Paper No.
02235
2002
CORROSION
I M P R O V E M E N T S ON DE W A A R D - M I L L I A M S CORROSION PREDICTION AND APPLICATIONS TO CORROSION M A N A G E M E N T
Bert.F.M. Pots, Randy C. John Shell Global Solutions (US) 3333 Highway 6 South Houston, Texas 77082 Ian J.Rippon, Maarten J.J.Simon Thomas, Sergio D. Kapusta Shell Global Solutions International B.V. Badhuisweg 3, 1031 CM Amsterdam The Netherlands Magdy M. Girgis Shell Canada, Ltd., Calgary Canada Tim Whitham Shell Aviation Limited, Wythenshawe UK
ABSTRACT
This paper describes corrosion rate prediction models for the main corrosion mechanisms of carbon steel in Exploration and Production service. The models succeed earlier work by De Waard, Milliams, and Lotz. The paper emphasizes that model accuracy is less of an issue than knowledge of the key corrosivity parameters and the quality of the corrosion control system. Models will be described for the following mechanisms: CO2 corrosion, CO2/H2S corrosion, HES corrosion, organic acid corrosion, oxygen corrosion, and microbiologically-induced corrosion. Application limits will be indicated. A good comparison with high-quality lab data is only possible for the CO2 corrosion mechanism. Computer programs will be described in which the corrosion prediction models are applied for front-end design and facility integrity management. Use of these programs during the lifetime of a facility provides a way of focusing on corrosion control issues and they are therefore essential tools for pro-active corrosion management. Copyright 2002 by NACE International. Requests for permission to publish this manuscript in any form, in part or in whole must be in writing to NACE International, Publications Division, 1440 South Creek Drive, Houston, Texas 77084-4906. The material presented and the views expressed in this paper are solely those of the author(s) and not necessarily endorsed by the Association. Printed in U.S.A.
1 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
INTRODUCTION
The ability to confidently predict the internal corrosion of E&P carbon-steel facilities is essential for both from-end design materials engineering and for managing lifetime integrity through optimum corrosion control. While this paper will describe an improved way of CO2 corrosion rate prediction as a follow-up of the Shell De Waard-Milliams-Lotz school 1'2'3, the paper will also consider other corrosion mechanisms. These other mechanisms are: CO2/H2S corrosion, H2S corrosion, organic acid corrosion, corrosion by other acids (particularly spent stimulation acids), oxygen corrosion, and microbiologicallyinduced corrosion. Prediction models for these mechanisms are expected to help the corrosion risk identification and assessment processes and to find the optimal means of corrosion control and management. For example, too often corrosion inhibition is seen as the solution for avoiding further corrosion damage, but obviously one needs to understand the corrosion mechanism before defining the control means. Another point of attention in the paper will be the link of corrosion prediction models to field application. Such a link is essential and it puts the model in the right perspective. It helps to quantify the requirements of corrosion control, such as, for example, corrosion inhibition availability, gas dehydration system availability or corrosion allowance. As such, the paper gives an overview of our latest position with regard to corrosion prediction and its application to corrosion control management. SWEET CORROSION
Model improvements are described, taking earlier work from the De Waard-Milliams-Lotz prior work as a base line 1'2'3. The quoted references are referred to as the 1991, 1993 and 1995 models, respectively. Different models are needed for Bottom-Of-Line (BOL) or bulk-water corrosion and Top-Of-Line (TOL) or dewing corrosion. In previous models, a factor was introduced in the bare-steel BOL corrosion rate for the prediction of dewing corrosion at stratified wavy flow conditions. This approach was abandoned as it was found that TOL corrosion bears no simple, direct relation with the bare-steel BOL corrosion rate. For TOL CO2 corrosion, the improved approach as described in Ref. 4 is now recommended. This approach assumes that, at scaling conditions, the TOL corrosion rate is controlled by the Fe 2+ super-saturation level that is required in the condensation water to maintain the iron carbonate scale. The main model parameter is the water condensation rate. Good agreement of the model with both lab and field data was obtained. It should be noted that high-quality TOL corrosion lab tests need durations of at least a number of months. The starting point for BOL corrosion rate prediction is the bare-steel corrosion rate, which corrosion rate is multiplied by factors to account for the effects of scale protection, oil wetting, and glycol/methanol presence. CO2 fugacity rather than partial pressure is used as a measure for the CO2 activity. Bare-steel C02 corrosion rate
Following the De Waard-Milliams-Lotz approach a more mechanistic approach for corrosion rate prediction was developed starting in 1995. In this approach, species transport and reactions in the flow and reaction boundary layers near the wall, as well as electrochemical reactions at the wall are considered using first principles. The Limiting Corrosion Rate (LCR) model presented in Ref. 5 forms the basis for describing the species transport and reactions. The LCR model gives the limiting corrosion
2 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
rate as set by the mass transport of corrosive species to the wall and the chemical reactions producing the aggressive species. The extension presented in this paper involves the combination of this LCR model with basic equations of electrochemical reaction kinetics following the mixed-potential theory:
J - kFeCFeeE/bF" + Z
kicie-E/b~ kici e-E/b' Jlim,i
(1)
i=1'21 +
