equal to the theoretical values indicated by the combustion reaction. In practice, apart from situations that require a reducing atmosphere, imperfect mixing with combustion air can be compensated for by adding excess air: this sur-

Density Molec. Kg/Nm3 W.

Gas

Symb.

Hydrogen

H2

2

0.090

Methane

CH4

16

0.717

Ethane

C2H6

30

1.356

Propane

C3H8

44

2.019

Butane

C4H10

58

2.703

Ethylene

C2H4

28

1.261

Propylene

C3H6

42

1.915

Butylene

C4H8

56

2.501

C. Oxide

CO

28

1.250

Oxygen Nitrogen Carb. Dioxide Sulph. Dioxide Water Dry air

O2 N2 CO2 SO2 H2O --

32 28 44 64 18 29

1.428 1.250 1.997 2.926 0.804 1.293

High-Cal. value Kcal/Nm3

Low-Cal. value Kcal/Nm 3

3050 12.770 9520 39.858 16820 70.422 24320 101.823 32010 134.019 15290 64.016 22540 94.370 29819 124.808 3020 12.644

2570 10.760 8550 35.797 15370 64.351 22350 93.575 29510 123.552 14320 59.955 21070 88.216 27840 116.560 3020 12.644

Oxygen requirement

Combustion products CO2 H 2O

N2

0.5

2.38

-

1

1.88

2

9.52

1

2

7.52

3.5

16.66

2

3

13.16

5

23.80

3

4

18.80

6.5

30.94

4

5

24.44

3

14.28

2

2

11.28

4.5

21.42

3

3

16.92

6

28.56

4

4

22.56

0.5

2.38

1

--

1.88

Tab. 1. Properties of gases involved in combustion under normal conditions (0 °C, 760 mm Hg, dry).

217


plus is in the order of 40-150%, 25-60% and 10-50% for solid, liquid and gaseous fuels respectively. Gaseous fuels are particularly advantageous in that they blend completely with the air, hence combustion can occur with air quantities only slightly above theoretical levels. Moreover, they can be burnt close to the products without causing any damage, do not produce ash, do not need preheating and thus burners of relatively simple design can be used. Gaseous fuels provide very high temperatures because of their high calorific value and because pre-heated air can be used; they also make control of heating and the in-kiln atmosphere more efficient. Natural gas is a mix of methane, ethane and limited quantities of other light hydrocarbons: it is found both in gas-only deposits and in oil deposits as a gaseous phase. Having a calorific value of around 8500 kcal/m3, it burns with a very pure flame and contains negligible quantities of sulphur. It is thus the most commonly used fuel in open flame kilns (where combustion gases come into direct contact with the product). Generally supplied to factories via pressurised piping, it goes through an intermediate decompression stage before reaching the burners. Liquid fuels also have certain advantages: easier batching and distribution throughout the various kiln sections, good mixing with the combustion air by way of atomisation, limited ash production, small efficient burners. To ensure efficient combustion two fuel-air mixing methods may be used: atomisation of the liquid in a fan-blown air stream (the most widespread system) the liquid is vaporised by using a part of its own combustion heat. The liquid is atomised by pumping it into a duct which ends in a nozzle. Burners may be fitted on the walls or the roof of the firing chamber; in the latter case burners that inject a given quantity of fuel at set time intervals are used. Ceramic kilns can also use liquid fuels such as the so-called light oils and oil distillates, generally available as either kerosene or naphtha: the latter is usually preferred on account of its price. Combustion is generally good, yet the combustion system as whole is an expensive one and maintenance is more frequent than with natural gas systems. At ambient temperature heavy oils generally have the consistency of tar and must therefore be preheated to at least 60 °C to make them fluid enough to circulate through the ducts and then heated to at least 110 °C to ensure good atomisation. Combustion is poorer and produces a considerable quantity of carbon residues, sulphur dioxides and other pollutants. Because of these problems heavy oils are only used where better fuels are either unavailable or unaffordable. Where fuels of better quality are unavailable, it may also be possible to use poor 218

Firing

gases of low calorific value, such as gases derived from the treatment of coke, or mixtures of hydrogen, methane and CO etc. The most suitable liquid fuel for modern burners is undoubtedly liquid petroleum gas (LPG). It is clean, of high calorific value and, in most cases, requires burners similar to those used with natural gas; however, storage and distribution systems are required, and a good understanding of its technological behaviour is necessary as it is commercialised in liquid form but used as a gas. This necessitates the installation of special vaporising units (electrical or with burner). Feeding the fuel that spontaneously evaporates inside the tank directly to the kiln is not feasible as it is impossible to keep the composition of the gas consistent (the more volatile elements evaporate first). To stop the fuel entering the distribution network while still in its liquid state and prevent any ice forming when the fuel is decompressed to the standard 500-600 mm H2O, it is good practice to follow the values given in the following tables (tables 2 and 3). By-weight percentage composition of LPG (propane/n-Butane)

Mains pressure BAR

Propane 100/0

0.5 0.75 1 1.5 2

6 7 8 9 10

90/10 6 7 8 9 10

80/20 6 7 8 9 10

70/30 6 7 8 10 15

60/40 6 7 9 14 20

50/50 6 9 13 19 24

40/60 9 12 16 22 27

30/70 12 16 20 26 31

20/80 15 19 23 29 34

n Butane 0/100

10/90 18 22 26 32 38

21 25 29 35 41

Tab. 2. Minimum LPG temperatures and pressures in burner feed lines. 3

Gas

3

Gas (m ) produced by the stoichiometric combustion of 1 m of gas with (dry/humid) air. CO 2 N2 H 2O Totale

Methane

0.998 / 0.998

7.470 / 7.470

1.934 / 2.046

10.491 / 10.603

Ethane

2.011 / 2.011

13.174 / 13.174

2.924 / 3.121

18.266 / 18.463

Propane

3.052 / 3.052

19.041 / 19.041

3.945 / 4.229

26.265 / 26.549

Butane

4.147 / 4.147

25.226 / 25.226

5.025 / 5.401

34.698 / 35.074

Tab. 3. Combustion products.

The temperature of the gas depends on by-weight percentage composition of the LPG, expressed as a mixture of propane and n-Butane and the effective pressure upstream from the kiln control unit (expressed in bars). Example: if the LPG is made up of 30% Propane and 70% n-Butane, and the pressure in the feed line is 1 bar, the temperature of the incoming gas must not be less than 20 °C for the safety margin to be acceptable. With the same composition at a pressure of just 0.5 bar, a temperature of 12 °C would be sufficient. 219


Where LPG is used its by-weight composition is not always known and varies over time: hence it is always advisable to use pressures of around 0.5 bar to prevent the formation of liquid phases or ice in the equipment or system. CERAMIC FIRING KILNS Heat exchange The kiln is the machine in which the tiles are fired to obtain the desired product characteristics. During the various stages of firing heat exchange takes place between the heat sources, the fluids, the kiln structure, the products being fired and the atmosphere surrounding them: this involves conduction (of limited importance), convection (the main form of heat transmission) and radiation (relevant only in the high temperature zones). Heat transmission by convection Convection transports the heat via air circulation (hot air rises, cool air descends), transmitting a quantity of heat which, other conditions remaining equal, is: Q = ρ . S . ∆T where S is the surface area in contact with the heat, ∆T is the temperature difference between ceramic body and gas and P represents a convection coefficient which varies widely depending on circumstances: this coefficient also indicates how heat exchange depends, to a limited extent, on the temperature difference between the air and the heated solids, while controlling the movement of that air is, instead, very important for heat exchange. Air circulation depends on the position of the heat source with respect to the fume intake duct, high speed burners and forceddraught circulation fans. Convection is also aided by loading the tiles so that the air can circulate freely between them; also, mounting high-speed burners on opposing kiln walls in a staggered pattern aids convection by producing turbulence and gas circulation at right angles to the direction of product feed. Heat is also propagated by way of radiation, transmitted via the infra red rays emitted by all hot bodies. The quantity of heat emitted depends on temperature as per the following equation: Q = σ (T24 T14) where T2 and T1 are the temperature of the emitting (usually flame or in-kiln walls) and the receiving material (usually tiles or cooling air) respectively. 220

Firing

Logically, the higher the kiln temperature the more important this phenomenon, to the extent that in some cases, especially where flames are in the immediate vicinity of the tiles, their surfaces are protected with refractory material. Heat transmission by radiation is typical of muffle kilns and where products are contained in saggars. It is, in fact, the refractory material that radiates its accumulated heat to the tiles. Radiation can also be used to accelerate cooling of items that have been fired at high temperature. In this case the radiation given off by the bodies is captured by silicon carbide tubes (with a high infrared absorption capacity) through which ambient air is circulated. Kiln construction Today, the most common type of tile firing kiln is the single layer type, where just one layer of material is passed through the kiln on a set of rollers. Its length corresponds to the required firing time. Double layer roller kilns are still used as they require less space (fig. 5). On entering the kiln the tiles are fed through zones of steadily increasing temperature (preheating) until they reach the maximum temperature zone. Subsequently the tiles enter a fast cooling zone where the gradient flattens out a little to prevent problems associated with the presence of quartz and steepens again until the kiln outlet, where temperatures vary between 40 and 60 °C. The firing cycle is defined as the time taken by the tiles to travel from one end of the kiln to the other and, together with cross sectional load capacity and number of channels, is a key determinant of output capacity. The tiles usually pass through the kiln at constant speed.

Fig. 5. Double layer roller kiln.

221


Generally, tiles are transported through the kiln on rollers. The kiln is divided into modules, each about 2 metres long, assembled end to end; they are made of tubular frames and sheet metal panels which support the insulating layer, made of refractory brick and insulating mortar on the floor and refractory slabs and Z-blocks on the walls and roof. Fig. 6 shows how kiln insulation consists of layers of different materials and illustrates the resulting insulation performance. Module composition varies enormously depending on production requirements: in general, single fire tile production involves kilns 60-120 m long, with effective cross sections of 1.3-3 m (usable area of 80-360 m2), sub-divided, approximately, as follows: Pre-kiln (10% of total kiln length) Preheating (31%) Firing (19%) Fast cooling (6%) Slow cooling (20%) Final cooling (14%)

Fig. 6. Illustration of kiln wall showing thermal insulation.

222

Firing

The vertical cross section of the chamber is kept to an absolute minimum so as to accelerate fume speed and thus increase fume-tile heat exchange. Air flow is provided by fans (there are usually six) which have, starting from the kiln infeed, the following functions: fume exhaust combustion air input rapid cooling air input slow cooling heat exchanger exhaust final cooling hot air exhaust final cooling cold air input. Fig. 7 shows side view, plan and cross-sections for a typical single layer kiln of the latest generation. This diagram also illustrates the efficient use of combustion gas; in most kilns the firing zone is characterised by a flow of fumes that runs against the direction of tile feed, the fumes being exhausted through the main stack located in the kiln entrance zone. This set-up is of prime energy management importance because it allows the heat generated in the firing zone to be used for preheating purposes; where toxic or undesired emissions are present in the preheating zone, they are sent to the stack directly for depuration treatment. Combustion system Gaseous or gassified fuels such as methane and LPG are burnt directly in the combustion chamber and then expelled as fumes into the kiln interior. The burners mix the fuel with air drawn in through ducts by fans. Such air may be defined as either primary (the burner comburent air) or secondary (used to fine-tune the firing atmosphere). In most cases, however, fuel-air mixing takes place directly at the nozzles: by varying the diameter of the primary combustion chamber outlet different crosssectional heat distribution patterns can be obtained. The extreme temperatures in the combustion chamber necessitate utilisation of specially designed materials, such as silicon carbide (SiC) or sometimes SiSiC. Adjusting the air flow makes it possible to obtain hot gases that exit the burner at speeds of 100 m/s and more, producing directional jets than maintain uniform temperature along almost their entire length, which usually corresponds to kiln width. As the burners are arranged in an alternating, staggered pattern above and below the rollers, fume turbulence is maximised, thus ensuring maximum heating uniformity. The traditional burner modulation systems are: fixed air / modulated gas modulated air / modulated gas.

223

224 Firing section

Fast cooling

Slow cooling

Fig. 7. Side view and plan of single layer kiln showing how cross section changes from zone to zone.

Kiln stack pre-heating section

Final cooling


Firing

In the former the quantity of air arriving at the burner, once adjusted, remains constant independently of changes in load or anything else which might alter the temperature in the firing zone: temperature settings are maintained via gas modulation only. With this system, it is obvious that the burner will operate at a correct gas/air stoichiometric ratio only within a narrow operating range, as the most common situation is one of excess air. This system is of simple design, gives good pressure stability in the kiln where there are gaps in the load, yet is costly in terms of fuel consumption. In the latter both gas and air can be modulated (at a constant ratio) to compensate for gaps in the in-kiln tile load or other factors that might alter temperatures in the zone. Temperature is maintained by varying air and gas flows simultaneously and consequently acting on the volumes inside the kiln. Note also that with this system the burners generally operate at a stoichiometric ratio which optimises fuel consumption. However, negative aspects such as pressure instability caused by gaps in the load, greater constructive complexity and difficulties in avoiding temperature peaks can cause serious problems, especially where kilns are particularly wide and long. High output (wide infeed) kilns and very fast firing cycles demand narrow tile size tolerances. Consequently the use of pulse-type systems with burners staggered on opposing kiln walls and adjustable flame distribution systems, which use refractory ducts or deflectors, are becoming increasingly common. Both these solutions optimise heat control, which, thanks to the development of ever-more reliable control electronics and measuring systems, limit temperature fluctuation to just a few degrees Celsius, thus providing excellent results in terms of wall to wall heating homogeneity. To sum up our description of the combustion system it should be pointed out that enormous efforts have been made to reduce energy consumption. This is highlighted by the simple fact that in 1980 it took 967 Kcal to produce 1 Kg of 300 × 300 mm glazed tiles with a water absorption of 5-6%, while today just 450-500 Kcal are needed. One simple, highly effective solution in this regard is use of pre-heated combustion air. Even where air is heated to just 100 °C fuel savings are in the order of 56%. More radical heat recovery (from the cooling zone) allows combustion air to be used at temperatures as high as 230-250 °C, resulting in savings of some 12-14% (see tab. 4). Without entering into the merits of different plant layouts, the following table shows specific consumption for a kiln with a 2.5 m wide infeed and a total length of 110 m, a 47 minute firing cycle, Tmax 1220 °C for the firing of porcelain tiles, shrinkage 8%, weight 20 Kg/m2, output 6400 m2/day (5350 Kg/h fired product): The heating sequence and the corresponding kiln structure can be described in terms of the reactions that occur at certain temperatures. The following paragraphs chart the progress of the tile through the various kiln zones.

225


COMBUSTION TEMPERATURA AIR ARIA COMBURENTE TEMPERATURE SPECIFIC CONSUMO CONSUMPTION SPECIFICO Kcal/Kg unfired) (Kcal/Kgcrudo)

30 °C

100 °C

160 °C

220 °C

485

456

440

420(-(13.4%) 420 13,4 %)

Tab. 4. Specific consumption in a single layer kiln as a function of combustion air temperature.

1 - Pre-kiln This is the kiln inlet, designed to eliminate any residual hygroscopic water left behind by the drying process, picked up during glazing or absorbed following unsuitable storage: maximum permitted water content is 2% by weight. Here, elimination of the zeolite water in the clay also begins. This part of the kiln does not usually have its own heating apparatus, but relies, instead, on transfer of fumes from the downstream firing zones via stack ducts located above and below the pre-kiln rollers. Temperature is monitored by a thermocouple in the roof. Pre-kiln temperatures range from 200 to 500 °C, but considering the speed of the cycle and the endothermal nature of evaporation the material itself reaches temperatures of just 50-200 °C. The only temperature regulation device is an air intake located in the roof and floor of the pre-kiln, adjustable by way of a gate valve. The vertical height of the kiln interior at the early end is minimised so as to accelerate fume speed and consequently increase fume-tile heat exchange. Both walls and roof are insulated with rigid slabs made of special fibres, anchored to the metal frame of the module. The floor, instead, is shielded with harder insulating bricks capable of withstanding the wear generated by the periodic removal of tile debris. 2 - Pre-heating Here, the key event is the degassing of the ceramic body, indispensable for the prevention of swelling, bubbling and cratering of the glaze/body and colour alterations. The relevant temperature range depends on the type of product being fired and, even more importantly, the type of glaze. Conceptually, preheating ends where glaze melting begins and the surface porosity of the tile begins to drop, causing a sudden fall in gas permeability. Pre-heating temperature ranges may be in the order of 500-700 °C or, if the glaze is of good quality, 500-1000 °C. For exceptionally high-fluxing bodies and glazes it may extend as far as 1100 °C. Pre-heating also involves the delicate allotrope transformation of quartz α in quartz β. This must be completed without the tension generated by the sudden increase in the volume of the ceramic body causing any breakages: in short, preheating must be efficient and suited to the characteristics of the tile. The pre-heating zone has a series of burners, mounted on the walls above and 226

Firing

below the rollers. Kilns for the second glaze fire do not have burners below the rollers, except in the firing zone. The burners are sub-divided into several multiburner sets on two kiln modules, staggered symmetrically along the right and lefthand sides of the kiln. Burner sets above and below the rollers are always adjusted separately. Compared to the pre-kiln, the pre-heating section is higher and modules are larger as they require thicker insulation. The innermost layer of the walls is made of insulating brick and a secondary layer of fibre. The roof is lined with lightweight refractory blocks, anchored by metal couplings to the tubular structure on the module frame. In addition to the blocks there is also a second layer of refractory fibres and a third layer of insulating cement. The floor consists entirely of various types of insulating brick or compressed refractory. To shield it from any mechanical impact, the floor is lined with thin, drymounted tiles made of dense refractory materials. Wall insulation is completed by plugging the roller seats with insulating fibre. 3 - Firing This is where temperatures breach the 1000 °C barrier and are at their maximum. Insulation: the innermost layers of the wall are made of refractory-insulating bricks of maximum refractory performance; insulating fibres complete the shielding. Note that insulation thickness varies as a function of the temperatures required for specific product characteristics. To insulate properly at temperatures of 1100-1250 °C three different layers of insulation are used, thus providing good kiln performance at optimised running costs. The entire firing zone is equipped with wall-mounted burners both above and below the rollers. It is in the firing zone that virtually all the final size, flatness, vitrification and glaze characteristics of the tile are established. Hence fine temperature control and a perfectly flat roller plane are essential, as tiles can soften considerably. The influence of the adjoining cooling zone also needs to be attenuated: this is achieved by inserting two barriers at the end of the firing zone, each consisting of a wall that cuts off the lower part of the firing channel between floor and rollers and rigid insulating fibre slabs (chicanes) inserted through an aperture in the kiln roof to divide the channel above the rollers. Inserting similar barriers in the firing or pre-heating zones, while often a source of frequent and sometimes costly maintenance work, rarely gives appreciable results. Hence they are only installed where essential (e.g. in very short kilns or kilns for products with special requirements). 4 - Rapid cooling This phase covers the interval between maximum firing temperature and about 600 °C. The process is designed to cool the tiles as fast as possible, but must not continue to the point at which allotrope transformation of quartz takes place (573 °C). 227


Because this stage involves a sharp drop in temperature and the return of both body and glaze to a solid state, it is undoubtedly one of the most critical in the entire firing process. Rapid cooling is achieved by way of blower pipes housed in the walls, both above and below the rollers: the air exits the pipes and enters the kiln chamber where required via a series of aligned holes, which ensure even air distribution over the load and are generally made of austenitic steel or, where exposed to very high temperatures, silicon carbide. Temperature control in the fast cooling zone is provided by thermocouples installed both above and below the rollers. Equally important is the heat exchanger, consisting of a series of steel pipes running transverse to the kiln below the roof; the air flowing through these pipes is actually the combustion air on its way to the burners. Hence the exchanger has the dual function of cooling the tiles and pre-heating the burner combustion air, normally to 100-120 °C. At these temperatures the air is already sufficiently expanded, is moisture-free and thus will not damage the kiln interior in the event of a burner shutdown. Wall insulation in the rapid cooling zone consists of an initial layer of refractoryinsulating brick and then a layer of insulating fibre. Insulation is thinner here as a result of the relatively low temperatures. 5 - Slow cooling This is the stage at which the highly delicate allotrope transformation of quartz takes place, an event that involves a sharp decrease in the volume of the ceramic body. As the above subtitle implies, cooling must proceed slowly and gradually so that this transformation takes place simultaneously throughout the tile: otherwise the tensions generated in the already-rigid ceramic tile can easily reproduce the characteristic cracking which, in medium-high vitrification, gives the fracture a smooth, shiny, conchoid look with sharp edges. Temperatures in this part of the kiln are in the order of 600-450 °C, corresponding to actual tile temperatures of 700-500 °C. Heat exchange is provided by a nest of tubes running widthways beneath the roof of the kiln; a fan draws in cold ambient air and circulates it through the tubes. The flow is routed so that air in adjoining tubes flows in opposite directions. Some tubes can be excluded by closing valves on the hot air ducts. As in the pre-kiln, the roof of the kiln is lower and the modules are smaller because lower temperatures require thinner insulation. The roof chamber is insulated with rigid fibre slabs anchored to the module frame while the walls are insulated with brick on the innermost layer and then fibre; the floor is made of insulating brick. In this part of the kiln the refractory plays a secondary heat insulation role, by ensuring that temperatures remain sufficiently high when there are gaps in the tile flow. The end of the slow cooling zone is marked by two transverse barriers, positioned closely together. Both consist of a wall sectioning off the below-roller level and austenitic steel 228

Firing

slabs slotted through apertures in the kiln roof above it; the extremities of these chicanes are hinged so that accidentally overlapped pieces can pass through. The purpose of these barriers is to control air counter-flow in the final cooling section. 6 - Final cooling At this point the tile has passed the critical quartz transformation point. Final cooling is thus designed to lower the latent heat of the product as much as possible. Cold air is introduced directly above and below the tiles by blower sets consisting of holed triple-tubes running widthways. Flow rates on individual sets can be adjusted by means of a gate valve on the air feed system. A second fan withdraws the air that inevitably heats on contact with the tiles, removing it via ducts in the kiln roof fitted with adjustable throttle valves. These intakes extend into the slow cooling zone: one is positioned between the two barriers separating the slow and final cooling zones and another is located upstream from the barriers themselves: this last intake is useful when the kiln is being heated and in establishing the necessary equilibrium regarding air volumes flowing towards the firing zone. Roller drive units The rollers are generally driven by a gear transmission system, usually a worm screw driving an angular gear. The series of motor gears is fixed by grub screws to a steel shaft, one for each kiln module. The drive shafts in two, three or four modules can be coupled together. The shaft (or group of shafts) is driven by a gearmotor and chain transmission system. Speed is adjusted automatically by software. Motor reducers piloted by frequency modulators are also becoming common. All gears are shielded by sealed casings which also act as oil bath sumps. Circulation of air volumes and pressure in the kiln Balancing the air volumes in the kiln is of considerable importance, as implied by the sheer size of the fans. The combustion air fan, the fast cooling fan and the final cooling air blower all introduce significant volumes of air into the kiln. Similarly, the fume fan and hot air removal fan in the final cooling zone extract significant volumes. The fume intakes at the start of the pre-kiln are arranged symmetrically above and below the rollers (see first cross section in fig. 7). They are not angled towards the incoming fume flow: they simply generate a vacuum towards which the fumes flow automatically. Intake flow is adjusted individually above the rollers and as a set below by throttle valves: these allow the user to balance flows and depressions above and below the rollers. The fume fan flow rate can be adjusted in several ways: usually, a manually controlled throttle valve is installed upstream from the fan intake. Alternatively, valves 229


can be adjusted by servomotors (controlled via the kiln control panel) or frequency modulators (inverters) on the fan itself: these systems may also include an automatic kiln pressure adjustment device. An ambient air intake on the fume stack is also present: this is usually a manual throttle valve and allows fume temperatures to be lowered when excessively high for the fan; it also acts as a fine adjuster of exhaust fume quantities. However, the systems which distribute fume suction between the above-roller and under-roller sections are of limited efficiency: even when some 10-12 metres from the intakes the fumes already tend to follow the natural route above the rollers. This tendency is aided by the intercommunication afforded by the interroller gaps (where uncovered by tiles). The exact amount of combustion air supplied to the burners (quantities usually vary between 5 and 35 m3/h) depends on requirements. Dilated by the high temperatures, this air represents a significant proportion of the volumes that need to be evacuated. To these volumes at least a part of the fast cooling air is added; the air blown into the fast cooling section is partially or wholly exhausted by the fume fan or the hot air intakes in the final cooling section. Where removed by the fume fan this air has a positive effect on fuel consumption because it is already heated (so less cold air is fed to the burners) and provides good oxygenation in the kiln interior (fig. 9). Where, instead, it flows towards the final cooling zone in the same direction as the tiles, this air can be of considerable importance in keeping the kiln interior hot in the event of extensive gaps in the tile load; this protects the leading row of tiles in the kiln from the effects of the volume changes associated with the quartz transformation (fig. 8). The quantity of air introduced is modulated automatically so as to maintain correct temperatures in the fast cooling zone; consequently, especially where tile feed is discontinuous and thermoregulator default parameters are not optimal, kiln pressure may oscillate. The final cooling air is blown in through widthways blower ducts arranged both above and below the rollers. The air exits through blowholes positioned perpendicular to the tile load. This final cooling arrangement optimises efficiency and enhances the opportunities for energy recovery as air volumes at 110-160 °C are easily employed elsewhere. Pressure is important for kiln regulation purposes but does not, in itself, have much influence on firing results (except for planarity).

Combustion air

Pre-kiln

Preheating

Cooling air

Firing

Fig. 8. Air circulation with partial exhaust of cooling air.

230

Fast cooling

Slow cooling

Final cooling

Firing

Cooling air

Combustion air Pre-kiln

Preheating

Firing

Fast cooling

Slow cooling

Final cooling

Fig. 9. Air circulation with conveyance towards main stack.

When a kiln pressure reading is taken it only refers to a single point in the entire firing channel: bear in mind that pressure is at its minimum at the fume extraction points in the pre-kiln and at maximum in the fast cooling zone on account of the large volumes of air being introduced. Within a given zone, pressure is higher at the roof level, minimum at floor level, the differences often being considerable. For purposes of convenience, kiln pressure is detected at the burner level, above the rollers just before the last firing zone module. Another, less technical yet more practical and meaningful method consists of localising the ± 0 pressure point between the partial vacuum of preheating and the pressure of the firing zone. The following diagram illustrates how pressure changes from zone to zone (fig. 10). In any case, kiln pressure should not exceed a limit of approximately 0.3 mm and draught should be balanced so that fumes produced in the firing zone are not drawn towards the kiln tile outlet: the latter can increase fuel consumption and cause degassing problems in pre-heating.

Slow cooling

pressure

Fig. 10. How pressure varies in the various kiln zones.

231

Final cooling

Fast cooling

partial vacuum

Slow cooling

Fast cooling

Firing

Cooling air output

Combust. air

Pre-heated air

To the stack

Preheating

Final cooling


Controls Proper running of a ceramic kiln involves a whole range of controls: composition of fumes exiting the stack, composition (or at least oxygen content) of the atmosphere in the kiln, fan speeds and flow rates etc. Yet the key parameters requiring monitoring are pressure and temperature. Pressure can be controlled with simple instruments that indicate the low levels of pressure/vacuum in the various kiln zones. Simple glass-tube manometers, with one end open and in communication with the ambient atmosphere and the other end connected to the firing chamber by a special heat resistant tube, can be used. Even water can be used as a filling medium, the difference in level with respect to equilibrium indicating the degree of pressure or vacuum. Inclined-tube manometers, which operate according to the same principle, are also used and give good resolution (up to 1/20 mm). Several temperature monitoring instruments exist, each using different principles of measurement. The main problem is getting an accurate picture of how the heat is distributed between the tiles and the kiln structure, in that measurement generally takes place without direct contact because the tiles are moving and the firing chamber is large and sealed. One of the simplest and most widespread high temperature measurement devices is the thermoelectric pyrometer, also known as a thermocouple; this consists of two dissimilar conductors joined together at their ends which, by way of the Seebeck effect, generate a thermoelectric voltage between the two junctions proportional to the temperature difference between the junctions. Different types of metal are coupled depending on the temperature interval being monitored. Up to 600 °C a copper and copper/nickel alloy match is usually employed; up to 1000 °C chrome/nickel alloy (cromel) and aluminium/nickel alloy (alumel) are used, while for higher temperatures, up to about 1600 °C, various platinum alloys with rhodium (Pt/Pt-Rh) are preferred. However, as mentioned, the problem lies in obtaining a contact measurement: thermocouples are generally used for continuous control of the various kiln zones and occasional material checks (e.g. by inserting them inside rollers to different inkiln depths). Optical pyrometers are also an effective means of measuring temperature. These instruments determine the temperature of a very hot surface from its incandescent brightness; the image of the surface is focused in the plane of an electrically heated wire, and current through the wire is adjusted until the wire blends into the image of the surface. Even better performance is attainable with IR ray measurement pyrometers. In this case proper calibration of the instrument, effected via the choice of a suitable emissivity (ε) coefficient, is extremely important. Tile temperature can be estimated by passing certain ceramic objects of a known temperature-shrinkage ratio through the kiln. Manufacturers generally use Buller rings with a porcelain body, which have a diameter of about 63.5 mm. After firing the ring is measured again and size diffe232

Firing

rences can be translated into temperature via a conversion table. However, such measurements are of limited use as they depend heavily on the amount of time taken for the rings to pass through the kiln, thus giving a relative rather than an absolute indication of temperature. Nevertheless, they are useful monitors of cross sectional temperature stability. Bear in mind that the unfired ring diameter is not calibrated so the effective difference between unfired and fired diameters must be recorded accurately, without conclusions being drawn on the basis of fired diameters only. Finally, specially shielded recorders can be passed through the kiln together with the tiles. Throughout the firing cycle, these receive and record signals from various connected thermocouples. Correct positioning of the thermocouples thus provides valuable information on the thermal inertia of the tiles and how temperature is distributed between their interior and exterior. An in-depth treatise of optimal kiln management, especially specific fuel and electricity consumption etc., is beyond the scope of this volume. However, the reader will undoubtedly find the heat balances in figures 11 and 12 useful as they provide a good overview of how energy is distributed and consumed. The diagrams illustrate tile manufacturing situations with combustion air at ambient temperature (fig. 11) and preheated to 200 °C (fig. 12).

17.3%

10%

14%

Endothermic reactions

Structural dispersion 70 kcal/kgP

1.62 Kg of fumes + air T=250 °C 89 Kcal/kgp

50 Kcal/kgp

1 Kg of mat to be fired

55,7%

Hot air 12 Kg

T= 126 °C 276 Kcal/kgp

1 Kg of fired mat. T= 90°C 15 Kcal/kgp

Roller Kiln

Combustion air

Fuel 500 Kcal/kgp

T=30°C e=1.10

Combustion air T = 30 °C

Indicative heat balance of a roller kiln

Fig. 11. Approximate heat balance for a roller kiln with combustion air at 30 °C.

233

3%


10.8%

12%

16.7%

Structural dispersion

0.82 Kg of fumes + air 45 Kcal/kg T=250 °C Kcal/kgp

Endother mic reactions

70 Kcal/kgp

1 Kg of mat to be fired. T=30 °C

50 Kcal/kgp

Hot air 10.4 Kg 240 Kcal/kgp

T= 126 °C

1 Kg of fired mat. T= 90°C 15

Roller Kiln

Combustion air 0.74 kg 56 Kcal/kg T=230 °C e=1.10

57%

3.5%

Fuel 420 Kcal/kgp

Combustion air preheated via recovery from cooling zones T = 230 °C

Fig. 12. Approximate heat balance for a roller kiln with combustion air at 200 °C.

Rollers The roller is one of the most important components in the kiln. Like the kiln itself, it has changed much in recent years. As firing temperatures have risen, kiln infeed widths have increased and firing cycles become faster, so there has been a gradual switch from metal rollers to ceramic rollers. It was the advent of firing temperatures above 1160-70 °C that led to the widespread abandon of metal rollers as they deteriorate quickly at such temperatures, even when made of expensive, quality steels such as lNCONEL 601. Moreover a comprehensive range of rollers is now available. Formulas, which may be both oxidic and non-oxidic, are optimised for the various kiln zones. Ceramic rollers provide excellent quality at relatively contained cost, thus making metal rollers virtually obsolete (they are, though, sometimes used at the kiln outlet). Metal rollers Initially, then, roller kilns used only metal rollers, from the simple Mannesman roller to the stainless steel roller, with rollers of increasing quality being used towards and in the firing zone. 234

Firing

Rollers were generally of the following quality: Mannesman for temperatures up to 300 °C Aisi 310 S for temperatures up to 900 °C Inconel 601 for temperatures up to 1170 °C. Metal rollers have the following advantages: they are easy to clean: because they have a much higher expansion coefficient than the ceramic materials that stick to them. Hence any ceramic dust tends to detach from the rollers when the roller is cooled suddenly. Simply extracting a hot roller will usually be enough to clean it. insensitive to thermal shock: the metal roller is unaffected by emergency kiln shutdowns. Heating requires fewer precautions, so kilns with metal rollers and internal fibre insulation can be switched on and off repeatedly, even weekly. they are long lasting: as long as there are no aggressive chemicals in the kiln atmosphere. they stay straight: this is especially important in kiln zones with substantial differences between above-roller and below-roller temperatures. This quality stems from good conductivity, which prevents differentiated heating and thus deformation. Of course, metal rollers have been abandoned because they also have several disadvantages: cost: generally three/four times more expensive than ceramic rollers. durability: limited where kiln atmospheres are chemically aggressive; sulphur compounds are particularly damaging as they react with the nickel in the steel to form low fluxing nickel sulphide; in extreme scenarios this reaction can perforate the roller in just 72 hours. impossibility of use at high temperatures on account of the fast deterioration and bending that takes place above 1160 °C, thus excluding their use in the manufacture of single fire wall tiles and porcelain products in general. impossibility of use in wide kilns as the rollers flex even under lightweight loads. As the disadvantages clearly outweigh the advantages recent years have seen almost total abandonment of metal rollers, except in: porous biscuit for double fire tile manufacture, in which the characteristics of the body materials (without sulphurs) cause them to stick to the rollers very quickly in the preheating zone. the last low-temperature part of the cooling zone, where the environment is clean, to increase tile separation on the rollers and facilitate the work of unloading machines. Since the above represent such a narrow range of the industrys output manufacturers generally prefer to optimise the bodies, improve tile underside cleaning performance and improve overall kiln control so as not to have to resort to the use of metal rollers. 235


Ceramic rollers Developments in ceramic tile manufacturing techniques and the advent of new tile types have driven parallel developments in the physical-chemical-ceramic compositions of the rollers that carry them. Formulation of new roller compositions for use with ever-higher temperatures, faster cycles and larger tiles soon became necessary to reduce tile deformation, especially flatness defects. Producers have responded to these needs by providing specific kinds of rollers that differ according to both temperature and the type of materials that the roller can be expected to convey. It should be pointed out that rollers of the same class may behave differently where there are differences in the raw material mix of the body being fired. Analogously, rollers of the same category made by different suppliers may have slightly different formulations and thus differences in durability. Rollers, then, are generally divided into the following classes: standard rollers semi-technical rollers technical rollers special rollers. As expense logically varies from one category to another, it has become common practice to use different types of roller in different kiln zones so as to contain installation costs. Raw materials and formulations Manufacturers use raw and semi-finished materials of high quality, which have a considerable influence on final cost. The following tables show standard compositions for the most commonly used rollers: ROLLER COMPOSITION MATERIE PRIME RULLO ROLLER RAW MATERIAL ALLUMINE ALUMINA MULLITE ZIRCONIO ZIRCONIA-MULLITE CLAYS ARGILLE KAOLIN CAOLINO TALC TALCO

STANDARD STANDARD SEMITECNICO o STANDARD COR- STANDARD MULLI- SEMI-TECHNICAL CORDIERITICO MULLITICO TECNICO DIERITIC ROLLER TIC ROLLER /TECHNICAL 50 55 60 65 50 55 10 - 15 10 15 10 20 10 15 10 15 15 20 10 15 5 10 -

Tab. 5.

