Intelligent indicator diagram production calculation method applicable to all working conditions

Abstract

The invention provides an intelligent indicator diagram production calculation method applicable to all working conditions, and mainly solves the problem of unstable production calculation accuracy caused by the influence of working conditions, wellbores, stratum conditions and the like when an existing rod-pumped well uses an indicator diagram to measure the yield. The method comprises the following steps: S1, establishing an indicator diagram model library; s2, screening single well indicator diagrams; carrying out production calculation on wells with different diagnosis results by adopting different schemes; s3, performing overlay analysis on all the indicator diagram data of the rod-pumped well on the day, performing indicator diagram value selection, selecting indicator diagram data with accurate indicator diagram production, determining the most reasonable indicator diagram to represent the working condition on the day, and taking the corresponding indicator diagram production result as the yield on the day; and S4, regularly verifying and comparing the daily yield obtained by indicator diagram production calculation by using a separator. According to the indicator diagram production calculation method, the accuracy of the production calculation result of the separator-free flow rod-pumped well can be improved; and meanwhile, a separator can be replaced for measuring oil, and the single-well yield change is mastered in time.

Description

Technical Field

[0001] The invention relates to the technical field of oil production engineering, and in particular to an intelligent power diagram production calculation method applicable to all working conditions. Background Art

[0002] The existing technology is the "effective stroke method" power diagram production measurement technology. Since it is unreasonable to extract the effective stroke from the indicator diagram under the condition of rod breakage or pump leakage, this technology is only applicable to pumping wells with normal pumping conditions or insufficient fluid supply. When measuring the production of the indicator diagram under normal pumping conditions or insufficient fluid supply, the accuracy of the single well power diagram production measurement is unstable due to the interference of various factors such as wax deposition, rod aging, and uncertainty of leakage. Therefore, it is necessary to develop a power diagram production measurement method that is applicable to all working conditions and effectively improves the accuracy of production measurement. Summary of the invention

[0003] The technical problem to be solved by the present invention is to overcome the problem in the background technology that the accuracy of production measurement is unstable due to the influence of working conditions, wellbore conditions, formation conditions, etc. when the existing pumping wells use dynamometer diagrams to measure production, and to provide an intelligent dynamometer diagram production measurement method applicable to all working conditions. The intelligent dynamometer diagram production measurement method applicable to all working conditions can improve the accuracy of the production measurement results of the separator-free process pumping wells; at the same time, it can also replace the separator to measure oil and timely grasp the changes in single well production.

[0004] The present invention solves the problem through the following technical solution: the intelligent power diagram production calculation method applicable to all working conditions comprises the following steps:

[0005] S1. Through big data learning of historical performance diagram data of pumping wells, the wrong performance judgment results are manually corrected and a performance diagram model library is established;

[0006] S2. Screening single well indicator diagrams; comparing the screened single well indicator diagrams with the established diagram model library to finally obtain a diagnosis result, and using different plans to calculate production for wells with different diagnosis results;

[0007] S3, by superimposing and analyzing all the power diagram data of the digitized pumping well on that day, selecting the power diagram value according to the analysis result, selecting the power diagram data with accurate power diagram production calculation, determining the most reasonable power diagram to represent the working conditions on that day, and the corresponding power diagram production calculation result is used as the daily production;

[0008] S4. Regularly use separators to verify and compare the daily production calculated from the power chart, and use other production calculation methods for wells with inaccurate power chart production calculation results.

[0009] Preferably, the specific steps of performing big data learning on historical performance diagram data of pumping wells and establishing a performance diagram model library in S1 include:

[0010] S11. Obtain historical performance diagram data of the pumping well;

[0011] S12. Classify the working condition diagnosis types into five major types according to the characteristics of the power diagram shape;

[0012] S13. Label the historical power map data according to five categories;

[0013] S14. Use the labeled historical work graph data to train the model through the tensorflow technical framework tool. After the model training, 35 models are generated to form a work graph model library;

[0014] S15. Use the trained model to test and adjust the model parameters in time to improve the accuracy if problems are found;

[0015] S16. Repeat training-testing-adjusting until the accuracy of the working condition diagnosis reaches the requirement, and obtain a power diagram model library with the accuracy of the working condition diagnosis reaching the requirement.

