Pitting Corrosion and Microstructure of J55 Carbon Steel Exposed to CO2/Crude Oil/Brine Solution under 2–15 MPa at 30–80 °C

This study aimed to evaluate the corrosion properties of J55 carbon steel immersed in CO2/crude oil/brine mixtures present in the wellbores of CO2-flooded production wells. The main corroded position of wellbore was determined and wellbore corrosion law was provided. Corrosion tests were performed in 30% crude oil/brine solution under the simulated temperature (30–80 °C) and pressure (2–15 MPa) conditions of different well depths (0–1500 m). The corrosion behavior of J55 carbon steel was evaluated through weight-loss measurements and surface analytical techniques, including scanning electron microscopy, energy dispersive spectrometer, X-ray diffraction analysis, and optical digital microscopy. Corrosion rate initially increased and then decreased with increasing well depth, which reached the maximum value of 1050 m. At this well depth, pressure and temperature reached 11 MPa and 65 °C, respectively. Under these conditions, FeCO3 and CaCO3 localized on sample surfaces. Microscopy was performed to investigate corrosion depth distribution on the surfaces of the samples.


Introduction
CO 2 is internationally recognized as a major greenhouse gas that accounts for approximately 65% of the total greenhouse gas emissions [1][2][3]. All countries currently attach considerable importance to environmental issues, particularly global warming caused by CO 2 . CO 2 is used as an oil-flooding agent worldwide because it can effectively reduce crude oil viscosity and residual oil saturation, dissolve gum in reservoirs, and increase permeability and crude oil recovery rate [4,5]. CO 2 flooding can reduce CO 2 -associated air pollution and greenhouse effects. Nevertheless, oil pipe failure caused by CO 2 corrosion has become a commonly encountered problem in oilfields and results in great economic losses and safety hazards while severely restricting the development of CO 2 flooding technology [4][5][6].
CO 2 corrosion and its control in oil casing and surface transmission pipelines have been important topics in the field of oil and gas exploration. Related studies have focused on the influence of environmental and material factors on corrosion behavior [7][8][9][10][11][12][13][14][15][16]. For example, the corrosive medium and environment have been identified as the deciding factors of corrosion rate and morphology. The influence of temperature and pressure on corrosion rate is mainly reflected by the changes that they induce in the protectiveness of the corrosion product layer [7][8][9][10][11][12][13][14][15][16][17][18]. These changes, in turn, lead to changes in the corrosion rate. With increasing temperature, the CO 2 corrosion rate of carbon steel

Experimental Materials
J55 carbon steel was processed into rectangular samples (50 mm × 10 mm × 3 mm, ϕ = 6 mm) for the weight-loss test and surface analysis. The chemical composition of J55 carbon steel is shown in Table 1. The samples were placed in acetone to remove surface oil and then immersed in ethanol for 5 min for degreasing and dehydration. The samples were collected, dried with cold air, packed in filter paper, and placed in a dryer for 4-7 h. The sizes and weights of the samples were measured within an accuracy of 0.1 mg. The corrosive medium comprised crude oil and brine. Crude oil was collected from the Chang-8 oil reservoir, a certain block in Changqing Oilfield, and the composition of simulated brine was based on the composition of the water produced in the Chang-8 oil reservoir. The compositions of crude oil and brine are shown in Tables 2 and 3, respectively.

Weight-Loss Corrosion Test
The corrosion test was performed in accordance with the weight-loss method with a PARR-4578 autoclave. The schematic of the test is shown in Figure 1. 1 L the mixture of crude oil and brine (v:v = 3:7) was added to the autoclave and purged with a small amount of N 2 for 120 min under 0.5 MPa to remove dissolved O 2 . Then, the mixture was subjected to 60 min of injection with high-purity CO 2 under 1 MPa to remove N 2 [19]. Finally, autoclave temperature was increased to the test temperature, and autoclave pressure was increased to the test pressure with high-purity CO 2 . The test conditions were maintained for 2 days at a running speed of 0.5 m·s −1 (200 r·min −1 ).

