CN115526086A - Carbonate reservoir water drive breakthrough time prediction method based on proxy model - Google Patents

Carbonate reservoir water drive breakthrough time prediction method based on proxy model Download PDF

Info

Publication number
CN115526086A
CN115526086A CN202211326213.4A CN202211326213A CN115526086A CN 115526086 A CN115526086 A CN 115526086A CN 202211326213 A CN202211326213 A CN 202211326213A CN 115526086 A CN115526086 A CN 115526086A
Authority
CN
China
Prior art keywords
water
reservoir
oil
breakthrough time
fracture
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.)
Pending
Application number
CN202211326213.4A
Other languages
Chinese (zh)
Inventor
孙致学
张灏
宋文铜
尹逸凡
王许强
李雪源
李昕睿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
China University of Petroleum East China
Original Assignee
China University of Petroleum East China
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by China University of Petroleum East China filed Critical China University of Petroleum East China
Priority to CN202211326213.4A priority Critical patent/CN115526086A/en
Publication of CN115526086A publication Critical patent/CN115526086A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/23—Design optimisation, verification or simulation using finite element methods [FEM] or finite difference methods [FDM]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/10—Numerical modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/08—Fluids
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/12—Timing analysis or timing optimisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
    • Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention relates to a carbonate reservoir water drive breakthrough time prediction method based on a proxy model, which comprises the following steps: s1: selecting a historical data set which discloses the actual production of a certain oil reservoir; s2: the natural water flooding of the oil reservoir is divided into four water flooding types according to the production dynamics of the oil reservoir; s3: carrying out simulation calculation on the related parameters of the oil reservoir by using a finite element conceptual model to obtain a numerical simulation sample data set; s4: constructing four water drive breakthrough time prediction sub-models; s5: constructing training sets of four water drive breakthrough time prediction submodels; s6: training each water drive breakthrough time prediction sub-model to finally obtain a trained prediction model; s7: and correcting the trained prediction model to obtain a final carbonate reservoir water drive breakthrough time prediction model. By using the technical scheme, the defects of long time consumption, low adaptability and low pertinence of the existing oil reservoir water drive breakthrough time prediction method can be overcome, and the specific development work of an actual oil well can be effectively guided.