[0016] Preferably, the five major types of working condition diagnosis include full type, missing corner type, handle type, long strip type, and other types.

[0017] Preferably, the screening method of the single well performance diagram in S2 is:

[0018] S21, collecting the power diagram data of a single well; the power diagram data of a single well includes basic production data, power diagram parameters, theoretical parameters, and power diagram;

[0019] S22, automatically analyzing and comparing the power diagram data of the single well to determine whether the basic production data, power diagram parameters, theoretical parameters and production reports are consistent, and if they are consistent, proceed to the next step of working condition diagnosis;

[0020] S23. Compare the dynamometer diagram of a single well that is consistent with the production report with the dynamometer model library to ultimately arrive at a diagnosis result; use different plans to calculate production for wells with different diagnosis results.

[0021] Preferably, the step 22 automatically analyzes and compares various dynamometer data of the dynamometer data of the single well using a computer program, and the computer program is dynamometer diagram precise metering system software, and the dynamometer diagram precise metering system software includes dynamometer diagram diagnosis and dynamometer diagram production calculation functions.

[0022] Preferably, the method of using different schemes to calculate production for wells with different diagnosis results in step 23 includes:

[0023] S231. For wells diagnosed as normal, a production calculation scheme combining the historical average pump efficiency method and the submergence correction method is adopted; the historical average pump efficiency is compared with the pump efficiency corresponding to the current submergence, and the historical average pump efficiency is corrected;

[0024] S232. For wells diagnosed as insufficient liquid supply or gas impact, the original effective stroke method is used to calculate production;

[0025] S233. For the wells diagnosed as interstitial leakage and tubing leakage, the pump efficiency is calculated according to the relationship between the pump efficiency and the submergence under different pump diameters, and it is determined whether the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well; if the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well, the pump efficiency of the well is directly set to 60% of the pump efficiency under normal working conditions; if the calculated pump efficiency does not exceed 60% of the pump efficiency under normal working conditions of the well, the calculated pump efficiency is directly used for power diagram production calculation;

[0026] S234. For a well diagnosed as having pump loss, the power diagram of the well under pump loss condition is compared with the power diagram under normal condition, and the ratio of reduction in the area of ​​the power diagram is calculated. The calculated ratio of reduction in the area of ​​the power diagram is multiplied by the production calculation result under normal condition to obtain the production calculation result of the pump loss condition.

[0027] S235. For wells diagnosed with broken rod or other operating conditions, production will not be counted and the production will be 0.

[0028] Preferably, the step 231 adopts a production calculation scheme combining a historical average pump efficiency method and a submergence correction method, compares the historical average pump efficiency with the pump efficiency corresponding to the current submergence, and corrects the historical average pump efficiency, including:

[0029] a1. Using the tensorflow technology framework tool based on the python platform in S1, the well history of the single pumping well is analyzed and counted, and the corresponding rules between the pump efficiency and the submergence of different pump diameters under normal working conditions are summarized, the upper and lower limits of the submergence and the corresponding pump efficiency are confirmed, and the corresponding relationship matrix between the submergence and the pump efficiency is established;

[0030] a2. Based on the established corresponding relationship matrix between submergence and pump efficiency, the reference value of the pump efficiency of the single well is calculated according to the pump diameter and submergence value of the single well in actual application, and then the average value of the pump efficiency reference value and the historical average pump efficiency is calculated as the corrected pump efficiency, and the corrected pump efficiency is used for power diagram production calculation.

[0031] Preferably, the specific items of the superposition analysis of all the power diagram data of the pumping well on the day in S3 include analyzing four aspects: overall stability, maximum and minimum load errors, effective stroke errors, and gap leakage judgment.

[0032] Preferably, the method S4 regularly uses a separator to verify and compare the daily production obtained by the power chart production calculation, including: comparing the power chart production calculation result with the oil value measured through the separator and the corresponding production calculation error, confirming the reasonable error range for different production levels, and the wells whose comparison error is within the reasonable range are regarded as having accurate power chart production calculation.

