E3S Web of Conferences 121, 01019 (2019) https://doi.org/10.
1051/e3sconf/201912101019
Corrosion in the Oil & Gas Industry 2019
Mathematical simulation of the rate of carbon dioxide corrosion
at the facilities of Gazprom dobycha Urengoy LLC
Aleksandr Yusupov1,*
1Gazprom dobycha Urengoy LLC, Novy Urengoy, Russia, 629300
Abstract. In Gazprom dobycha Urengoy LLC, as in other oil-and-gas production enterprises, there are
problems of increased equipment wear due to corrosion. A special role there plays CO2 corrosion. Despite
the homogeneity of the extracted fluid and even chemical composition of the working medium, the nature
and intensity of corrosion damage to pipelines and equipment varies over a wide range, due to different
thermobaric parameters of well operation. To determine parameters influencing the rate of corrosion most
different methods of statistical analysis were used. The paper provides a methodology for compiling a
mathematical model and assessing its reliability. As a result, the equation of carbon dioxide corrosion in
relation to the conditions of Achimov deposits of Urengoy oil, gas and condensate field was obtained. The
type of the obtained equation was chosen according to the model of the classical de Waard-Milliams
carbon dioxide corrosion equation. The model proposed by the authors describes the processes of carbon
dioxide corrosion more reliably than the de Waard-Milliams equation does. The disadvantage of the
developed model is that it does not reliably describe the speed of corrosion in wells with corrosion rates,
significantly exceeding the average values for all wells studied.
1 Introduction carbon dioxide is reduced to the level of 0.1 MPa but the
presence of free water and the stratified flow conditions
In October 2009, the gas condensate facility was cause the corrosion process along the lower pipe
launched developing Achimov deposits new for PJSC generatrix.
Gazprom. Achimov deposits are characterized by At the facilities of Gazprom dobycha Urengoy LLC
abnormally high reservoir parameters: pressure is up to exposed to carbon dioxide corrosion a corrosion
60 MPa, and the temperature is up to 106°C. monitoring system is implemented, which allows to
After 5 years of operation the first corrosive damage measure the corrosion rate in different parts of the gas
to the inner surface of the wellhead equipment elements collection system [4,5]. According to the measurement
and pipelines were identified. Figure 1 shows one of the results, the following pattern was noted. In areas up to
defects. The cause of these defects was carbon dioxide the pressure regulator, characterized by high
corrosion. temperatures and pressures, the rate of corrosion flow is
higher than in the areas after the pressure regulator,
where thermobaric parameters are lower. This fact is
explained by the fact that the rate of carbon dioxide
corrosion depends on the temperature and partial
pressure of the gas, which in turn depends on the
operating pressure.
A large number of theoretical, empirical and semi-
empirical models exist to describe the corresponding
dependencies [6,7]. The model de Waard-Milliams is
considered the most popular and - already - classic [8,9].
However, the use of this model for the conditions of
Fig. 1. Photo of corrosion defect of a tool flange.
Achimov deposits gives excessive corrosion rates
relative to the actual corrosion rates. Based on this, the
The partial pressure of carbon dioxide at the authors of the article faced the task of developing their
wellhead exceeds 0.21 MPa. Thus, the environment has own model describing the rate of carbon dioxide
a high degree of aggressiveness in accordance with corrosion in relation to the conditions of Achimov
various regulatory documents (NACE SP 0106-2006, deposits.
GOST R 51365-2009, SТО Gazprom 9.3-011-2011)
[1-3]. In gas collecting pipelines, the partial pressure of
*
Corresponding author: [email protected]
© The Authors, published by EDP Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0
(http://creativecommons.org/licenses/by/4.0/).
E3S Web of Conferences 121, 01019 (2019) https://doi.org/10.1051/e3sconf/201912101019
Corrosion in the Oil & Gas Industry 2019
2 Materials and Method
Percentage of explained
To carry out a multi-factor analysis assessing the impact 40 38.5%
variances
of various parameters on the corrosion rate, the 30
following data were selected: the results of chemical 18.9%
analyses of formation water samples, the results of gas 20
condensate studies, the average values of pressure, 10.6%
temperature and flow rates for each well, corrosion rates 10 7.8%
obtained by gravimetric method. All data were
summarized in a matrix consisting of 72 rows and 28 0 1 2 3 4 5 6 7 8 9 10
columns, where 72 is the number of observations; 28 - Dimensions
the number of factors that could have a possible impact Fig. 2. Main components with percentage of total variation in
on the corrosion rate. data.
The generated data array has undergone a
preprocessing procedure, including: Further, the contribution (in percent) of each of the
1. Elimination of missing observations. As part of analyzed factors to the first two main components was
this operation, we removed from the data set estimated, in other words, how strongly the analyzed
observations, in which at least one of the factors was factor affects the main components. After that, a
missing. For example, wells that lack data on the correlation map was drawn in the software system
chemical composition of water were excluded from the (Figure 3).
data set.
