Commission of the European Communities
technical steel research Properties and service performance
Measurement and interpretation of dynamic loads in bridges Phase 3 Fatigue behaviour of orthotropic steel decks
Synthesis Report EUR 13378 EN
3 Commission of the European Communities
technical steel research Properties and service performance
Measurement and interpretation of dynamic loads in bridges Phase 3 Fatigue behaviour of orthotopic steel decks
Edited by:
A. Bruls Service 'Ponts et Charpentes' Université de Liège Quai Banning 6 B-4000 Liège
Contract No 7210-KD/119/201 /317/411 /609/807 (1 July 1986 to 31 December 1988)
Synthesis report
Directorate-General Science, Research and Development
1991
PARI [UROP Biblioth. N.C.
EUR 13378 EN
Published by the COMMISSION OF THE EUROPEAN COMMUNITIES Directorate-General Telecommunications, Information Industries and Innovation L-2920 Luxembourg
LEGAL NOTICE Neither the Commission of the European Communities nor any person acting on behalf of the Commission is responsible for the use which might be made of the following information
Cataloguing data can be found at the end of this publication
Luxembourg: Office for Official Publications of the European Communities, 1991 ISBN 92-826-0532-9
Catalogue number: CD-NA-13378-EN-C © ECSC-EEC-EAEC, Brussels • Luxembourg, 1991 Printed in Belgium
SUMMARY. This research, carried out with the financial help of the ECCS, concerned the fatigue strength of orthotropic steel decks of road bridges. It followed two phases that were concerned with the collection of traffic data and measurement of stresses produced in bridges. Fatigue tests under constant and variable amplitude were carried out on stiffener-plate connections, stiffener-stiffener connections with U and V shapes, and stiffener cross-beam connections. From the tests results and calculations some conclusions can be drawn which are directly usable in bridge design. However, some unexpected behaviour occured and some connections need more investigation.
Résumé. Cette recherche, réalisée avec l'aide financière de la CECA, concernait la résistance à la fatigue des dalles orthotropes de ponts-routes. Elle faisait suite à deux phases qui se sont concentrées sur la collecte de données relatives aux charges du traffic et aux contraintes produites dans les ponts. Les essais de fatigue sous amplitude constante et variable ont été réalisés sur les assemblages raidisseur en U-tôles, les assemblages raidisseur-raidissèur en U et en V, les assemblages raidisseurs-entretoise. Les résultats des essais et les calculs ont permis de tirer des conclusions directement applicables au calcul des ponts-routes. Néanmoins des comportements imprévisibles s'étant manifestés, certains assemblages demandent des investigations complémentaires.
Zusammenfassung. Dieses Forschungsvorhaben wurde mit finanzieller Unterstützung der EGKS durchgeführt und betraf das ErmUdungsverhalten orthotroper Platten von Strassenbrüchen. Es folgte auf zwei Phasen, in denen hauptsachlich Daten über Verkehrslasten sowie Beanspruchungen von Brücken gesammelt wurden. Ermüdungsversuche unter konstanten und variablen Amplituden wurden für folgende Verbindungen durchgeführt : U-Längsverstelfung/Blech, U-und V-Längsversteifung, und Längsversteifung/Quertrager. Die Ergebnisse der Versuche und der Rechnungen erlaubten Schussfolgerungen, die direkt für die Auslegung von Brücken anwendbar sind. Aufgrund unerwarteter Verhaltenswesen der untersuchten Bauteile sind jedoch zusatzliche Forschungen notwendig.
This report is a synthesis of the final reports performed by each Laboratory that has participated to the common research :
1. H.LEHRKE, Fraunhofer Institut fUr Betriebsfestigkeit, Bartningstrasse H7
_
6100 Darmstadt - Germany. [1] 2. A. BRULS, E. POLEUR, Service "Ponts.et Charpentes", Université de Liège, 6, quai Banning - 1)000 Liège - Belgique. [2] 3. A. BIGNONNET, I.R.S.I.D. 78105 Saint-Germain-en-Laye - France, and J. CARRACILLI, B. JACOB, L.C.P.C., 58, Bd. Lefebvre - 75732 Paris - France.
C3] k.
S. CARAMELLI, P. CROCE, M. FR0LI, L. SANPAOLESI, Istituto di Scienza delle Costruzioni, Università di Pisa, Via Diotisalvi, 2 - 56126 Pisa - Italy.
cu 5. H. KOLSTEIN, J. DE BACK, Stevin Laboratory, Universiteit van Delft, 2628 CN Delft, Nederland. [5] 6. C. BEALES, Transport and Road Research Laboratory, Old Wokingham Road, Crowthorne, United Kingdom [6].
Coordination
: A. BRULS
IV
CONTENTS Page SUMMARY
111
RESUME
111
ZUSAMMENFASSUNG
111
1.
INTRODUCTION
1
1.1
Orthotropic steel deck
1
1.2
Details tested
1
2.
METHODOLOGY OF FATIGUE PREDICTION
5
2.1 2.2
Traffic loads and effects on bridges Classical life calculation
5 6
2.3
The fracture mechanic approach
9
3.
CONNECTION STIFFENER-PLATE
12
3.1 3.2 3.3 3.4 3.5 3.6 4. 4.1 4.2 4.3 4.4 4.5 4.6
Types of connection Stress determination Fatigue testing and results Fatigue life calculation Crack propagation and lifetime calculations Conclusion CONNECTION STIFFENER-STIFFENER Types of connections Stress determination Test results of the University of Pisa Test results of the T.U. DELFT Comparison with other research programs Conclusions
12 12 15 19 20 22 40 40 40 42 45 51 51
5.
CONNECTION STIFFENER-CROSSBEAM
70
.5.1 5.2 5.3 5.4 5.5
Types of connection Stress determination Test results of TRRL Test results of the T.U. Delft Test results of the L.B.F.
70 70 72 75 77
5.6
Conclusions
79
6.
ORTHOTROPIC DECK TO CROSSBEAM CONNECTION
106
7.
APPLICATIONS
111
8.
CONCLUSIONS
112
BIBLIOGRAPHY
115
1. INTRODUCTION. 1.1. ORTHOTROPIC STEEL DECKS (Fig. 1.1). Orthotropic steel decks are used in bridges with long spans and in movable bridges in which dead weight has to be as low as possible. The upper part of these decks is composed of a plate on which the traffic runs. This plate is covered by a thin surfacing (7 to 12 mm) or an asphalt surfacing (MO to 70 mm). Longitudinal stiffeners are welded to the under part of the deck plate, approximately
300 mm. apart, parallel
to the direction of the
traffic lanes. They are usually closed sections with trapezoidal or "V" shapes although open sections are sometimes used. The stiffeners transmit loads to crossbeams to which they are connected. Crossbeams are normally spaced at 3 to 5 meters and are connected to main girders or diaphrams. Orthotropic
steel decks are very
sensitive
to fatigue damage
because they are directly subjected to the actions of wheel loads which give rise to stress ranges which are high compared to the dead load stresses especially if the surfacing is thin. In order to understand the fatigue behaviour of orthotropic steel decks, the path of the traffic loads must be considered. Under the action of wheels, the deck plate acts as a beam on elastic supports (the stiffener webs). The elasticity of the supports decrases with the spacing of the crossbeam. (support
The web of the stiffeners are reaction)
and
to
bending
moment
subjected if
to normal forces
stiffeners
have
closed
sections. Longitudinally
the
stiffeners are subjected
to shear forces,
bending and torsion moments. The butt-welds connecting the longitudinal stiffeners and the stiffener to crossbeam connection bear these effects. 1.2. Details tested. This research concerns
the study of
the fatigue
strength of
orthotropic decks with closed stiffeners. Bibliographical researches and bridge examinations made during the two first phases have enabled us to define details which are most sensitive to fatigue damage (Fig. 1.1). -1 -
* detail 1
stiffener to deckplate connection,
* detail 2
stiffener to stiffener connection,
* detail 3
stiffener to crossbeam connection,
* detail H
bolted connection crossbeam-orthotropic plate.
For and/or
by
each
detail,
measurement
in
the the
stress
field
was
studied
laboratory. Afterwards
by
calculation
constant
amplitude
fatigue tests were carried out to define the S-N curves. Finally specimens were
tested
under
variable
amplitude
load
cycles
simulating
traffic
effects.
Distribution of the work between laboratories was as follows : * detail 1 :
stiffener to deck plate connection :
- University of Liège analysed local stresses with the help of a finite band program and tested the connection under constant and variable amplitude loading ;
- IRSID carried out constant amplitude
tests on this connection
with a different welding procedure to that in the University of Liège.
-
L.C.P.C. developed and applied a fracture mechanics model.
* Detail 2 : stiffener to stiffener connection :
-
University of Pisa measured and calculated stresses on a full size orthotropic deck ;
-
University
of Liège calculated stress histograms produced by
traffic loads using the results of the 1st and 2nd phases ;
-
University of Pisa and T.U. Delft tested different designs of this detail.
* Detail 3 : Stiffener to crossbeam connection. - T.R.R.L. measured stresses in a full
size orthotropic deck and
in an actual bridge ; - L.B.F. calculated stresses in the web of the crossbeam
with
the help of a finite element program. - University of Liège calculated stress histograms produced by traffic loads , using the results of the 1st and 2nd phases. - T.R.R.L, T.U. Delft and L.B.F. tested different designs of this detail. * Detail 1J : bolted connection crossbeam-orthotropic plate. - University of Liège tested this detail. Using the traffic load data from the 1st and 2nd phases and the fatigue data from the current research, it has been possible to make some important conlusions about the fatigue behaviour of orthotropic steel decks.
Longitudinal deck plate butt weld
Transverse deck plate butt weld
Deck plate
/ detail ^
Longitudinal stiffener to deck plate
Longitudinal stiffener to crossbeam weld
Longitudinal stiffener
detail ( 5 )
Longitudinal stiffener splice welds
Crossbeam to deck plate weld
Figure 1.1
detail © stiffener to crossbeam connection (alternative connections)
Main welded connections in a typical o r t h o t o p i c bridge deck
Crossbeam
Diaphragm
2. METHODOLOGY OF FATIGUE PREDICTION. 2.1. Traffic loads and effects on bridges. During
the
first
and
second
phase
of
the
research
[7][8],
measurements of traffic loads and stresses were carried out on 1 -4 highway bridges. Different types of bridges and spans from 13 to 1000 m were examined, most of them included orthotropic steel decks. The traffic on these bridges was recorded and the number and types of commercial vehicles identified and classified in a uniform scheme. Axle loads, were measured by weighbridges, and axle spacings, separation between vehicles and lateral wheel positions were also recorded. Additional sensors simultaneously recorded the stresses produced in structural components like main and cross girders, longitudinal
stiffeners and deck plates.
The recorded stress histories were analysed using Level-crossing and Rain-flow cycle counting methods. The resulting spectra were assessed in terms of their fatigue damage potential with reference to constant stress range fatigue design curves from existing or draft codes. Special attention was paid to the possibility of calculating by means of finite elements, and to the temperature dependent effect of the asphalt surfacing in reducing the stresses. Based on the experimental data a method of computer simulation of traffic loads and stresses induced was developed and tested. It has proven a suitable tool not only to reproduce the experimental data, but also to derive more general results in that any uniform or hypothetical traffic situation may be studied for the bridges investigated or for any components of bridges described in terms of influence lines. The main conclusions were : 1. The level of the measured axle and vehicle loads are much higher than the allowed loads. Despite this observation, the measured stresses are never higher than 90 N/mm2 ; which is not very much. 2. The number of vehicles is so high for some traffic, that exclusing the vibration effect, the number of cycles produced during a life of 100 a
years may reach 10° • which is very high.
3. The recording of a traffic flow comprising about 10 000 axles higher than 10 kN will be sufficient to establish a spectrum reliably and to allow a fatigue calculation. Two types of vehicle, the articulated 2-axled trailer with a 2-axled semi trailer or with a 3_axled semi trailer, have been found to produce about 40 to 50 % of the total fatigue damage on Dutch bridges, although they
represent
only
6
to
18
?
of
the total number
of
commercial
vehicles. i\. As steel decks support high and frequent local stress ranges caused by wheel loads, a main part of the work was concentrated on orthotropic steel decks. During
the 2nd phase of the research,
the most
aggressive traffic was
recorded in the Netherlands on the Rheden bridge, where the frequency of articulated lorries was highest. Therefore, in this work the Rheden traffic was mainly considered for the determination (by simulation) of the stress spectra used in fatigue tests. During the period of the third phase, new measurements of traffic
flow were made
(independently of this work) in
France, Germany and Italy. These data suggest that the load of the vehicles has not changed very much, but the number of articulated lorries and the number of loaded lorries has increased. The results obtained by using the Rheden traffic are usable, but the number of cycles are higher for some recent traffic [9]. It is clear that the development of steel bridges needs the knowlegde of the fatigue behaviour of the details of an orthotropic deck under variable amplitude
load and a high number of cycles ; thus it needs the fatigue
design curve for stress ranges beyond 2.10
cycles.
2.2. Classical fatigue life calculation. 2.2.1. Miner's rule. During stress
spectra
necessary
to
the
life of
induced translate
by
a bridge, traffic.
the
actual
its components
For
the fatigue
stress
spectra
are subjected to calculation
by
a
stress
it is range
histogram. Cycle counting methods such as Rain-flow or Range-Pair are commonly used to produce stress range histrograms ; mean stress is not usually considered. A bridge is thus influenced by variable stress ranges, occuring randomly, from the vehicle loads running on the bridge. The fatigue
behaviour
of the
curves. -6 -
details
are
characterised by S-N
Fatigue damage is calculated using Miner's rule
D = Z AOl-0 with
(ÜL) N.
n. = number of cycles in the stress range histrogram corresponding to Aa. measured or calculated during a time t. N,
= number of cycles corresponding to Aa. in the S-N curve. That is the number of cycles with a stress range of Aa. producing failure.
If D < 1 : no failure D - 1 : failure The remaining fatigue life is — g —
t.
In Eurocode the fatigue classification of a detail is defined by the value Aa
corresponding to N
C
=2.10
cycles.
c
Eurocode 3 considers
SN curves with two slopes (Fig. 2.1) :
N.Aa3 = este = 5.106. Ao 3 N.Aa
5
6 5 = este - 5.10 . Aa D
N « »
if
Aa £ AaQ
if
Ao^ D
> Aa i AaT L
if
Aa.
> Aa
Li
where, Aan corresponds to N =5.10 cycles ; u o g AaTcorresponds to N = 1 0 cycles. L
Li
The fatigue damage corresponds to Aa
DE =
D
Z
U
n. . Aa5,
(_i
lower
L_) +
Aa=Aa L 5.106.Aa 1
The
"
part of the curve
5 D
n. . Ao\
E Aa
i
Ao
(_j__J_)
D
5.10
6
.Aa3 D
(m = 5) is important
experience large numbers of low stress cycles.
- 7
in bridges which
2.2.2. Existing codes. To design a structure, designers have to find information in codes about actions, strengths and calculation methods. For the fatigue assessment of bridges most of the codes propose the Miner's rule as the calculation method. The action (traffic) is idealised by one or a set of standard vehicles and the strengths are presented in the form of S-N curves. A. Action. The action to consider in the fatigue assessment of bridges is the action of traffic. During the first and
second phases of the research,
measurements were made of traffic including the frequencies of vehicle types and the level of loading of lorries. The traffic were compared to try to detect local, regional or international influences. It is not possible to define a spectrum taking all these influences into account. In existing codes actions are defined by one, two or three of the following load models : - one vehicle moving on an influence line allows the calculation of the highest stress range (NBN 5 [10], BS 5^00 [11]) ; - one vehicule moving on an influence line allows the calculation of the stress range histogram (.La., n. ) by a cycle counting method (BS 5*100). - a set of vehicles moving on the influence line allows the calculation of the stress range histrogram (Aai, ni) by a cycle counting method (BS 5100). B. Strength. The different SN curves existing in codes are presented in Fig. 2.1. The main remarks are : - for constant amplitude loading, the S-N curves often have a slope of -1/3 (m = 3) with an endurance limit. Previously the endurance limit corresponded to N = 2.10 ,now it is associated with N - 5.10
(NBN)
[10], EC3 [12]) or 107(BS 5^00 [11], NEN [13]). - for variable amplitude loading, the S-N curve is continued beyond the endurance limit with a slope corresponding to m (NBN), m+2 (BS 5^00) or 2 m -1 (EC3). o
-
EC3 proposes a cut-off value at N - 10
below which the variable
amplitude loading does not produce any fatigue damage. -8 -
I f the maximum value of the stress range A o m a x is below the endurance limit, it is assumed that there is no fatigue damage (NBN).
C. Calculation. A first verification, which
is simple
and conservative, is to
calculate the maximum stress range for the detail under the load model. This value is compared with the endurance limit (NBN 5) or with charts of limiting stress which vary according to the bridge span and detail class (BS 5^00). The calculated stress range must be below the limiting stress for the detail to be acceptable by this method. The second verification is more precise. I t involves a fatigue damage calculation on the basis of the (Ao,
n ) histogram defined by i
ii
the load model, using the Miner's rule. 2.3. The fracture mechanic approach. The only method
suitable
for bridge
design requires a Miner's
calculation, but the crack growth evaluation
in an existing bridge
needs a fracture mechanic approach. The classical Miner's calculation is very simple and does not take into account the changes in the structure when a crack is propagating, or
the
mean
stress
level
onto
which
the
stress
variations
are
superimposed. Hence this model is very sensitive to the choice of the SN
curves. For
generally
much
road
bridges
smaller
than
in
which
the
the
permanent
stress
variations
stresses,
the
are
computed
lifetimes are highly dependent on the high endurance end of the SN curves.