Description

Carbonate reservoir water-drive breakthrough time prediction method based on proxy model
Technical Field
The invention relates to the field of prediction of water drive breakthrough time of an oil reservoir, in particular to a carbonate reservoir water drive breakthrough time prediction method based on a proxy model.
Background
Carbonate oil gas is hidden in the first place in the reserves and the yield of oil gas in the world, and is a research object with high attention at home and abroad. Large areas of carbonate rock strata are distributed in China, and a huge innovation space exists in the aspect of geological research of carbonate rock oil and gas reservoirs. The exploratory reserve of part of carbonate reservoirs is very high, but the physical property difference of the reservoirs is extremely strong and the exploitation difficulty is extremely high due to the complex distribution relationship of underground cracks and matrix horizons. At present, the common development problems of the crack-pore type carbonate oil reservoir include that the rapid rising of the water content cannot be controlled, the water breakthrough of an oil well is too early, so that the yield of part of the oil well is reduced too early and too fast, the development effect is poor, and the flooding of the oil well is a leading factor for reducing the yield. A large amount of cost is consumed for simulating the influence of the distribution relation of the fracture-matrix layer, the reservoir characteristic parameters and the fracture attributes (length, width, opening degree and trend) on the water breakthrough time of the oil well by using a conventional digital-analog method.
At present, the methods for predicting the natural water drive breakthrough time of an oil reservoir mainly comprise the following steps: (1) an indoor macroscopic three-dimensional physical simulation method. (2) A numerical simulation prediction method for breakthrough time of natural water flooding of an oil reservoir. Based on the carving of an effective reservoir body of an oil reservoir and a three-dimensional physical property geological model, the oil-water two-phase mathematical model of the oil reservoir is subjected to differential solution according to an equivalent seepage medium theory, the oil-water motion rule at any moment in the oil reservoir is simulated, and the dynamic evolution process of the 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 advantage that the influence of actual parameters such as reservoir geological conditions, oil-water properties and the like can be considered. (3) water-breakthrough method. The method comprises the steps of according to a sandstone reservoir bottom water reservoir development rule and conventional sandstone reservoir water cone prediction, dividing oil well development stages, establishing a corresponding relation between accumulated liquid production and well head indexes, and determining the production stage according to the existing accumulated liquid production to determine the natural water drive breakthrough time. And (4) analyzing historical data and counting. The method for predicting the natural water drive breakthrough time of the carbonate reservoir comprises a reservoir opening engineering method, a physical-analog-digital method and a statistical analogy method, and can theoretically analyze the influence factors of the natural water drive breakthrough time of the reservoir and predict the natural water drive breakthrough time of the reservoir. Unfortunately, the above techniques are not well suited and targeted for the actual development of oil reservoirs, and it is difficult to effectively guide the specific development of oil wells.
Chinese patent No. CN106150486B, "method for identifying water breakthrough rule of fractured reservoir", proposes a method for determining water flooding time according to the range of included angle Φ formed in different water-containing stages. The patent does not relate to a method for predicting the water breakthrough time of the production of a multilayer carbonate reservoir oil well, wherein the water breakthrough time refers to the water flooding breakthrough time.
A Chinese patent with the patent number of CN110020495A discloses a method and equipment for predicting water breakthrough time of a weathering crust karst fractured-vuggy reservoir oil well, and provides the method and equipment for predicting the water breakthrough time of the weathering crust karst fractured-vuggy reservoir oil well based on a data mining algorithm. The patent does not relate to the influence relationship of the distribution relationship of the cracks and the matrix horizon on the water breakthrough time, namely the water drive breakthrough time.
Disclosure of Invention
Aiming at the problems in the prior art, the technical problems to be solved by the invention are as follows: how to accurately predict the breakthrough time of the natural water drive of the oil reservoir.
In order to solve the technical problem, the invention adopts the following technical scheme: a carbonate reservoir water drive breakthrough time prediction method based on a proxy model comprises the following steps:
a carbonate reservoir water drive breakthrough time prediction method based on a proxy model comprises the following steps:
s100: selecting and disclosing an actual production historical data set of a certain oil reservoir, wherein the actual production historical data set comprises oil reservoir production dynamics, oil reservoir recovery ratio, oil reservoir parameters and oil reservoir fluid parameters;
s200: according to the production dynamics of the oil reservoir, natural water flooding of the oil reservoir is divided into four water flooding types, all data in the actual production historical data set are divided into four types of original data which correspond to the four water flooding types one by one according to the production dynamics of the oil reservoir, and the four water flooding types comprise: type I, type II, type III, and type IV;
s300: carrying out data simulation calculation on oil reservoir parameters and oil reservoir fluid parameters by using a fracture-pore type carbonate reservoir oil-water two-phase flow finite element conceptual model to obtain a numerical simulation sample data set;
s400: constructing a carbonate reservoir water-drive breakthrough time prediction model, wherein the carbonate reservoir water-drive breakthrough time prediction model comprises four water-drive breakthrough time prediction sub-models, the four water-drive breakthrough time prediction sub-models correspond to four water-drive types one by one, and each water-drive breakthrough time prediction sub-model is a multilayer sensor neural network;
s500: respectively constructing training sets of four water drive breakthrough time prediction submodels, and combining the first-class original data and a numerical simulation sample data set to obtain a training set of a water drive breakthrough time prediction submodel corresponding to the first-class original data;
s600: setting the maximum iteration times of training, taking a training set corresponding to each water drive breakthrough time prediction submodel as input, training each water drive breakthrough time prediction submodel by using a training module in MATLAB, stopping training when the training reaches the maximum iteration times, obtaining four trained water drive breakthrough time prediction submodels, and obtaining a trained carbonate reservoir water drive breakthrough time prediction model;
s700: correcting the trained carbonate reservoir water drive breakthrough time prediction model by utilizing the reservoir recovery ratio to obtain a final carbonate reservoir water drive breakthrough time prediction model;