2. Transformation of qualitative features into
1.0
numerical (binary) ones. For example, one of the layers
from which the fluid was extracted was assigned the
value 0, the other layer – 1.
3. Application of bootstrapping method. Due to the 0.5
Dim2 (18.9%)
large amount of data available for multi-factor analysis,
the bootstrapping method was used to artificially
increase the amount of information and improve the 0.0
quality of the future model [10,11]. The essence of the
method consists of forming a set of samples based on a
random selection with repetitions and ultimately the -0.5
entire sample assumes the normal form of distribution,
which allowed further use of standard methods of
mathematical statistics and data analysis. -1.0
After preliminary data processing for each analyzed
-1.0 -0.5 0.0 0.5 1.0
factor, mean values and mean square deviation
Dim1 (38.5%)
(hereinafter referred to as MSD) were found. For some Fig. 3. Correlation map.
factors, the MSD exceeded the average value of the
factor, which indicates a large level of spread in values. The correlation map indicates positively correlated
The next step was to conduct a factor analysis, the factors (grouped together), negatively correlated (located
purpose of which is to find the factors that most affect in opposite quadrants), as well as the level of variability
the corrosion rate. For its implementation, the principal and significance of factors: the location closer to the
component method was chosen as the most frequently edge of the circle indicates greater variability, while the
used due to ease of use and transparency. This method location closer to the center indicates lesser variability.
was used to exclude the maximum number of factors The obtained diagram shows that the corrosion rate in
from the model, because a large number of variables in 2016 (Vcorr16) and in 2017 (Vcorr17) has a relationship
the mathematical model makes it variable, and the with the temperature (Ту) and flow rate (Qpl), also has a
reliability of such a model may be in doubt. positive correlation with the first main component with
In the RStudio software product, a diagram was built the operating pressure (Pу) and, accordingly, the partial
(Figure 2), estimating the contribution of the total pressure of carbon dioxide (PСО2). Some other factors
variance in each of the main components. that have a positive correlation with the corrosion rate
The higher the resulting column, the more variations (water content – QH2O, the content of hydrocarbon
include the main component being analyzed. The main components C5+в – PС5В) have an overestimated mean
components are ordered in descending order. Thus, the square deviation and cannot be included in the model.
first two main components account for about 60% of the Thus, the hypothesis that the maximum influence on the
total number of variations. corrosion rate is exerted by the temperature, pressure and
flow rate of wells was previously formulated. To
confirm this hypothesis, we started the construction of a
multidimensional regression model.
2
E3S Web of Conferences 121, 01019 (2019) https://doi.org/10.1051/e3sconf/201912101019
Corrosion in the Oil & Gas Industry 2019
Further, also using RStudio software, a model was theoretical distribution is well suited with the observed
compiled. For this, all factors included in the original data.
matrix were used. The result of this work was a
multidimensional regression model presented in the form
of a table indicating the regression coefficients, MSD for 6
Standardized residuals
11
each factor, the values of the Student's t-test, p-values 22
(significance level) for each criterion. All of the above 4
criteria allow us to estimate the statistical significance of 111
each factor and further exclude the least significant 2
factors from the model.
On the basis of the obtained model, it was decided to 0
reduce the model, in particular, all factors were excluded
from the model except for the temperature, pressure and -2 -1 0 1 2
flow rate of the well. After that, a model was built that Theoretical Quantiles
includes only the above factors (Table1). lm(Vcorr~T+P+Q)
Table 1. The results of calculating the parameters of the linear Fig. 5. Q-Q diagram of the model before detecting anomalous
regression model in the first approximation. values.
Estimate Std. Error t value Pr(>|t|) A very close correlation with the line, except for the
Intercept -0.3318 0.0911 -3.642 0.0004 anomalous points (№ 11, 22, 111), indicates that the
Temperature 0.0098 0.0029 3.358 0.0011 distribution of samples is normal.
Pressure 0.0021 0.0029 0.728 0.4680 The third diagram (Figure 6) shows the dependence
Flow rate 0.0021 0.0030 0.684 0.4953 of the degree of influence of each observation on the
result of the corrosion rate on the error for each
observation.
The next stage of the study was to assess the
reliability of the model. In this stage, we built three
diagrams. 6
Standardized residuals
Using the Gauss – Markov theorem, an array of data 11 1
22
was formed, on which the first diagram was built (Figure 4 0.5
4), which allows us to evaluate the correctness of the 111
choice of the model type. 2
0
1.0 Cook’s distance
11 -2
0.8 22
0.6 0.00 0.02 0.04 0.06 0.08 0.10 0.12
Residuals
1 Leverage
0.4
lm(Vcorr~T+P+Q)
0.2
Fig. 6. Graphic interpretation of Cook’s distance for the model.