In
the
fracture
mechanic
approach,
the
crack
propagation
is
computed for each stress cycle, taking into account the present state of the structure and hence the time history of the stress variations. There are numerous laws describing this evolution ; the Paris one has been chosen because of its simplicity and its ability to describe the crack propagation
in this type of structure. The crack propagation
speed is written as : da ——
m
■ c ÅK
t
where
c and m are
material
and
AK
depending
on
the
parameters
the
intensity
factor,
stress range and on the local
geometry of the detail. 9
stress
depending on the
In order to account for a threshold, under which no crack may be propaged, the Paris law is modified as da
_ dN
ra
=
c' (AK - AKs )
'
The threshold plays a similar rule than the classical fatigue limit. If the stress are both positive and negative, only tension is considered for the crack propagation. But in the real bridges, it is generally assumed that
the residual
stresses
induce
high forces, and the whole stress
variation are considered. The crack propagation time is obtained by integrating this formula from the initial crack length a to its .length a at the failure. The number of o r stress cycles involved is :
N = N
1 + _! C'
where N
a
ƒ
a
r
m ' (AK - AK ) da
O
is the number of cycles for the crack initiation. The failure
criteria adopted here is : a
0.5e, if e is the plate thickness ; it
corresponds to the loss of rigidity of the structure. The problem is then to compute the AK values. The cracked beam theory shows that AK may be generally written as : AK = /la.Ao.f(a), where f(a) is a function of the structure geometry and Aa is the stress range in the uncracked section studied. For the orthotropic deck stiffener-plate connections, the following formula may be adopted :
AK = /HaCA F + 2a_ A, F + _*1 A„ F„ + ^ ì L A F + Ü _ A„ F„] oo f 11 33 8 ^ ^ 2 2 2 3 í where the F. are amplification factors depending on the crack geometry but independent on the loading
case. The A.
are
the coefficients of the
smoothing polynominal of the stress diagram. This model has been applied by the LCPC to the stiffener-plate connection behaviour in the chapter 3.
10
s.ioV
•N(log)
BS5¿00
Ao. (log)
N(log)
NEN 2063 Ao ,
AT
(log)
AOc AOn. AOL
4
(log)
^"""""^IT i "■"■"■»«^ ■ "^««^
2m1
cutoff
cutoff Nllog)
2.10*
S.10*
1o'
2.10' S .V*
EC3
10*
EC3 Figure 2.1 : S-N curves
11 -
3. CONNECTION STIFFENER-PLATE. 3.1. Types of connection. 3.1.1. General considerations. The main parts of the stiffener-deck plate connection are the deck plate, the stiffener and the weld between them. Usually the deck plate is between
10 or 12 mm thick and the
surfacing thickness is around 10 mm or around 60 mm depending of the material-. The stiffeners studied
are
closed
trapézoïdal
or "V" shaped.
Stiffener dimensions are about 300 mm wide, 250 mm high and 6 mm thick. They are typically placed 300 mm apart. Because of the closed section of stiffeners, welding is carried out on only one side of the stiffener web. Other types of open stiffeners that were used in earlier bridges are not considered in this research. 3.1.2. Welding prodecures. Welding procedures have evolved
: originally the procedure was
manual metal arc (tests carried out by Maddox [1*1] and JANSS [15]), now it is often automatic submerged arc welding. Procedures are still evolving to obtain a smaller lack of penetration (tested in this work). Several tests have been carried out at IRSID to optimize the welding procedure. The influence of the edge preparation has been checked and comparisions made chamfer. The
between edges
welding
energy
chamfered
was
adjusted
at 60° or 45° and without to
minimize
the
lack
of
penetration. From several sets of welding tests it has been shown that without edge preparation and without chamfer, a satisfactory penetration (lack of penetration £ 1 mm) is obtained for a welding energy of 20 Kj/cm, (figure 3.1), even with a gap of 2 mm between the top of the stiffener web and the deck plate. 3.2. Stress determination. 3.2.1 . Measurements and calculations. Stresses in the deck plate to stiffener connection are induced by the direct effects of the wheel load, the orthotropic steel deck behaves as a beam on elastic supports (stiffener webs) (Fig. 3.2.).
12 -
During the two first phases, the stresses near the weld were measured under traffic loads and with a test vehicle of known axle load. The stresses were influenced by the temperature and by the distribution of wheel loads, both in magnitude and transverse position. The temperature affects the stiffeners of the surfacing and its composite action steel plate.
with the
Another factor is that it is not possib le to measure a
stress at the crack initiation point. In order to have a general approach to the behaviour of the welded connection under a wheel load, calculation program.
method.
The
frame
Stresses used
for
it is necessary to develop a stress
have the
b een calculated b y a finite b and calculation
has
the
geometry
of
orthotropic decks found in Belgian bridges (Fig. 3.3.a.). The points where stresses are calculated are located in the neighbourhood of the weld (Fig. 3-3.b) ; points A' , 3' : in the deck plate ; points C' , D' : in the weld. Axial and b ending stresses are calculated in the cross section in which they are the highest (section 0 - axis 1 - Fig. 3.3.a). With the finite band program it is possible to study the influence of the following parameters : - longitudinal location of wheel : longitudinal influence lines are drawn in Fig. 3.11. It appears that at points A' and B' (deck plate) stress values change sign at a certain distance from section 0. Thus the stress amplitude at point A' and 3' is higher than the maximum stress obtained when the wheel is on axis 1 ;
- transverse location of wheel : results are presented in the form of transverse influence lines (Fig. 3.5 and 3-6). ■ surfacing thickness : two surfacing thicknesses are considered : - deck without surfacing : load is not distributed ; - deck with a 60 mm surfacing thickness : load distrib utes through the thickness at an angle of 1)5° ; no composite effect is considered. - dimensions of wheel contact area : different sizes of wheel are studied.
13
3.2.2. Stress histograms. The calculation of the stress histograms used for the variable amplitude tests were made using the simulation program of Liège [2][7] [8]. For each axle of the traffic the program chose at random a transverse position on the deck plate corresponding to a value of the transverse influence line. This value, multiplied by the load of the axle gave the stress induced. The data introduced in the simulation program are : 1. Traffic : the vehicle axles of the Rheden traffic were divided into four groups according to their wheel type. 2. Transverse distribtion : the transverse distribution of the vehicles was obtained from measurements made during the 1st and the 2nd phases. 3. Transverse influence lines : calculated in section 3.2.1. The histogram obtained is presented in table 3.1 and figure 3.7.. It was used for the variable amplitude tests. 3.2.3. Equivalent stress range. To compare the variable amplitude test results with the constant amplitude ones, the applied stress spectra were analysed according to Miner's rule in two different ways, each using an S-N curve with a slop of -1/3. a. An equivalent stress range La was calculated in a way that n-cycles of that stress range have the same fatigue damaging potential as n-cycles of the stress spectrum, using a third power relationship ; z -T7T n . Âio .l' ) 1 / 3 MPa E n. b. The equivalent values proposed by the University of Liège [25] in the report of the 2d phase [25], corresponds to the centre of gravity of the damage distribution (see Fig. 3.7) ;
*°ae
"
(
Z n<Åo<
Ao - _ _ L _ _ 3 m I niAoi
MPa
3
nm
I niAoi J
A°m 3 in which : Ao^ - individual stress range rU
- individual number of cycles corresponding to Aoj
Ao
ra " equivalent stress range
nm
- number of cycles belonging to
Ao m This definition is not much influenced by a cutting of the high number or low cycles which produces not much damage. In this report both method are used. 14
3.3. Fatigue testing and results. 3.3.1. Test specimens. The tests presented in this section are : * previous tests : - W.I. (Maddox) [11] - CRIF (Janss) [15] * tests of the University of Liège [2] * tests of I.R.S.I.D [3]. The
loading
mode
and
the
geometrical
characteristics
of the
specimens are given in figure 3.8. and Table 3.2. The material used is of the type E 36—^. General welding procedure characteristics are : - no edge stiffener preparation - horizontal position - one run - no preheating - no postheating. - Specific conditions : W.I. : manual arc welding C.R.I.F. : manuel arc welding I.R.S.I.D. : automatic welding (submerged arc). U.Lg. : automatic welding. The geometrical characteristics of the welds are given in table 3.2. 3.3.2. Test results presentation. Depending on the stress distribution in the deck plate and in the trough,
as
well
as
the
weld
quality
(penetration,
throat,
thickness,
undercut, ...) fatigue cracking may occur either : a) from the weld toe on the deck plate, point A fig. 3.9., developing in the deck plate. b) from the weld root in the stiffener, point D fig. 3.9., developing in the throat of the weld. To determine the stress distribution in the specimen, static tests were carried out. It is noted that in the median axis of the specimen the loading is biaxial with o2/o1 - I/3 (o1 is in the direction of the bending stresses). This phenomenon has not been taken in account in the S-N curves because designers do not consider biaxial stresses.
15
From these tests the nominal stresses to be used in the fatigue SN diagrams were derived (figure 3.9). extrapolation of the stress in the deck plate to the weldtoe, Ao. definition generally used when cracks propagate through the deck plate from the weld toe,
extrapolation of the stress
in the trough
to the weld root Ao , S
definition generally used when cracks propagate through the weld from the root of the weld. Four different failure criteria have been used : N
» crack detection by strain gauge ;
N
=■ first visible crack ;
N
= measurable charge in stiffness of the specimen, or 25 mm. long crack.
N. = end of test. The results of the tests made at IRSID and at the University of Liège are given in tables 3.3. Tests with failure in the deck plate and in the root weld are considered separately. 3.3.3. Failure in the deck plate Aorf. 3.3.3.1. Constant amplitude tests : These tests determine the fatigue strength of the deck plate at the weld toe. Results are given in Table 3.3.a and plotted in Fig. 3.10. Main conclusions are : There are no significant differences between specimens with
a 2 mm gap
between the top of the sti f f ener web and the deck plate and specimens with no gap, provided the lack of penetration is less than 2 mm. The mean Wöhler curve is determined with R
= 1 and R 0 0,1 in the
weld root for the specimen tested at IRSID. Aod
26777 N
1/3
(m = 3 imposed) .
As the standard deviation is 24 N/mm2, the characteristic value for 97,5 ? is Ao = 163 N/mm2 for Nc 2 106 cycles, de The two experiments conducted at IRSID show a lower fatigue resis tance at R = 0,1 than R » 1. However, two similar tests performed at Liège do not indicate such a detrimental effect.
16
In
only
two
U.Lg.
corresponding to R
specimens
did
failure
occur
in
the
deck
plate
» 0 (only tensile).
3.3.3.2. Variable amplitude tests : The loading histogram used for these tests is the stress spectrum calculated in section 3.2.2. (table 3.I- and Fig. 3.7). This histogram simulates traffic effects. Loads were applied at random to the test specimen. Two tests were equivalent
carried
values
(Ao m
out. They are plotted in Fig. 3.10 at their , n ) m
equivalent values correspond
calculated
from
the
histogram. The
to the centre of gravity of the damage
distribution, see figure 3.7. Results are similar to those obtained with constant amplitude loading. 3.3.I. Failure in the weld Ao . s 3.3.1.1. constant amplitude tests. Tests results are given in Table 3.3.b and plotted in Fig. 3.11. Main conclusions are : - the fatigue strength increases significantly when using automatic welding,
this
technique
allows
larger
penetration
and
throat
thickness at the weld. - The mean S-N curve is determined : Ao - 17258 N~ 1 / 3 and the characteristic value for 97,5 $ is Ao - 111 N/mm2 for N - 2.10 cycles, se c - The tests show the importance of R
ratio. To obtained the failure
in the weld it was necessary to have more tension than compression at the root of the weld : - 1
< 0 in the ULg specimens » 0,1 in the IRSID specimen.
3.3.1.2. variable amplitude fatigue tests : The loading histogram used for these tests is the stress spectra calculated in Section 3.2.2. modified to obtain failure in the weld R
s
- -0,5 instead of - 2,0 at the root for the highest stress range.
- 17 -
The results are plotted in Fig. 3.11. at their equivalent values corresponding to the centre of gravity of damage distribution. Results
are
similar
to
those
obtained
under
constant
amplitude
loading. 3.3.5. Comparison with previous research. Results from the Welding Institute [1*1], CRIF [15], University of Liège [2], and IRSID [3] are compared in Figure 3.11 in terms of nominal stress range in the trough at the weld root La
(any other representative 3
stress could be used) versus the number of cycles to failure of the trough to deck plate connection M . The main difference is that in the Liège University, WI and CRIF experiments cracks initiate at the weld root and the failure occurs in the weld, while in the IRSID experiments cracks initiate at the weld toe in the deck plate and the failure occurs in the deck plate. The results of the tests at IRSID and at the University of Liège are given in tab. 3-3. As shown in tables 3.2., the specimens
tested by CRIF and by
Maddox, welded by manual arc welding, confirm the importance of lack of penetration. It is clear from
figure 3.11 that the fatigue resistance
increases significantly when using submerged arc welding, this technique allowing larger penetration and throat' thickness of the weld. Nevertheless it is shown that cracks initiate at the weld root even with a lack of penetration
of
2
mm.
However
if
the
welding
operation
is
properly
optimised, the lack of penetration can be. limited below 1 mm. and, for alternate
or
repeated
tensile
bending
in the deck plate, the cracks
initiate at the weld toe, and the fatigue resistance is improved.
\
f.
V
/ y
18 -
3.H. Fatigue life calculation. The
characteristic stress range deduced from the fatigue tests, defined
for N
c
» 2.10
cycles following Eurocode 3, are :
— 114 N/mm2, if the crack occurs in the weld (point D) ; sc Aa. - 163 N/mm2 if the crack occurs in the plate (point A ) , dc If the traffic composition measured at Rheden is considered, the number of Aa
lorries required to cause failure may be calculated (see 3.2). The results are given in table 3.4. Putting there data in perspective, the traffic flows recorded on highways during the 1st and 2nd phases generally comprise between 1000 and 4000 lorries during a working day. Such flows produce, after 100 years, between 20 and 80.10
lorries.
The following comments may be made : 1. The fatigue life calculated in the deck (Ao ) is always a little higher than
in the weld
(Aa ) : the higher
fatigue strength is partially
offeset by higher stress ranges produced by the traffic loads ; 2. A surfacing of 60 mm gives a fatigue life 3 to 5 times longer
than the
unsurfaced deck ; 3. An increase in the deck plate thickness of 1 mm gives a fatigue life around twice as long. 4. An increase in the thickness of the stiff ener reduces the fatigue life a little. 5. The
given
fatigue
lives
are
pessimistic,
because
the
transverse
position of the traffic flow considered in the calculation is in the most damaging position. 6. The given fatigue lines are too pessimistic for surfaced decks where the calculation doesn't take in to account a composite effect, that exist mainly by low temperature. We may conclude that the required thickness of the deckplate depends of the expected number of lorries, and the thickness of the surfacing.
- 19
3.5. Crack propagation and lifetime calculations. The fracture mechanics model has been applied to the stiffenerplate connection of various structures : - the IRSID test specimens, - the two temporay bridges of Montlhery and Choisy-le-Roi, - the bridge of Caronte. 3.5.1. Test specimens. The calculation of the crack propagation lifetime for the test specimens presented in § 3.3. was made by the LCPC to calibrate and check the model. The tested and computed lifetimes are compared in table 3.5. Various stress levels have been considered from 150 to 3*40 MPa and the mean stress described by the ratio R presented in § 2.3. was given either by R = -1 or R = 0.1. The lifetime calculation was made for 2 initial crack lengths : 0,05 and 0,H mm. The very good welding conditions (automatic optimised welding), are likely to lead to small initial defects and the initial crack length adopted here is between 0,05
and 0,1 mm. In existing bridges, it is
generally assumed that this length may rise to between 0,3 and 0.5 mm. The agreement between the tests and the model predictions are fairly good, given the usual uncertainties in such experimentation. 3.5.2. Montlhery bridge. In this case the stress diagram is fitted in this case by the polynomial : o(z) = 7.533 - 2.706 z + 0.481 z 2 - 0.0788 z s + 0.0055 zk The
lifetimes
are
calculated
using
the
measured
stress histograms,
recorded in 1978 during the first phase of this study. Results are given in the tables 3-6 and 3.7. In order to compare the results with the classical Miner's factor AK
approach, the lower limits of the stress intensity
are chosen as functions of the fatigue limits (at 5 million
cycles) of the S-N curves : AK
= f (a , a ) , following the last formula
of §.2.3. For this bridge the fairly poor welds may lead to the adoption of well class iJ5 or 50 MPa and the lifetimes calculated from the fracture mechanics model will be longer than the calculated using Miner's rule. For good welds (classes between 63 and 71 MPa), both models give the same
-20
both models give the same results for this stress distribution. The sensitivity of the fatigue life to the weld class is much higher in the Miner's model than in the fracture mechanics one ; the latter seems to be more realistic. In any case, the short lifetimes found here show that this bridge was not designed for such a heavy and dense traffic. 3.5.3. Bridge of Caronte. The stress diagram here is fitted by the polynomial : a(z) - 6.0211 - 3.287 z + 1.075 z 2 - 0.190 z3 * 0.0123 z** Table 3-7 shows the computed lifetimes for various R and AK . The stress variations are those measured in the second phase of this research. Due to the low stress variations, no damage is expected for classes above 71 MPa and
the lifetimes are
always
very long. With the Miner's rule, the
lifetimes become short for the lowest classes... 3.5.1. Bridge of Choisy-le-Roi. The stress diagram is represented by the same polynomial as for Montlhery, because the structure is identical and the loading similar. In this case two traffic flows have been used for computing the stress time history by the LCPC's program CASTOR. One is the existing traffic on the bridge, before an increase forecast after the opening of a new motorway section, and the other was assessed in Angers (RN 23), representing the future traffic on the bridge. The influence surface of the transverse stress in the deck plate along the stiffener-plate connection is given in figure 3.12. Tables 3.8 and 3.9 give the fatigue lives computed for both types of traffic, various AK seen.
.The influence of the initial crack length can be
S
The choice of AK = 2 and a - 0.3 mm. seems to be realistic. Under s 0
the existing traffic the fatigue life is approximately 200 years, while under the predicted traffic it falls to 1)2 years (assuming well class 63 MPa). The results obtained by Miner are very close for this class, but again the calculations are very sensitive to the fatigue limit. Figure 3.13 shows the crack propagation versus time, for various AK and s two traffic flows. It is clearly seen
for the
that the crack length (and the
damage) does not grow as a linear function of time or number of cycles, as predicted by the Miner's law.
21 -
3.6. Conclusion. The specimens used in the experiments were welded with automatic submerged arc welding in an industrial situation. The first step in this work consisted of optimising the welding parameters. It has been shown that full
penetration,
(lack
of penetration
less than
1 mm) can nearly be
achieved without edge preparation. In these conditions, there was no signifiant difference in fatigue behaviour for specimens with a 2 mm gap between the top of the stiffener web and the deckplate and those with no gap.
