CN109918722B - Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving - Google Patents
Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving Download PDFInfo
- Publication number
- CN109918722B CN109918722B CN201910090214.5A CN201910090214A CN109918722B CN 109918722 B CN109918722 B CN 109918722B CN 201910090214 A CN201910090214 A CN 201910090214A CN 109918722 B CN109918722 B CN 109918722B
- Authority
- CN
- China
- Prior art keywords
- water
- oil
- karst
- breakthrough
- water breakthrough
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention provides a method for predicting water breakthrough time of an oil well under data driving, which aims at the problems of poor practicability and low prediction accuracy of the existing method for predicting the water breakthrough time of an oil well of a fracture-cavity type oil reservoir, and belongs to the field of prediction of the water breakthrough time of the oil well. The method comprises the following steps: collecting production data of all water-breakthrough oil wells of the oil field under study; screening high-production typical wells with water exposure, and dividing the high-production typical wells according to the karst background; counting water-breakthrough early warning parameters of the water-breakthrough high-production typical wells under each karst background; establishing a water breakthrough time prediction multivariate regression equation corresponding to each karst background; training a neural network model for predicting water breakthrough time under each karst background; and verifying the accuracy of the multiple regression method and the neural network method by dividing the karst background, and selecting a prediction method with higher accuracy under each karst background. The method can effectively avoid the problem that the accurate geological data of the fracture-cavity type oil reservoir is difficult to obtain, and can be popularized to the fracture-cavity type oil reservoirs with various karst backgrounds.
Description
Technical Field
The invention relates to prediction of water breakthrough time of a fracture-cavity carbonate reservoir oil well, in particular to a method for predicting water breakthrough time of a fracture-cavity reservoir oil well under data driving.
Background
The fracture-cavity type oil reservoir has a complex oil-water relationship, the yield of an oil well is often reduced sharply after water breakthrough, and the water logging of the oil well is a leading factor causing the yield to be reduced. The water breakthrough time of the oil well is accurately predicted, so that the waterless oil recovery period of the oil well is increased after the water breakthrough time is delayed, and the method becomes a key technology for improving the recovery efficiency of the fracture-cavity oil reservoir. The water breakthrough early warning time coincidence rate of the existing oil well water breakthrough early warning technology is about 63.1%, and from the aspects of application range and early warning accuracy, the water breakthrough early warning technology still has a larger promotion space and needs further deep research and improvement.
At present, the following methods are mainly used for predicting the water breakthrough time of a fracture-cavity oil reservoir oil well. (1) The theoretical formula method of water breakthrough time of the fracture-cavity type bottom water reservoir oil well comprises the following steps: referring to a conventional sandstone reservoir, the method is based on the principles of bottom water coning and high-speed non-Darcy seepage. The method has strong theoretical performance, but has high measurement precision required by parameters, and is difficult and serious in practical application. (2) The numerical simulation prediction method of the water breakthrough time of the oil well comprises the following steps: the method is based on the carving of an effective reservoir body of a fracture-cavity oil reservoir and a three-dimensional physical geological model, differential solution is carried out on an oil-water two-phase mathematical model of the fracture-cavity oil reservoir according to an equivalent seepage medium theory, the oil-water motion rule at any moment in the fracture-cavity oil reservoir is simulated, and the dynamic evolution process of formation, lifting and breakthrough of a water cone of bottom water of the oil reservoir is simulated, so that the breakthrough time of the bottom water in an oil well production interval is predicted. Compared with a simplified analysis method, the method has the advantages that the influences of actual parameters such as oil reservoir geological conditions, the shape, the scale and the oil-water property of the fracture-cavity unit can be considered, the precision of a simulation result depends heavily on the precision of a fracture-cavity unit geological model and the values of the parameters, and all parameters required by the simulation of the total-precision numerical value are difficult to obtain due to the multiple leakage and emptying of the fracture-cavity reservoir section, so that the precision of the prediction of the water breakthrough time of the oil well applying the method is influenced. (3) Water-seeing method: the method comprises the steps of predicting a water cone of a conventional sandstone bottom water reservoir according to a sandstone bottom water reservoir development rule, dividing oil well development stages, establishing a corresponding relation between accumulated liquid yield and well head indexes, and determining the production stage according to the existing accumulated liquid yield to determine water breakthrough time. However, the conventional sandstone bottom water reservoir development rule is not applicable to fracture-cavity type reservoirs, and the prediction result is not ideal.
