A method for detecting the minimum inner diameter of a casing

Abstract

The present invention relates to a method for detecting the minimum inner diameter of a casing, belonging to the technical field of well logging in oil and gas field engineering. First, a well for measuring casing deformation is made, and the minimum inner diameters of casings with different deformation degrees in the well for measuring casing deformation are measured. Then, a tubing string is lowered into the casing of the well for measuring casing deformation, and an electromagnetic logging tool is arranged in the tubing string to measure the pulsed eddy current response signals at different well depths in the well for measuring casing deformation. Next, a data model between the pulsed eddy current response signals and the minimum inner diameter of the casing is established by using a machine learning algorithm. Through the way of sample training, a relatively accurate data model between the pulsed eddy current response signals and the minimum well diameter of the casing is obtained, and the non-linear relationship between the pulsed eddy current response signals and the minimum well diameter of the casing is more accurately fitted. Finally, the pulsed eddy current response signals of the casing of the well to be measured are brought into the above model, so as to accurately obtain the minimum inner diameter value of the casing without removing the tubing string.

Description

Technical Field

[0001] The present invention relates to a method for detecting the minimum inner diameter of a casing, belonging to the technical field of well logging in oil and gas field engineering. Background Art

[0002] In modern oil and gas well industry, the downhole string structure of oil and gas production wells mainly includes casings and tubing. Among them, casings are used to maintain the stability of the wellbore and separate oil and gas layers, and are cemented and sealed with the wellbore using cement slurry; the tubing, as the oil and gas transmission channel, is located inside the casing. Affected by factors such as formation stress, perforation, production layer pressure change, and downhole fluid corrosion, the casing, as a protective layer, often has various degrees of defects, such as extrusion, deformation, etc. These defects will directly affect the safety of downhole operations, oil and gas production, and production efficiency.

[0003] The minimum inner diameter of the casing is a key parameter in the detection of casing deformation in oil and gas wells. Its value directly reflects the degree of deformation of the casing and is an important basis for evaluating the integrity of the oil and gas wellbore. Commonly used downhole string minimum inner diameter detection methods at home and abroad include ultrasonic downhole video logging, eagle eye video logging, multi-arm caliper logging, and eddy current electromagnetic logging.

[0004] Among them, ultrasonic downhole video logging, eagle eye video logging, and multi-arm caliper logging belong to caliper imaging logging methods, which respectively adopt the measurement principles of ultrasonic, optical, and mechanical arm expansion, and can only measure a single layer of tubing or casing. In oil and gas production wells, the tubing is placed inside the casing. Due to the limitation of the well structure, when testing the inner diameter deformation of the casing, the tubing needs to be removed first so that the instrument can directly contact the inner wall of the casing for measurement. Generally, the depth of the tubing in oil and gas wells exceeds 3000m. The operation of removing the tubing has a long construction period and high cost. Especially under high-pressure and high-sulfur conditions, permanent packers are used in the downhole string. When removing the tubing, in order to avoid hydrogen sulfide leakage, a series of measures such as well killing, temporary plugging of the oil and gas layer, and milling the packer need to be taken for operation. The process is complex, the construction time is long, and the operation risk is high. There may even be situations such as the tubing and downhole tools falling into the well, the production layer being killed, and being unable to resume production in the later stage, resulting in limited application of logging technologies such as ultrasonic downhole video logging, eagle eye video logging, and multi-arm caliper logging.

[0005] Eddy current electromagnetic logging is based on the principle of electromagnetic mutual induction and adopts an indirect measurement method. An electromagnetic coil is used to excite a group of bipolar current signals to form a primary magnetic field around the coil. When this magnetic field encounters a ring-shaped medium, an eddy current ring is generated, and then a secondary magnetic field is formed. By receiving the magnetic field signal during the interval of the transmitted excitation signal, according to the distribution and magnitude of the eddy current electromagnetic field in the tubing and casing, the defect conditions of the single-layer tubing structure of the tubing and casing and the double-layer tubing structure of the tubing + casing can be measured. It is an effective method for detecting casing defects in oil and gas wells at present, especially having obvious advantages in detecting casing defects in high-sulfur gas wells. According to the measurement principle, the test data directly collected by the eddy current electromagnetic logging tool can only characterize the magnitude of the electromagnetic field of the casing defect and cannot directly obtain the minimum inner diameter of the casing. It is necessary to perform feature extraction and processing calculations on the test data to achieve quantitative calculation of the minimum inner diameter of the casing, providing a reliable basis for formulating measures such as later workover, operation, and well sealing.