236

Firing

CRYSTALLINE PHASES IN THE ROLLERS

Tab. 6.

Firing the rollers at different cycles and temperatures causes crystalline phases to form in different quantities and qualities, thus determining the characteristics of the finished roller. The most common causes of roller breakage Rollers essentially break because of: Mechanical problems Being made of ceramic, rollers are rather fragile and mechanical stresses such as knocks or excessive pressure can damage them even when new. Hence they must be handled with care, starting with their removal from the packaging. Another task requiring due care and attention is cleaning: this is done using a special machine, which needs to be adjusted accurately so at not to stress the roller. There is also a risk of breakage when the roller is inserted in the kiln: if it has not been thoroughly dried the sudden release of water could seriously damage it. Fatigue As mentioned above, commercially available rollers differ in terms of mineralogical constitution and maximum working temperature. Roller manufacturers supply tables that illustrate these limits: yet temperature is really no more than an indicator as to operational limits as the latter also depend on roller load, kiln characteristics and the nature of the material being fired. Mechanical stress on the roller can be calculated by way of the following formula: σ (N/mm2) = M/W where M = (P1 L/8) + [P2 (2L-I)/8] π De4-Di4 W = 32 De M = static moment (Nmm) W = modulus strength (mm3) P1 = roller weight between supports (N)

(

)

237


P2 L I De Di

= = = = =

load weight (N) distance between sample holders (mm) load width (mm) external diameter (mm) internal diameter (mm).

Chemical aggression This is the main cause of early roller breakage and can occur inside the kiln during routine operation, when rollers are extracted for cleaning purposes, in the event of an emergency stop or during a scheduled maintenance stop. The damage generally takes the form of a longitudinal crack a few tens of centimetres long or transverse cracks which separate the roller into several cylinders, each 15-20 cm long. Breakage usually occurs in a kiln section some 10 m long, located between the preheating zone and the start of the firing zone, where temperatures range from 680 °C to about 880 °C and salt deposits may be observed both on the roof and the walls. The composition of these salts, mostly K2SO4 and K3Na(SO4)2, indicate the presence of alkaline and sulphur vapours in the kiln atmosphere: the former originate from the material being fired while the latter have their origin mainly in the fuel. In-depth studies where the raw materials were clays with a high soluble salt content, made up of sodium sulphate and lesser quantities of calcium sulphate and complex potassium and sodium salts (SO3 = 0.8%), have been carried out. The rollers were attacked by the alkalies, mainly potassium, resulting in the formation of crystalline phases such as leucite, kalsilite and sanidine, totally absent in the rollers before use. Some theories sustain that alkaline aggression occurs when the potassium sulphate is deposited on the exposed surface and the pores of the rollers where it reacts with the mullite contained therein as follows: 3Al2O3.2SiO2 + 3K2SO4 + 10 SiO2 à 6KAlSi2O6 + 3SO3 mullite leucite Precautions Under normal working conditions roller duration can be maximised and performance optimised simply by observing the following. Always dry the rollers in as uniform a manner as possible before introducing them into the kiln. It is advisable to dry them in the pre-kiln as placing them above or next to the kiln is inadequate. To prevent moisture infiltration always store rollers in a warm, dry place, never outdoors. To ensure homogeneous heating along the entire length of the roller, insert it into the kiln module as fast as possible; insertion must only be carried out by properly qualified personnel. To prevent dispersion of heat via the kiln exterior or when it is necessary to degas the roller itself, it is good practice to plug it, at one end only (drive side), with ceramic fibre. 238

Firing

– Hot roller rotation speed during extraction and insertion should be as close as possible to routine running speed. After extraction the roller should be kept turning for a few minutes. – Extracted rollers must not come into contact with cold metal parts; they should be handled with tools insulated with ceramic fibre. To stop rollers, especially technical rollers, cooling too quickly, they should be blanketed with ceramic fibre. – In maximum temperature zones, rollers removed for cleaning purposes should be replaced with new ones. Cleaned rollers should always be re-inserted towards the beginning of the firing zone, taking the place of still-clean rollers that can then be shifted to maximum temperature zones. Ideally, the temperature of the roller should not drop below 500 °C during these tasks. The chemical aggression caused by the “chimney effect” inside the hollow roller interior can be attenuated by plugging it with ceramic fibre. – To minimise roller cleaning and limit (as far as is possible) the formation of crusts, the rollers could be engobed. The back of the tiles, instead, must be engobed. The following types of engobe may be used. Back engobe for carbonate base porous tiles and porcelain tiles (can also be used for roller engobing): ANHYDROUS ALUMINA 80.0% KAOLIN 20.0% Tile back engobe for vitrified tiles (only to be used for tiles, never rollers): ANHYDROUS ALUMINA 12% 88% MAGNESITE (MgCO3) Universal engobe for rollers: BROKEN ROLLERS KAOLIN – – – –

80% 20%

N.B.: Preparation and application methods are similar for all types of engobe: Grinding residue: 1.0% on 45 micron mesh Glue: 0.3% of CMC Water: 70-90% Application density: 1100-1200 g/litre

Problems caused by rollers The question of dirty rollers aside, it should be pointed out that even clean rollers can cause problems regarding the feed of the material through the kiln. Firing zone The material moves forwards as illustrated in fig. 13.

239


Cooling zone The material is conveyed as illustrated in fig. 14, or worse still, overlapped. The problem can be resolved by acting either on the firing zone or the cooling zone. Firing zone Here, two kinds phenomena can slow down the material in the middle of the kiln with respect to the material at the sides. The first is a natural, temperature-related effect where pyroplasticity causes the roller to sag. Even a variation of just a few tenths of a millimetre will slow down product feed in the centre of the kiln. This behaviour worsens at higher temperatures. The other phenomenon stems from the way in which the rollers themselves are manufactured; the rollers are fired in a suspended, vertical position. At high temperatures this arrangement can cause the material to flow downwards and give a roller cross section of the type illustrated in fig. 15. The roller is then cut (as illustrated by the dotted lines) to obtain the desired length. The difference in central tile speeds in the firing and cooling zones can compensate for each other and produce a more even line of tiles at the exit. If the new roller arrangement still fails to produce the desired results (i.e. improper tile feed), biconical rollers which maximise peripheral speed at the centre and reduce it at the sides can be used. These rollers should be used in sets of 5-8 pieces per module towards the firing zone, for a total maximum of 30-40 rollers, starting from the zone where central slowing is first observed.

Fig. 13. Material feed in the firing zone.

Fig. 14. Material feed in the cooling zone.

240

Firing

Fig. 15. The ceramic roller manufacturing process.

Cooling zone With regard to direct air cooling a situation of the following type may occur (see fig.16). Differences in temperature above and below the rollers cause their deformation (bending upwards), thus creating an effect opposite to that observed in the firing zone. It follows that the tiles will then advance faster at the centre of the kiln than at the sides. Homogenising cooling above and below the rollers should limit the problem. T2

Where T2 >T1

T1

Fig. 16. Roller deformation in the cooling zone.

241


242

Sorting, packaging and palletizing lines

Chapter VII SORTING, PACKAGING AND PALLETIZING LINES

Introduction Before analysing the machines and devices used in the sorting department it is important to point out the following: it is located at the end of the actual tile manufacturing process, after the tile has already acquired all its physical and aesthetic characteristics and does not in any way influence these characteristics as it exists only for control and classification purposes. The sorting department allows the product to be sub-divided and packaged on the basis of both qualitatively objective and subjectively settable parameters according to the ceramic companys position in the market. In other words, the parameters which form the basis of tile classification have much to do with commercial considerations that vary from one company to the next, thus making them relative rather than absolute. For a long time much of the work was manual, being entrusted to the skill of the individual because of the difficulties of designing and building reliable, efficient artificial vision systems; in recent years the latter have been the focus of intense technological research, with new image acquisition systems and ever-more sophisticated sensors being developed. Hence it is extremely difficult to forecast how sorting will develop in the near future. Analysis and classification of tiles Visual checks: to highlight tile defects to allow sub-division according to shade. Dimensional and geometric controls allow: sub-division according to size classes qualitative analysis of shape defects. It is thus necessary to distinguish between a defect (i.e. that changes the finished tile price) and dimensional or colour differences; the latter do not actually alter the value of the tile provided sorting produces lots or packs of similar shade and size.

243


Further details can be obtained by consulting the relevant international standards (ISO 13006) and the section on finished product classification (see fig. 1). Sorting department configuration Far from consisting of just one machine for a certain product size, type and output rate, the sorting department consists of a versatile series of devices. The following paragraphs analyse the characteristics of the main machines used in sorting, packaging and palletising. Feed lines Connection to upstream machines may consist of either a direct link to the kiln(s) or to fired material storage systems; in both cases the product is fed by conveyors which do not necessarily provide a consistent flow, so at the sorting infeed point, other devices modify this into one which is smooth and uniform, eliminating discontinuities or non-uniformities and producing a continuous, homogeneous gapfree tile flow. Achieving this is essential, as any glitches in the flow will if the sorting rate is to be maintained need to be compensated for by increasing the speed of the tile conveyor, with obvious problems for the human sorter who consequently has less time to examine the tile. Furthermore, the feed line usually features other devices such as pressure wheels to remove weak tiles (usually at the kiln outlet), brush and fan cleaning systems, tile turntables (to position rectangular tiles correctly) and vertical compensators to manage brief sorting line stoppages. Visual inspection station The visual inspection station (fig. 2) is the zone in which qualitative tile analysis is carried out by one or two operators. The tiles flow through this zone as a continuous, gap-free, uniform carpet (see above paragraph). Good illumination (incident and/or diffused lighting) allows the inspector to pick out defects. At this stage the main aim is to identify those caused by production errors (black points, glaze droplets, chipping); scanning for defects is accompanied by analysis that aims to divide the tiles into groups of consistent shade. This allows manufacturers to supply packs of near-identical tiles that produce the required aesthetic results when laid. After viewing the tiles the inspector marks them so that they can subsequently be divided into uniform stacks. They generally use fluorescent marker pens (the mark subsequently being read by a U.V. photocell) or virtual systems which use encoder-controlled shuttles or magnetic turntables. With the latter system the tile is allocated a code and tracked as far as the stacking station, where it is then placed in the code-related stack. However, continuous observation and identification of defects or shade variations is no easy task as inspector performance inevitably varies over time and maintaining consistency is difficult. 244


Aesthetic criteria 1st QUALITY The surface of the tile must be completely intact; aesthetic evaluation must be carried out on a sample of at least 1 m2 or a minimum number of tiles as established by standards. It must involve observation of the surface of the tile at a distance of 1 m under 300 lx lighting conditions. The result is expressed as the percentage of defective tiles.

1st Quality requisites Percentage of tiles with defects (max 5%).

2nd QUALITY Must involve a test sample of at least 1 m2 or 30 tiles. Evaluation must involve observation of the surface of the tile at a distance of 2 m under 300 lx lighting conditions.

2nd Quality requisites Percentage of tiles with defects (max 5% of 3rd quality).

3rd QUALITY Tiles of 3rd Quality are defined as those that do not meet 1st and 2nd Quality requisites.

Functional criteria

1st Quality requisites The tiles must satisfy 1st Quality requisites as defined by relevant standards for individual product classes. 2nd Quality requisites Dimensions Tiles must satisfy requisites as defined by standards for individual product classes, with a maximum permissible deviation of 25% with respect to the tolerances indicated by the above standard. Physical and chemical properties The tiles must satisfy the requisites of the specific standard. 3rd Quality requisites Tiles of 3rd Quality are defined as those that do not meet 1st and 2nd Quality requisites.

Fig. 1. Sorting criteria as defined by standards.

245


Fig. 2. Manual inspection station (bench).

Given the need to employ different workers and considering the vast number of variables, classification of output according to parameters that are not always objective but of great commercial importance is extremely delicate. This human factor is highlighted when an increase in output makes it necessary to employ two observers at a sorting station; here, the product is fed along two lines, each assigned to a different worker; the problem here is that their separate, subjective evaluations are blended at the packaging stage. Manual defect and shade sorting, then, is an essential yet tricky task yet one for which a reliable automated alternative has yet to found. However, human error can lead to mistakes or slow down production: the output rate of a sorting line is, in fact, not so much dictated by automated cycle speeds but, rather, by human limitation. It should thus come as no surprise that the ceramic industry is following the development of automatic tile control systems with keen interest. Dimensional control Immediately downstream from the sorting bench comes the automatic dimensional and geometric control unit, consisting of two different sections (fig. 3): a device for the control of size and geometric defects a flatness control device. The former checks tile dimensions and classifies them within ranges set by the machine operator: variations in tile size do not, in themselves constitute defects and do not necessarily alter the technical specifications or commercial value of the product. The important thing is to sub-divide them so that tiles of significantly different size are not laid next to each other. 246


Fig. 3. Size and planarity control unit.

Optic fibres and digital linear sensors take a snapshot of the tile, which is then compared to a master sample; precision is in the order of 0.1 mm and repeatability is good, as is stability, despite the movement generated by the conveying system. Other measurement solutions use photodiodes and encoders that give time/speed readings, which are then interpolated to calculate a distance: precision is again in the order of a tenth of a millimetre. Measurements are carried both longitudinally and transversely (fig. 4). Measurements X1, X2 and X3 identify the sides parallel to the direction of tile travel, while measurements Y1, Y2 and Y3 concern sides perpendicular to it. Thus not just one measurement but a whole series of them are made; these are processed to attribute a single size value to the tile so that it can then be assigned to a group (i.e. a calibre class) of similarly sized tiles. The successive approximations method This method is characterised by use of an algorithm that ensures good immunity from disturbances linked to measurement grouping (fig. 5). It is based on the setting of central sizes that represent the central value of a size class. Calculating the difference between the central sizes and the X1-2-3/Y1-2-3 readings leads to compilation of a table, the concept of which is illustrated in tab. 1: the top row shows the central sizes set while the first column shows the measured axis size. The other columns show the difference between the central size and the various X or Y measurements. Hence each field shows a deviation; the maximum absolute value in each column (i.e. maximum deviation from the corresponding central size measurement) is considered (bottom row) and the minimum of such maximum variation identifies the tile class. 247


Fig. 4. Measurements made by the size checking unit. Central sizes

Tile

12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123 12345678901234567890123

Test method

Fig. 5. The central sizes dimensional evaluation method.

Column maximum

Tab. 1: X1 = measured size value, C1 = central value of set size.

248

Size class


An example with actual numerical data is provided in tab. 2. Tenths of millimeter

Column maximum

Tab. 2.

Here, four different size classes have been set (top row). The minimum out-of-size value is 12, the lower central size is chosen and the tile is thus assigned to C2 (4493). Average side method (as per EN 98) The advantage of this method lies in the extreme intelligibility of assignment. As above, size classes are defined, this time as simple dimensional intervals (fig. 6) with minimum and maximum values for each size class. The size of the tile is the average of the tile size measurements: = (X1+X2+X3+Y1+Y2+Y3)/6. The class size of the tile is simply given by the interval to which the abovecalculated value belongs. Maximum side method Like the previous test method, this one is based on the definition of thresholds (minimum and maximum) and assignment of detected values within certain intervals: in this case that value is simply the largest of the X1-2-3/Y1-2-3 measurements. Modern control units allow the user to set any one of the above test methods as desired. Flatness defects are detected via telemetric sensors, calibrated with a master plate to provide an ideal reference: its comparison with the actual tile surface reveals defects and deviations (both statistical and local) and the tile is classified accordingly (i.e. deviation from master plate planarity constitutes a defect). The limits of deviation for each class of tile can be set by the operator. The figure below illustrates the most common defects detected by the abovedescribed unit (fig. 7). 249


Tile

12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890 12345678901234567890

Test method

Size

Size class

Classification threshold

Determination of the size class requires the introduction of classification thresholds with the following characteristics: – thresholds must form consecutive intervals that determine the size class. – the number of classification thresholds is equal to the number of programmed size classes minus one. Size class

•

Classification threshold

Determining size: The size of the tile is calculated as the average of the tile size measurements: size = (X1+X2+X3+Y1+Y2+Y3)/6

All that needs to be done is to assign the calculated size to the relevant interval. As the table shows, size classes C1 and C6 are open-ended intervals: all tiles smaller than S1 and larger than S5 will be class C1 and class C6 respectively.

Fig. 6. Average side method.

Another recently implemented test involves measuring the squareness (orthogonality) of adjoining sides. This can be measured directly as per the corresponding ISO standard (i.e. by measuring the extent of deviation from the ideal position for each side see fig. 8).

250


Different size sides

Positive/negative edge curvature (lunette)

Squareness (orthogonality)

Upturned or downturned corners

Diagonal and edge concavity

Diagonal and edge convexity

Warpage (global coefficient)

Deformation range on sides

Diagonal concavity

Diagonal convexity

Fig. 7. The most common ceramic tile dimensional defects.

+/- readings

Fig. 8. Apparatus for measuring size, squareness and edge shape as per standards.

251


Other devices give a squarerness evaluation by calculating the difference of the diagonals, assuming that the edges are parallel. This last quality check is particularly useful downstream from the edge grinding units used to rectify porcelain tile sides; since porcelain tiles are laid without any inter-tile gap precise squareness is essential. Automatic tile inspection Introduction For some years now industrial automation has been synonymous with improved product quality and savings in terms of time and money. The ceramic floor and wall tile industry is no exception to this rule and has extensively automated just about every stage of the production process. Sorting remains the only stage of that process to remain almost exclusively manual, with inspectors classifying tiles according to shade and defects. Workers are usually seated at an inspection station upstream from the actual sorting line, where they can examine all the tiles moving past them. A panel in front of the selector shows samples of the current product with different colour shades against which the tiles are compared and classified. At the same time the worker also has to pick out defects which will influence the quality category. On the basis of his/her judgement, the worker then marks the tile with a code: this is read by a downstream sorting machine which automatically packages the tiles accordingly. Hence inspection and classification outcomes are entirely dependent on the workers skill and intelligence. To start with, it is evident that the operators capacity to perform this monotonous task lies in his/her ability to focus on anomalies and defects while ignoring that which is correct. The human intelligence factor is often a discerning one, allowing identification of small and sometimes new defects, even on tiles that have complex decorations and/or surfaces, where the distinction between what is acceptable and what is a defect is far from obvious. Yet human control is also inherently problematic, not so much because people are incapable of detecting faults but, rather, because it is difficult to maintain concentration (i.e. objectivity and performance) over time. The introduction of automatic shade control and defect detection systems has the evident advantage of eliminating the above problems while raising quality and lowering costs. This explains why, since the early 90s, several specialist companies and consortiums have been focussing their efforts in this direction. The goal that these companies have set themselves is an ambitious and difficult one, as the current trend is towards tiles that mimic the aesthetics and performance of natural stone, or tiles of a deliberately random, heterogeneous pattern and texture that are often difficult to distinguish from actual defects. An automatic tile sorting machine is obviously intended to mimic the tasks performed by the inspectors, who, by careful observation, extract global information 252


about the tile to identify the shade, and local information, used to track down defects. As the tiles pass by the inspector needs to maintain a certain level of concentration: if anomalies are detected concentration levels are suddenly heightened. However, in some cases, reliance on sight alone can trick the worker and distort his/her performance. Particles, dust, detritus or drops of liquid can sometimes fall onto the tile along the conveyor. At first such tiles appear sub-standard, but more accurate inspection will show that the defect can simply be wiped off. In this event the inspector usually counter-checks by running his/her gloved hand over the surface of the tile. Nevertheless, the most important human faculty in tile inspection remains sight, occasionally helped, in borderline cases, by touch. This visual information allows the inspector to evaluate the tile and assign it to a certain class. Automation of this process, then, involves imitating these human senses and replicating certain intellectual functions: hence the employment of vision systems that see the objects being examined and subsequently take the necessary sorting decisions. A standard vision system uses cameras, which, suitably piloted, can, in fact, acquire an image of the item being inspected and then send it to a processing unit for evaluation. However, standard vision systems cannot emulate a sense of touch and can thus easily be led astray by tiles that are dirty when they pass under the cameras. As this can lead to errors and wastage, a vision system requires installation of a tile cleaning unit upstream from the actual point of inspection. The complexities of designing and building an efficient, automated vision system the ultimate goal of which is to replace the human operator and reproduce his/her decision-making skills requires expertise in fields such as optics, image processing and artificial intelligence. Yet despite these difficulties, automatic tile sorting systems have been available for some years now and numerous ceramic production plants have installed them to good effect. Tile vision systems: characteristics The first task of any vision system designed to replace a human operator is, then, to replicate the human faculty of sight with cameras. High resolution video cameras are required in order to acquire the detailed images needed for accurate detection of often small defects on sometimes very large tiles. The second step is the decisional one, carried out by a processor; here, through utilisation of complex algorithms, the machine attempts to imitate the human decision-making process. The third step is to communicate the processing outcome for every single image to the automatic packing devices. To do this the inspection system marks 253


the tiles with codes similar to those applied by human sorters which are easily recognisable by standard reading machines on the line. Whatever the employed technical, optical and image processing solutions, nearly all tile inspection systems have several aspects in common. Since tile classification takes into account both shade differences and defects it follows that inspection systems are generally separated into two separate modules. This distinction stems from the fact that these different problems need to be dealt with in different ways. An understanding of how inspectors evaluate shade differences and defects requires an analysis of the lighting conditions under which the tiles are observed. Similarly, an automatic system needs to acquire, via the camera, images which contain information useful for identification of certain characteristics and/or tile surface defects. Hence a proper combination of lighting conditions and optics is essential: a wrong choice can seriously compromise the efficiency of the vision system. During shade control, the inspector observes the tiles from above and lighting is diffused. This overhead view allows him/her to evaluate overall shade while picking out details that constitute decoration defects. Automated systems generally employ the same principle. The shade analysis module generally features diffused lighting and the camera is mounted perpendicularly over the tile conveyor (fig. 9). This set up is largely explained by the desire to reproduce the same conditions that work so well for the human eye. Classification of tiles according to shade is a colorimetry problem (i.e. measurement of tile colour and subsequent identification of shade as belonging to a defined class). In processing the camera-acquired image, a tile sorting vision system must also be able to replicate the human faculty of colour perception. Similarly to the human eye, automated systems perceive colour via the three parameters of hue, saturation and intensity, which define the so-called HSI space representing colour. In speaking of colour we are really speaking of hue. Hue allows us to distinguish between colours such as green and yellow. Hue is nothing more than the perception of colour experienced by an observer when he/she is exposed to different wavelengths of reflected light. Image perception begins in the eye (fig. 10). The muscles in the eyeball modify the lens of the cornea to focus the image on the retina (i.e. the membrane lining the interior of the eyeball). The retina contains millions of photoreceptors, of which there are two types with different image perception functions. These photoreceptors convert brightness and colour information into impulses that pass along the optic nerve to the brain where they are perceived as images. Light-sensitive photoreceptors which allow you to see even under dim lighting conditions are called rods. These do not contribute to colour recognition: that task, 254


Fig. 9. Automatic vision system: illumination method.

Fig. 10. Human eye structure.

instead, is performed by the receptors concentrated in the fovea, the cones. There are three types of cone (known, in simplified terms, as red, green and blue), each of which responds to different wavelengths of light. The intensity of cone response vis-à-vis wavelength is illustrated in fig. 11. Note that the response peak of the cones does not always correspond to the colour with which they are identified. Figure 11 shows that the prevailing perception for wavelengths between 430 and 480 nanometres is blue, between 500 and 550 nanometres green and over 610 nanometres red. The concept of saturation refers to the degree of purity of the colour (i.e. nonmixing with white light). A non-saturated colour appears wan and faded while a saturated colour is intense and vivid. For example, red is a high-saturation colour 255


Wavelength (nm)

Fig. 11. Wavelengths of principle light components detected by the cones in the human eye.

and pink a very low one. For pure colours, that is, those that do not contain white light, the degree of saturation ranges from 0 to 100%. Intensity, instead, expresses the quantity of light reflected and perceived by the observer. Analysis of these parameters, then, allows recognition of shade and identification of any shift away from a shade. The colour video cameras used in tile inspection vision systems supply images that are expressed as an RGB (red, green, blue) mix. These need to be converted into HSI (hue, saturation e intensity) format as it is the latter which lead to accurate identification of the shade. More importantly for the tile manufacturer, analysis of local variations in hue, saturation and intensity makes identification of colour defects such as stains, impurities or glaze droplets possible. An automatic shade sorting system has many advantages compared to manual sorting. For example, a worker is usually able to identify even slight differences in colour between two tiles that appear within his field of vision simultaneously, yet not always able to judge or remember the colour of the tile in absolute terms. An automatic system, instead, ensures consistent shade identification judgement. This consistency is also guaranteed by the fact that many of the available systems feature calibration mechanisms which ensure that tiles are seen by the cameras under conditions that allow constant reproduction of colour over time. This is why automatic shade sorting is characterised by a stability and objectivity unattainable with manual sorting. On the module dedicated to the detection of surface defects, instead, the best results are obtained by viewing the tiles under direct reflection. Fig. 12 illustrates how the light emitted by an artificial source normally (i.e. 256


Fig. 12. Surface defect and the consequent light scatter effect.

where there are no defects) reflects off a tile surface and how the light is, instead, scattered by a surface defect. When it comes into contact with a defect the light is diverted from its normal reflection route and only a fraction of it is reflected into the camera lens. Thus an optical anomaly in the acquired image is used to identify a shape surface defect. Once again, machine developers have attempted to mimic the mechanisms employed by human operators, where sorters perceive the presence of surface defects as anomalous reflections of the light used to illuminate the tiles on the sorting bench. A selector will often shift his/her viewpoint of the tile so as to catch defects by looking at the reflected light from different angles. From the above it follows that automatic systems need to adopt two different illumination-image acquisition modules: one for shade analysis and the other for defect analysis. The defects module uses simple black/white high-definition cameras so as to identify imperfections even small ones such as lumps and pin-holes even on large tiles. Where defect classification is done manually the results depend on the operators skill and speed. The ever-pressing need for higher output rates means that selectors are very often forced to work under conditions that compromise their efficiency. For example, the product is often conveyed across the sorting bench at a speed that gives them very little time to analyse shade and discover defects (especially on smaller tiles). Moreover, the tiles on the conveyor belt are often very close to each other and the operator sees them pass by without any intervening gaps as if the control process were a continuous one; consequently, he/she is unable to inspect the tile edges. Even where tiles are properly separated when they reach the sorting bench it is still impossible to inspect all the edges. Some automatic systems get round this problem by using special optical arrangements to acquire images which, through utilisation of special algorithms, allow identification of glaze shrinkage, chipping and flaking at the corners, even where such defects are small and can escape the attention of a hard-pressed selector. These systems also help manufacturers pick out damage invisible from above which does not 257


compromise the aesthetics of the tile but can compromise bending strength (i.e. small cracks on edges). Just as workers are put through a training period before they are allowed to inspect specific products, so automatic systems have to be taught how to perform certain tasks. Machine training consists of acquiring a set of representative samples of a certain shade; subsequently the machine can automatically extract those characteristic parameters that identify the product in question and which will be used by the image processing algorithms during inspection. The automatic sorting machine supervisor has the task of setting shade and defect tolerances (minimum size, minimum contrast, declassification thresholds etc.). It is also possible to monitor any drifting away from the shade and set the machine so that it issues a shade warning when a certain threshold is reached, thus allowing the supervisor to intervene. The entire system can be controlled via a user-friendly graphic interface that does not require personnel with specialised computer skills. Installing an automatic sorting system From the above it is clear that the best place to install an automatic tile sorting system is on the sorting line itself in place of the traditional selection station, upstream from the packaging machines. If there are some products for which automatic sorting is as yet unsuitable then both automatic and manual sorting units will be required. In this event the manual inspection station should be installed downstream from the automatic system so that all sorting operations can be transferred to manual (machine off) or the unit can assist the manual workers (machine on). This arrangement allows the inspector to see the marking deposited by the automatic system and correct, where necessary, any evaluation errors. Whatever the layout, insertion of an automatic system on an existing sorting line must take into account the available space. Compatibility of system dimensions with available space is obviously less problematic where the system is part of a new sorting line. Generally speaking, nearly all automatic sorting systems have their own tile feed belts because the tiles must flow under the image acquisition units under specific conditions (i.e. at constant speed and as undisturbed as possible by the vibration produced by the rest of the line). Definition of relative speeds between infeed/outfeed conveyors and the automatic sorting system conveyor must ensure that the tiles are spaced apart properly so that the cameras can acquire a separate image for each incoming tile. Separation can be achieved mechanically by running the sorting system conveyor belt faster than the infeed belt so that tiles on the upstream conveyor can still be conveyed up against each other but as soon as they reach the sorting system belt they are accelerated to produce the necessary gap. 258


Sorting system conveyor speeds may be as high as 40 metres/minute (with peak speeds of 60 metres/min on systems with particularly fast image processing units). Outfeed belt speed should be set so as not to brake the tiles exiting the sorting system. A mechanical alignment unit is installed at the sorting system conveyor belt inlet: this ensures that the leading edge of the tile passes under the image acquisition unit at 90° to the direction of travel. If tiles pass under the cameras at the wrong angle an additional image processing stage is required to shift the tiles back to their orthogonal position. From a software viewpoint this is a serious complication and inevitably slows down inspection rates. The pre-sorting approach section should be equipped with guides that channel the tiles into the actual alignment unit; the infeed conveyor should be level with the sorting system conveyor for a distance equivalent to some 3-4 times the size of the tile. Evaluation of the space required for installation of a sorting system on an existing line should also take into account the space needed for the positioning of a cleaning system with brushes and blower/suction units; this will need to be installed upstream so that the tiles are already clean when they pass under the image acquisition unit (remember that dirt and dust can modify inspection results). Ceramic production plants are generally high temperature environments: since electronic devices are generally destabilised by high temperatures, an air conditioning system should be installed to keep temperatures in the automatic sorting area below 40 °C. Defects detected by an automatic sorting system Both surface defects and shade errors are generally classified in terms of type and severity. Within the context of a single image a defect is usually detected as an anomalous contrast and/or alteration in the chromatic content of the image. Yet there are limits to this defect detection capacity. One such limit is the minimum size of the defect, dependent on the resolution with which the image is acquired. Each image is a matrix of pixels (or points), each of which corresponds to an actual physical dimension on the cameras field of vision. If, for example, the field captured by the camera in one direction (e.g. transverse to tile motion) is 600 mm long and the corresponding number of pixels is 2000 then the image resolution in this plane will be 0.3 mm. The same reasoning can be applied to the length parallel to the direction of motion. On the basis of these two resolutions every pixel corresponds to a certain surface area. Logically, defects any smaller than this area cannot be seen in the image. In general it can be said that a defect, in order to be detectable, must cover an area of at least 0.5 mm2. Another limit is represented by defect contrast. The higher the contrast of the 259


defect the more likely it is to be detected. Reliable defect detection is therefore linked to the contrast that such defects generate on the image under optimum lighting conditions. Moreover, the more complex the surface of the tile the higher the complexity of the algorithms needed to detect defects: for example, heavily structured tiles may have rough and/or smooth zones deliberately intermingled with matt and/or gloss zones. The same argument holds for the detection of colour defects where tile decoration has an extensive chromatic component mixed in with random elements. Just as size and contrast are key identification factors, so is the type of defect. Difficulties can arise when only a part of the defect, not all of it, is detected. This may happen when only a portion of it generates sufficient contrast. To obviate this difficulty special algorithms which allow the defect to be reproduced in its entirety are used. Once the tile has been inspected and all the detected and identified defects have been evaluated, the system assigns the tile to a category on the basis of the userapplied settings. Automatic sorting systems: performance and advantages From a performance viewpoint (i.e. number of tiles inspected per minute) an automatic inspection system guarantees better results than those provided by manual solutions. Size (side parallel to direction of travel, in cm) Vision system Operator on single bench

20 125 120

25 115 96

30 100 85

33 90 75

40 80 65

45 60 52

50 50 45

The average values in the above table compare manual and automatic sorting system performance. Given the continuous progress in both the hardware and software fields the problem of inspection speed is negligible. Despite the complexity of image processing operations, utilisation of more and more powerful microchips in parallel with artificial intelligence algorithms allow high-speed analysis of every single tile. The widespread availability of such enormous computing potential means that there are now many systems capable of examining even complex tiles. The product types which best lend themselves to automatic inspection are: glazed single and double fire tiles with repetitive, random, marble-like patterns on smooth or structured surfaces porcelain tiles (glossy, polished and rough) natural stone.