Weight-Loss Corrosion Test
The corrosion test was performed in accordance with the weight-loss method with a PARR-4578 autoclave. The schematic of the test is shown in Figure 1. 1 L the mixture of crude oil and brine (v:v = 3:7) was added to the autoclave and purged with a small amount of N2 for 120 min under 0.5 MPa to remove dissolved O2. Then, the mixture was subjected to 60 min of injection with high-purity CO2 under 1 MPa to remove N2 [19]. Finally, autoclave temperature was increased to the test temperature, and autoclave pressure was increased to the test pressure with high-purity CO2. The test conditions were maintained for 2 days at a running speed of 0.5 m·s −1 (200 r·min −1 ). The corrosion rate of the corroded steel was determined through the mass-loss method in accordance with the ASTM G1-03 Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens [20]. The samples were immediately rinsed with distilled water. Acetone was used to remove crude oil from the surfaces of the samples after corrosion induction. Then, the samples were immersed in acid cleaning solution (500 mL of HCl and 3.5 g of hexamethylenamine brought to volume with water to 1000 mL) for 10 min. At the same time, the corrosion products were removed from the surfaces of the samples, and the samples were collected from the acid cleaning solution. The acid cleaning solution was thoroughly rinsed off from the surfaces of the samples with distilled water. The samples were then twice immersed in ethanol for cleaning and dehydration, collected, placed on filter paper, dried with cold air, packed in filter paper, and placed in the dryer for 4-7 h. The corrosion rate of the corroded steel was determined through the mass-loss method in accordance with the ASTM G1-03 Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens [20]. The samples were immediately rinsed with distilled water. Acetone was used to remove crude oil from the surfaces of the samples after corrosion induction. Then, the samples were immersed in acid cleaning solution (500 mL of HCl and 3.5 g of hexamethylenamine brought to volume with water to 1000 mL) for 10 min. At the same time, the corrosion products were removed from the surfaces of the samples, and the samples were collected from the acid cleaning solution. The acid cleaning solution was thoroughly rinsed off from the surfaces of the samples with distilled water. The samples were then twice immersed in ethanol for cleaning and dehydration, collected, placed on filter paper, dried with cold air, packed in filter paper, and placed in the dryer for 4-7 h. Finally, the samples were weighed to within an accuracy of 0.1 mg. Corrosion rate was calculated using the following formula: where r corr is average corrosion rate (mm·y −1 ); m and m t are the weights of the test sheet before and after the experiment, respectively (g); S is the area of the whole surface in contact with the solution (cm 2 ); ρ is the density of the tested steel (g·cm −3 ); and t is the duration of immersion (h). Each test was performed with three parallel samples. The mean corrosion rate error was calculated on the basis of the results for the three parallel samples.

Microstructure Observation
After corrosion induction, samples were extracted from the autoclave and rinsed with distilled water and acetone. The surface microstructures of the corrosion product layers on the surfaces of corroded samples were analyzed through SEM with FEI Quantu 600F microscope (Hillsboro, OR, USA). The elemental compositions of the corrosion product layers were determined with OXFORD INCA energy 350. The compositions of the corrosion product characterized were determined through XRD by using Bruker D8 XRD (Billerica, MA, USA).

Statistics of Corrosion Depth Distribution
The surface depths of the corroded samples cleaned using acid cleaning solution were visualized using an OLYMPUS DSX500 optical digital microscope (Tokyo, Japan). Sample surfaces were subjected to grand horizon 3D image capture under bright-field mode with the adjacent visual threshold splicing mode. As shown in Figure 2, images for the analysis of surface corrosion morphology were acquired at eight observation points on both sides of the same sample through nine-field splicing under 200× magnification and 10% coincidence rate. The display heights of the 3D images were adjusted to the maximum pitting height to ensure that the different ranges of corrosion depths in the same set of images can be represented by different colors. Identical corrosion depth ranges were represented by the same color. Finally, the 3D images were converted into the contour diagrams of corrosion depth distribution.  (1) where rcorr is average corrosion rate (mm·y −1 ); m and mt are the weights of the test sheet before and after the experiment, respectively (g); S is the area of the whole surface in contact with the solution (cm 2 ); ρ is the density of the tested steel (g·cm −3 ); and t is the duration of immersion (h). Each test was performed with three parallel samples. The mean corrosion rate error was calculated on the basis of the results for the three parallel samples.

Microstructure Observation
After corrosion induction, samples were extracted from the autoclave and rinsed with distilled water and acetone. The surface microstructures of the corrosion product layers on the surfaces of corroded samples were analyzed through SEM with FEI Quantu 600F microscope (Hillsboro, OR, USA). The elemental compositions of the corrosion product layers were determined with OXFORD INCA energy 350. The compositions of the corrosion product characterized were determined through XRD by using Bruker D8 XRD (Billerica, MA, USA).