[0006] For example, in the invention patent with the application number 201811217762.1 and the invention name "A method for detecting the deformation degree of the inner diameter of the downhole tubing casing", by establishing a standard well for casing deformation and using the least squares linear fitting method, a mathematical model between the inner diameter size value of the casing at different well depths and the electromagnetic response characteristics of the casing at the corresponding well depths is obtained, realizing the quantitative detection of the deformation degree of the inner diameter of the casing without removing the tubing. However, in this method, a linear fitting method is adopted, while the pulsed eddy current response signal and the minimum well diameter of the casing are actually non-linear relationships. And when establishing the fitting model, only the maximum electromagnetic response value and the inner diameter size of each casing deformation section are used to establish the fitting relationship, without considering the electromagnetic response characteristics of the entire casing deformation section, resulting in the measured inner diameter of the downhole tubing casing not conforming to the actual inner diameter of the downhole tubing casing. Summary of the Invention

[0007] The purpose of the present invention is to provide a method for detecting the minimum inner diameter of a casing to solve the problem that the current measurement results are inaccurate when using a linear fitting relationship for minimum inner diameter detection.

[0008] The present invention provides a method for detecting the minimum inner diameter of a casing to solve the above technical problems. The detection method includes the following steps:

[0009] 1) Statistically analyze the casing deformation types and deformation degrees of each measured well in the study area. According to the casing deformation types and deformation degrees of each measured well in the study area, make several deformed casings with different deformation types and different deformation degrees to form a casing deformation well, and measure the minimum inner diameters of the casings with different deformation degrees in the casing deformation well;

[0010] 2) Lower a tubing into the casing of the casing deformation well, and set an electromagnetic logging tool in the tubing to measure the pulsed eddy current response signals at different well depths in the casing deformation well;

[0011] 3) Establish a non - linear relationship model representing the relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing using a machine learning algorithm. The non - linear relationship model is trained with samples composed of the minimum inner diameters of casings with different deformation degrees in the deformed casing well and the corresponding pulsed eddy current response signals;

[0012] 4) Lower an electromagnetic logging tool into the downhole tubing to be measured, measure the pulsed eddy current response signals at different well depths, and input the pulsed eddy current response signals at different well depths of the downhole tubing to be measured into the non - linear relationship model to obtain the minimum inner diameter of the casing corresponding to each well depth of the downhole tubing to be measured.

[0013] The present invention uses a machine learning algorithm to obtain a data model between the pulsed eddy current response signal and the minimum inner diameter of the casing. Due to the non - linear relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing, the machine learning algorithm can, through sample training, obtain a relatively accurate data model between the pulsed eddy current response signal and the minimum inner diameter of the casing, more accurately fit the non - linear relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing, and achieve accurately obtaining the minimum inner diameter value of the casing without removing the tubing.

[0014] Further, to more accurately describe the relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing, the machine learning algorithm is a regression random forest algorithm, and the non - linear relationship model is obtained by fitting with classification and regression trees.

[0015] Further, to avoid the problem that the electromagnetic logging tool is interfered and the measurement is inaccurate, the method further includes performing baseline drift correction on the measured pulsed eddy current response signal.

[0016] Further, the baseline drift correction uses the least - squares method.

[0017] Further, to obtain complete three - dimensional electromagnetic information in different well depths of the deformed casing well, the electromagnetic logging tool uses an eddy current electromagnetic logging tool with three - direction probes.

[0018] Further, to reduce the dimension of the original signal, retain the vast majority of information in the signal, and eliminate the mutual influence between some dimensions of the signal, the method further includes using the principal component analysis method to extract features from the signals of each probe of the eddy current electromagnetic logging tool at each depth, selecting the first principal component of each probe as the statistical feature, combining the first principal components of each probe to obtain a combined feature, and using the combined feature as the pulsed eddy current response signal corresponding to that depth.

[0019] Further, the deformed casing well is formed by connecting a number of deformed casings with different deformation types and different deformation degrees in series and then lowering them into the measured well. Description of the Drawings

[0020] Figure 1 is the flowchart of the method for detecting the minimum inner diameter of the casing of the present invention;

[0021] Figure 2 is a schematic diagram of the measured data of the minimum inner diameter of the casing at each well depth sample point in the casing deformation well of the embodiment of the present invention;

[0022] Figure 3 is a schematic structural diagram of an eddy current electromagnetic logging tool with three-direction probes used in the embodiment of the present invention;

[0023] Figure 4 is a schematic diagram for comparing the minimum inner diameter detection result of the well to be measured with the corresponding multi-arm caliper measurement result in the embodiment of the present invention;

[0024] Among them, 1 is the longitudinal probe A, 2 is the first transverse probe B, 3 is the second transverse probe C, 4 is the auxiliary structure, and 5 is the centralizer. Detailed implementation manners

[0025] The following further describes the detailed implementation manners of the present invention with reference to the accompanying drawings.