260


The advantages offered by an automatic tile sorting system are: replacement of manual sorting inspection and quality control of all tiles a continuous kiln-to-packaging line reduced tile shade stock requirements guaranteed objectivity/repeatability over time, unattainable with manual sorting 24 hr/day sorting significant reduction in sorting staff requirements lower fixed production costs.

Because automated sorting can be applied to an extensive product range, with all the consequent advantages, the ceramic industry has obviously shown great interest in such systems; their commonplace installation would inevitably revolutionise what is still the most labour-intensive part of the production process. Automatic sorting systems: other uses When a ceramic plant begins a production run the ideal outcome would, of course, be production of 1st Quality tiles only. However, the reality of manufacturing is somewhat different. For reasons that are evident, the tiles that reach the sorting department may well be 2nd Quality, 3rd Quality or even rejected. The higher the percentage of low quality tiles the lower the earning capacity of the plant: lower earnings, wasted raw materials and energy, waste disposal and stock management problems all combine to reduce profitability. The more frequently this happens the greater the need to identify and eliminate the causes of defects at source. Of course, end-of-line systems do not actually solve problems of low plant efficiency: by the time the tiles reach them the production process is already over. However, installing automatic sorting systems downstream from specific stages of the production process lets manufacturers identify errors before they can compromise the final outcome. Hence a sort of distributed quality control system can be built up along the entire production line. For example, an automatic sorting system might be positioned downstream from the drier to check dried, unfired tiles or downstream from the first kiln in a double fire plant. Such systems must be able to detect mechanical defects such as cracks, corner chipping, edge imperfections and surface faults such as contamination, sunken and raised areas, and remove such sub-standard pieces from the cycle. Consequently, the manufacturer avoids wasting any further energy/raw materials on tiles that would certainly have been rejected at the end of the production process. Note also that the sooner the defective pieces are identified, the greater the chance of recycling them: an unfired tile is 100% recyclable, while disposing of a rejected, finished tile is time-consuming and costly. By analysing defects at different stages of the production process manufacturers can track down their causes and increase production profitability by making corrections and eliminating them at intermediate stages of the production cycle. 261


Classification and stacking When it exits the control area every aspect of the tile has been examined and, whether by manual or automatic means, has been classified according to shade and quality (the order of classification might be shade A 1st Quality, Shade B 1st Quality, Shade A 2nd Quality, shade B 2nd Quality, 3rd Quality, Reject). The dominant shade is not usually marked by the operator, only those shades which deviate from it. A photocell upstream from the stacking zone, reads and decodes the markings. The size measuring device assigns the tile a size and a quality class. The planar device assigns the tile a quality class. At this point the control and management station has all the necessary tile classification information: the final category will depend on the more negative of the evaluations: for example, a tile deemed to be of 1st Quality shade and 1st Quality flatness that has, however, failed to meet size criteria, will be rejected; likewise, a tile that is perfect in terms of size and flatness will be rejected if declassed because of surface defects. The classified tile is conveyed to the stacking zone by two lateral feed belts (capable of carrying different sizes). Here, the tiles are sub-divided according to the quality classes in the table above. All tiles of the same class are directed to the same stacker, which extracts them from the feed belts and produces a stack that can then be sent to the downstream packaging station. Stacking generally involves one of two different methods: direct stacking indirect stacking. In the former the tile is usually extracted from the belts by a pair of pneumatically-driven push pads that deposit it on the underlying stack being formed; the tileto-stack drop gap is, at this point, no further than the thickness of a single tile (see fig. 13). Coordinated movement of the pneumatic cylinder and the stacker plate ensures both maximum productivity and a soft tile landing, in that the relative speed between tile and underlying stack is minimized to prevent any detrimental impact or scraping, especially for delicate materials. Precision of movement is provided by sophisticated software which eliminates the need for any interposing buffer between stack and tile. The latest direct stacking machines have replaced DC motors with standard 3-phase motors controlled by an inverter; this solution maintains high performance standards yet provides greater simplicity of operation and more readily available spare parts. In indirect stacking the tile is still removed from the belts by (2 or 4) pads operated by a pneumatic cylinder, but the tile is not dropped directly onto the stack from a minimal height. It is, instead, handled by intermediate loaders which cover the gap between extraction height and the placement height a gap that is much wider than 262


WALL

FLOOR 1

FLOOR 2

FLOOR 3

FLOOR 4

+

+

+

+

1st Quality – shade A – size 1 1st Quality – shade A – size 2

+

+

+

+

1st Quality – shade B – size 0 1st Quality – shade B – size 1

+

+

+

+

+

+

+

+

Super 1st Quality – shade A – size 0 1st Quality – shade A 1st Quality – shade B 1st Quality – shade A – size 0

* *

+

1st Quality – shade B – size 2 1st Quality – shade C – size 0

+

+

+ +

1st Quality – shade C – size 1 2nd Quality – shade A

*

+

+

2nd Quality – shade B 3rd Quality

*

+

*

+

2nd Quality - shade A – size 0+1 2nd Quality - shade B – size 0+1

+ +

2nd Quality - shade C – size 0+1 2nd Quality - shade A – size 0+1+2

+

2nd Quality - shade B – size 0+1+2 3rd Quality – shade A + B – size tol. 3rd Quality – shade A + B + C – size tol. 4th Quality – shade a + B – size tol.

+

+

+

+

+

+ +

+

Possible sub-division criteria.

Fig. 13. Stacking.

263


with direct stacking. Drive units are generally of the DC step-by-step type. Another characteristic of these devices regards the number of stacking stations. There may be: lines with dynamic stackers: in this case the number of stackers is equal to the number of quality classes + 2 reserve stackers used during unloading. A turret arrangement ensures that all stations take the various quality classes alternately, thus dividing the work load and spreading wear evenly among the stackers. lines with stacking stations which feature double autonomous movements: once one stack is complete that same station is immediately ready to take new tiles; in this type the number of stations is equal to the number of quality classes. Whatever the methods of tile extraction, stack formation and on-conveyor placement the precise function of the machine is always the same: to gather classification information and proceed with subdivision of the tiles into stacks of the same class that can then be sent to the downstream packaging station. Packaging The final product can be packaged either automatically or semi-automatically; the latter combines both machinery and labour, while the former only uses personnel to monitor and feed the machine with materials. The purpose of the packaging process is to enclose stacks of the same class of tiles in a cardboard container that will hold them together, protect them and makes it easy to handle and transport them up until the time they are laid. For small-medium size tiles the wrap-around system consisting of a series of stack pushers, carton pick-up/carry mechanisms and flap fold/seal devices is generally used. Note that the pack can be wrapped and sealed with the final flap (4point gluing) or can be closed symmetrically along the centre line of the box (5point gluing). Some machines allow manufacturers to adopt both solutions, thus augmenting flexibility and versatility (fig. 14). The wrap-around case has the advantage of producing little scrap and thus keeps costs low, yet has its limitations in terms of tile size and is difficult to use where the wrap-around box blank is longer than 2500 mm. Within those limits, however, it maintains a good degree of rigidity and long-lasting handling stiffness, thus aiding good machine reliability. Producers of medium, large and extra-large tiles are increasingly using the cover packaging method, in which the cardboard covers and wraps the stack from above, leaving its underside exposed. This eliminates the problems associated with case length, yet there is inevitably a higher percentage of scrap (see fig. 15). Wrap-around and cover packaging units may be used separately or together; in the latter case, the manufacturer generally employs a mix of packaging machines that allow him to switch from one method to another, thus making the line suitable for a wide range of tile sizes (figs. 16-17-18). 264


Fig. 14. Wrap around packaging: the 4 and 5-point gluing systems.

CARDBOARD SURFACE AREA INCLUDING SCRAP

CARDBOARD AREA / TILE SIZE GRAPH 900000 800000 700000 636804

COVER

600000 512656

500000 440896 400000 300000

315240 277300

419904

404128 342720

302984

302984

200000 100000 0 150

200

300

400

500

TILE SIZE

Fig. 15. Comparison of wrap-around and cover packaging performance.

265

600

WRAPAROUND BOX BLANK


Fig. 16. Cover packaging.

Fig. 17. Wrap-around packaging.

Fig. 18. Packaging machine.

266


Another important factor concerns the type of glue used to seal the cases: depending on the temperature of the tiles themselves, either cold or hot-glue units are used. The most common of the two, hot-gluing, consists of units which melt the glue and guns which apply it at the required points; adhesion is excellent, the system is easy to keep clean and performance is steady and reliable. However, it needs to be used at temperatures that bring the glue to the softening phase, otherwise good adhesion between the materials being joined is compromised; the limitations of this method are highlighted when packaging is carried out immediately downstream from the kiln outlet and the tiles are still quite hot (maximum 60-65 °C). Under these circumstances cold vinyl glues applied with high-pressure spray guns are used; in this case adhesion times are faster than the previous solution, yet as results are, in any case, inferior to those obtained with hot glue, cold gluing needs to carried out in a pressing tunnel in which the flaps, once sprayed, are squeezed by special devices to ensure good adhesion and thus prevent any subsequent opening of boxes and the resulting downtime. Whatever the sealing method, the packaging machine will provide the manufacturer with a box of tiles that is ready to be coded and palletised. Printing and labelling zone On exiting the packaging machine the box is transferred to the printing zone where it is marked with essential product data, as required by standards. This information is generally printed on a part of the box where it can be read after the boxes have been stacked on the pallet (usually on the side of a rectangular or square box or on the upper side of the box). An ink-jet printer is generally used. Black ink is preferred as it is less likely to fade in sunlight (this is important as boxes may be left in the sun for some time). On-box information generally includes a description of the product, its class (size, shade, quality) and sometimes a bar code that is used either by the ceramic company itself or later on at the distribution stage. Bar code application requires more sophisticated devices: either the code is printed on self-adhesive labels which are then applied to the boxes or special high definition printers apply the bar code on the box directly; as application and, more importantly, reading results cannot be guaranteed, a photocell is usually positioned immediately downstream from the bar code printer for an immediate check: a warning is given where the code cannot be read. Palletizing zone After printing comes the palletizing station. The latter generally consists of a 3-axis cartesian-coordinate robot which picks up the packs and arranges them on pallets (fig. 19). 267


Fig. 19. Palletization robot.

To complete the classification process begun during the stacking phase, all boxes of the same code are deposited on the same pallet. Hence the number of palletizing stations will be equal to the number of product classes, plus a reserve station which allows the process to continue uninterrupted when a pallet is full and needs to be removed with the aid of a manual transpallet or automatic shuttle. Generally speaking, there are two types of palletization unit: those that run on floor rails and those on overhead rails. The former are more widespread as they can be adapted to the changing needs of the ceramic market more easily and at lower cost (i.e. the length of the rails is directly proportional to the number of quality classes on the sorting line). The boxes are transferred to the palletizing zone by conveyors with controlledspeed motors that allow accumulation of packs, synchronisation of different infeed rates and consequently smooth pick-and-place operations. The box is usually transferred the same way up as it was packed (i.e. resting stably on the side of greatest surface area): on reaching the pick-up point the boxes are rotated to the vertical (90°), a more suitable pick-and-lift and palletization position (fig. 20). Note that there is both a need to provide maximum box stability and, at the same time, to deposit the tiles on their edges so that other boxes stacked on top do not damage them: pallet loading may be seen as the art of compromise between these two needs. Hence the two key factors are tile size and pack width. If, for example, we take a pack containing ten 300 × 400 mm tiles, each 10 mm thick, it will be placed on the pallet on its 400 × 100 side to give a pack height of 300. The pack stability equation is empirically given by H (mm) × 0.2 + 40 = (mm of pack width). Thus, in this case, we have 300 × 0.2 + 40 = 100 although other nongeometrical factors (such as the type and quality of the pallet on which the product is to be loaded) come into play. Packs cannot, of course, be deposited 268


Fig. 20. Pack picker.

widthways individually as they could topple. In this case and this generally involves tiles larger than 330 × 330 mm configuration is dictated not only by the palletizer but also the upstream sorting line, on which a divider and stacker unit is installed to produce a stack of tiles twice as high as normal: this is split into two equal parts that are then packaged adjacently and given the same code. Immediately downstream from this point another unit stacks the two boxes and joins them together either with glue or a plastic strap (fig. 21). Another, much less widespread alternative to turning the pack 90° at the fixed pick-up point is to turn it to the ideal palletization position immediately at the start of the roller conveyor on the palletizer station; the palletizer thus moves towards the pack and lifts it only when it is as close to the destination pallet as possible; while this method tends to speed up loading operations it requires a more extensive base to ensure good pack stability during movement in front of the line of pallets (sometimes tens of metres long). Moreover, this logic is unsuitable for simultaneous lifting of packs with different codes as they obviously need to be released in different places. Going back to the pick up and deposit process in a more general sense, the pack is lifted, turned (where necessary) and then deposited on the pallet so that the printed product information and the top of the tile face outwards and is thus visible. Pallet load layout varies as a function of product type, characteristics, size, transport and presentation requirements. This clearly complicated task is managed by the sorting computer, specifically designed for simplicity, speed and user-friendliness (fig. 22). 269


Fig. 21. Divider-stacker unit.

Fig. 22. Visual information display facility.

The above information largely concerns standard sizes. With much larger tiles, sometimes defined as slabs (600 × 900, 900 × 900, 600 × 1200 mm), the packaging and palletizing processes are less standardised. When it comes to packaging, some of these products are packed individually and manually. Subsequently the tiles are laid directly on the pallet; overlaying them would produce stacks of considerable weight requiring specially designed heavy duty palletizers. Note that while palletization may, in many production plants, rely on utilisation of semiautomatic systems, there are plenty of factories where the use of dedicated labour makes sense, thus making all those rotation, tilting and automatic pack coupling mechanisms quite unnecessary. Human intelligence is, in fact, perfectly capable of evaluating, on a case by case basis, just what needs to be done to ensure pack or pallet load stability and should not be underestimated. Data exchange between sorting unit and palletizer usually occurs by way of clean contacts and the triggering of a FIFO (First In First Out) mechanism: each pack exiting the packaging machine is entered into a memory (the FIFO memory), while each pack picked by the palletizer exits that memory (i.e. leaves the FIFO). In 270


other words the palletizer knows, for example, that there are 10 incoming packs, that the first of these is code 1, that the second is also code 1, that the third is code... and so on; every time a pack exits the packaging machine the relative class code is sent to the robot which adds it to the chain, just as it removes it from the chain when it picks and places it. The palletizer must be informed of every change in the memorised chain otherwise loading errors will occur and packs of different characteristics will be mixed together. The limitations of this system show up where the conveying line between sorting and palletizing is particularly long, as this makes it easier for the above-described errors to creep into the process. Such problems can be prevented by controlling the type of material to be palletised directly: packs exiting the packaging machine are marked with an inkjet printer (usually a high contrast spot a few millimetres in diameter) and the spot is then scanned at the pick-up point to provide class code information. Alternatively a bar code containing all the palletizing data can be applied and read by a scanner at the pick-up point. If the latter system is adopted both printer and scanner will need to be of much higher quality than those used with the spot method. Sizing a sorting and palletizing line A sorting department is not a machine, but, rather, a set of devices some of which are indispensable, other optional; they can be configured in innumerable ways to produce the set up which best suits the needs of the manufacturer. The key points to be taken into consideration are product type, output rate and tile size range. Equally important is overall plant efficiency (i.e. full utilisation of machine production capacity, dependent not only on the above factors but also on line configuration and proper training of use and maintenance personnel). In other words, while it is important to configure the devices appropriately it is also important to utilise them in a manner that ensures good overall plant efficiency, with as little downtime (caused by maintenance problems and improper use) as possible. Company policy also needs to be taken into consideration: is the goal one of maximum automation or is a more labour-intensive approach more practical? If the latter is true then we can expect to see lines characterised by semiautomatic packaging and manual palletization: the high labour-intensiveness of material transfer operations will provide good flexibility as the capacity of workers to adapt to change is more easily managed than the corresponding software changes on a fully automatic system. On the other hand, however, the workers capacity to carry out a certain function inevitably establishes an output ceiling. As a rough guide, then, we can calculate a tile flow rate in front of the worker of about 20 m/min (rates for double fire and decorated products drop to as low as 15 m/min) with a corresponding packaging rate of 4/5 packs per minute: the line will inevitably be built around these figures. 271


Lines with automatic packaging and palletizing units are a more complex affair in which the human contribution is limited to the selection station; here the operator observes the tiles passing by him/her and evaluates their shade and quality, classifying them with an on-tile mark (for more detailed information on the importance of this process and the general lack of objectivity vis-à-vis evaluation of shade and surface faults see the chapter on defects). Establishing a priori how many tiles an operator can analyse per minute is no easy task: it depends on tile size (yet is not strictly proportional to total surface area), on the type of defects being sought out, on attentiveness in shade evaluation, on the need to check any variation in shade, on product type, on the companys strategic goals and priorities (maximum attentiveness to defects or minimisation of waste) plus many other variables that are too numerous to be listed here. Bearing this in mind a selection speed of 20 m/min should only be taken as a rough upper limit (assuming continuous, uniform flow, with averagely skilled workers and average sorting criteria). Given this, manufacturers should configure sorting and palletizing lines so as to make the most of the individual selectors abilities (or pair of selectors where there are dual tile feed lines or two stations). The best configuration is the one which allows optimum employment of human skills without under-utilisation of automatic units. The manufacturer also needs to decide whether or not it is appropriate to install devices such as stack dividers, glue/strap box joiners, turn-and-pair pick up units, double palletizing pickers and empty pallet transfer units. The right combination is the one that works best as a team, providing plant simplicity, maximisation of efficiency and minimisation of both general and specific running costs. Production control software Production management in a ceramic company is greatly influenced by the enormous number of codes needed to identify products of widely varying type, size, colour and characteristics. A ceramic company often distinguishes its wares according to variations that occur within a certain product: for example, a product may be divided into 1st Quality or 2nd Quality and they, in turn, can contain different shades and sizes that give rise to even more codes. Following this explosion in the number of codes and the resulting complications in stock control the industry has seen widespread application of information technology to enhance automation of production information management. The information does not need to be created: it already exists as part of the sorting, packaging and palletizing line logic, where qualitative and size-related subdivision occurs and pallets containing homogeneous products are loaded. Software enables the user to compress that information, organise it and make it readily available (e.g. as a networked database). 272


Sorting line configurations for porcelain tiles The success of porcelain tiles has greatly influenced and continues to do so sorting and palletising departments, even if indirectly: while the actual machine functions are unaffected by the different physical characteristics of porcelain tiles, indirect changes have been brought about because they are subject to additional post-production processes and have generated a trend towards larger and larger tiles (fig. 23). With porcelain tiles, a significant amount of output is subject to polishing and edge grinding: these are pre-sorted and stacked separately and can be diverted at different stages along the line: the options listed below should be selected according to the percentage of porcelain tiles the manufacturer needs to divert to these polishing and/or grinding stations or the size of the production run. 2

3

4

1

Fig. 23. Porcelain tile sorting line layout.

Tile outlet (fig. 23 - 1) Once identified, tiles to be polished are sent to a flat bed loader. They are then unloaded by a transfer unit positioned at the head of the polishing line. This set up allows the user to divert all the tiles or just some of them, the others being sent directly to the packaging and palletizing department. Stacker outlet (fig. 23 - 2) Device integrated into the machine body and positioned at the sorting line stacker outlet. Expels the stacks to be polished onto a holding belt before they enter the wrap-around unit. These are then loaded onto beds or pallets and transferred manually to the lines in the polishing department. As above, the other stacks are conveyed to the packaging station directly. Wrap through-passage (fig. 23 - 3) All output is polished together: in this case the wrap-around unit can be set up so that the stacks pass through without being packaged. 273


This device avoids early wear of the wrap-around unit as the tiles transit without rubbing against the box closure mechanisms. The stacks are then sent to the palletizer which picks them (but does not, of course, rotate them) and positions them on a pallet which is subsequently transferred to the polishing station: thus palletising of both packaged tiles and those to be polished is performed simultaneously. Size and squareness controller (fig. 23 - 4) This machine classifies tiles by size and shape. Software-controlled dimensional checks determine both the size and shape defect classes. Analysis of orthogonality (squareness) also makes this unit ideal for checking tiles exiting an edge grinder; it is therefore especially suitable for installation on sorting and packaging machines downstream from squaring machines. Another consequence of the growing demand for porcelain tiles is the progressive shift towards larger and larger sizes: tiles measuring 450 × 450 or 600 × 600 mm are now commonplace, 600 × 900 or 900 × 900 mm ones are not unusual and ultra-large slabs measuring 600 × 1200 mm are now making their appearance. Correspondingly, sorting machines have been enlarged and reinforced and the extra care required in packaging has seen manufacturers preferences switch towards cover rather than wrap-around machines.

274

Firing

Chapter VIII POLISHING

Introduction Ceramic tiles, whether porous or vitrified, floor or wall, are increasingly subject to various edge grinding, flattening and polishing operations at the end of the traditional production cycle; the aim is to enhance the aesthetics of the tile and add value. The tiles most likely to be put through such processes are: PORCELAIN TILES: GLAZED (including single fire products) UNGLAZED POROUS WALL TILES: GLAZED (single and double fire). Past developments and current trends Polishing as an additional production process first made its appearance in the mid-80s, being used exclusively for the surface treatment of granito porcelain (salt and pepper) tiles. Towards the end of the decade, as tiles steadily got larger, manufacturers began grinding the edges in order to reduce the number of size classes and thus minimise stocks. At the same time manufacturers began using large grains, micronised powders, flakes and pelletized or sintered glaze grains to produce macro granito, veined and variable-shade effects that necessitated surface polishing and, to a greater extent, squaring, in that such tiles were often marred by geometric variations. Subsequently, polished and edge-ground (or rectified) wall tiles and even glazed porcelain tiles (often confused with a well vitrified single fire product) appeared on the market, the latter being polished with more sophisticated lapping and ageing techniques designed to create specific effects. Here the aim is to create (especially on floor tiles) products that resemble the materials seen in old churches, castles and so on, where decades of wear and waxing have modified the original appearance to give a homely or rustic look that is, paradoxically, in huge demand in todays high-tech society. Randomisation of many production processes (body preparation, press filling, die texturing, unfired and fired tile cutting and machining, mix-and-match packaging etc.) has done much to recreate that sought-after antique look on the individual tile but manufacturers are still a long way from giving the laid floor or wall 275


that same individuality as the gaps between tiles still highlight the fact that they have been made using modern manufacturing techniques. Lines and machines for the rough flattening - polishing - squaring of porcelain tiles With machines for traditional tile sizes the focus is on productivity: todays manufacturers require lines capable of polishing 200-250 m²/h (1600-2000 m²/ shift), a figure that looks set to increase. The work cycle ROUGH FLATTENING (surface levelling) Finished tiles are not always perfectly flat; rough flattening them provides the manufacturer with products of uniform thickness. This is carried out using diamond-tipped rollers, locked in chucks (mandrels) that feature vertical adjustment (usually motor-driven and shown on the display) to facilitate positioning. Normally, the chucks have a pneumatic system which maintains constant onpiece pressure and also raises the roller during line stoppages and repositions it automatically when the line restarts. Diamond-tipped rollers need to be very sharp so as to reduce on-tile pressure to a minimum, thus protecting the fragile ceramic from risk of breakage. Rough flattening leaves the surface with a scoured appearance because diamond tipped rollers are, at this stage, of the rough grain type. Combined roller + satellite heads + roller or roller + turntable + roller systems are generally used. On traditional tile sizes a layer of material some 0.7-0.8 mm and 0.6-0.9 mm thick is removed for salt and pepper and penetrative soluble salt products respectively. On veined marble-like or coarse grained tiles where 1.0-1.2 mm or more may be removed the purpose of rough flattening is not only to level the surface but also, and most importantly, to bring out the desired pattern, colour or motif. On large tiles the removed layer grows to 1.0-1.2 mm (on account of curvature). To calculate how much material needs to be removed note that where flatness variations are in the order of +/- 0.5 mm it is necessary, in order to create a perfectly flat surface, to remove 0.5 × 2 = 1 mm plus at least another 0.2 mm. Grain size ranges from 50-60 mesh on the first stations to 140-180 mesh on the last. In terms of final product quality and line productivity this stage is the most critical. Line speed is 6-7 m/min. New, more efficient rough flattening systems eliminate roller replacement at the product change-over stage: more recent machines, in fact, can be used with all tiles in the 300 × 300 - 600 × 600 mm range and even those measuring 300 × 600, 450 × 900 mm etc. without having to change the diamond-tipped tools. Rough flattening rollers (figs. 1 and 2b) have spiral, diamond-tipped abrasive ridges; initially about 1 cm thick, they wear down completely after rough flattening 276

Polishing

Fig. 1. Flattening-edge grinding machine and close-up of diamond-tipped roller.

some 80,000 100,000 m². New diamond-tipped rims are provided by specialist suppliers and old ones can be regenerated. The latest machines allow the abrasive tools to be changed and maintained safely without stopping production since all zones in each roller station are independent and have a system which raises the chucks well clear of the tiles. The above is true for all lines up to and including fine polishing: hence they are usually equipped with several heads more than are strictly necessary; alternatively the machine work programs feature machine speed adjustment mechanisms that calculate current machine requirements and optimise the qualitative-quantitative efficiency of the system. Because there is a gap of 2-3 mm between the ends of the rollers and the outer tile edge parallel to the direction of product feed (this clearance is necessary because if the roller were the same width as or wider than the tile it would wear unevenly and inevitably chip the tile) it is necessary to flatten the two strips of tile at the edges parallel to the direction of product feed. This can be done by: a) rotating the tile 90° on a turntable after the first few rollers and then continuing through the second group or b) inserting a coarse grinding disc head between the roller stations (see fig. 2a). Solution a), given the same number of rough flattening stations, takes up more space. A rough flattening machine producing 1200 m²/shift normally consists of: 3 diamond-tipped coarse grinding rollers 1-2 diamond-tipped flat plane satellites 3 diamond-tipped finishing rollers. 277


Fig. 2a. Rough flattening with/without tile rotation.

COARSE GRINDING-POLISHING A combined system of flat plane satellites or conical heads + tangential heads (fig. 3). The aim of coarse grinding/polishing is to eliminate any surface roughness that may have been caused by rough flattening.

After rough flattening

After coarse surface grinding

The polishing heads must be perpendicular to the tile surface. Flat plane rotating disc units (from 5 to 7) oscillate slightly, allowing them to adapt to irregularities without exerting excessive pressure or working at the wrong angle; they are thus not perfectly perpendicular but may tilt by 0.5° - 1° vertically and to a same or different degree horizontally. Each tool traces a different concentric area with respect to the vertical abovetile axis and so only a portion of each tool is in contact with the surface at any one time. This helps keep it clean and stops it being clogged by the removed material. To ensure that homogeneous removal also takes place at the sides parallel to the direction of feed the support or, rather, the upper tool-holder beam, must oscillate across the path of the tiles. Where heads with diamond-tipped tools are used, the first coarse grinding stations are in the order of 50/60, 60/80, 80/100 mesh. If, instead, abrasive SiC tool heads are used the coarse grinding grain may start at 46 mesh. Generally speaking, coarse grinding-polishing ends with a grain size sequence of 120, 150, 220 mesh. Feed rates range from 5 to 10 m/min, depending on the material, the thickness of the removed layer, tile size and line type. A coarse grinding-polishing station with an output of 1200 m²/shift usually consists of: 2-3 flat plane rotary discs 9-10 tangential heads. 278

Polishing

Fig. 2b. Transverse roller grinder.

Fig. 3. Machine for coarse grinding and polishing followed by fine polishing; the figure below illustrates different types of head (tangential, cylindrical and satellite).

POLISHING Polishing employs tangential heads to give the tile a mirror-like finish without any scratches, shadows or signs of machining. Tangential heads are similar to those used for coarse grinding, but the abrasive material is of much finer grain size (starting at 400 mesh and ending with an extremely fine 1200-1500 mesh; the polishing tools are usually made of composite mortar based on sintered oxides). Polishing is generally performed in the second section of the coarse grindingpolishing machine. As in the initial section, the support top or tool-holder beam oscillates transversely to the direction of product feed. Tiles are fed through the machine at a speed of about 6-7 m/min. 279


SQUARING - BEVELLING (or SIZING-CHAMFERING) A combined system of tangential rough flattening grinders, frontal rough flattening grinders (fig. 4) and tilted chamfering grinders (fig. 5) with diamond-tipped tools. Tangential grinders carry out rough flattening, removing up to 3-4 mm on each side (6-8 mm total), while frontal grinders are used to smooth the roughness generated by the former. Chamfering units (angled edge grinders) eliminate the very sharp corners on the face of the tile. On porcelain tiles they operate at 45°. Because all four sides of the tile need to be ground, machines (fig. 6) generally feature a turntable between two sets of grinding machines. The intensity of the machining process requires a particularly effective piece holding system. Dual push-pull belt sets both above and below the tile are generally used. Similarly, the need for tight squareness and parallel side tolerances calls for very accurate tile alignment, usually provided by a pusher device. Product feed may be as fast as 10-15 m/min. A squaring-chamfering section with an output of 1200 m²/shift normally consists (on each grinding side) of: 2 rough tangential grinding wheels 2 fine grinding wheels 1 chamfering unit.

Fig. 4. Tangential and frontal edge grinder.

280

Polishing

Fig. 5. Angled edge grinding machine for chamfers.

Fig. 6. Squaring machine. DIAMOND-TIPPED ABRASIVE, GRAIN SIZE ANSI B74 16 1971) SIEVE SETACCIO USA FEPA USA FEPA

99.9% passing Passante

20/30 D852 30/40 D602 (35/40) D501 40/50 D427 50/60 D301 60/80 D252 80/100 D181 100/120 D151 120/140 D126 140/170 D107 170/200 D 91 200/230 D76 230/270 D64 270/325 D54 325/400 D46 (35/40)non is not an (35/40) è uno ANSI standard standard ANSI

1270 920 770 650 455 384 271 227 197 165 139 116 97 85 75

Aperture Max % Apertura Min Max. Apertura Max% Max. Aperture Min.% % retained retained passing % sup. % sup. passante

920 650 541 455 322 271 197 165 139 116 97 85 75 65 57

8 600 90 8 429 90 8 429 90 8 302 90 8 255 90 8 181 90 10 151 87 10 127 87 10 107 87 11 90 85 11 75 85 11 65 85 11 57 85 15 49 80 15 41 80 APERTURE IN(IN MICROMETRI APERTURE MICRONS)

281

8 8 8 8 8 8 10 10 10 11 11 11 11 15 15

2% Max. max 2% passing passante

429 302 302 213 181 127 107 90 75 65 57 49 41


Glazed wall tile squaring lines Main characteristics: high quantities of material removed from the sides (sometimes more than 1 cm per side, thus reducing overall size by 2 cm) squaring with diagonal tolerances of 0.1-0.2 mm perfect side finishing (no glaze chipping) small, adjustable-angle chamfer high productivity (tile feed 22-24 m/min, output greater than 5000-5500 m² per 20 h day) versatility (number and type of grinding wheels can be varied). Initially, attempts were made to perform this task with cutting machines but poor results led to the introduction of more powerful squaring machines, conceptually similar to those used for porcelain tiles. Standard configuration for each grinding side: 3 diamond-tipped tangential grinding wheels for the removal of large quantities of material 1 diamond-tipped frontal grinding wheel 1 diamond-tipped resinoid grinding wheel 1 chamfer grinding wheel, capable of following surface irregularities and adjustable so that shallow, almost horizontal chamfers can be made without removing the entire glaze layer. Polishing - satin finish - semipolishing lines for glazed, unglazed and third firing ceramic These machines have been designed for small (10 × 30 cm) and medium (60 × 60 cm) tile production lines with output rates higher than 600 m²/shift. They carry out: mirror-like POLISHING of the relief on third fire glazed materials. Individual (not flat-plane disc) frontal heads, featuring shock absorption and SiC or diamond-tipped tools with diameters of 150-180 mm are used. Polishing requires the upstream presence of diamond-tipped rough flattening heads. SEMIPOLISHING: consists of small imperfectly flat movable polishing heads as opposed to mirror-like polishing of glazed, unglazed and especially relief surfaces: the frontal SiC or diamond-tipped tools oscillate to adapt to the surface outline. SATIN FINISH-LAPPING of unglazed porcelain tiles and non-gloss glazed materials: this technique enhances lustre by reducing surface roughness, but does not provide a mirror-like finish. The frontal heads mount semi-rigid rubber brushes impregnated with SiC.