The first term represents the anodic current related to iron dissolution and the second term the total cathodic currem; the temperature-dependem k's are the charge transfer reaction rate constants, the b's the Tafel slopes, c's the concemrations (CFe=I), and jlim'S the limiting currems from the LCR model. Protons (i=1) and carbonic acid (i=2) are assumed to be cathodically active. At the corrosion potential E=Eco~, the anodic and cathodic currents are equal (/'=0), and equal to the corrosion current. The model is referred to as LCR+ model. For the charge transfer kinetics, rule-of-thumb data for iron dissolution (b=40 mV/decade) and proton or carbonic acid reduction (b=120 mV/decade) were taken. It was assumed that charge transfer for protons and carbonic acid are the same. This leaves only one empirical parameter in the model, which is one reaction rate constant k plus its Arrhenius coefficient. Equal anodic and cathodic k constants can be used, as the zero potential is arbitrary. The electrochemical reaction rate constant was determined from glass-cell potentiodynamic curves. Validation of the bare-steel corrosion model with lab data
The new LCR+ CO2 corrosion model was compared with "high-quality lab" data generated at IFE (Institute for Energy Technology, Norway) in the Kjeller Sweet Corrosion projects and at the authors' Amsterdam lab. Only data were considered where scale formation was absent. High quality means in particular that 1) the solution chemistry is controlled, particularly the iron level, and 2) the flow defined. On top of this quality requirement, it was ensured that experimental data points were evenly distributed over the parameter space. Assuming that pCO2, temperature, pH and flow velocity are the main parameters, the 4-dimensional space made up by these parameters was divided in 72 windows or categories, see Table 1. For each category a maximum of 3 experimental data points was allowed. Figure 1 compares the LCR+ model with the lab data. For reference purposes, Figure 2 compares the Shell 1995 model 3 with the same data. Given the fact that the LCR+ model is based on firstprinciples equations with only one empirical parameter, the agreement with the data is remarkably good and the agreement is better than for all our previous models. Additional advantages of the mechanistic approach are that effects of concentrations (including pH), temperature, and flow follow basic physics laws. An interesting observation is that for conditions where the rate-determining step is the hydration of CO2 in the reaction boundary layer, corrosion rates from the LCR+ model come close to those of the De Waard/Milliams nomogram, see Figure 3. The hydration regime prevails for medium flow velocities (-~ 1 m/s) and buffered conditions (iron-saturation). For other conditions (other velocities and/or other pH's), deviations between the two models can be considerable and the nomogram should not be used. From a scientific point of view, the above-described mechanistic modeling is based on a number of large simplifications. For example, hydration in a cementite network (if present) is ignored and the
3 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
actual electrochemical kinetics are far more complicated than assumed. Nevertheless, the LCR+ model is seen as a good compromise. As will be later explained in this paper, uncertainties in key parameters, including operational ones, are the accuracy-determining factor, not the model.
Corrosion protection by corrosion product scale In previous modeling, corrosion protection by a corrosion product scale was assumed to be present for temperatures above the scaling temperature. In 1996, a paper 6 demonstrated that the main parameter, for a protective scale to form, is the ratio of the iron carbonate precipitation rate and the corrosion rate (scaling tendency). Not only temperature was shown to be important for scale formation, but also pH, flow, and Fe 2+ level. Stable and protective scales can form at temperatures well below the old scaling temperature, which scales can be difficult to remove mechanically. On the other hand, scale protection may not be present at temperatures well above the old scaling temperature, particularly for high flow, low pH and low Fe 2+ level. A full description of the scaling tendency model can be found in Ref. 6, whereas additional information can be found in Ref. 7. The corrosion rate when a corrosion product is present follows this dependence" CR oc k[Fe 2+ ]supersat
(2)
where Fe 2+ supersat is the iron super-saturation level at the wall and k mass transfer coefficient. Taking credit for scale protection for cases with formation water carries a risk, as porous mixture scales may form with little protection or scale disturbers may be present. Therefore, no credit should be taken for scale protection for conditions with formation water.
Oil wetting Credit can be taken for the entrainment of water by the oil phase or wetting of the steel wall by the oil phase. While in earlier work a critical flow velocity for full water entrainment was used of 1 m/s, field measurements of the water distribution in Main Oil Lines (MOL) in Oman in 1994 showed that it is safer to assume a higher critical velocity of 1.5 m/s. Full protection (fou=0) applies to water cuts below 40%. Partial protection applies to water cuts above 40% (fo~t = water cut). However, below a flow velocity of 1.5 m/s, no credit should be taken for oil wetting as one has to assume that water will separate (fo~t=l). No a-priori credit can be taken for protection by hydrocarbon condensate. Condensates require considerably higher flow velocities before full water entrainment occurs. It was found that in flow loop tests, even at flow velocities as high as 2.5 m/s, there might still be a free water phase at the bottom of the line. The above rules of thumb are conservative. The actual behavior of oil-water mixtures depends on a large number of physico-chemical properties and this behavior is impossible to predict without carrying out lab or field tests with the actual oil-water mixture under the precise conditions of interest.