For
alternate
bending
(which
best
represents
the
loading
in
bridges) with a lack of penetration not greater than 1 mm., the cracks initiated at the weld toe in the deck plate. Work at the University of Liège with similar specimnes but with a lack of penetration of 1,5 to 2,5 mm.
lead
to crack
initiation at the root of the weld
if the tensile
stresses were higher than the compressive stresses (a m a x > /a
/) at this
point. Therefore if the lack of penetration is not greater than 1 mm., it is possible to exclude cracks in the weld, and to assess the risk of crack initiation at the weld toe. The
test
results
allow
the
required
plate
thickness
to
be
determined, depending of the expected lorry traffic flow, and the thickness of the surfacing (see table 3.1)). It is now possible to choose between the thickness of the surfacing and the thickness of the deck plate in order to increase the fatigue life.
22
;
Stresses 0.
i
Number
Û«/isaa ) 70 60 50 40 30 20 10
-
SO 70 60 50 40 30 20
0 - 10 0 -10 / -0 -20 / -10 -30 / -20 -40 1 -30 -50 / -¿0 -60 / -50 -70 / -60 -50 / -70 -90 / -SO -100 / -50 -110 / -100 -120 / -110 -130 / -120 -140 / -130 TOTAL
13 19 61 222 635 1175" 2751 5176 | 12627 6629 ¿348 2343 1668 1C68 570 365 1S9 83 20
9 7 2
Stressranges ha.
(N/sra2)
10 20 30 40 50 60 70 SO 90 100 110 120 130 140 150 160 170 ISO 190 200 210
- 20 - 30 - 40 •- 50 - 60 - 70 - SO - 90 - 100 - 110 - 120 - 130 - 140 - 150 - 160 - 170 - ISO - 190 - 200 - 210 - 220-
40000
-
Number
6767 4163 24S9 1650 1090
872 601 441 314 253 142 134 81 43 17 21 11 7 5 2 2
19150
TABLE 3.1. : Stress and stress-range histograms at point D' (variable amplitude tests)
- 23
TABLE GEOMETRIC
3 .2 CHARACTERISTICS (nun)
i
CHARACTERISTICS
W.I.
CRIF
U.LG.
OF TEST
td
11
12
12
- stiffener thickness
ts
12
6,35
6
6
- stiffener width
B
6
305
300
3Ò0
322
b
150
109
109
- stiffener height
h
212
230
250
- location of the load
L
250
226
85
- location of the support
S,
85
335
335
335
337
337
335
Manuel
Autom.
< 0,5
< 0,5
S
- type of welding
2 Manuel
- gap - lack of penetration - fillet weld
5 to 6
3,0 to 4,5
1,5 to 2,5
4,5 to 5
3,5 to 4,5
4,7 to 6
SPECIMENS
IRSID
- deck plate thickness
1
ro
GEOMETRIC
Autom. 0 or 2 1 5,5 to 6,5
Lab.
Ao. d
Aos
deck plate
(HPa)
(Ml»a)
1 1 1 1 1 1 1 1
115 142 107 167 200 200 250 250 283
II
—1
150 170 200 200 240 240 300 300 340
II
0,1 0,1
100 240
150 180
™x
200 240 240 300 300 300 340 340
167 200 200 250 250 250 283 203
0 0
215 199
198 105 227 255
IRSID H H H H ii ■i
M
IO Ol
"s
II
II
_ i
II
1 1 1 1 1 1
II II II II II II
U.Lg. U.Lg. U.Lg. U.Lg.
amp.var. var.amp.
TABLE 3.3.a.
»1
N2
N3
»A
Cycles
464000
>5700000 >5670000 5031000 1676000 1069000 1211000 591600 560000 527000
750000 22100
950400 66000
1065600 232000
850000 208000 403000
4007000 1200000 1900000
4293000 1562000 2170000 440000 796000 447900 517000 470750
3000000 947000
4570000 1432000
135000
.1047000
245000
248000 345000 135000 120000
470000
522000
712000 434700 410000 405000
1678000 1315000 753000 791000
Failure in the deck. (Ao . at point A ) .
Lab.
R
S (weld root)
Ao
Ao
d (MPa)
s (MPa)
N,
N
2 Cycles
N,
IRSID
0,1
240
180
1.230.400
U.Lg.
- 0,63
296
240
- 0,71
290.000
302
245
276.000
- 0,96
333
280
282.000
- 1,0
150
400.000
152
2.100.000
139
7.600.000
- 0,57
144
1.850.000
- 0,61
128
>14.750.000
102
5.600.000
144
>18.000.000
- 0,58
142
- 0,34
282
1.610.000
225
0
157
510.000
136
0
224
8.900.000
182
0
206
610.000
177
1.315.000
0)
U.Lg.
Var.amp.
228
U.Lg.
Var.amp.
• 465.000
252
465.000
TABLE
3.3.b.
Failure at the weld root(Åo
at
point D),
TABLE
3.4.
Fatigue life Connection Stiffener -
Number of lorries (10 6 )
Thickness (mm) Plate
Stiffener
12
6
12
Deck plate
surfacing
à
4a d
0
2a
20
6
60
71
75
13
6
0
44
45
13
6
60
199"
219
14
6
0
89
103
14
6
60
494
647
14
7
0
62
73
14
7
60
297
393
°s ,
27
Initial
R
0*
"'M
(MPa)
0
D
-1
(mm)
-1
-1
0.1
-1
-1
IRSID tests
F.M model Propagation
Total
150
0.05 0.10 0.15 0.20 0.40
7286000 3650000 2388000 1766000 908200
>5700000
17Ò
0.05 0.10 0.15 0.20 0.40
5031000 2555000 1672000 1236000 635500
>5700000
200
0.05 0.10 0.15 0.20 0.40
316600Ó 1608000 1052000 777900 399900
240
0.05 0.10 0.15 0.20 0.40
r
-1
dei set a
NUMBER OF CYCLES AT FAILURE
Propagation
5831000 1676000 4293000
2031000 729000 3443000
1884000 957000 626000 463000 238000
1069000 1211000 1562000 2170000
1076000 1274000 1767000
240
0.05 0.10 0.15 0.20 0.40
262000 133000 87000 65000 33000
232000
209000
300
0.05 0.10 0.15 0.20 0.40
998000 507000 332000 245000 125900
591600 560000 796000
315000 548000
340
0.05 0.10 0.15 0.20 0.40
698000 355000 232000 172000 88100
470750
350000
Tableau 3. 5 : Comparison of the tested and computed lifetimes
28 -
MONTLHERY CLASS CECM 36 40 45 50 56 63 71 80 90
1
AK s (MPav^T)
1.127
1.257 1.430 1.604 1.777 1.994 2.254 2.557 2.861
CARONTE
Miner
M.R.
6 8 14 21 33 55 103 199 406
30 31 34 38 44 55 78 159 to
1 1
AK
^ (MPavnT) 1.117 1.246 1.417 1.589 1.761 1.976 2.233 2.534 2.835
Miner 37 57 99 172 315 600 1393 2835 8058
Table 3.6 : Computed lifetimes by two models for various S-N curves and Ak (bridges of Montlhéry and Caronte)
a
MONTHLERY 0
AK
(mm)
, ,
c eu 1 1
(MPa/ -"S
CARONTE
)
AK
1.430
1 .604
0. 1
>>
>>
>>
0. 2
88
118
>>
0. 3 0. 4
34 22
38 23
0. 5
17
0. 6
2.254
seuil
1.589
1 >>
(MPavnïï) 2.233 >>
>>
>>
78
371
>>
33
166
436
17
21
109
187
14
14
16
83
125
0. 7
12
12
14
69
96
0. 8
11
11
12
59
79
0. 9
10
10
11
53
68
KO
9
9
9
48
60
Table 3.7 : Lifetimes computed for various a and AK o s
- 29 -
M.R. 1.88 218 269 371 fa> w u b> W
Initial crac k a (mm)
Lifetimes Traffic
0
Table 3.8
Choi s y T r a f f i c Angers
362 165 104 76 61 51 45 40 36 33
0. 1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.
(years)
82 38 24 16 13 11 9 8 8 7
Lifetimes computed (AK = 0) for various a t
CLASS
T r a f f i c CHOISY
AK i
(Choisy).
o
Traffic ANGERS
CECM
MPav'm
Miner
M.R.
Miner
M.R.
36 40 45 50 56 63 71 80 90
1. 127 1 .257 1.430 1 .604 1 .777 1 .994 2.254 2.557 2.861
26 37 60 92 138 224 401 768 1469
107 112 121 132 147 176 234 361 754
6 8 13 21 31 51 94 187 368
24 26 28 30 34 42 59 115 333
Table 3.9 : Comparison of both models (Miner and F.M., Choisy)
30
Chanfrein Angle f i l Tension ce (V) I n t e n s i t é (A) Avance (cm/mn) Energie (kJ/cm)
Figure 3.1 : Macrography of weldments f o r welding conditions
deck p l a t e
optimized
wheel action
'stiffener
wheel action
JBBHB Figure 3.2 : Behaviour of c r t h o î r o p i e s t e e l deck
- 31
: : : : : :
sans 60' 29 550 50 19,1
- .Y
ly
rrf
c
a 01
x axe
Li
¥
'axe 1
r-
point where stresses are calculated ^-x
L-kl OX VIEW
g f
\J
gqq \ f
thickness :12mm \J
V J L th'ckness =6mn,
7 955 109 95.5 2100
Figure 3.3a : Structure used for the calculation
|
(N/mm)
NJSls
A uoe t
real structure
//I calculation s t r u c t u r e
Figure 3.4 : 0 influence line 0 for a 10kN wheel circulating on axis 2 Figure 3.3b : Detail
32 -
Figure 3.5 :0 A Influence lines fop a wheel laxlsl ; y=0) transverse repartition I
co co
Figure 3.6 :o0-lnfluence lines fop a wheel laxlsl ; y=0) transverse repartition
►X
i
ni
Ini
Aoms98N/mm 0,3-
-—EL =0,115
Ini
0,2-
dornage distribution (m = 3)
Figure 3.7 : Stress Range Histograms at Point D' (variable amplitude test)
34 -
Figure 3.8 : Test specimen and loading
poste 5 Stress evolution in the trough : point D
Stress evolution in the deck plate : point A
Od 1 ("Pa) poste 4
C U ("Pa)
Fs lOkN
400_
paste 3 F.IOkN
_90.
_70
i i i i i
0 1 Í » UÖ
i
15
»
-1—
d (mm;
£5
50
Figure 3.9 : Strain gauging of the specimen and nominal stress extrapolation
35
►
d (mm)
Aod i 400
"1
I I I I NU
1
I l I I 1111
I
I I I I 1111
1—I
I I I Ili
2
i i i 8im n 8 IO
320 — CU
280
CL
240 200
LU CD co
ΠCC CO CO LU
az
CO
160
mo 120 100
Ê 0 80 L ® 70 : + : © 60
ULg R=0 Ao Cte ULg Ao Var. IRSID R=-1 IRSID R=0.1
50 40 <
10
y
1—1 I II 111.1. 5 2 3 456 8
io5
2
3 456 8
R
io6
2
3 4 56 8
7
io7
3456
CYCLES N
Figure 3.10 : Crack initiation at the weld toe : A o d
Aos 400 320 -— CU
280
CL
240 200h
co
LU LD 2: CE
CO
CO CO LU
160 140 120 100 80
oc
70
cn
60 50 40
10
CYCLES N Figure 3.11 : Crack initiation at the weld root : Ao<
s> S
Ve^
Figure 3.12 : Influence surface of the transversale stress along the s t i f f e n e r - p l a t e connection
38
ş
traffic : Choisy
crack len gth [m] 0.005
"
r i ■ ■
;
0.004
—
K -
0.
— — -- --
K -
1.127
K -
1.604
-
2.254
,'
/ /
-
-
—
-
• ;
'
/
;
/
/
;
!
/ /
, ;
i
'
■
'
'
/ 1
0.001 '
i
/
/ /
■
-'
/ '
0.002 "
'
'
/
/
'
l ,
/
'
0.003
'
/ ,"
' /
-
. ■"
, , . , ' .
0
, ,300 time [years]
200
100
Crack propagation depending on the A K S
crack len gth [m] 0 . 0 0 5 t—
AKS=
1.604
Choisy Angers J
0.004
/
0.003 •
0.002
0.001
0
50
100
'
mti e [years] 150
Crack propagation for the two t r a f f i c s
Figure 3.13 : Evolution of the crack length with the tim e lor the cycles num ber) - 39 -
1». CONNECTION STIFFENER-STIFFENER. H.1. Types of connections. In most
large orthotropic steel bridge decks field splices are
necessary, because transportation of the complete bridge from the shop to the site is seldom possible. In general, both longitudinal and transverse field splice have to be made. In the longitudinal splices only the deck plate has to be connected. This is mostly done by butt welding and as the weld is accessible both from above and from below, a good quality weld can be achieved. The same applies to the transverse butt splice weld in the deck plate. However, at the transverse splice, the longitudinal ribs have to be connected as well (figure il.1). With closed ribs, which are usually used in modern bridges, the most appropriate way of splicing is by welding, but as
the welds
can
only be made from the outside
in an unfavourable
overhead position, the quality of those welds will be dubious. Depending on the location of the splice in the deck, the load on the splice can have a fluctuating part due to the traffic load, dominating the static loading, so conditions frequently recently
for
fatigue
in a bridge been
found.
damage
are
present.
Stiffener
splices
occur
deck and fatigue cracks in this connection have An
investigation
into
the
fatigue
behaviour
of
stiffener splices is therefore required. The Dutch and the Italian partners investigated the fatigue behaviour of the field splices in this ECSC research. The Italians studied a triangular shape of the trough ; and
trapezoidal shaped troughs were tested by the
Dutch. A triangular shape has been studied by Cunninghame in 1982 [22]. In
1988 a Japanese
IIW
document was published
which
contained
fatigue
results on trapezoidal ribs [23]. Test specimens and fatigue results of the ECSC research are compared and presented by plotting S-N curves. In these graphs the Eurocode fatigue design curves are used as a reference. M.2. Stress determination. 4.2.1. Measurements on a Deck Panel in Pisa. The
specimen
tested
was
a full size portion of an orthotropic
bridge deck (see fig. 4.2). The steel plate, 2000 mm wide, was stiffened
- 40 -
longitudinally
by
three
torsionally
stiff ribs with
triangular cross
section, spaced at 600 mm centres, and by two flat stiff eners (200 x 12 mm), to reproduce the transverse continuity of the deck. The stiffeners were made of 6 mm. plate which had been cold formed. These passed through the cross beams to which they were welded with fillet welds. The cross beams were a double »T' section laid at a centre distance of 3500 mm. The
12 mm thick web of the crossbeams had 25 mm radius circular
cut-outs at the apex of the troughs. A 100 kN proof load was applied, using a hydraulic jack, through a 200 x 300 x 50 mm thick reinforced neoprene plate : the load was controlled by a force ring gauge. Forty five loading measured
positions were
tested
on the deck
; strains were
at each location. Strains were measured using electrical strain gauges located on
the deck plate, at the apex and on the webs of the stiffeners. The deflections were measured at 88 points located under the stiffeners and the cross-beam. Diagrams and tables of the deflection and stress measurements are presented in [J|]. 1.2.2. Calculation of the stresses. In order to study the static behaviour of the deck undergoing examination, an extension of the classic HUber's method was adopted [18] where the variable section of the crossbeams is taken into consideration, as well as their shear strain. Using the simplifications introduced by Pelikan and Esslinger [19], one obtains [20] the definition of a simple analytical model to calculate the influence surfaces of the continuous orthotropic deck on flexible cross beams. The differential equation that governs the problem was resolved using the Levy's method [21]. The sections of the theoretical deflection influence surfaces of the central ribs an the stress influence surfaces at the apex of the central rib are presented in
- 41
The sample tested was also analysed using F.E.M. The code used was MARC implemented on an IBM 3090 computer, whilst the finite element adopted was a thin shell element with eight nodes with zero degrees of freedom (element n°72 of the MARC library). The load (100 kN) was applied on a 20 x 30 cm. rectangular surface with its centre coinciding with that of the deck. A comparison
between
the results
obtained
in this way and the
analytical and experimental ones, reveals a close correspondence. Thus, for theoretical investigations, the analytical method was chosen, because of its accuracy and its simplicity. Some theoretical and experimental influence lines are compared in figure 4.3. (deflections) and in figure 4.4. (stresses at the apex of the central
rib). In the diagrams, .the theoretical curves are shown as a
continuous line, while the experimental results are shown as a dotted line. A study of the results reveals a close agreement between analytical and measured values. 4.2.3. Stress spectra. The stress spectra was obtained by means of the University of Liège simulation program. The level crossing and the Rain-flow histograms were calculated for the Rheden traffic on the influence line of the detail tested
at T.U. Delft
(Fig. 4.5.
: bending moment at mid-span of a
continuous beam). 4.3. Test results of the University of Pisa. 4.3.1. Fatigue tests on type ' B' specimens. 4.3.1.1. Test specimens. The Fe 51 OC specimens, 2000 mm long, are made of a triangular rib obtained by the cold forming of a 6 mm thick steel plate welded to a top plate 600 mm wide and 12 mm thick (figure 4.6). Along the centre line of each specimen, a type I or a type II joint was shop fabricated using the same procedures as those used on site. In Figure 4.6 shows details at the tested joints the procedure. In type I joints, the ribs stop about 100 mm short of the end of the deck plate.