282

Polishing

While these lines aim to provide flexibility, both in terms of performance and tile size, they are generally less productive than their porcelain tile counterparts. Designed to partially process (i.e. lap or mirror polish only a portion of the surface area) medium-large tiles (60 × 60 cm); output rates are not high enough to allow full-tile polishing of medium sizes (30 × 30 cm). Appendices To illustrate the above concepts, there follows a description of a standard rough flattening-polishing-cutting line for large porcelain tiles (1200 × 1800 mm). Output 800 m²/shift, fired tile dimensions 1200 × 1800 mm: ROUGH FLATTENING: 3 diamond-tipped coarse grinding rollers 2 flat plane satellite heads with diamond-tipped tools 3 diamond-tipped finishing rollers. COARSE GRINDING-POLISHING-FINE POLISHING: 18-head machine with: 2-3 flat plane satellite heads 15-16 tangential heads. LONGITUDINAL AND TRANSVERSE CUTTING: completely automatic, multi-disc cutting machines to divide slabs up into smaller elements (for panelling/kitchen tops/bathroom tops). SQUARING/SIZING (on each grinding side): 2 frontal (finishing) grinding wheels 1 chamfer grinding wheel. SPECIAL PROCESSES: Special cutting machines to make trims from polished slabs. Numerical control shaper to make kitchen and bathroom tops, table tops and other items. Water treatment and recycling Recovery/re-utilisation of porcelain tile polishing sludge The polishing of porcelain stoneware produces large quantities of sludge. Annual production in Italy alone was, in the year 2000, estimated at 70,000 tons and that figure looks set to increase in line with rising national and international demand. At present, porcelain tile polishing sludge is temporarily stored by the ceramic company prior to use/treatment or consignment to external enterprises that then dispose of it in controlled dumps. As this material largely consists of valuable porcelain tile powder, these solutions are clearly unacceptable both from an environ283


mental and an economic viewpoint. New waste disposal legislation in Italy has recently forced the ceramic industry to meet certain requirements, namely: prevention and reduction of the quantity and hazardous nature of refuse, to be achieved by using clean technologies, economic incentives, environmental certification, ecocompatible products and containment of waste disposal activities, all to be carried out in compliance with strict safety standards. Polishing sludge Being a wet process, polishing produces very large quantities of aqueous suspension (e.g. a line producing 70 m2/hour of 60 × 60 cm tile requires 1800 l/min). Environmental benefits arise from reduced consumption as a result of recycling waste water after clarification, flocculation and settling; the recycling plant is essentially made up of two parts: one for purification of waste waters (clarifier) and another for treatment and drying of sludge (filter press). During clarification the particles in the water sediment out; this process is accelerated by the use of chemical coagulants (ferric chloride and aluminum chloride) that cause the particles to agglomerate and flocculating agents (organic polyelectrolyte) that cause those agglomerates to grow, creating flakes (or flocs) of a size that inevitably drop out of suspension. Subsequently, the sediment is filter-pressed to reduce its moisture content to around 40% and then dried. The sludge is a by-product of wet grinding the porcelain tile surface with abrasive silicon carbide elements (grinding wheels). The resulting residue consists of dust from the tile body (quartz, mullite, zirconium silicate where used as an opacifier agent) and calcite, silicon carbide and alkaline-earth oxides from the grinding wheels themselves and wash water. The material has the consistency of mud or slush, about 2-3 kg of sludge being produced for each square metre polished. On account of its composition, not environmentally harmful, sludge is classified as special inert residue. Chemical analysis (see tab. 1) shows a composition similar to that of the tiles themselves (note the high alumina and silica content). The high percentages of calcium and magnesium oxides derived from the bonding agent in the polishing wheels is accompanied by a high percentage of carbon (elementary chemical analysis) thus indicating the presence of SiC. Loss on ignition is high owing to decomposition of the hydroxides and the hydrated derivates of the divalent metals. X-ray diffractometry analysis highlights the main crystalline phases. Some of these are representative of the materials used in the porcelain tile body: particularly evident are the peaks associated with quartz (SiO2), zirconium silicate (ZrSiO4) and mullite (3Al2O3.2SiO2). Also evident are other crystalline phases such as silicon carbide (SiC), calcite (CaCO3) and soluble magnesium compounds such as magnesium oxychloride (MgOHCl) and periclase (MgO), which can be traced to the bonding agents in the abrasive wheels. Mineralogical analysis (see tab. 2) highlights the presence of the compounds associated with the bonding agent magnesite (MgCO3), magnesium sulphate 284

Polishing

(% weight)

OXIDES

Tab. 1. Chemical composition of polishing sludge.

COMPONENTS COMPONENTI MgCO3 SiC MgCl2 MgSO4 pomice

(% (%weight) peso) 55 14 28 2 1

Tab. 2. Mineralogical analysis of the abrasives analysed.

(MgSO4), magnesium chloride (MgCl2) and the actual abrasive: silicon carbide (SiC) and pumice. Analysis of abrasives with different grain sizes has highlighted the correlation between grain size and binding agent chloride content. The percentage of chloride, in fact, is inversely proportional to the size of the silicon carbide grains embedded in the wheels (see tab. 3).

Cl/kg grinding wheel (g)

ABRASIVE

Tab. 3. Chloride content in abrasive wheels as a function of grain size.

285


Difficulties in using the sludge Porcelain tile polishing sludge cannot simply be recycled into the production cycle because it is a source of undesirable side effects: 1) increased apparent viscosity and thixotropy of slips during wet grinding on account of the soluble salts (mostly chlorides). 2) increased fusibility and deformation during firing on account of calcium and magnesium compounds derived from the grinding wheel binding agent (magnesium oxychloride cement). 3) high porosity stemming from the considerable quantities of gas generated at 1100-1150 °C following decomposition of the silicon carbide. 4) emission of volatile chlorine compounds during firing. The presence of volatile chlorine compounds in the sludge is particularly problematic in that their release as gas during firing can cause corrosion inside the roller kilns. Thermogravimetric (TGA) analysis of samples subject to heating cycles ranging from 20 to 1200 °C shows a significant drop in weight between 600 and 800 °C, which correlates to volatilization of the chlorine compounds and carbonates. Confirmation of this phenomenon has come from chemical analysis of samples treated at different temperatures in which the remaining amount of chloride is measured. As table 4 shows, beyond 600 °C there is a significant drop in the quantity of chlorides (74%), thus confirming the TGA results. Summing up, then, it is impossible to recycle the polishing sludge back into the porcelain ceramic body. To prevent rheological problems in the slip the tiles could be extruded, thus allowing manufacturers to skip the wet grinding stage; to prevent firing problems the sludge could be used with materials that are highly porous, so that they consolidate at temperatures lower than the combustion temperature of silicon carbide. Once the cause of the grinding wheel chemical composition problem has been

Tab. 4. Chloride values in sludge at different temperatures.

286

Polishing

identified, it may also be possible to modify the binding agents, switching to those with a low magnesium oxychloride content. If it is not possible to change the type of grinding wheels it might be possible to separate out the material produced during the first (diamond-tipped) stage of surface treatment, which does not contain silicon carbide or magnesium compounds, thus reducing the volumes in play. Alternatively, the sludge could be used to produce low temperature refractory materials. Some research programs have focussed on the possibility of recovering this inert sludge via heat treatment to produce new, high porosity insulating materials.

287


288

The environmental impact of the ceramic industry

Appendix 1 THE ENVIRONMENTAL IMPACT OF THE CERAMIC INDUSTRY

Like all industries involved in the transformation and processing of raw materials, the ceramic industry inevitably has an effect on the surrounding environment. In other words, a ceramic production plant is an open system which draws on the environment for its supply of: raw materials water fuels electricity and introduces the following back into it: finished products gaseous emissions solid and slurry waste waste waters heat energy noise. An approximate illustration of this two-way flow of materials and energy is provided in fig. 1. It should be pointed out that a ceramic production plant has less polluting potential than factories in many other industries in that a considerable proportion of the pollutants, especially solid particles, can be efficiently filtered at relatively low cost to avoid releasing them into the atmosphere. It also recycles a considerable amount of its scrap and de-watered sludge and much of the waste water. Moreover, because of the changes which take place during firing, solid residue and sludge are generally rendered inert. Nevertheless, the problem should certainly not be underestimated, especially where industrialisation leads to a concentration of factories within a given area, inhibiting the effective dispersion of pollutants, an indispensable requisite for partial self-cleaning of the environment. In Italy, the Sassuolo area is a case in point, and has provided the industry with valuable experience as regards the pollution problems associated with the manufacture of ceramic tiles resulting in enormously improved environmental protection measures. Thanks to prevention-oriented public health policies and increasing awareness among the business community, pollutant emissions have been reduced significantly. The results are evident: compared to 15 years ago, airborne dust concentrations in ceramic manufacturing areas are now just a quarter of what they were, and lead concentrations just one twentieth. 289


NOISE

Gaseous emissions

FILTRATION

Raw materials CERAMIC TILE PRODUCTION PROCESS

Water

Energy

Finished product

Fuel Electricity

Filtration

Production waste/residue

Filtration or separation of waste/residue

Waste water

Fig. 1. Incoming-outgoing material and energy flows in a ceramic tile plant (from Piastrelle Ceramiche & Ambiente - EdiCer 1995).

Technological innovation deserves much of the credit for this progress: for example, the widespread adoption of rapid cycle roller kilns has controlled and reduced the fluorine compound emissions associated with the firing of clayey materials. The introduction of these kilns also led to reformulation of glazes and significantly lower lead content in frits. The involvement of plant engineers and colour manufacturers has resulted in a rational subdivision of roles and responsibilities, essential if ambitious environmental protection and workplace health goals are to be achieved. Anti-pollution measures can be taken at different levels, both upstream and downstream from the actual process that generates the pollutants. 290


Upstream measures include: use of materials with low environmental impact (e.g. reduction of boron and other harmful components in glazes). rational use of fuels with high combustion efficiency and low sulphur content. plant engineering solutions that allow rationalisation of energy consumption (e.g. cogeneration plants for the combined heat/electricity production). appropriate technologies (e.g. powder transport systems that prevent dust dispersion). responsible management of procedures and processes of high environmental impact (e.g. mill and glazing line washing). frequent checking of process efficiency, especially on heat treatment units (kilns, driers, spray driers): extensive use of information technology to control operational parameters within the various departments. Downstream solutions include: installation of efficient pollutant entrapment and separation systems. incorporation of best technology available, cost-compatible abatement technologies at the plant design stage. research into more efficient systems for dispersion of purified waste (air, water or solid wastes) into the environment. research into the potential recycling of waste both internally and in other industries. The most efficient control system remains effective monitoring and long term recording of key pollution data: pollutant emission factors upstream and downstream from separation units per unit of output. pollutant concentration in production process effluent. abatement system efficiency. Bear in mind that a serious approach should focus not only on specific pollutants (mainly dust, fluorine, lead and boron), but also needs to take into account problems that affect the world as a whole (energy resources, greenhouse gases etc.). Finally, companies would also be well advised not to underestimate the competitive advantages to be gained from a serious pollution control policy, as such efforts enhance corporate image and provide additional advertising leverage; this is especially so in fast-growing markets where customer awareness and quality perception evolves almost daily. Pollutants in raw materials for bodies SILICA: found just about everywhere in the ceramic factory, as most raw materials are rich in SiO2. From a toxicological point of view silica can, where particles of a certain size are inhaled, give rise to chronic illnesses (silicosis). 291


FLUORINE: in quartz and feldspars in quantities of 0.0002-0.042%, and up to 0.3% in clays (0.05-0.17% in Italy), and up to 2% and more of micaceous materials, such as F- (vicariant of OH-, which has similar dimensions, in the clayey lattice), F2, HF, H2SiF6, SiF4 (reaction with SiO2 from the destroyed lattice) and/or alkaline fluorines in particulate form. SULPHURS: mineral sulphurs, such as Pyrite FeS2 (disassociation beyond 300 °C) and SO42- sulphates, which disassociate beyond 800 °C. ARSENIC: 2-15 mg/kg (ppm) in porcelain tile bodies, up to 100-150 mg/kg in some pigments. CHLORINE: in water and clay. ORGANIC SUBSTANCES: C and H-based, and nitrogen-containing substances in clays (vegetable residues, humic acids). ORGANIC ADDITIVES: used for rheological correction during body preparation (acrylates etc.). Pollutants in glazes FRITS: Pb, B, As (50-600 ppm and chromophor metal in some minerals). GLAZES AND PIGMENTS: various toxic/harmful elements. Nickel, cadmium, chromeVI, cobalt salts etc. ORGANIC SUBSTANCES: fixers and fluidisers, vinyl resins, CMC, ...screen printing vehicles: solvents (glycols and ethylene and propylene polyglycols); thickeners (waxes, glycerol starch, other starches); fluidizers (polyacrylates); fixers: (CMC, polyvinyl alcohol, starches). Pollutants in gaseous emissions These vary in nature, depending on the raw materials in the body and the glaze, the type of fuel used in the kiln and drier and the applied firing/drying curves. Fig. 2 provides a summary of their characteristics (PV = dusts). Most emissions are produced during the firing process: dispersion of most of the above substances in the form of gas-carried solid particles (fumes) is possible, as is direct vaporisation of the low-fluxing substances. Firing, of course, involves fuel combustion, and that means emission of gaseous products such as nitrogen oxides and carbon oxides. Where natural gas is used, CO2 emissions are in the order of 0.23g/Kcal. If small quantities of sulphur are also contained in the fuel, sulphur oxide emissions can become substantial. 292


The following are also important: Prod.

Stage of production

Main tasks Dry grinding

Body raw mat. prep.

Wet grinding Spray drying

Pressing

Ceramic tiles

Drying Glaze grinding Glaze prep. and glazing

Glazing Glazed tile dry-blowing Porcelain tiles Biscuit only

Firing Glaze only Single fire Smoothing Polishing

Fig. 2. Gaseous emission pollutants (from Piastrelle Ceramiche & Ambiente - EdiCer 1995).

BORON COMPOUNDS Present in frits, boric acid (H3BO3) evaporates during firing. Boron compounds can deposit in chimneys, causing encrustation. ARSENIC COMPOUNDS Small quantities are found in fumes separated with bag filters: they accumulate in the solid reagent (to be disposed of) and sometimes exceed legal limits (according to Italian legislation, this could result in separated residues being reclassified as toxic harmful waste as opposed to special waste, thus increasing waste disposal costs significantly). AMMONIUM COMPOUNDS Generated from nitrogen-containing organic substances and, it is thought, from the NOx produced during firing. Some (e.g. NH4Cl) compounds condense and form deposits in chimneys. 293


CHLORINE COMPOUNDS These mainly consist of the chlorines freed during firing following the decomposition of chlorine compounds in raw materials (mostly clays). ORGANIC SUBSTANCES Aldehydes, benzenes, xylenes, 1,4-dioxanes, n-derivates, alcohols, ketones, esters. The presence of organic substances in fumes, mainly introduced during glazing, is linked to the in-kiln counterflows. Organic substances evaporate during preheating (and may crack), mixing with the fumes without actually becoming hot enough to burn. Such substances may be carcinogenic (aromatics) or, more often, produce an odour that is perceived even at extremely low concentrations. Atmospheric pollution A ceramic plant, then, releases many different substances into the air, and in widely varying quantities; their significance as regards environmental impact and thus prevention and containment correlates closely to their degree of toxicity. Proper evaluation of the impact these emissions have on the area surrounding the factory requires more than just observance of the standards in force and demands application of strict procedures. More than just simple pollutant measurements, these procedures need to define: 1. Specific stack or chimney outflows (Nm3/kg product) 2. Temperature (for hot emissions) 3. For each of the pollutants: MASS FLOW = mass of pollutant emitted per unit of time (g/h). EMISSION FACTOR (EF) of a pollutant = the mass of pollutant emitted per unit of product (kg): defined by the relationship between Mass Flow of the pollutant (g/h) and hourly output (kg/h). Key parameters: UEF = Average emission factor upstream from separation plant (g/kg product). DEF = Average emission factor downstream from separation plant (g/kg product). i = efficiency in terms of % reduction in the separation plant. c = concentration of pollutant in the separated emissions (mg/Nm3). Limits for the main pollutants in gaseous emissions take into account toxicity. Limits according to Italian law (1995) are as follows:

294


Emissions Emissione Grinding - Pressing Macinazione-Pressatura Spray-drying Atomizzazione Smaltatura Glazing Cottura Firing fusori FritForni melting kilns CLASSIFICAZIONE CLASSIFICATION Sostanze cangerogene, Carcinogenic, teratogen, teratogene e/osubstances mutagene mutagen

Inorganic substances Sostanze inorganiche mainly in powder form presenti prevalentemente come polveri

Sostanze substances inorganichein Inorganic presenti sotto forma diform gas o gaseous or vapour vapori

Sostanze organiche Organic substances

Polveri Powders 30 30 10 5 30

3 MaxConc. post-treatment (mg/Nm Max. dopoconcentration depuraz. (mg/Nm ) 3) Pb ---0.5 5

INQUINANTE POLLUTANT Be As Cr6+ Co Ni Acrilonitrile Acrylonitrile Cd Se Sb Cr3+ Mn Pb Cu Sn V F NH3 NOx SOx Acetaldeide Acetal Formaldeide Formaldehyde Acreoline Acreolina Propionaldeide Propionaldehyde Alcol metilico Methyl alcohol Acetone

FLUSSO MASSA MASSDI FLOW ≥ 0.5 g/h ≥ 5 g/h ≥ 5 g/h ≥ 5 g/h ≥ 5 g/h ≥ 25 g/h ≥ 1 g/h ≥ 5 g/h ≥ 25 g/h ≥ 25 g/h ≥ 25 g/h ≥ 25 g/h ≥ 25 g/h ≥ 25 g/h ≥ 25 g/h ≥ 50 g/h ≥ 2 Kg/h ≥ 5 Kg/h ≥ 5 Kg/h ≥ 0.1 Kg/h ≥ 0.1 Kg/h ≥ 0.1 Kg/h ≥ 2 Kg/h ≥ 2 Kg/h ≥ 4 Kg/h ≥ 0.5 Kg/h ≥ 0.1 Kg/h < 0.5

PolveriTotal totali powders

F ---5 5

LIMITE (mg/Nm33)) LIMIT (mg/Nm 0.1 1 1 1 1 5 0.2 1 5 5 5 5 5 5 5 5 250 500 500 20 20 20 150 150 600 50 150

Emissions limits (as per Italian law): the above limits are minimums; maximum values are double that figure. For carcinogenic substances etc. the indicated limits represent both minimums and maximums.

Prevention and separation Preventing the emission of dust and gaseous effluents into the atmosphere begins with proper selection of raw materials and plant engineering solutions. Both should be chosen so as to minimise environmental impact, contain pollutant dispersion and optimise the exhaustion of gasses in a way compatible with workplace health/safety and optimal energy consumption. Production plants should therefore feature closed conveying/transfer and exhaust systems, specially designed and built to minimise dust dispersion, and kilns specifically designed to reduce fuel consumption (and thus combustion emis295


sions) to a minimum: where necessary, kilns should also have systems for the recovery/combustion of the vapours released during preheating etc. Of course, in a competitive industrial context, these goals are not easily achieved: hence the use of filtration plants is inevitable. Gaseous emission filtration systems are designed to: abate dust purify fumes abate the organic substances produced by combustion. While a detailed description of the machines used to achieve these goals is beyond the scope of this book, a summary of the main working principles behind them is undoubtedly useful; for example, in dust abatement it is possible to use: Dust abatement SYSTEM Cyclones Venturi separators Electrostatic units Bag filters

OPERATING PRINCIPLE Agglomeration via turbulence Agglomeration via turbulence Agglomeration Separation

SETTLING Centrifugal force Inertial impact and diffusion Electrostatic attraction Inertial impact +interception

Bag filters Given their versatility, they are used extensively by the ceramic industry: they can also be used at quite high temperatures (about 170 °C). With bag filters, one of the most important parameters is the inevitable drop in filtered vapour flow rate, caused by the increasing resistance opposed by the filtration element. That resistance has two components: the resistance of the fabric itself (constant) and the resistance offered by the layer of dust that gradually builds up on the bag interior. To ensure proper filter performance that layer is periodically removed using mechanical shakers, flow inverters, compressed air etc. Bag filter sizing/performance is summarised in the following table:

296


Parametro di Sizing parameter dimensionamento Velocità di filtrazione Filtration speed Perdite di carico Flow rate loss

Unità di misura Unit of measure

Valori tipici Standard values

m/min m/s m/min m/s Kpa kPa mm H2O

0.8 ÷ 1.2 0.02 0.01 ÷ 1.8 ÷ 2.4 0.04 0.03 ÷ 3.8 1.5 ÷ 300 150 ÷

Fume filtration This includes filtration plants capable of trapping both the gaseous (e.g. fluorine) and solid particulate (e.g. lead, dust) pollutants contained in industrial gaseous emissions. The latter are treated as per the above-described dust abatement methods while gaseous pollutants are generally treated in systems that ensure their wet or dry absorption in an active reagent (sometimes just water). The contact reactor may take the form of a nebulisation tower, a dynamic Venturi-type system, a plate tower or other system in which fumes exiting the stack are scrubbed with water (to which suitable reactive absorbents are often added); such systems should have as extensive an exchange surface area as possible and prevent the spray or liquid particulate exiting the filtration unit itself being drawn in. In dry systems the gas exiting the stack is passed between suitably sized particles of chemical absorbent, or through filters of extensive surface area (honeycomb structures or similar) that react intensely with the gaseous pollutant. A typical example of this is entrapment of fluorine via dry or wet calcium hydroxide/calcium carbonate absorbents (the absorbent outnumbering the pollutant by a ratio of at least 3:1). 2HF + Ca(OH)2 → CaF2 + 2 H2O 2HF + CaCO3 → CaF2 + H2O + CO2 The main problem with these filtration methods, whether wet or dry, lies in difficult control of the reaction products (calcium hydroxides or calcium carbonates enriched with calcium fluoride); recycling them back into ceramic bodies is difficult because of rheological side effects and the presence of calcium carbonate, which alters the fusibility characteristics of the body. Abatement of organic substances As seen, a considerable quantity of organic substances at the kiln inlet is commonplace, especially with glazed tiles; in addition to those contained in the raw 297


materials, normally kept to a minimum through careful selection in order to eliminate black core defects, many organic substances are added during the production process in the form of rheological additives (during grinding), screen printing vehicles, thickeners and fixers etc. During firing, all raw materials are subject to reactions: these range from simple evaporation to oxidative dissociation and varying degrees of pyrolysis, depending on the complexity of the polymeric chain making up the molecules. While reactions differ in complexity, it can safely be said that carbon-based substances rarely give rise to the simple combustion reaction: CH4 +2 O2 → CO2 + 2H2O. Instead, because of the extremely high heating rate inside ceramic kilns, the combustion process forms a series of organic compounds that, on account of their toxicity or because they have an easily perceived unpleasant smell, need to be trapped and eliminated. Entrapment can usually be carried out using a dry-bed or fluid-bed process via absorbent (or active) carbon, or by way of chemical reactant aqueous solutions in scrubbers, in which the vapours to be purified meet a counter-flow water surface. However, these absorption systems provide low performance and are too complex to be used effectively by the ceramic industry. Post-combustion is a much more efficient method of pollutant removal, yet is complicated from a plant engineering viewpoint on account of the high fume flow rate and is costly. It involves capturing the fumes to be treated (or, at least, a part of them) and burning them at high temperature so as to cause complete combustion of any remaining organic substances. A part of the heat energy used in this process can nearly always be recovered and used for other purposes. While abatement is never perfect, efficiency ratings of 85-95% can be achieved, depending on the organic substance in question. This does not, of course, guarantee the complete destruction of odorous substances, which can still be perceived at the parts per billion (ppb) level and even lower. Generally speaking, where the problem of organic pollutants is a pressing one (e.g. bricks lightened with polystyrene or wood chippings), a reduction in the pollutant content of the fumes is best achieved via careful kiln design: the vapours released during preheating (400-600 °C) should be partially recuperated and directed to the actual firing zone where they dissociate completely. Water pollution Obviously, the tile manufacturing process: a) uses the greatest amount of water during the wet grinding of raw materials and the preparation and application of glazes. b) waste water emissions are almost exclusively associated with glaze preparation and application. 298


Both the composition and flow rate of such water emissions are extremely variable, being influenced by the quantity/type of output, glaze type, glaze application system and other difficult-to-quantify factors (organisation of production, departmental layout, extent of automation in washing etc.). This explains why there is no linear correlation between specific water consumption and output type and/or potential. Water inflow/outflow balances are, rather, influenced by the type of production technology, especially as regards raw material grinding. Fig. 3 shows a hypothetical water inflow/outflow balance for a generic production cycle, while fig. 4 is more specific and shows how water is used in a ceramic production plant. Note how waste waters can largely be traced back to glaze preparation and application, as the waste waters from the raw material wet grinding mill (and spray drier) and press cooling water are already fully recycled. The evident economic and ecological advantages of recycling are highlighted by comparing filtration water flow rates and overall factory water requirements. Fullscale recycling reduces water supply requirements by 86% in double fire processes and 48% in single fire ones. In the latter, nearly all companies use recycled water, even where only partially filtered, for wet grinding of bodies. In double firing with dry grinding, recycled water is used to wet the raw materials and, above all, to wash glazing lines. In this case, though, the percentage of companies making use of full-scale water recycling is lower. Before waste waters from the various stages of the production process can be discharged into the environment surrounding the factory (streams, rivers etc.) or re-utilised, they must be clarified. Observation of the parameters determining discharge acceptability (maximum concentration in the effluent) (tab. 1) provides information as to the pollutants that need to be removed by such treatment.

299


KEY Gaseous effluent

Water requirement Outgoing material

Water Production process

Incoming material

Water for production

Waste water

Water balance in a production process:

The W symbols indicate specific flow rates in m3/1000 m2 of tile for a production process: WT: water requirement. Water used both as raw material and process fluid. WMi: water contained in incoming material. WG: water formed by chemical reactions during the production process. WV: outgoing water, as vapour in gaseous effluents or dispersed as vapour in workplace. WMu: outgoing water contained in finished product. Ws: Waste waters.

Fig. 3. Water inflow/outflow in a manufacturing cycle for ceramic tiles and glaze products (from Piastrelle Ceramiche & Ambiente - EdiCer 1995).

300


Water used as:

Prod.

Phase

Main operations

Raw material

Water released as:

Process fluid

Vapour in gaseous effluent

Waste waters

Dry grinding Wetting Body raw material prep.

Wet grinding

W

Spray drying

W

Ceramic tiles

Pressing

C

Drying Glaze grinding Glaze prep. and Glazing application

W W

Tile blowing Porcelain tiles Biscuit only Firing Glaze only Single fire

Products for glazes (from glaze suppliers)

Smoothing Polishing

W

Frit production

Colorant and compound prod.

C

W

Oxide grinding

C

Spray dying

C

Fig. 4. How water is used and recycled in the ceramic tile production process (W = washing, C = cooling. From Piastrelle Ceramiche & Ambiente - EdiCer 1995).

301


Type

Parameter/element

Chemical-physical parameters

pH Coarse materials

Dispersed solid substances

In surface water

In sewage systems

Absent

Absent

Sedimentable materials (ml/l) Total materials in suspension (mg/l)

80-200 < 40% of value upstream from system

Aluminium, Al (mg/l) Boron, B (mg/l) Cadmium, Cd (mg/l) Chromium III, Cr (mg/l) Chromium VI, Cr (mg/l) Iron, Fe (mg/l) Manganese, Mn (mg/l) Metals and non-metals Nickel, Ni (mg/l) is solution and suspension Lead, Pb (mg/l) Copper, Cu (mg/l) Selenium, (mg/l) Tin, Sn (mg/l) Zinc, Zn (mg/l) Fluorines in solution, F (mg/l) Chlorines in solution, CI (mg/l) Sulphates in solution, SO4= (mg/l) BOD5 (mg/l)


COD (mg/l)


Organic substances

Tab. 1. Parameters determining discharge acceptability (maximum concentration in effluent).

302


Water treatment (separation) Fig. 5 illustrates a standard separation plant. Here, effluent undergoes chemical-physical treatment via the use of reactive alkalines (normally soda in aqueous solutions), inorganic flocculants [for the most part FeCl3, AlCl3 or Al2(SO4)3] and organic coagulants (anionic polyelectrolytes of high molecular weight). Initial clarification takes place in the sedimentation tanks, efficacy depending on particle size and density (see table below). Particle size

Sedimentation rate (m/s)

Type

Gravel Coarse sand

Time required for sediment to drop 1 m 1 second 10 seconds 19 seconds 8 minutes 2 hour 8 day 2 years 200 years

Fine sand Bacteria Clay Colloidal substances

Grating Waste waters

Accumulation Pre-sedimentation

Homogenisation Sludge Purified waters Claro-flocculation

REAGENTS: 1. Neutriliser [Ca(OH)2] 2. Coagulant (e.g. FeCI 3) 3. Flocculant

Sludge

Fig. 5. Water treatment plant.

303


In the claro-flocculation section the flakes grow and settle out from the ascending water current; this process is aided by the filtration effect of the sludge layer that forms beneath because the water slows down as it rises until it equals the sedimentation speed of the larger suspended particles; these thus reach a dynamic equilibrium and form a filtrating layer capable of withholding the smaller particles too. The thus-purified water exits the upper section of the settling tank (normally by way of a drop spillway); sludge, instead, is periodically removed from the bottom of the separator and may be reintroduced into accumulation tanks where it can thicken further. The type of plant shown in the figure is the most widespread type, although many others exist. Such plants generally provide manufacturers with relatively good heavy metal separation performance. However, given the high incoming concentrations, that performance does not always guarantee compliance with legal limits. The parameter which seems to have the greatest effect on heavy metal separation efficiency is pH (in addition, of course, to retention time, which depends on the size of the plant). In 90% of cases, infeed pH oscillates between 7.5 and 8.5, owing to the basic hydrolysis of the raw materials contained in the water. The pH is carefully regulated by adding reactants so as to control concentration of, for example, lead and zinc in the filtered waters. As pH increases, Zn (OH)2, present as precipitate, tends to re-dissolve, passing into solution as a zincate anion, while for acid-level pH solubilisation of the Zn salts with formation of Zn2+ occurs. Lead behaves similarly: in acid environments it is in solution as a cation, while for decidedly basic pH values re-dissolution of the precipitated lead hydroxide occurs and plumbites form. Optimum pH levels for precipitation of Pb and Zn lie between 8 and 9, especially with regard to zinc. Where pH values are higher than 9 undesirable post-reactions can occur, and where higher than 10 the re-dissolution of precipitates becomes commonplace. The main technical limit in this sort of plant is an almost total ineffectiveness against boron and its compounds. Boron control is a tricky problem in that separation systems specifically designed to tackle this pollutant, such as those which use ionic exchange resins, are difficult to apply to the ceramic production process and would not, in any case, provide cost-effective reliability or duration. In addition to high investment and running costs, utilisation of such systems is hampered by the need to install resin regeneration equipment and dispose of the eluates containing high concentrations of boron. As an alternative, it may be possible to use easily removable ionic exchange units, which can be regenerated by specialist enterprises, although the usual disposal method for boron-containing waste water is to use it for wet grinding of the body. 304


Porcelain tiles Over recent years, management and purification of waste waters has been made even more difficult by the increasingly common practices of polishing and edgerectifying the fired tile. These operations require considerable quantities of water (800-1000 m3 of water/m2) and the sludge separated from the resulting effluent is difficult to reuse in that it contains a lot of fired, very hard ceramic material, and large quantities of abrasive granules (artificial diamond, corundum, SiC) and materials from the abrasive blocks (metal or MgCl2-based), all in quantities ranging from 2.5 to over 10 g/l. Moreover, demand for surface smoothing/polishing of glazed products is on the increase, thus complicating re-utilisation of sludge and recycling of purified water even further (see Chapter 8, Water treatment and recycling). Solid waste and residues Solid waste, whether associated directly with manufacturing operations or the filtration of gaseous/water emissions, is produced at just about every stage of the ceramic tile production cycle. Such waste may be classified as illustrated in table 2. A better understanding of the situation can be provided by taking a look at the changes that occurred between the 80s and the late 90s, when internal recycling became far more common. Over that time enormous improvements were made regarding the amount of waste introduced back into the environment. Waste output in the Sassuolo area was, in past years, about 40,000 tons/year of which about 50% was directed to landfills. The amount being dumped has now been slashed to about 2000 ton/year: this improvement is even more impressive when one considers that total sludge output doubled over the same period on account of increased ceramic production. The current situation as regards recycling of individual solid waste categories is more or less as follows: BODY PREPARATION powders total recycling PRESSING powder and scrap total recycling DRYING powders and scrap total recycling GLAZING line sludge [Average production of (dry) sludge = 12% of applied glaze] 68% Fired material chamotte (grog) 85% Exhausted depuration calcium hydroxide 40% Fired material smoothing/polishing sludge almost 0% Concentrated boron solutions (from ionic resin exchange) none Re-utilisation capacity, as with the waters, depends on how separation of the recovered solid is carried out, and thus the extent to which salts, chromophor components or hard materials have accumulated. 305


Refuse/residue Prod.

Phase

Production

Depuration Powders (sludge) trapped by gaseous emissions filtration systems

Ceramic tiles

Body raw material prep.

Pressing

Unglazed, unfired scrap

Drying

Unglazed, unfired scrap

Glaze prep. and appl.

Unfired/unfired scrap (unfired glaze on unfired body)

Powders (sludge) trapped by gaseous emissions filtration systems

Unfired/fired scrap (unfired glaze on fired body)

Water depuration sludge (chemicalphysical plant)

Frit, glaze leftovers and scrap etc.

Concentrated boron solutions from water filtration (ionic exchange process eluates, concentrated by membrane processes)

Glaze trimmings/scrapings

Firing

Glaze products (from glaze suppliers)

Smoothing Polishing

Powders trapped by gaseous emissions filtration systems

Exhausted calcium hydroxide from fume filtration

Fired scrap (*)

Powder (sludge) trapped by gaseous emissions filtration systems Fired scrap (*) Water filtered sludge (chemicalphysical plant)

Frit production

Water filtered sludge (chemicalphysical plant) Production leftovers Exhausted calcium hydroxide from fume filtration

Colorant and compound prod.

(*) includes pieces rejected during subsequent sorting stage.

Tab. 2. Classification of refuse by stage of production cycle (from Piastrelle Ceramiche & Ambiente EdiCer 1995).