[0026] The present invention first fabricates a casing deformation well and measures the minimum inner diameter of the casing with different deformation degrees in the casing deformation well; then, an oil pipe is lowered into the casing of the casing deformation well, and an electromagnetic logging tool is arranged in the oil pipe to measure the pulsed eddy current response signals at different well depths in the casing deformation well; then, a data model between the pulsed eddy current response signals and the minimum inner diameter of the casing is obtained by using a machine learning algorithm. Due to the non-linear relationship between the pulsed eddy current response signals and the minimum inner diameter of the casing, the machine learning algorithm can, through the way of sample training, obtain a relatively accurate data model between the pulsed eddy current response signals and the minimum inner diameter of the casing, and more accurately fit the non-linear relationship between the pulsed eddy current response signals and the minimum inner diameter of the casing, so as to accurately obtain the value of the minimum inner diameter of the casing without removing the oil pipe. The detection process of this method is as Figure 1 shown and includes the following steps:

[0027] 1. Fabricate a number of deformed casings with different deformation types and different deformation degrees to form a casing deformation well.

[0028] Statistically analyze the casing deformation types and degrees of deformation of each measured well in the study area. Through statistical analysis of the historical well diameter monitoring data of each measured well in the study area, including the interpretation results of historical well diameter data and historical well repair construction records, as well as the geological data of the study area, etc., obtain the casing deformation types and degrees of deformation of each measured well in the study area, and form simulated casing deformation wells. According to the casing deformation types and degrees of deformation of each measured well in the study area, fabricate a number of deformed casings with different deformation types and degrees of deformation to form casing deformation wells, and connect them in series to form a simulated casing. Lower the simulated casing into the measured wells (i.e., real wells) in the study area to form simulated casing deformation wells.

[0029] As another implementation method, a number of deformed casings with different deformation types and degrees of deformation can also be individually lowered into the measured wells in the study area without being connected in series.

[0030] 2. Measure the minimum inner diameter of the casings with different degrees of deformation in the casing deformation well.

[0031] In this embodiment, the multi-arm caliper logging method is used to measure the inner diameter size values of the casings at different well depths in the simulated casing deformation well. Among them, the well diameter measuring device used in the multi-arm caliper logging method is multiple groups of mechanical arms that expand outward from the center of the instrument. Using the multi-arm curves measured by the multiple groups of mechanical arms, obtain the inner diameter size values of the casings at different well depths.

[0032] In this embodiment, the MFC-24 twenty-four-arm caliper logging tool is used to measure the inner diameter size values of the casings at different well depths in the casing deformation well, and use R = min 0≤i≤24 r i (h) to further obtain the minimum inner diameter of the pipe string casing at the corresponding well depth. Among them, h represents the well depth, with the unit of m; R represents the minimum inner diameter of the casing at a well depth of h, with the unit of mm; r i represents the inner diameter value of the casing measured by this multi-arm measurement, with the unit of mm. The measurement data of the minimum inner diameter of the casing at each sample point is as Figure 2 shown; it can be seen from Figure 2 that since the minimum inner diameter of the casing in the casing deformation well will only change significantly when there are defects such as deformation and bending in the casing, the distribution of the minimum inner diameter data of the casing is uneven.

[0033] As another implementation method, other well diameter imaging logging methods can also be used to measure the inner diameter size values of the casings at different well depths in the simulated casing deformation well, such as: ultrasonic downhole video logging method or eagle eye video logging method.

[0034] 3. Lower a tubing string into the simulated casing in the simulated casing deformation well, and lower an eddy current electromagnetic logging tool into the tubing string to measure the electromagnetic response characteristics of the casings at different well depths in the simulated casing deformation well, and perform preprocessing and feature extraction.

[0035] 3.1 In the present invention, the tubing is centered and lowered into the simulated casing of the casing deformed well. In order to obtain the complete three-dimensional electromagnetic information in space at different well depths of the casing deformed well, an MTD-J eddy current electromagnetic logging tool is lowered into the tubing to measure the pulsed eddy current response signals at different well depths of the casing deformed well. The eddy current electromagnetic logging tool used has three direction detection probes, A, B, and C. The spatial distribution of the three direction probes used is as Figure 3 shown. The spatial axes of the first transverse probe B2 and the second transverse probe C3 are horizontal and perpendicular to each other, and the spatial axis of the longitudinal probe A1 is horizontal and vertical. The eddy current electromagnetic logging tool also includes an auxiliary structure 4 and a centralizer 5.