42
In type I joints, the ribs stop about 100 mm short of the end of the deck plate. The two ends of the top plate - one of which has the backing strip welded to it (S1 welding) - are positioned with a gap of about 6 mm, and then automatically welded. The
missing
rib element
is then
inserted and manually welded
in the
overhead position (backing strip weld S3). In type II joints, the top plate is about 100 mm short of the end of the stiffener. The
stiffener webs are butt welded, with complete penetration manually
with coated electrodes. First, the internal part of the ribs is welded in ascending
vertical position. The root of the
internal weld is ground
before the external weld is placed in the overhead position. The joint is completed
with
execution
of
the the
insertion S1
flat
of
the missing
position
welding
top and
plate the
portion,
manual
the
overhead
remaining welding S2 between the top plate and the ribs. All the welds were checked by means of visual and magnetic controls. The butt weld, were also
100 % X-rayed and repaired where found (once only)
to be unacceptable according to the UNI 7278 Italian norms. The fatigue tests at constant stress amplitude, were carried out on nine type I joint specimens and on eight type II joint specimens. The specimens were simply supported (at 2400 mm span) and the fatigue load applied at two points 200 mm apart of the centre line using a pulsating hydraulic jack with a frequency of 1 Hz. The load was applied though 50 x 60 mm. rectangular pats of neoprene, 10 mm. thick, between the mobile jack head and the top plate. The strains at the apex of the ribs were measured using electrical strain gauges, so placed as to determine the nominal stress amplitude without taking into consideration the presence of local peaks of tension. During the tests the minimum nominal stress was kept at a constant 1.5 kN/cm2 for all of the specimens. Each test was stopped when failure was reached,
as recognized by the
specimen's loss of stiffness (an increase of one centimetre in the maximura deflection under load), or when eight million cycles had been completed without breaking. 4.3.1.2. Experimental Results. In table 1.1. the type of joint, the nominal stress range at the lower part of the weld and the number of cycles to failure are shown for each test.
43
In specimen 1 the first crack appeared in the fillet weld of the backing strip, while a second crack appeared and propagates in the weld ; in specimen 2, cracks initiated in both welds but only one crack propagated. In specimens 3, 6, 7, 10, 11, 12, 13, 15, 16 and 17 cracks started in the weld, at the apex of the rib, and propagated in the weld. In specimens 4, 8, 14 cracks started at the apex of the rib in the parent metal and propagated in the plate. In specimen 5 the crack started in the weld at the apex of the rib and propagates both in the weld and in the plate. Specimen 9. no cracks occured in the rib, but three longitudinal cracks appeared in the top plate : two at the position of the rib to top plate connections and the third in the middle of the plate. In figure 4.7 the results obtained and the mean life curves concerning type I and type II joints are reported. 4.3.2. Fatigue tests on type 'A' specimens - (full size panel). 4.3.2.1. Test specimens. In order to check the application of the fatigue results obtained on type B specimens to the real deck joints, fatigue tests were carried out on two large specimens with type I joints which were obtained by cutting the specimen used in the static'test across at the centreline (see fig. 4.2). The panel was supported on two crossbeam with an overhanging canteliver section which was loaded by two concrete blocks of 54 kN total weight. The pulsating fatigue 'load, applied through a 20 x 30 cm. rectangular plate placed in the middle of the span, caused a nominal stress range at the lower apex of the central rib weld equal to 22.5 kN/cm2 and a minimum stress of 1.5 kN/cm2, reference being to the stress induced by the ballast and the deal weight of the panel. 4.3.2.2. Experimental results. The first specimen failed after 240000 cycles, the second one after 260000 cycles. In both cases the crack initiated in the weld at the apex of the rib and propagated in the weld toe. In figure 4.7, the results obtained on type B specimens with type I joints and those obtained on type A specimens are compared. The comparison reveals close agreement, within normal experimental limits, between the results obtained on the two types (A and B) of specimens. The slightly lower fatigue life noticed on type A specimens is probably due to the higher level of residual stresses present in these specimens. - 44 -
4.4. Test results of the T.U. DELFT. 4.4.1. Tests specimen. The specimens for the bending tests were single rib specimens as depicted in figure 4.8. Two types of field splices were selected ; Type A : Butt splice with backing strips with varying root gap (0, 2 and 4 mm). Type D : Butt splice with a V-groove and backing strips (root gap 4 mm). The ribs were a rolled trapézoïdal section (steel grade Fe 510) the dimensions of which conformed to F.K.H. - Trapezprofile nr. 2/325/6. As the usual spacing of the ribs is 600 mm, the width of the deckplate in the specimens was also 600 mm, in order to get the same position of the neutral axis as in an actual bridge deck. During the fabrication of the test specimens, the welding conditions specimens
on a real made
with
bridge
deck
were
special
care
under
imitated
because
favourable
test
conditions
would not be representive. The
ribs were welded
on a 8400 mm. wide plate, at the proper
spacing of 600 mm., in a downhand position. Each rib was in two parts with a gap in between to make the splice. Then this assembly was turned over and
the field splices were welded hampered by the adjacent ribs as it
would be on site. The welds in the bottom of the ribs were made in the overhead
position
and
the welds in the webs of the ribs were made by
upward welding. After completion of the splices, the assembly was cut into fourteen test specimens and the end plates were fixed with fillet welds. The test specimens were made by a fabricator with experience in making orthotropic steel bridge decks.
4.4.2. Testing and measuring equipment. As
mentioned,
a
four
point
bending
test was
chosen
to study
fatigue in the rib splices. Due to the fact that tests in the region of ten million cycles were planned, two test rigs were built. Part of the experiments, mostly the variable amplitude tests, were carried out using servo hydraulic test equipment operating in closed loop control with load feedback. The main part of the constant amplitude tests was executed with loading equipment from Losenhauser. To avoid any secondary effects, all supports in the test rigs were provided with hinges or roller bearings.
- 45
Each
test specimen was
instrumented
with a number of strain gauges,
varying from 9 - 52 locations, before testing. Strain measurements were carried out dynamically to check the applied stress range. Furthermore, strains were monitored 2H hours a day to obtain information about and location of crack initiation. Measurements of crack growth were carried out during a number of tests. This was done by periodic visual inspection with a magnifying glass. It was only possible to measure crack length at the surface of the specimens. As far as possible four stages in fatigue failure are expressed in number of cycles : NI : Moment of crack initiation given by 10 % reduction in strain measured by the gauge nearest to the crack. N2 : Moment of visuable crack becomes visible. N3 : A surface crack length of 50 mm. N4 : End of test with extensive through cross section cracking (leading to loss of specimen stiffness causing limitation of the actuator stroke) and/or
loss of symmetry
(causing unacceptable
side
load on the
actuator bearing). i|.i*.3- Constant amplitude tests. i».11.3.1. Tests results. The fatigue results of the constant amplitude tests are presented in figure H.9.
and table H.2. by plotting the a S-N relationship, on a
log-log scale. The normal stress in the bottom of the rib in the middle of the splice, due to pure bending, was chosen as the main stress parameter in this figure. In all the specimens, the crack began in a weld between the rib and the splice plate. In most cases, the starting point was located just above the bend at the bottom of the rib. At this location the stresses are about 20 % lower than in the middle of the bottom weld,
where cracks
might be expected to start because of the higher bending stresses. In one case the crack started in the side of the web at a location where the stresses are about 70 % lower than in the middle of the bottom weld. With some specimens it was possible to continue the testing for some time to study the behaviour of the crack. The cracks propagated more in the side weld than in the bottom weld, eventually the crack in the side weld propagated into the base metal of the rib. - 46 -
To study the unexpected crack initiation points, some of the cracked specimens were cut open. Inspection of the inside revealed local defects, such as bad penetration and slag inclusion. The crack location can be further explained by the welding procedure. The welds started are finished just above the bend near the bottom of the rib and a tack weld was also placed at this location. Depending on the weld geometry and root gap of the weld, the first visible crack N2 was noticed at 51 - 98 % of the number of cycles to failure of the welded connection (N-U). The moment the invisible crack was detected by measuring strains (N1) varied between 5 and 75 % of the number of cycles failure (NU). For two specimens no cracks were detected by gauge but one of the welds of these tests specimens failed very suddenly over 50 - 75 % of the total length of the weld. This phenomenon can be explained by the poor quality of these particular welds. 1.1.3.2. Effect of the root gap. Comparing the results of the fatigue tests from figure 1.9. it appears that a weld with a zero root gap results in a fatigue life far below the weld with a root gap of 1 mm. (a factor 18). A welded detail with a root gap of 2 mm. gave variables results. At a level of 163 MPa it gave a fatigue life comparable with the fatigue life of a detail with a 1 mm. root gap but at a lower stress range level, about 110 MPa, the fatigue life was a factor of 12 lower (Fig. 1.10). 1.1.3.3. Effect of the weld geometry. Changing the weld geometry by using a V-groove did not give the expected improvment. However, it is clear that it is easier for the welder to make a better weld using a V-groove as a butt weld. The disadvantages are that both sides of the detail have to be prepare and it is necessary to use more welding material. 1.1.3.1. Fatigue limit. From the results it appears that for the specimens with a root gap of 1 mm. as well as those with a V-groove, no fatigue cracks were discovered at a level of about 90 MPa after testing for ten million cycles. It can therefore be concluded that for this type of detailing we are almost
47 -
in the neigbourhood of the constant amplitude fatigue limit. For specimens with zero root gaps however, several cracks were discovered in the bottom side of the rib at a very early stage at a stress range level of 83 MPa. Due to the bad results for this detail it is expected that there will be no fatigue limit. 4.4.3.5. Comparison with previous Dutch tests. In figure 4.10. the test results of the specimens with a root gap of 4 mm. are compared with previous Dutch tests [24]. It appears that the fatigue failures of the specimens tested at the high stress range levels fall between the scatter band of the previous tests. It can therefore be concluded that these fatigue results can be used for a weld classification for this detail. 4.-4.3.6. Weld classification : S-N curves. Ignoring the fact that cracks did not initiate at the bottom side of the rib, it can be concluded that for the rib splice with a root gap of 4 mm., a weld class 80 (according to the Eurocode 3) can be recommended. If the crack initiation point is taken into account the classification reduces to a weld class 63. This values are to compare with the values given in last draft of the Eurocode. 4.4.3.7. Calculated fatigue life. Using the simulated stress spectrum for the field welded splices (fig.4.5) and assuming that the detail can be classified as class 80 according
to
Eurocode 3, a fatigue life
of
about 75 years can be
calculated. Further optimisation of the detail seems to be unnecessary. However in a lot of existing bridges welds were made with small root gaps. Here the fatigue lives will be very short ; first failures have already been discovered. To assist with the maintenance of these bridges it is necessary to know how to repair those welds in an economic way and how to calculate the remaining fatigue life after repair. 4.4.4. Variable amplitude tests. 4.4.4.1. Load spectra. In the working group meetings it was decided to use load spectra from the earlier ECSC Phase 1 and 2 research for the variable amplitude tests of the third Phase. In the computer assessment programme of the University of Liege, it was decided to use the traffic flow of the Rheden
48 -
Bridge. Results
of
the
simulation
for
the
field
splice
in a rib are
given in section 4.2.3. Analysis of the spectrum (figure 4.5) showed that the stress range < 18 MPa caused only 7 ? of the total damage of the sepctrum. However, these stress range classes amount, to 84 % of the total number of cycles. So leaving out these classes saves a lot of testing time. The remaining stress ranges (18 MPa - 82 MPa) had to be raised a level above the assummed constant amplitude fatigue limit. For testing techniques it was
necessary
to
leave
out
the
highest
stress
range
classes,
which
accounted for only 5 J of the total damage. As well as, the simulated load spectra, for one of the specimens a measured spectrum was used. A review of all sepctra used is given in table 4.3. To
compare
amplitude
the
ones,
variable the
amplitude
applied
tests
stress
spectra
results were
with
the
constant
analysed'according
to
Miner's rule in the two different ways explained in section 3.2.3« 4.4.4.2. Tests results. The fatigue results of the variable amplitude tests are presented in figure 4.11. by plotting the S-N relationship, on a log-log scale. The location where the cracks initiated and the crack development were the same as those found in the constant amplitude tests. The first visible crack N2 was noticed in all cases at about 80 % of the number of cycles
to
failure
of
the
welded
connection
(N4). The
moment
the
(invisible) crack was detected by measuring strains (N1), varied between 69 and 76 % of N4. 4.4.4.3. Comparison with C.A. - tests. - Test specimens with a root gap of 4 mm.•under simulated spectra. There is a good agreement between the constant amplitude tests and the variable amplitude tests using the Miner's calculation. Furthermore no cracks were found in a specimen A.1.8. loaded to twice the simulated spectrum
after
48
million
cycles. The maximum
stress
range
of the
spectrum of this specimen was 132 MPa. - Test specimen with a root gap of 4 mm under measured spectra. Specimen
A.1.6. was
tested
with
the measured
spectrum
of
the Forth
Bridge. Comparing the fatigue strength with the constant amplitude test results, the spectrum showed to some degree a better fatigue life.
- 49
* Test specimen with V-groove under simulated spectra. The fatigue endurance of this specimen D1.1.-V was about 8 times that calculated
from
the constant
amplitude
test
on this
type
of weld
geometry using Miner. Comparing the fatigue behaviour of the relevant speoimens, the fatigue behaviour of the V-groove specimen is almost H times
that
of the specimen without the v-groove. More tests on this
detail are necessary to confirm this difference. Comparing the results it appears that applying the measured spectrum
(A.1,6)
resulted
a very
small
difference.
Using
the measured
Spectrum (specimen A.1.6) instead of the simulated spectrum gives not a significant difference in fatigue life. i\.i\.5.
Conclusions. Cracks
in
these
welds
always
started
on the inside
where
inspection is impossible. When a visible crack is located the total weld will fail very soon. Repairing the weld must therefore be done as soon as possible. - Constant amplitude test ; The
results
welded
of the tests showed that the root gap of these kind of
connections
must
have
a minimum
width of 4 mm. to achieve a
welded detail class 80 according to Eurocode 3. If a weld can be made without penetration defects and slag inclusions the classification may be higher. For the test specimens with a root gap of 4 mm., no cracks were
found
at a stress range
of 90 MPa, (compared with the 60 MPa
constant amplitude fatigue limit of class 8 0 ) . So a knee-point at two million cycles instead of five million cycles is possible. Changing the weld geometry did not give the expected improvement, however the number of tests (two) is too small to draw a definite conclusion. " Variable amplitude tests t For the traditionally welded connection with a root gap of H mm., there is good aggreement between the constant amplitude and variable amplitude tests, using Miner's
calculation.
In this part of the programme the
V-groove (one test) resulted in a much better result than the welded connection necessary
tested
under
to confirm
a constant
amplitude
load.
More
tests are
this difference. The difference in fatigue life
using a simulated spectrum seemed to be small.
50 -
H.5. Comparison with other research programs. In addition to the ECSC research described in this chapter, the fatigue behaviour of field splices in ribs of orthotropic steel decks has been studied by Cunninghame [22] and Yamada [23]. A triangular shape was studied by Cunninghame in 1982 and in 1988 a Japanese IIW document, published by Yamada, contained fatigue results for trapezoidal ribs. The results of the tests are given in figures 4.12 to 4.14.. From the UK-research it can be seen that, with the exception of one test, all results are situated above the class 125 of the Eurocode S-N curve. As with the Dutch results, the location of the fatigue cracks was unexpected. All the failures initiated at the root of the weld on the flat web of the stiff ener rather than at the more highly stressed apex.
Examination
of
several
fracture
surfaces
showed
no
single
initiation point. It is concluded that the crack initiation point is determined by residual stresses due to the welding procedure. Gathering the
"Pisa-tests"
(type
I)
and
the
"UK-tests" together
it can be
concluded that for triangular shaped ribs containing a weld with backing strip a Eurocode class 112 can be considered (figure 4.12). Confirming the Dutch research, the great influence of the size of root gap
on
the
fatigue
strength
was found
in the Japanese research.
Furthermore the fatigue strength was affected by the residual stresses in the direction
of the rib which
depends largely on the welding
sequence. Comparing the results with the Eurocode S-N curves, it can be concluded that field splices with a root gap of 3 mm. or more are above the class 71 curve. Gathering the "Delft-tests" and the "Japan-tests" together it can be concluded that for trapezoidal shaped ribs containing a weld with a backing strip and a root gap greater than 3 mm., a Eurocode class 71 can be considered (figure 4.13). However with small root gaps the classification can drop to 36 or less (figure 4.14). 4.6. Conclusions. A lot of
data are available concerning the field welded rib
joints. A first comparison showed ; - The butt weld connection gives lower fatigue behaviour as the connection with braking strips (fig. 4.7) ; but the fatigue behaviour of this last connection seems influenced by local defects, that are depending of difficulties of realization in a bridge working site.
- 51
The
better
behaviour
of
the
triangular
ribs
compared
with
the
trapezoidal ones, for a connection with bracking strip and a root gap greather than 3 mm. : ¿0
»112
c
N/mm 2 for trapezoidal ribs (that corresponds to
= 71
Äa
N/mm2 for triangluar ribs ;
c Eurocode 3 prescription).
- The reduction of the fatigue strength when the root gap is smaller < 36 N/mm 2 .
3 mm. ; La -
The
than
influence
of
residual
stresses
and
defects
due
to the welding
procedure. - Using Miner's calculation there seemed to be a good aggreement between the results of the constant and variable amplitude tests. More
detailed
analyses
of
the available
experimented
data are
needed to give a general conclusion for design and practical recommendations. From the above, it follows that for the orthotropic deck details whose dimensional characteristics are described, it would be desirable to define in the codes, S-N curves which take into account the type of joint, the presence of defects and here control and, above all, the construction procedures starting details
and
thus
the
level
point of welding). which
do not
of
The
satisfy
residual
S-N
curves
minimum
stress should
quality
(for example
: the
be lower for those
requirements. Naturally,
because of the complexity of the problem, the conclusions proposed here do not pretend to be in any way definitive, but rather a reference point for future investigation.
52 -
Speci-.en Jí.
o-rsnçe Cycles ac I KW/en = ] failure
J oin z
Type
1
I
19.50
->
I
22.50
13S0000 53S0OO i
3
II
22.50
251000
4
II
19.50
939700
I]C
17.50
321000
5
IT
17.50
459500
7
I
22. 50
463000
S
II
14.00
6040000
3
I
17.50
8100000
10
TT
19.50
335000
11
I
19.50
1200000 "
Ţ
22.50
13
II
22.50
305000
14
1 1
17.50
1930000 1657000
5
»
!
1
1
15
I
19.50
16
II
12.50
17
I
19.50
Test results of Pisa TABLE 4 . 1 .
- 53
1
1 1
S560CO
7073000
1622000
SPECIMEN VELD
n
l
V
n
n
3 -
4
At>r[KPa]
CA?[ B3]
'M
> 0.833
150
•
0.833
9.085
9.085
9.298
-
> 9.298
• -
>29.582
A.1.3.