Statistics of Corrosion Depth Distribution
The surface depths of the corroded samples cleaned using acid cleaning solution were visualized using an OLYMPUS DSX500 optical digital microscope (Tokyo, Japan). Sample surfaces were subjected to grand horizon 3D image capture under bright-field mode with the adjacent visual threshold splicing mode. As shown in Figure 2, images for the analysis of surface corrosion morphology were acquired at eight observation points on both sides of the same sample through nine-field splicing under 200× magnification and 10% coincidence rate. The display heights of the 3D images were adjusted to the maximum pitting height to ensure that the different ranges of corrosion depths in the same set of images can be represented by different colors. Identical corrosion depth ranges were represented by the same color. Finally, the 3D images were converted into the contour diagrams of corrosion depth distribution.

Corrosion Law of the Deepening Well
The CO 2 -flooded well in Chang-8 Oil Reservoir of a certain block in Changqing Oilfield was taken as an example. The experimental well had a depth of 1550 m, a well-head temperature of 30 • C, pressure of 2 MPa, and well-bottom temperature of 82 • C. Temperature and pressure distributions as a function of well depth are shown in Figure 3. Tests were performed at 0 m (2 MPa, 30 • C), 240 m (4 MPa, 40 • C), 580 m (7 MPa, 50 • C), 800 m (9 MPa, 55 • C), 1050 m (11 MPa, 65 • C), and 1500 m (15 MPa, 80 • C). The liquid produced by the well had a water cut of 70%. The crude-oil composition of the liquid is shown in Table 2. The chemical composition of water produced by the well is shown in Table 3.  Table 2. The chemical composition of water produced by the well is shown in Table 3.  Figure 4 shows the appearance of samples after the removal of corrosion scales. Figure 5 shows the average corrosion rate and the maximum corrosion depth of the samples corroded at different well depths. The average corrosion rate initially increased and then decreased with well depth, except for 1050 m. The general equations for the anodic and cathodic reactions of CO2 corrosion in deoxygenated solution are shown as Equations (2) and (3), respectively [9].
Temperature and pressure increased as well depth increased. Increasing temperatures reduced the viscosity and protective effect of crude oil on the sample surfaces while intensifying mass transfer between the samples and corrosive medium and accelerating corrosion. Moreover, CaCO3 and FeCO3 deposits generated from the reaction between CO3 2− and HCO3 − and between Ca 2+ and Fe 2+ in the liquid gradually increased and inhibited corrosion development by forming a protective layer on the sample surfaces [8,9]. The increase in CO2 pressure reduced system pH and is conducive for the formation of the protective corrosion product layer [21]. Hence, corrosion acceleration and inhibitory effects simultaneously occurred in the system. At 580 m, the maximum corrosion depth sharply increased, and corrosion type shifted from uniform to local corrosion because the corrosive environment transformed from a CO2/crude oil/brine environment to a subcritical CO2/crude oil/brine environment [22]. The scaling ability of the solution decreased, the surface of the sample could not be completely covered by precipitates, and the progress of the anodic reaction could only be partially prevented [23]. These effects resulted in local corrosion. Therefore, the average corrosion rate was low (1.7658 mm·year −1 ), whereas the maximum corrosion depth was high (164.358 μm). Under increasing temperature and pressure, the corrosive environment transformed to the supercritical CO2/crude oil/brine environment and was dominated by corrosion. Hence, the maximum average corrosion rate was observed at the depth of 1050 m. The possible reduction in the base level of corrosion depth measurement may have reduced the maximum corrosion depths of the  Figure 4 shows the appearance of samples after the removal of corrosion scales. Figure 5 shows the average corrosion rate and the maximum corrosion depth of the samples corroded at different well depths. The average corrosion rate initially increased and then decreased with well depth, except for 1050 m. The general equations for the anodic and cathodic reactions of CO 2 corrosion in deoxygenated solution are shown as Equations (2) and (3), respectively [9].
Temperature and pressure increased as well depth increased. Increasing temperatures reduced the viscosity and protective effect of crude oil on the sample surfaces while intensifying mass transfer between the samples and corrosive medium and accelerating corrosion. Moreover, CaCO 3 and FeCO 3 deposits generated from the reaction between CO 3 2− and HCO 3 − and between Ca 2+ and Fe 2+ in the liquid gradually increased and inhibited corrosion development by forming a protective layer on the sample surfaces [8,9]. The increase in CO 2 pressure reduced system pH and is conducive for the formation of the protective corrosion product layer [21]. Hence, corrosion acceleration and inhibitory effects simultaneously occurred in the system. At 580 m, the maximum corrosion depth sharply increased, and corrosion type shifted from uniform to local corrosion because the corrosive environment transformed from a CO 2 /crude oil/brine environment to a subcritical CO 2 /crude oil/brine environment [22]. The scaling ability of the solution decreased, the surface of the sample could not be completely covered by precipitates, and the progress of the anodic reaction could only be partially prevented [23]. These effects resulted in local corrosion. Therefore, the average corrosion

