Rolf Nyborg and Arne Dugstad Institute for Energy Technology P.O. Box 40, N-2027 Kjeller, Norway

Paper No. 09283 2009 Top of Line Corrosion with High CO 2 and Traces of H 2 S Rolf Nyborg and Arne Dugstad Institute for Energy Technology P.O. Box 40, N-2027 Kjeller, Norway rolf.nyborg@ife.no Tim G. Martin (Ret.) ExxonMobil Development Company 12450 Greenspoint Drive Houston, TX 77060, USA ABSTRACT Top of line corrosion has been studied in a flow loop where moist gas is circulated and water is condensed on 1.8 m long carbon steel pipes with external cooling. The condensed water is collected at the end of each test section. This setup enables measurement of top of line corrosion on large surfaces with low water condensation rates. An experiment with high CO 2 partial pressure and traces of H 2 S in the gas showed that most of the iron dissolved by corrosion precipitated as a porous iron sulfide film which did not offer good corrosion protection. Thin, protective iron carbonate films were formed close to the steel surface. This showed that the presence of even small amounts of H 2 S can change the top of line corrosion mechanism considerably compared to pure sweet conditions. Keywords: CO 2 corrosion, H 2 S corrosion, top of line corrosion, condensation, gas pipeline INTRODUCTION Top of line corrosion can occur when water condenses in the upper part of wet gas pipelines. The condensing water will have a low ph and high corrosivity, but becomes rapidly saturated with corrosion products, leading to increased ph and possible formation of protective corrosion product films. The water chemistry in the thin film of condensed water in the top of the pipeline can be very different from the bulk water phase in the bottom of the line. Laboratory studies of top of line corrosion were performed already around 1990 1, 2, and several cases of top of line corrosion in gas Copyright 2009 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, Copyright Division, 1440 South creek Drive, Houston, Texas 777084. The material presented and the views expressed in this paper are solely those of the author(s) and are not necessarily endorsed by the Association. Printed in the U.S.A. 1

pipelines were reported a few years later 3, 4. Common factors for these cases were excessive cooling of the gas, high water condensation rates and presence of organic acid in the gas, with 300 to 2000 ppm acetic acid in the produced water. The acetic acid is transported in the gas phase and will condense together with the water and increase the solubility of iron and the top of line corrosion. Water condensing in the top of a wet gas pipeline will form small droplets or a thin film on the steel surface. The condensed water can become rapidly supersaturated with corrosion products, resulting in increased ph and iron carbonate film formation. The top of line corrosion rate then becomes dependent on the water condensation rate and the amount of iron which can be dissolved in the condensing water 1, 5. Several top of line corrosion models have been developed based on this dependence of water condensation rate and iron carbonate supersaturation 4, 5, 6. The presence of acetic acid in the gas may increase the top of line corrosion rate, as it increases the amount of iron which can be dissolved in the condensing water. Top of the line corrosion models have also been developed from a more detailed mechanistic viewpoint 7, 8. For sweet systems the top of line corrosion (TLC) rate can be estimated from the water condensation rate and the concentration of iron in the condensed water. The condensing water will have a low ph and high corrosivity. However, when the water condensation rate is low, the corrosion products accumulate rapidly in the condensed water and increase the ph until the water is saturated with respect to iron carbonate. When the water becomes supersaturated with iron carbonate, precipitation of protective iron carbonate films is possible. The top of line corrosion rate will in most cases be controlled by the ability to transport dissolved iron away from the surface with the condensed water. For systems with even small amounts of H 2 S the top of line corrosion models are not directly applicable, as corrosion product films will be dominated by iron sulfide instead of iron carbonate. The mechanisms for supersaturation and protective film formation are very different from a situation dominated by iron carbonate. Due to this an experimental project was initiated in order to study top of line corrosion under conditions typical for a planned gas pipeline with 10 bar CO 2 and 2 mbar H 2 S. FLOW LOOP EXPERIMENTS FOR TOP OF LINE CORROSION STUDIES A flow loop designed for top of line corrosion studies was used to study the effect of small amounts of H 2 S on top of line corrosion. A schematic drawing of the top of line corrosion loop system is shown in Figure 1 and 2. A photograph of the loop is shown in Figure 3. Moist gas is circulated from tank 2 (high pressure reservoir), through the test sections and into tank 1 (low pressure reservoir) where a water/gas ejector system pumps the gas back to the high-pressure reservoir, tank 2. In this way the gas is circulated using only a liquid pump and an ejector, without the need to use compressors. This minimizes the risk for leakages and oxygen ingress in the loop. Both tanks are half filled with a water phase, but only the gas phase is circulated in the loop. The gas is thoroughly mixed with the liquid in the ejector, ensuring that the gas which is fed into the test sections is saturated with water vapor. Three 1.8 m long thick walled carbon steel pipes with 55 mm inner diameter are exposed to the moist gas. This gives a large carbon steel surface where condensation can occur. The pipes were 2