1. 2.
• -
Ï50 105 105 90 90
A.1.4.
1.
0.180
0.180
0.245
A.1.4.
2-
0.239
0.244
> 0.245
233 233
A.2.1.
J#
.
.
> 0.668
163
A.2.1.
2. 1.
-
0.668
A.2.2.
-
> 0.779
163 110
2 2 2
A.2.2.
i »
0.713
-
0.779
110
2
A.3.1.
1.
0.035
0.039
0.445
A.3.1.
1.
0.035
0.029
-
•
153 153 153 S3 83 83 83
0 0 0 0 0 0 0
1. 2.
A.1.1. A.1.1.
1. 2.
A.1.2. A.1.2. A.1.3.
A.3.1. 0.157
A.3.2.
• -
0.175
>29.582
0.445 > 0.465 0.327
0.180 .
A.3.2.
0.180
-
0.327
A.3.2.
0.203
0.253
0.327
A.3.2.
0.223
0.240
> 0.327
-
-
0.728
-
0.798
-
-
> 0.798
D.l.l.
1.
D.l.l.
2.
D.I.2.
1.
D.I.2.
2
to n
-
'V i'lì î
4:
u 4ƒ 4J
H'
>10.990 . >10.990
*
- nominal stress ranga bottoaside trough. - number of cycles x 10 .
n. - initiation by strain gauge. n, - first visual crack. n, - length of crack equal to 50 va. n, - failure of the veld. 4 Table 4.2. Results constant amplitude tests
54 -
lêo
1 J
STVÜLATE3 S7ECTP.A; TABLE 4.3. VtRTABLE AMPLITUDE TESTS ! T-U. £1 ITT NUMSES
^i(iooo)
KVM3ER
lM?ft)
CYCLES
TEST SPECIMEN A.1.5.
SPEC??.*,: in Forth ijr
i(1000)i" ? B j
0?
O? CYCLES
y-AS'L-10
A.1.7,
A.1.8.
TEST SPECIMEN A.1.6.
D.I.l.v. E130
80
£0
40
40E6
£0
saia
'EB
66
44
3456
£8
5570
56
72
48
3366
56
5410
104
78
52
2502
104
5120
112
E4
55
2034
112
«620
120
50
60
1618
120
3500
128
56
64
1818
128
3700
136
102
£3
1215
135
3330
144
108
72
1341
144
2 POO
152
114
76
1044
152
2530
160
120
SO
£23
160
1970
168
126
£4
909
168
1670
176
132
88
810
175
1550
1E4
138
52
774
1E4
500
192
144
96
855
192
1050
200
150
100
792
500
208
156
104
6A0
216
162
108
440
224
168
112
330
232
174
116
480
240
180
120
310
248
186
124
110
256
192 .
128
120
264
198
132
[M?a] -
135
101
68
145
¿S [M?al a1 ' n - 45610 a
153
115
77
1S2
I n t - 613SO to
Vl03)
621 455 . 558 414 342 171 1E0
208 216 224 232 260 243 256 264
31095
15840 Dll.V
A.15
4760
1000 1200
4750
1.420 1.200
5150
1230 1350
5400
1.470
n 2 (103) n (103)
1270 6270
1.620
6250
5190 n 4 (103)
666
200
1460
6500
48.100
55 -
1.770
PAVEMENT DECK PLATE
LONGITUDINAL RIB LATERAL RIB RELD WELDED JOINT
MAIN GRIDER or STRINGER
DECK PLATE BACKING STRIP FIELD BUTT WELD JOINT LONGITUDINAL 'ROUGH RIB CONNECTING RIB
F 5Ure
'
4 1 :
-
Tthotropfc , t „ i
56
bridg
e deck
B
? -y-
*yf ■i-
'T" I
i
i -! A
::;
A
I' 'T"
?
B 3SO0/2
3S00
m
UJ section A-A
i
I ' ! i l
i
section B-B Figure 4.2 : Type A specimen (PISA) 57
X(cm]
x=525cm
WUmm)
0 1
r—
1>
y=90cm 4.3a
1050
1
X . '
2 3 x=52 5cm
W ' (mm)
y=60ciT
4.3b
1
0
!
iI
1
•
i
1 2 3 W
—•<>s
^"^^J
(mm)
Y = Çft ì
1050 - — w ». X(cm) X
1
v=pn rm
ìrm
4.3c
0 1 2
■
\ ■■
1050
1
■
<
3
i\
/_
i
x=58 3,3cm
y=60 cm
4.3d
Figure 4.3 : Comparaison between t h e o r e t i c a l and experimental r e s u l t s (deflections)
58
^X
0
'""
\
s
1050 — * X
10
O'lkN/cm2)
x=525cm y=90cm ¿.¿a
0
r
5
ii
1050
1
^s
S
^X(
10
(kN/c mj
0
>:=525c m y=(30cm
UMb
:K:
_0 5 10
OÎ(kN/cm2)
x = 700cm y=90c m LAC
Figure 4.4 : Comparaison between theoretical and experimental results (stresses)
59
1050^ X{cmJ
n¡[%] f i
i
.
63.5
.1. 1
70 60
0.2
Rain flow histogram Rheden traffic
• ■
50
16.2 Nummer of Cycles 1%)
■
CO , 30 20-
l_ S ' niula
10'
ed
used in tests
, 1
5
N//7V7J
i
n5
2
D 3 [%]
/ 56 H
82 1
D3 -DANA CEI'/.} 88.0
7.0
5.0
5. 5.
JÌ
4 3-2.. 1-.
o2
V
r-
sA
Lr S],
ln_
IB
6t;
In.- 6138
I
62
^
h
i
¿a- - 33.8 N/nm
:
-j
ASm » 38.3 N/nm , a
m
- 4450
En.- 37.740 ¿O- - 18.7 N/ma2 i
e
2 ¿SB
- 40.5 N/mm , D
Figure 4.5 : Detail t e s t e d in Dalft - 60 -
H - 4091
NU2
l
Js^M22_Äi)
\S
-^o»- l!
-j*l4
j . í j
! monjol
"Ţwe/ch'nţ
fc^
h ■*■■<■
AL
Type 1 SECTION •• / -A
I 100 I fl» !
^-•r
SECTION •■ A-A
Type 2
MS'
V&tjp&.-X /£%>S
■r-M™
Figure 4.6
- 61 -
co
1■ 1
=
i
1
i
i
JlliillUil.llilJiii 1 1 J 1i 1i 1t I 1 1 1 1 1 : 1 1 |
[_
^~^
■
— —
O
^
■ CD ■ LO
T
i
a
i
M M
^■ ro
i i
Q) O Q) O i Oi O i
■
H E" H
CO LU r^ CD I r o* " 1 CJ
^
_ ~ _, —. _
■ 3*
i a < 1 M
>i
>,
1 H
>,
• «+ • '
■ i
ru
~
—
i
i i
— —* —^
/
/5>
,
Z
to
OJ
ai Q.
CO
O
>■.
•* •o c
ro*—' —
/
CO
•
4
^mm
C 3 1_>
/
u A / 7 /
OJ >
ro
A
—
w
o
Z?
j
^^
>
to LO
i
zz.
ni
LO
—
— ZT
01 Q. 4
— ro
C OJ QJ
OJ
LO
O
—
—
C0'F~^
•
—
fc_ — CO — LO — 3* m m ~ ro
™ *
""
1
J2
ra Q. E o C»;
OJ QJ C 3
O
o
rj*
1
i
i
i
\
O OJ CO
D O OJ
o 3"
o o
o
OJ
OJ
LO
\
ImiinitliiiniiiiliMilii i i 1 i i i i 1 i i r r 1 O OJ
O O
•—4
O CD
a r—
o
CO
(edW) JS 30Nby SS3ÜÍS
62
o
LO
o =r
~o
II
3525 1000
v
V
r^\ 1500
-4
1500
1-
3300
6
r i W E^sl
TYPE A. A-1 root g ap = 4. A-2 ,. =2. A-3 .. =0.
^=3
\\\\\\\N
l„root gap
TYPE D.
■ƒ30x4
Sx ^
tl
K
-ƒ30x A a=3
N\WN\N
550
300
325
WELD AREA BACKING STRIP
Figure 4.8 : Selected types of field splices in the ECSC research
- 63
I !. J
fr ¡5 —
X
< c co
■ <
co O o ,
ce o
• A.
m
1 1 1
^
1
1— t/]
^ »-
' co
í»» IO
CO
K:
o
o o 1—
co
o o ce m UJ UJ t— CO
o a. o en t— O
zsz t— Cd
o o 00
r»
—i
(O
—
i—
IO
UI
a o o o ce
s
3 ■ »
3
to ni c_ 4
o CO CM
) C 5
>-
IO
co
O I J J
t—o: cc»— o o
cc UJ
co UJ
I
c
?*" I a> o
" m i n i m i n o O -«■
<
O (O
o
o
CM rO
O
co CM
D [
i i i i i i i i i i i o o
o co
CM
—
O
en ,
r»
Li_
IO
O
IO
X
o
O
o
o
to
■»■
i
i o CM
i
i
i
i
o o
o o>
o oo
d W ] (3QIS H0I108 TVNinON)
64
i
3 o N V cJ
»—COCO
I S S 3 el 1 S
-*•
ce
IO
UJ 0Q
ni 1
•• a\
«* 0J
c_ 3 CTI a.
F I E L D - W E L D E D RIB JOINTS OF O R T H O T R O P I C ( S - N Curves A I I
400
"I
1
according
STEEL 3RIDGE
DECKS
to EUROCODE 3 )
I I I I III
"I—I
I I I I II
T
I
I I I I III
1
1
I M M b
CONSTANT A M P L I T U D E TESTS E
360 320
F.H.K.-TRAPEZPROFIEL 2/325/6 Fe 510 j= TYPE A : Butt spile« with back-up strips Z A ROOT CAP 4 m m (ECSC HOLSTEIN 1989)1 4 m m (TUD TROUP 1974) -
280 o 240 200 180 160 o m O) UI
— run-ouI
—
140 120
100 O
-<
OC
90 80 70
CO
co
60
LU CU
t—
CO
50 40
10
.TU-DELFT
Stevin LAB0RAT0 Steel S T R U C T U R E I
i i i i i 11 3 4 5 679
10" NUMBER
OF C Y C L E S
I
10" (UNTIL TOTAL FAILURE)
10'
H -r-l-M-H 4 5 679
10
[N]
Figure 4.10 : Comparison with previous research T.U.D.
> STST-005.GRA
FIELD-WELDED RIB JOINTS OF O R T H O T R O P I C (S-N Curves
A I I
«O 360
I
320
I
I
according
I I I I III
to
I
I I I I III
240
Ş
200
M
180
Ü o
160
T
I I I I I III
1 I I I I II-
1
VARIABLE A M P L I T U D E TESTS E
CONSTANT AMPLITUD E TESTS
F.H.K.-TRAPEZPROFIEL 2/325/6 Fe 510 ^ TYPE A : Bult spilet with back-up «trips Z TUD ECSC-RESEARCH KOLSTEIN 1989 SIMULATED LOADSPECTRA RHEDEN TRAFFIC * EQUIVALENT STRESS USING SLOPE 3 A EQUIVALENT STRESS DEFINED BY BRULSl
280 Sb
DECKS
EUROCODE 3 )
T
REGRESSION LINES
STEEL. BRIDGE
— runout
ai
_ co
HO 120
UI
LU
O
100 90 80
CC
co co LU
70 * 60
cc I—
co
50
JU-DELFT
Stevin LABORATORY Steel S T R U C T U R E S
40
10
4
1 2
' i i iii n 3 4 5 67 9
s
J 2
10 NUMBER OF CYCLES
1 ' I I I I II 3 4 5 6 7 9
J—1 l l i i n 3 4 5 6 7 9
10
10
(UNTIL TOTAL FAILURE)
Figure 4.11 : Fieldwelded rib joints of orthotroplc steel bridge decks
J
I 3
[N]
I I l l ii 4 5 679
R
io8 — - >
STST-006.GRA
F I E L D - W E L D E D R I B J O I N T S OF O R T H O T R O P I C (S-N Curves
/\ I
400 p -
"I—I
according I I I I I I
360
DECKS
to EUROCODE 3 )
1
I—I I I I I I I
1
I
I I I I Ml
CONSTANT
320
\
1
i
AMPLITUDE
1 I I I II;
TESTS E
TRIANGULAR STIFFENER
— , 280 o ^ 240 r:
STEEL BRIDGE
TACK WELD
TYPE A : ROOT CAP 6 - 12 rrm. O TRRL-RESEARCH CUNINGHAUE 1982
—
-RESEARCH SANPAOLESI 1989— — run-out
200
180 a o
160
o m
140
i—
1
0) 1
_* -t z
120
3 O
z
100
LxJ O
90
-z.
80
a:
70
CO CO
60
Ol CO
50
LTU-DELFT
40
10
S t e v i n LABORATORY S t e e l STRUCTURES I I i i i ii n 3
4 5
6 7 9
10
J
I 3
I I I I I II 4 5 6 7 9
10
-I 3
I I I I I II
4 5
6 7 9
10'
J 3
I.I I I I I
4 5
6 7 9
10
NUMBER OF C Y C L E S (UNT.L TOTAL FAILURE) [N] > Figure 4.12 : FlelrJ-welded rib Joints of orthotroplc steel bridge decks- Triangular stiff ener
8
F I E L D - W E L D E D R I B J O I N T S OF O R T H O T R O P I C (S-N Curves a c c o r d i n g
/\ ¡ I
400 360
1
I
I lililí
\
I
1—I
I I II
240
Z
200
I
180
i
160
1
I I I I I III
1
1 — i|II l b
CONSTANT AMPLITUDE TESTS E
320
t±
DECKS
to EUROCODE 3)
TRAPEZOIDAL
280 O CL
STEEL BRIDGE
STIFFENER P > 1 mm. TROUP 1974 EARCH HOLSTEIN 1989 PAN YAUADA 1989 — run-ou I
140 O)
120
co
o -z. «=t Q:
co co LU
CE h— co
100
-
90
-
80
-
70
-
El-
60
50
JU-DELFT
40
io
S l e v i n LABORATORY Steel STRUCTURES 1 l ' i i i m 3 4 5 679 10
J
1—I l l l i M 3 4 5 679 10
J l l i l i ll 3 4 5 679 10
N U M B E R OF C Y C L E S (UNTIL TOTAL FAILURE) Figure 4.13 : Field-welded rib Joints of o r t h o t r o p i c s t e e l bridge decks
_L
J I I I I II 3 4 5 679 8 10
[N]
—-> STST-001.GRA
F I E L D - W E L D E D R I B J O I N T S OF O R T H O T R O P I C (S-N C u r v e s a c c o r d i n g A I I I I
400
1
I I I I I II
C
200
Î
'80
3
160
£
140
I I I I I II
I
1—I I I I I
TRAPEZOIDAL STIFFENER : ROOT CAP < 3 mn. ECSC-RESEARCH KOLSTEIN 1989 EARCH JAPAN YAIIADA 1989
280
240
"I
CONSTANT AMPLITUDE TESTS E
320
Q.
DECKS
to E U R O C O D E 3 )
360
o
STEEL BRIDGE
I
— run-oul
i
O) (O
*
120
100 O
90 80 h
en co co
UJ
I— CO
70 60 50 40
10
TU-DEL S I ev ¡ n S t e e l S" I
L
I I 1 I II
10
I I I I II
10"
4 5 6 7 9
10
NUMBER OF C Y C L E S (UNTIL TOTAL FAILURE) Figure 4 ,14 : Field-welded rib Joints of orthotropic steel bridge decks
I
2
3
I
I 1 I I I
4 5
6 7 9 o
10
[N]
-—> STST-002.GRA
5. CONNECTION STIFFENER-C ROSSBEA M.
5.1. Types of connection. Bridges with orthotropic steel decks built in Europe used trapezoidal or 'V' shaped longitudinal stiffeners. Sometimes in the past these longitudinal stiff eners were butted up to the transverse stiffeners. Fatigue failure of these trough to crossbeam welds occurred in heavily trafficked bridges in less than 20 years. In other cases, the longitudinal stiffeners passed through cut-outs in the crossbeams and the connections were no longer weakened by load carrying welds at" each side of the crossbeam. Until now, no failures of the stiffener through crossbeam connections have been reported, however, information about the fatigue classification of different types is unknown. Tests at TRRL were intended to establish the fatigue behaviour of three connections typical of the later types. They are illustrated in Fig. 5.1. Recent research [26] suggests an improved form of the cut-outs in the crossbeam, see type ' R', Fig. 5.2. Tests at Delft were intended to get an agreement of stress distributions and fatigue behaviour of three types of connections (Fig. 5.2) and not to define a fatigue design curve. The work of LBF was to study stress fields in the crossbeam in different shapes of cut-outs in order to reduce stress concentrations (Fig. 3.3 and 3.1). 5.2. Stress determination. 5.2.1. Measurements on a deck panel at TRRL. The three types of connection shown in Fig. 5.1. were incorporated in the central crossbeam of a 15.2 m long by 3.1 m wide deck paneL Strain gauges were installed around each connection, on the web of the trough and on the crossbeam, 15 mm. from the root of the weld. The main gauges installed around connections 'A' and ' B' are shown in Fig. 5.1. Static loads were applied to the panel through a single wheel and the influence surface of stress was obtained for each gauge position. The panel was unsurfaced.