Microstructures and Compositions of Corrosion Scales
The SEM images of the samples after corrosion at different temperatures and pressures are shown in Figure 6. The results for the spectral analysis of the corroded sample surfaces are shown in

Microstructures and Compositions of Corrosion Scales
The SEM images of the samples after corrosion at different temperatures and pressures are shown in Figure 6. The results for the spectral analysis of the corroded sample surfaces are shown in

Microstructures and Compositions of Corrosion Scales
The SEM images of the samples after corrosion at different temperatures and pressures are shown in Figure 6. The results for the spectral analysis of the corroded sample surfaces are shown in Table 4. Before CO 2 reached a supercritical state, the samples exhibited dense surface coatings that mainly consisted of FeC 3 and FeCO 3 and almost lacked CaCO 3 [14,17,19,24]. The average corrosion rates and maximum corrosion depths of the samples were low. After CO 2 reached the supercritical state, loose surface coatings that mainly consisted of FeCO 3 and CaCO 3 formed and failed to provide effective surface protection to the samples [14,19,24]. Thus, the average corrosion rates and maximum corrosion depths of the samples increased.  Table 4. Before CO2 reached a supercritical state, the samples exhibited dense surface coatings that mainly consisted of FeC3 and FeCO3 and almost lacked CaCO3 [14,17,19,24]. The average corrosion rates and maximum corrosion depths of the samples were low. After CO2 reached the supercritical state, loose surface coatings that mainly consisted of FeCO3 and CaCO3 formed and failed to provide effective surface protection to the samples [14,19,24]. Thus, the average corrosion rates and maximum corrosion depths of the samples increased.    Figure 7 shows the XRD spectra of the surface layers of the corroded samples immersed in CO 2 /crude oil/brine mixtures under the given temperatures and pressures at different well depths. FeCO 3 is the main product of the CO 2 corrosion of carbon steel [7][8][9][10][11][12][13][14][15][16][17][18][19]25]. Similarly, the corrosion product layer that formed on samples immersed in CO 2 /crude oil/brine mixtures mainly comprised CaCO 3 and FeCO 3 complex salts. The specific composition of the product layer may be attributed to the isomorphous substitution of metal cations during CO 2 corrosion [23] Figure 7 shows the XRD spectra of the surface layers of the corroded samples immersed in CO2/crude oil/brine mixtures under the given temperatures and pressures at different well depths. FeCO3 is the main product of the CO2 corrosion of carbon steel [7][8][9][10][11][12][13][14][15][16][17][18][19]25]. Similarly, the corrosion product layer that formed on samples immersed in CO2/crude oil/brine mixtures mainly comprised CaCO3 and FeCO3 complex salts. The specific composition of the product layer may be attributed to the isomorphous substitution of metal cations during CO2 corrosion [23] Figure 7. XRD spectra of the surface layers of the corroded samples. Figure 8 shows the results for the corrosion depth analysis of the sample corroded in 30% crude oil/brine at the well depth of 800 m. Figure 8a,b were obtained through optical digital microscopy under 200× magnification with nine-field splicing and 10% coincidence rate. Figure 8c shows the contour diagram of the corrosion depth distribution of Figure 8a, which was transformed from Figure 8b. The size of one contour diagram of corrosion depth distribution was 7612 μm × 7612 μm,  Figure 8 shows the results for the corrosion depth analysis of the sample corroded in 30% crude oil/brine at the well depth of 800 m. Figure 8a,b were obtained through optical digital microscopy under 200× magnification with nine-field splicing and 10% coincidence rate. Figure 8c shows the contour diagram of the corrosion depth distribution of Figure 8a, which was transformed from Figure 8b. The size of one contour diagram of corrosion depth distribution was 7612 µm × 7612 µm, and the total image area observed was 57.94 mm 2 , which accounted for 42% of the sample surface area. Therefore, this corrosion depth analysis method can accurately reflect the surface conditions of the samples after corrosion. Figure 9 shows the contour diagrams of corrosion depth distribution at different observation positions after the removal of corrosion scales from the sample corroded in 30% crude oil/brine at the well depth of 800 m.