[0033] Preferably, the reasonable error range of the separator oil value and the corresponding different production levels is:

[0034] ① The oil value of the separator is ≤3t, and the production error does not exceed ±50%;

[0035] ②3t<separator oil value≤5t, production error shall not exceed ±30%;

[0036] ③5t<separator oil value≤50t, production error shall not exceed ±20%;

[0037] ④50t<separator oil value≤100t, production error shall not exceed ±10%;

[0038] ⑤ The oil measurement value of the separator is >100t, and the production error does not exceed ±10t.

[0039] Compared with the above background technology, the present invention has the following beneficial effects:

[0040] In the current situation where pumping wells are accelerating their digitalization, a large amount of dynamometer diagram data can be obtained from a single well in a single day. The present invention can fully transform the quantitative change of the dynamometer diagram data into a qualitative change in auxiliary production, thereby realizing intelligent dynamometer diagram production calculation under all working conditions. It can replace the original dynamometer diagram production calculation technology and improve the accuracy of the production calculation results of pumping wells with separator-free processes. At the same time, it can also replace the separator oil measurement, save construction costs and human resources, and timely grasp the changes in single well production. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 The present invention is a flow chart of the intelligent power diagram production calculation method applicable to all working conditions. Specific implementation method:

[0042] The present invention will be further described below in conjunction with the accompanying drawings:

[0043] like Figure 1 As shown, an intelligent power diagram production calculation method applicable to all working conditions includes the following steps:

[0044] Step 1: Use the tensorflow technology framework tool based on the python platform to conduct big data learning on the historical performance diagram data of the pumping wells, manually correct the incorrect performance judgment results, and establish a performance diagram model library; the specific methods include:

[0045] S11. Obtain historical performance diagram data of the pumping well;

[0046] S12. Classify the working condition diagnosis types into five major types according to the characteristics of the power diagram shape;

[0047] S13. Label the historical power map data according to five categories;

[0048] S14. Using the labeled historical power diagram data, the model is trained through the tensorflow technical framework tool. After the model training, 35 models are generated to form a power diagram model library; the power diagram model library includes 35 power diagram models in five categories, which serve as the basic database for intelligent working condition diagnosis; the five categories of working condition diagnosis types include full category, missing corner category, knife handle category, long strip category, and other categories.

[0049] S15. Use the trained model to test and adjust the model parameters in time to improve the accuracy if problems are found;

[0050] S16. Repeat training-testing-adjusting until the accuracy of the working condition diagnosis reaches the requirement, and obtain a power diagram model library with the accuracy of the working condition diagnosis reaching the requirement.

[0051] The big data learning is achieved by collecting a large amount of historical power diagram data in the early stage, classifying the historical power diagram data, using the classified power diagram data to train the model through the tensorflow technical framework tool, establishing a power diagram model library with the generated model, and performing model evaluation, repeatedly adjusting the model parameters, and finally enabling the computer to have the function of automatically classifying and judging the historical power diagram data of the pumping well.

[0052] Step 2: Screen the single well indicator diagram; compare the screened single well indicator diagram with the established diagram model library to finally obtain the diagnosis result, and use different plans to calculate the production for wells with different diagnosis results;

[0053] The screening method of the single well indicator diagram comprises:

[0054] S21, collecting the power diagram data of a single well; the power diagram data of a single well includes basic production data, power diagram parameters, theoretical parameters, and power diagram;

[0055] S22, automatically analyzing and comparing the power diagram data of the single well to determine whether the basic production data, power diagram parameters, theoretical parameters and production reports are consistent, and if they are consistent, proceed to the next step of working condition diagnosis;

[0056] Among them, a computer program is used to automatically analyze and compare various dynamometer data of the dynamometer data of a single well. The computer program is the dynamometer diagram precise metering system software, and the dynamometer diagram precise metering system software includes dynamometer diagram diagnosis and dynamometer diagram production calculation functions.