70 -
By superposition, longitudinal influence lines of stress were calculated for the vehicle types described in the British Standard code of practice for
fatigue, BS 5*100 part
method,
histograms
vehicles distributed
of
10 [11], Using the Rain-flow cycle counting
stress
ranges
across the
were
deck
calculated
(BSS^OO, clause
for
one million
C.1.*0 with the
centreline of the distribution directly over the trough. The stress spectra for the important gauge positions around connections 'A' and ' B' are given in Table 5.1 • 5.2.2. Measurements on a UK bridge. Strain gauges were installed around a trough to crossbeam connection on a heavily trafficked UK bridge. The instrumented trough was located under a slow lane wheel track. Gauges were installed at the positions of high stress indicated in the panel tests (see Fig. 5.1). Strains were recorded continuously for a two week period during which the asphalt
temperature
ranged
from
-1.6°C to 18.0°C (mean 7.6°C). Stress
histograms for the 8 gauges are given in Table 5.2. 5.2.3. Local stress calculations and measurements in the web of the crossbeam at LBF. The stress level in the web near to the cut-outs is proportional to the shear forces in the crossbeam and is practically independent of bending moment. Additionally, these stresses are influenced by the loads introduced locally into the crossbeam. Therefore the fatigue problem at the two critical points (notches in the cut-outs and end of weld between the stiffener and the web near to the cut-outs) may be investigated by using a simply supported crossbeam together with short pieces of the stiffeners and of the deck plate as a test specimen and as a model for calculation (Fig. 5.15). This crossbeam model may be loaded by definited shear forces to simulate stress distributions and stress time histories similar to those in bridges under real load conditions. Two shapes of cut-outs in the web of crossbeams were investigated, one which is commonly used (type II) and the other, an "improved" design proposed by Haibach [26] after investigating crossbeams of railway bridges (type I). The stresses due to shear forces in the beams at both types of cut-outs were computed and measured, Fig. 5.3, 5.4, 5.5. As a result the stress distribution around the cut-outs was found to be antimetrie with respect to their plans of symmetry. The level of maximum stress at the notches of both cut-outs is nearly the same but the volume of highly. - 71 -
stressed material is much
larger in the conventional cut-outs. At the end
of the welds (the other critical point of the connections) the new shape produces significantly lower stresses than the conventional shape. 5.2.^. Stress Spectra. The stress spectra were obtained using the simulation program of the University of Liège. This program calculates the level crossing and the Rain-flow histograms. Two influence lines are tested with the Rheden traffic : *
influence
line
stiffener
on
concerning
the
support
stresses of
a
due
to bending moment at the
continuous
beam.
This
has been
measured by the T.R.R.L (Fig. 5.6. and table 5.5). *
influence line of the support reaction of a continuous beam. This has been given by the L.B.F. (Fig. 5.7.).
5.3. Test results of TRRL. 5.3.1. Test specimens. Full-scale test specimens were manufactured comprising a 1500 mm length of deck plate, a single trough and a central crossbeam. Two types of specimen were made with the detailing of the trough to crossbeam connection representing
the 'A'
and
' B'
connections
of Fig. 5.1. Six millimetre,
single pass, Manual Metal Arc Welds were used for the trough to crossbeam connection. Inspection of the welds showed them to be representative of those that could be found on a typical bridge. The specimens were loaded in a reaction frame test rig, illustrated in Fig. 5.8. Strain gauges were installed at the high stress locations around the weld (and 15 mm from the wel root) in identical positions to the gauges on the test panel. The loading on the specimens was arranged to produce a similar
distribution
of
stress around the connection as that determined
from the tests on the deck panel with the wheel load in the most damaging position.
- 72
Five type 'A' and six type 'B' specimens were tested at constant amplitude. A further
four type
' B'
specimens were tested at variable
amplitude. In all cases a 25 mm crack was defined as failure of that part of the specimen. 5.3-2. Test results - type 'A' connection. Fatigue cracks developed at four locations (see Fig. 5.9), though not necessarily all in the same specimen : crack a was a weld toe failure through the trough plate at the bottom of the weld, initiating within 25 mm of the weld end. This crack was expected from the high stresses found at this point in the panel tests. Crack b was in the crossbeam plate at the bottom of the weld. Crack c was in the trough plate at the top of the weld. It occurred in all 5 specimens, on both sides of the trough and on both sides of the crossbeam. In all cases cracks initiated at very low endurances. Stresses were found to be much higher at this location than had been expected ; the stresses were confirmed by measurements on the panel. Crack d was in the crossbeam plate at the top of the weld. It occured in only three specimens and at much longer endurances than crack c. It is regarded as a secondary order crack. For cracks a, b and c the endurances for a 25 mm crack are given in table 5.3 and plotted in Figure. 5.10, against the stress measured by the strain gauge adjacent to the crack location. Estimated stresses at crack b were used for four of the specimens. The data are compared with Eurocode S-N curves. It is concluded that class 50 is appropriate for the failure in the trough plate at the top of the weld (crack c) and class 125 for the weld toe failure at the bottom of the weld (crack a). A high classification is indicated for crack b from the estimated stresses at this point. 5.3.3. Test results - type '3' connection. Out of six specimens, the two tested at the lowest stresses (95 and 100 N/mm2) were uncracked after 11,7 and 13,2 million cycles respectively. The remaining four specimens all suffered weld toe failures through the trough plate as expected (see Fig. 5.8.). Cracks initiated near the apex of the trough. In one specimen, a second crack developed in the toe of the weld at the crossbeam.
73
The endurance for a 25 mm crack is given in table 5.1 and is plotted in Fig. 5.11. against the stress in the trough plate (at the apex of the trough) adjacent to the weld. The results indicate a weld class 80. Four specimens were tested under variable amplitude loading. The applied stress spectrum was derived from the static load test data (section 5.2.1) for the gauge on the apex of the trough under the vehicle loading from the Rheden Bridge (Section 5.2.1). The level of the spectrum was raised to obtain failures in a reasonable timescale. The spectrum used in the tests if given in Table 5.5.. It has an equivalent stress range La
of
2
97.1 N/mm . where Ao - [(1/Zn.) x E(n. a.')] e i i i
1/3
The results of the variable amplitude tests are also shown in Fig. 5.11.» plotted at the level of the equivalent stress range. The four results are below the mean line of the constant amplitude tests the average endurance suggesting that a fatigue life based on the Palmgren-Miner summation would be optimistic by a factor of about 2. However, three out of the four results are within the 95 % confidence limits for weld class 80. 5.3.1. Fatigue life calculations. For each type of specimen, strain gauges were installed in identical positions on the deck panel and on the fatigue test specimens. It is therefore possible, using the stress spectra calculated from the static tests on the deck panel (Table 5.1.) and the weld classifications determined from the constant amplitude fatigue tests, for corresponding gauges, to calculate the fatigue lives of the connections for BS5100 traffic loading : Type 'A' - crack a, bottom of weld through trough - > 120 years - crack b, bottom of weld through crossbeam - > 120 years - crack c, top of weld through trough - 5 years. Type ' B' - crack through trough plate at apex of trough - 13 years. The lives quoted are for a 2,3 % probability of failure and for one
million
HGVs per annum. There
is no
influence
from
bridge deck
surfacing in these calculations. Lives were also calculated using the stress spectrum obtained from the measurements on the bridge. For crack c the life is calculated to be 280 years. This assumes that the traffic flow across the bridge and the temperature
of the bridge deck surfacing for the two week measurement
period is typical of that throughout the year. - 74 -
In fact the traffic flow is
known
to be only about half a million HGVs and the surfacing
temperature
was below the annual mean. Consequently,
the
though failures
connection
in service
fails
to
meet
the
UK
design
requirements
are not expected to occur within the lifetime
(120 years) of the bridge. The
fatigue
life of 13 years calculated for the type '3' connection from
the constant amplitude tests would be reduced to around 6 years using the average endurance data from the variable amplitude tests. 5.H. Tests results of the T.U. Delft. Two series of tests were executed. In the first series the fatigue load was about 50 % higher than in a real bridge. In all cases no cracks were
found after testing
for
at least
12 millions
cycles. In the second
series of tests the specimen of the first series were cut into two pieces and were tested at a much higher level (at least H times the first loading c a s e ) . The testrig in that case was the same as used by the Transport and Road Research Laboratory. 5.^.1. Test specimens. The
specimens
for
the
bending
tests
of
the
first
series
were
single rib specimens. The ribs were rolled trapézoïdal sections of steel grade Fe 510. The dimensions of the cross section were equal to a F.K.H. Trapezprofile nr. 2/325/6. As the usual spacing of the ribs is 600 mm., the width of the deckplate in the specimens was also 600 mm.,
in order to get
the same position of the neutral axis as in an actual bridge deck. Three types of specimen were made with the detailing of the trough to crossbeam connection These
representing
the ' S' , •T' and ' R'
connections were manually
connections
of figure 5.2.
welded. By cutting the test specimen in
two pieces the specimens of the second series were made in the same rig as used at TRRL (Fig. 5.8). The first
test
series were executed with loading equipment from
Losenhauser. The second series were carried out using servo hydraulic test equipment operating in closed loop control with load feedback.
- 75
Each test specimen was instrumented with a number of strain gauges before testing. Strain measurements were carried out dynamically for 2k hours a day to obtain information about the time and location of crack initiation. Measurements of crack growth were carried out during a number of the tests. This was done by periodic visual inspection with a magnifying glass. It was only possible to measure crack length at the surface of the specimens.
Where possible, four stages of fatigue failure are expressed (see section 4.M.2) : 5.1.2. First series of tests. As mentioned before no cracks were found in the first series of tests at a load range of 50 kN. From these results it appears that the highest
stress
is
measured
in
the
connections
without
cut-outs. The
stresses measured on the connections with the new design of the cut-outs are a little bit higher than the design with the old cut-outs. 5.^.3. Second series of tests. A review of the test results is given in Table 5.6. In this table the following parameters are given : - the fatigue load ; - number of cycles for each stage of fatigue failure in the crossbeam and the weld. The results plotted on Fig. 5.13 and 5.1*4 correspond to N3. The location of the cracks is shown in Fig. 5.12. The cracks appeared in the crossbeam are influenced by the test specimen and the testing, they are not considered here. The two specimens type ' S' suffered weld toe failures through the trough plate as expected. Cracks initiated near the rounding of the botton-web of the trough. Fig. 5.13 shows that the fatigue behaviour is better than the highest class of the Eurocode curves.
- 76 -
In
type
'T1
-
connection,
fatigue
cracks
developed
in
the
crossbeam in the rounding of the cut-out and the weld toe through the tough plate at the botton of the weld. For the crack initiating at the weld, the endurance is plotted against the stress measured at 12 mm from the weld toe (Fig. 5.11). Just as the type ' T' - connection, in type ' R' connection, fatigue cracks developed in the crossbeam in the rounding of the cut-out and at the weld toe through the trough plate at the bottom of the weld. For the crack initiating measured
at
at
the
12 mm
weld, from
the
endurance
is
plotted
against
the
stress
the weld toe (figure 5.11). It seems that also
fatigue behaviour of this detail is a little better as for connection T. 5.1.5 Conlusions. Connection type S, without cut-outs has a better fatigue behaviour as
connections
type
T
and
R
with
cut-outs.
The
fatigue
behaviour
of
connections type T and R is very near. This conclusions are in opposite of the results obtained at the V stiffeners (section 5.3) : type B, without cut-outs gives a fatigue behaviour near at type T and R, but type A, with cut outs at the apex, gives a higher fatigue behaviour as type B.
5.5. Test results of the L.B.F. 5.5.1. Test conditions. Three test specimens were fabricated, nearly the same scale as the orthotropic
decks of
stiffeners, and commonly
used
real bridges, each
consisting
of
a cross beam, 6
a deck plate. One of the specimens had cut-outs of the
shape
and
two the improved shape. When the specimens are
loaded as shown in Fig. 5.15, four of the cut-outs are stressed at nearly
- 77 -
the same level. This is due to symmetry and the constant shear force. Therefore
up to four results can be expected from each test specimen,
provided cracks are early detected early and repaired. Three fatigue
tests were performed one with constant amplitude
loads applied to a specimen with the new shape of cut-outs and two using identical
load sequences of variable
amplitude
(derived
from
measured
traffic loads) applied to the other two specimens one with the old and one with the new cut-outs. 5.5.2. Test results. Test results are given in Fig. 5.16, 5.17, 5.18 and 5.19 cracks occured at both critical points of the connection specified earlier. At the
end
of
the
welds
between
the
longitudinal
stiffeners
and
the
crossbeam, cracks were observed only in the specimens with the new shape of the cut-outs, although this shape was developed
especially to reduce
the stresses at this point. The stresses due to the external loads are actually lower as shown by computation and measurement, Fig. 5.3, 5.4. However, it seems that the new
shape increases residual stresses at the
end of the welds due to unequal heating during welding. High residual stresses allow fatigue cracks to develop ever when the stresses from the applied loads are low. With crack initiation, the residual stresses are removed and the craks do not grow (this was observed during the tests). In contrast the cracks initiated at the notches of the cut-outs grew and would have destroyed the specimens if not repaired. These cracks occured at the cut-outs of the new shape after 3-5 times more load cycles than
at
the more
commonly
used
cut-outs,
which
means
a
significant
improvement of fatigue life. However high residual stresses globally distributed in all three specimens were observed without any doubt. Their influence on fatigue is so strong that cracks occured at notches stressed only in compression due to the external loads, while other notches remained crack free, although stressed in tension at the same level. Therefore it is not clear, whether the increase of life time is due to the new shape of the cut-outs alone or at t ti
least partially to a more favourable residual stress distribution.
v ./ 78 -
5.6. Conclusions. Three types of stiffener to crossbeam connection were tested at TRRL, LBF and Delft : welded all round and with conventional and 'improved' cut-outs in the crossbeam around the apex of the stiffener. Both trapezoidal and 'V' shaped stiffeners were tested. In one of the research programmes
the main
aim was to study
the fatigue behaviour of the
crossbeam by applying loads to produce shear forces in the cross beam. In the other two programmes the fatigue behaviour of the welds was studied by applying bending stresses to the longitudinal stiffener. Despite difference in the shape of the longitudinal stiffeners, the weld sizes and probably different residual stresses due to different weld
procedures, a minimum weld classification of class 80 following
Eurocode 3 can be considered for the welded all round connection. However, a
higher
classification
may
(trapézoïdal stiffener without
be
considered
for
cut outs) as for
connection
type
S
connection type A
(triangular stiffener with cut-outs at the apex). Conclusions for joints in trapézoïdal stiffener are provisional, because only results for two tests for each type of connection are available. Most of the variable amplitude test on this type of connection gave a good agreement with the constant amplitude tests applying Miner's method of damage summation. Because of the difference modes of failure from the difference types of welded connections with cut-outs in the crossbeam it is not possible at present to define an overall classification for this detail. For specific details, classifications are given in previous sections where possible. The improved form of cut-outs did not result in a better fatigue behaviour as had been expected. Cracks occured in unexpected locations in the crossbeam connection in the new design despite low calculated stresses at this point. Weld toe failures occurred at similar endurances in the conventional design. For the failure in the crossbeam plate the new design was an improvement over the conventional design but it was difficult to manufacture this shape of cut-out without producing notches.
79
For
the
'V*
shaped
stiffeners,
the
connection
with
cut-outs
behaved better than the fully welded type. This was not the case with the trapezoidal stiffeners. A possible explanation is that residual stresses may be higher at the apex of the 'V' stiffener than in the bottom of the trapezoidal stiffener. Therefore a cut-out at the apex of the stiffener which avoids residual stresses in this area gives a greater improvement in fatigue behaviour for the 'V' stiffener.
To
give
a
general
classification
for
stiffener
to
crossbeam
connections it would be necessary to refer to nominal stresses rather than stresses at specific designer.
Further
points
analysis
which of
the
cannot be easily results and
needed to give practical design 'recommendations.
- 80 -
calculated by the
additional
research
is
STRESS RANGE
OF CYCLES
(N/mm )
CONNECTION 'A' GAUGE 13
CONNECTION 'A1 GAUGE 88
CONNECTION 'A1 GAUGE 90
CONNECTION 'B' GAUGE 49
0-10
6477245
6118059
8456885
3119612
10-20
481807
718470
520645
1254028
20-30
355667
279695
288755
582721
30-40
255613
317424
100389
302745
40-50
83593
175765
0
146634
50-60
22933
61490
0
205557
60-70
4063
38862
0
78214
70-80
0
3899
0
21212
80-90
0
0
0
11404
90-100
0
0
0
1969
°R
09
NUMBER
2
- For 1.000.000 HGVs Centre of distribution of vehicles over centreline of trough TABLE 5 . 1 .
STRESSES
FROM
TESTS
ON DECK
PANEL
STRESS
NUMBER OF CYCLES
RANGE GAUGE NUMBER
°R 2
(N/mm )
1
00
ro i
1
2
3
4
5
6
7
8
8-12
15383
14680
20851
17866
14454
16098
12-16
26099
20709
4614
10408
11301
11785
5038
10518
16-20
12632
16227
1790
6254
5613
3642
2071
6064
20-24
876
. 6787
10108
3515
2236
747
1153
24-28
3535
455
2650
3341
2116
992
227
682
2569
28-32
1463
229
936
1698
325
32-36
664
1082
36-40
537
490
34 3
438 330
1600
80 21
29 0 0
157
40-44
127
5
186 28
518 120
177
0
0
60
44-48
21
0
2
0
42
0
0
15
48-52
9
0
1
0
4
2
0
52-56
2
1
0
0
0
1
0
0
56-60
0
0
0
2 0
0 0
0
1
1 0
TABLE 5.2.