This is important because the latter have a marked effect on the rheology of the suspensions into which they are introduced, on body and glaze colour, on surface appearance and body vitrification curves. Outsourced recycling, while complicated by strict standards governing the handling and transport of toxic-harmful waste, may be worthwhile: however, outsourcing is more commonplace in the production of brick (where recycled waste accounts for 306


5-20% of the body), expanded clay (60-90%) and cement (not more than 2% but total quantities are very high). Where recycling internal or external is impossible, refuse and solid waste must be disposed of properly in a dump. Where such waste does not contain toxic substances it is stocked directly, yet where, as is more likely, it contains more than the allowed amount of toxins, it undergoes stabilisation/solidification treatment; this consists of acid digestion followed by mixing with cement, which fixes the components in a sort of chemical-mechanical bond via a gradual crystallisation and polymerisation process. The flow diagram below provides an indicative illustration of the solid waste treatment cycle: CERAMIC TILE PRODUCTION PROCESS

Recycled back into the production process

Refuse/residue

Controls

Storage Refuse

Residue Pre-treatment

Transport

Rendered inert (toxic and harmful wastes)

Re-utilisation in other products /processes

Dump

307


308


Appendix 2 PRODUCTION CONTROLS

Volume I underlined the importance of proper ceramic body formulation, illustrating key analysis techniques. Where such techniques are applied and interpreted properly, they provide a good understanding of each raw material vis-à-vis the desired technological characteristics of the body. Additionally, once selected, all raw materials, whether plastic or non-plastic, need to undergo suitable quality control procedures in order to routinely monitor their technological characteristics. A description of the main quality controls follows: this list should be taken as a general guideline, actual tests being applied on a case by case basis that depends on the degree of sample homogeneity. Controls on raw materials or bodies Pre-analysis preparation of raw material samples should include the following: wet grinding in laboratory mills and evaluation of the quantity of water needed to produce a slip of a standard, set viscosity. Grinding should always be controlled according to the residue on 63 or 45 micron screens. sieving of slip through a 1000-1600 mesh/cm2 screen. drying of the suspension. breaking up of the dried material in hammer mills or manually in a mortar and pestle. wetting of the powders and sieving at 90 mesh/cm2. ageing to homogenise the moisture. pressing of the powders in a 4-ton hydraulic laboratory press, with a specific pressure of at least 300 kg/cm2, preferably with a cylindrical die, ∅ 50-70 mm, or, if rectangular, at least 50 × 100 mm. With non-plastic materials samples can be prepared by mixing them with a standard quantity (e.g. 30%) of a standardised illite clay: this is sometimes preferable to the practice of adding an organic binder (usually CMC in wetting water) to the non-plastic material.

309


Sample tiles are tested to establish the following: 1a) Post-pressing expansion (%) A side of the pressed sample tile is measured with a gauge accurate to 0.01 mm, giving (L2). The difference between the latter and the die measurement (L1) multiplied by 100 then divided by L1 gives post-pressing expansion: L2 - L1 % E = × 100 L1 where E = Post-pressing expansion Expansion is considered to be low when less than 0.4%, medium between 0.4 and 0.7% and high above 0.8%. 1b) Pressed tile bending strength (MOR) Bending strength is expressed in kg/cm2, or, rather, in N/mm2. Tests are carried out on samples of at least 50 × 100 mm using 3-point testers (see fig. 1).The following formula is applied: 3×P×L BS = 2 × h2 × b where BS = bending strength (N/mm2) P = reading on scale (N)

Fig. 1. Apparatus for measurement of bending strength.

310

Production controls

h = minimum sample thickness (mm) L = distance between the two blades supporting the sample (mm) b = sample width at the point of breakage (mm). Results are generally evaluated as follows (kg/cm2) = 10.2 × N/mm2 Green BS Dry BS

LOW <6 <20

MEDIUM 6-12 20-30

HIGH >12 >30

Samples dried at 110 °C in laboratory ovens are tested for the following: 1c) Drying size variation (usually shrinkage) (%) A gauge accurate to 0.01 mm is used to measure the just-pressed sample (L2) and the dried piece (L3). The difference between the two measurements multiplied by 100 then divided by the pressed sample size (L2) gives drying shrinkage; the usual reference is the pressed tile size: L3 -L2 % Re = × 100 L2 where Re = Drying size variation 1d) Dried tile bending strength Procedure as per 1b). The dried samples are then fired in (electric muffle) kilns at various temperatures (1060°-1220 °C) or in a gradient kiln at one hour cycles at suitable temperatures (e.g. 1100-1150-1200 °C). Tests on fired samples are carried out to evaluate the following: 1e) Loss on ignition (%) The sample is fired at 1060 °C and weighed. The difference between dried tile weight (P1) and fired tile weight (P2) is multiplied by 100 and divided by (P1) to give L.O.I. (%): P1-P2 % PF = × 100 P1

311


1f) Colour and aspect of tile surface Variations in shade at different firing temperatures are recorded and a description of any impurities, specks, cratering etc. on the surface of the material is made. 1g) Firing size variation (usually shrinkage) (%) A side of the fired sample is measured with a gauge accurate to 0.01 mm (L4). The corresponding size of the die cavity (L1) is also measured and the difference between the two measurements is divided by L1 and multiplied by 100: L4 - L1 % R = × 100 L1 where R = Firing size variation 1h) Water absorption (%) Samples fired at various temperatures are weighed (P1) and then placed in a water bath and boiled for a given time. They are then left to cool to ambient temperature, excess water is wiped off the surface of the tile and they are weighed again (P2). Absorption (WA%) is calculated by way of the following formula: P2 - P1 % WA = × 100 P1 where WA = water absorption The results of all the above tests are then summarised on the raw material report cards. These should be filed carefully and updated as necessary (via periodic structural, chemical, particle size distribution, thermal analyses etc. effected by external facilities). Non-plastic raw materials, instead, are put through routine chemical analysis and sometimes X-ray diffractometry, softening point and heating microscope melting behaviour tests (see tab. 1 - tab. 8). The record cards illustrated on the following pages summarise the main physical and mechanical characteristics of the plastic and hard materials used to produce ceramic tiles. Of course, attaining consistency in terms of product characteristics, plant efficiency, product performance and in-depth understanding of the technical characteristics of the raw materials will require (in the factory test lab or in external facilities) a routine programme of regularly performed controls on all semi-finished items at every stage of the production cycle.

312

Production controls

SACMI (laboratory) PLASTIC CLAY RAW MATERIAL SPECIFICATIONS

Tab. 1. Red firing plastic clay, with incipient fusibility characteristics.

313


SACMI (laboratory)

RAW MATERIAL SPECIFICATIONS

Tab. 2. White firing plastic clay, with fairly refractory firing characteristics.

314

Production controls

MEDIUM PLASTICITY CLAY RAW MATERIAL SPECIFICATIONS

Tab. 3. Clays with good unfired characteristics. Vitrification is gradual without sudden melting.

315


LOW PLASTICITY CLAY RAW MATERIAL SPECIFICATIONS

Tab. 4. Clay with limited plastic properties. Low unfired bending strength. Refractory during firing.

316

Production controls

FELDSPATHIC SAND RAW MATERIAL SPECIFICATIONS

Tab. 5. Non-plastic materials, with medium potassium and sodium oxide values.

317


POTASSIUM FELDSPAR RAW MATERIAL SPECIFICATIONS

Tab. 6. Non-plastic materials, with high potassium oxide content (feldspathic rock with a high orthoclase concentration).

318

Production controls

SODIUM FELDSPAR RAW MATERIAL SPECIFICATIONS

Tab. 7. Non-plastic material with high sodium oxide content (feldspathic rock with a high albite concentration).

319


QUARTZ RAW MATERIAL SPECIFICATIONS

Tab. 8. Non-plastic refractory material, with a very high SiO2 content.

320

Production controls

The most important of these are described below. Note that the quantities to be controlled (i.e. moisture, density etc.) can be changed in accordance with the precision of the measuring instrument (i.e. scales). Body preparation department controls To maximise plant efficiency and optimise output the Quality Control staff should carry out regular checks and tests in the body preparation department. Controls on slips at the mill outlet or slips in secondary holding tanks: 2a) Residue Two hundred g of slip is sieved at 10,000 mesh/cm2 (equal to 63 µm or even finer for glazes, using 16.000 mesh/cm2, equal to 45 µm). The residue is dried to constant weight, then weighed: Dry residual weight in g Residue % = × 100 dry weight in slip Control frequency: every mill load. 2b) Slip density = weight/volume One hundred cc of suspension is sampled with a calibrated cylinder or container: D. (g/l) = weight in grams × 10 Control frequency: every mill load. 2c) Slip water content (%) One hundred g of slip is sampled and dried at 110 °C to constant weight. The dry weight is subtracted from 100 to give the percentage of water in the slip. Automatic thermobalances, which need only about 10 g of slip, can also be used. Control frequency: every mill load. 2d) Viscosity For information on the rheological properties of aqueous suspensions such as viscosity, shear stress, thixotropy etc., refer to Chapter 6 in Volume 1. Note that comparison between different outflow viscosity measurements can be affected by thixotropy. For production line controls a Gallenkamp or Engler viscometer should be used; this is used to calculate the ratio between the time it takes the slip to flow out of a high precision outflow funnel/cup (fig. 2) via a calibrated hole (usually 4 mm) and fill a 100 cm3 graduated cylinder and the time it takes for water to do the same: 321


Fig. 2. Viscosity measuring apparatus (Ford Cup: can be used to measure Engler viscosity).

°E = T1 / T2 where °E = viscosity in degrees Engler T1 = slip outflow time, in seconds (this value is equal to Ford cup viscosity and used for glazes) T2 = water outflow time, in seconds. Control frequency: every mill load. Spray dried powder controls 3a) Moisture content Calculation is similar to that described in 2c), the sample being dried in an oven. Testing is done approximately every 4 hours. Speedy devices, consisting of a metal bottle with seal plug, pressure gauge and weighing scales are generally used. The scales are used to measure the quantity of powder to be placed in the bottle-reactor; a quantity of calcium carbide is placed inside the bottle with the sample (this must not be mixed with the powder until the bottle has been hermetically sealed). The bottle is then shaken. The % moisture content can be read off directly as it is directly proportional to the pressure created in the bottle by the release of acetylene: 2H2O+ CaC2 → C2H2+ Ca (OH)2 Other instruments are available: for example special weighing scales with fast infrared or microwave heaters allow the moisture content of spray dried powder 322

Production controls

samples (taken from the feeder belt or press hopper) to be measured in just a few minutes. IR or MW technology also allows use of continuous control, contact-free instruments. Control frequency: every half hour. 3b) Bulk density One litre of spray dried powder is weighed: bulk density is expressed in g/cm3. Control frequency: once a day. 3c) Particle size distribution A sample of spray dried powder is dried to constant weight. Then 200 g of dry product is weighed out and passed through a system for measuring particle size distribution by sieve analysis (fig. 3) on 600-125 µm screens. Control frequency: every day on an average sample.

Fig. 3. System for measuring particle size distribution by sieve analysis.

Press department controls Maintenance of optimum plant efficiency and output requires that press department staff carry out the following controls: Powder controls 4a) Moisture content (as described in 3a) Calculated using a SPEEDY device (fig. 4) Control frequency: every 2 hours. 323


4b) Soft powder filling Evaluated by measuring the distance between the first fall punch (deepest position) and the level of the die. Control frequency: once per day per press. It is then necessary to re-check the particle size distribution of the body using the vibrating set of sieves (as this may have altered during silo storage, sieving or transport on conveyor belts). Tile controls 4c) Thickness A gauge is used to measure thickness on all 4 sides (marked A, B, C, D) of all the tiles from a pressing cycle. Comparison gives information on die cavity filling and punch settling performance. If defects are detected more accurate controls can be carried out with the aid of a penetrometer. Control frequency: once every two hours. 4d) Penetrometer measurement The instrument illustrated in fig. 5 applies a standard load (about 3-4 kg) on a cylindrical needle of selected cross-sectional area. It records the extent of penetration into various parts of the pressed tile using an electronic or mechanical micrometer. The greater the number of control points the better the overall picture of pressing consistency: the minimum requisite is that readings be taken at the four corners and in the middle of the tile. They should not be taken any closer than 15 mm to the edge. More accurate pressed tile bulk density checks can be made using a mercury density measuring system (fig. 6).

Fig. 4. Speedy moisture content measuring device.

324

Production controls

4e) Weight of tiles from every press cavity All the tiles from a press cycle are collected, marked and weighed separately. Weight comparison provides useful information for the optimisation of die filling. Control frequency: every 2 hours. 4f) Green bending strength (MOR) Bending strength is calculated using the apparatus illustrated in fig. 1. All the tiles in a pressing cycle are analysed (see 1b). Together with other parameters such as moisture content, tile thickness, penetrometer readings and weight, this parameter provides useful information for the optimisation of die filling. Control frequency: every 2 hours. 4g) Post-pressing tile dimensions All four sides of all the tiles from a pressing cycle are measured and the average size is calculated. The results provide data for the calculation of post-pressing expansion and are indispensable for calculating drying shrinkage. Control frequency: every 2 hours. 4h) Compression ratio This is the ratio between the thickness of the powder fill and that of the pressed tile. Control frequency: once per press per day.

Fig. 5. Penetrometer.

Fig. 6. Mercury density measuring system.

325


Controls on dried tiles Dried tiles need to be tested as follows: Controls at the drier inlet 5a) Moisture content at inlet A representative batch of tiles from a car (double fire) or a batch corresponding to a press cycle is weighed at the drier inlet (P2); it is weighed again after being dried to constant weight in an oven (P1). The difference between the two weights divided by the initial weight and multiplied by 100 gives the average moisture content (%) of tiles being fed into the drier. P2-P1 % MC = × 100 P2 Control frequency: twice per shift. Controls at the drier outlet 5b) Average weight of dry tile A set of tiles is dried to constant weight, then weighed; that weight is divided by the number of tiles in the set to give average dry tile weight. Control frequency: twice per shift. 5c) Moisture content at outlet Procedure as per 5a), using tiles exiting the drier. The dried tile can be taken only if the sample drying cycle has been working properly. Control frequency: twice per shift. 5d) Dry bending strength (MOR) Bending strength is calculated by using the apparatus and methods described above on a set of cold, dried tiles (see 1b). Control frequency: twice per shift. 5e) Drying shrinkage A set of tiles representative of either a drier cross-section or a pressing cycle is sampled. All their sides are measured before and after with a gauge accurate to 0.01 mm and average tile size is calculated. Then the following formula is applied: D1-D2 % Re = × 100 D1

326

Production controls

where Re = Drying shrinkage D1 = post-pressing size D2 = post-drying size. Controls on the biscuit (double fire only) The following controls must be carried out: 6a) Average weight Two sets of tiles representative of the kiln cross-section are sampled then weighed. The obtained weight divided by the overall number of tiles in the sample gives average tile weight. Control frequency: once a day. 6b) Water absorption (%) The above samples are weighed separately (P1) then boiled for 2 hours and left to cool to ambient temperature. Excess H2O is wiped off with a cloth and the tiles are re-weighed (P2): WA is given by the formula: P2 - P1 % WA = × 100 P1 Control frequency: once a day. 6c) Bending strength Bending strength is calculated by using the above-mentioned apparatus and methods with sample sets taken as per 6a (see 1b). Control frequency: once a day. 6d) Size variation % The size of each tile (L2) is measured with a gauge: size variation (%) with respect to the original die cavity dimension (L1) is given by: L2 - L1 % R = × 100 L1 Control frequency: once a day.

327


Fig. 7. Dilatometer for thermal expansion measurements.

6e) Biscuit thermal expansion coefficient Calculated with the aid of a dilatometer (fig. 7): periodically used to check for proper glaze-body match. Controls in the glazing department Control Laboratory staff should carry out periodic checks to maintain plant efficiency and optimise production. Controls to be carried out on the glazing line. 7a) Glaze density and Ford cup viscosity (as described above in 2b and 2d). 7b) Applied weight Several tiles are sampled downstream from each glaze application unit. With double fire products the glaze used to be scraped off and weighed; a faster and more reliable method is now used (with double and single firing) in which the pre and post-application weight difference is calculated. An efficient alternative is to use stainless steel trays exactly the same size as the tiles to collect and then weigh the glaze (or engobe) directly as it is applied on the line. The weights applied by each individual application unit are then compared with production standards to ensure that glazing remains within acceptable parameters. 328

Production controls

Controls on the finished product Finished products should undergo the following checks: 8a) Average weight Two sets of tiles representative of the kiln cross-section are weighed. Total weight divided by the number of tiles in the sample gives average tile weight. Control frequency: once a day. 8b) Water absorption (%) Samples are take as per 8a) and boiled for 2 hours and left to cool under water to ambient temperature. Excess H2O is wiped off with a cloth and the tiles are reweighed (P2): WA is calculated as per the formula given in 6b). Control frequency: once a day. 8c) Size variation % The dimensions of each tile and the corresponding die cavity dimensions are measured with a gauge. Size variation is calculated as per the formula in 6d). Control frequency: once a day. 8d) Bending strength Method as described previously. The test is carried out on a set of samples representative of the kiln cross-section. Once a day. 8e) Crazing resistance One or more sets of tiles representative of the kiln cross-section are taken. All the samples are placed in an autoclave and subject to pressure/temperature increase checks as per international standards. The tiles are then removed and checked for crazing by spreading special inks or methylene blue over the glazed surface. Control frequency: at least once a day. 8f) Resistance to acid/alkaline aggression Chemical aggression tests are carried out on several samples of each product, taken from stocks and the kiln, as per international standards. Control frequency: weekly. 8g) Abrasion resistance (for floor tiles) Abrasion resistance tests are carried out on several samples of each product, taken from stocks and the kiln. Control frequency: weekly. The results of the above controls, carried out at various stages of the production 329


cycle, are generally recorded monthly and filed by the Technological Laboratory for reference purposes. Laboratory controls on raw materials for frits and colours All the controls described below must be carried out in parallel and under the same conditions with a standard laboratory sample. The laboratory must therefore have a comprehensive sample set consisting of all currently used materials. 9a) Viscous or low fusibility frits The fastest control method is to produce a tablet and measure its fusibility and to glaze a tile so as to reveal surface characteristics such as glossiness, transparency, opacification, matting, pin holes. Where available, utilisation of a colorimetric analysis spectroscope with relevant software (e.g. CIE Lab), is advisable. The control is carried out by grinding 100 g of frit in parallel with a standard sample, under the same residue and density conditions. A tile is then glazed and the remaining quantity used for the tablet test using the same weight. The samples are then fired in parallel, according to the cycle and maximum temperature deemed optimum for the standard frit. 9b) Fluxing frits As above, the most important control is the tablet (or button) test which demonstrates fusibility; the tile glazing test is less important as behaviours diverge less in this regard. With these materials it may be useful to evaluate fusibility in numerical terms and this can be done empirically by applying the following formula: Length × Button Width Fusibility (relative index) = Length × Standard Button Width With both viscous and fluxing frits it is advisable to carry out controls on several samples from several bags in proportion with supplied amounts. If, instead, frits are produced internally by the ceramic company, constant controls (every 2-4 cycles for rotary melting kilns or every 4-6 hours for continuous tank melting kilns with a capacity of 2-3 tons) are advisable. If considerable differences, visible to the naked eye, are found it may be necessary to increase control frequency. 9c) Colorant (calcined pigments and natural oxides) There are several colorant control methods. The most practical, reliable ones are:

330

Production controls

1 - Introduce the pigment in a percentage of 1% in a glossy zirconium opaque glaze. Then glaze the tile (always in parallel with the standard sample), fire and run a visual check on the results. 2 - Introduce the pigment in a percentage of 3% in a glossy transparent glaze. Then proceed as described above (1). 9d) Quartz and feldspars Introduce the raw material being analysed into a glossy viscous transparent or opaque frit in quantities of 15-20%. Glaze a tile, fire and run a visual check on hardening power. Example: Parts Parts Standard quartz 20 Quartz being analysed 20 Standard kaolin 5 5 Standard opaque glossy frit 75 75 The percentage of the raw material to be introduced into the frit depends on the employed frit, firing temperature and firing cycle. This value must be sufficient to bring the frit to the gloss-matt hardening transition point and allow evaluation of the contaminants of the raw material. 9e) Wollastonite Wollastonite (calcium silicate) is a low temperature matting agent. It is this matting property which needs to be controlled. This test is effected by introducing Wollastonite into a zirconium opacified glossy frit in percentages of 20-25% (minimum value in which it starts to matt, sufficient to evaluate contaminants). 9f) Kaolin Two types of control are recommended for this material: 1st test: suspending power test. Methodology: introduce 5% of kaolin into a frit, grind and empty into a graduated cylinder. Agitate and check sedimentation every 15 minutes. 2nd test: particle size distribution and purity test. Methodology: screen 200 g of kaolin (dispersed in about 1 litre of water) at 16,000 mesh/cm2, run a visual check for impurities, dry and weigh the residue. 9g) Zirconium silicates The aim of this test is to control opacifying power, an essential characteristic of this raw material. Percentages are established on a case by case basis as the minimum values needed for good opacification (usually 10% in a transparent glaze).

331


9h) Calcined or anhydrous alumina Two checks need to be run on this raw material: 1st test: particle size distribution control on different screens. 2nd test: hardening power test: introduce alumina into a zirconium opacified viscous frit in percentages of 15-20%. 9i) Corundum and Sands(Zirconium-Rutile-Quartz) The main control concerns particle size distribution, which must always be in line with optimal standard values. A test with 20-25% in the glaze is useful for contaminant evaluation. 9l) Carbonates: Ca - Mg - Ba - Sr Checks on these raw materials require measurement of the quantity of CO2. A test using 15-20% of the tested material in an opaque glossy glaze is useful to verify the presence of contaminants. 9m) Tin Cerium Titanium oxides Control their opacifying power by introducing them in percentages of around 510% in a glossy viscous transparent frit. 9n) Zinc oxide Check its matting power by introducing it into a medium-fluxing frit in percentages of 20-25%. 9o) Raw materials soluble in water or in any case used only in melting These materials are checked by means of grain size distribution or a melting test. Two laboratory melting cycles are carried out: in the first all the raw materials are standard and in the other all the raw materials except those being tested are standard. The thus-obtained frits are used to glaze two tiles. Check results visually. General criteria for quality control tests on raw materials for glazes All the above tests should be carried out ensuring that: the glaze composition should be sensitive to variations in the product being analysed. melting, grinding, application and firing conditions must be consistent. all results should be recorded on analysis charts (one per raw material), to be filed in chronological order of supply.

332

Defects

Appendix 3 DEFECTS

All production processes are interlinked. Items carry with them a history from earlier processes which will influence their subsequent performance. In this sense, firing, being the last of a series of processes (edge grinding and polishing aside) and the one in which key physical, chemical and structural changes take place, highlights both earlier latent defects as well as those induced by firing itself. Being able to recognise a defect, at any stage of the production process, and knowing how to identify its causes are skills that are largely acquired by experience: they depend on extensive hands-on experience and a capacity to see the whole production line from raw material storage to sorting as a single organism and require both global and specific knowledge of the production plant and a truly wide-spectrum understanding of ceramics. Acquisition of such abilities is essential to the rapid elimination of production problems; identification of a defect and rapid deduction of its origin allows manufacturers to act fast through carefully targeted machine adjustments and thus limit the costs associated with product rejection or declassing. In general (especially on fired tiles), defects are of the following sort: structural: flatness, orthogonality, cracks, laminations surface: wrinkles, holes, bubbles, particles, flakes, crazing aesthetic: stains, shade or decoration defects process: defects associated more with the handling and flow of materials than the product itself - overlapping tiles, chipping, sticking, jamming etc. While any defect can usually be traced to a specific stage of the production process, it generally has a series of concomitant causes, which, beginning with the choice of raw materials, contribute to its emergence. A rough summary of how such causes are interlinked is given in the following table (tab.1).

333


DEFECT PREPARATION Black core Size

PRESSING

DRYING

GLAZING

FIRING

CAUSE Low grinding residue Incorrect particle size distribution (grinding residue) High H2O absorption Low Grinding too coarse, insufficient bending strength moisture Stains, cratering, specks Screening or iron removal errors Black core Pressure too high or not homogeneous Small side cracks Ejection too fast or imbalanced Size Incorrect filling Dimensional defects Incorrect press settings Low bending strength Low pressing force. Low moisture content Explosions at kiln inlet Pressing force too high, lamination Cracks Incorrect regulation, high temperature Explosions at kiln inlet Residual moisture content high, incorrect regulation Glaze defects Incorrect drier outlet temperature Soluble salt stains Unsuitable drying cycle Cracks Unsuitable temperatures, tiles knocked along line Explosions at kiln inlet Excessive wetting Cracks in middle of tile Unfired tile stored too long Stains, holes, cratering Incorrect application, glaze not aged correctly Planarity defects Poor glaze-body thermal expansion match Colour, surface defects Engobe of poor quality Side cracks Pre-heating too fast Side cracks cooling cracks Cooling too fast Explosions at kiln inlet High kiln inlet temperature Planarity defects Poor temperature control Black heart, glaze defects Unsuitable cycle, erroneous atmosphere Size Incorrect temperature distribution Surface stains Accumulations of glaze condensate Craters, holes etc. Preheating temperature too high

Tab. 1. Possible links between defect and production department.

334

Defects

To help understand the causes and effects in the above table the following pages provide a more in-depth explanation of defects and how they correlate with the various stages of the production process. Defects associated with raw materials Defects attributable to the presence of impurities in the raw materials can be split into three main groups: Specks caused by the presence of impurities, usually particles of carbonacious minerals, pyrite, pyrolusite, carbon and mica. Specks caused by impurities absorbed during the manufacturing process such as particles of iron, drops of oil or grease or small fragments of rubber. More widespread defects caused by the presence of other kinds of impurities, such as organic matter and efflorescences. Carbonate particles (calcite and/or dolomite). With monoporosa products the presence of coarse particles of calcite and/or dolomite may lead to craters or holes. The larger the particles, the more extensive these defects. With single fire tiles, the defect depends not only on the presence and size of the such particles but also the melting point of the glaze. Those with low melting points tend to enhance the defect (fig. 1).

Fig. 1. Surface defect caused by coarse grains, degassing during firing.

The decomposition of minerals like calcite during firing is illustrated by the following reaction: CaCO3 > CaO+CO2 á 335


This reaction occurs at approximately 800-900 °C, depending on particle size. When particles are small, decomposition is complete and calcium oxide reacts with the other components to form calcium silicates or aluminosilicates (ghelenite, wollastonite, anorthite) of varying complexity. If, instead, particles are coarse, carbonate decomposition may be incomplete or, rather, the reaction of calcium with silica and alumina may remain incomplete. The post-firing presence of free calcium oxide may lead to expansion phenomena when the material comes into contact with moisture (expansion is caused by body components that have not reacted with CaO). The reaction is: CaO + H2O > Ca(OH)2 This phenomenon, where concentrated and localized, may form a hole of a very distinctive shape (fig. 2). When, instead, the defect is more generalized and diffused, latent tile expansion can cause adhesion failure of the tile after laying.

Fig. 2.

After grinding, it is possible to check for the presence of coarse calcite and/or dolomite particles by screening (e.g. via a 200 micron mesh) and then analysing the washed residue under a microscope. Adding a drop of diluted hydrochloric acid can also confirm the presence of carbonates. The exact nature of the particles can be determined precisely by analysing the area with an electron microprobe (SEM) or using more traditional chemical analyses. In this case, the exact nature of the elements in the agglomerate can be identified and the mineral can be identified. To eliminate or reduce the frequency and presence of such defects, it is good practice to minimize the percentage of calcite and/or dolomite particles larger than 120 microns. Using a suitable engobe as an interface between the tile body and the glaze may also be very effective. 336

Defects

Pyrite particles Pyrite, an iron sulphide, may be present as an impurity in some types of clayey materials. The most probable defect associated with coarse particles of this mineral is specklike colouring of the body and/or glaze (fig. 3). The decomposition reaction of pyrite during firing is: 2FeS + 7/2 O2 > 2SO2 + Fe2O3

Fig. 3.

Where thermal gradients are shallow, decomposition occurs between 400 and 500 °C. Where, instead, firing cycles are rapid, and if there is insufficient oxygen in the firing atmosphere, oxidation remains incomplete and different reactions may occur. In both slow traditional firing and fast single-firing, the defects associated with the presence of coarse pyrite particles take the form of atypical coloured specks. If, in the case of fast firing, the oxidation reactions remain incomplete, or if the formed sulphur dioxide causes condensate inside the kiln, problems may be caused by the gas attacking the vitreous surface of the tiles (brightness loss). This problem does not, though, depend on the morphology (size) of the pyrite particles but the total amount of pyrite and conditions inside the kiln. As with carbonate particles, coarse particles of this mineral can be identified by washing the grinding residue left on a 80 mesh (approx. 180 micron) screen and analysing it. In most cases, this defect can be prevented by lowering the size of the pyrite particles to below 180 micron through use of suitable slip preparation sieves. As above, an intervening layer of engobe between tile body and glaze may be effective in solving the problem. 337


Pyrolusite particles Pyrolusite (manganese dioxide) and goethite (iron hydroxide), when present as coarse grains, lead to the formation of spots on the surface of the glaze (fig. 4). As above, the particles can be identified with a microscope after screening and washing, again on a 80 mesh sieve. Final results can, as with the above minerals, be improved by using suitable slip screening sieves (i.e. those able to separate 180 micron particles). A layer of engobe between tile body and glaze may be effective in reducing the severity and frequency of this problem.

Fig. 4.

Carbon particles Coarse particles of carbon in the post-grinding residue can cause concentrated specks similar to those caused by carbonates. These defects stem from combustion of the carbon during firing and the formation of CO2 or CO as per the following reactions: C + ½ O > CO CO + ½ O2 > CO2 Fe2O3 + C > 2FeO + CO The carbon monoxide formed reduces the iron minerals in the composition and produces, together with silica and alumina, dark coloured glasses (fig. 5). In fast double firing, the defect may be arise from incomplete combustion of the largest carbon particles, which could finish burning during the second fire and so produce cratering or small dark holes. Identification of the carbon particles that cause the above defects involves washing and screening the grinding residue (at 230 mesh/63 micron) for consistency and then observing it under a microscope.

338

Defects

Fig. 5.

An approximate weight percentage of carbon particles can be obtained via calcination of the residue after washing (after removal of carbonates by acid, e.g. CaCO3). If it is not possible to replace the raw materials that are the source of the defect, it may be better to grind it finer and sieve it through finer screens up to 120 mesh (2400 mesh/cm2 equal to 125 microns). An intervening layer of engobe between body and glaze may also be useful in limiting the problem. Specks caused by impurities absorbed during the manufacturing process Iron and/or iron oxide particles Iron and/or ferrous oxides can contaminate raw materials during extraction or during the subsequent mechanical handling. Body contamination is also possible during preparation processes and can occur in mills, storage silos, transfer elevators and cooling systems etc. The defect itself usually takes the form of a spot or agglomerate on the surface. A useful problem identification feature is the texture (softness or roughness) of both these defects. A metallic particle, partially dissolved in the glaze, with a surface that is rough to the touch, probably comes from the kiln cooling zone; a metallic particle enveloped in the body or glaze-body interlayer will definitely have other origins (fig. 6). If, after macroscopic analysis and evaluation of the different characteristics of the defect (position, colour, surface appearance) the causes are still unidentified, microscopic analysis of samples removed from the defective areas is carried out. Analysing the spot with an electron microprobe provides comprehensive information as to the chemical nature of the sample, without which it is virtually impossible to identify which part of the production cycle is responsible. 339


Fig. 6.

However, attention should focus on the following: cleaning and tidiness (this can help identify contaminants) preventive maintenance preventive replacement of old plant components. Other impurities Other types of contaminant such as paint, grease, oil, wood or rubber particles can occur throughout the process (fig. 7). Where spray dried powders contain traces of unburned fuel (naphtha), the appearance of a defect, consisting of raised dots (some mm in diameter) over a certain area of the tile, is likely (fig. 8). This defect becomes more evident when fluxing glazes and firing cycles with very brief preheating times are used. As above, there are no general solutions to these problems, which will have to be analysed and the appropriate action taken.

Fig. 7.

340

Defects

Fig. 8.

More widespread defects caused by the presence of other kinds of impurities, such as organic matter and soluble salts. Organic material Clayey materials may contain various percentages of organic matter up to 5% or more. When ceramic tiles with compositions containing a significant percentage of organic matter are fired, a dark centre known as black core may appear (fig. 9). In single fire floor (and wall) tiles this defect often occurs when very fast firing cycles or very high pressing pressures are used.

Fig. 9.

Silk-screen printing can augment the defect as it involves the use of organicbased vehicles (glycols or polyglycols), thus increasing the amount of organic matter. As stated above, black core problems depend both on the firing cycle and the characteristics of the raw materials. This chapter mainly focuses on how the problem is influenced by the nature of the raw materials and the impurities contained in them. The black core problem results in a multitude of effects: tile swelling (bubbles in the interior) pyroplastic deformation deterioration of the technical and aesthetic characteristics of glazes shade changes in the body and glaze. 341


Tile swelling Depending on its severity the defect is either localised or present right across the tile (fig. 10). It mostly affects vitrified materials and is caused by the low temperature formation of vitreous phases in the body such as the reduction of ferric to ferrous oxide in red firing materials or surface melting, which prevents the escape of the gases that develop inside the piece (CO2 and CO), and areas of the surface are deformed.

Fig. 10.

If these organic materials oxidize prior to formation of the vitreous phase, the defect is not seen. With white firing materials, instead, where there is little iron in the body (< 1%), deformation is largely caused by the presence of gases (CO2 and CO) that are unable to escape because of melting at the surface (or the use of glazes with a low melting point). In addition to swelling, trapped gases inside the tile may form reduced phases close to the glaze surface grey or black in colour. Pyroplastic deformation Shows up as both tile geometry and flatness deformations of varying severity. As mentioned above, black core defects derive from the joint presence of a liquid phase and trapped gasses. It can also be said that in areas where black core appears the percentage of liquid phase is higher and viscosity is lower than in other areas of the tile. For that reason, pyroplastic deformation occurs more often in vitrified products where porosity is lower than 3%. Deterioration of the technical and aesthetic characteristics of glazes A phenomenon that frequently accompanies black core is the appearance of a considerable quantity of gas bubbles in the glaze layer (fig. 11). This considerably reduces abrasion and stain resistance on account of widespread microporosity in the glaze and worsens pin holing. These gas bubbles are caused by gases (CO, CO2, H2) in the glaze layer that were generated by the oxidation reactions of organic matter originally present in the tile.

342

Defects

Fig. 11.