sand blasted before the exposure to remove mill scale and ensure a clean steel surface prior to the experiment. The configuration of the test pipes is shown in Figure 2. The test pipes are cooled by water running through cooling coils mounted on the outside of the test sections, and the condensation rate can be controlled by varying the flow through the cooling coils. The corrosive condensed water collects in the bottom of the test pipe. In order to avoid corrosion in the bottom of the pipe, the bottom segment of the test pipe is painted. The paint covers about 20 % of the surface. At the inlet and outlet of the test pipe, the condensed water is taken out in small cyclone separators. The cyclones are drained periodically, and this gives a measure of the amount of water condensed in each test section between each draining. This is then used to calculate the water condensation rate in each test section. Different condensation rates can be achieved in each test section. This is done by first estimating the necessary temperature drop across a section, then varying the flow of the cooling medium to achieve the estimated temperature drop. The volume of condensed water removed from the loop is then used to calculate the actual condensation rate. The cooling rate is fine tuned during the first couple of days to achieve the desired water condensation rate by sampling over defined periods of time. The iron concentration in the condensed water samples is measured with a photometric technique. Small amounts of H 2 S is added as a sodium sulfide solution to the tanks in the loop, and the H 2 S content in the loop is measured by sulfide analyses of the water in the tanks in the loop. Sulfide is added periodically to replace sulfide consumption, based on frequent measurements of the sulfide content in the loop water. Acetic acid can also be added as an acetate solution to the tanks in the loop, and the total acetate/acetic acid content is measured by titration. The H 2 S and acetic acid content in the circulating gas is calculated with an in-house solubility program that takes temperature, CO 2 and H 2 S partial pressure, ionic strength and dissolved iron carbonate into account. The test pipes are inspected visually before and after exposure and the corrosion films are analyzed by SEM (Scanning Electron Microscopy). Samples for analyses are cut 20 cm downstream the inlet, in the middle of the pipe and 20 cm upstream the outlet of each test pipe. The corrosion rate is calculated from the total amount of dissolved iron, based on SEM measurement of corrosion film thickness and composition and the measured condensation rate and concentration of Fe 2+ in the condensed water. EFFECT OF TRACES OF H 2 S ON TOP OF LINE CORROSION Corrosion Rates in Low Temperature Experiment with Traces of H 2 S Top of line corrosion experiments at 30 to 85 C, around 1 bar CO 2 and presence of acetic acid have been performed previously for another pipeline project in the loop described above 9. The loop was rebuilt to enable experiments with higher CO 2 partial pressure and traces of H 2 S to be performed for the project described here. The experiment described in this paper was performed at 25 C with 10 bar CO 2 and 2 mbar H 2 S partial pressure. Acetic acid was added to the system in order to obtain 3