STRESSES
FROM
MEASUREMENTS
ON
BRIDGE
17
0
CRACK a NORTH SIDE
CRACK b SOUTH SIDE
NORTH SIDE
CRACK c SOUTH SIDE
NORTH SIDE
SOUTH SIDE
STRESS CYCLES STRESS CYCLES STRESS CYCLES STRESS CYCLES STRESS CYCLES STRESS CYCLES SPECIMEN RANGE TO RANGE TO RANGE TO RANGE TO RANGE TO RANGE TO NUMBER Oi FAILURE o=, FAILURE FAILURE Oii» FAILURE OT» O n FAILURE FAILURE (N/mm a ) (N/mm z ) X10d XIO* (N/mm 3 ) X10& (N/mm z ) (N/mm*) XIO* xio<- (N/ram=) XIO*
00
u
IA
125
>12.4
113
)12.4
137-
>12.4
124-
>12.4
2A
ISO
5.2
142
> 6.3
164*
> 6.3
150-
3A
200
1.8
192
> 2.5
219-
2.3
5A
144
1.4
175
2.1
158-
2.1
8A
160
6.3
151
> 7.0
176
> 7.0
39*
6.3
50-
8.2
5.8
47
3.0
59
4.0
210*
> 2.5
73
1.8
63
2.4
192*
1.8
56
2.7"
54
2.9"
> 7.0
63
2.9
63
5.2
177
denotes estimated stress denotes extrapolated cycles
TADLE g-3 FATIGUE TEST RESULTS - TYPE 'A' SPECIMENS
T.R.R.L
SPECIMEN NUMBER
23 33 43 63 7B 8B
a)
103 12B 133 14B
TABLE SA
Ca (N/RWli*)
95 150 125 115 100 200
CYCLES TO FAILURE X10* >11.70 0.90 2.80 1.67 >13.20 0.47
SPECIMENS TESTED AT CONSTANT AMPLITUDE
SPECIMEN NUMBER
b)
STRESS RANGE
MAXIMUM STRESS RANGE Oi
(N/rem3) 245 245 245 245
CYCLES TO FAILURE XI0* 2.12 0.7S 1.86 2.75
SPECIMENS TESTED AT VARIABLE AMPLITUDE
FATIGUE TEST RESULTS - TYPE -'B ' SPECIMENS TRRL
- 84
STRESS RANGE N/mro2
CLASS CENTRE
CYCLES
N/iran2
ni
50-60
55
23710
60-70
65
17440
70-80
75
12590
80-90
85
10920
90-100
95
9080
100-110
105
6850
110-120
115
5290
120-130
125
3440
130-140
135
3690
140-150
145
2340
150-160
155
"980
160-170
165
1410
170-180
175
570
180-190
185
710
190-200
195
340
200-210
205
90
210-220
215
240
220-230
225
70
230-140
235
50
240-250
245
10
TABLE 5. 5 S T R E S S IN
THE
SPECTRUM TRRL
- 85
USED
TESTS
"T&ft-l
TESTSPECIMEN
LOAD
NR
[kN]
TYPE
v vy \J
u \j \j
LOCATION OF
NUMBER OF CYCLES (xlOE06)
Nl
CRACKS
N2
N3
N4
1.658 0.809
1.732 0.879
2.461 2.461
1.157 1.125
1.184 1.157
1.272 1.272
VELD
NORTH SOUTH
CROSS BEAM
NORTH SOUTH
VELD
NORTH SOUTH
CROSS BEAM
NORTH SOUTH
WELD
NORTH SOUTH
0.080
0.271
0.474
1.190 1.190
CROSS BEAM
NORTH SOUTH BOTTOM
0.660 0.090 0.480
0.771 0.174 0.653.
1.190 0.711
1.190 1.190 1.19P
VELD
NORTH SOUTH
0.040
0.241
0.250
0.572 0.572
0.016 0.016
0.025 0.025 0.087
0.564 0.087 0.112
0.572 0.572 0.572
0.050
0.187
0.348
0.700 0.700
0.208 0.103
0.606 0.397
0.700 0.700
0.175
0.351
0.856 0.856
0.369 0.084
0.676 0.376
0.856 0.856
SI 200
S2 200
R2 200
Rl 267 CROSS BEAM
NORTH SOUTH BOTTOM
VELD
NORTH SOUTH
TI 266 CROSS BEAM
NORTH SOUTH
VELD
NORTH SOUTH
0.100
T2 267 ■ CROSS BEAM
NORTH SOUTH
Nl : Moment of crack initiation by 10 X strain fall of, measured in the strain gauge nearest to the crack. N2 : Moment of visuable crack initiation. N3 : A crack indicating the number of cycles when a surface crack length of ± 50 mm is reached. N4 : End of extensive through cross section cracking.
Tablé S.'£..
R eview of the second series results. T.U. Delft.
86 -
Type 'C' Type 'B' Type 'A' Deck plate
00
Web of trough Welded connection under test
Figure 5.1 : Types of longitudinal/transverse stiffener connectlon-TRRL And main gauges in test (specimen 'A'and 'B'and in the bridge studied (specimen 'C'
and weld Icul oul )
00
co
and weld (cut out)
Figure 5.2 : Types of longitudinal / Tramsverse stlffener connection
•
measured at ~ 2 mm distance to the edge
D measured on the edge load F = 200kN s. Fig. 5.21)
b^
calculated
00 (O
Figure 5.3 COMPARISON OF COMPUTED AND MEASURED STRESSES AT THE CUT-OUTS OF THE NOW USUALLY USED SHAPE (TYPE II)
o %
c. calculated calculated
^
load F = 200kN Is. Fig. 5.21 J
/
CO I
/
S
-compression
D D
\
\
\
X
\ D
measured on the edge
x
measured at - 2 mm distance to the edge
measured at ~ 2 mm distance to the edge
Figure 5.4
COMPARISON OF COMPUTED AND MEASURED STRESSES AT THE CUT-OUTS OF THE SHAPE PROPOSED TO BE BETTER {TYPE I )
(O
measured stress Figure 5.5 STRESS DISTRIBUTION IN THE NARROWST SECTION BETWEEN THE CUT-OUTS OF THE SHAPE PROPOSED TO BE BETTER
___. influence line / V " X N . used for —» //' ^, the simulation./^' ' • ^ .
20
15 ■■
I 76.21 (152.AI
10 ■■
vw f/r^X
/
(226.6) (30 4.81
v \ a/
(381.0)
%
/
/
\
\
5"
\
. 1 if
/
\\W/ mi /
■■ 3
H 1 1—I—' I I i \ W / / l 1 I I 1 3 7 11 15 19 25 \ W :\':I/-K / 3 S A1 « fop a 20kN wheel W
h—I—I—I—I—I—I—h 59 M 51 CO K 67 7 75 75
y
"N/mm2 influence lines measured a t t h e T.R.R.L. with a 20kN wheel Histogram Rain flow NI/NTOT
cu--
Rheden Traffic 0.3 •
0.2■
0.1
N/mm
' rT") n 11111111 50
100
150
H—» 250
200
Fatigue damge ,.DI 10 + D3 1/0.SE*03 n D z0.3A2n 3 o
9 8 ■•
7 E 5 4 3 2 1 ■
r£ 50
100 AS
Ek_DL 150
N/mm' 200
4
250
Figure 5.6 : Detail t e s t e d at T.R.R.L. 92
Influence line
Histogram Rain flow NI/NTOT 0.5 ■
Rheden Traffic
0.4 ■
0.3 •
0.2 •
0.1 •
kN
100
200
300
¿00
500
Fatigue damge M
D I
10• D3 1/0.1E.05
g .. nm
=0.108 naj,
8 7 ■ 6 S ■
¿ 3 ■ 2 ■ 1 ■
=d£
100
iL
200
300
UL.
kN ¿00
Figure 5.7 : D etail tested at LBF
93
H SOO
■
Hydraulic actuator
Reaction frame
Figu
re
5.8 : Fatigue test rig 94
Soirth si d«
TYPE A SPECIMEN
South si dì
Fatigue crack at toe of velo
Figure 5.9 : Location of cracks in fatigue t e s t specimens
95
lOÜO
Mean line 9Ş% confidence Limirs STRESS RANGE N/mm2
o • x
crack a crack b crack c
Eurocode class 125
100
(O O)
Eurocode class 50
J CYCLES
Figure 5.10 : Fatigue tests at constant amplitude - TYPE A* SPECIMtNS
l i l i IU'
1000'
300
Ş-ÎLii'Unviîş rem Evnocohiz cLÃãs 'uu
[-lean l i n e V57. c o n f i d e n c e
1 i m i LK
Test::» a t c o n f i a n t . ;>mi> I. ¡Lude T e s i s a t v a r i a b l e ampi I U n i e
HANGE (ll/mma)
100
CO
50
APPLJED «IMÍCTUWM
10
"1
1
1
1" I
i l l
i
1—i—i—i
i i i
10
CYCLES
Figure 5.11 : Tests on TYPE B SPECIMEN
TYPE S Crack
TYPE T Erack
TYPE R
Crack
Figure 5.12 : Location of the cracks in the fatigue tests specimens
- 98
TROUGH TO CROSSBEAM CONNECTION OF ORTHOTROPIC STEEL BRIDGE DECK (SN Curves according
io
A I I
c -7 ro
n. °
I
I I I III
"I—I
T
I I I I II
OJ »
O
■a
ro o n>
<
320
THEORETICAL NOMINAL STRESS RANGE
280
TUDECSC Research •. TYPET: Connection w i t h
240
.— i
-J
-*■ CU 3 UD CD
'old*
cutout«
A TYPER: Connection w i t h ' n e w ' c u t o u t « ô
TYPES:
Connection w i t h o u t
cutout«
— runout
200
(/)
3
180
V) O ft U
160
2 O K
140 120
m 100
ro a f ro <"
i—i i i 11 i i 1—i—n-rrru CONSTANT AMPLITUD E TESTS E
oj
S'Y1 *~ I
i
360
C O í^5 C O
CD
-h
1—r
H m
«^f ro -•-
CO CO
400
l o EUROCOD E 3 )
Lu O
z:
90 80
< 70
co co
60
IJJ
ce i— co
50
40
10
TU-DELFT S t e v i n LABORATORY S t e e l STRUCTURES I
l I l l LLU 3 4 567 9
i.
10 NUMBER OF CYCLES
J
I
I I I I II
3 4 5 67 9
.J 3
I I I I I II 4 567 9
10 (AT VISUAL CRACK INITIATION)
7
J 3
10
l_.L _ U ± 4 5 67 9
10
[N] CRB.GRA
TROUGH TO CROSSBEAM CONNECTION OF ORTHOTROPIC STEEL BRID GE D ECKS (SN Curves UD
c ro
A
■p
s« QJ UD UD
*•
s ir
,?"> TD
£ M1
a" .
UJ
o
o
268 KN
U TYPET: Connection w l l h ' o l d '
eutouti
A TYPER: Connection w i t h 'now'
cut—out»
200 180 160
K.
140
> ■ij
120
u.
= wel
fM *m^
LÜ
O
^
ro
Q-
-z. < er
50
CO CO
Ld ro
LOAD RANGE 200 TUDECSC Research
•»run—oul
iÜ o o
"i i r~i i 11 n i—t—i-inrnT CONSTANT.AMPLITUDE TESTS E
o. 240
- * ■
T ro -s w -J c o «—
1—i—r-rrnr
280
ro <"
3
n-nrm
to EUROCOD E 3 )
320
~& o
i
according
360
ui
f
400
100 90 80 70
'Ao 60
o: r
co
12 //////////s.
JU-DELFT 50
S t e v i n LABORATORY S t e e l STRUCTURES I
40
10
l l I I LLLL 3 4 567 9
•
J
I I 3 4 567
10
NUMBER OF CYCLES
9
J 2
I I I I I I II 3 4 567 9
10 (AT A
CRACK LCNGTH OF 50 MM)
2
J_J__L.LJJXl 3 4 5 6 7 9 o
10
10
[N]
> CRU-12.GRA
en uu Q LU
en
o
Ü_
en
LU CL
LO LO dl DI
load
o
to
crack crack crack crack crack crack crack
no. no. no. no. no. no. no.
1 2 3 U 5 6 7
after after after after after after after
205 339 339 3U 350 350 350
000 cycles 000 cycles 000 cycles (no crack growth observed) 000 cycles 000 cycles 000 cycles 000 cycles
T : region with tension stresses C : region with compression stresses
Figure 5.16 : RESULTS OF CONSTANT LOAD AMPLITUDE TEST WITH A SPECIMEN HAVING CUT-OUTS OF THE SHAPE PROPOSED TO BE BETTER (TYPE I )
load
o co
crack no. crack no. crack no. crack no. end of test
Figure 5.17
after 2 8 U 200 cycles after 3 ¿09 900 cycles after ¿ 5¿1 000 cycles after 5 780 300 cycles after 5 926 600 cycles
T : region with tension stresses C : region with compression stresses
RESULTS OF THE VARIABLE LOAD AMPLITUDE TEST WITH THE SPECIMEN HAVING CUT-OUTS OF THE NOW USUALLY USED SHAPE (TYPE I I )
load
o
crack no. 1 crack no. 2 crack no. 3 crack no. U crack no. 5 crack no. 6 end of test
Figure 5.18
after after after after after after after
6 6 6 6 13 16 22
728 728 728 728 385 089 129
000 000 000 000 000 000 000
cycles cycles cycles cycles cycles cycles cycles
(no (no (no (no
crack crack crack crack
growth growth growth growth
observed) observed) observed) observed)
T : region with tension stresses C : region with compression stresses
RESULTS OF THE VARIABLE LOAD AMPLITUDE TEST WITH A SPECIMEN HAVING CUT-OUTS OF THE SHAPE PROPOSED TO BE BETTER (TYPE I )
800
I
I I I I lili
I I i IUll 1 — T results of variable amplitude tests Pmax = 500 kN P.. = 15 kN
S-N test results P = 500 kN Pu= 15 kN
kN
¿85
lililí
U cracks
¿00 -o o o
AP K = en o
UI
c a
B-¿-
equivalent load En,
k =5 k =A
¿-E¿ e- -E¿-k = 3
APM
Un, A Pi"
200
equivalent load cycles k En,i AP *- ' :i
EAR' estimated S-N curve slope k = 5
100 10 4
1—I I I H i l l 2 3 4 56 8 1 0 5
1 — ' ' ' m i i 2 3 4 56 8 1Q6
i 2
i i i M MI 3 4 56 8 1Q7
i 2
i i i i nn 3 4 56 8 i n 8( 10 cycles
Figure 5.19 : RESULTS OF THE CONSTANT AND VARIABLE LOAD AMPLITUDE TESTS (TYPE I )
6. ORTHOTROPIC DECK TO CROSSBEAM CONNECTION. 6.1. Introduction. The bolted connections tested are used to attach orthotropic deck plates
to supporting
I section crossbeams. This type of construction is
found in harbow arens where the bridges carry both road and rail traffic. These orthotropic steel decks have to be dismountable. These bolted connections are subjected to complex forces resulting from angular rotation of the supports (lever effect), from vertical forces (tension or compression) and from horizontal strains These forces are
difficult to calculate because
(axial shortening).
they depend on element
stiffness and secondary effects. Tests
were
carried
out
on
different
specimen
configurations
subjected to forces representing those found in bridges. In addition, tension-bending tests on bolts were carried out.
6.2. Results from the University of Liège on the connection. The problem was solved using two approaches : * introduction
of
fexible
elements
to
reduce
secondary
bending
stresses. Tests gave the solution presented in Fig. 6.1. ; *
using
elements
with
sufficient
stiffness
to
reduce
secondary
effects. 6.3. Results of fatigue tests on 8.8 bolts. These tests were carried to study the influence of the following parameters on the fatigue strength of the bolts : * stiffness of the beams with influences the angular rotation of bolt heads and thus influences the bending stresses in the bolts ; * prestressing ; * introduction of neoprene element ; Characteristics of tested bolts are : * nominal diameter : 12 mm ; * measured net section : 8*4,3 ram2 * measured ultimate stress : 896 N/mm2 * Brunell hardness : 280 to 295. - 106 -
Test beams are : * series 1 : IPE 160
- without prestressing.
* series 2 : IPE 160
- with prestressing.
* series 3 : IPE 160
- with neoprene and steel element.
* series H : HEM 120
- without prestressing.
* series 5 : HEM 120 - with prestressing.
A specimen of series k is shown in Fig. 6.2. and test results are presented in Fig. 6.3» Prestressing
has
favourable
effects
of
reducing
the
angular
rotation and the stress variation. Figure
6.3.
shows
that
the
Eurocode
3
classification
is
conservative. It is lower than the most unfavourable tests (flexible beams without prestressing). Before to have general conclusion, more test are necessary.
- 107
]
Q max :80kN
Coupe. A A
Q min : 5kN
plat
205
12
raidisseur o 00
L_ 100/100/10 -Ó
-masmaamaat
1363
rondel le neoprene frette 6Q/3Q/7
Figure 6.1
ctìT\
neoprene frette 80/80/7
t
rondelle d'acier 024
70mm
rondelle d'acier#24 ep 2mm
point de rupture du boulon i
»
♦ Figure 6.2
109
Boulon 8.8-012 400
I
i I I I 111
I
T T
TT
320
1 profilé IPE 160
— 280 Cu CL 240
3 profilés IPE 160»néoprène X 5A
Í.D
4 profilés HEM 120
X 5B SC
200
2 profilés IPE ICO prestressed bold. 5 p r o f i l s HEM 120 -
CO ÜJ CD
Œ ÛC
mo AC
• • •—*-
120
3A
en hm
3D 3C
100
tB 2A
CO CO UJ
N.B. : for s e r i e s 2 and 5 drawn s t r e s s ranges are not actuals but are obtained as if bolts where not presstressed.
i 60
80 70 60 5040
10
i
i i M m
3 4 568
I
1 I I I III
3 4 56 8
i i i i i in 3 4 56 8
10
Figure 6.3
I
2
i i i i i m 3 4 568
CYCLES N
10
7. APPLICATIONS. The sponsors
of this research are the European Community for Cool
and Steel, the National Autorities responsible for bridge building and bridge builders. All the results obtained in the laboratories participating to the research have been widely disseminated in each country . Some of the information is already
used
in
the
new
versions
of
the
design rules and national
recommandât ions. The fatigue strength
of the connection stiffener-plate given in
the Belgian Standard NBN 5 has been changed, taking in account the new results. The experts busy with the redaction of Eurocode 3, have also been informed of the new results. The conclusions of chapter 3, concerning the minimum value of the root gap of the connection stiffener-stiffener is now a rule for the bridge builders in Netherlands, Italy and France. Designer welder and superviser are informed, and pay particular attention to this very important detail. In Germany, the new shape of the cut-outs in the web of the cross-beam is a part of the design rules of Deutsche Bundesbahn following the
investigations
informations
of
concerning
Haibach the
and
fatigue
Plasil
[26]. The
behaviour
research
of this shape
gives
in road
bridges. Orthotropic decks are excluded form the UK bridges design code [11] but the code advises the designer to seek specialist advice. That advice is frequently provided by the TRRL, so that the results of research are passed on directly
to bridge
designers. In addition, designs are
subject to technical approval by the Department of Transport who receive report of all TRRL research. The results construction.
have
already
been
applied
in
some
bridges
in
The new automatic welding procedures studied by IRSID and LCPC for the connection stiffener-plate have been applied to the fabrication of the orthotropic steel deck of two large bridges in France : Pont de Cheviré on the Loire river : Pont de Normandie on the Seine rive. Submerged arc welding was used, without edge preparation and the lack of penetration below 1 mm. - 111 -
Analytical stiffener-plate,
methods
weld
to
developed
be
enable
calculated
the stress range
in the
for various stiffener shapes.