Shade changes in the body and glaze Where the firing environment is not oxidising enough, the formation of black core may, in both floor and wall tiles, alter the colour of the product. Colour changes stem from the reduction of the iron oxide in the body composition, shifting the red shades, typical of this oxide, towards green or brown. With wall tiles containing a high percentage of calcium, however the colour changes from burnished orange to greenish beige. With single fire tiles the change in colour may stem from different degrees of oxidation of any copper, manganese and iron in the glazes, causing a shift towards duller shades. Dynamics of black core formation Black core defects are usually attributed to the presence of carbon residues from thermal decomposition of organic matter in the 300-800 °C temperature range. Some researchers have suggested that it is the presence of reduced iron in the grey and black zones of the tile centre that characterizes the defect. Moreover, they have recently come to the conclusion that the components which cause black core are elements such as reduced carbon and iron oxides (FeO and Fe3O4). The former is formed by pyrolysis of organic matter, the latter by reduction, mainly of the carbon, from Fe2O3, present in the initial clay compositions. The chemical reactions, as well as the sequential dynamics of the black core formation are as follows: Loss of clay crystallization water Hydroxyl (OH) groups, which are freed at different temperatures depending on the kind of clayey mineral, can act as reducers.

343


Thermal decomposition and reaction of organic materials As stated earlier, the organic matter dispersed in clays mainly consists of humic acids and carbon. When these materials are heated CO2 develops (below 500 °C) as the molecule is split by the following reactions: R-COOH > RH + CO2 (g) R-OH > RH + H2O (g) where R-COOH R-OH RH g

= = = =

organic molecule with carboxyl functional group organic molecule with hydroxyl functional group product of the decomposition reaction gaseous substance.

If heating continues, the resulting solid residue (coke-like coal) reacts vigorously between 500 and 700 °C and produces a nuclei polycondensation reaction. Above 700 °C the following reactions may occur: 2 (R-H) > R-R + H2(g) R-H > R + H2 (g) where R-H = is the result of the initial decomposition reactions R-R = is the result of the dehydrogenation reaction. The clayey materials are essentially aluminosilicates capable of initiating a catalysis which aids the pyrolysis reaction that occurs in the 400-500 °C range. Hence the following reactions between carbon oxide produced by decomposition of the organic matter and any other available gases: C(s) + H2O (g) > CO (g) + H2 (g) CmH2m+2 > mC (s) + H2 (g) where CmH2m+2 = a hydrocarbon formed from the decomposition reactions s = solid material. Without oxygen, the steam generated by previous reactions may combine with hydrocarbons or the existing free carbon. Oxidation temperature and rate will depend on the type of carbon and on the organic matters in the clayey raw materials. As a consequence of the above reactions, when clayey mixtures containing organic impurities are progressively heated, first the clay crystallisation lattice water

344

Defects

and then, starting from 550-600 °C, a mixture of gases made up of H2O, CO2, CO and H2 are released. Reduction of iron oxide Iron is usually contained in clayey materials as hematite (Fe2O3) and less frequently as goethite [FeO (OH)], limonite or carbonate. Iron carbonate decomposes at 460 °C as per the following reaction: FeCO3 (s) > FeO (s) + CO2 (g) At approximately 400 °C iron hydroxide changes into ferric oxide, which is stable in an oxidizing atmosphere (assuming the partial pressure of oxygen to be higher than its decomposition pressure). Without oxygen, hematite can, instead, progressively decompose and lose oxygen: Fe2O3> hematite

Fe3O4> magnetite

FeO> wustite

Fe (7) metallic iron

On the other hand, Fe2O3 iron oxide, with carbon oxide and hydrogen (CO and H2) can be reduced even further at temperatures above 550 °C. Working conditions that help eliminate black core problems Back core problems can be reduced by: Avoiding the use of high percentages of raw materials (clays) containing organic substances and/or inorganic compounds such as (bivalent) Fe. There is no clear-cut upper limit for these materials as development of black core depends on many factors (firing cycle, tile size and thickness, pressing pressure, type of glaze used). Formulating compositions that give the ceramic tile sufficient permeability to improve oxidation reactions during the pre-heating phase. In this case too, coming up with specific solutions that aid the development of high permeability lattices and thus the escape of reaction gases is no easy task. Nevertheless, the production process must adopt all those measures, where compatible with other parameters, that help bring about this situation. Optimisation of the firing curve Last but not least come in-kiln firing conditions. With some materials the prevention of black core requires that the firing curve includes a zone of constant or slowly rising temperature between 300 and 700 °C. This aids decomposition and oxidation and also facilitates outflow of gases before the glaze melts and traps them. Bear in mind that the above temperature ranges may shift depending on body characteristics and the glaze softening point. 345


Modification of the kiln atmosphere in the pre-heating zone It has been demonstrated that increasing the partial pressure of oxygen in the pre-heating zone reduces or even eliminates black core; this can be done by increasing the volumes of primary combustion air being fed to the burners. Efflorescences Efflorescences are salt concentrations which appear on the surface of fired tiles when more than a certain percentage of soluble salts is present in the body. High concentrations are caused by migration of salty solutions through the porous lattice of the tile, causing their accumulation, over-saturation and precipitation in areas where water evaporation is fastest. Soluble salt concentrations remaining equal, the problem is more evident on tiles with more permeable bodies. Soluble salts have various origins: they may be an intrinsic element in the raw materials or form during firing as chemical reactions take place among the different components or between the latter and the combustion gases surrounding the tiles. Generally speaking, the soluble salts that cause efflorescences are already present in the clayey raw materials or originate from: gases produced by combustion (SO3) water used in production (sulphides, chlorides, nitrates). In addition to the most soluble sulphates and chlorides in the raw materials, even low-solubility sulphates such as Ca and Mg and others can cause efflorescences. Manufacturing defects attributable to the presence of soluble salts appear as yellow-brownish stains, holes or bubbles on the surface of the glaze or, in some cases, as detachments of the glaze from the body (fig. 12). Yellow-brownish staining may be caused by the presence of iron and manganese soluble salts.

Fig. 12.

346

Defects

Sometimes the above defects also depend on: hygrometric state of the ambient air: high levels of humidity and high ambient temperatures tend to concentrate the salts on the perimeter or the corners of the tile. firing conditions: the in-kiln dynamics of air volumes and combustion air turnover are critical factors in eliminating/limiting aggression on the glaze. The importance of eliminating or minimising the raw materials containing these salts is self-evident. If this is not possible corrections may be attempted as follows: try to precipitate the soluble salts in the form of insoluble salts with double exchange reactions. In the case of sulphates the addition of barium carbonate during grinding may produce good results. Added amounts of these salts must be batched stoichiometrically vis-à-vis the percentages of the already present salts so as not to worsen the situation. in double firing: fire the biscuit at higher temperatures to aid fixing of SO3, Cl ions in the ceramic body (this can reduce efflorescence but sometimes worsen glaze defects). vice versa, reducing the biscuit firing temperature can, in some cases, increase its porosity. This greatly reduces glazed tile defects because, as the water evaporates, the soluble salts are concentrated inside the biscuit and do not affect the glaze surface. Problems attributable to the technological characteristics of the body A proper treatise of this complex subject is beyond the scope of this volume: problems associated with technological aspects of the body differ widely depending on the product (floor or wall tiles) and the employed technology (double or singlefiring). Nevertheless, the following pages summarise some of the more general concepts associated with the various product families below: A) White firing vitrified floor tiles B) Red firing vitrified floor tiles C) Porcelain floor tiles D) White firing porous wall tiles E) Red firing porous wall tiles. Product families A, B and C are generally obtained through single fire technology. Families, D and E can, instead, be produced using either double or single fire processes; the problems will differ. The most frequent problems affecting the different body types are:

347


Body deflocculation problems Deflocculation difficulties, with all product families, may necessitate the use of higher percentages of water at the body grinding stage. Such problems may be caused by: the nature of the minerals in the clays very fine particle-size distribution of clayey particles organic materials (humic acids or other) in excessive percentages high percentages of soluble salts. Pressing problems Affecting all groups. While actual defects may be different their technological causes are generally similar: mineralogical origin of the clays making up the bodies excessively fine particle-size distribution of the clays or other minerals particles flat (two dimensional), acicular or fibrous plasticity or non-plasticity of the body causing: lamination or peeling low bending strength values dirty moulds. Lamination involves powder stratification, which prevents the release of the air contained in the powders themselves and therefore inhibits pressing performance. Low post-pressing bending strength results in frequent tile breakages in both double and single fire plants. Excessively dirty moulds result in costly cleaning-related stoppages, common with steel dies but relatively rare with rubber-lined or resin ones. Drying problems These differ depending on production technology (single or traditional double firing). However, their causes are generally to be found in: the nature of the clayey minerals the excessively fine particle-size distribution of the clays atypical behaviour of pressed tiles during the drying phase (high expansion and low bending strength). These aspects or conditions of the raw materials making up the body may lead to the following production problems: excessive shrinkage during drying low post-drying bending strength. The defect is more evident where single fire technology is used as the tile is exposed to more mechanical stress, especially on the glazing lines. 348

Defects

Generally, the bending strength of a ceramic product after drying should be higher than 25 kg/cm2. Other defects are influenced by: drier management (drying curve) moisture content of tiles entering the drier hygrometric / general conditions of air. Problems associated with the nature of the raw materials that appear after firing Like much of the previous subject matter, an in-depth treatise of these problems would require a book in itself. The variables that depend on sintering or vitrification are many and their interaction is highly complex. Hence only a general description is given. Good vitrification or sintering of the body evidently depends on the quality of the raw materials, their ratios and reactivity among the different components (oxides). It is obvious the characteristics such as: high bending strengths low shrinkage product-compatible porosity wide firing range mainly depend on the origin of the raw materials and their reactivity. Low dimensional stability during the firing process is, without doubt, more frequent in vitrified materials (groups A, B, C) where dimensional variations are more evident. The associated problems are ones of geometry (orthogonality), dimensional variation, sides (squareness) etc. The narrower the firing range, the more evident the defects. While the origin of the problem may also lie in the production process (pressing, firing etc.), it is the nature of the raw materials (i.e. the organic content and clay particle size) and thus body composition that play the key role. The stability of a vitrifiable composition is, in fact, regulated by the extent of the vitreous flows being formed and their viscosity. It is known that white firing bodies have wider stability ranges than red firing ones because of the nature of the glasses being formed and their higher viscosity. The phenomena is highly complex and dependent on many interlinked chemical-physical factors. Body composition or, rather, the nature of the raw materials, has much influence on the severity of the defect. Other defects associated with raw materials a) Excessive drying shrinkage This usually causes several small edge cracks some 1 to 1.5 cm long, many usually appearing on each side. This defect is typical of many high-plasticity clays and generally appears because the low bending strength of the green product is unable 349


to withstand the mechanical stress caused by shrinkage. For most pressed tiles the dry shrinkage limit is around 0.25-0.30%. This defect may result from clays extracted from different parts of a quarry or from their insufficient homogenisation. Shrinkage can be checked via simple laboratory tests. b) Low green tile bending strength This defect tends to cause frequent breakages in both single and double fire products either when tiles are turned over or loaded into the rapid driers. However, the resulting unfired waste can be reintroduced into the mill and is fully recyclable. To obviate this problem bending strength should be no lower than 0.7-0.8 N/mm2. c) Low dried tile bending strength Causes frequent breakages at critical stress points on glazing and conveying lines. Dry bending strength should normally be no lower than 1.8-2.5 N/mm2. The defect has its origins in the clay percentage or low plasticity. d) Excessive free quartz An increase in the percentage of free quartz in a body formula increases its expansion coefficient, the alpha-beta inversion at 575 °C and the corresponding mechanical stress. Dilatometric analysis will identify quartz variations. If the quartz does not react (and change into silicate) during firing, coolinginduced creep can also occur. e) Low fired bending strength Results in inferior finished product resistance and customer complaints of wastage during tile laying. Insufficient reaction and sintering of the clay and non-clay components in the body lie at the heart of this problem, which may arise because of changes in either percentage or particle size distribution. As with the above defects, routine checks on technological characteristics generally prevent this problem. Body preparation defects a) Insufficient grinding The effects of insufficient grinding differ depending on whether the body has a single component or is an actual compound. In the former case there is a slight change in surface appearance and reduced bending strength, in many cases only slight and difficult to detect. The latter case is a more serious affair, with coarser particles delaying important reactions (i.e. carbonate degassing). 350

Defects

The usual outcome is that the tile falls outside required physical-mechanical tolerances. Prevention and correction of this defect generally involves checking slip sieving residues and, when possible, the particle size distribution pattern. b) Inconsistent moisture content in powders to be pressed This defect usually occurs when the powder to be pressed has not been mixed in draining it from the various silos or stored long enough for moisture content to homogenise throughout the silo. The phenomenon is more frequent in dry grinding processes. The first effect is a tendency towards lamination; green and dry bending strengths are also found to be inconsistent. With spray-dried powders the problem is less serious because the moisture is concentrated inside, not outside, the particles. Moisture control is usually performed on powders sampled from different points of the conveyor belt over a time span of approximately one minute. Differences in moisture content can be reduced by storing spray dried products for at least 24 hours and dry ground wetted ones for 36-40 hours. c) Excessive moisture content in powders to be pressed The first effect of excess moisture is that the powder tends to stick to the die, thus requiring more frequent mould cleaning. Note that in fast fire processes excess powder moisture causes, indirectly, black core. d) Low moisture content in powders to be pressed Low moisture content makes pressing more difficult and can lead to lamination and de-moulding problems. Note that powders with a low moisture content tend to have higher pressing expansion. This increased expansion results in greater friction against the die liners when the lower punch ejects the tile. It can also cause low green bending strength and subsequent breakage later on in production. e) Excessively fine particle size of the powder to be pressed Causes lamination defects as a result of the low bulk density of the powder and the considerable air that needs to be expelled. f) Excessively coarse particle size This gives a rough tile surface with a granular appearance. Further troubles include frequent dirtying of the dies. g) Particle size separation Particularly seen in the production of porcelain tile, this problem occurs where particle segregation in the press hopper causes the outer die cavities to be filled with grains coarser than those in the inner cavities. The defect is extremely detrimental when various coloured powders are mixed (e.g. salt and pepper). 351


Generally speaking, it is necessary to use spray-dried powders of uniform particle size distribution (in the 0.3-0.5 mm range). h) Encrustation lumps These appear because condensate forms in the storage silos, the elevators and the hoppers. As moisture content and density in these lumps differ from those in the rest of the powder, their firing behaviour is also different. In the most evident cases, small cracks form around the lump on account of shrinkage differences. In less evident cases small depressions form on the surface of the glaze because of lower absorption in the areas affected by the lumps. To eliminate the defect the powder is passed through finer mesh sieves prior to the press feed point and efforts are made to prevent the lumps forming in the first place. Pressing defects Dirty dies The problem shows up as surface defects on the just-pressed product and can remain after glazing and firing (fig. 13, left). The quality of the die surface is highly important as any wear-induced roughness causes powder to stick to the die more easily. Rubber-coated dies help prevent this problem and can drastically reduce the frequency with which cleaning needs to be carried out.

Fig. 13.

352

Defects

Lamination The defect consists of foliated stratification of the material that can sometimes be seen only after glazing and firing (fig. 14).

Fig. 14.

It has several possible causes: 1) When the first entry speed of the punch is excessive and does not allow air to escape from the powder. 2) When the punch lifting time between the first and the second entry is too brief to allow escape of the air. 3) When die construction is erroneous and clearances (which should take into account the working temperature of the die) between the liners and punches are too narrow. Defects caused by ejection difficulties This defect usually appears as: cracks running from the sides to the middle of the tile, but without necessarily affecting the edges themselves (fig. 15). planarity defects. There are various causes: 1) The upper punch moves too fast during ejection. 2) The die liners are not properly tapered, thus preventing normal expansion of the piece. 3) The interval between the end of pressing and the beginning of ejection is too long. 4) Excessive mould wear. 5) Unsuitable back pattern. Flash on tile edges (residues compacted between liners and punch) This defect is generally observed at the end of the brushing and fettling opera353


tions carried out on the collecting station. It can occur on both the face and back of the tile. The cause lies in excessive mould wear. The flash that forms during pressing on the edges of the surface to be glazed is usually removed by brushes, while the larger flashes on the back of the tile are removed by scrapers. Small bits of flash sticking on the surface to be glazed can cause visual defects and selection losses (fig. 13, right). Filling defects Non-uniformity of filling depends on filler box regulation and causes different densities at different points on the tile usually detected by thickness checks and bulk density measurements with either a penetrometer or mercury densometer.

Fig. 15.

Excessive pressure In the fast single fire process excessive pressure can cause black core. The particles are packed so tightly that the gas produced by the (sometimes incomplete) combustion of the organic substances is unable to escape. Glaze and glazing defects There follows a description of the most commonly encountered defects. Defects caused by the glaze a) Excessive or insufficient grinding Over-grinding has a negative effect on glazes with very high surface tension such as zircon whites where shrinkage causes cuts or cracks to appear on the glaze before melting. The effects of unusual particle size distribution are many. For example, an overground rustic corundum glaze will have a non-vitrified surface on account of excessive refractoriness, while excessive grinding of a glaze to which zircon sand materials have been added also produces cracks that do not close during melting. 354

Defects

Considerable changes in surface appearance aside, excessive grinding always produces a glaze of higher water absorption capacity, thus causing drying problems. When glaze grinding residue is too high (materials ground for a short time) the glaze is difficult to sieve and normally has low viscosity and settles easily. Problems can be avoided by screening and checking residue at all mills before the material is used in production. b) Lack of plasticity Plasticity, where it refers to a glaze, is an improper term, yet one widely used to indicate its capacity to adhere to a tile body. The problem is often difficult to detect on the green tile as it generally appears during the initial stages of firing, when the glaze first starts to shrink and then melt; because it does not stick to the body properly, it tends to crawl away from the body. In single-firing, adhesives of the C.M.C. type often need to be used. This adhesive is also useful in resolving glaze detachment problems in the double fire process. Plasticity can also be improved by using ball clays, kaolins (these can be added in percentages of 8-10%) and bentonites (up to maximum 2%). The above materials, in addition to exerting a plasticising action, also play an important role as suspension agents. Proper suspension is important in that it ensures consistent application density and prevents separation of the larger particles by settling. c) Poor firing stability Some glazes are highly unstable, sensitive to very small temperature or firing environment changes. This sensitivity often results in tiles of different shades. d) Excessive fusibility The origin of this problem lies in the composition of the glaze itself. Excess fusibility can cause surface bubbling (pin holes) at the tile sides or where the glaze layer is thinner. e) Incomplete frit melting A problem that clearly has its origins in the frit and normally causes two types of defect: 1) surface bubbles and other surface defects caused by the escape of gaseous products that have not fully decomposed during melting. 2) partial solubility of components that have not chemically bonded completely. It is thus important that the supplied frit be subject to in-factory quality control checks.

355


Application defects a) Glaze detachment The fired tile has areas of varying size where the glaze has crawled away from the surface during firing. These zones are generally surrounded by a rounded, thickened border. Such glaze movements can sometimes appear on the sides or edges of the tile; in other cases they take the form of extensive, thickened slicks in the midst of unglazed zones or may appear as circular holes of varying dimensions. Factors in preventing this problem include the quantity of plastic material in the glaze, application thickness, glaze-body adhesion, low grinding times and the unfired strength of the glaze. b) Glaze shrinkage Tensions between ceramic body and glaze may generate small fractures in the latter. Subsequently, where the melted glaze is highly viscous, these fractures are not sealed; typical of majolica with a high zirconium silicate content, the problem is much less frequent in most lead compound glazes. Moreover, where melted glaze surface tension is high, the glaze shrinkage problem is proportionately high. Glaze shrinkage (by way of surface tension) is attributable to: improper type or quantity of plastic material inclusion of special components. The presence of materials such as zinc oxide, talc, magnesium carbonate, marble, aluminium hydrate, all characterised by extensive drying and firing shrinkage, may be the source of such defects. These materials hold significant quantities of water and influence glaze viscosity. Over-grinding Excessive grinding increases glaze shrinkage and causes loss of adhesion. Presence of soluble salts The use of frits or raw materials of marked solubility can lead to phenomena similar to those described earlier. The defect tends to appear when manufacturers use very fine-ground glazes that have been left in slip form for some time. Overlapping applications Glazes not normally subject to shrinkage may produce this problem when: the base glaze is dusty or prevents the subsequently applied glaze adhering properly. the second application re-wets the just-dried first layer and causes it to lift off the body. the two glazes are incompatible on account of surface tension and viscosity during firing. 356

Defects

c) Interaction with production processes Note that these defects often have joint causes and may originate from a combination of several factors independent of the glaze in itself. Poor glaze-body match Incompatibility between body and glaze expansion coefficients can cause serious defects. During cooling molten glaze interacts with the body; if the thermal expansion coefficients of body and glaze differ considerably tensions may arise, causing cracks in the now hardened glaze (crazing-flaking). a) Crazing and flaking If the thermal expansion coefficient of the glaze is higher than that of the body, the glaze, as it cools below the transformation point, contracts to a relatively greater extent and thus stretches. If the resulting tension is higher than the modulus of elasticity of the glaze, breakage occurs. Known as crazing, this defect appears as hairline fractures. Vice versa (i.e. when, during cooling, the thermal expansion coefficient of the body is higher than that of the glaze), the glaze is compressed. This causes it to fracture into numerous flakes that detach from the tile along the fracture lines, hence the term flaking. Flaking, nevertheless, is a relatively infrequent phenomenon as compression resistance is some ten times greater than tensile strength. Experience has shown that it is actually preferable to have glazes that are slightly compressed as this helps contain any increase in body volume caused by moisture absorption prior to, during and after laying. Quite often, crazing appears as a series of localised fractures of circular shape, mainly at the sides of the tile. b) Glaze-body interaction Also deserving of mention are the expansion phenomena caused by interaction between body and glaze. Acceptable reactivity between body, engobe and glaze is indispensable if they are to adhere to each other properly and avoid separation due to edge impact, more frequent with double fire than single fire tiles on account of their lower reactivity. Glazed surface defects Tile surface defects take a multitude of forms. They rarely have one clearly definable cause but generally stem from a number of interacting negative factors. These defects rarely affect the functionality of the tile but can seriously compromise its appearance. 357


a) Pin holes on the glaze These are a frequent problem and one of the most difficult to eliminate characterised by post-firing minute holes, hollows or craters scattered across the surface in varying concentrations. The holes are created during firing by gaseous bubbles bursting through the surface tension of the molten glaze. Under certain conditions this phenomenon does not have any lasting effects as the molten glaze automatically flows back into the pit and neatly seals it. The key gas escape factors are viscosity and surface tension. Glaze viscosity at its maturing temperature is an important characteristic as it determines the capacity of the glaze to spread over the body and form a uniform layer. More fluid glazes allow more complete, faster evacuation of the gasses entrained in the vitreous mass. Firing conditions being equal, high viscosity glazes are far more susceptible to large-size pin holing. Surface tension also pays a key role: higher tensions tend to aid re-absorption of residual bubbles during cooling. Nature and formation of gaseous emissions a1) Defective fritting Some glaze constituents do not react perfectly during preparation of the frit (especially at the melting point) and thus retain a certain instability, re-interacting with the other constituents of the vitreous mass during the second fire and generating gas. Frit instability or heterogeneity is particularly deleterious in rapid firing cycles as the available reaction times during the second fire are very short indeed. Consequently, gasses may generate defects on the molten surface that do not disappear at the cooling stage. a2) Additions in the mill With rapid firing cycles and where materials need to be added to the mill, it is important not to use raw materials with degassing or reactivity characteristics that clash with firing duration and/or temperature. Failure to observe this rule can lead to dot-like craters, holes and pin holes etc. a3) Over-firing Over-firing of the glaze can cause pin holing. Glazes with a high alkaline or boron oxide content are particularly susceptible to this defect. These oxides, in fact, tend to volatilise if over-fired, thus giving rise to the offending bubbles. a4) Salts These are usually sulphates and carbonates (especially soluble ones), which can originate from the body, the grinding water, the pigments, additives or the glaze itself, and can cause pin holing. 358

Defects

Specks in the glaze may have accidental causes such as: Contamination of the glaze during the production cycle Air retention during application Imperfect drying Incomplete firing.

b) Depressions (dimples) Glossy glazes, especially zirconium whites, often have their surface appearance marred by evident dimples. Appearing as a shallow circular hollow, the defect is rarely more than 5-6 mm in diameter. Such imperfections may have their origins in frit grains of large particle size, usually the result of imperfect sieving. During melting these grains behave differently from the rest of the vitreous mass. On double fire tiles depressions may also be caused by significant variations of water absorption in the body. c) Relief grains Grains standing out from the surface of the glaze are frequently a consequence of contamination that occurs during the production process. The defect may also stem from an in-mill addition of zirconium silicate or other hard materials that have not been suitably ground. In this case the defect is characterised by easily visible, small, white, protruding specks. d) Loss of gloss or mattness Finished tiles sometimes lack the intended shine; instead, their glossiness varies from one part the tile to another. In other cases the problem appears as variations in intensity or shade of colour. Where a highly fusible glaze is applied on a particularly porous body (especially where the latter is highly siliceous) it can react so intensely with the body that nearly all of it is absorbed. This aesthetically weakens the vitreous phase in the glaze to the extent that the final appearance is significantly altered. Concentration of those non-vitreous, low-reactivity compounds in the glaze usually gives rise to opaque and matt surfaces. This defect is frequently accompanied by the appearance of numerous tiny holes on the surface of the glaze, giving it a sponge-like texture. We have seen how an excessive tendency toward volatilisation of the alkaline oxides can cause dots to form. Glazes sometimes have undissolved phases, usually made up of silica or oxides (calcium, barium, aluminium, magnesium, tin or chromium) as well as spinels. If these usually opalescent phases are present on the surface or just beneath it, the glaze may occasionally be characterised by pseudo-mattness. Devitrification is a rather complex process and often creates problems. Surfaces intended to be uniformly matt may have relatively glossy patches or, vice versa, 359


glossy glazes may be marred by duller areas. Earlier on we stated that devitrification, while dependent on glaze composition, is also heavily influenced by the firing cycle and, even more so, the cooling pattern. Generally speaking, where manufacturers using conventional cycles intend to prevent devitrification, the ratio between the SiO2 and basic oxides in the glaze should be 3:1. Nevertheless, even where this ratio is maintained (and even where ratios theoretically more efficient in preventing devitrification are used) the problem can still arise. This is because not all basic oxides behave the same way. Excessive calcium, magnesium or zinc oxide contents are notorious causes of devitrification. Moreover, the behaviour of the various devitrifying oxides during cooling differs enormously. Alumina matt glazes, for instance, are highly sensitive to temperature differences at this stage. Hence these glazes or those with high ZnO, CaO or TiO2 contents have a tendency, under certain conditions, towards non-uniform surface mattness or glossiness. The most common scenario is where the matt finish varies from one tile to another, an occurrence largely due to variation of the firing curve in different sections of the kiln. Slower cooling, in fact, gives complete devitrification while higher firing temperatures can increase shininess. This explains why gloss, satin and matt glazes all appear sensitive to devitrification phenomena when firing cycles are unsuitable. Other phenomena are closely correlated with devitrification problems. One of the most common is found in single layer rapid firing kilns where the leading edge of the tile has a markedly different appearance compared to the rest of it; with matt or crystallised glazes it looks shiny or bubbled. While evidently connected to cooling especially direct cooling this defect stems from the extreme sensitivity of these glazes vis-à-vis variable cooling temperatures. In such cases the thermal inertia of the kiln walls or even the tile mass itself may be sufficient to cause that non-uniformity. Depending on circumstances, these differences can then give rise to an overly rapid increase in viscosity with consequent bubbling, lack of devitrification and the appearance of shiny areas on the tile. e) Sulphur The damaging effects of sulphates contained in the tile body, the raw materials, the grinding water or the sulphur compounds in kiln fumes give rise to a whole series of finished tile defects. The most common are opalescence, surface halos, pin holes, bubbling, devitrification and creasing (wrinkling) at the sides. These defects usually appear more frequently when highly alkaline compositions are used. The alkali are usually introduced into the glaze as frits. Nevertheless, when these frits are insufficiently rich in silica, boric oxide or alumina, or if they have a certain reactivity due to incomplete fritting, the alkali may remain open to attack by any sulphur oxides. Reactions of the above type (i.e. determined by sulphur in the composition or the kiln fumes) are also likely to occur during fritting, although tests have yet to confirm this hypothesis. 360

Defects

The amount of sulphuric and sulphurous oxide contained in the combustion fumes varies depending on firing conditions. Oxidising atmospheres, particularly during the initial stages of firing, tend to accentuate sulphuric oxide production. The concomitant presence of high quantities of water vapour (from removal of water from the tiles and decomposition of hydrates) can form sulphuric acid. This attacks any basic components of the glaze (especially lead, calcium, magnesium) not securely bonded to the silica or other acid components of the glaze. Sulphates extremely damaging as they do not decompose easily at high temperatures thus form. These salts generally tend to form thin films on the surface of the tile, giving rise to devitrified areas. When these compounds are concentrated (e.g. at the edges), creasing or bubbling occurs. The phenomena described thus far could be detrimental in rapid double firing. The influence of stains The introduction of colouring oxides can generate defects on the surface of the glaze after firing. Such phenomena can generally be traced to unsuitable or incomplete pigment preparation, or interaction of the latter with some glaze components. A description of the most common production plant problems follows. a) Volatilisation Some colouring oxides are subject to volatilisation phenomena, giving rise to numerous defects on the tile surface during firing. Chromium oxides, for example, are used extensively in ceramics on account of their excellent colouring power and relative stability. Nevertheless, pure chromium oxide has a strong tendency to generate halos because of the sheer intensity of its colouring power, despite the fact that its vapour pressure is low and its volatilisation ratio not particularly high. Halos can be eliminated via calcination and subsequent oxide washing. b) Calcination Calcination is a key phase in the pigment preparation, its purpose differing according to the raw materials in question and the final product type. Generally speaking, it can be said that, in its simplest form, calcination aims to decompose the constituent components of a colour at high temperature, causing them to release gas. If noncalcinated colours are used in glazes they inevitably cause numerous defects such as specks, bubbling, halos etc. Where combined with the formation of crystalline lattices, the calcination function is a much more complex one. These lattices are the product of well mixed raw materials; during calcination the latter give rise (with the aid of catalysts) to crystalline neo-structures inside which chromophore ions are present by way of ionic exchange. The calcination process can also be used to obtain solid oxide solutions with 361


improved characteristics as regards the original, individual components. A typical example of this is calcination of chromium oxide and aluminium hydrate to obtain a solid solution of the two oxides; this improves stability and reduces volatilisation of the chromium, especially at high temperatures. c) Washing After calcination the colour may still contain a certain quantity of salts subject to volatilisation. To eliminate these the finely ground oxide is washed, using water and additives that aid leaching of the soluble compounds. In the above-described case washing removes that part of the chromium oxide which would otherwise fail to enter into formation of the solid solution and would be water-soluble, its diffusion subsequently causing halos. Washing is also indispensable when calcination aims to create a crystalline structure and uses excess amounts of catalyst and chromophore salts. The production of zirconates is a classic example. With, for instance, turquoise zirconate, the process begins with zirconium oxide, quartz, vanadium pentoxide and catalysing agents. At the end of calcination the crystalline zircon lattice forms, with ionic diffusion of vanadium ions. This is usually achieved by introducing an excess of vanadium pentoxide. This, together with the catalysts only, needs to be washed out after calcination. If the catalysts are not removed they can reduce the colouring power of the oxide and, because they are usually soluble salts, can cause the relevant defects when added to the glazes as described above. Equally detrimental are vanadium salts that have failed to bond inside the lattice. As vanadium is a powerful fluxing agent, concentration of its salts during drying generates aggregation zones; when fired, the latter have different fusibility characteristics compared to the surrounding vitreous mass. Hence glazes with a high percentage of colours containing vanadium salts may be damaged by post-firing dimples and non-uniform colour. If the surface is screen printed reactivity is accentuated and some zones may be poorly coloured. d) Reactions with contaminants While colouring oxides generally have a certain inertia vis-à-vis the glazes in which they are dispersed, their sensitivity to the chemical composition of the glaze is far from negligible. This tends to highlight defects caused by colours reacting with glaze components or, more frequently, contaminants. The behaviour of red-pink oxides with a SnO2 - CaO - SiO2 - Cr2O3 composition is a classic example. These oxides are extremely sensitive to reduction. The presence of reducing contaminants such as carbon, iron or other metallic particles can cause alterations to the colouring lattice; the chromium chromophore ion is separated from the crystalline matrix by the reducers, thus freeing the tin oxide contained in the lattice. Consequently, the contaminated area is completely de-coloured. The surface of the finished tile is thus dotted with small white specks caused by the opacifying action of the tin oxide. Where the contaminant particle is quite large a dark spot appears at the centre 362

Defects

of the affected zone. These defects are usually caused by inefficient removal of iron from the glaze. Use of magnets and the addition, in small percentages, of oxidants such as cerium oxide usually resolve the problem. Glaze application defects a) Drops falling on tiles One of the most evident and frequent defects where glaze is applied in disc or spray gun booths, it is caused by the fall of drops of glaze condensate; these can form at any point on the inner wall of the booth, often because the air exhaust suction is insufficient and the pulverised glaze thus condenses on its inner walls. The defect can occur even where pressure is very low. Likewise, with discs, the defect may be caused by condensate forming directly in the disc unit because of imperfect atomisation. The problems may also be caused by incorrect positioning of the protective screens, which, instead of channelling the glaze, allow it to splash and form drops that inevitably fall onto transiting tiles. Vibration transferred to the booth by other items of machinery contributes to the problem. Wet drops can generally be reduced or eliminated by increasing suction rates and by keeping all exhaust ducts and filters clean so as to maintain efficiency. b) Glaze clots (dry agglomerates) Clots, like drops, are caused by anomalous booth exhaust rates. Glaze settles on the inner surface of the booth and forms clots that subsequently detach from the wall and fall onto the surface of the transiting tiles. The defect appears when exhaust rates are excessive and the glaze hitting the walls tends to dry and encrust. Glazes with a high percentage of plastic components are particularly susceptible to drying, especially where applied at high density. Engobes, owing to their high plastic component, also tend to cause this defect. Vibration obviously accentuates this inconvenience as it shakes the crust off the wall. The simplest way to prevent this problem is to wash the booths according to a suitable schedule. Excessive suction or vibration, should, of course, also be eliminated. c) Stripes and waves with waterfall and bell units Stripes originate from imperfect sieving of the glaze and can appear as imperceptible lines or scratches or even unglazed streaks. In the first case the defect will probably escape the attention of glazing line inspectors and only become noticeable on the finished product; in the second case the defect is evident and the tile is rejected immediately. The cause is usually a clot or grain that has formed in the orifice, thus modifying glaze flow at that particular point or even blocking it altogether. In the latter case a tear is seen in the waterfall or veil. 363