300 ppm free acetic acid in the condensing water. These conditions are representative for the inlet of the carbon steel part of the pipeline system which was the basis for this study. Because of the low temperature very low water condensation rates were expected, and water condensation rates of 0.006 and 0.003 g/m 2 s were used in the experiment. The very large surface of the carbon steel test sections where the condensation took place enabled measurement of these very low water condensation rates. As the test specimens are the whole 1.8 m long test pipes it is not possible to obtain direct weight loss measurements of the corrosion rate. The average corrosion rate in each specimen is calculated from the sum of dissolved iron measured in the condensed water and the amount of iron remaining in the corrosion films, as shown in Table 1. The duration of this experiment was 6 weeks. The corrosion film weight was measured by cutting small pieces of the test pipes and measuring the weight difference when the corrosion films were stripped off chemically. This is done by stripping in an inhibited acid solution consisting of concentrated HCl containing 50 g/l SnCl 2 and 30 g/l SbCl 3. This solution removes corrosion products efficiently without attacking the underlying metal. The corrosion film weight was measured on several smaller pieces cut from the 1.8 m long test tubes. Other pieces were studied in the SEM with the corrosion films intact. The SEM examination showed that the corrosion film consisted almost entirely of iron sulfide with an Fe:S ratio very close to 1:1. This makes it possible to calculate the amount of iron in the corrosion film. The corrosion rate calculated by this method was 0.17 mm/y for the highest water condensation rate of 0.006 g/m 2 s and 0.10 mm/y for 0.003 g/m 2 s condensation rate. The calculation of iron dissolved in the condensed water and remaining in the corrosion film showed that about 90 % of the iron dissolved by corrosion remained in the corrosion film. This is much higher than in previous top of line corrosion experiments in pure sweet systems, where the corrosion films consist of iron carbonate 9. In the present experiment the corrosion films consist primarily of iron sulfide, which precipitates much more easily than iron carbonate, and less dissolved iron is transported away with the condensed water. The iron content in the condensed water was measured several times, and decreased somewhat during the experiment. The variation in top of line corrosion rate during the experiment has been estimated from the variation in iron content in the condensed water and is shown in Figure 4. This indicates a slight decrease in corrosion rate during the experiment, with values around 0.15 and 0.07 mm/y at the end of the experiment for the two test sections. It should be noted that this variation in top of line corrosion rate is uncertain, as it depends on the assumption that the fraction of dissolved iron that is transported away with the condensed water is constant during the experiment. The condensing water will have a low ph value due to the high CO 2 partial pressure, but the ph value will increase due to corrosion products dissolving in the condensed water. With a measured dissolved iron content around 500 ppm in the condensed water the in-situ ph in the collected condensed water will be in the range 5 to 5.5. 4

Corrosion Product Films in Top of Line Corrosion Experiment with Traces of H 2 S When the experiment was finished and the test specimens were removed and cut for examination, a thick black corrosion film appeared. This film spalled and cracked when the specimens dried after being removed from the loop. In some areas most of the film fell off the specimens, while other pieces were kept with the film intact for further examination. Figure 5 and 6 shows the appearance of the corrosion film as seen by SEM examination on cross sections of the specimens. Figure 5 shows a detail of an area where the thick film is intact, while Figure 6 shows an area where the thick film has fallen off. In all specimens there were some areas where the iron sulfide film was adhering to the steel surface and other areas where the FeS film spalled off. Much of the spalling occurred when the specimen had dried up after the end of the experiment, and in some areas the film fell off during storage after smaller pieces had been cut out. In the areas where the thick film remained the film thickness was typically in the range 50-100 µm. The film weight measurements and SEM studies were concentrated on areas where the film remained on the surface. Element analysis of the films in the SEM showed that the corrosion film consisted almost entirely of iron sulfide with an Fe:S ratio very close to 1:1. This applies for the whole film in Figure 5 for instance, except for the very thin (5-10 µm) inner part of the film closest to the steel surface, which appears darker and more compact. This inner part of the film contains less than 10 % sulfur and much oxygen, and is identified as primarily iron carbonate. In Figure 6 only this thin iron carbonate film remains on the surface, while the thick and more porous iron sulfide film has spalled off after the exposure. The film in Figure 6 contains almost no sulfide, and is clearly a predominantly iron carbonate film. It is likely that this iron carbonate film has formed as a result of sulfide depletion in the condensed water close to the steel surface, as the small amounts of sulfide in the condensed water can be consumed by precipitation of iron sulfide in the outer part of the corrosion film. It should be noted that also other top of line corrosion studies with a few mbar H 2 S has shown corrosion films consisting of iron sulfide 10. The iron dissolved by corrosion will diffuse away from the steel surface in the condensed water film, and dissolved H 2 S will diffuse towards the steel surface. When the dissolved iron diffusing outwards meets the H 2 S diffusing inwards, iron sulfide will precipitate. However, there is so little H 2 S available in the system that the H 2 S is consumed by iron sulfide precipitation before it reaches the steel surface. This results in a porous, fluffy and non-adherent iron sulfide film. At the steel surface there will be no H 2 S left for iron sulfide precipitation, and iron carbonate precipitation becomes possible. A very thin and protective iron carbonate film was observed at the steel surface, beneath the porous and non-protective iron sulfide film. The formation of this iron carbonate film will be influenced by the presence of H 2 S in the system, as the iron carbonate film may become thicker and more protective when there is no H 2 S present which can drain ferrous ions from the surface and cause precipitation of iron sulfide instead of iron carbonate. The main effect of H 2 S in this system is then to precipitate iron dissolved by corrosion as a porous, non-protective iron sulfide film, and thereby act as a sink for Fe 2+ ions. This can make it possible to dissolve more iron in the condensed water, resulting in a higher top of line corrosion rate than under pure sweet conditions. The protectiveness is governed by a thin iron carbonate film close to the steel surface in the same way as in pure sweet systems. 5