Water sweep-out Sweep out of water from low points was found from flow loop tests to correlate with the Froude number
4 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Fr=I
Po VL ~ 0.65 ApgD
(3)
where Po represents the oil density, Ap the oil-water density difference, g the acceleration due to gravity, D the hydraulic diameter of the liquid (oil and water) flow field, and VL the liquid flow velocity. From theoretical work, it was found that the Froude number is also a measure for water dispersion, the critical Froude number for full water entrainment being Fr~2. Alcohol factor
Glycol or methanol when applied for hydrate prevention purposes simultaneously work as corrosion rate reducer. They can lower the un-mitigated corrosion rate by an order of magnitude, see Ref. 3 for the alcohol factor equation. MIXED SWEET/SOUR CORROSION
In the 1990's, concerns by a number of subsidiaries with respect to the corrosivity of slightly sour conditions initiated an R&D program in which the effect of H2S was investigated. Autoclave experiments were carried out in the mixed sweet/sour regime, for partial pressures ratios of CO2 and H2S between 20 and 500, see Figure 5. Characteristic for this regime is the formation of a mixed scale of iron carbonate and iron sulfide. After elaborate testing it was found that the highest pitting rate is never worse than the sweet corrosion rate. Therefore, application of a sweet corrosion model for the mixed sweet/sour corrosion regime appears justified, though it may be conservative.
SOUR CORROSION
In the sour corrosion regime (pCO2/pH2S<20), field corrosion rates as high as 20 to 30 mm/y have been seen in our fields, notably in Canada. Prompted by these high corrosion rates, a start was made with the development of a sour corrosion model. For its development, tests are being carried out in autoclaves (~200 ml) at various levels of CO2 and H2S (1 to 10 bar), various levels of chlorides (1000 to 100000 ppm), with and without elemental sulfur. Figure 6 shows the type of corrosion coupon used. While a magnetic stirrer is used to generate flow, the flow velocity is limited to very low values (<1 cm/s) to avoid the sulfur starting to float around and losing contact with the coupon. The low flow is also believed to simulate worst case conditions. Based on the experiments carried out so far, the following preliminary model was proposed:
CR - Fp CR~wee,
(4)
where Fp is the pitting factor. The pitting factor is a function of elemental sulfur presence and chloride level in the water, see Table 2. As oxygen can lead to the production of free, elemental sulfur in sour service, presence of oxygen is interpreted as having elemental sulfur present. In the sour regime, the various factors and effects require re-consideration. An overview of our current suggestions is given in Table 3.
5 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Top-Of-Line corrosion Currently no TOL corrosion model is available for the sour regime. Use of the sweet corrosion model is seen as a conservative approach, assuming there are no agents that can destroy the TOL iron sulfide scale.
Scale factor The slowing of corrosion by iron sulfide scale poses the greatest challenge in sour corrosion modeling. Formation of an iron sulfide scale is one of the main characteristics of sour corrosion, even at relatively low temperatures and explains the low field corrosion rates often seen. Corrosion in the field relates to failure of the iron sulfide scale, rather than the absence of scale. However, no method is currently available to distinguish between protective and non-protective situations. Obviously, the CO2 corrosion scale factor for iron carbonate is of no use in the sour corrosion regime.
Comparison with data Figure 4 shows a comparison of the sour corrosion model with our lab data. Similar and worse results were obtained with other (commercial) sour corrosion models demonstrating that considerable further work is required before an acceptable sour corrosion model will be available. The current models are far too conservative. The scatter in Figure 4 is explained by differences in scale protection. The protection will differ between tests, even for the same test conditions. In most tests, the scale will slow corrosion to levels well below the bare-steel (CO2) corrosion rate. Only for cases with a strong disturbance of the scale by free sulfur and/or chlorides, will corrosion rates be close to or higher than the bare-steel (CO2) corrosion rates. Another shortcoming of the current sour corrosion modeling is that the sweet corrosion rate is taken as the reference whereas H2S and ~ee sulfur are both corrosive towards carbon steel without requiring other corrodents. This issue needs further attention. The current model does not work for systems containing no CO2.