0.0
-0.2 In addition, the diagram shows a dotted line of
0.0 0.1 0.2 0.3 Cook's distance (in the upper right corner). The closer to
Fitted values these lines the observation is, the more this observation
lm(Vcorr~T+P+Q) shifts the predicted corrosion rate from the actual value.
Fig. 4. The dependence of predicted values of rate of corrosion In accordance with this diagram, there are anomalous
on the standardized residues. observations (№ 11, 22, 111), which coincide with the
previous two diagrams.
The resulting dependence indicates that the model Further, using the Cook’s distance determination
type (linear relationship) was chosen correctly. In the method [12], anomalous corrosion rate values were
case of a different nature of the desired dependence, the revealed, creating a high variability of the regression
trend line (red line) would have a geometry that is model. Cook's distance shows the difference between the
different from the linear one. Also on this diagram, it is calculated coefficients of the regression equation and the
clearly visible that there are anomalous points (No. 1, 11, values that would be obtained by excluding the
22) that do not fall on the model. The greater the spread corresponding observation. If all Cook’s distances are
of points from the trend line, the less adequate is the not the same, which indicates the adequacy of the model,
resulting model. we can assume that this observation shifts the estimates
The following diagram (Figure 5) is a scattering of the regression coefficients. Figure 7 shows the results
diagram of observed and expected (standardized) values of the algorithm in identifying anomalous values for
with a corresponding specified distribution. If the corrosion rates. Numeric designations correspond with
observed values fall on a straight line, then the the index identifier of a well with an anomaly. Through
3
E3S Web of Conferences 121, 01019 (2019) https://doi.org/10.1051/e3sconf/201912101019
Corrosion in the Oil & Gas Industry 2019
this method, four anomalous observations located above 3 Results
the red line were excluded from the model.
Thus the equation describing the dependence of
corrosion rate on the pressure and temperature
10
0.4 conditions for Area 2 of the Achimov deposits in the
Urengoy oil, gas and condensate field can be written in
0.3
Cooks'd
the form proposed by de Waard-Milliams as follows:
0.2 2377
73 lg 𝑉𝑉𝑐𝑐𝑐𝑐𝑐𝑐 = 6.37 − + 0.52 lg 𝑃𝑃𝐶𝐶𝐶𝐶2 (2)
21
0.1 𝑇𝑇
1
0.0 Figures 8-9 show a graphical comparative analysis of
0 20 40 60 80 100 the actually measured corrosion rates with the predicted
Index
Fig. 7. Graphic interpretation of the Cook's distance for the corrosion rates for the two models.
model.
10
The next step was a similar construction of the Actually Vcor
8
Corrosion rate
regression model in the absence of anomalous P(CO2) = 0.25 MPa
observations (Table 2). 6 P(CO2) = 0.20 MPa
P(CO2) = 0.15 MPa
Table 2. The results of calculating the parameters of the linear 4
regression model after eliminating anomalous observations.
2
Estimate Std. Error t value Pr(>|t|) 0
Intercept -0.2011 0.0538 -3.737 0.0003 20 30 40 50 60
Temperature 0.0064 0.0017 3.633 0.0004 Temperature, °С
Pressure 0.0030 0.0018 1.685 0.0951 Fig. 8. Dependence of the predicted corrosion rate according to
Flow rate -0.0001 0.0018 -0.060 0.9524 the de Waard-Milliams model from the temperature.
Based on the evaluation of the statistical significance 1,0
of each factor, it was decided to exclude the flow rate Actually Vcor
Corrosion rate
0,8
from the model due to the lack of a sufficient level of P(CO2) = 0.25 MPa
significance for the above predictor - 0.9524. After that, 0,6
the model was rebuilt and its reliability was evaluated P(CO2) = 0.20 MPa
0,4
similarly to the original model. After rebuilding, the P(CO2) = 0.15 MPa
0,2
dispersion of the actual values from the predicted
corrosion rates visually decreased, which is an indicator 0,0
of the adequacy of the model obtained. 20 30 40 50 60
It is known that in the classical equation of an Temperature, °С
estimation of carbonic acid corrosion of de Waard- Fig. 9. Dependence of the predicted corrosion rate of the
Milliams (1) there are the same variables as in the model developed model on temperature.
obtained by authors. The only difference is that in the de
Waard-Milliams equation, the pressure is represented as 4 Discussion
the partial pressure of carbon dioxide but because the
CO2 content is the same for all wells, respectively, the In the model, all members are statistically significant
operating pressure is directly proportional to the partial predictors. The value of the determination coefficient
pressure of CO2. (R2) for the obtained model is 0.47. Taking into account
the fact that all the initial data were obtained in field
1710 conditions, the model can be considered sufficiently
lg 𝑉𝑉𝑐𝑐𝑐𝑐𝑐𝑐 = 5.8 − + 0.67 lg 𝑃𝑃𝐶𝐶𝐶𝐶2 (1)
𝑇𝑇 adequate.