Examination of the corrosion films formed in the experiment showed that the protectiveness under conditions is governed by a thin iron carbonate film close to the steel surface in the same way as in pure sweet systems. It is likely that this iron carbonate film has formed as a result of H 2 S depletion in the condensed water close to the steel surface, as the small amounts of H 2 S in the condensed water can be consumed by precipitation of iron sulfide in the outer part of the corrosion film. The main effect of H 2 S in this system is to precipitate iron dissolved by corrosion as a porous and only slightly adherent iron sulfide film which does not offer any good protection, as the small amount of dissolved H 2 S easily becomes depleted by iron sulfide precipitation before it reaches the steel surface. The H 2 S will then act as a sink for Fe 2+ ions, thus enabling dissolution of more iron in the condensed water and a higher top of line corrosion rate. The mechanism for protection of top of line corrosion appears to be in principle the same as for pure sweet systems: precipitation of a protective iron carbonate film due to supersaturation of iron carbonate in the condensed water. Prediction of Top of Line Corrosion with Traces of H 2 S The present models for top of line corrosion are developed for sweet systems without presence of H 2 S and are based on iron carbonate precipitation 4-8. They are not necessarily valid for situations with even small amounts of H 2 S, where iron sulfide will be the dominant corrosion product. Model predictions for the present conditions with 25 C, 10 bar CO 2, presence of acetic acid and 0.006 g/m 2 s water condensation rate give a corrosion rate of 0.06 mm/y when the presence of H 2 S is not taken into account 5. The corrosion rate measured in the experiment described here was 0.17 mm/y for this condensation rate, or about three times higher than the predicted top of line corrosion rate for pure sweet conditions. A possible explanation for this can be that the principles of the sweet top of line corrosion model may still be applied since the protection afforded seems to be governed by iron carbonate precipitation also for these conditions, but the top of line corrosion rate is increased by the presence of small amounts of H 2 S due to the effect of H 2 S on removing Fe 2+ ions from the solution and enabling faster iron dissolution. It should however be noted that the top of line corrosion model is uncertain and not well tuned to the present conditions with low temperature and very low water condensation rate. At 10 bar CO 2 partial pressure as used in the experiment the presence of acetic acid in the condensed water results in an increase in the iron solubility and hence the predicted top of line corrosion rate by about 40 % compared to similar conditions without acetic acid. This is a marked effect, but much smaller than at low CO 2 partial pressures, where the presence of acetic acid has a much stronger effect. For corresponding conditions at 1 bar CO 2 the same amount of acetic acid would more than double the iron solubility and the predicted top of line corrosion rate. The results from the present experiment indicates that the top of line corrosion rate should increase with the water condensation rate for cases with traces of H 2 S much in the same way as for purely sweet conditions. However, it should be noted that top of line corrosion in the presence of H 2 S has not been studied in detail, and much more work is needed before top of line corrosion mechanisms and models can be relied upon in the presence of H 2 S. In any case dedicated laboratory testing should be the preferred way to validate a specific design if top of line corrosion is considered important. 6