[0057] S23, using the single well indicator diagram that is consistent with the production report to compare with the diagram model library, and finally obtaining the diagnosis result; using different schemes to calculate production for wells with different diagnosis results. The method of using different schemes to calculate production for wells with different diagnosis results includes the following steps:

[0058] S231. For wells diagnosed as normal in working condition, a production calculation scheme combining the historical average pump efficiency method and the submergence correction method is adopted; the historical average pump efficiency is compared with the pump efficiency corresponding to the current submergence, and the historical average pump efficiency is corrected; the specific method includes the following steps:

[0059] a1. Based on the Python platform, the tensorflow technology framework tool is used to analyze and count the well history of single pumping wells, summarize the corresponding rules between pump efficiency and submergence of different pump diameters under normal working conditions, confirm the upper and lower limits of submergence and the corresponding pump efficiency, and establish the corresponding relationship matrix between submergence and pump efficiency;

[0060] a2. Based on the established corresponding relationship matrix between submergence and pump efficiency, the reference value of the pump efficiency of the single well is calculated according to the pump diameter and submergence value of the single well in actual application, and then the average value of the pump efficiency reference value and the historical average pump efficiency is calculated as the corrected pump efficiency, and the corrected pump efficiency is used for power diagram production calculation.

[0061] S232. For wells diagnosed as insufficient liquid supply or gas impact, the original effective stroke method is used to calculate production;

[0062] S233. For the wells diagnosed as interstitial leakage and tubing leakage, the pump efficiency is calculated according to the relationship between the pump efficiency and the submergence under different pump diameters, and it is determined whether the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well; if the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well, the pump efficiency of the well is directly set to 60% of the pump efficiency under normal working conditions; if the calculated pump efficiency does not exceed 60% of the pump efficiency under normal working conditions of the well, the calculated pump efficiency is directly used for power diagram production calculation;

[0063] S234. For a well diagnosed as having pump loss, the power diagram of the well under pump loss condition is compared with the power diagram under normal condition, and the ratio of reduction in the area of ​​the power diagram is calculated. The calculated ratio of reduction in the area of ​​the power diagram is multiplied by the production calculation result under normal condition to obtain the production calculation result of the pump loss condition.

[0064] S235. For wells diagnosed with broken rod or other operating conditions, production will not be counted and the production will be 0.

[0065] Step 3: Through computer programs (such as dynamometer precision metering system software) Perform superposition analysis on all the power diagram data of the digitized pumping well on that day. The specific items of superposition analysis include analysis of overall stability, maximum and minimum load errors, effective stroke errors, and gap leakage judgment; select the power diagram value according to the analysis results, select the power diagram data with accurate power diagram production, determine the most reasonable power diagram to represent the working conditions on that day, and use the corresponding power diagram production result as the daily output;

[0066] Step 4: Regularly use the separator to verify and compare the daily production calculated by the power chart, and use other production calculation methods for wells with inaccurate power chart production calculation results;

[0067] Among them, the method of regularly using a separator to verify and compare the daily production obtained from the power chart production calculation includes: comparing the power chart production calculation results with the oil value measured through the separator and the corresponding production calculation error, confirming the reasonable error range for different production levels, and the wells with comparison errors within a reasonable range are considered to have accurate power chart production calculations.

[0068] The reasonable error range of the separator oil value and the corresponding different production levels is:

[0069] ① The oil value of the separator is ≤3t, and the production error does not exceed ±50%;

[0070] ②3t<separator oil value≤5t, production error shall not exceed ±30%;

[0071] ③5t<separator oil value≤50t, production error shall not exceed ±20%;

[0072] ④50t<separator oil value≤100t, production error shall not exceed ±10%;

[0073] ⑤ The oil measurement value of the separator is >100t, and the production error does not exceed ±10t.

[0074] Example 1

[0075] In combination with the Xing 8-1-140 well of the Daqing Oil Production Plant No. 5, the intelligent power diagram production calculation method applicable to all working conditions of the present invention is further explained.

[0076] (1) Preliminary screening of correct power diagram data

[0077] A set of power diagram data was collected every half hour on the day of Xing 8-1-140 Well, for a total of 48 sets. Every time a set of data was collected, it was compared with the production report to determine whether the basic production data, power diagram parameters, theoretical parameters and production report were consistent. If they were consistent, the next step of operating condition diagnosis was performed. If they were inconsistent, they were defined as erroneous power diagram data. Among the 48 sets of data on the day, a total of 3 sets of data were defined as erroneous power diagram data, and no operating condition judgment and power diagram production calculation were performed;

[0078] (2) Intelligent working condition judgment

[0079] After a set of real-time collected power diagram data is determined to be correct power diagram data, the power diagram is compared and analyzed with the power diagram model library to obtain a working condition diagnosis result. On that day, 45 sets of correct power diagram data diagnosed a total of 45 working conditions, including 41 gas effects and 4 severe gas effects;

[0080] (3) Power diagram production

[0081] 45 groups of correct power diagram data of Xing 8-1-140 well were used to substitute the production calculation plan of the corresponding working condition according to the intelligent working condition diagnosis results of each group of data. One power diagram production calculation result was obtained for each group, and a total of 45 power diagram production calculation results were obtained.

[0082] (4) Select the value of the power diagram for the day

[0083] After all 48 groups of power diagram data were collected on that day, all correct power diagram data of Xing 8-1-140 well were superimposed and analyzed, and the power diagram value was selected according to the analysis results. The most reasonable group of power diagram data was determined to represent the working conditions of that day, and the corresponding power diagram production calculation results were used as the daily production;

[0084] (5) Comparison and verification of the accuracy of power diagram production

[0085] After selecting the value of the power chart, it was confirmed that the main operating condition of the Xing 8-1-140 well on that day was affected by gas, and the power chart production result was 8.7t / d. The separator oil measurement result of the Xing 8-1-140 well on that day was 9.2t / d, and the power chart production error was 5.43%. According to the comparison rules, for a single well with a daily liquid production of more than 5t and less than or equal to 50t, the power chart production error should not exceed ±20%, confirming that the power chart production of the well is accurate.

Claims

1. An intelligent power diagram production calculation method applicable to all working conditions, Features: The following steps are involved: S1. Through big data learning of historical performance diagram data of pumping wells, the wrong performance judgment results are manually corrected and a performance diagram model library is established; S2. Screening single well indicator diagrams; comparing the screened single well indicator diagrams with the established diagram model library to finally obtain a diagnosis result, and using different plans to calculate production for wells with different diagnosis results; S3, by superimposing and analyzing all the power diagram data of the digitized pumping well on that day, selecting the power diagram value according to the analysis result, selecting the power diagram data with accurate power diagram production calculation, determining the most reasonable power diagram to represent the working conditions on that day, and the corresponding power diagram production calculation result is used as the daily production; S4. Regularly use separators to verify and compare the daily production calculated from the power chart, and use other production calculation methods for wells with inaccurate power chart production calculation results.

2. According to the intelligent power diagram production calculation method applicable to all working conditions as described in claim 1, Features: The specific method of performing big data learning on the historical performance diagram data of the pumping well and establishing a performance diagram model library in S1 includes the following steps: S11. Obtain historical performance diagram data of the pumping well; S12. Classify the working condition diagnosis types into five major types according to the characteristics of the power diagram shape; S13. Label the historical power map data according to five categories; S14. Use the labeled historical work graph data to train the model through the tensorflow technical framework tool. After the model training, 35 models are generated to form a work graph model library; S15. Use the trained model to test and adjust the model parameters in time to improve the accuracy if problems are found; S16. Repeat training-testing-adjusting until the accuracy of the working condition diagnosis reaches the requirement, and obtain a power diagram model library with the accuracy of the working condition diagnosis reaching the requirement.

3. According to the intelligent power diagram production calculation method applicable to all working conditions as described in claim 1, Features: The five major types of working condition diagnosis include full type, missing corner type, handle type, long strip type and others.

4. According to the intelligent power diagram production calculation method applicable to all working conditions as described in claim 1, Features: The screening method of the single well indicator diagram in S2 is: S21, collecting the power diagram data of a single well; the power diagram data of a single well includes basic production data, power diagram parameters, theoretical parameters, and power diagram; S22, automatically analyzing and comparing the power diagram data of the single well to determine whether the basic production data, power diagram parameters, theoretical parameters and production reports are consistent, and if they are consistent, proceed to the next step of working condition diagnosis; S23. Compare the dynamometer diagram of a single well that is consistent with the production report with the dynamometer model library to ultimately arrive at a diagnosis result; use different plans to calculate production for wells with different diagnosis results.