s800: selecting a crack-pore type carbonate reservoir to be predicted, and judging the water drive type of the reservoir by utilizing the reservoir production dynamics of the crack-pore type carbonate reservoir to be predicted;
s900: and (3) taking the reservoir parameters and reservoir fluid parameters of the fractured-porous carbonate reservoir to be predicted as the input of the final carbonate reservoir water-drive breakthrough time prediction model, and outputting to obtain the water-drive breakthrough time prediction result of the fractured-porous carbonate reservoir to be predicted.
Preferably, the production dynamics of the oil reservoir in S100 comprises oil well pressure, water content, oil-water front and oil reservoir recovery.
The pressure and the water content of the oil well are data which are easy to collect in the oil field, the oil-water front is a parameter which can be calculated and most directly reflects the breakthrough speed of the natural water drive, the three parameters of different oil deposit types have large difference, and the oil deposit water breakthrough types can be finely divided into four types by utilizing the pressure, the water content and the oil-water front of the oil well.
Preferably, the reservoir parameters in S100 include fracture-matrix horizon distribution relation characteristic parameters, reservoir characteristic parameters, and fracture attribute parameters.
Compared with other water drive breakthrough time prediction methods, the method considers the influence of the distribution relation of the fracture-matrix layer and the fracture attribute on the water drive breakthrough time, the fracture is often a seepage channel, and the distribution mode and the distribution rule of the fracture have great influence on oil reservoir water channeling and oil well water breakthrough, so the distribution relation of the fracture-matrix and the fracture property need to be considered.
Preferably, the characteristic parameters of the distribution relation of the fracture-matrix layer position comprise the number of fracture matrix layers, the number of matrix rock layers and the layer thickness; the reservoir characteristic parameters comprise porosity and permeability; the fracture attribute parameters comprise fracture opening and fracture density.
Compared with other water drive breakthrough time prediction methods, the method combines more factors, the reservoir is considered to be mostly a complex superposed reservoir, the superposed mode of the fracture and the matrix is complex, and the prediction of the water drive breakthrough time of the reservoir is greatly influenced, so that the qualitative analysis of the related parameters of the fracture-base layer is particularly important.
Preferably, the parameters of the oil reservoir fluid in S100 include oil-water viscosity, density and phase permeability data.
The viscosity, density and phase permeability of oil and water can accurately reflect the flow state of underground fluid under the oil reservoir condition.
Preferably, the specific method for obtaining the numerical simulation sample data set in S300 is as follows:
s310: establishing a finite element conceptual model of oil-water two-phase flow of a fracture-pore type carbonate reservoir by utilizing COMSOL software;
s320: calculating an oil-water two-phase mass conservation equation in a matrix in the fracture-pore type carbonate reservoir, wherein the calculation expression is as follows:
Figure BDA0003912147680000041
wherein phi is m Is the porosity of the matrix; s mo The oil saturation of the matrix; s mw The substrate water saturation; t is time; v. of mo Is the velocity vector of the matrix oil phase, in m.s -1 ；v mw Is the matrix water phase velocity vector in m.s -1 ；q mo As a substrateOil phase sink item; q. q of mw Is a matrix water phase source and sink item;
wherein v is mo 、v mw An oil-water two-phase motion equation in the matrix is expressed as follows:
Figure BDA0003912147680000042
wherein K is the permeability tensor; k is a radical of formula ro Relative permeability of the oil phase; k is a radical of formula rw Relative permeability of water phase; mu.s o Is the oil phase viscosity; mu.s w Is the viscosity of the water phase; p is a radical of mo Is the pressure of the oil phase fluid in the matrix; p is a radical of mw Is the aqueous fluid pressure within the matrix;
s330: substituting the formula (2) into the formula (1) to obtain an oil-water two-phase control equation in the matrix of the fracture-pore type carbonate reservoir, wherein the calculation expression is as follows:
Figure BDA0003912147680000043
s340: calculating an oil-water two-phase mass conservation equation and an oil-water two-phase motion equation in the fracture, wherein the calculation expression is as follows:
Figure BDA0003912147680000044
Figure BDA0003912147680000045
in the formula, phi f Is the fracture porosity; p is a radical of fo Is the oil phase fluid pressure in the fracture; p is a radical of fw Is the aqueous fluid pressure within the fracture; s fo Is the fracture oil saturation; s. the fw The fracture water saturation; t is time; v. of fo Is the velocity vector of the fissured oil phase, m.s -1 ；v fw Is the fracture water phase velocity vector, m.s -1 ；q fo Is a source sink of fracture oil phase; q. q.s fw Is a crack water phaseA source and sink item; p is a radical of formula fo Is the oil phase fluid pressure in the fracture; p is a radical of fw Is the aqueous fluid pressure within the fracture;
s350: substituting the formula (5) into the formula (4) to obtain a control equation of oil-water two phases in the fracture-pore type carbonate reservoir, and calculating the expression as follows:
Figure BDA0003912147680000051
s360: obtaining a flow comprehensive equation of the water phase and the oil phase of the fracture-pore type carbonate reservoir according to the formula (3) and the formula (6), wherein the expression is as follows:
Figure BDA0003912147680000052
wherein S is w Indicating the water saturation of the reservoir, S o Representing the oil saturation of the reservoir, q w Representing the water sink of the reservoir, q o Representing oil phase source sink entries of the oil reservoir;
s370: calculating a pressure equation and a saturation matrix equation of oil-water two phases of the fracture-pore type carbonate reservoir, wherein the expression is as follows:
Figure BDA0003912147680000053
wherein, P o Denotes the oil phase fluid pressure of the reservoir,. Phi. Denotes the overall porosity of the reservoir, p c ' represents p c The first derivative of (a); wherein p is c Indicates capillary force in Pa, p c ' and p c The expression of (a) is as follows:
p c (S w )＝p o -p w ； (9)
Figure BDA0003912147680000054
wherein p is o Indicating oilHidden oil phase fluid pressure, p w Representing the water phase fluid pressure of the reservoir;
s380: and (3) obtaining saturation fields of the fractured-porous carbonate rock oil reservoir at different moments by using a formula (8), recording the moment as the water breakthrough time of the oil well under the oil reservoir condition when the water saturation of the oil well of the oil reservoir is not 0, and taking the set of all water breakthrough time data as a numerical simulation sample data set.
Preferably, the specific steps of correcting the trained carbonate reservoir water drive breakthrough time prediction model in S700 are as follows:
s710: calculating the trained carbonate reservoir water drive breakthrough time prediction model by utilizing COMSOL software to obtain the corresponding prediction recovery ratio of the model;
