Stabilizing pH in Troll pipelines solves glycol-regen problems

Stein Olsen
Statoil Research Centre
Trondheim

Olav Lunde
Statoil
Bergen

Arne Dugstad
Institute for Energy Technology
Kjeller, Norway

Iron from corrosion in the two 36-in. gas-condensate pipelines from the Troll field to shore caused problems in the glycol regeneration units as a result of iron precipitation.

Although the amount of iron represented no significant corrosion problem in the pipelines, the solution to the problem in the glycol-regeneration units was to reduce the corrosion rate by use of pH stabilization.

NaOH was injected to achieve a pH value of 7.4, which in turn reduced the iron content to 10 ppm, corresponding to a maximum corrosion rate of less than 0.1 mm/year.

Moving corrosive liquids

Troll is a large gas-condensate field offshore Norway originally developed by Shell and operated by Statoil. Maximum production rate from the field, which started up in mid-1996, is up to 100 million standard cu m/day (MMscmd) in water roughly 300 m deep.

To minimize the size of the offshore production facilities, Statoil decided to perform only simple processing offshore and the main processing on shore. The offshore process was designed to remove the free water only. Condensate and gas were then mixed and fed into the pipelines.

As more water condenses in pipelines with temperature decreases, corrosive fluids are transported in two 36-in. pipelines from the platform over a distance of 65 km to shore. Troll is a sweet gas field with a CO2 content of 0.3%. Pipe inlet and outlet conditions are given in Table 1 [12,901 bytes].

To control hydrate formation, operator Statoil injects monoethylene glycol (MEG). Following evaluation of different corrosion-mitigation methods, Statoil decided to use MEG solely with a large corrosion allowance in the pipe steel. This resulted from the fact that the corrosivity was low in the system, the CO₂ pressure being only 0.04 MPa.

The target was to reach the manageable corrosion rate of less than 0.2 mm/year. The amount of MEG had to be increased more than what was needed for hydrate control in order to reach that level. The amount of MEG to be injected into the pipelines was estimated to be 3 cu m/million cu m of gas.¹ This was a lean MEG with a maximum 10% water content.

The water in the system is a mixture of condensed water from the gas and the water in the injected MEG. Because of the corrosion process in the pipelines, iron and bicarbonate will be formed. In addition, some carry-over of other salt must be expected from the platform process.

In the MEG regeneration process, these salts will remain in the glycol and accumulate or precipitate. Some increase of salts was thus expected. A simplified sketch of the system is shown in Fig. 1 [57,978 bytes].

During production, precipitation of iron led to a severe problem in the regeneration units.

Precipitation took place in the heat exchangers and boilers when the pressure was reduced and the temperature increased. Some of the iron formed scales on the surfaces in the heat exchangers and the boilers and some precipitated as particles in the bulk phase. These particles recirculated in the low flow areas like tanks, drums, and slug catcher.

The estimated amount of corrosion products was 20 tons the first year of operation.

This was regarded not as a corrosion problem in the pipelines but a process problem. Statoil evaluated three methods for solving this problem which led to selection of a method based on reduction of iron production, i.e., stop corrosion in the pipelines by use of pH stabilization.

Stabilizatiion

The pH-stabilization technique was introduced by Elf Aquitaine Production as a new method for corrosion control in corrosive sweet gas/condensate environments. The method has been used in Italy, The Netherlands, and Norway.

In sweet corrosion, protective corrosion product films play an important role. The pH is one of the most important parameters in this context. Stabilization of pH is based on increasing the pH significantly and achieving conditions favorable for formation of protective corrosion product films.

Laboratory experiments and field data show that the corrosion rate will be very low under such conditions, typically well below 0.1 mm/year. Prediction by use of corrosion models is irrelevant for these high pH conditions.

Because of high pH, this method is strictly limited to conditions with condensed water only. In case formation water with calcium is produced, a high pH will immediately lead to scaling.

Norway's Institute for Energy Technology (IFE) has conducted extensive experiments² to study the effect of pH stabilization. A pH shift of 1.5-3 units compared to the "pure" condensed water pH is usually recommended. The actual target pH will depend on the CO₂ partial pressure and the glycol concentration.

In temperatures greater than 40° C., the corrosion rate immediately decreased to less than 0.1 mm/year. For lower temperatures, the reduction in corrosion depended on the metal surface state.

On freshly grounded specimens, only a slight reduction in corrosion rate could be detected after 4 weeks of exposure. But if the specimens were covered with an oxide, the corrosion decreased immediately.