It can be seen that, existing methods for predicting water breakthrough time of oil wells of fracture-cavity reservoirs are not complete, and refer to seepage theory of conventional sandstone reservoirs more or less, however, fracture-cavity reservoirs and sandstone reservoirs have great difference in geological structure and flow pattern, seepage theory has poor applicability in fracture-cavity reservoirs, and in field practice, it is difficult to obtain accurate parameters such as pore seepage of fracture-cavity reservoirs, resulting in poor prediction effect.
Disclosure of Invention
Aiming at the problems that the existing prediction method for the water breakthrough time of the oil well of the fracture-cavity type oil reservoir is poor in practicability and low in prediction accuracy, the invention provides the prediction method for the water breakthrough time of the oil well under the data driving.
In order to achieve the purpose of the invention, the technical scheme adopted by the invention is as follows:
a method for predicting water breakthrough time of a fracture-cavity oil reservoir oil well under data driving comprises the following steps:
s01, collecting historical production data of all water-exposed oil wells in a researched block;
s02, screening high-production typical wells under the condition that the accumulated oil yield is more than 3 ten thousand tons or the initial daily oil yield is more than 30 tons, and determining the karst background of each well;
s03, counting water breakthrough sensitive parameters of a typical oil well with high water breakthrough yield under each karst background;
s04, performing partial correlation analysis on the water breakthrough sensitivity parameter statistical data to determine main control parameters, and establishing a water breakthrough time prediction multiple regression equation corresponding to each karst background by using the main control parameters;
s05, training a water breakthrough time prediction neural network model of each karst background oil well;
and S06, carrying out accuracy test on each karst background of the water breakthrough time prediction methods of the multivariate regression method and the neural network method by using the verification well data, and selecting the prediction method with higher accuracy under each karst background as the water breakthrough time prediction method of the oil well with the karst background.
Further, in step S01, the oil well production history data required to be collected includes, but is not limited to, total oil production, initial daily oil production level, and days of production in the dry oil recovery period.
Further, in the step S03, the water breakthrough sensitivity parameters include 10, which are respectively the reserve size, the reservoir characteristics, whether the heavy oil well is in use, the bottom water development degree, the mountain advance depth, whether the oil pressure is in use, the instantaneous capacity, whether the water breakthrough is sporadic, the production degree, and the oil pressure retention rate; three verification wells are reserved for each karst in the step, do not participate in the following steps S04 and S05, and are only used for the prediction accuracy test in the step S06.
The karst background types include but are not limited to weathering crust karsts, fracture control karsts, composite karsts.
Further, in step S04, the specific steps of performing partial correlation analysis and establishing a multiple regression equation include:
taking 10 water-breakthrough sensitive parameters statistically obtained in the step S03 as independent variables, taking the production days in the waterless oil recovery period collected in the step S01 as dependent variables, taking a statistical table for each karst background, and respectively importing the oil well parameter statistical tables under each karst background into SPSS business software;
calling a partial correlation analysis module of SPSS software, and setting other nine parameters as control variables when performing partial correlation analysis on the correlation between the 10 water-breakthrough sensitive parameter data and the production days in the anhydrous oil recovery period; screening out water-breakthrough sensitive parameters with the significance level lower than 0.05 according to the partial correlation analysis result, wherein the parameters can be regarded as water-breakthrough main control parameters;
and calling a linear regression module of SPSS software, taking the water breakthrough master control parameter screened in the last step as an independent variable and taking the production days in the anhydrous oil extraction period as a dependent variable, and establishing a corresponding linear regression equation for each karst background.