s720: setting a representation deviation value Z (e) of the oil reservoir by using a probability function;
s720: calculating a difference value between the predicted recovery ratio and the real recovery ratio, and if the difference value is smaller than Z (e), obtaining a final carbonate reservoir water drive breakthrough time prediction model; if the difference value is larger than Z (e), correcting the trained carbonate reservoir water drive breakthrough time prediction model by using a random number function, and specifically comprising the following steps:
s721: processing the recovery factor variables in the trained carbonate reservoir water drive breakthrough time prediction model by using a srand (unknown) time (NULL) function to obtain different recovery factor seed values; wherein the unknown is a variable type, and a recovery factor variable is selected; time (NULL) is a time value;
s721: the different recovery seed values are "randomized" using the rand () function, the expression is as follows:
X i ′＝kX i ； (11)
k＝rand()/16383.5； (12)
wherein k represents a correction coefficient, X i ' denotes the recovery factor after the i-th correction, X i Represents the ith recovery factor seed value, i =1, \8230;, n;
s722: calculating a deviation value S, wherein the specific expression is as follows:
Figure BDA0003912147680000061
wherein the content of the first and second substances,
Figure BDA0003912147680000062
representing the average recovery of all predicted results, and n represents the prediction times;
s723: and when S is less than Z (e), obtaining a final carbonate reservoir water drive breakthrough time prediction model, otherwise, returning to S721.
Compared with the conventional water flooding breakthrough time prediction method, the method carries out model construction based on COMSOL commercial software, and fully realizes the construction of a multi-crack-bedrock spreading model; based on the agent model method, water drive breakthrough time prediction and model characteristic link analysis are carried out on a typical characteristic oil reservoir under the conditions of considering a crack-bedrock spreading mode, physical properties and the like, a corresponding typical characteristic agent model is generated by training and learning, the water breakthrough time is accurately predicted by combining the real physical properties of the oil reservoir based on the agent model, and the prediction precision is improved; and then, result pre-judgment is carried out based on the deviation value theory, so that the oil reservoir scale high-precision water breakthrough time prediction simulation is realized.
Compared with the prior art, the invention has at least the following advantages:
1. the existing prediction method for the water drive breakthrough time of the oil reservoir consumes a large amount of calculation cost, and by using the technical scheme of the invention, the operation time can be greatly reduced, and the prediction accuracy is improved. According to the scheme, water drive type division is carried out on existing oil reservoir actual measurement data, then actual data are processed by utilizing an existing data simulation technology, and a data simulation sample set is obtained; then, combining data corresponding to different water flooding types with a numerical simulation sample set to serve as a training set for integral model prediction; and constructing a fracture-pore type carbonate reservoir water drive breakthrough time prediction model by taking the MLP multilayer perceptron neural network as a baseline model, performing model training by using a corresponding training set, performing condition verification on the trained model, and performing further model correction when the trained model does not meet the verification conditions to obtain a final prediction model. Model correction based on the deviation value theory is to carry out result pre-judgment in principle, and high-precision prediction of oil reservoir dimensions is realized.
2. According to the scheme, the influence of the distribution relation of the fracture-bed rock layer on the breakthrough time of the oil reservoir natural water drive is considered for the first time, in the step 3, the characteristic parameters of the distribution relation of the fracture-bed rock layer include the number of fracture matrix layers, the number of bed rock layers and the layer thickness as consideration factors, the parameters are reflected in the Comsol numerical simulation modeling process and the data collection process of an actual oil field, the parameters are used as input data for training MLP, and the prediction accuracy of the proxy model is improved.
3. Compared with the conventional water flooding breakthrough time prediction method, the method carries out model construction based on COMSOL commercial software, and fully realizes the construction of a multi-crack-bedrock spreading model; the water drive breakthrough time prediction and model characteristic link analysis are carried out on the basis of a proxy model method under the conditions that a typical characteristic oil reservoir considers a crack-bedrock spreading mode, physical properties and the like, a corresponding typical characteristic proxy model is generated by training and learning, the water drive breakthrough time is accurately predicted by combining the real physical properties of the oil reservoir on the basis of the proxy model, and the prediction precision is improved; and then, result pre-judgment is carried out based on the deviation value theory, so that the prediction simulation of the reservoir scale high-precision water drive breakthrough time is realized.
Drawings
FIG. 1 is a schematic view of the overall process of the present invention.
Fig. 2 is a schematic diagram of an MLP framework.
Fig. 3 is a flow chart of digital-analog model establishment.
FIG. 4 is a two-dimensional diagram of a Comsol model.
FIG. 5 is a Comsol model oil and water saturation diagram.
Fig. 6 is a flow chart of recovery correction.
Detailed Description
The present invention will be described in further detail below.
The method comprises the steps of analyzing the oil-water two-phase flow characteristics of the underground fractured-porous carbonate reservoir, simplifying and developing on the basis of a general water breakthrough time prediction model, constructing a fractured-porous carbonate reservoir proxy model, obtaining a numerical simulation sample by utilizing a finite element model simulation orthogonal test, solving parameters in the proxy model, correcting the model by utilizing recovery ratio in consideration of the influence of the development degree and the extraction degree of bottom water on the water breakthrough time, so that the model is more in line with the actual application requirements, obtaining a fractured-porous carbonate reservoir water flooding breakthrough time prediction model meeting the real-time simulation requirement, and carrying out real-time simulation calculation on the early and late natural water flooding breakthrough time and predicting the water flooding breakthrough time by matching with the characteristic parameters of the fractured-porous carbonate reservoir.
Example 1: the effectiveness of the technical method is further illustrated by taking a Tahe oil reservoir T unit as an example, and the implementation mode of the invention is explained as follows:
referring to fig. 1-6, a carbonate reservoir water drive breakthrough time prediction method based on a proxy model comprises the following steps:
s01: water breakthrough types are classified into 4 types according to production dynamics such as well pressure, water cut, oil-water front, etc., wherein the four water drive types are described as follows: type I: the water content rises to 40-60% dramatically in two years at the initial stage of well opening, and the expression is t 1 ＝A 1 x f +B 1 P+C 1 f w (ii) a Type II: two years after the well is opened, a first anhydrous recovery period is carried out, the water content is 0, then the water content slowly rises, the pressure of the oil well is kept at 70-90% of the original formation pressure, and the expression is t 2 ＝A 2 P+B 2 x f +C 2 f w (ii) a Type III: after two years of well opening, a first anhydrous recovery period is carried out, the water content is violently increased to 40% -60%, and the expression is t 3 ＝A 3 x f +B 3 P+C 3 f w (ii) a Type IV: two years after the well is opened, the pressure of the oil well is slowly reduced to 70-90% of the original formation pressure, but the water content is kept stable at 20-40%, and the expression is t 4 ＝A 4 P+B 4 x f +C 4 f w . The following edge and bottom water types are respectively corresponded: the side bottom water is close to the well mouth and is communicated with the channel with the advantages of crack and the like; the water-avoiding height is certain, and reservoir cracks are not developed; has a certain water-avoiding heightThe degree and the cracks are relatively developed; the edge bottom water is mainly local sealing water. The actual production data of the oil reservoir are divided into four types and are preprocessed: including outlier processing, and data normalization.
S02: the input parameters include: reservoir parameters such as fracture-matrix layer distribution relation characteristic parameters, reservoir characteristic parameters and fracture attribute parameters, wherein the number of fracture matrix layers, the number of sandstone layers, the thickness of the layers, the porosity is 0.2, the permeability is 100mD, the fracture opening, the fracture density and the like; fluid parameters, crude oil viscosity 7.75mPa/s, density 0.84g/cm 3 Viscosity of injected water of 0.48mPa/s, density of 1.02g/cm 3 Phase permeability data, etc. The method mainly comprises two parts of actual production historical data and numerical simulation sample data. The numerical simulation sample utilizes COMSOL software to construct a fracture-pore type carbonate reservoir conceptual model, the COMSOL software is shown in figure 4, the COMSOL software is the prior art, the fracture-pore type carbonate reservoir oil-water two-phase flow finite element conceptual model is a known and trained existing model, a fractured reservoir adopts a discrete fracture model, a PDE module is used for calculating the saturation of each phase, the lower boundary is an inflow interface, the upper boundary is an outflow interface, the maximum value of the water saturation of each point of the outflow interface is equal to 1 as a termination condition, the calculation time at the moment is recorded as water breakthrough time, the water breakthrough time is used as a natural water flooding breakthrough time sample for solving the natural water flooding breakthrough time prediction model, and the result is shown in figure 5.
And (3) combining an oil-water two-phase mass conservation equation and a motion equation in the matrix and in the crack, deducing an ordinary differential equation to obtain an oil-water two-phase control equation only related to saturation and pressure, and arranging the control equation into a matrix form to realize the calculation of the oil-water two-phase flow considering the discrete crack in Comsol software.
S03: in order to realize the construction of a natural water-drive breakthrough time prediction model of the fracture-pore type carbonate reservoir with high dimensional variables, an MLP (multilayer perceptron) neural network is utilized through characteristic analysis of oil-water two-phase flow of the fracture-pore type carbonate reservoir based on a developed mathematical model. The fracture-pore type carbonate reservoir oil-water two-phase flow finite element conceptual model is an existing and trained model; multilayer perceptron neural networks are prior art;
s04: firstly, inputting reservoir physical parameters such as fracture-bedrock horizon distribution relation characteristic parameters, reservoir characteristic parameters, fracture attribute parameters and fluid parameters, oil-water viscosity, density, phase permeability data and the like, and outputting water content (f) w ) Oil well pressure (P), oil water front (x) f ) Judging that the tower river oil reservoir T unit belongs to the type II based on the front edge distance well position, the pressure maintaining level and the water content change trend; solving the mapping relation expression of the corresponding type water-breakthrough time prediction model as t 2 ＝A 2 P+B 2 x f +C 2 f w 。
S05: using 80% of water breakthrough time samples of different concept models obtained by oil deposit actual production data and a numerical simulation method as an initial data set of a training proxy model and 20% as a test set, and using MATLAB to solve a mapping relation expression of a fracture-pore carbonate reservoir natural water drive breakthrough time prediction proxy model constructed in S03, wherein parameters to be solved mainly comprise A 2 ，B 2 ，C 2 (ii) a The training module in MATLAB is prior art.
S06: correcting each type of agent model based on secondary control factors such as recovery ratio, and the specific process is shown in FIG. 6, and representing a deviation value Z (e) by using a probability function, wherein the representation deviation value is an allowable error value provided according to experience; if the deviation value of the model recovery ratio and the real oil reservoir recovery ratio is less than Z (e), the agent model meets the operation requirement; if not, the agent model is corrected by using a random number function, and then the judgment is repeated until the deviation value requirement is met.
The random number function mainly utilizes the srand () function and the rand () function in the C language, the seed value is changed through the srand () function, usually the srand (signed) time (NULL)), so that the rand () function realizes further "randomization" according to different seed values, because the time is always changed, the seed value is always changed, the "randomization" is that the corresponding random number is returned corresponding to different seed values after the rand () function is used, the seed is continuously changed, the returned rand () function value is more random, the range of the obtained random number is between 0 and 32767, the programming of the random number function with the range of k between 0 and 2 is k = rand ()/16383.5, finally, the visible water time is 1.5a, and the deviation value of the proxy model is less than 5%.
The method disclosed by the invention deeply researches the fracture-pore type carbonate reservoir, and overcomes the defects that the existing natural water-drive breakthrough time prediction method is long in time consumption, low in applicability and low in pertinence in the actual development of an oil field, and can effectively guide the specific development of an oil well based on the natural water-drive breakthrough time prediction method of the fracture-pore type carbonate reservoir based on the neural network model.
Finally, the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting, although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions may be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all of them should be covered in the claims of the present invention.