ORGANIC ACID CORROSION Two field cases of organic acid corrosion damage in NAM in the Netherlands initiated the modeling of organic acid corrosion. The first field case (1994) concerned the failure of a carbon steel inlet nozzle of an offshore flash vessel on a sweet natural gas platform. The nozzle failed in less than two years, which corresponds to a corrosion rate of more than 3 mm/y. The partial pressure of CO2 was almost zero. The second field case (1998) concerned the failure (leak) of a water/condensate drainpipe in an onshore sweet natural gas plant. Failure occurred al~er one year of service, which corresponds to a corrosion rate of about 5 mm/y. In both cases significant amounts of organic acids were present (700 ppm and 150 pprn, respectively). The currem model for organic acid corrosion is a limiting corrosion rate model similar to the CO2 corrosion model and assumes the rate-determining steps are the diffusion of un-dissociated organic acid and protons. Both contributions were needed in order to explain the field cases. Application of the model to other situations often results in excessive corrosion rates. Work is in progress to clarify this issue.
6 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
CORROSION BY OTHER ACIDS
Spent stimulation acids are another threat to E&P carbon steel facilities. Corrosion rates can be predicted with the "proton part" of the CO2 corrosion model, assuming the pH is known. OXYGEN CORROSION
The current oxygen corrosion model is based on assuming that the rate-determining step is the diffusion of oxygen over the diffusion boundary layer. Standard Sherwood mass transfer calculations are carried out to calculate the corrosion rate. M I C R O - B I O L O G I C A L L Y - I N D U C E D CORROSION
In the early 1990's, one of the authors began to develop an approach to assist ranking of oil transport pipelines for susceptibility to micro-biologically influenced corrosion (MIC). The outcome of this work was a "predictive flowchart" where the susceptibility of a pipeline could be assessed on the basis of details of water chemistry and pipeline operational parameters. MIC of mild steel manifests as a localized, pitting form of attack, following the development of a surface biofilm. The anaerobic environments of oil transport pipelines often support the growth of biofilms, which almost invariably contain sulfate-reducing bacteria (SRB), a major cause of MIC. SRBcontaining biofilms generate H2S, which precipitates iron sulfide within the biofilm. These sulfides are cathodic to the steel and may greatly exacerbate corrosion at anodic sites. The physico-chemical and biological interactions, which take place between the biofilm and the environment in a mild steel pipeline, are very complex. They may be interdependent on other processes not directly related to biological activity. This makes modeling the suite of corrosion phenomena that constitute MIC an extremely difficult task. Prediction of pipe wall penetration rates attributed to MICinduced localized corrosion is unreliable, because of the uncertainty of the onset of pitting and whether it proceeds at a constant rate. Nevertheless, the original predictive flow chart was translated into a MIC corrosion rate predictor, realizing that corrosion rates can be uncertain. Corrosion rate values are very approximate. The main idea is to link the risk of MIC and the influencing parameters. Predicting the susceptibility of a pipeline to MIC is facilitated by considering 1) whether the pipeline environment can support microbial activity and biofilm formation, and 2) how pipeline operational parameters affect the microbiology. The premise is that if an active biofilm can form inside a pipeline, then that line may suffer from biofilm-associated corrosion. The MIC corrosion rate calculation is based on the following equation
(5)
CR =C x F p
with
F=f,
xLx
(6)
...... .fo
where C is a constant (C=2 mm/y), the f s are factors for the various influencing parameters, and p is a power law index (0.57). An overview of the factor values is given in Table 4.
7 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Influencing parameters The following parameters influence MIC: Presence of water: if the crude and pipeline are free of water, there will be no MIC. Water wetting: MIC is only to be considered at separated flow conditions; if all water is entrained in the oil, the risk of MIC is significantly reduced. pH of water: SRB growth is largely confined between pH's of 5 and 9.5. Salinity or total dissolved solids (TDS): most SRB will grow best at < 60 g/1TDS. SRB may still grow if TDS > 60 g/l, but this must be confirmed by growth in a nutrient medium with the same TDS concentration. Temperature: different types of SRB can grow at temperatures between 4 and 110 °C. If the temperature is above 45 °C, it is necessary to confirm that SRB isolated from the pipeline can grow at the pipeline temperature. Microbial activity is generally limited by extremes of these parameters. Although certain microorganisms can grow outside the normal limits, they are usually confined to extreme habitats. The limits for pH and temperature are well established for bacteria that could occur in oil transport pipelines, but tolerances for salinity are less well understood. The important consideration with these parameters is whether bacteria isolated from a system can grow under the prevailing conditions. For example, in a pipeline carrying high salinity produced fluids, it may be possible to isolate bacteria using standard isolation media, but these media may have a considerably lower salinity than the water phase in the pipeline. There may be no significant bacterial growth at pipeline conditions and therefore no significant impact with respect to MIC. Demonstration of growth at pipeline conditions is therefore an essential part of the assessment of susceptibility to MIC.