Following this calculation, the shape of the stiffener was changed in the design of the Kruiskansburg in Antwerpen (reducing of the thickness of the stiffenerweb and increasing of the height) in such a way as to increase significatly the fatigue life. The
results
obtained
at
the
stiffener-stiffener connection have been
University
of
Pisa
for
the
used in the preliminary studies
for the design of the Messina Bridge in Italy (suspended steel bridge, with a span of 3300 metres and with six lanes for road traffic and two rail ways). The results obtained by the testing of the bolted connection of the orthotropic deck to cross-beam are used for the design of repairing the Oosterweelburg in Antwerpen and the design of the new Kruiskansbridge.
8. CONCLUSIONS. The first
two phases
of this common
reserach
programme were
concerned with the collection of traffic data needed to determine loads on road
bridges.
This
third
phase
concerned
the
fatigue
behaviour
of
orthotropic steel decks. It was necessary to analyse the fatigue behaviour of all the connections in the orthotropic deck. Because of the scale of the problem, it was not possible for all the testing and analysis to be carried out in a reasonnable timescale in one laboratory or one country. A common research programme was therefore formulated and the work distributed between seven laboratories situated in six countries of the European Community. For some connections improvements in the design are suggested : for
others,
unexpected
behaviour
occurred
which
require
further
investigations. Stiffenei—pkate connection. A welding procedure has been defined which gives a much better fatigue resistance and excludes cracks in the weld.
- 112
The
calculation method
developed
allows alternative
stiffener
shapes to be assessed which may lead to an improved deck design. A fracture mechanics model has been developed and calibrated using the experimental results. Such a model, which can be used on a micro computer, has already been used successfully to study fatigue problems on existing bridges. Stiffener-stiffener connection The
tests
on the stiffener-stiffener
connection with backing
strips showed the importance od the root gap before welding and indicated a better behaviour of triangular compared with trapezoidal stiffeners. Triangular stiffeners butt welded from both sides dit not give the expected improvement in fatigue strength over the butt weld made from one side on a backing strip. This
important
information can be
directly
applied by bridge
builders. However, further investigations are required to clarify these results and assess the influence of construction procedures and residual stresses. Fatigue
cracks are
expected to develop in this connection on
existing bridges in a short time. Repair procedures should therefore be developed as a matter of urgency. Stiffener crossbeam connection. Test results from the stiffener-crossbeam connection indicate a minimum
fatigue
classification
of
Eurocode
class 80. However, cracks
occured where they were not expected and more information is required on the shape of the cut-outs in the crossbeam. A improvement
cut-out in
around
fatigue
the
strength
apex over
of
the
triangular
a welded
rib
gave
an
all round connection.
However, cracks developed unexpectedly at small cut-outs near the deck plate. Conversely, for trapezoidal ribs, cout-outs around the apex of the rib reduced the fatigue strength, the "optimised" shape of cut-out proposed for Germam railway bridges dit not give the expected improvement in fatigue strength in these tests.
113
Before new tests are carried out, a computer analysis of the stress distribution at the critical points is necessary to explain this behaviour an to assess new shapes of this connection. Bolted connections. The
fatigue
tests
on
bolted
connections
show
that
the
classification given in Eurocode 3 corresponds to the worst case. This connection, which occurs frequently in steel bridges, needs further investigation in order to define more realistic S-N curves.
Results from tests carried out under variable amplitude loading generally showed good agreement with lives estimated using the Miner rule. It may out
be concluded
mainly
under
that
constant
in furture, fatigue amplitude
loading.
tests can be carried These
tests
are more
economical. Only a small number of tests under variable amplitude loading are then required to confirm the application of the Miner rule for each particular case.
This work gives a lot of information on orthotropic steel decks which can be used directly by bridge designers and builders. However, some problems remain which require furhter investigation.
114
BIBLIOGRAPHY [1] Dr. Lehrke "Messung und Interpretation von Dynamischen Lasten an StahlbrUcken". Forschungsvereinbarung Nr. 7210-KD/119 - Abschlussbericht - Phase 3. Fraunhofer - Institut rur Betrieksfestigkeit. [2] A. BRULS - E. POLEUR. "Mesures et Interprétation des Charges dynamiques dans les Ponts". Recherche n° 7210-KD/201 - Rapport final - Phase 3. Service Ponts et Charpentes, Université de Liège. [3] A. BIGNONNET (IRSID) and CARRACILLI, B. JACOB (L.C.P.C.) "Mesures et Interprétation des charges dynamiques dans les Ponts". Recherche n° 7210 - KD/317 - Rapport final - Phase 3. Institut de Recherches de la Sidérurgie Française. Laboratoire Central des Ponts et Chaussées. [H] S. CARAMELLI, P. CROCE, M. FROLI, L . SANPAOLESI. "Misure ed Interpretazioni dei Carichi dinamici sui Ponti". Convenzione n°7210 - KD/¿)11 - Relatione technica finale - Fase 3 Instuto di Scienza delle Costruzioni, Università di Pisa. [5] H. KOLSTEIN, J. de Back. "Mesurements and Interpretation of dynamic loads in Bridges". Agreement Number : 7210-KD/609 - Final Report - Phase 3. Delft University of Technology - Stevin Laboratory. [6] C. BEALES. "Measurements and Interpreation of dynamic Loads in Bridges". Agreement Number : 7210 - KD/807 - Final Report - Phase 3. Transport and Road Research Laboratory. [7] DE BACK, A. BRULS, J. CARRACILLI, E. HOFFMANN, L. SANPAOLESI, J.P. TILLY and J.M. Zaschel. "Measurements and Interpretation of dynamic Loads on Bridges". Synthesis Report - Phase 1. Commission of the European Communities. EUR 775^ FR, EN, DE (1982). [8] E. HAIBACH ,J. De BACK, A. BRULS, J. CARRACILLI, B. JACOB, M.H. KOLSTEIN, J.PAGE, M.R. PFEIPER, SANPAOLESI, TILLY, ZACHEL, HOFFMAN. Measurements and Interpretation of dynamic Loads on Bridges". Synthesis Report - Phase 2. Commission of the European Communities. EUR 9759 FR, EN, DE
(1986). - 115 -
[9]
A. BRULS, B. JACOB, G. SEDLACEK and all. Traffic data of the European countries. Eurocode on action - Part 12 : Traffic loads on Bridges. Report of working groupe 2.
[10]
NBN5 "Ponts en acier - projet 1987". Institut Belge de normalisation, Bruxelles.
[11]
BS 5400 - Steel, concrete and composite bridges. Part 10 : Code of practice for fatigue. British Standards Institution, London, 1980.
[12]
Eurocode n°3 : Design Steel Structures. Final Draft. December 1988.
[13]
NEN 2063. "Arc Welding - Fatigue loaded structures - Calculation of welded joints in unelloyed and low-alloy steel up to an including Fe 510 (Fe 52)". Nederlands Normalisatie Instituut march 1988. Delft.
[14]
MADD0X S.J. "The fatigue behaviour of tropezoïdal stiffener to deck plate welds in orthotropic bridge decks". T.R.R.L. Supplementary Report 96 U.C.
[15]
TH0NNARD - JANSS. "Comportement en fatigue des dalles orthotropes avec raidisseurs trapézoïdaux. CRIF : Section Métallique - MT lol - Août 1985. D.E. NUNN AND J.R. CUNINGHAME. "Stresses under wheel loading in a steel orthotropic deck with V Stiffeners". TRRL Report 59 U.C.
[16]
[17]
D.E. NUNN AND J.R. CUNINGHAME. "Stresses under wheel loading in steel orthotropic deck with trapézoïdal stiffeners". TRRL Report 53 U.C.
[18]
HUber M.H. "Die Grundlagen einer rationellen Berechnung der kreuzweise bewerthn Eisenbetonplatten". Zeitschrift der Osterreiches Ingenieur und Architekten-Vereines, n.30, 1914.
[19]
Pelikan W., Esslinger M. : Die Stahlfahrbahn, Berechnung und Konstruction". M.A.N. Forschungsheft, 7, 1957.
[20]
CROCE P. : "Linfluenza della variabilità di sezione nelle nervature di una piastra ortotropa". Atti dell'Istituto di Scienza delle Costruzioni, n.252, vol. XVIII, Pisa, 1988.
[21]
LEVY M. : "Sur l'équilibre élastique d'une plaque rectangulaire". Compte rendu acad. S c , 129 1989.
[22]
CUNINGHAME J.R. Steel bridge decks : Fatigue performance of joints between longitudinal stiffeners. TRRL. Report. LR 1066.
116
[233
K. YAMADA Fatigue Strength of Field-Welded Rib Joints of Orthotropic Steel Decks. IIW. Doc. XIII 1282-88. Department of Civil Engineering, Nagoya University.
[24]
TROMP : Fatigue of field splices in ribs of orthotropic steel bridges decks. Report 6-7*1-15. Stevin Laboratory - Steels structures - Delft University of Technology 197*1.
[25]
A. BRULS Mesures et interprétation des charges dynamiques dans les ponts. 2ème phase. - REcherche CECA. Rapport EUR. 8864.
[26]
E. HAIBACH, PLASIL. Untersuchen sur Betriebsfestigkeit von Stahlleichtfahrbahnen mit Trapezhohlsteifen im Eisembahnbrückenbau. Der Stahlbau 53~S 269-27*1. 1983.
117
European Communities — Commission EUR 13378 -
Measurement and interpretation of dynamic loads in bridges Phase 3: Fatigue behaviour of orthotropic steel decks
A. Bruls Luxembourg: Office for Official Publications of the European Communities 1991 - VI, 117 pp., num. tab., fig. - 21.0 x 29.7 cm Technical steel research series ISBN 92-826-0532-9 Catalogue number: CD-NA-13378-EN-C Price (excluding VAT) in Luxembourg: ECU 10
This research, carried out with the financial help of the European Coal and Steel Community, concerned the fatigue strength of orthotropic steel decks of road bridges. It followed two phases that were concerned with the collection of traffic data and the measurement of stresses produced in bridges. Fatigue tests under constant and variable amplitude were carried out on stiffener-plate connections, stiffener-stiffener connections with U and V shapes, and stiffener cross-beam connections. From the test results and calculations some conclusions can be drawn which are directly usable in bridge design. However, some unexpected behaviour occurred and some connections need further investigation.
il IIIMIIIIIIIIIIKI II
Venta y suscripciones • Salg og abonnem ent • Verkauf und Abonnement • Πωλήσεις και συνδpoμές Sales and subscriptions • Vente et abonnements • Vendita e abbonamenti Verkoop en abonnementen • Venda e assinaturas BELGIQUE / BELGIË Moniteur belge / Belgisch Staatsblad Rue de Louvain 42 / Leuvenseweg 42 1000 Bruxelles / 1000 Brussel Tél. (02) 512 00 26 Fax 511 01 84 CCP / Postrekening 000-2005502-27 Autres distributeurs / Overige verkooppunten Librairie européenne/ Europese Boekhandel Avenue Albert Jonnart 50 / Albert Jonnartlaan 50 1200 Bruxelles / 1200 Brussel Tél. (02) 734 02 81 Fax 735 08 60
Llibreria de la Generalität de Catalunya Rambla deis Estudis . 1 1 8 (Palau Moja) 08002 Barcelona Tel. (93) 302 68 35 302 64 62 Fax 302 12 99
FRANCE Journal officiel Service des publications des Communautés européennes 26. rue D esaix 75727 Paris Cedex 15 Tél. (1) 40 58 75 00 Fax (1) 40 58 75 74
Jean D e Lannoy Avenue du Roi 202 /Koningslaan 202 1060 Bruxelles / 1060 Brussel Tél. (02) 538 51 69 Télex 63220 UNBOOK B Fax (02) 538 08 41 CREDOC Rue de la Montagne 34 / Bergstraat 34 Bte 11 / Bus 11 1000 Bruxelles / 1000 Brussel
DANMARK J. H. Schultz Information A/S EFPublikationer Ottiliavej 18 2500 Valby Tlf. 36 44 22 66 Fax 36 44 01 41 Girokonto 6 00 08 86
IRELAND Government Publications Sales Office Sun Alliance House Molesworth Street Dublin 2 Tel. 71 03 09
Bundesanzeiger Verlag Breite Straße Postfach 10 80 06 5000 Köln 1 Tel. (02 21) 20 29-0 Fernschreiber: ANZEIGER BONN 8 882 595 Fax 20 29 278
GREECE G.C. Eleftheroudakis SA International Bookstore Nikis Street 4 10563 Athens Tel. (01) 322 63 23 Telex 219410 ELEF Fax 323 98 21
YUGOSLAVIA
Imprensa Nacional Casa da Moeda, EP Rua D . Francisco Manuel de Melo. 5 P-1092 Lisboa Codex Tel. (01) 69 34 14
Privredni Vjesnik Bulevar Lenjina 171/XIV 11070 - Beograd Tel. 123 23 40
Distribuidora de Livros Bertrand, Ld. a Grupo Bertrand, SA Rua das Terras dos Vales. 4-A Apartado 37 P-2700 Amadora Codex Tel. (01) 49 59 050 Telex 15798 BERD IS Fax 49 60 255
TURKIYE Pres D agitim Ticaret ve sanayi A.Ş. Narlibahçe Sokak No. 15 Cagaloglu Istanbul Tel. 512 01 90 Telex 23822 D SVO-TR
UNITED KINGD OM H M S O Books (PC 16) HMSO Publications Centre 51 Nine Elms Lane London SW8 5D R Tel. (071) 873 9090 Fax GP3 873 8463 Telex 29 71 138
or by post
Sub-agent:
Government Stationery Office EEC Section 6th floor Bishop Street Dublin 8 Tel. 78 16 66 Fax 78 06 45
Alan Armstrong Ltd 2 Arkwright Road Reading, Berks RG2 OSO Tel. (0734) 75 18 55 Telex 849937 AAALTD G Fax (0734) 75 51 64
AUTRES PAYS OTHER COUNTRIES ANDERE LÄND ER Office des publications officielles des Communautés européennes 2, rue Mercier L-2985 Luxembourg
Tél. 49 92 81 Télex PUBOF LU 1324 b Fax 48 85 73 CC bancaire BIL 8-109/6003/700
CANADA ITALIA Licosa Spa
BR D EUTSCHLAND
PORTUGAL
Via Benedetto Fortini, 120/10 Casella postale 552 50125 Firenze
Tel. (055) 64 54 15 Fax 64 12 57 Telex 570466 LICOSA I CCP 343 509
OSTERREICH
Renouf Publishing Co. Ltd
Manz'sche Verlags und Universitätsbuchhandlung Kohlmarkt 16 1014 Wien Tel. (0222) 531 61-0 Telex 1 1 25 00 BOX A Fax (0222) 531 61-81
Mail orders — Head Office: 1294 Algoma Road Ottawa. Ontario K1B 3W8 Tel. (613) 741 43 33 Fax (61 3) 741 54 39 Telex 0534783
Subagenti: Libreria scientifica Lucio de Biasio AEIOU Via Meravigli. 16 20123 Milano Tel. (02) 80 76 79 Herder Editrice e Libreria Piazza Montecitorio. 117-120 00186 Roma
Tel. (06) 679 46 28/679 53 04 Libreria giuridica Via XII Ottobre. 172/R 16121 Genova Tel. (010) 59 56 93
BTJ Box 200 22100 Lund Tel. (046) 18 00 00 Fax (046) 18 01 25
Ottawa Store: 61 Sparks Street Tel. (613) 238 89 85 Toronto Store: 211 Yonge Street Tel. (416) 363 31 71
UNITED STATES OF AMERICA SCHWEIZ / SUISSE / SVIZZERA OSEC Stampfenbachstraße 85 8035 Zürich Tel. (01) 365 51 51 Fax (01) 365 54 11
UNIPUB 4611-F Assembly D rive Lanham, MD 20706-4391 Tel. Toll Free (800) 274 4888 Fax (301) 459 0056
GRAND-DUCHÉ D E LUXEMBOURG ESPANA Boletín Oficial del Estado Trafalgar, 27 28010 Madrid Tel. (91) 44 82 135
AUSTRALIA Abonnements seulement Subscriptions only Nur für Abonnements
MAGYARORSZAG Agroinform
Messageries Paul Kraus
Központ: Budapest I.. Attila út 93. H-1012
MundiPrensa Libros, S.A. CastellÓ. 37 28001 Madrid Tel. (91) 431 33 99 (Libros) 431 32 22 (Suscripciones) 435 36 37 (D irección) Télex 49370-MPLI-E Fax (91) 575 39 98
1 1. rue Christophe Plantin 2339 Luxembourg
Tél. 499 88 88 Télex 2515
Levèlcím:
Sucursal: Librería Internacional AED OS Consejo de Ciento. 391 08009 Barcelona Tel. (93) 301 86 15 Fax (93) 317 01 41
SDU Overheidsinformatie Externe Fondsen Postbus 20014 2500 EA 's-Gravenhage Tel. (070) 37 89 911 Fax (070) 34 75 778
Fax 499 88 84 44 CCP 49242-63
Budapest, Pf.: 15 H-1253 Tel. 36 (1) 56 82 11 Telex (22) 4717 AGINF H-61
NEDERLAND POLAND Business Foundation ul. Wspólna 1/3 PL-00-529 Warszawa Tel. 48 (22) 21 99 93/21 84 20 Fax 48 (22) 28 05 49
Hunter Publications 58A Gipps Street Collingwood Victoria 3066
JAPAN Kinokuniya Company Ltd 17-7 Shinjuku 3-Chome Shinjuku-ku Tokyo 160-91
Tel. (03) 3439-0121 Journal D epartment PO Box 55 Chitóse
Tokyo 156 Tel. (03) 3439-0124
NOTICE TO THE READER All scientific and technical reports published by the Commission of the European Communities are announced in the monthly periodical 'euro abstracts'. For subscription (1 year: ECU 92) please write to the address below.
o a
o
Price (excluding VAT) in Luxembourg: ECU 10
*
*** * ÕÕ . EUR
OFFICE FOR OFFICIAL PUBLICATIONS OF THE EUROPEAN COMMUNITIES L-2985 Luxembourg
ISBN
TS-ñBb-DSBE-T
||| llllll llllllll IIIINIIIII