[0036] 3.2 The signals obtained in step 3.1 are preprocessed by a baseline drift removal method based on the least squares method to obtain the preprocessed pulsed eddy current signals. The input is the collected original pulsed eddy current signals. The sample data in this embodiment contains 24-dimensional data, including the depth data X, and the pulsed eddy current signal data contains 23-dimensional data. The data dimensions from different detection probes A, B, and C are 11 dimensions, 6 dimensions, and 6 dimensions respectively. U = [U1…U 23 . For the baseline drift existing in the eddy current signals at different depths of the 23-dimensional pulsed eddy current signals, the least squares fitting method (a, b) = argmin a,b (‖aX + b - U‖ 2 ) is used to calculate the vectors a and b, realizing the removal of baseline drift;

[0037] 3.3 The preprocessed pulsed eddy current signals obtained in 3.2 are obtained. The pulsed eddy current signal data contains 23-dimensional data. The data dimensions from different detection probes A, B, and C are 11 dimensions, 6 dimensions, and 6 dimensions respectively. The data is expressed as the following matrix:

[0038]

[0039] Among them, U a , U b , U c represent the signal components measured by the A, B, and C probes respectively; the principal component analysis method is used for the signal components of different probes to extract the first principal component components and combine them as the combined feature quantity D = [D1, D2, D3]. Among them, D1, D2, and D3 represent the first principal component components obtained by the principal component analysis of the eddy current signals U a , U b , U c of the A, B, and C probes respectively.

[0040] The present invention utilizes the different detection capabilities of each probe to extract features from the signals of each probe respectively using the principal component analysis method, and selects the first principal component of each probe that can fully characterize the minimum inner diameter information of the casing as the statistical feature, and combines them to obtain the combined feature. Selecting the first principal component of each probe to obtain the combined feature not only reduces the dimension of the original signal, but also retains the vast majority of the information in the signal, and at the same time eliminates the mutual influence between some dimensions of the signal. The above combined feature fully characterizes the casing state information and provides an effective feature for the quantitative analysis of the minimum inner diameter of the casing.

[0041] 4. Construct a quantitative analysis data model between the pulsed eddy current response signal and the minimum inner diameter of the casing.

[0042] In this embodiment, a quantitative analysis data model between the pulsed eddy current response signal and the minimum inner diameter of the casing is constructed based on the regression random forest method. The training set data is the combined feature quantity extracted from 3201 groups of pulsed eddy current response signals and the minimum inner diameter value of the casing at the corresponding well depth. The training data set E = {(U1, R1), (U2, R2),..., (U 3201 , R 3201 )}, where U i represents the combined feature quantity at a certain depth of the casing deformation well, and R i represents the minimum inner diameter value of the casing corresponding to this well depth. Based on the random forest regression method, a quantitative analysis data model between the pulsed eddy current response signal and the minimum inner diameter of the casing is constructed through training: the combined feature quantity U i of the sample at a certain depth and the minimum inner diameter R i corresponding to this sample are input into the regression random forest algorithm, and the algorithm fits the non-linear relationship between the combined feature quantity and the minimum inner diameter of the input sample by establishing a classification regression tree; after inputting 3201 groups of samples for training, a mathematical model of the non-linear relationship between the pulsed eddy current response signal U and the minimum inner diameter R of the casing based on the regression random forest algorithm is established.

[0043] The present invention uses random forest regression for quantitative analysis detection to obtain a data model between the pulsed eddy current response signal and the minimum inner diameter of the casing. Due to the non-linear relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing, the random forest method can obtain a relatively accurate data model between the pulsed eddy current response signal and the minimum inner diameter of the casing by training a classifier containing multiple decision trees, and more accurately fit the non-linear relationship between the pulsed eddy current response signal and the minimum inner diameter of the casing.

[0044] As other embodiments, the random forest algorithm in this embodiment can also adopt other machine learning algorithms, such as neural network and other algorithms.

[0045] 5. Lower an electromagnetic logging tool into the downhole tubing to be measured, measure the pulsed eddy current response signals at different well depths, and input the pulsed eddy current response signals at different well depths of the downhole tubing to be measured into the non-linear relationship model to obtain the minimum inner diameter of the casing corresponding to each well depth of the downhole tubing to be measured.