Impact of nutrient status of water phase The following nutrients have an influence on the corrosion rate: Sulfate ion concentration: SRB growth is severely reduced at sulfate levels < 10 mg/1 unless an altemative electron acceptor is present. Total carbon (C) from fatty acids such as formate, acetate, propionate, & butyrate: SRB growth is severely restricted if utilizable carbon is < 20 mg/1. Nitrogen (as utilizable N): SRB growth is severely restricted if utilizable nitrogen is less than 5 mg/1 Critical C:N ratio: SRB growth is the most prominent if C:N <10. Nutrient availability directly influences microbial activity and biofilm formation both qualitatively and quantitatively. Carbon and nitrogen are the two of the most important nutrients, which can be readily quantified. Sulfate is also important for SRB. It is the principal respiratory electron acceptor, which can ultimately have a significant influence on MIC. There are threshold concentrations below which these nutrients cannot be effectively utilized by microorganisms. When one of the nutrients essential for microbial activity falls below the threshold, this activity ceases until more nutrients become available. It should be noted that a water analysis indicating a nutrient deficiency does not mean that no MIC can take place, if previously nutrients were available. MIC associated with a biofilm that was formed when nutrients were available, but which subsequently faces a nutrient shortage, may continue for some time.
8 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Additionally, the ratio of carbon to nitrogen is important. A ratio of carbon to nitrogen of approximately 10" 1 is required to sustain microbial growth. The ratio will indicate the limiting nutrient, useful information when looking at options for controlling microbial activity (e.g. a production chemical could change the ratio in such a way as to encourage microbial activity). Some bacteria can survive and even grow in the complete absence of sulfate, they simply use an alternative electron acceptor, such as nitrate. This concept has been utilized to prevent reservoir souring from water floods. However, this form of metabolism is unlikely to be significant in oil transport pipelines. Apart from sulfate, other oxidized sulfur species may be utilized, such as thio-sulfate or bisulfite. Hence if there are sources of these species present in the water, then they may represent an important increase in the potential for sulfide generation if the level of sulfate is very low. Note that the threshold values given for nutrients are based on rather few data. These values should be substantiated with more experimental data to obtain a higher level of confidence since oil field waters often contain key nutrients close to threshold limits.
Impact of flow rate Flow rate will directly influence the nature ofbiofilm formation and the rate of nutrient delivery. As flow rate increases, biofilms become less bulky and more adherent. However, above a certain threshold, the initiation of biofilm formation is significantly limited. The transition zone is considered to be between 2 and 3 m/s. At the other extreme, stagnation is often associated with the severest MIC incidents. A pipeline's flow history could be important in determining current status of the line and subsequent risk of MIC.
Impact of debris on the bottom of the pipeline Presence of debris exacerbates MIC, because the presence of debris can create stagnant conditions in the bottom of low velocity lines, where biofilms can be established.
Impact of pigging frequency Pigging can potentially have a significant effect on biofilm and MIC. The more frequent the pig runs, the less time biofilm has to recover. A pigging frequency of at least once every two weeks will be very effective in controlling biofilm. No detailed data are available on different pigging intervals versus biofilm recovery rates or MIC, although there is some published experience of the impact of pigging as a treatment for MIC. The extent to which pigging is effective in controlling biofilm-associated corrosion depends on how well a corrosion process was established prior to the pigging program. Since pigging has a significant impact on the occurrence of MIC, the frequency should take a prominent place in the prediction module.
Impact of oxygen ingress The highest biofilm-associated corrosion rates are seen when oxygen periodically accesses the sulfide-rich, anaerobic region within SRB biofilms.
9 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Impact of a biocide treatment Biocides are used to control unwanted microbial activity. If a biocide is used in a pipeline, the risk of a failure due to MIC should be significantly reduced, or even eliminated. It is essential that the biocide is selected carefully and the treatment regime is appropriate and applied properly and consistently. For practical purposes, the inclusion of biocides as a parameter in a predictive scheme would need to be based on the assumption that the correct biocide has been selected and applied in such a way that it can function as intended. With this in mind, it is possible to make some simple rules regarding the type of biocide and the application dose and frequency for different types of water chemistries, watercuts etc.
Impact of operational history of pipeline The following operational history parameters have an influence on MIC" •
Operational age of pipeline in years
•
Total operational downtime:
Long periods of stagnation promote MIC.
Other parameters Other parameters may also play a role in MIC. So far, these parameters have not been included in the model. An example is production chemicals. Production chemicals can impact MIC in a number of ways. They may be a source of nutrients to stimulate microbial activity. They may interact with and reduce the efficacy of biocides being used to treat a bio-fouled system. Corrosion inhibitors could act to reduce MIC in some circumstances. Previous studies have shown that production chemicals can indeed affect the performance of some biocides, and vice versa. Phosphorus and nitrogen-containing chemicals have been shown to act as nutrient sources. Since production chemicals are often used continuously in oil transport pipelines, this parameter needs to be addressed in a future predictive module.
