Go To Review

Calibration of an Integrated Model for Prediction of Corrosivity of CO₂/H₂S Environments

Vamshi R. Jangama
Sridhar Srinivasan
CLI International, Inc.
Houston, TX USA
e-mail software@intercorr.com

A study of field data from 15 wells was performed to analyze and develop correlations for corrosion prediction in carbon steels. These data were made available by six oil and gas production companies from more than 10 different fields. The data was analyzed to determine parametric relationships and develop an understanding of field corrosion as a function of critical parameters. The field observations were compared with a corrosion prediction computer model, Predict, to assess the accuracy of the model and to develop a basis for model calibration.

One of the most fundamental issues in current day corrosion research is assessment of corrosion rates in steels and determination of corrosivity of typical operating environments in oil and gas production. Such an assessment requires an understanding of the role of primary environmental and metallurgical variables and underlying mechanisms of corrosion as well as synthesizing different types of information to account for various parametric and inter-parametric effects [1]. Ability to evaluate field parameters and predict corrosion rates can increase the safety of the operation and also the profitability by cutting down on workovers that greatly affect the economy of the industry. Corrosion prediction models need to account for field behavior in terms of critical operating parameters and also be efficient and cost-effective to be useful to the industry. In this study field data from 15 different wells have been analyzed to observe underlying trends that can provide parameteric relationships for corrosion prediction.

Information for this analysis has been provided by six oil and gas production companies from fields mostly in the Gulf of Mexico and the North sea. Data available for each of the wells include information about [2]:

• Well Documentation
• Corrosion Control
• Selection procedure for inhibitors

Information for this study was selected from the well documentation and corrosion control. Well documentation contains information about:

• Location
• Depth
• Gas Analysis (H₂S and CO₂ content)
• Water (Cl^- content and Production Rate)
• Liquid Hydrocarbon information (GOR & Weight)
• Pressure
• Temperature
• Flow rate
• Tubing size, grade and coating information.

Corrosion control information available contained:

• Corrosion rates
• Corrosion monitoring methods
• Inhibitor treatment methods and results of the treatment.

Aforementioned field parametric data was used as input to the integrated computer model (Predict™) for corrosion prediction and estimates for corrosion rates were obtained. Table 1 shows the information used to obtain a corrosion rate estimate for each of the 15 wells. Table 1 also shows predicted pH, predicted corrosion rate and actual reported corrosion rate for each of the wells.

Well Number	H2S, bar	CO2, bar	Chloride, ppm	Temperature, oC	GOR, m3/m3	W/G, m3/Mm3	Dew Pt. OC	Inhibition method	Inh. Eff %	pH	Pred. CR, mmpy	Actual CR, mmpy
1	1.0345	6.8965	30000	107	5882	16.76	107	Squeeze	85	4.12	0.23	0.19
2	1.0345	0.8276	30000	79	1069.5	5.59	79	Squeeze	85	4.67	0.008	0.038
3	0	0.5517	15000	79	1782.5	16.76	79	Squeeze	85	5.20	0.018	0.013
4	4.138	3.45	15000	121	1069.5	11.17	121	Squeeze	50	4.13	0.62	0.38
5	10	19.31	42000	171	1069.5	16.76	171	Squeeze	85	3.55	5.38	5.08
6	24.83	34.48	160000	135	3565	5.59	135	Squeeze	95	3.24	2.05	0.76
7	6.9 e-2	5.31	15000	193	1069.5	21.23	176	Squeeze	85	4.28	0.013	0.006
8	0	2.07	8000	138	1584.3	8.94	138	Continuous	95	4.70	0.056	0.0025
9	0	4.48	60000	138	1069.5	3491.6	138	Continuous	95	4.36	0.475	0.305
10	4.8276	20.69	15000	193	1069.5	44.13	166	Squeeze	95	3.62	0.015	0.051
11	3.448	93.1	15000	191	7130.12	419	179	Continuous	95	3.03	9.03	7.62
12	6.9 e-2	17.24	15000	160	11764.7	1.12	160	Squeeze	85	3.78	0.224	0.19
13	0	6.621	10000	99	445	30%	Not Per	Squeeze	85	4.19	0.67	0.61
14	0	4.138	30000	79	2762.9	3.74	79	Squeeze	85	4.33	0.145	0.43
15	0	8.276	15000	193	1069.5	335.2	179	Continuous	85	4.1	0.028	0.004

A technical description of the reasoning in the Predict system is given in Ref. 1. PREDICT is a software tool that provides an assessment of environment corrosivity and prediction of corrosion rates for steels in typical production environments. The program incorporates a numerical/heuristic model that integrates the effects of a complex set of environmental parameters to provide a corrosion rate assessment based on extensive literature data, lab testing and field experience. A flow chart delineating the hierarchical reasoning structure of the Predict system is given in Figure 1a. The Predict system has been implemented as a Windows-based computer program with an interface as shown in Figure 1b. Based on data specified for different parameters, the system will instantaneously display the following results:

• System pH
• Predicted Corrosion Rate called Corrosion Index (in mpy or mmpy)
• A textual recommendation in the results box indicating whether the predicted corrosion rate is within the specified allowance for the particular system.
• A corrosion index bar that graphically represents the corrosion rate.