This defect can be eliminated, or at least limited, by properly sieving the glaze on the feed line and that recycled from recovery trays. Glaze containers should also be covered to ensure no contaminants get in. A similar problem is waves on the glaze surface, straight with the waterfalls and slightly curved with bell units. They may be-caused by: low glaze density or viscosity drop speed too high for tile transit speed glazing machine frame or floor vibration. Air bubbles in the glaze tear the veil and cause oval-shaped voids. With highly plastic glazes washing needs to be carried out at least twice a day to eliminate the crusts that form on ducts and sieves. d) Irregular disc application Applying the glaze by atomising it with centrifugal discs is perhaps the most commonly used technique in ceramics. Unfortunately this equipment is often troublesome, especially with larger tiles. To begin with, glaze application on the right and left hand sides of the tile is asymmetrical, a problem usually resolved by using two sets of discs that rotate in opposite directions, each applying 50% of the required glaze. Another defect caused by the discs, especially on large tiles, is the appearance of so-called stripes, caused by the glaze being deposited in alternating strips of different thickness. These problems generally stem from poor balancing of the disc unit, thus causing it to oscillate rhythmically and produce stripes. This explains the trend towards discs of smaller diameter and precision disc machining to ensure fine-tuned balance. Glazing machine manufacturers are currently engaged in intensive research activities aimed at solving both streak and droplet problems. Following the recent shift towards very large tiles with smooth, uniform surfaces, waterfall and bell units are making something of a comeback as they do not produce the above problems and in any case produce smoother surfaces than disc units. e) Non-uniform cup application This application device is almost no longer used but certain problems/solutions may be relevant to other devices. The main defects caused by this type of application unit are: 1) More glaze is deposited on edges parallel to the direction of transit than elsewhere. especially on large tiles. The problem can be resolved by increasing cup rotation speed or increasing the clearance between cup and tiles. 2) Excess glaze is deposited on one side parallel to the direction of transit. In this case the manufacturer should check that the cup axis is perpendicular and that both cup and tile are horizontal. The cup also needs to be centred over the tile flow perfectly, and glaze feed should also be perfectly central. 364

Defects

3) If a low density glaze is used less glaze may be deposited on the leading and trailing edges of the tile. To get round this problem the tiles can be arranged as a continuous carpet under the cup drop device. 4) Slightly arched streaks may be noticed crosswise to the direction of tile transit. This usually occurs either because the cup is not perfectly centred or because tile transit speed is excessive with respect to cup rotation speed. Decreasing transit speed or increasing cup rotation speed and adjusting weight usually solves the problem, although this does slightly alter the shape of the drop. 5) Drops parallel to tile motion. Mainly caused by too acute an angle of on-tile drop impact; reducing rotation speed, lifting the cup or inserting diaphragms along the drop fall trajectory usually resolves the problem. 6) If preparation glazes with intense colours are used or silk-screen printing is carried out prior to application of a very liquid glaze (scorza), application should be carried out at considerable distances so that it dries well and avoids any contamination of the glaze used for the cup. Moreover, consistent application of the glazed requires that absorption be kept constant before it is actually applied. f) Glaze on back of tile (dirtying) This defect is caused by glaze sticking to the support items (refractory slabs, rollers etc.). Problems can stem from incorrect adjustment of a waterfall orifice or bell (excessively fluid glaze, low tile conveying speed, violent fall of the glaze etc.) or may be associated with the application method itself, as with cup glazing where cloudy effects require very low density glazes. Where the problem depends on the glaze application method, the glaze can be removed from the back of the tile by brushes. Cleaning of fast single firing products is particularly difficult because using dry wire brushes wears the still soft tile and wet brushes cause water absorption. In this case engobes are applied after glazing to form a refractory layer between roller and tile (applications of water repellent before glazing are rarely used). This problem can drastically reduce the output capacity of a fast single firing plant. Not only that, it can also raise maintenance costs considerably on account of high roller replacement rates. g) Inefficient side fettling or cleaning Very frequent with both single firing and double firing. The problem is generally caused by incorrect adjustment of guides and fettling devices, or because the glaze is not yet dry enough. The defect can also damage the entire length of the tile side where fettlers are improperly regulated. Guides should have only a limited piece alignment effect and should be angled so as to only contact the bottom edges of the tiles. Lining the guides with Teflon generally gives good results: Teflon has waterrepellent properties and limits glaze adhesion.

365


h) Glaze removal on unfired tiles Tiles may often collide with one another as they pass through the glazing line; these knocks tend to detach or chip the glaze and the defect then shows up during firing as glaze shrinkage. The situation is critical where gap-free tile carpets form (i.e. when the tiles bunch up as conveyor speeds slow). Generally speaking, this occurs before the silk-screen printing machines and before the single layer kiln loading or line compensators. Glaze detachment problems can at least partially be solved by augmenting bodyglaze adhesion with glues while mechanical impacts can be reduced by minimising speed differences from one conveyor section to the next. i) Shade differences caused by non-uniform glaze weight Inconsistent glaze weight/spread performance give rise to differences in colour and surface appearance, thus making sorting necessary. Such differences do not, in themselves, necessarily constitute defects. However, incorrect sorting does constitute a defect, and an excessive number of shades makes proper sorting extremely difficult. Shade differences that originate on the glazing line may be caused by variations in glaze density, absorption, body temperature or partial clogging of the glaze ducts. Density can be kept constant by checking it at fixed intervals; a density check is also compulsory whenever new glaze is added to the feed tank. Glaze application weight needs to be closely monitored when absorption time changes. Partial clogging of the glaze ducts is especially frequent in nozzles and holed tubes. Such problems are best prevented by proper sieving of the glaze and scheduled washing of the equipment and ducts through which it flows. l) Too much/little water This affects density and thus causes some of the defects described above. Note that too little water, especially with disc glazing, leads to a very grainy, rough surface of poor spreading performance during firing. The outcome is a glaze with a hammered look. Vice versa, an excessive amount of water causes small craters; cuts on the edges may also be noticed, especially where multiple glaze coats are applied. This inconvenience is particularly frequent in on-glaze flame effects, because the spray gun colour requires a perfectly dry attachment base. On a wet surface it tends to move and produce the above-mentioned defects. Another technique often affected by excessive amounts of water is waterfall washing after application of repellent and glaze, thus resulting in high water quantities being absorbed by the biscuit.

366

Defects

Decoration defects Recent years have seen intense development of screen printing techniques, with associated ink preparation problems and difficulties in finding suitable vehicles for machines that have very different operating principles. A whole subset of defects is linked to these difficulties. The main ones, for flat screen printing, are: a) White line on tile side closest to worker This defect is caused by the lack of screen printing paste on the side of the tile closest to the operator and parallel to the direction of tile transport. The screen can be cleaned but the defect often reappears just a few tiles later. If paste residue is within the proper range and no dry areas are observed on the screen, proceed as follows: Check that the screen is not too close to the tile Check that the squeegee is aligned properly Check that tile height is aligned with belt support guide Check for proper squeegee pressure. b) White line on tile on side opposite worker (as above) Check that the home position of the squeegee is high enough; it must not press the screen against the tile edge when the tile exits (if it does adjust the shuttle cam) Check synchronism of movement between belt and shuttle. c) Accumulation of glaze at corners This defect is usually caused by an excessive screen-tile gap. Check that the tile is properly aligned between belt and support guide Lower the screen over the corner where inconsistent accumulation is observed. d) Pattern/motif not centred properly Check that the screen is positioned and secured properly Check that the belts grip the piece firmly. e) Tile sticks to screen Check fixative application is working properly Check height of screen above tile. N.B. Above-tile screen clearance is usually set to 4-5 mm but this can vary as a function of: Screen tension Screen fabric Screen printing paste viscosity.

367


f) Screen breaks and/or splits This usually occurs over tile edges or corners and is caused by: Excessive on-screen squeegee pressure Excessive glaze residue Screen with poorly catalysed gelatine (emulsion). A well made screen usually has a working life of some 2000-2500 m2 (depending on tile size). Firing defects with effects on the glaze a) Temperature differences Usually associated with tunnel kilns, but can easily occur on other kiln types. The result is differences in tile shade or, sometimes, variations in glaze surfaces. The problem is a serious one because it causes almost imperceptible shade differences between adjoining tiles. This makes sorting particularly difficult. b) Lack of oxidation This does not mean that ambient air in the kiln is completely reducing, but, rather, that there is not enough excess kiln air to ensure so-called scrubbing (air washing). Hence fume passage zones form, often sufficiently concentrated to cause a defect; this generally consists of eggshell or bubbles on tile corners. There may also be zones where the air stagnates and is full of glaze vapours. This nearly always causes surface opacity, sometimes accompanied by pinholes. In both cases the defect can be corrected by increasing air turbulence and volumes, thus increasing stack exhaust rates and, consequently, air intake. c) Incorrect fume flow Normally appears as surface opacity and can degenerate into eggshell. Originates from improperly regulated pressures that cause some of the fumes from firing zones to follow the tiles towards the outlet instead of immediately being drawn towards the stack. When these tiles cool, condensate can deposit on the surface of the still molten glaze and thus be incorporated in it. This problem can be controlled by: increasing the air aspirated towards the stack increasing counter-flow fume speed to stack increasing the volume of direct cooling air. d) Over/Under-firing of glazes Excess and insufficient firing often generate similar outcomes (as with fusible and refractory glazes); the former generates an outer cordon characterised by bubbles or other defects while the latter can lead to the formation of matt or hammered-looking surfaces. 368

Defects

With floor tile glazes, the matter may be more complex as it is not always possible to establish whether the tile has been fired too much or too little. For a single firing glaze with a high percentage of added coarse material and a highly fusible vitreous base, for example, excessive firing can cause separation of the added material, (which remains on the surface) from the glaze (absorbed by the body). The result is a rough, hard surface, that initially brings to mind an under-firing defect. In such cases a good understanding of glaze types and firing behaviour is required. e) Cycles too fast or too slow for the glazes Like bodies, glazes may also be affected by fast firing cycles; these can stop the glass maturing or halt development of desired colour and/or texture. With simple glazes like transparent or zircon white frits, a hammered appearance may be noticed because the glaze has no time to spread properly. Floor tile glazes nearly always maintain the same surface area as when they are unfired; when disc-applied they are consequently never smooth enough. Despite the cost, manufacturers usually prefer to correct these defects by increasing glaze meltability rather than extend firing cycles. The latest bell application techniques give green materials a smoother surface without manufacturers having to use the fluxing agents that, in fast single firing, affect the geometry of the body. Eggshell and opacity often occur especially with highly fusible glazes on the edges because of sublimation of components that are then missing from the composition of the glaze itself. Since the cycle cannot be shortened, the amplitude of the firing curve is reduced and the duration of maximum temperature is limited.

369


Firing defects Tile bursts in the pre-kiln (explosions) Abrupt evaporation of water can cause tiles to explode. The problem originates from excessive water content, be it residual after drying/glazing or reabsorbed during green tile storage. Firing with water contents above 2% is inadmissible; if necessary, drying should be extended. As a guideline, drying temperatures in the 250-300 °C range are appropriate. However, such a degree of caution is only required in critical situations: pre-kiln temperatures are usually in the order of 350500 °C. To contain temperatures, which rise significantly when gaps appear in production, the air intakes in the pre-kiln channel (through which the fan draws ambient air) can be opened. The occasional explosions of just-glazed tiles with a high concentration of stratified water just below the glaze is not particularly significant here, nor is the bursting of pieces with de-airing defects (flaking or peeling). Breakage during preheating Characterised by fracture lines with ragged edges running from the outer zones of the tile towards its centre; the glaze tends to penetrate the fracture and round its edges (fig. 16). Breakage is caused by the cracking that occurs when edges heat up and shrink faster than the rest of the tile. Single fire breakages of this sort are relatively rare and are usually associated with extreme thermal gradients. The problem may also appear as cracks some 20-30 mm long; these are nearly always multiple and never appear at the corners. They tend to occur at points on the kiln cross-section where temperature increases occur first. The critical range lies between 700 and 900 °C. Very hot flames under the rollers are the most common cause. Reducing the temperature of the first/second under-roller burner set by 20-30 °C usually resolves the problem; increasing air flow towards the burners also helps as this cools the flame.

Fig. 16.

370

Defects

Small, almost imperceptible cracks can also appear on lower tile edges at the sides of the kiln load. These are usually the forerunner of much more evident cracking during cooling. Pre-heating breakages can also occur at around 800-900 °C because the load comes into direct contact with cold air. This happens when burners have been shut down or when rollers have been removed and apertures inadvertently left open. To all intents and purposes these are actually cooling breakages on already-stiff pieces that have just passed through a critical temperature range. Preheating damage may also take the form of multiple, tightly packed, short cracks on the upper edges (fig. 17), sometimes almost covered by the glaze. This defect sometimes involves the leading edge only (i.e. the part of the tile most exposed to the flow of gasses towards the stack).

Fig. 17.

The problem lies in preheating gradients that are too steep for pieces already made susceptible to such defects by sharp or ragged, poorly compacted edges, high stratified moisture contents beneath the glaze and low body plasticity. Worn press punches can also contribute to the problem, as can intense kiln reagulation for concave deformation at the preheating stage. Broken corners can also occur during preheating; the line of fracture often gives the detached corner a characteristic pistol shape (fig. 18). The point at which this defect occurs can be identified simply by inspecting the bottom of the firing channel: a heap of broken pieces usually indicates where the trouble lies. Lowering temperature a few tens of degrees usually solves the problem. Increasing air flow to the burners may also help as this cools the flame. The broken pieces must be dropped to the kiln hearth through a trap to stop them obstructing the load.

Fig. 18.

371


Other defects also characterised by ragged fracture lines and rounded edge glaze are often mistaken for preheating faults; their true cause, however, is generally found upstream from the firing process. a) Long, curving, occasional cracks occurring randomly across the load (fig. 19).

Fig. 19.

Caused by mechanical stress: crucial checkpoints include screen printing, row accumulation units, conveyors in general and unfired tile box loading/unloading units. b) Cracks some 10-30 mm long at the corners, occurring randomly across the load. Usually caused by inter-tile collisions during row formation. An accumulation of shards and corner chippings on the floor usually identifies the critical conveyor point. c) Cracks in the glaze, parallel to one or more sides of the tile (fig. 20). d) Similarly, fissures may form inside the piece, again parallel to the sides (fig. 21). Caused at the pressing stage during ejection of the piece from the die cavity. e) Slightly divaricated cracks some 20-30 mm long, not localized at any particular point of the kiln cross-section. Their emergence depends on how long the unfired tiles have been stored. Often seen together with the longitudinal fissures described in d) (fig. 22).

Fig. 20.

372

Defects

Fig. 21.

Fig. 22.

The origin lies in longitudinal breakage that occurs during ejection from the die; excess water applied during glazing stratifies in the longitudinal fissure and cracks on the glazed surface then appear during the initial stages of preheating. f) Small cracks on the glazed surface, often so small as to be confused with pin holes or on-glaze depressions (fig. 23). The cause lies in defective ejection from the press die cavity or, more frequently, application of excess water during glazing.

Fig. 23.

g) So-called chicken claw crack (fig. 24). If the crack does not go all the way through the tile it is most likely the result of a de-airing fault at the pressing stage (e.g. lamination). Vice versa, where the crack 373


Fig. 24.

penetrates the full thickness of the tile, it may have been caused by an impact received from below during conveying: for example a pulley with an overly small belt will jolt against the underside of the tile as it passes over. h) Laminations causing detachment or even explosion of large plaques from the surface of the body and consequent crumbling of the piece (fig. 25).

Fig. 25.

This stems from poor de-airing during pressing. i) Cracks that look just like the preheating cracks described earlier (fig. 16). Where they can all be traced to the same press cavity, the real cause lies in pressing. Such defects do not correlate with any particular part of the kiln load but are generally found on the same area of each tile. Check that the die cavity and/or punch are in good condition. Similar cracks can also be produced at the drying stage: they do not appear on any particular zone of the kiln load but they are consistently localised on a specific zone of the drier load. In this case the cracks can often be highlighted at the drier outlet with liquid colorants (e.g. kerosene). In second fire, complete tile breakage can occur; where this is at random across the load and seen to intensify at certain moments, it is attributable to existing lesions in the body that compromise resistance to the tensions generated during preheating. 374

Defects

Cooling breakage (or “cooling cracks”) This damage originates from excessive tensions generated by non-uniform dimensional variations. It is usually associated with the very high cooling ratios that accompany the inversion of quartz at about 573 °C and results in markedly different temperatures from one area of the tile to another. Damage risk is higher where the body contains significant quantities of free quartz, tiles are thick and large, and cooling cycles are rapid. Rapid cooling gradients or changes in gradient caused by discontinuous tile feed also increase the likelihood of damage. The cracks themselves tend to be curved with sharp, clear-cut edges, as with broken glass. Fissures are never open and are, in fact, often difficult to spot without a coloured liquid; they always run from the outer edge of the tile towards its centre. In non-porous products the fracture has a smooth, shiny, conchoid look. Tapping the tile produces a dull, muffled noise. Cooling is a three-stage process and takes place in three different kiln sections. Initial RAPID COOLING lowers piece temperature to 650-700 °C. Subsequent SLOW COOLING gradually cools tiles to at least 550-500 °C. A third FINAL COOLING stage cools them to the point where they can be handled at the kiln outlet. Any attempt to correct cooling problems should take into account the following: 1 Temperatures detected by thermocouples in the firing kiln are much lower than the actual temperature of the moving tiles. 2 Pieces at the outsides of the load cool down much faster than pieces in its middle: this difference is accentuated by steep cooling gradients. 3 Temperatures at the sides of the tile are lower than in its middle; this difference, accentuated by steeper cooling gradients, is the true cause of tile breakage. 4 Barriers (chicanes) separating firing and cooling zones must be intact; efficient zone separation is essential for stable regulation of both firing and cooling. Identifying the point of tile failure If cooling breakages occur repeatedly in specific zones of the load and specific situations the manufacturer is provided with useful information on the regulation error being made. a) Breakages are localised at one or both extremities of the roller conveyor. This indicates that damage is occurring in rapid cooling, which is either too intense or extended. Correction involves raising the in-zone temperature and closing off one or more blowers so that each tile spends less time directly under the air jets. Reducing the stack aperture and thus increasing hot air intake can also help. A further measure consists of closing the first heat diffuser segment in slow cooling. 375


The above-roller thermocouple in the fast cooling zone usually gives a temperature reading of around 570-600 °C, temperatures being lower where cycles are faster. Tiles exiting rapid cooling can be checked for proper temperature with the aid of a portable thermocouple; this is inserted inside the last roller of the zone. A reading of 600-620 °C is ideal. b) Breakages are mainly localised at the centre of the kiln. Indicative of problems at the start of the final cooling zone: centrally positioned tiles cool more slowly and may pass under final cooling jets before their temperature has dropped below 573 °C. The problem can be corrected by enhancing cooling performance in the first two cooling stages and/or delaying final cooling. The dissipators should be opened wider. A moderate increase in stack draught is also useful, as is reduced hot air intake (especially the initial intake) and closure of the initial cold air blower in final cooling. The temperature of the last thermocouple in the slow cooling zone should be around 480-500 °C. c) Breakages occur on the first tiles to enter the kiln after a significant gap in product flow and persist for several minutes. Gaps cause cooling temperatures to drop: a reduction of just 50-70 °C can cause problems. Check that, with the air modulation valve in the rapid cooling section closed, the blowers are not excessively pressurised. It may be necessary to reduce heat exchanger efficiency by partially opening the by-pass valve. Check the slow cooling section and make sure the heat dissipator modulating valve closes completely when temperature drops. Generally speaking, it is best to modify kiln volumes by aspirating large volumes of hot air. Check that the thermoregulators in rapid and slow cooling are working properly. d) Breakages occur after significant load gaps, not on the first tiles but those following on some minutes after. The problem is usually transitory. As per c). It is especially important to limit maximum air pressure in rapid cooling. Crazing Similar in aspect but very different in origin and nature, crazing is often confused with cooling cracks. It takes the form of sharp-edged fracturing of the glaze only and often shows up only with the aid of marker dyes or certain lighting conditions. 376

Defects

Crazing is not necessarily evident at the kiln outlet either; it can appear at a much later stage during long-term de-tensioning, or during re-hydration of the hygroscopic body. Crazing fracture patterns are highly distinctive: lines do not necessarily run from edge to middle and tile edges are often unmarred. The cause of crazing lies not in cooling errors but a poor glaze-body expansion match that stretches the glaze. Tests can be carried out by putting samples through repeated autoclave cycles according to a specific protocol. A more empirical way of identifying susceptibility to crazing is to lay tiles with cement: if the glaze-body match is incorrect crazing will appear. However, the test can take a long time and is not well controlled. Uniformity of shrinkage over the kiln cross-section Attainment of uniform firing is especially difficult where kilns have wide inlets and are used to produce large tiles. The extent of shrinkage and, more importantly, the body vitrification range, are crucial in highlighting the effects of even small temperature differences across the firing channel. Temperature differences can be quantified and pinpointed by measuring pieces pressed in the same die cavity and fired in different load positions (all tiles must be oriented the same way with respect to the press outlet). After marking the tiles to indicate their position (fig. 26) all the right and then all the left side pieces are compared.

Fig. 26.

An example is evaluated as follows: + 0.5 mm on left side of kiln with respect to middle tile + 0.2 mm on right side of kiln with respect to middle tile. An indistinct comparison between right and left sides includes pressing errors (0.4 mm on right-hand side of piece). When checking the tile dimensions comparisons should be made between tiles all across the kiln section and care should be taken to note any effect resulting from pressing errors which overlap with firing variations.

377


a) The most frequent occurrence is larger tiles in the cooler areas nearer to the outside edges of the kiln cross-section. Several causes combine to create the problem (fig. 27):

Fig. 27.

378

Defects

Flame temperature at the nozzle is lower than further away where combustion is complete. Re-circulation of colder ambient air on account of the Venturi effect generated by flame speed. Fumes move more slowly near the walls than in the middle of the channel because of friction (flow towards chimney). Stronger irradiation of centre of the load by perpendicular walls. Outward dispersion of roller heat by conduction. Aspiration of cold air via improperly plugged roller seats in the firing zone. Low-speed cold air blow movement, especially above rollers. The problem can be corrected on several fronts (fig. 28).

Fig. 28.

Kiln pressure (see relevant paragraph) must be positive in the firing zone (at roller height) so as to prevent cold air infiltration; however, pressure should not be too high. Insertion of one or more deflectors at the beginning and inside the firing zone diverts more fumes towards the tiles near the walls. Using a suitable number of burners with semi-radial dispersion above and below the rollers is a tried and tested system that compensates for colder flame areas near burner nozzles and any intake of colder ambient air. To fine-tune, air can be dosed to the burners to modify flame temperature and length. Flame deflector screens can also be positioned in front of firing zone burners beneath the rollers (fig. 29). b) Irregular shrinkage across the cross-section (known as candy paper). Characterised by shorter tile sides 60-70 cm from the walls and longer ones adjacent to them (fig. 30). Here, despite the similarities with case a), erroneous flame geometry (especially on burners above the rollers) is usually the culprit. Considerable modifications to 379


Fig. 29.

Fig. 30.

flame speed and geometry are usually required. An easy and often effective solution is to shut down a burner in those sets that markedly influence size; alternatively, the flow of burner air can be modified. c) Other, differently configured, irregularities can also occur; depending on the situation, their solution may require use of the above solutions or even their direct opposites. Glaze brilliance and shade In terms of shine, satin effects, opacity, spread and shade the final outcome of glazing essentially depends on the high temperature firing stage. Higher temperatures or extended firing time results in more compact vitrification of the glaze and thus improves shine. Longer times at relatively lower temperatures give a satin look and better (flatter) spread. Much, of course, depends on the nature of the glaze, the limitations of which are inevitably highlighted by pin holes and micro-bubbles if temperatures are too high. The appearance of satiny effects (clouding of the glaze at certain points of the load) indicates a combustion defect on one or more burners in the high temperature zone; a check on (and possible increase in) the flow of combustion air is generally required. Shade changes across the load are the result of temperature gradients. Lustre 380

Defects

applications, for instance, are particularly sensitive to overheating and can easily vanish. Where strongly reactive colouring oxides are used (iron, selenium etc.), longer firing is known to disturb colour development. Hence iron oxide veers towards green where firing lasts longer and red where firing is shorter and sharper. Pin holes - holes - bubbles in the glaze These are caused by small gas bubbles that expand in the glaze and burst, leaving a small crater on the surface. Gas bubbles have numerous causes. Sectioning a defective tile and taking a close look at its interior can provide valuable information. Using a magnifying glass or pocket microscope, it is good practice to observe the glaze along the line of fracture, paying special attention to the nature of the trapped bubbles. Bubbles are very small, densely packed and close to the surface. This defect stems from very low melting viscosity or over-firing, causing the glaze to boil. Generally attributable to overly hot flames on the last above-roller burners or, more simply, too high a firing temperature (fig. 31).

Fig. 31.

Bubbles observed in body of glaze only; bubbles that are forming, incomplete or touching the body are not observed (fig. 32).

Fig. 32.

Indicative of high melting viscosity or low quality glaze containing impurities. May also indicate a glaze preparation (milling) or application (bubbling) error. The problem cannot be solved by adjusting the kiln. This defect also occurs when glaze is applied to an excessively hot body, thus causing water vapour to form between body, engobe and glaze. 381


In addition to those in the glaze, other bubbles are seen to be forming in the body and rising off it (fig. 33).

Fig. 33.

The most likely cause in this case is incomplete degassing prior to glaze softening: black core, especially if it is stratified near the glaze or takes the form of specks scattered through the body. carbonate degassing problems, an eventuality more or less limited to monoporosa products: the defects are larger than pin holes and the underlying engobe is frequently exposed. pyrite degassing with SO2 emission: in this case the tiles need to be held at 10401050 °C for several minutes (frequently solved by fine sieving of the body slip). Confirmation that degassing is involved in the pin holing problem is usually provided by a simple double fire test: some unglazed pieces are sampled at the drier outlet and then fired to fully degas them. They are then passed through the glazing stations, ensuring that each inter-station time lapse is long enough to let water be absorbed or evaporate. The samples are then re-fired: absence of bubbles in the glaze confirms that body degassing was the cause. There may also be larger holes in the glaze. These have various causes: late degassing of body carbonates, especially where their particles are quite coarse. In this event the 920-980 °C firing time needs to be extended. Body grinding and composition must be checked. thin, compacted flakes of body, especially from the sides of the die box, work their way into the tile body or flakes/powder stick to its surface at the time of ejection from the press: in the kiln this causes micro-peeling. The only solution is to ensure that dies are kept clean by suction/blower devices. Pin holes, holes and bubbles in the glaze may also be caused by defects in the glaze or its application: grinding and sieving viscosity and rheology CMC deterioration incompatible glaze-engobe melting points

382

Defects

high screen print stickiness compromises adhesion of glaze to underlying engobe or body glazing on excessively hot body inadequate spacing between glazing units poor glaze-body adhesion overlaying of incompatible glazes. In some cases extending the firing time towards 500-600 °C may be useful; it will, in any case, be necessary to eliminate the cause of the problem by correcting glaze and glazing technology. Degassing All ceramic bodies lose a considerable amount of weight during firing. Mainly by the loss of residual process water and the water in the clays. The rest is attributable to gasses and vapours produced by the combustion of organic contaminants, the decomposition of carbonates (especially CaCO3) and sulphur compounds etc. Gas generation occurs during preheating, a stage that ends with the formation of the first vitreous phases, which rapidly reduce permeability. This generally occurs at 980-1000 °C in single fire operations and around 1100 °C with porcelain tiles. High temperatures also generate vapours (fluorine, chlorine etc.); while they do not necessarily cause product defects, they often attack the refractory and steel parts of the kiln structure. Other vapours released by the glaze (lead, boron compounds etc.) can easily condense on the kiln walls and roof, resulting in the fall of droplets and damage to transiting tiles. The theory behind degassing is a simple one: the aim is to raise the material to a temperature at which gasses and vapours develop as quickly as possible and then hold the temperature for as long as it takes for all the gasses to escape. The best temperature is the highest possible (but, of course, still below the body and glaze vitrification point). Degassing problems depend on several factors: composition, purity and particle size of raw materials thickness, density and moisture content of pressed tile fusibility of glaze and frit characteristics and quantities of screen printing inks and their media (vehicles). The prevailing opinion is that a richly oxygenated preheating environment aids degassing; this is true only in part and in any case needs further explanation. For instance, the flow of exiting gas may stop ambient oxygen penetrating the piece. However, it is true that burners fed with excess air supply all the required energy and high volumes at low temperature: this makes overheating and thus early development of the vitreous phase and reduced permeability less likely. 383


Similarly, good fume draught is helpful. The first degassing benefit is not the draught-induced vacuum, which might erroneously be thought of as sucking the gases out of the tile, but the considerable quantities of hot volumes drawn in from the firing zone; these allow a reduction in gas flow to the pre-heating burners (and thus lower flame temperatures). Degassing in double fire processes Since the biscuit has already been fired once the second fire is usually free from degassing problems. Nevertheless, problems may arise with biscuits that have been put through rapid firing cycles and then left on hold for some time after application of the glaze. The problem may be caused by water (from glaze, or absorbed by hygroscopy) reacting with calcium oxide (CaO) residues left over from the first fire (in longterm storage it could transform into CaCO3 again). The resulting defects take the form of evident, foamy bubbles and bare sections of body, often confused with glaze adhesion problems. In this event it is good practice to extend the preheating time below 550-950 °C both by easing the temperature gradient and slowing the cycle. Black core This defect has already been dealt with in the section on defects associated with raw materials (see Appendix 3: Defects, Organic material). It takes the form of a halo, ranging in colour from yellow-green to grey or black, in the tile interior. The problem can also appear as simple black specks dispersed throughout the body. Black core is caused by incomplete combustion of organic residues in the body or by subtraction of oxygen from iron oxides: in any case, it is always a reduction phenomenon. Black core would not, in itself, constitute a problem (often normal in extruded products) were it not associated with other defects such as: pin holes and bubbles in the glaze glaze porosity and its stain resistance differently shaded halos and spots appearance of back pattern on the face of the tile dimensional and planarity defects. The most suitable firing range for elimination of black core generally lies between 880 and 960 °C. While the bulk of good quality glazes have a softening point of around 980-1000 °C, note that a few low-fluxing glazes seal pores at just 680-700 °C; where the latter are used the firing cycle will need to be slowed down considerably. High-temperature glazes and bodies allow preheating to be extended towards 1100 °C. No less important than optimal temperature is duration of exposure to effective degassing temperatures. Maximum efficiency is obtained by keeping temperatures high at the start of preheating and low in the approach to the firing zone: 384

Defects

here, it may be useful to blow air through switched off burners (taking care not to generate significant temperature differences above and below the rollers as these can change tile planarity). Good degassing requires abundant volumes, both from the firing and cooling zones and the burners. Hence any cooling adjustment that exhausts a lot of hot air (cooling stack) is to be avoided. Degassing problems also raise energy costs. The use of more refined raw materials and tile technology should thus be attentively evaluated. Black core can appear as: a localised halo in the core of the tile or as scattered specks (fig. 34) an under-glaze halo that fades towards the centre of the tile (fig. 35). Glaze reaches its softening point too quickly and prevents degassing of the body.

Fig. 34.

Fig. 35.

A localised halo at the core of the tile that fades towards the glaze (fig. 36).

Fig. 36.

385


High temperature preheating time too brief or under-roller temperature at end of preheating is too high. Halos limited to areas of lenticular cross-section (fig. 37).

Fig. 37.

Highlights non-uniform pressed piece density caused by a die filling defect or accumulation of fine powder. Where intensity is low the defect can generally be resolved by holding the optimum degassing temperature long enough. Rather than halos, there may be evident swelling of small, circumscribed areas. Such cases are generally sporadic (see fig. 38).

Fig. 38.