The field data with methyl diethanolamine (MDEA) showed, however, very low iron content in the water from all systems with pH stabilization. This has also been the case for systems partly operated at low temperatures. This indicates that pH stabilization works for low temperatures on real pipe surfaces.

The pH stabilization method is well suited to be used in combination with glycol because the pH stabilizer will remain in the regenerated glycol. This means that there is no need for continuous renewal.

The last chemical used in field application is MDEA, which is chemically stable under the conditions in a regeneration unit. In laboratory tests, sodium salts like NaOH and NaHCO₃, which are more environmentally acceptable, are studied.

If NaOH is used, this will react with the CO₂ and form bicarbonate which will increase the pH. In a practical situation, one should select what is most convenient and economical.

NaOH injection

The pH based on the CO ₂ partial pressure and condensed water is about 4.2. But because of the corrosion process in the system, the actual pH at the outlet of the pipeline before pH stabilization was about 6.

Results from the laboratory testing led to a decision to increase the bicarbonate concentration to achieve pH 7.4 in the section of the pipeline operating at 20° C. This required a total amount of 50 mol/l. bicarbonate in the pipeline. This was based on calibration curves made for both the actual CO₂ pressure of 0.04 MPa and for 0.1 Mpa (Fig. 2 [104,720 bytes]).

The data for 0.1 MPa CO₂ were made to simplify the laboratory procedure when pH was checked in water samples. The curves are based on a content of 7 g/l. of NaCl, the actual content in the outlet of the pipelines before the injection of the pH stabilizer. The dashed curves in the figure represent values with no NaCl.

For practical reasons, it was decided to inject 21.9 wt % NaOH as pH stabilizer. The solution was injected in batches into the lean MEG tank which is operated at ambient pressure.

There were some uncertainties regarding local precipitation and foaming when the pH was to be increased. The pH was thus increased in several steps to allow detection of any problems. The total program lasted more than 1 month before the target pH of 7.4 was reached. No problems occurred. The development of iron, bicarbonate, and pH in the slug catcher is shown in Figs. 3 [90,910 bytes] and 4 [107,806 bytes].

Measurement

Iron content was measured at several points in the system to determine the total iron mass balance. To allow discrimination between dissolved iron and iron precipitated in the bulk phase, measurements were taken of both total and dissolved iron.

Fig. 3 indicates that the major part of the iron coming from the slug catcher was in the form of dissolved iron. Measurements in the lean MEG tank showed, however, that a large amount of the iron was in fact transported out of the glycol units and recirculated to the pipeline as solid iron salts. Thus, only a small fraction of the iron precipitated as scale in the MEG trains.

The particles leaving the MEG trains must thus be either settled out in the supply line to the field, settled out in the Troll pipelines or in the slug catcher, or redissolved when exposed to CO₂ in the main pipelines.

Redissolving is the most likely scenario because laboratory experiments showed that almost all the particles were dissolved when the liquid was exposed to CO₂. This shows that the main part of the particles must be Fe(OH)₂ because the solubility of FeCO₃ and Fe₃O₄ is too low to account for the high concentration of the dissolved iron.

Evaluation

The original corrosion evaluation was based on prediction of the potential corrosivity. ³ For the actual temperatures, it must be assumed that good protective corrosion product films will not be formed. Two different types of corrosion attacks are likely to occur: "top of line" corrosion and corrosion in the bulk phase in the bottom of the line.

The most important rate determining parameters in top-of-line corrosion are condensation rate, temperature, and CO₂ partial pressure. The water that will condense on the inner pipe wall will rapidly be saturated with corrosion products. The pH in the water will therefore increase and more or less protective corrosion product films will form and cover the steel surface.

These films can reduce the corrosion rate. A constant corrosion rate is obtained when the corrosion rate has been reduced so much that it is balanced by the rate at which corrosion products are transported away from the surface by the condensing water.

Experiments have shown that the corrosion rate can be estimated when the condensation rate and the solubility of iron carbonate in the condensed water are known, and a simple model has been developed.⁴ The approach used for prediction of the top-of-line corrosion in the present evaluation and the condensation rates were calculated from a temperature profile in the pipelines. The results are shown in Fig. 5 [72,182 bytes].

It is assumed that the gas is saturated with water vapor, that all the water condenses out on the wall (90% of the total wall area), and that cold spots have twice the normal condensing rate. It can be seen that the corrosion rate will be less than 0.1 mm/year. The results are very conservative.

The corrosion rate prediction in the bulk phase in the bottom of the line before injection of the pH stabilizer is based on the Norsok M-506.⁵ Important parameters in such a prediction are CO₂ partial pressure, pH, temperature, velocity, and glycol concentration. These parameters are determined as a function of distance from the Troll platform.