Law of Corrosion Depth Distribution
Materials 2018, 11, x FOR PEER REVIEW 9 of 13 and the total image area observed was 57.94 mm 2 , which accounted for 42% of the sample surface area. Therefore, this corrosion depth analysis method can accurately reflect the surface conditions of the samples after corrosion. Figure 9 shows the contour diagrams of corrosion depth distribution at different observation positions after the removal of corrosion scales from the sample corroded in 30% crude oil/brine at the well depth of 800 m.  The frequency density distribution of corrosion depth on the surface of the sample corroded at the well depth of 800 m is shown in Figure 10. The class intervals used to represent corrosion depth ranges in Figure 10 are the same as colors used to represent corrosion depth ranges in Figure 9. The plot of frequency density was bell-shaped with bilateral symmetry, wherein high values clustered in the center of the plot and low values clustered at both ends of the plot; these characteristics are indicative of typical Gaussian distribution [27,28]. The frequency density distribution maps of samples corroded at different well depths shown in Figure 11 exhibit similar characteristics of Figure 10. Materials 2018, 11, x FOR PEER REVIEW 9 of 13 and the total image area observed was 57.94 mm 2 , which accounted for 42% of the sample surface area. Therefore, this corrosion depth analysis method can accurately reflect the surface conditions of the samples after corrosion. Figure 9 shows the contour diagrams of corrosion depth distribution at different observation positions after the removal of corrosion scales from the sample corroded in 30% crude oil/brine at the well depth of 800 m.  The frequency density distribution of corrosion depth on the surface of the sample corroded at the well depth of 800 m is shown in Figure 10. The class intervals used to represent corrosion depth ranges in Figure 10 are the same as colors used to represent corrosion depth ranges in Figure 9. The plot of frequency density was bell-shaped with bilateral symmetry, wherein high values clustered in the center of the plot and low values clustered at both ends of the plot; these characteristics are indicative of typical Gaussian distribution [27,28]. The frequency density distribution maps of samples corroded at different well depths shown in Figure 11 exhibit similar characteristics of Figure 10. The frequency density distribution of corrosion depth on the surface of the sample corroded at the well depth of 800 m is shown in Figure 10. The class intervals used to represent corrosion depth ranges in Figure 10 are the same as colors used to represent corrosion depth ranges in Figure 9. The plot of frequency density was bell-shaped with bilateral symmetry, wherein high values clustered in the center of the plot and low values clustered at both ends of the plot; these characteristics are indicative of typical Gaussian distribution [27,28]. The frequency density distribution maps of samples corroded at different well depths shown in Figure 11 exhibit similar characteristics of Figure 10 Table 5 shows the fitting parameters for the curve shown in Figure 11. The parameters were obtained by using the Gaussian model (Equation (6)) multicurve global mode in origin 9.0 and a correlation coefficient of 0.9886. Corrosion depths followed Gaussian distribution. The physical interpretation of corrosion depth distribution revealed that y0 = 0 and A = 1 in Equation (6).
where y is the probability density function; y0 is the offset, y0 = 0; A is the area, where A = 1; x is corrosion depth (μm); xc is the expected value (μm); w is twice the standard deviation; xc determines the location of the distribution curve; and w determines the amplitude of the distribution curve. Table 6 shows the results of variance analysis through the Gaussian model used to fit the curve shown in Figure 11. The Prob > F value of less than 0.01 indicates that the frequency density distribution curve in Figure 11 shows Gaussian distribution. In Gaussian distribution, xc indicates the average value of the random variables and represents the average corrosion depth. The corrosion of the sample surface may have reduced the datum plane during image acquisition. Moreover, the trend followed by xc with the change in well depth differed from that followed by the average  Table 5 shows the fitting parameters for the curve shown in Figure 11. The parameters were obtained by using the Gaussian model (Equation (6)) multicurve global mode in origin 9.0 and a correlation coefficient of 0.9886. Corrosion depths followed Gaussian distribution. The physical interpretation of corrosion depth distribution revealed that y0 = 0 and A = 1 in Equation (6).