CONCLUSIONS The corrosion product films formed in top of line corrosion in a mixed acid gas can be quite complex. In conditions with even very small amounts of H 2 S, the corrosion product film changes from iron carbonate to iron sulfide dominated and existing top of line corrosion models are not valid. Very limited data is available at other H 2 S concentrations. In the absence of this data laboratory testing under pipeline design conditions is advisable and this testing should include exposure times which are long enough to demonstrate mature film growth rates. Top of line corrosion studies performed under conditions at 25 C with 10 bar CO 2 and 2 mbar H 2 S partial pressure and presence of acetic acid resulted in measured top of line corrosion rates close to 0.1 mm/y. Most of the iron dissolved by corrosion precipitated as a 50-100 µm thick, porous and fluffy iron sulfide film which did not offer good corrosion protection. Thin, protective iron carbonate films were formed close to the steel surface, probably as a result of sulfide depletion in the condensed water close to the steel surface. The main effect of small amounts of H 2 S may be to act as a sink for ferrous ions, thus enabling dissolution of more iron in the condensed water and a higher top of line corrosion rate than under pure sweet conditions. In situations with higher H 2 S content this sulfide depletion may not be seen, and the mechanism for top of line corrosion rate may then be totally different. The present observations at low H 2 S content should not be generalized to situations with higher H 2 S content in the gas. The measured top of line corrosion rate in the presence of small amounts of H 2 S was higher than estimated by a top of line corrosion model for a pure sweet situation without H 2 S. The top of line corrosion rate seems to increase with the water condensation rate also for cases with traces of H 2 S. Top of line corrosion in the presence of H 2 S is not understood in detail and is not possible to predict reliably with the present knowledge. ACKNOWLEDGMENT The authors want to thank ExxonMobil Development Company, Chevron Australia, Shell Global Solutions and Gorgon Upstream Joint Venture for technical discussions and permission to publish this paper. REFERENCES 1. Stein Olsen, Arne Dugstad, "Corrosion under Dewing Conditions", CORROSION/91, Paper No. 472, (Houston, TX: NACE, 1991). 2. Rolf Nyborg, Arne Dugstad, Liv Lunde, "Top-of-the-Line Corrosion and Distribution of Glycol in a Large Wet Gas Pipeline", CORROSION/93, Paper No. 77, (Houston, TX: NACE, 1993). 3. Yves Gunaltun, D. Supriyam, J. Achmad, "Top of Line Corrosion in Multiphase Gas Lines. A Case History", CORROSION/99, Paper No. 99036, (Houston, TX: NACE International, 1999). 7

4. Yves Gunaltun, Dominique Larrey, "Correlation of Cases of Top of Line Corrosion with Calculated Water Condensation Rates", CORROSION/2000, Paper No. 00071, (Houston, TX: NACE International, 2000). 5. Rolf Nyborg, Arne Dugstad, "Top of Line Corrosion and Water Condensation Rates in Wet Gas Pipelines". CORROSION/2007, Paper No. 07555, (Houston, TX: NACE International, 2007). 6. Bert Pots, Edwin Hendriksen, "CO 2 Corrosion under Scaling Conditions - The Special Case of Top-of-Line Corrosion in Wet Gas Pipelines", CORROSION/2000, Paper No. 00031, (Houston, TX: NACE International, 2000). 7. Frédéric Vitse, Srdjan Nesic, Yves Gunaltun, Dominique Larrey de Torreben, Pierre Duchet- Suchaux, "Mechanistic Model for the Prediction of Top-of-the-Line Corrosion Risk", CORROSION/2003, Paper No. 03633, (Houston, TX: NACE International, 2003). 8. Ziru Zhang, Dezra Hinkson, Marc Singer, Srdjan Nesic, "A Mechanistic Model of Top of the Line Corrosion", CORROSION/2007, Paper No. 07556, (Houston, TX: NACE International, 2007). 9. Tore Roberg Andersen, Anne Marie K. Halvorsen, Arne Valle, Gry Pedersen Kojen, Arne Dugstad, "The Influence of Condensation Rate and Acetic Acid Concentration on TOL Corrosion in Multiphase Pipelines", CORROSION/2007, Paper No. 07312, (Houston, TX: NACE International, 2007). 10. Alvaro Camacho, Marc Singer, Bruce Brown, Srdjan Nesic, "Top of the Line Corrosion in H 2 S/CO 2 Environments", CORROSION/2008, Paper No. 08470, (Houston, TX: NACE International, 2008). TABLE 1. Top of line corrosion experiment at 25 C with 10 bar CO 2 and 10 mbar H 2 S. Test section A B Condensation rate g/m 2 s 0.006 0.003 Average iron content in condensed water mg/l 410 500 Iron dissolved in condensed water g/cm 2 0.0011 0.0007 Corrosion film weight g/cm 2 0.022 0.013 Iron in corrosion film g/cm 2 0.014 0.009 Total dissolved iron g/cm 2 0.015 0.009 Average corrosion rate mm/y 0.17 0.10 8

FIGURE 1. The test loop used for top of line corrosion experiments. FIGURE 2. Setup of the test sections. 9

FIGURE 3. Top of line corrosion loop showing two of the test sections. Corrosion rate mm/y 0.2 0.1 Section A Section B 0.0 0 200 400 600 800 Time h FIGURE 4. Estimated development in corrosion rate. 10

FIGURE 5. Thick iron sulfide film from Test section A. FIGURE 6. Thin iron carbonate film in Test section A. 11