Further, in the step S05, a neural network toolkit of a matlab platform is used, the 10 water-breakthrough sensitive parameters of the water-breakthrough well obtained through statistics in the previous step are used as an output layer of the neural network, the water breakthrough time of each oil well is used as an output layer, two hidden layers are set up, and a 10 × 6 × 3 × 1 neural network model is trained for each karst background and used for predicting the water breakthrough time of the water-breakthrough wells.
The method comprises the steps of firstly collecting historical big data of a work area to be researched, then screening main control factors based on statistical results of water-breakthrough sensitive factor data of existing water wells, establishing respective water-breakthrough time prediction multiple regression equations aiming at different karst backgrounds, simultaneously training neural network prediction models under the karst backgrounds, then respectively substituting verification well data of the karst backgrounds into the corresponding multiple regression equations and the neural network models, and preferably selecting a method with high accuracy of prediction results as the water breakthrough time prediction method of the oil wells with the karst backgrounds.
The invention realizes the crossing from 'qualitative' to 'quantitative' of the prediction of the water breakthrough time of the fracture-cavity oil reservoir oil well; the accurate prediction of the water breakthrough time of the oil well can realize the 'passive water prevention' after the 'abnormal signal' is broken through by the water cone and the 'active water control' is pushed forward to the cone supporting period, and the 'lead' is made for the control measures of the water production of the oil well.
Drawings
FIG. 1 is a flow chart of the operation of the present invention;
FIG. 2 is a schematic diagram of an embodiment SPSS partial correlation analysis;
FIG. 3 is a neural network model based on matlab platform of an embodiment;
Detailed Description
The present invention will be described in further detail with reference to the following embodiments and the accompanying drawings.
Examples
In order to further explain the effectiveness of the technical method, the embodiment of the invention is further explained by combining the practical result of the Ordovician fracture-cavity carbonate reservoir in the Tahe oil field, the working process of the invention is shown as figure 1, and the specific steps are as follows:
s01: in this example, water breakthrough time prediction was performed for the wells with no water breakthrough in tahe field 6, pavu field, and 12. Historical production data was therefore collected for all water-met wells (879 total) in tahe field 6, tuppon and 12.
S02, screening high-yield typical wells under the condition that the accumulated oil yield is more than 3 million tons or the initial daily oil yield is more than 30 tons, and dividing the high-yield typical wells according to the karst backgrounds, wherein the karst backgrounds of the Tahe 6 area, the Topu Taiwan area and the 12 area are weathering crust karst, breaking control karst and composite karst respectively, so that 47 mouths, 160 mouths and 155 mouths of the high-yield typical wells with the weathering crust karst, the breaking control karst and the composite karst are obtained.
And S03, carrying out water breakthrough sensitivity parameter statistics on the water breakthrough wells under the karst backgrounds. The statistical water breakthrough sensitivity parameters are 10 in total, including the reserve scale (ten thousand tons), the reservoir characteristics (crack type, crack hole type and karst cave type), whether the heavy oil well is (no 0, 1), the bottom water development degree (no 0, 1), the mountain entering depth (m), whether the acid fracturing is (no 0, 1), the production degree, the oil pressure retention rate and the instantaneous capacity (m) 3 And d), whether water is sporadically seen or not (no 0, 1). And 3 wells are reserved for each karst background, and 9 wells are used for the final prediction accuracy test.