LINK TO FIELD APPLICATION A link between the above corrosion rate prediction models and field application can be achieved using the following "corrosion control availability" equation7'9'1°: C R = F x C R m + (1 - F) x CR,,
(7)
where F is the availability of the corrosion control system (fraction of time in a year that the system is complying with all its targets), CRm the residual corrosion rate at mitigated conditions, and CR,, the unmitigated corrosion rate obtained from the corrosion rate prediction model. The equation expresses that the field corrosion rate will be the average of the mitigated corrosion rate over the periods that the corrosion control system is available (up and working as expected) and the un-mitigated corrosion rate over the periods where there is no compliance with the corrosion control targets. Experience is that the industry generally has not determined the corrosion control availability, i.e. ignoring how good (or bad) corrosion control is or has been. Without this knowledge it is impossible to
10 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
predict a field corrosion rate. The information about the quality of corrosion control has great value and can for example be used to optimize corrosion control or inspection planning. This paper does not address how to derive the availability for the various service types (e.g. oil/gas multiphase or separated oil) and the various types of corrosion control (e.g. corrosion inhibition or gas drying). For inhibited systems, one could use for example information such as inhibitor pump availability, injection rates, and residuals versus target to prepare an estimate of the corrosion inhibition availability. For dried system, information such as dew points and free water received after pigging are useful input. For systems that need corrosion control, the availability equation shows that the accuracy of the corrosion rate prediction depends both on the value of the availability and the model corrosion rate. Another source of uncertainty is the accuracy and/or unavailability of the input parameters for the model calculations. Often the uncertainty in the availability controls the overall accuracy. This makes further refinement of model prediction to be of a lower priority than obtaining good data for the availability and other key parameters.
COMPUTER PROGAM FOR CORROSION RATE PREDICTION The above-described models are incorporated in the computer program HYDROCOR. The program quantifies the corrosivity of the operating conditions associated with the transportation of wet hydrocarbons in carbon steel pipelines. An example input/output screen for the CO2 corrosion mechanism is shown in Figure 7. With the program the operational conditions can be identified under which a corrosive, wet hydrocarbon system can be transported in a pipeline of carbon steel. The program checks for the presence of free water considering the presence of water vapor in the gas phase, dissolved water in the hydrocarbon liquid phase, and any free water entering either from production (formation, condensation) or injection (alcohol). Based on whether free water is comacting the pipeline, corrosion rates are calculated along the pipeline profile. This calculation requires a number of additional models, which are also incorporated in the program. Multi-phase flow A multi-phase flow model evaluates the main characteristics of the flow, such as flow pattern, pressure gradient, liquid and water holdups, phase velocities, hydraulic diameters, and wetted perimeter, etc. The generic flow pattern is slug flow, comprising of a slug part and a film part. In the limit of no slug part, the (infinite) film can represent stratified wavy or annular dispersed flow. In the limit of no film, the infinite slug part can represent dispersed bubble flow or single-phase liquid flow. Corrosion rates are evaluated for both slug and film parts for both bottom and top of the pipe and time-averaged. Optionally, the elevation profile can be considered. Alcohol/water phase distribution In wet gas pipelines, any alcohol injected will be diluted with condensation water. Alcohol and water mass balances for each pipe section, taking into account evaporation and condensation processes, simulate this dilution process. Water/alcohol condensation A condensation model is included to calculate the condensation rate of water and alcohol at the wall, which is an important parameter for the TOL corrosion model.
11 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
Temperature gradient A heat transfer model is included for calculating the temperature profile along the pipeline.
Water chemistry A water chemistry model is included for calculating the pH, iron super-saturation, and iron carbonate precipitation. Based on the iron mass balance, the change of dissolved iron concentration and related parameters (pH) are calculated. The above computer program is being extensively used throughout the company of the authors. Many corrosion engineers have been trained to use the program. Such training is essential as results must be interpreted and applied properly. With the additional training, the computer program was found to be very helpful in the evaluation of design options and improves the confidence in the understanding of corrosion mechanisms. As such it is not a stand-alone tool, but should be seen as an integral part of the overall systems approach. The impact of our computer program on project design in terms of savings that were achieved is impressive and rtm in the multi-million US Dollars range for many individual projects ~. These savings can be credited in part to the availability of this versatile tool.
PIPELINE RISK-BASED INTEGRITY ASSESSMENT The standalone version of the above-described corrosion prediction program has been applied mainly to new designs. It has also been incorporated in a pipeline risk-based integrity assessment program, PIPE-RBA. This provides the opportunity to assess the integrity status of existing pipelines, improve inspection planning, and create a feedback loop for improved corrosion control. This program integrates corrosion rate prediction with defect assessment, risk-based principles, and data management. A more extensive description can be found in Ref. 8.