Claims (7)

1. A carbonate reservoir water drive breakthrough time prediction method based on a proxy model is characterized by comprising the following steps: the method comprises the following steps:
s100: selecting and disclosing an actual production historical data set of a certain oil reservoir, wherein the actual production historical data set comprises oil reservoir production dynamics, oil reservoir recovery ratio, oil reservoir parameters and oil reservoir fluid parameters;
s200: according to the production dynamics of the oil reservoir, natural water flooding of the oil reservoir is divided into four water flooding types, all data in the actual production historical data set are divided into four types of original data which correspond to the four water flooding types one by one according to the production dynamics of the oil reservoir, and the four water flooding types comprise: type I, type II, type III, and type IV;
s300: carrying out data simulation calculation on oil reservoir parameters and oil reservoir fluid parameters by using a fracture-pore type carbonate reservoir oil-water two-phase flow finite element conceptual model to obtain a numerical simulation sample data set;
s400: constructing a carbonate reservoir water-drive breakthrough time prediction model, wherein the carbonate reservoir water-drive breakthrough time prediction model comprises four water-drive breakthrough time prediction sub-models, the four water-drive breakthrough time prediction sub-models correspond to four water-drive types one by one, and each water-drive breakthrough time prediction sub-model is a multilayer sensor neural network;
s500: respectively constructing training sets of four water drive breakthrough time prediction submodels, and combining one type of original data and a numerical simulation sample data set to obtain a training set of a water drive breakthrough time prediction submodel corresponding to the type of original data;
s600: setting the maximum iteration times of training, taking a training set corresponding to each water drive breakthrough time prediction submodel as input, training each water drive breakthrough time prediction submodel by using a training module in MATLAB, stopping training when the training reaches the maximum iteration times, obtaining four trained water drive breakthrough time prediction submodels, and obtaining a trained carbonate reservoir water drive breakthrough time prediction model;
s700: correcting the trained carbonate reservoir water drive breakthrough time prediction model by utilizing the reservoir recovery ratio to obtain a final carbonate reservoir water drive breakthrough time prediction model;
s800: selecting a crack-pore type carbonate reservoir to be predicted, and judging the water drive type of the reservoir by utilizing the reservoir production dynamics of the crack-pore type carbonate reservoir to be predicted;
s900: and (3) taking the reservoir parameters and reservoir fluid parameters of the fractured-porous carbonate reservoir to be predicted as the input of the final carbonate reservoir water-drive breakthrough time prediction model, and outputting to obtain the water-drive breakthrough time prediction result of the fractured-porous carbonate reservoir to be predicted.
2. The method for predicting the water drive breakthrough time of the carbonate reservoir based on the proxy model as claimed in claim 1, wherein: and the production dynamics of the oil reservoir in the S100 process comprises oil well pressure, water content, oil-water front and oil reservoir recovery ratio.
3. The carbonate reservoir water drive breakthrough time prediction method based on the proxy model of claim 2, wherein: the reservoir parameters of the reservoir in the S100 comprise fracture-matrix horizon distribution relation characteristic parameters, reservoir characteristic parameters and fracture attribute parameters.
4. The method for predicting the water drive breakthrough time of the carbonate reservoir based on the proxy model as claimed in claim 3, wherein: the characteristic parameters of the distribution relation of the crack-matrix layer position comprise the number of layers of a crack matrix, the number of layers of bedrock and the layer thickness; the reservoir characteristic parameters comprise porosity and permeability; the fracture attribute parameters comprise fracture opening and fracture density.
5. The method for predicting the water drive breakthrough time of the carbonate reservoir based on the proxy model as claimed in claim 4, wherein: the oil reservoir fluid parameters in the S100 comprise oil-water viscosity, density and phase permeability data.
6. The method for predicting the water drive breakthrough time of the carbonate reservoir based on the proxy model as claimed in claim 5, wherein: the specific method for obtaining the numerical simulation sample data set in S300 is as follows:
s310: establishing a finite element conceptual model of oil-water two-phase flow of a fracture-pore type carbonate reservoir by utilizing COMSOL software;
s320: calculating an oil-water two-phase mass conservation equation in a matrix in the fracture-pore type carbonate reservoir, wherein the calculation expression is as follows:
Figure FDA0003912147670000021
wherein phi is m Is the porosity of the matrix; s. the mo The oil saturation of the matrix; s mw The substrate water saturation; t is time; v. of mo Is the velocity vector of the matrix oil phase, in m.s -1 ；v mw Is the matrix aqueous phase velocity vector, unit m.s -1 ；q mo Is a matrix oil phase sourceA collection item; q. q.s mw Is a matrix water phase source and sink item;
wherein v is mo 、v mw An oil-water two-phase motion equation in the matrix is expressed as follows:
Figure FDA0003912147670000022