The model developed by the authors describes the
Therefore, the next stage of the work was the processes of carbon dioxide corrosion in relation to
construction of a model according to the type of the de objects of Achimov deposits with a better correlation
Waard-Milliams equation. Refined model parameters are than the de Waard-Milliams model, despite the fact that
presented in Table 3. the de Waard-Milliams equation has a theoretical basis
on which the results of laboratory studies are imposed.
Table 3. Parameters of the regression model built by the type
of the de Waard-Milliams equation. The resulting equation is of practical importance at
the stage of commissioning of new wells to predict
Estimate Std. Error t value Pr(>|t|) corrosion rates until it is possible to actually measure the
Intercept 6.37 1.9966 3.191 0.0019 corrosion rate.
1/T -2377 612.0577 3.885 0.0001
ln(P) 0.52 0.2974 1.754 0.0826
4
E3S Web of Conferences 121, 01019 (2019) https://doi.org/10.1051/e3sconf/201912101019
Corrosion in the Oil & Gas Industry 2019
5 Conclusion 7. R. Nyborg, CO2 corrosion models for oil and gas
production systems, Corrosion/2010, Paper No.
1. A model describing the processes of carbon 10371 (Texas, TX: NACE International, 2010)
dioxide corrosion in relation to objects of Achimov 8. C. de Waard, D.E. Milliams, Carbonic Acid
deposits of the Urengoy oil, gas and condensate field corrosion of steel, Corrosion 31, No. 5, p. 175 (1975)
was developed;
2. To identify the main factors affecting the corrosion 9. C. de Waard, U. Lotz, Prediction of CO2 corrosion of
rate, a statistical analysis was carried out using the carbon steel, Corrosion/1993, Paper No. 69
principal component method and the construction of (Houston, TX: NACE International, 1993)
multidimensional regression models. It is established 10. S. Streukens, S. Leroi-Werelds, Bootstrapping and
that the factors most correlated with the corrosion rate PLS-SEM: A step-by-step guide to get more out of
for the conditions of Achimov deposits are temperature your bootstrap results, European Management
and pressure; Journal 34 (6), p. 618-632 (2016)
3. An anomaly search algorithm was implemented 11. S. Basu, Improved small sample inference
based on the Cook's Distance determination. Using this procedures for epidemiological parameters under
algorithm, four observations were excluded from the cross-sectional sampling. Journal of the Royal
total sample; Statistical Society: Series D (The Statistician) 50 (3),
4. After exclusion of the anomalous wells an updated p. 309-319 (2001)
regression model was built by the type of de Waard- 12. Cook R. Dennis, Detection of Influential
Milliams equations; Observations in Linear Regression, Technometrics
5. The proposed model describes the processes of 19, No. 1, p. 15-18 (1977)
carbon dioxide corrosion more accurately than the de
Waard-Milliams equation;
6. The disadvantage of the developed model is that it
does not describe the corrosion rate in wells with
abnormal values of the corrosion rate.
List of symbols
Vcor corrosion rate, mm/y
PCO2 partial pressure, bar
T temperature, K
References
1. NACE Standard SP 0106 (latest revision), “Control
of internal corrosion in steel pipelines and piping
systems” (Houston, TX: NACE)
2. GOST R 51365-2009 Oilfield Mining wellhead.
General specifications (Standartinform, Moscow,
2015)
3. System standarts of Gazprom 9.3-011-2011.
Corrosion protection. Inhibitory protection against
corrosion of field objects and pipelines. Primary
requirements (Gazprom Expo, Moscow, 2011)
4. V. Yu. Artemenkov, A. Yu. Koryakin, D. V.
Dikamov, E. O. Shishkov, A. D. Yusupov,
Organization of corrosion monitoring at the sites of
the second section of the Achimov deposits of the
Urengoy oil and gas condensate field, GAS Industry
of Russia, Spesial Issue No. 2, p. 74-78 (2017)
5. A. Yu. Koriakin, V. F. Kobychev, I. V. Kolinchenko,
A. D. Yusupov, Conditions of the carbon dioxide
corrosion on the production facilities of achimovskie
deposits, methods of monitoring and forecasting,
GAS Industry of Russia, No. 12, p. 84-89 (2017)
6. S. Nesic, Key issues related to modelling of internal
corrosion of oil and gas pipelines, A review,
Corrosion Science 49, p. 4308-4338 (2007)