where y is the probability density function; y0 is the offset, y0 = 0; A is the area, where A = 1; x is corrosion depth (μm); xc is the expected value (μm); w is twice the standard deviation; xc determines the location of the distribution curve; and w determines the amplitude of the distribution curve. Table 6 shows the results of variance analysis through the Gaussian model used to fit the curve shown in Figure 11. The Prob > F value of less than 0.01 indicates that the frequency density distribution curve in Figure 11 shows Gaussian distribution. In Gaussian distribution, xc indicates the average value of the random variables and represents the average corrosion depth. The corrosion of the sample surface may have reduced the datum plane during image acquisition. Moreover, the trend followed by xc with the change in well depth differed from that followed by the average Figure 11. Frequency density distribution map of the corrosion depths of samples corroded at different well depths. Table 5 shows the fitting parameters for the curve shown in Figure 11. The parameters were obtained by using the Gaussian model (Equation (6)) multicurve global mode in origin 9.0 and a correlation coefficient of 0.9886. Corrosion depths followed Gaussian distribution. The physical interpretation of corrosion depth distribution revealed that y 0 = 0 and A = 1 in Equation (6).
where y is the probability density function; y 0 is the offset, y 0 = 0; A is the area, where A = 1; x is corrosion depth (µm); x c is the expected value (µm); w is twice the standard deviation; x c determines the location of the distribution curve; and w determines the amplitude of the distribution curve.  Table 6 shows the results of variance analysis through the Gaussian model used to fit the curve shown in Figure 11. The Prob > F value of less than 0.01 indicates that the frequency density distribution curve in Figure 11 shows Gaussian distribution. In Gaussian distribution, x c indicates the average value of the random variables and represents the average corrosion depth. The corrosion of the sample surface may have reduced the datum plane during image acquisition. Moreover, the trend followed by x c with the change in well depth differed from that followed by the average corrosion rate, especially when the average corrosion rate was high. By adding/subtracting a constant value, the frequency density curve for new random variables generated from random variables with Gaussian distribution was transformed into the translated frequency density curve for former variables in the x-direction without changing the shape of the frequency density curve. Therefore, the absence or presence of the datum plane during image acquisition will not affect the w of the fitting results for Gaussian distribution. The trend followed by w with the change in well depth was the same as that followed by the maximum corrosion depth. This similarity indicates the existence of a strong linear correlation between w and corrosion type, as illustrated in Figure 12. Moreover, w was positively related to the span of the frequency density curve. A small w value represents a density distribution curve with a narrow range and limited corrosion depth distribution. These characteristics indicate uniform corrosion. A high w value represents a density distribution curve with a broad range and corrosion depth with broad distribution. These characteristics indicate local corrosion. corrosion rate, especially when the average corrosion rate was high. By adding/subtracting a constant value, the frequency density curve for new random variables generated from random variables with Gaussian distribution was transformed into the translated frequency density curve for former variables in the x-direction without changing the shape of the frequency density curve. Therefore, the absence or presence of the datum plane during image acquisition will not affect the w of the fitting results for Gaussian distribution. The trend followed by w with the change in well depth was the same as that followed by the maximum corrosion depth. This similarity indicates the existence of a strong linear correlation between w and corrosion type, as illustrated in Figure 12. Moreover, w was positively related to the span of the frequency density curve. A small w value represents a density distribution curve with a narrow range and limited corrosion depth distribution. These characteristics indicate uniform corrosion. A high w value represents a density distribution curve with a broad range and corrosion depth with broad distribution. These characteristics indicate local corrosion.

Conclusions
Based on the observed corrosion behavior of J55 carbon steel in CO2/30% crude oil/brine mixtures under the simulated conditions of different well depths (0-1500 m), we conclude the following: (1) The average corrosion rate of J55 carbon steel initially increased and then decreased in the CO2/crude oil/brine environment as partial CO2 pressure increased. Corrosion type shifted from uniform to local corrosion; (2) The main corrosion products on the surfaces of J55 carbon steel were FeCO3 and CaCO3;

Conclusions
Based on the observed corrosion behavior of J55 carbon steel in CO 2 /30% crude oil/brine mixtures under the simulated conditions of different well depths (0-1500 m), we conclude the following: (1) The average corrosion rate of J55 carbon steel initially increased and then decreased in the CO 2 /crude oil/brine environment as partial CO 2 pressure increased. Corrosion type shifted from uniform to local corrosion; (2) The main corrosion products on the surfaces of J55 carbon steel were FeCO 3 and CaCO 3 ; (3) The distribution of corrosion depth obeyed Gaussian distribution, and w was positively correlated with the maximum corrosion depth.