Go To Review
Experience Survey on Corrosion Monitoring and Mitigation Techniques for Sweet Well Production
Russell D. Kane and Sridhar Srinivasan
CLI International, Inc.
Houston, Texas USA
email: rdk@clihouston.com
Abstract
Experience Survey on Corrosion Monitoring and Mitigation Techniques for Sweet Well Production
Russell D. Kane and Sridhar Srinivasan
CLI International, Inc.
Houston, Texas USA
email: rdk@clihouston.com
Abstract
A review of field case histories was conducted to assess corrosion monitoring and mitigation techniques used in sweet (CO2-containing) oil and gas wells. Particular emphasis was placed on (1) obtaining information on both successes and failures in wells having high partial pressures of carbon dioxide and (2) analysis of the field data based on available laboratory data and predictive models for corrosion. In general, correlation between various surface corrosion monitoring techniques and actual down-hole corrosion and pitting severity determined by inspection and caliper surveys was poor. Pitting rates were found to be between 2 and 15 time the general corrosion rates reported for various cases. The ratio of pitting to general corrosion rate was lower for cases where the general corrosion rates were low (<0.5 mmpy) and increased to values of 12 for cases involving high general corrosion rates. Good correlation was obtained between predicted and actual corrosion rates for five cases with relatively complete data that were evaluated. In all but one case, predicted corrosion rate was greater than that observed.
Many hydrocarbon producing reservoirs contain acid gases (i.e. carbon dioxide and hydrogen sulfide) that are generally associated with increasing the severity of corrosion. For example, in one field used in this study, the carbon dioxide partial pressure was approximately 3.5 MPa with about 10 ppm hydrogen sulfide. The gas to oil ratio (GOR) was 200 M.m3/m3 (Million cubic meter/cubic meter), making this an oil production system. The water cuts were generally low (<30 percent) and the typical well depth being about 3000 meters. Corrosion problems were variable and difficult to predict. Many wells appeared to suffer little or no corrosion whereas others in the same area were found to have severe corrosion due to scattered and unpredictable pitting.
Based on the aforementioned operating scenario, a survey of field experience was conducted. The survey focused on fields with high levels of carbon dioxide. Particular attention was given to assessment of down-hole corrosion and pitting tendencies.
Case histories of both successes and failures in high CO2 oil and gas wells were solicited from producing companies and well service companies. A total of six companies participated in this survey. Table 1 shows data requested from survey participants. Utility of the data submitted by the companies was limited by the fact that information on all the categories of data requested was not available in all cases. At least partial information was obtained for a total of 66 wells.
Well Conditions
These wells were primarily located in the Gulf Coast region and the North Sea. In most cases, data were obtained for single, packed-off completions. About half of the data was from wells having a depth less than 5,000 meters and half came from wells with greater depth. About one half of wells surveyed had CO2/H2S ratio in the range of 0 to 1,000 and the other half a CO2/H2S ratio greater than 1000. Approximately 30 percent of the wells had water production rates in the range 1 to 20 bbl/day (low water production), 20 percent in the range 20 to 100 bbl/day and the remaining half of the wells had water production rate above 100 bbl/day. The wells were primarily gas condensate wells split evenly above and below a GOR of 20000 M.m3/m3. Bottom hole temperatures of these wells were about evenly split in the three categories: <150 C, 150 to 177C and >177 C.
Corrosion Control and Monitoring Practices
General corrosion rates varied in these wells from essentially zero to 1 mmpy with corrosion inhibition. Pitting rates in problem wells were highly variable and in the range of 3 to 15 mmpy. In one case, a pitting rate as high as 125 mmpy was reported. General corrosion rates were usually estimated by using iron counts in the produced water while pitting rates were commonly estimated from a multiplying factor time the general corrosion rate. The most credible corrosion rate data typically came from inspection of pulled tubing or from down-hole coupons or caliper surveys.
Many companies utilized more than one method of monitoring down-hole corrosion. Methods included: (1) iron counts, (2) caliper surveys, (3) down-hole mounted coupons, (4) radioactive sleeves, (5) inspection and (6) electrical corrosion probes. The most common methods were iron counts followed by corrosion coupons with both being used about 50 percent of the time. Caliper surveys and inspections were only utilized in about 25 percent of the cases.