wherein K is a permeability tensor; k is a radical of ro Relative permeability of the oil phase; k is a radical of formula rw Relative permeability of water phase; mu.s o Is the oil phase viscosity; mu.s w Is the viscosity of the water phase; p is a radical of mo Is the fluid pressure of the oil phase in the matrix; p is a radical of mw Is the aqueous fluid pressure within the matrix;
s330: substituting the formula (2) into the formula (1) to obtain an oil-water two-phase control equation in the matrix of the fracture-pore type carbonate reservoir, wherein the calculation expression is as follows:
Figure FDA0003912147670000023
s340: calculating an oil-water two-phase mass conservation equation and an oil-water two-phase motion equation in the crack, wherein the calculation expression is as follows:
Figure FDA0003912147670000031
Figure FDA0003912147670000032
in the formula, phi f Is the fracture porosity; p is a radical of formula fo Is the oil phase fluid pressure in the fracture; p is a radical of formula fw Is the aqueous fluid pressure within the fracture; s fo Is the fracture oil saturation; s. the fw The fracture water saturation; t is time; v. of fo Is the velocity vector of the fissured oil phase, m.s -1 ；v fw Is the fracture water phase velocity vector, m.s -1 ；q fo Is crackedOil-sewing phase source sink; q. q.s fw Is a source and sink item of fracture water; p is a radical of formula fo Is the oil phase fluid pressure in the fracture; p is a radical of formula fw Is the aqueous fluid pressure within the fracture;
s350: substituting the formula (5) into the formula (4) to obtain a control equation of oil-water two phases in the fracture-pore type carbonate reservoir, and calculating the expression as follows:
Figure FDA0003912147670000033
s360: obtaining a flow comprehensive equation of the water phase and the oil phase of the fracture-pore type carbonate reservoir according to the formula (3) and the formula (6), wherein the expression is as follows:
Figure FDA0003912147670000034
Figure FDA0003912147670000035
wherein S is w Indicating the water saturation of the reservoir, S o Representing the oil saturation of the reservoir, q w Representing the water source sink of the reservoir, q o Representing an oil phase source sink of the oil reservoir;
s370: calculating an oil-water two-phase pressure equation and a saturation matrix equation of the fracture-pore type carbonate reservoir, wherein the expression is as follows:
Figure FDA0003912147670000041
wherein, P o Denotes the oil phase fluid pressure of the reservoir,. Phi. Denotes the overall porosity of the reservoir, p c ' represents p c The first derivative of (a); wherein p is c Indicates capillary force in Pa, p c ' and p c The expression of (c) is as follows:
p c (S w )＝p o -p w ； (9)
Figure FDA0003912147670000042
wherein p is o Representing the oil phase fluid pressure, p, of the reservoir w Representing the water phase fluid pressure of the reservoir;
s380: and (3) obtaining saturation fields of the fracture-pore type carbonate reservoir at different moments by using a formula (8), recording the moment as the water breakthrough time of the oil well under the reservoir condition when the water saturation of the oil well of the reservoir is not 0, and taking the set of all water breakthrough time data as a numerical simulation sample data set.
7. The carbonate reservoir water drive breakthrough time prediction method based on the proxy model as claimed in claim 6, wherein: the concrete steps of correcting the trained carbonate reservoir water drive breakthrough time prediction model in the S700 are as follows:
s710: calculating the trained carbonate reservoir water drive breakthrough time prediction model by utilizing COMSOL software to obtain the corresponding prediction recovery ratio of the model;
s720: setting a representation deviation value Z (e) of the oil reservoir by using a probability function;
s720: calculating a difference value between the predicted recovery ratio and the real recovery ratio, and if the difference value is smaller than Z (e), obtaining a final carbonate reservoir water drive breakthrough time prediction model; if the difference value is larger than Z (e), correcting the trained carbonate reservoir water drive breakthrough time prediction model by using a random number function, and specifically comprising the following steps:
s721: processing the recovery factor variables in the trained carbonate reservoir water drive breakthrough time prediction model by using a srand (unknown) time (NULL) function to obtain different recovery factor seed values; wherein the unscheduled is variable type, and the recovery factor variable is selected; time (NULL) is a time value;
s721: the different recovery seed values are "randomized" using the rand () function, the expression is as follows:
X i ′＝kX i ； (11)
k＝rand( )/16383.5； (12)
wherein k represents a correction coefficient, X i ' denotes the recovery factor after the i-th correction, X i Represents the i-th recovery factor seed value, i =1, \ 8230;, n;
s722: calculating a deviation value S, wherein the specific expression is as follows:
Figure FDA0003912147670000051
wherein, the first and the second end of the pipe are connected with each other,
Figure FDA0003912147670000052
representing the average recovery of all predicted results, and n represents the prediction times;
s723: and when S is less than Z (e), obtaining a final carbonate reservoir water drive breakthrough time prediction model, otherwise, returning to S721.
CN202211326213.4A 2022-10-27 2022-10-27 Carbonate reservoir water drive breakthrough time prediction method based on proxy model Pending CN115526086A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211326213.4A CN115526086A (en) 2022-10-27 2022-10-27 Carbonate reservoir water drive breakthrough time prediction method based on proxy model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211326213.4A CN115526086A (en) 2022-10-27 2022-10-27 Carbonate reservoir water drive breakthrough time prediction method based on proxy model