S04: performing partial correlation analysis by using SPSS software and establishing a multiple regression equation, which comprises the following specific steps:
(1) Taking 10 water-breakthrough sensitive parameters obtained by statistics in the step S03 as independent variables, taking the production days in the waterless oil recovery period collected in the step S01 as dependent variables, namely, each high-yield typical oil well corresponds to 11-dimensional data (namely 11 variables), making an EXCEL statistical table, wherein each karst background is a statistical table, and all oil well parameter statistical tables under the karst background are imported into SPSS business software when partial correlation analysis needs to be carried out on which karst background;
(2) And calling a partial correlation analysis module of SPSS software, performing partial correlation analysis on the correlation between the 10 water-breakthrough sensitive parameters and the production days in the anhydrous oil recovery period, and setting other nine parameters as control variables when performing partial correlation analysis on each parameter, as shown in FIG. 2. Analysis results show that under the background of the weathering crust karst, parameters with the significance less than 0.05 are the development degree of bottom water, whether the bottom water is sporadic, whether the oil is thick, whether the oil is acid fracturing or not and the extraction degree; under the background of composite karst, the parameters with the significance less than 0.05 are the type of a reservoir body, whether acid fracturing exists, the extraction degree, the bottom water development degree and the instantaneous capacity; under the background of the control failure karst, the parameters with the significance less than 0.05 are instantaneous productivity, extraction degree, bottom water development degree, oil well energy maintenance level and whether acid fracturing exists. These parameters are the main control parameters of oil well water breakthrough under various karst backgrounds.
(3) Calling a linear regression module of SPSS software, taking the water breakthrough master control parameter screened in the last step as an independent variable and taking the production days in the anhydrous oil extraction period as a dependent variable, and establishing a corresponding linear regression equation for each karst background as follows:
6, predicting a multivariate regression equation of karst water breakthrough time of the weathering crust in the area:
T＝645.754+351.476X 1 -268.36X 2 -334.58X 3 +193.688X 4 -13.347X 5 R 2 ＝0.435；
the 12-region composite karst water breakthrough time prediction multiple regression equation:
T＝680.21-755.543X 1-1 -678.979X 1-2 -677.776X 2 -14.221X 3 -361.256X 4 +1.583X 5 R 2 ＝0.475；
the Tofu platform intermittent control karst water breakthrough time prediction multivariate regression equation:
T＝601.725+3.045X 1 -12.734X 2 -197.484X 3 -9.67X 4 +65.007X 5 R 2 ＝0.479；
s05: based on a matlab platform, a neural network prediction program is compiled, 10 water-breakthrough sensitive parameter data of water-breakthrough wells obtained through statistics in the previous step are used as an output layer of a neural network, the water breakthrough time of each oil well is used as an output layer, two hidden layers are set up, a 10 multiplied by 6 multiplied by 3 multiplied by 1 neural network model is trained aiming at each karst background and used for predicting the water breakthrough time of the water-breakthrough wells, and the prediction program is shown in figure 3.
S06: and (3) carrying out accuracy test on the karst divided backgrounds of the multivariate regression method and the neural network method by using the previously reserved verification well data of 9 different karst backgrounds. The calculation method of the prediction accuracy comprises the following steps:
table 1 validation results of two prediction methods under each karst background.
The prediction results are shown in table 1, and it can be seen that the neural network method has high prediction accuracy in the weathering crust karst and composite karst backgrounds, and the multivariate regression method has high prediction accuracy in the discontinuous control karst backgrounds. Therefore, a neural network method is selected as a water breakthrough prediction method of weathering crust karst and composite karst background oil wells, and a multiple regression method is selected as a water breakthrough prediction method of the intermittent control karst background oil well.
The above-mentioned embodiments are merely illustrative of the preferred embodiments of the present invention, and do not limit the scope of the present invention, and various modifications and improvements of the technical solution of the present invention by those skilled in the art should fall within the protection scope defined by the claims of the present invention without departing from the spirit of the present invention.