The intensity of the phenomenon means that it cannot be resolved via kiln adjustment. Swelling is usually caused by drops of machine oil, carbon particles formed by poor spray drier combustion or other body contaminants (even very wet or high density grains). Planarity defects Precise diagnosis of planarity problems requires in-depth observation of the defects. Deformation profiles, small differences between one area of the load and another, variations caused by specific events and circumstances and links with other problems are all, in the light of product characteristics, essential if a reasonable hypothesis is to be developed and then tested. Planarity defects are not necessarily caused by malfunctioning or incorrectly adjusted kilns; more properly, it should be said that the kiln is often unable to eliminate errors that occurred upstream from it. A poor dilatometric (thermal expan386

Defects

sion) match between body, glaze and engobe is a frequent cause; alternatively, the problem may be associated with heterogeneous tile density or the low structural strength of very molten, very thin or specially shaped pieces. However, that is certainly not to say that proper kiln adjustment never solves the problem: it often does. Sometimes it merely attenuates the defect and sometimes it has no effect at all: in any case, close collaboration between kiln operators and technologists is essential if an effective plan of action is to be developed. Controlling tile planarity largely involves control of the consequences of linear shrinkage during firing. Imagine a tile cross-section: if the face of the tile is subject to higher temperatures than the back then it will shrink more, causing the tile to bend and become concave; vice versa, if the underside of the tile is exposed to more heat, the back of the tile contracts more and makes the tile convex. Only a small area of a convex/concave tile actually rests on the rollers. Once the body reaches its softening point, the unsupported part of the tile, tends to collapse under its own weight: this secondary deformation works in the opposite direction to the initial one caused by temperature differences. Because it takes time for the tile to soften and collapse, this secondary effect takes place at the same time as the curvature effects of thermal expansions occur at the beginning-middle of the firing zone. Because the edges of the tile heat up faster than the middle, an appreciable temperature difference during the brief final firing stage can accentuate concavity or convexity at the corners. Planarity problems can also result from non-uniform cooling of tile face and back: this changes both the moment in which the glaze locks to the body and influences the relationship between glaze and thermal body expansion. Smooth, clean, straight rollers are essential for orderly in-kiln movement of tiles and the prevention of mechanically-induced warping and planarity deformations. There follows a description of the most common planarity defects. Downturned corners All four tile corners droop (usually the outermost 3 cm or so); the rest of the surface is essentially flat or slightly convex (fig. 39). The defect is distributed throughout the load and is constant; deformation is slightly reduced at the ends of the loaded tiles (head and tail). Corrections should be made in the final firing zone by raising temperature above the roller plane and reducing temperature below it. Diagnosis should be effected with care as it is easy to confuse this defect with others, different in nature, that are similar at first glance but quite different when observed more closely (figs. 40-41).

387


Fig. 39.

Fig. 40.

Fig. 41.

Upturned corners All the tile corners are upturned (outermost 3 cm); the rest of the surface is essentially flat or slightly concave (fig. 42).

Fig. 42.

The defect is distributed evenly throughout the load and is constant; deformation is slightly reduced at the ends of the loaded tiles. Corrections should be made in the final firing zone, where it will be necessary to lower the temperature above the roller plane and increase temperature below it. Diagnosis requires caution as it is easy to confuse this defect with others, different in nature, that are similar at first glance (figs. 43-44).

388

Defects

Fig. 43.

Fig. 44.

Convexity Homogeneous downward bending of all tiles, with similar deformation profiles on all four sides (fig. 45).

Fig. 45.

One of the most frequent and difficult planarity problems as the defect stems from glaze-body expansion incompatibility. Factors such as engobe, glaze-body thickness ratio, tile density, tile size, degree of vitrification, back pattern and others all influence this problem. The temperature curve in itself rarely causes convexity: only a central-final firing zone characterised by higher below-roller temperatures is capable of doing this. However, this is not to say that kiln adjustment cannot provide solutions; in many cases results are rather good. Nevertheless not all products respond to adjustment the same way, hence the need to distinguish between different product types. For instance, with single fire floor tile we can hypothesise: a) Correction in Rapid Cooling Zone Below-roller direct blowing is intensified at the start of rapid cooling. This reduces convexity throughout the piece, especially on sides parallel to the rollers. By cooling the body earlier the latter contracts before the glaze locks and opposes it. This difference between sides parallel to the rollers and those at right angles to them may be explained by the fact that they undergo different cooling processes. Parallel sides (leading and trailing tile sides) are cooled all in one go 389


while sides at right angles to them are cooled from one end to the other as the tile advances (fig. 46).

Fig. 46.

Meaningful correction results can only be achieved with properly designed final cooling zones: there should be a fair concentration of blowers beneath the roller planes, especially in the first section, and each blower must have a certain output rate; it should also be possible to shut off corresponding above-roller blowers. Additionally, more hot air can be aspirated into the cooling area; this air stratifies above the rollers and keeps the glaze hot for longer, thus delaying its attachment to the body. Removing the rollers immediately above the bottom blowers also helps. Avoid overly dense packing of tiles in rapid cooling (i.e. maintain space between lines). Similar results are also attainable with unglazed porcelain tiles, especially large ones. However, no discernable advantages have been observed with porous products (monoporosa, biscuit, glazed double firing). Considerable temperature differences (hot tiles on top of the rollers and cold air under) can cause evident roller deformation. This, in turn, generates asymmetry and overlapping of tiles. The only way out of this quandary is to use specially designed rollers with high heat transmission coefficients. b) Correction in the Firing Zone Takes the form of a firing curve structured to give higher below-roller temperatures in the initial section, from 1050 °C to just beyond peak temperature. The difference is usually 20-30 °C, but may require peaks of 50-60 °C. Convexity correction is always slightly more marked on sides at right angles to the rollers than on those parallel to them (fig. 47).

Fig. 47.

The dynamics of correction may be illustrated as follows: Stage 1 Higher temperatures beneath the roller cause the bottom of the tile to shrink faster than the top: the piece thus becomes convex (fig. 48). 390

Defects

- t° – SHRINKAGE + SHRINKAGE

+ t° Fig. 48.

As the tile shrinks it also becomes soft. Stage 2 In passing from one roller to another the convex tile rests on its leading and trailing edges (black areas): the central section is suspended (fig. 49).

Convex tile rests on leading and trailing edges only (black area).

Fig. 49.

The whole of the weight rests on the points of contact (fig. 50). ~8cm – SHRINKAGE + SHRINKAGE

Support zones

Fig. 50.

Gravity causes the tile to flex and droop. Stage 3 Firing continues: the face of the tile is now exposed to temperatures and shrinkages similar to those on the back, thus causing concavity (fig. 51). = t° = SHRINKAGE

= t° Fig. 51.

391


Sides parallel to the rollers are rather more uniformly supported (or suspended) and anti-convexity correction is visibly modest. Over-correction can produce a defect characterised by upturned leading and trailing edges affecting sections some 8 cm long (fig. 52) ~8 cm

~8 cm

Fig. 52.

or the priest hat deformation (fig. 53).

Fig. 53.

Higher above-roller temperatures in the final firing and/or maximum temperature zone may also prove useful, possibly involving shutdown of burners below the rollers. Excessive correction in this sense can, however, produce the priest hat effect and produce upturned corners (fig. 54).

Fig. 54.

Concavity Homogeneous upward curving of all tiles with similar deformation profiles on all four sides (fig. 55). 392

Defects

Fig. 55.

A somewhat rare defect: if it were caused by an erroneous glaze-body expansion match the glaze would be stretched and crazing would result. Concave deformation is often transitory in nature, being caused by accidental kiln shutdown or macroscopic errors in the firing curve. Correction at the start of the firing zone Takes the form of a firing curve structured to give higher above-roller temperatures in the first section, from 1050 °C to just beyond peak temperature. The temperature difference may be as high as 50-60 °C. The resultant correction is spread uniformly over the four sides. The dynamics of concavity correction are similar to those for convexity. Stage 1 Higher temperatures above the roller plane cause the face of the tile to shrink sooner than the back: the piece thus becomes concave (fig. 56).

+ T° Fig. 56.

- T°

Shrinkage is accompanied by softening. Stage 2 In passing from one roller to another, the concave tile rests on its middle; the remaining peripheral areas are suspended (fig. 57).

Constantly suspended part of tile (black areas)

Fig. 57.

393


The whole weight of the tile is supported on the central part in contact with the rollers (fig. 58).

Suspended areas

Fig. 58.

As the suspended areas press down the tile tends to flex and flatten out (fig. 59). + shrinkage % – shrinkage %

Fig. 59.

Stage 3 Firing continues: the back of the tile is now exposed to temperatures and shrinkages similar to those on the face, thus causing convexity (fig. 60).

= T° = shrinkage %

= T° Fig. 60.

Exaggerated corrective action can produce a central asymmetric hump between the leading and trailing edges (fig. 61) (i.e. only central part of tile is convex).

Fig. 61.

Correction at the end of firing Higher below-roller temperatures in final firing and in the maximum temperature range are also useful; this may involve the shutdown of burners above the rollers. 394

Defects

Excessive correction in this sense can, however, produce deformation (fig. 62).

downturned edges

roller effect (porpoising) Fig. 62.

Correction in the rapid cooling zone Concentrating blown air above the rollers at the start of the zone is useful. Maximum efficiency is obtained by pointing the blowers not at the tiles but towards the kiln roof and bunching the tile rows as close together as possible by decreasing rapid cooling zone roller speed. This accelerates hardening/locking of the glaze, thus increasing the effect of glaze-body thermal expansion coefficient differences. Roller effect A classic single-fire defect (fig. 63). 8 cm

3 cm

Fig. 63.

The defect only involves the sides of the tile at right angles to the rollers; the two sides parallel to them are perfectly straight or only slightly convex. A central area, generally flat or slightly concave, is bracketed by evidently upwardly bent strips, some 8 cm wide, on the leading and trailing edges. The last 3 cm of those strips curve downwards.

395


This deformation has a number of possible causes: 1. Below-roller heating in the firing zone (or at least in its central and final section) is excessive. The piece thus becomes convex on account of relatively more intense underside shrinkage; as the tile advances only its leading and trailing edges are supported by the rollers (fig. 64).

Fig. 64.

The now decidedly soft piece is unable to sustain its own weight and the central section of the tile collapses. If this is the case deformation is resolved by increasing temperature above the rollers and decreasing temperature to the same extent below them: correction should begin at the end of the firing zone and then retreat back towards its beginning. 2. Greater below-roller heating at the start of the firing zone causes relatively greater shrinkage of the tile underside, thus making it convex (fig. 65).

- T° SHRINKAGE

+ T° Fig. 65.

Shrinking is accompanied by softening. Because only the leading and trailing edges of the tile rest on the rollers the piece deforms as described in the CONVEXITY paragraph and on exiting the kiln will be concave, but with a tighter radius over the outermost 8 cm of the leading and trailing edges (fig. 66). Such defects are easily confused with others (CONCAVITY or UPTURNED CORNERS) and often erroneously corrected by increasing below-roller temperatures in the final firing zone (fig. 67).

8 cm

8 cm

Fig. 66.

396

Defects

3 cm

Fig. 67.

Because the tiles remain in this section for only a short period correction influences the corners only, thus causing the outermost 3 cm of the leading and trailing edges to become convex. In this event deformation can be resolved by: FIRST: increasing the temperature above the rollers at the end of the firing zone and then lowering it to the same extent below them. The resulting deformation is as illustrated in fig. 66. THEN: increasing the temperature above the rollers at the start of the firing zone and lowering it to the same extent below them. 3. There is a somewhat rarer third scenario, limited to products fired at very high temperatures. Known also as the ROLLER EFFECT, it is fairly evident but has gentler curves (fig. 68).

Fig. 68.

Solving this sort of problem requires an altogether different kind of approach. Temperatures below the rollers need to be increased considerably and temperatures above them reduced: this needs to be done from the beginning of firing onwards and inevitably involves an intermediate priest hat stage. Priest hat Frequently observed where glaze and body expansion coefficients are significantly different (i.e. mainly with glossy glazes and grains, especially where applied thickly, in monoporosa and floor tile). Characterised by a central convex area with a peak at the geometric centre of the tile and upturned extremities, particularly over the last 3-4 cm at right angles to the rollers (fig. 69). 397


Fig. 69.

Where the defect is mild the ends appear flattened. The radius of the central, convex curve remains unchanged (fig. 70).

Fig. 70.

Priest hat can also be caused by an attempt to resolve convexity (by increasing above-roller temperatures at the end or in the central section of the firing zone). In this event bringing the temperature differences back within more acceptable limits and applying a more suitable anti-convexity stratagem will usually suffice. May also be caused by excess temperatures below the rollers in the initial firing zone. There are two possible approaches: bring temperature differences back within more acceptable limits. force the existing adjustment by increasing the temperature differential, extending it into the central firing zone and, if necessary, into the final firing zone with higher below rollers temperatures. Note that where such adjustment is performed on products without any particular expansion match problems, the end result is the roller effect (fig. 71). which represents the exact opposite of “priest hat”

Fig. 71.

Where priest hat develops at the start of firing there often follows asymmetric warping caused by irregular feed, pieces bumping into each other and overlapping. These secondary deformations tend to mask the priest hat itself and can delay its identification. Problem solving thus needs to be prioritised: first priest hat, then warping. Given their associations, the two are often eliminated simultaneously.

398

Defects

Asymmetric deformation or warping Irregular deformation, mostly occurring on sides at right angles to the rollers and corresponding to specific positions on the load (fig. 72).

Fig. 72.

The appearance of tiles like those in the illustrations should be taken as a warning sign that the rollers need cleaning: this is, in any case, always the first measure that needs to be taken. This defect always stems from tiles which are not lying flat on the rollers. There are two possible causes: 1. Evident heat-induced tile deformation destabilises the tiles, causing them to pitch, roll and rotate around their centre and advance or retreat with respect to the other tiles in the row (fig. 73).

Fig. 73.

They thus tend to overrun the inter-tile gap (spacer) and bump into each other or overlap. The best approach is to adjust the temperature curve and eliminate or attenuate the differences that caused tile deformation in the first place. 2. The quality and characteristics of rollers in the firing zone. Ceramic rollers, now commonplace, must meet certain dimensional requisites (inter-roller gap, temperature range, load bearing capacity, geometric uniformity) and have certain qualitative characteristics (hot bending strength, modulus of elasticity etc.). Observation of any kiln shows that rollers flex constantly under the weight of the tiles. However, because load density is never fully homogeneous or continuous, gaps between rows and poor tile alignment sometimes allow rollers to straighten out momentarily. 399


It thus follows that where properly arranged, aligned, straight tiles jump from a flexed or bent roller to a straight one and vice versa, asymmetric non-identical deformations can result. The problem is intensified by worn, dirty rollers and the random mixing of rollers of different quality. Solutions: Adjust tile feed density so as to minimise inter-row gaps: a compact load makes it harder for individual tiles to leave the formation and thus re-arrange it. Depending on planarity priorities, use temperature settings that prevent concavity in preheating, and avoid tiles moving around and becoming disordered or out of line. Use specially shaped rollers that are thicker in the middle: these correct and compensate for the arch-shaped rows that stem from relatively higher speeds at the sides of the kiln. A number of these rollers, properly distributed, can improve the situation considerably. In the rapid cooling zone, the beginning of which is especially critical, the often evident roller deformation produced by temperature differences between the heated part of the load and the simultaneously cooled part can be disastrous. The solution lies in reducing roller blowing to a minimum. Stability and reliability can also be increased by using rollers of high thermal conductivity (in silicon carbide or blends of silicon carbide and refractory) that deform little if at all. Sometimes, the only answer is to replace a number of rollers with others of better quality, especially in the firing zone (less bending under load). Such defects have no easy diagnosis or solution. The best approach is to analyse the situation from several angles and apply a blend of possible solutions. Other deformations regard only porcelain tile manufacture. Well-defined strips that droop at the leading and trailing ends (fig. 74) is one such example. This problem is more frequent with thin, highly vitrified products.

Fig. 74.

This defect is caused by intense softening and inter-roller distance: as the tile passes from one roller to the next, overhanging portions droop under their own weight. The only solution is to drastically reduce below-roller temperature at the end of the firing zone; this may involve shutdown of some burners and/or use of blown air. Vitrification is affected only slightly. A wide variety of evident, widespread warping is more likely to be seen on double 400

Defects

filling products (even where pressing results seem good); the problem, however, is also seen with traditional die filling. Rapid cooling, with abundant quantities of air blown through above-roller tubes only, sometimes provides an effective solution. Directing air jets upwards and compacting the load by slowing down roller speeds increases the effectiveness of this adjustment. Nevertheless, in other cases it may be useful to use only blowers below the rollers and direct the jets upwards at the tiles. Hard and fast rules are elusive and manufacturers will need to evaluate problems and take action on a case by case basis. Monoporosa The complexities of the monoporosa firing process have yet to be fully understood and are still a cause of much debate. However, attentive observation in a variety of situations has already produced a wealth of experience that allows the monoporosa technician to solve at least the most common problems. Convex planarity (Monoporosa) Convex planarity occurs during cooling. It is particularly evident during the allotropic β→α quartz conversion at around 573 °C. This phenomenon is easily observed by removing a tile via an inspection hatch near the rapid cooling zone: hold the piece with a pair of tongs and it bends before your eyes. The extent of deformation largely depends on the tensions generated by the different expansion coefficients of body, glaze and engobe, their relative thickness, the extent of back pattern protrusion, pressed tile density, ambient moisture content and tile ageing etc. The role played by individual parameters varies from product to product. Fig. 75 shows how expansion varies over the firing range. It goes without saying that quality tiles are only obtained where such parameters are optimised. As long as tile design error remains within reasonable limits there are several ways of nullifying it via kiln adjustment. Bear in mind that the porous nature of the body leads to absorption of atmospheric moisture and, consequently, moisture expansion. If that expansion overstretches the glaze the end result is crazing. Hence reasonable error means that the tile body must still contract more than the glaze during cooling, thus ensuring that latter is compressed, albeit minimally, under all circumstances. Because porous bodies can expand on absorbing moisture, the tile must leave the kiln with its glaze under sufficient compression so that it always remains so throughout the life of the tile. Otherwise crazing results when the glaze comes under tension. 401

EXPANSION %


C° Fig. 75.

Correction in Preheating The term PREHEATING usually refers to the low temperature zone in which degassing occurs, as opposed to the high temperature zone where melting, shrinkage and thus planarity control take place. With monoporosa that distinction is no longer applicable because, between 920 and 1020 °C calcium carbonate decomposes and generates considerable CO2 outflow and, at the same time, the most important expansion event of the entire firing process takes place. Anti-convexity measures generally involve application of higher temperatures below the rollers, a measure that is most effective from 950 °C onwards. Similar measures are often taken at lower preheating temperatures; rarely successful on their own, they do provide considerable benefits in conjunction with other adjustments. Planarity differences on sides parallel to the rollers are minimal, if they exist at all. No substantial differences in results have been observed by increasing/decreasing burner air volumes or increasing/decreasing flame speed. However, a gradual approach towards 950-960 °C has proved moderately beneficial. Under such temperature conditions direct in-kiln observation shows that the convex tiles pitch on the rollers (fig. 76).

Fig. 76.

402

Defects

Excessive temperature differences above and below the rollers can cause unacceptable deformities; if they are excessive the tile will be deformed at the kiln outlet as illustrated in fig. 77:

Fig. 77.

Temperature differences between above and below-roller areas may be extended as far as the 1000-1080 °C range. In this case marked concavity will probably be observed on sides parallel to the rollers and the trailing edge is always more concave than the leading edge (fig. 78): the reason for this is not yet known.

Fig. 78.

Immediately after CONVEX deformation the tile becomes visibly concave, a 3040 mm strip all the way round the perimeter of the tile appearing raised; the middle of the tile is affected to a far lesser extent. Correction in firing Anti-convexity action here generally involves application of higher temperatures above the rollers, a measure that essentially becomes effective after 1100 °C. Temperature differences between the upper and lower kiln areas are in the order of 5-30 °C. Results are fairly uniform over the four sides of the tile but a little less so over the load cross-section. Applying these temperature differences influences the part of shrinkage produced by high temperature sintering (fig. 79):

403

EXPANSION %


C°

Fig. 79.

As with floor tiles, the top of the tile receives more heat and shrinks more, thus making the piece concave (fig. 80): + Temperature + Shrinkage

– Temperature – Shrinkage Fig. 80.

Excessive temperature differences between the above and below-roller areas thus causes exiting tiles to have the defect illustrated in fig. 81.

Fig. 81.

Correction at the end of the firing zone The tail end of the firing zone (usually the last 2-4 m and, more rarely, the last 6 m) provides excellent convexity correction opportunities; results are usually homogeneous, within the required limits, and do not generate the deformation seen in fig. 82.

404

Defects

Fig. 82.

Correction usually involves shutting down pairs of burners (left-right) below the rollers, 1-4 or at most 6, and then feeding large volumes of air through them (up to 40-50 m3/h). Rather than following the logic of temperature difference = shrinking difference, this correction attenuates the effect of the body expansion on the relatively lower glaze expansion, thus delaying the moment in which the glaze adheres to the body. With the body incandescent (over 1100 °C) and the glaze molten (i.e. liquid), cooling of the body from beneath is begun in advance. The body starts to contract as per its expansion curve and the liquid glaze follows it without offering any resistance. Correction ends when the glaze stiffens (cools) and starts to follow its own expansion curve. This technique allows a part of body expansion to occur before glaze adhesion, the practical equivalent of a better glaze-body expansion match. It is not, of course, a system that allows for any increase in glaze heating (i.e. no increase in above-roller temperature allowed). If this were the case the higher temperature = greater shrinkage rule would apply, resulting in the deformation illustrated below (fig. 83).

Fig. 83.

An identical technique can be applied with floor tiles, but later, during rapid cooling. Non-uniform planarity across the kiln load in monoporosa Less efficient anti-convexity correction near the kiln walls is a common problem: the more marked the corrective adjustments the more evident this drop in efficiency becomes. This is explained by the fact that, in preheating, temperature differences between the above and below-roller areas are attenuated by the empty spaces that form near the walls (between tile load and wall) and, in the above-roller part of the firing zone, by the fact that it is physically impossible for the burners to heat the 405


area near the walls in exactly the same way as they heat the central zone. The sum of these two phenomena thus reduces correction efficiency. Solutions a) Fairly high kiln pressures should always be maintained; ideal firing zone pressures at burner level in the above-roller part of the firing zone are around 0.3-0.4 mm (water column). Note that this pressure must be generated by damping the exhaust fume flow, not by limiting hot air intake in the cooling zone. In other words, the solution works if there is an albeit modest flow from firing to cooling, not vice versa. b) Appropriate use of the chicanes fitted on the roof can improve matters; these need to be in good condition and no higher than 80-100 mm. It is essential that high temperature zones where higher above-roller temperatures have been set be defined by lowered chicanes. Conversely, any transverse walls below the rollers at the start of or inside the firing zone need to be eliminated. c) Installing semi-radial burner units above the rollers in the firing zone also helps; these must be arranged and adjusted so as not to compromise glazing results. d) Generally speaking, gradual rapid cooling settings, with blowers emitting no more air than is strictly necessary, are to be preferred. The above also applies to double fire processes (first fire and second glost fire). Non-uniform planarity over time (Monoporosa) Small planarity variations caused by small or difficult-to-monitor changes in production parameters and/or raw materials are, over a period of hours and days, quite normal: for the kiln operator, the resultant corrections are all in a days work. Macro alterations at the front of a load following a gap in production are an altogether different matter. The alterations that take place in an empty kiln (or, more correctly, one in which the carpet of tiles that physically separates the above and below-roller zones is absent) are easily understood. The burners under the rollers produce more heat than usual and the ones above them less. The newly introduced load thus enters a kiln in which energy distribution has been radically altered and effects on planarity are inevitable. Active adjustment of specific burner gas flow valves has been seen to reduce deformation, the downside being that the problem then seems to last longer. It is, perhaps, best to accept this inconvenience as inevitable, leaving gas modulation valves to operate freely, keeping deformation within as narrow a time span as possible; rejection of the first 10-20 rows following a significant gap in production is also routine practice. If post-gap tile deformation persists well beyond those first 10-20 rows, thermoregulator and gas modulating valve reaction times need to be speeded up.

406

Defects

Non-uniformity of size across the load cross-section in monoporosa While monoporosa is characterised by firing shrinkages of around 1%, heterogeneous shrinkages of considerable magnitude are far from rare, especially with large tiles. As with floor tiles, the defect takes the form of increased size at the sides of the kiln load. However, the dynamics behind the defect are quite different: with floor tiles the root cause is lower temperatures near the firing zone walls. With monoporosa, instead, the problem lies in overly-fast heating of the load sides within the 950-1000 °C range, during the fast contraction that correlates with carbonate decomposition, and the way in which expansion and shrinkage overlap and interfere with each other (fig. 84). temperature expansion – temperature + shrinkage +

EXPANSION

SHRINKAGE

Edge shrinkage is obstructed by expansion of the centre: = the edge shrinks less than it should

EXPANSION

Edge expansion is boosted by greater residual expansion of centre = the edge remains larger.

Fig. 84.

There is no clear-cut solution to this problem. Experience has shown that spacing the tile rows apart helps as it homogenises the gap situation at the sides of the load. This evens out the size variation by bringing all the tiles towards the larger end of the range. Some success has also been obtained by adjusting pre-heating burners to reduce temperatures at the load sides. Surer results, however, are to be had by reducing temperatures beneath the rollers, especially in the 950-1000 °C range (if planarity control so allows). 407


It has yet to be understood why size defects are a constant with certain body formulations and completely absent with others (fig. 85).

Fig. 85.

Comparison of expansion curves for problematic and problem-free bodies highlights that: The problem is common where bodies have a high percentage of calcium carbonate. Size problems exist where the contraction associated with carbonate decomposition is more marked. Further, marked expansion following carbonate decomposition is decidedly negative. Production contamination Cases vary considerably and tracing the cause of the defect is no easy task. Analysis of the circumstances under which the phenomenon is more marked, frequency and distribution across the load, glaze depth at which contamination is found, size, shape and colour all provide useful clues. A small microscope will prove indispensable. In-kiln contamination may be sub-divided into the following groups. a) Vitreous contamination b) Ceramic contamination c) Ferrous metal contamination d) Non-ferrous metal contamination e) Carbon contamination. Note that a contaminant particle produced by the kiln can never sink below the thickness of the glaze. Its exact position is found by sectioning the tile and examining the cross-section under the microscope. Contamination by grinding wheels, welding machines and debris near the production line during assembly and maintenance work is an altogether different matter. 408

Defects

a) Vitreous contamination Appears as round blotches (drops) of considerable size (1-40 mm in diameter), mostly yellow, green or brown. The phenomenon initially appears occasionally and then becomes more frequent with each passing day. Increases in intensity following significant gaps in the tile feed and substantial variations in the final preheating zone. Dependent on the presence of strong contaminants (mainly lead and boron) in glazes and frits. The vapour from the frits condenses on the roof and walls of the kiln at temperatures between 850 and 1000 °C. Excessive accumulation or lower viscosity caused by high temperatures causes it to drip onto the material in transit: the blotches have an evidently vitreous appearance. Note that condensate formation cannot be prevented by kiln adjustment: significantly increasing flow rates gives some improvement, but the practical benefits are limited and the resultant increase in energy consumption is considerable. When vitreous contamination begins to compromise sorting yield, the roof must be cleaned. The rate of condensate accumulation is sometimes high and frequent cooling of the kiln for cleaning purposes (done by chiselling and scraping) is infeasible. The alternative is to raise temperature (by 100-150 degrees C) in the 900-1000 degree °C area for about 2 hours. This sharply decreases viscosity of the condensate, causing it to run off in large quantities. Needless to say, the underlying rollers must be protected from the inevitable fouling and must thus be removed beforehand. However, if the rollers are in poor condition removal is inadvisable: in this event they should be protected by a ballast (i.e. unglazed tiles) and the rollers should be moved backward and forward for the entire cleaning period. Be careful to switch the forward/backward movement on and off frequently enough to prevent activation of the relevant safety device that shuts off the gas valves. To facilitate the increase in temperature, keep the final cooling fan running and, if effective, stop the hot air suction fan. Monitor the kiln carefully, especially the inlet where, because of the increase in pressure, there is a risk of damage (photocells, cables, rubber rollers etc.). In any event, cleaning the roof by raising temperature is an expedient that should be used sparingly. In the long run, repeated heating can compromise insulation because the refractory material becomes impregnated (note that the refractory must be cleaned with the kiln cold and that impregnation reduces cleaning efficiency). In the long run, the only proper solution to the condensate problem is an intelligent selection of compatible glazes. Condensates of a powdery aspect may also be observed at low temperatures in the pre-kiln and in preheating. Here, roof and walls can be cleaned with a jet of compressed air passed through a metal tube: the latter is inserted through the side wall hatches under the roof. Combining water with compressed air provides a more efficient car-wash spray clean. Kiln cooling and roller extraction are unnecessary. b) Ceramic contamination Ceramic contamination appears as a relatively glossy, small, rounded dark spot, perceptible when scratched with a fingernail. 409


Observation with a microscope reveals an aggregate-like aspect typical of ceramic material and no halo: cavities are absent. This is the fine dust that accumulates as dunes anywhere there is turbulence and then detaches as small flakes because of excessive accumulation, vibration or thermo-dimensional variations. Flakes may also be produced by overly compact loads (tiles pitching against each other as they move over the rollers). Always keep the area around the kiln clean to prevent circulation of dust, bearing in mind that the kiln interior is a low pressure area in which air currents from all directions converge; this is because large volumes are aspirated by fans and discharged from the chimneys and because of ascending currents produced by the hot mass of the kiln. Installing good filters at the combustion air and rapid cooling fan inlets is an effective preventive measure; these filters must be kept scrupulously clean. The kiln areas involved in this problem are: Kiln inlet Loading machine structures and guards above the tiles, where dust collects. Blowing of the material, especially where autonomous fans are used: fit fans with filters and periodically clean blower tubes. Pre-kiln Kiln inlet door, above the load: brush and vacuum-clean. Roof, fume outlets and walls above the roller plane should be cleaned with compressed air (via a metal tube inserted through special hatches). Edges of the ambient air intake in the roof at the end of the pre-kiln: brush and vacuum-clean. Burner Only burners above the roller plane are involved. However, those below the rollers still require regular cleaning to ensure their efficiency. Sand the heater and electrodes and blow with compressed air. Clean burner casing with compressed air, tap the air valve and the flex hose fitting and clean with compressed air. Rapid cooling Where contamination originates here it is rigid or just partly molten, rough to the touch and always on the surface. Clean the blowers periodically: blow air through them simply by detaching the flex hose from one side and knock them to detach contamination. Blow air through the air regulation valve and the flex hose. Directing the below-roller blowers downwards can lift dust and waste which inevitably settles on the back of the tiles. However, such contamination ends immediately after the blowers are re-oriented. In rare cases, and only in double fire operations, the unscraped, under-tile flash 410

Defects

produced by presses with worn punches tears off during the rapid cooling phase and falls onto the still soft glaze. This type of contamination takes the form of fragments of elongated shape with ragged edges sticking to the still molten glaze. Debris from exploded tiles and the waste produced when workers walk over the kiln roof or adjust the chicanes and tube valves can also cause problems (but only occasionally). Blowing in final cooling is never a source of contamination as this area of the kiln is already a long way from any molten glaze. c) Ferrous metal contamination Appears as a black speck characterized by a small brown ring around a black nucleus, clearly visible with the microscope; quite frequent but very rarely produced by the kiln. To identify the cause section the tile and identify where the problem lies. If the contamination is produced by the kiln, the defect will always be superficial and poorly amalgamated with the glaze. Possible sources of contamination: The kiln inlet door (above the rollers) and the piping above it. Fume outlets in the kiln, especially where the system is subject to frequent power failures that cause formation and stagnation of acid condensate and moisture. Rub down and brush the door. Pierce the piping insulation at several points and tap with a hammer and chisel so that any unstable encrusting breaks off. The rare event of oxidation inside the rapid cooling piping should be monitored by placing special filters at the consumption points and inspecting pipes internally. It is inadvisable to direct the above-roller blowers at tiles directly as contamination does not sink into the glaze and remains very rough to the touch. d) Non-ferrous metal contamination Shows up well under microscopic observation on account of its glossy crystalline appearance, reminiscent of cast iron fragments, and the absence of the brown ring produced by partial oxidation of the iron. Possible sources of contamination include all the exposed stainless steel and refractory parts in the firing channel above the load and therefore: Fibre plate anchors in the pre-kiln roof. Generally small particles which include body and condensate, these increase markedly in intensity and density in post-gap loads as a consequence of the increase in pre-kiln temperature. Clean with compressed air, using a metal tube inserted through the doors. As soon as the kiln cools brush thoroughly.

411


Burners Particularly heavy duty working conditions can cause the heater to overheat and deteriorate or wear out. Contamination produced in this manner is then put into circulation by the fast flame. Contaminant particles are always small and round, their frequency and diffusion remaining constant and uniform. The same applies to excessively long starting electrodes: in this case shortening the electrodes, thus eliminating the part that shows signs of carbonisation, usually suffices. Rapid cooling blowers These produce very small powder-like black specks; contamination increases sharply following gaps in tile feed. These metal specks remain on the surface of the tiles and are rough to the touch. Contamination is caused by deterioration of the blowers above the roller surface, usually the 2 or 3 closest to the firing section. Blower deterioration is caused by overheating of the steel: where cooling is poorly adjusted deterioration can happen quickly. Regulate rapid cooling so as to ensure minimum blower cooling when the blown air modulation valve is open to minimum: set pressure no lower than 10 mm approx (water column). Under no circumstances should rapid cooling fan be stopped during the transitory phases of kiln heating or when there are gaps in the load. Avoid exposing blowers to fumes in the firing section by adjusting chicane height between firing and cooling accordingly: inspect chicanes regularly to make sure they are intact. Avoid cooling settings that suck firing fumes towards the rapid cooling zone. If blower degradation is within reasonable limits, detaching the flexible pipe from one side and blowing out the accumulated waste will suffice. For more pronounced degradation, remove the blowers, brush them and knock them. Beyond a certain limit, especially if blowers are evidently deformed, their replacement becomes absolutely necessary. Lining those blowers most at risk with ceramic fibre prevents contamination with waste produced on the surface exposed to the firing channel. Today, silicon carbide blowers are often used. While costly, this material avoids the above problems. Blower tubes made of refractory alumino-silicates can be broken by thermal shock: when breakage occurs jamming of the tiles is inevitably involved. Heat exchanger Only rarely a source of contamination, usually because of the care and attention involved in its design. This type of contamination can easily be identified by the shape of the metal flakes: lamellar with sharp edges. In this event it is necessary to brush down the exchanger (with the kiln cold, of course). 412

Defects

e) Carbon contamination Infrequent and limited to cases where use of fuels such as L.P.G. or liquid fuels results in poor combustion and thus formation of carbon in the burners. Appears as small black specks, opaque and ringless (without halo). Regular cleaning of the burners is essential, as is monitoring and correction of air-fuel regulation.

413

Printed September 2002 by Tipografia Moderna di Ravenna for Editrice La Mandragora of Imola

Sacmi Vol 2 Inglese - II Edizione

Recommend Documents