[0046] In this embodiment, before salvaging the tubing in Well D405-3 of a certain gas field, the pulsed eddy current response of the casing at different well depths of the casing deformation well was measured in the tubing, and a total of 1067 sets of on-site experimental data were obtained. Using the quantitative analysis data model obtained in Step 4, the calculation results of the minimum inner diameter of the casing at the corresponding well depths of the casing deformation well were obtained.

[0047] After obtaining the calculation results of the minimum inner diameter of the casing at the corresponding well depths of the casing deformation well by using the method of the present invention, after salvaging the tubing in this well, then directly measure the minimum inner diameter of the casing at the corresponding well depths after salvaging the tubing in this well by using the multi-arm caliper logging method, and obtain the comparison chart of the minimum well diameter calculation results of this well and the measured data results of the multi-arm caliper, as Figure 4 shown. It can be seen that the calculation error between the minimum well diameter calculation results obtained by the method of the present invention and the multi-arm caliper data is basically less than 10 mm.

[0048] In the embodiment of the present invention, the average prediction error of 1067 sample points is 1.69 mm, and there are 1043 sample points with an error less than 10 mm, accounting for 97.75% of the total sample points; comparing with the application number 201811217762.1, the invention name is "A method for detecting the deformation degree of the inner diameter of downhole tubing casing", the present invention can make the calculation error of the minimum inner diameter of the casing at the same depth of the same well < 5%. Therefore, the method of the present invention can more accurately fit the non-linear relationship between the pulsed eddy current response signal and the minimum well diameter of the casing, realizing an average prediction error of the minimum inner diameter of the sample points < 2%. And even if the data distribution of the minimum inner diameter of the casing is uneven, the method of the present invention still has good generalization ability. Among all sample points, the sample points with an error less than 10 mm account for 97.75% of the total sample points; the method of the present invention can more accurately quantitatively calculate the minimum well diameter value of the well casing and has good application prospects.

[0049] Those of ordinary skill in the art can understand that the above are only preferred examples of the invention and are not used to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, for those skilled in the art, they can still modify the technical solutions described in the foregoing examples, or perform equivalent replacements for some of the technical features. All modifications, equivalent replacements, etc. made within the spirit and principle of the invention shall be included within the protection scope of the invention.

Claims

1. A method for detecting the minimum inner diameter of a casing, characterized in that, The detection method includes the following steps: 1) Statistically analyze the casing deformation types and deformation degrees of each measured well in the study area. According to the casing deformation types and deformation degrees of each measured well in the study area, fabricate a number of deformed casings with different deformation types and different deformation degrees. After connecting a number of deformed casings with different deformation types and different deformation degrees in series, lower them into the measured well to form a simulated casing deformation well, and measure the minimum inner diameter of the casings with different deformation degrees in the simulated casing deformation well; 2) Lower a tubing string into the casing of the simulated casing deformation well, and install an electromagnetic logging tool in the tubing string to measure the pulsed eddy current response signals at different well depths in the simulated casing deformation well; 3) Use a machine learning algorithm to establish a non-linear relationship model characterizing the relationship between the pulsed eddy current response signals and the minimum inner diameter of the casing. The non-linear relationship model is trained with samples composed of the minimum inner diameter of the casings with different deformation degrees in the simulated casing deformation well and the corresponding pulsed eddy current response signals; 4) Lower an electromagnetic logging tool into the tubing string of the well to be measured, measure the pulsed eddy current response signals at different well depths, and input the pulsed eddy current response signals at different well depths in the tubing string of the well to be measured into the non-linear relationship model to obtain the minimum inner diameter of the casing corresponding to each well depth in the tubing string of the well to be measured.

2. The casing minimum inner diameter detection method according to claim 1, wherein The machine learning algorithm is a regression random forest algorithm, and the non-linear relationship model is obtained by fitting with classification and regression trees.

3. The minimum inner diameter detection method of the casing according to claim 1, characterized in that, This method also includes performing baseline drift correction on the measured pulsed eddy current response signals.

4. The method for detecting the minimum inner diameter of the casing according to claim 3, characterized in that, The baseline drift correction adopts the least squares method.

5. The minimum inner diameter detection method of the casing according to any one of claims 1-4, characterized in that The electromagnetic logging tool uses an eddy current electromagnetic logging tool with three-direction probes.

6. The method for detecting the minimum inner diameter of a casing according to claim 5, characterized in that This method also includes performing feature extraction on the signals of each probe of the eddy current electromagnetic logging tool at each depth respectively using the principal component analysis method, selecting the first principal component of each probe as the statistical feature, combining the first principal components of each probe to obtain a combined feature, and using the combined feature as the pulsed eddy current response signal corresponding to that depth.

Images