The temperature decreases rapidly, whereas the glycol concentration decreases as more water condenses. These parameters influence the corrosion rate in an opposite direction. The maximum predicted corrosion rate based on lean MEG injected 3 cu m/ million std. cu m of gas is about 0.2 mm/year.

As shown in Fig. 3, the iron content in the slug catcher varies with time. An average amount is about 100 ppm. As a conservative approach, it is assumed that 50 ppm is coming from corrosion in the pipelines and the rest from the injected glycol.

In order to transfer this into an actual corrosion rate, the area of the wetted surface must be determined. From field experience in gas-condensate pipelines, enhanced corrosion can be expected locally near and in the welds.

With the low corrosivity due to large amounts of glycol in the system, however, this effect will be insignificant in this case. An amount of 50-ppm iron represents less than 5 kg/day iron from each pipeline at maximum production. If only the lower half of the pipe circumference is contributing to iron release, less than 1% of the pipeline can corrode in order to achieve an average corrosion rate greater than 0.2 mm/year.

This is unlikely because the conditions are unfavorable for formation of protective corrosion product films which means that corrosion is likely to occur along the whole pipeline. The conditions before injection of pH stabilizer were thus not regarded as any corrosion problem in the pipelines but as a process problem only.

After injection of pH stabilizer, the amount of iron measured in the MEG/water mixture in the slug catcher was less than 10 ppm. More than half of the iron is in particle form, and it is assumed that this iron is recirculated in the glycol loop.

The contribution from corrosion is therefore probably less than 5 ppm, and the corrosion rate is very low. It is difficult to predict the exact corrosion rate because the area of the attacked surface is unknown.

Two cases

The corrosion rate is predicted for two cases:

• One assuming 5% of the pipe periphery being attacked.
• One based on the corrosion taking place where the temperature is 20-40° C. This part of the pipeline was chosen because it is the temperature range in which the uncertainty in the predicted corrosion rates is highest. There are no laboratory data for which NaHCO₃ has been used as a pH stabilizer. The corrosion rate is calculated to be far less than 0.1 mm/year in both cases.

It is likely that a protective film can be damaged during operation. This can be caused by pigging or by detachment. Detachment can arise as a result of internal stresses generated during film growth or due to large pressure and temperature fluctuations. The question is whether the film will recover when damages have occurred.

Recovery of the film depends among other things on the dissolved iron content in the pipeline. It is assumed that the iron content in the bulk phase decreases with time as the steel surface becomes better and better protected. That means that the driving force for film formation should also be reduced, but only if galvanic effects are not taken into account.

Scratching experiments of protected surfaces showed that the potential increased and a galvanic current passed between the scratched and the covered part of the surface. The acceleration of the anodic reaction rate in the scratched area increases the relative Fe²⁺ concentration, while the relative reduction in CO₃^2- is much less affected by the high bulk concentration of carbonate and bicarbonate.

That means that the supersaturation level of FeCO₃ on the scratched surface can be high, perhaps higher than the concentration when the steel was exposed the first time. Many experiments confirmed that protective films were reformed in scratched areas. High corrosion due to damaged protective films is therefore not likely in the Troll pipeline.

After injection of the pH stabilizer, Statoil discovered that the reflux system in the boilers started to corrode. This was probably the result of the higher pH that led to higher release of CO₂ in the boilers which increased the corrosivity in the reflux circuit.

Acknowledgment

The authors thank the partners in the Troll license (Statoil, A/S Norske Shell, Norsk Hydro Production AS, Saga Petroleum AS, Elf Petroleum Norge AS, Norske Conoco A/S, and Total Norge AS) for allowing publication of this work.

References

1. de Waard, C., Lotz, U., and Milliams, D.E., "Predictive Models for CO2 Corrosion Engineering in Wet Natural Gas Pipelines," Corrosion/91, Paper No. 577, NACE 1991.
2. Dugstad, A., Dr nen, P.E., "Efficient Corrosion Control of Gas Condensate Pipelines by pH-stabilisation," Corrosion/99, Paper No. 20, NACE 1999.
3. de Waard, C., "Corrosion Control in the Troll Wet Gas Pipelines," NITO/NKF International Conference "Corrosion and Materials Offshore," Oslo, Sept. 14-15, 1994.
4. Olsen, S., and Dugstad, A., "Corrosion under Dewing Conditions," Corrosion/91, Paper No. 472. NACE 1991.
5. Norsok model, http://www.nts.no/norsok/m/m50601/m50601.htm.

The Authors