Publications (1)

Publication Number Publication Date
CN115526086A true CN115526086A (en) 2022-12-27

Family

ID=84703514

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211326213.4A Pending CN115526086A (en) 2022-10-27 2022-10-27 Carbonate reservoir water drive breakthrough time prediction method based on proxy model

Country Status (1)

Country Link
CN (1) CN115526086A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116796611A (en) * 2023-08-22 2023-09-22 成都理工大学 Method for adjusting bridge buckling cable force based on flagelliforme algorithm and artificial neural network
CN117687096A (en) * 2024-02-02 2024-03-12 中国石油大学(华东) Proxy model construction method for predicting small-scale fracture-cavity distribution

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116796611A (en) * 2023-08-22 2023-09-22 成都理工大学 Method for adjusting bridge buckling cable force based on flagelliforme algorithm and artificial neural network
CN116796611B (en) * 2023-08-22 2023-10-31 成都理工大学 Method for adjusting bridge buckling cable force based on flagelliforme algorithm and artificial neural network
CN117687096A (en) * 2024-02-02 2024-03-12 中国石油大学(华东) Proxy model construction method for predicting small-scale fracture-cavity distribution
CN117687096B (en) * 2024-02-02 2024-04-05 中国石油大学(华东) Proxy model construction method for predicting small-scale fracture-cavity distribution

Similar Documents

Publication Publication Date Title
CN115526086A (en) Carbonate reservoir water drive breakthrough time prediction method based on proxy model
CN112360411B (en) Local well pattern water injection development optimization method based on graph neural network
CN109543828B (en) Water absorption profile prediction method based on small sample condition
US11308413B2 (en) Intelligent optimization of flow control devices
CN110096718B (en) Method for obtaining volume of karst cave in carbonate reservoir
CN111815773B (en) Three-dimensional complex geologic model label manufacturing method suitable for machine learning algorithm
CN112016212B (en) Reservoir longitudinal heterogeneity evaluation method based on seepage control equation
CN109684685B (en) method for analyzing productivity and reservoir stability under pressure-reducing exploitation condition of hydrate in multilateral well
CN106407503B (en) Forecast Means of Reservoir Fractures and device
Ma et al. Multiscale-network structure inversion of fractured media based on a hierarchical-parameterization and data-driven evolutionary-optimization method
CN112576248A (en) Method for evaluating and predicting early productivity of bottom water gas reservoir
CN111950112B (en) Carbonate reservoir dynamic analysis method suitable for bottom closure
Bahari et al. Intelligent drilling rate predictor
CN114036774A (en) Determination method, calculation method and determination system for quasi-relative permeability of semi-filled fracture-cavity reservoir
CN108664677B (en) Oil and gas well production data analysis method
CN110486008B (en) Parameter interpretation method and system for radial composite oil reservoir
CN111950111B (en) Dynamic analysis method suitable for open-bottom carbonate reservoir
CN111677486A (en) Compact oil carbon dioxide huff and puff simulation method and device and storage medium
CN106934075B (en) Drilling fluid density determination method and static equivalent density determination method
CN115526114A (en) Hydraulic fracture morphology inversion method and system based on ensemble Kalman filtering method
CN110905484A (en) Method for calculating communication degree between wells of fracture-cave type carbonate reservoir
CN111625925B (en) Ternary combination flooding injection-production optimization method based on chromatographic separation
CN113361111A (en) Partitioning method of low-permeability reservoir characteristic model
CN114139432A (en) Fractured reservoir CO using neural network technology2Flooding simulation method
CN106600693A (en) Four-dimensional oil saturation geological modeling method

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