Figure 1a: Predict System Flow Chart

Figure1b -- PREDICT Start up Screen

A typical Predict consultation will involve the following five steps:

1. Specification of pH related data: At the outset, the system determines a corrosion rate only if the operating environment is acidic or has aeration. If the specified environment has no acid gases or there is sufficient buffering to produce a pH higher than 7.0, the system will predict zero or very low corrosion rates, except under conditions of aeration. So, the first step in consulting the system involves specification of the acid gas (H₂S and CO₂) partial pressures as well as the bicarbonate content of the environment.
2. Temperature/Gas-Water ratios: Temperature has a significant impact on corrosion rates since precipitation of corrosion products and scaling are functions of temperatures. Corrosion rates typically increase with increasing temperature, though in CO₂ dominated systems, FeCO₃ scaling at higher temperatures can produce significant protection against further corrosion. If the Gas to Oil Ratio indicates gas dominated conditions (as opposed to an oil dominated system) the system uses the water to gas ratio and the dew point as means to determine availability of an aqueous medium to measure corrosion. So, depending on the value entered for the Gas to Oil Ratio, the system will let you specify the relevant water-related parameters. If the Gas to Oil Ratio is less than 5000 scf/bbl (which denotes an oil well), the system uses the water cut and oil persistency to determine the wetness effect.
3. Chlorides/oxygen/sulfur: Chlorides and sulfur typically make corrosion worse if the process has been initiated by the presence of acid gases. Their role, while not as critical as that of H₂S or CO₂, is significant because these parameters can significantly increase corrosion rates in mildly corrosive systems. Presence of oxygen beyond 20 ppb even in mildly acidic systems can lead to significant corrosion rates, especially with high chlorides and high flow rates at elevated temperatures.
4. Velocity/Type of flow: Flow parameters are very critical in both determining and controlling corrosion effects. Erosion corrosion as well as the protection (or the lack of it) from corrosion films is very much a function of fluid velocity. Velocity has a significant impact on mass transport within the corrosion boundary layer and also impacts a corroding system's ability to form protective scales.
5. Inhibition/corrosion allowance: Inhibition choices in the system allow the user to select applicable methods of inhibition for vertical or horizontal flow and determine the extent of corrosion mitigation. In some cases, the system might provide no protection due to inhibition because of high velocities or chloride concentrations. The system's rules assess the appropriateness of method of inhibition delivery for a given set of conditions.

Analysis of the data can be divided into 3 broad aspects, namely,

• Collection of data,
• Evaluation of each data set
• Cumulative analysis considering different aspects of each of the wells

Individual wells were studied to locate any discrepancies in information provided. One example of such a discrepancy is well # 4 which was reported to have no water in the system, but a corrosion rate of 15 mpy. Since an aqeous medium is needed to support any corrosion a water to gas ratio of 11.17 m³/M.m³ (2 bbl/Mscf) of gas was assumed, otherwise the computer model would predict no corrosion. Only one (Well # 13) of the 15 wells is an oil dominant system and has a water cut of 30% with a non-persistent oil film. H₂S and CO₂ partial pressures were calculated from the mole percent of each of the components and the total pressure of the system.

Of the 15 wells in the study 9 wells have H₂S content varying from 0.069 to 24.83 bar (1 to 360 psia) where as the others had no H₂S. The water analysis information available contained only the chloride count in ppm, further, a constant amount of 100 ppm of bicarbonate was assumed to be present in each of the cases. This was considered to be a good estimate for gas wells. Also a constant fluid velocity of 20 ft/s (6.1 m/s), a reasonable value for gas wells, was assumed for all the wells as this information was not reported. Wells that did not have information about the efficiency of inhibition were assumed to have 85 % inhibitor efficiency which was observed to be a typical value.

Figure1 shows a plot of CO₂ partial pressure and pH versus reported corrosion rate. A second order polynomial for CO₂ partial pressure shows a very strong correlation with the corrosion rate. The pH on the other hand, does not seem to show a good correlation with the corrosion rate. This can mostly be attributed to the fact that the water analysis information contained only the chloride ion concentration and also a constant concentration of bicarbonate ions was assumed for the analysis. These observations support the idea that corrosion is not a simple phenomena that can be predicted by just one parameter.

Figure 3 shows a plot of the predicted corrosion rate versus the reported corrosion rate for all the 15 wells. The 45^o line shows the case for a perfect fit where as the linear trend line shows the case for the present study. Predicted corrosion rate shows a correlation coefficient of 0.9805 which is very good for the number of data points used in the study.

Figure 4 shows a plot similar to Figure 3 for wells which have H₂S in the system. The high correlation coefficient (0.9808) indicates agreement between the trend line and the data spread. The number of data points in this plot are 9 and the correlation is clearly good for the data. Figure 5 shows a plot of the predicted and reported corrosion rates for wells that do not have any H₂S in the system. The correlation coefficient is 0.6993, which while not as high as that seen in Figure 3, is an acceptable basis to propose linearity between observed and predicted values. The number of data points used in this figure is 6 and it is possible that additional data can improve the correlation coefficient in this case.

Field observations were compared with theoretical predictions of a computer model to estimate model accuracy and develop a basis for model calibration. Reasonable agreement illustrated between field (actual) and predicted values supports the correctness of the model's underpinnings. Further calibration with field data will be useful in calibrating the model to provide predictions consistent with field observations.

1. S. Srinivasan and R. D. Kane, "Prediction of Corrosivity of CO2/H2S Production Environments", Paper No. 96011, Corrosion/96, NACE International, Denver, CO, March 1996.
2. S. Srinivasan and V. Jangama, Internal report on field data analysis, CLI International, Inc., Houston, Texas, 1996.
   

(0)