Inhibition was used to minimize down-hole corrosion in all but one well with carbon steel tubing. The most common inhibition method was batch treatment employing tubing displacement which was used in about 60 percent of the cases. The next most common method was continuous inhibition which was used in about 30 percent of the wells. These were typically deeper and high production rate wells (>150 C) where limited inhibitor film persistence can be expected. The three wells that were not treated had stainless steel tubing (two had 13Cr martensitic stainless steel and the third 28Cr austenitic stainless steel). In the only untreated well with carbon steel tubing, tubing perforation was observed in less than one year due to pitting corrosion despite the report of no (zero) water production. This was in high production rate well with significant morphological deviations (surface bends and corners) and may have resulted from erosion corrosion.
Major Factors Influencing Corrosion
Based on the work of DeWaard and Milliams, the partial pressure of CO2 has been known to have a strong influence on corrosivity of the environment and the ensuing corrosion rates.1 (Figure 1). In most cases, this survey found that corrosion penetration rates did not follow these predictions. A very qualitative correlation of predicted and actual corrosion rates was obtained with a combination of acid gas partial pressure, water production rate up to 0.69 MPa. However, no data was available on the bicarbonate levels in the produced water or on the nature of the liquid hydrocarbon phase. Absence of data for these critical parameters in the survey highlights the need for more complete and reliable field data. Information about bicarbonates should be considered particularly important given the criticality of pH in corrosion estimation. The acid gas partial pressure and bicarbonate level in the produced water combine to determine the pH and, in general, corrosivity will increase with decreasing pH of the system.
One of the more important correlations obtained from this survey was the assessment of pitting rates versus general corrosion rates. As shown in Figure 2, maximum penetration rates from pitting could vary from 2 to 12 to 1. Figure 2 indicates that in cases where the corrosion rate of steel is relatively low (<0.5 mmpy) the pitting rate was from 2 to 4 times the general corrosion rate. However, for cases where the general corrosion rates were high, the pitting rate increased to about 12 times the general corrosion rate.
Assessment of Field Data with Integrated Corrosion Model
A comparison of field corrosion rate data and corrosion rates predicted by an integrated computer model (PREDICTTM)2 on a random sampling of five cases from the data set is shown in Figure 3. These were cases where the data was complete in terms of field conditions and corrosion rate information. The cases evaluated included three gas condensate wells with a high CO2/H2S ratio, and two sweet oil wells (no H2S). Assumptions were made about data not reported in order to facilitate utilization of the computer model. In all cases, the fluid velocity was assumed to be in the range 7 m/sec and bicarbonate content was assumed to be nil. For cases where batch corrosion inhibition was utilized, an inhibitor efficiency of 75 percent was assumed.
Figure 3 shows a plot of the minimum and maximum field corrosion rates obtained from various field monitoring methods along with the corrosion rate predicted from the computer model. It can be seen from this information that even with the assumptions made to compensate for lack of complete field data, reasonably good prediction of the trends in corrosion rate were obtained. Generally, the corrosion rates predicted by the computer model were somewhat higher than actually observed in four out of five cases evaluated.
Assuming no error in corrosion rate reported, for the case where the predicted corrosion rate was lower than the maximum corrosion rate observed, it is very likely that the actual field conditions differed from those reported (used in the computer model). For example, the assumed inhibitor efficiency was 75 percent which is reasonable for many batch treatments. However, it is possible that flow effects or ineffective chemical treatment resulted in a situation wherein the actual inhibitor efficiency is much less than that assumed. Additionally, a possible reason for the prediction of higher than actual corrosion rates in most cases could be due to service conditions that were less severe than assumed for prediction purposes. An example of such a situation is one where in, the beneficial influence of the bicarbonate or a possible persistent oil phase is not accounted for. Therefore, it can be seen that when using corrosion prediction models for assessment of corrosion severity, it is critical to have accurate and relevant data.
Corrosion prediction and field-based correlation is critical to
understanding and modeling corrosion behavior. Some conclusions
that can be drawn from the analysis presented in this paper are,