REFERENCES
C. de Waard, U. Lotz, and D.E. Milliams, "Predictive model for CO2 corrosion engineering in wet natural gas pipelines, CORROSION/91, paper no. 577, (Houston, TX: NACE International, 1991) C. de Waard and U. Lotz, "Prediction of CO2 corrosion of carbon steel", CORROSION/93, paper no. 69, (Houston, TX: NACE International, 1993) C. de Waard, U. Lotz, and Ame Dugstad. "Influence of liquid flow velocity on CO2 corrosion; a semi-empirical model", CORROSION/95, paper no. 28, (Houston, TX: NACE Intemational, 1995) B.F.M. Pots and E.L.J.A. Hendriksen, "CO2 corrosion under scaling conditions- The special case of Top-Of-Line corrosion in wet gas pipelines", CORROSION/2000, paper no. 31, (Houston, TX: NACE International, 2000) B.F.M. Pots, "Mechanistic models for the prediction of CO2 corrosion rates under multi-phase flow conditions", CORROSION/95, paper 137, (Houston, TX: NACE International, 1995) E.W.J. van Hunnik, B.F.M. Pots, and E.L.J.A. Hendriksen, "The formation of protective FeCO3 corrosion product layers in CO2 corrosion", CORROSION/96, paper no. 6, (Houston, TX: NACE International, 1996)
12 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
S.D. Kapusta, B.F.M. Pots, and R.A. Connell, "Corrosion management of wet gas pipelines", CORROSION/99, paper no. 45, (Houston, TX: NACE International, 1999) M. Simon Thomas et al., "Deterministic pipeline integrity assessment to optimize corrosion control and reduce cost", CORROSION/2002, paper no. 02075 (Houston, TX: NACE International, 2000) B. Hedges, D. Paisley and R.Wollam, "The Corrosion Inhibitor Availability Model", CORROSION/2000, paper no. 34 (Houston, TX: NACE International, 2000) 10 I.J.Rippon, "Carbon Steel Pipeline Corrosion Engineering: Life Cycle Approach", CORROSION/2001, paper no. 55 (Houston, TX: NACE International, 200 l) 11 I.J.Rippon and M.J.J.Simon Thomas, "Selection of Corrosion Control Options to Optimise Production Field Development", CORROSION/2002, paper no. 02278 (Houston, TX: NACE International, 2002)
13 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
TABLE 1 DIVISION OF PARAMETER SPACE pCO2 Temperature pH Flow velocity
<0.2 bar <20 °C Un-buffered <0.2 m/s
>0.2 bar and < 5 bar >20 °C and <60 °C Buffered >0.2 m/s and <5 m/s
>5 bar >60 °C
>100 °C
>5 m/s
TABLE 2 PITTING FACTOR Fp USED IN SOUR CORROSION MODEL. F~
Chlorides CI-< 500 ppm 500 ppm < C l <5000 ppm C1- > 25000 ppm
Sx0 0.73 1.1 2.6
Sx=l 1.7 3.8 6.1
TABLE 3 EFFECTS AND/OR FACTORS CONSIDERED IN CO2 AND H2S CORROSION REGIMES Effect / factor Top-Of-Line (TOL) corrosion Fugacity Glycol no 02 Methano 1 no 02 Glycol + 02 Methanol + 02
C02 corrosion regime Yes
Scale
Yes, F
Presence oil Presence condensate Oxygen presence Failures of scale or inhibitor film
Yes, F
H2S corrosion regime Use CO2 TOL corrosion model No (F=I) No (F=I) No (F = 1) No ( F - l )
No (F=I) Treat O2 like free S No (F=I) Same as for C02 corrosion Same as for CO2 corrosion Treat like free S Same as for C02 corrosion
14 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
TABLE 4 OVERVIEW OF FACTORS FOR VARIOUS INFLUENCING PARAMETERS IN MIC
pH between 5 and 9.5? Total Dissolved Solids (TDS) <60 g/l? If TDS>60 g/l, do SRB grow? Temperature (T) between 10 and 45 °C? If T>45 °C, do SRB grow? Sulfate > 10 mg/l? Total Carbon (C) from fatty acids >20 mg/l? Nitrogen (as utilizable N) > 5 mg/l? C:N ratio <1 O? Flow velocity < 1 m/s Flow velocity = 2 m/s Flow velocity = 2.5 m/s Flow velocity = 3 m/s Debris on bottom of pipeline Pigging frequency never Pigging frequency 13 wks Pigging frequency 4 wks Pigging frequency 1 wk Prolonged oxygen ingress > 50 ppb Biocide routinely used? Operational history: • age pipeline < 0.5 yr • age pipeline > 0.5 yr & downtime--1 wk • age pipeline > 0.5 yr & downtime - 50 wks
I !