Claims (5)
1. A method for predicting water breakthrough time of a fracture-cavity oil reservoir oil well under data driving is characterized by comprising the following steps:
s01, collecting historical production data of all water-exposed oil wells in a researched block;
s02, screening high-production typical wells under the condition that the accumulated oil yield is more than 3 ten thousand tons or the initial daily oil yield is more than 30 tons, and determining the karst background of each well;
s03, counting water breakthrough sensitive parameters of a typical oil well with high water breakthrough yield under each karst background;
s04, performing partial correlation analysis on the water breakthrough sensitivity parameter statistical data to determine main control parameters, and establishing a water breakthrough time prediction multiple regression equation corresponding to each karst background by using the main control parameters;
s05, training a water breakthrough time prediction neural network model of each karst background oil well;
and S06, carrying out accuracy test on each karst background of the water breakthrough time prediction methods of the multivariate regression method and the neural network method by using the verification well data, and selecting the prediction method with higher accuracy under each karst background as the water breakthrough time prediction method of the oil well with the karst background.
2. The prediction method of claim 1, wherein in step S01, the historical data of oil well production required to be collected includes total oil production, initial daily oil production level, and days of production in dry oil recovery.
3. The prediction method according to claim 1, wherein in the step S03, the water breakthrough sensitivity parameters include 10 parameters, which are respectively the reserve size, the reservoir characteristics, whether the heavy oil well is in the bottom water, the bottom water development degree, the mountain advance depth, whether the acid fracturing is in the acid fracturing state, the instantaneous capacity, whether the water breakthrough is sporadic, the production degree, and the oil pressure retention rate; three verification wells are reserved for each karst in the step, do not participate in the following steps S04 and S05, and are only used for the prediction accuracy test in the step S06.
4. The prediction method according to claim 3, wherein in the step S04, the specific steps of performing partial correlation analysis and establishing a multiple regression equation are as follows:
taking 10 water-breakthrough sensitive parameters statistically obtained in the step S03 as independent variables, taking the production days in the waterless oil recovery period collected in the step S01 as dependent variables, taking a statistical table for each karst background, and respectively importing the oil well parameter statistical tables under each karst background into SPSS business software;
calling a partial correlation analysis module of SPSS software, and setting other nine parameters as control variables when performing partial correlation analysis on the correlation between the 10 water-breakthrough sensitive parameter data and the production days in the anhydrous oil recovery period; screening out water-breakthrough sensitive parameters with the significance level lower than 0.05 according to the partial correlation analysis result, wherein the parameters can be regarded as water-breakthrough main control parameters;
and calling a linear regression module of SPSS software, taking the water breakthrough main control parameter screened in the last step as an independent variable and taking the production days in the anhydrous oil extraction period as a dependent variable, and establishing a corresponding linear regression equation aiming at each karst background.
5. The prediction method according to claim 4, wherein in step S05, a neural network toolkit of a matlab platform is used, the 10 water-breakthrough sensitive parameters of the water breakthrough well obtained by statistics in the previous step are used as output layers of the neural network, the water breakthrough time of each oil well is used as an output layer, two hidden layers are set up, and a 10 × 6 × 3 × 1 neural network model is trained for each karst background to predict the water breakthrough time of the water breakthrough wells.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910090214.5A CN109918722B (en) | 2019-01-30 | 2019-01-30 | Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910090214.5A CN109918722B (en) | 2019-01-30 | 2019-01-30 | Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109918722A CN109918722A (en) | 2019-06-21 |
| CN109918722B true CN109918722B (en) | 2022-11-18 |
Family
ID=66961037