5. According to claim 4, an intelligent power diagram production calculation method applicable to all working conditions, Features: The step 22 automatically analyzes and compares the various dynamometer data of the single well using a computer program, and the computer program is the dynamometer accurate metering system software, and the dynamometer accurate metering system software includes dynamometer diagnosis and dynamometer production calculation functions.

6. According to claim 4, an intelligent power diagram production calculation method applicable to all working conditions, Features: The method of using different schemes to calculate production for wells with different diagnosis results in step 23 includes: S231. For wells diagnosed as normal, a production calculation scheme combining the historical average pump efficiency method and the submergence correction method is adopted; the historical average pump efficiency is compared with the pump efficiency corresponding to the current submergence, and the historical average pump efficiency is corrected; S232. For wells diagnosed as insufficient liquid supply or gas impact, the original effective stroke method is used to calculate production; S233. For the wells diagnosed as interstitial leakage and tubing leakage, the pump efficiency is calculated according to the relationship between the pump efficiency and the submergence under different pump diameters, and it is determined whether the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well; if the calculated pump efficiency exceeds 60% of the pump efficiency under normal working conditions of the well, the pump efficiency of the well is directly set to 60% of the pump efficiency under normal working conditions; if the calculated pump efficiency does not exceed 60% of the pump efficiency under normal working conditions of the well, the calculated pump efficiency is directly used for power diagram production calculation; S234. For a well diagnosed as having pump loss, the power diagram of the well under pump loss condition is compared with the power diagram under normal condition, and the ratio of reduction in the area of ​​the power diagram is calculated. The calculated ratio of reduction in the area of ​​the power diagram is multiplied by the production calculation result under normal condition to obtain the production calculation result of the pump loss condition. S235. For wells diagnosed with broken rod or other operating conditions, production will not be counted and the production will be 0.

7. According to claim 6, an intelligent power diagram production calculation method applicable to all working conditions, Features: The step 231 adopts a production calculation scheme combining a historical average pump efficiency method and a submergence correction method, compares the historical average pump efficiency with the pump efficiency corresponding to the current submergence, and corrects the historical average pump efficiency, including: a1. Using the Python-based tensorflow technology framework tool, the well history of single pumping wells was analyzed and counted, and the corresponding rules between pump efficiency and submergence of different pump diameters under normal working conditions were summarized. The upper and lower limits of submergence and the corresponding pump efficiency were confirmed, and the corresponding relationship matrix between submergence and pump efficiency was established. a2. Based on the established corresponding relationship matrix between submergence and pump efficiency, the reference value of the pump efficiency of the single well is calculated according to the pump diameter and submergence value of the single well in actual application, and then the average value of the pump efficiency reference value and the historical average pump efficiency is calculated as the corrected pump efficiency, and the corrected pump efficiency is used for power diagram production calculation.

8. According to claim 1, an intelligent power diagram production calculation method applicable to all working conditions, Features: The specific items of superimposed analysis of all the power diagram data of the pumping well on the day by S3 include analysis of four aspects: overall stability, maximum and minimum load errors, effective stroke errors and gap leakage judgment.

9. According to the intelligent power diagram production calculation method applicable to all working conditions as described in claim 1, Features: The method S4 of regularly using a separator to verify and compare the daily production obtained by the power chart production calculation includes: comparing the power chart production calculation result with the oil value measured by the separator and the corresponding production calculation error, confirming the reasonable error range of different production levels, and the wells with comparison errors within the reasonable range are considered to have accurate power chart production calculation.

10. The intelligent power diagram production calculation method applicable to all working conditions according to claim 9, Features: The reasonable error range of the separator oil value and the corresponding different production levels is: ① The oil value of the separator is ≤3t, and the production error does not exceed ±50%; ②3t<separator oil value≤5t, production error shall not exceed ±30%; ③5t<separator oil value≤50t, production error shall not exceed ±20%; ④50t<separator oil value≤100t, production error shall not exceed ±10%; ⑤ The oil measurement value of the separator is >100t, and the production error does not exceed ±10t.

Images