Factor when true 1 1 0.2
Factor when false 0.001 0.2 0.0001 0.2 0.2 0.2 0.2 0.2 0.4
~0.6 ~0.1 ~0.01
~0.3 ~0.001 ~0.0001 0.2 1
2
15 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
1000 t., /
~/
i
f
~F"
i-
j / ; z "
" ""
100
,t" 1
I
i //
O
/
,--
O L_
"
10
L_
it./
ill
O 0 "10 0
t
" w i / t 'II.l
"
0
f
-I
¢
/
,,hi i
J"
// .-"
-- i I"
li
.I
• • •
L_
/
//
_ivy/
/ " I t ~ . i,. ,q f
i
/
/
IFE fit or 1995 model I I I I III
0.1 O. 1
1
10
1 O0
1000
Measured corrosion rate, mm/y
Comparison of predicted CO2 corrosion rates of 1995 model (IFE fit) with measured laboratory data. Upper and lower lines refer to +50% and - 33% deviation. FIGURE
1 -
1000 .~/
/
/ j-
s
• f
i-
"--
E E
100
.. ll
_-.7
j
L_
e~ , -O O L_
~= ~,, am /.-, ~IP /
.,,
L_
,,,
• ,
"0 /
L_
1 i
2
t/ I ~ J /
/
Fr-
" / , u~"
~
"T J,
l ,
///
il
10
O 0 "0
a.
,L
It,/
,(
IF/
/'--
jr
AF
~
mS 0
o
I
/
/
o f
.i
i-
.," /"
II
tI"
..-"
m'
"
i
/.~U>S /
j j
/
" / S
9 model o.1 0.1
1
10
100
1000
Measured corrosion rate, mmly
Comparison of predicted CO2 corrosion rates of LCR+ model with measured laboratory data. Upper and lower lines refer to +50% and - 33% deviation. FIGURE 2-
16 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
100 pC02=10bar
~'
10
Y
S 8
..........~ ~ ~
Y ~
0.1
PC02=1bar pCO2=0.1bar
l pC02=0.01bar
0.01 0
25
50
100
75
Temperature, °C
FIGURE 3 - De Waard-Milliams nomogram (1991/1993) corrosion rate (closed markers) and LCR+ model (open markers) corrosion rates as function of temperature for various CO2 partial pressures for iron-saturated solution at flow velocity of 1 m/s (hydration dominated regime)
100
E
10
E ....-
"o o
E
n~ O O <~
1
O
00
0.1 0.1
1
10
100
S C A N lab e x p e r i m e n t s , m m l y
FIGURE 4 - Comparison of sour corrosion model with autoclave lab corrosion rates
17 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
/ PH2S
REGIME / SOUR /
H2S
CO 2 + H2S REGIME
/
C£MIXED
~ I pOO~pH= s=500 [
00 2 REGI?E
pCO2
FIGURE 5 - Corrosionregknesin CO2/H2Scorrosion.The CO2 corrosionmodelcoversthe CO2 reg~e and mixed CO2+H2S regime. The sour model aims at corrosion rate prediction in the H2S regime (pCO2/pH2S<20).
~k •
dt
L
C~
~: •
~
/ii:iiii i:i:~i:~i,:~~, ~,~I¸~:~i~i ,
FIGURE 6 - Corrosion test coupon used for sour corrosion testing, showing sulfur deposited on coupon prior to mounting into autoclave
18 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.
P I P E L I N E S Vemiolt ~.Ol ID Imperialunits I["1 Expertmode i!-"1Elevationprofile MULTCASE.XLS • 36" wet gas pipeline Norway PIPELINE MAIN OUTPUT" for LCR.99 0.25 Flow pattern Stratified wavy Pipe inside diameter m 0.854 Max BOL corrosion "mm/y Pipeline length km 70 Max TOL corrosion mm/y o.ot Iiuquiaho,aup " 0.74 OPERATING CONDITIONS Maximum occurs at m 127 IIMixturevelocity Its 8.47 1.00 IIFilm velocity Its :2.70 Pressure inlet bar 110 Scale protection factJ 50 Oil protection factor 1.00 llRelative film length 1.00 Temperature inlet °C 0.11 llWatercut 14.7% Ambient temperature" °C 7 Glycol protection factor CO2 content gas mole% 0.37 Scaletinhibitor failure 0 IIpH 4.57 HzS content gas mole% 0 - 7 0.3 Gas flow rate -rain Sm3/d 45 - 6 0.25 Liquid HC type .... - 5 NGUoil flow rate m3/d 840 0.2 Water flow rate m3td 130 - 4 WATER ANALYSIS 0.15 - 3 Water chemistry No formationwater V 0.1 Bicarbonates ppm 0 - 2 Organic acids ppm 0 - I 0.06 t"
M~.ILTI-PHASEGAS/OIL
C O R R O S I O N in Database: Case
~11
,
,
mo,,,il II
i
CORROSION CONTROL
0 Annual wet operation % of time 100 0 10000 20000 30000 40000 50000 60000 70000 80000 Inhibitor availability % 0 --e-Distance, m Alcohol type " I MEG iv -IpH Alcohol concentrabon w% 85 Corrosion rate bottom, mm/'y F I G U R E 7 - Input/output screen for CO2 corrosion m e c h a n i s m I
I
I
I
I
I
I
TM
•
•
19 Sajid Aslam - Invoice INV-673395-6ZVD2M, downloaded on 6/11/2013 5:16:07 AM - Single-user license only, copying and networking prohibited.