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910090214.5A Active CN109918722B (en) | 2019-01-30 | 2019-01-30 | Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109918722B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110674469B (en) * | 2019-09-27 | 2023-04-14 | 长沙理工大学 | Hydrological frequency calculation method suitable for arid karst-deficient data area |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103362500A (en) * | 2013-08-06 | 2013-10-23 | 中国石油大学(华东) | Detection system and method for cracks and caverns of fractured-vuggy carbonate reservoirs along with drilling |
| CN105631529A (en) * | 2015-10-28 | 2016-06-01 | 中国石油化工股份有限公司 | Edge water gas reservoir water breakthrough time prediction method |
| CN105956712A (en) * | 2016-05-18 | 2016-09-21 | 恒泰艾普（北京）能源科技研究院有限公司 | Fracture-vug type carbonate reservoir water breakthrough early warning method |
-
2019
- 2019-01-30 CN CN201910090214.5A patent/CN109918722B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103362500A (en) * | 2013-08-06 | 2013-10-23 | 中国石油大学(华东) | Detection system and method for cracks and caverns of fractured-vuggy carbonate reservoirs along with drilling |
| CN105631529A (en) * | 2015-10-28 | 2016-06-01 | 中国石油化工股份有限公司 | Edge water gas reservoir water breakthrough time prediction method |
| CN105956712A (en) * | 2016-05-18 | 2016-09-21 | 恒泰艾普（北京）能源科技研究院有限公司 | Fracture-vug type carbonate reservoir water breakthrough early warning method |
Non-Patent Citations (2)
| Title |
|---|
| Prediction of water breakthrough time for oil wells in low-permeability bottom water reservoirs with barrier;Zhao Xinzhi et al.;《PETROLEUM EXPLORATION AND DEVELOPMENT》;20120801;第39卷(第4期);全文 * |
| 塔河油田缝洞型油藏高产井见水特征及水淹预警机制的建立;杨占红等;《西部探矿工程》;20140315;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109918722A (en) | 2019-06-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11542801B2 (en) | Optimized design method for temporary blocking agent to promote uniform expansion of fractures produced by fracturing in horizontal wells | |
| CN108229811B (en) | Method for evaluating water injection effect of fractured-vuggy carbonate reservoir | |
| CN102747991A (en) | Method for determining single-layer output of commingled producing well | |
| CN112343587A (en) | Ultra-low permeability reservoir dominant seepage channel identification and characterization method | |
| US20200340356A1 (en) | Method for evaluating difference in gas injection effect of gas injection wells in carbonate reservoir | |
| CN105160414A (en) | Method and device for predicting type of whole oil reservoir production area | |
| CN109236273B (en) | Dynamic data processing method for oil field development and production | |
| CN107895092B (en) | Inter-well communication quantitative evaluation method based on complex nonlinear injection-production modeling | |
| CN105426620B (en) | A kind of oil reservoir remaining oil Dominated Factors quantitative analysis method and device | |
| CN107194101A (en) | Fractured horizontal well crack is worn counts diagnostic method layer by layer | |
| CN111379562B (en) | Water-controlled coal mining method and device under composite water body | |
| CN109918722B (en) | Method for predicting water breakthrough time of fracture-cavity type oil reservoir oil well under data driving | |
| CN111950112A (en) | Dynamic analysis method for carbonate reservoir suitable for bottom sealing | |
| CN109670729A (en) | A kind of top plate aquifer water well evaluation method | |
| CN111155980B (en) | Water flow dominant channel identification method and device | |
| CN113742962B (en) | Shale reservoir attribute three-dimensional modeling method based on horizontal well | |
| CN110863814A (en) | Method and device for determining single-well subsection specific fluid production index of ultra-thick type cuttings limestone oil reservoir | |
| CN111950111A (en) | Dynamic analysis method for carbonate reservoir suitable for bottom opening | |
| CN112035993A (en) | Method for testing and evaluating carbonate reservoir with constant pressure at bottom | |
| CN111611704B (en) | Method and device for establishing water-drive series curve plate and readable storage medium | |
| CN111461458B (en) | Method for predicting SAGD (steam assisted gravity drainage) production profile in middle and later stages of production well | |
| CN115130268A (en) | Oil reservoir injection-production scheme design method and system, storage medium and electronic equipment | |
| Alfred et al. | Midland Basin Wolfcamp resource assessment using well spacing normalized recovery per section | |
| CN112392456A (en) | Well pattern layout rationality judgment method and well pattern layout optimization method | |
| CN114764546A (en) | Method for predicting residual oil of fault block oil reservoir based on generation of antagonistic neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |