CN109490983B - A method and system for automatic fitting of reservoir geomechanics parameters - Google Patents

Abstract

The present invention relates to a method and system for automatic fitting of geomechanical parameters of reservoirs, which comprises the following steps: generating a random field about mechanical parameters through the Karhunen-Loeve expansion method; importing the random field into an Abaqus finite element simulator Stress calculation; adjust the random number in the Karhunen-Loeve algorithm through the optimization algorithm to achieve the fitting of the formation stress simulation result and the actual observation value within the error range; use the random number optimized by the genetic algorithm to obtain more accurate geomechanics parameter field. Using the Karhunen-Loeve expansion method to realize the generation of random fields, using Abaqus software to simulate the stress distribution under formation conditions, and using the geomechanical parameters fitted by the optimization algorithm, a method for automatic fitting of geomechanical parameters is established, which can automatically Fitting reservoir geomechanical parameters and obtaining a parameter field conforming to statistical distribution characteristics is of great guiding significance for engineers to quickly and accurately obtain the spatial distribution of reservoir geomechanical parameters.

Description

A method and system for automatic fitting of reservoir geomechanics parameters

technical field

The invention relates to an automatic fitting method, in particular to an automatic fitting method and system for reservoir geomechanics parameters.

Background technique

Due to the heterogeneity of underground reservoir rocks, the same reservoir has different mechanical properties with different locations, and thus has different in-situ stress distributions. For oil and gas reservoir production, the mechanical parameters present highly heterogeneous and anisotropic spatially, and there is a large deviation between the traditional homogeneous model and the actual formation. If the distribution of reservoir mechanical parameters can be accurately described, it will not only help to identify geological sweet spots, but also be of great significance for guiding fracturing construction in the later stage.

Contents of the invention

The purpose of the present invention is to overcome the shortcomings of the prior art, provide a method and system for automatic fitting of reservoir geomechanics parameters, and solve the problem of large deviation between the traditional homogeneous model and the actual formation.

The object of the present invention is achieved through the following technical solutions: a method for automatically fitting reservoir geomechanical parameters, which comprises the following steps:

Generate random fields about mechanical parameters by Karhunen-Loeve expansion method;

Import the random field into the Abaqus simulator for reservoir stress simulation;

Through the optimization algorithm, the simulation results and the actual values are fitted within the allowable error range, and then the optimal geomechanical parameter field is obtained.

The step of derivation of the Karhunen-Loeve expansion method needs to be completed before performing the step of generating a random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method.

After completing the derivation process steps of the Karhunen-Loeve expansion method and before performing the step of generating a random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method, it is also necessary to complete the Karhunen-Loeve according to the derivation implementation process Related procedures of the expansion method.

The specific content of the step of generating the random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method is as follows:

Generate N mutually independent Gaussian random variables ξ _i (θ);

Calculate the integral equation of the covariance matrix to obtain the eigenvalue λ _n and the eigenfunction f _n ;

Substitute the eigenvalue λ _n , the eigenfunction f _n and the Gaussian random variable ξ _i (θ) into Take N-order truncation to realize the generation of a random field;

Repeat the above third step P times to realize the generation of P random fields;

After P random field generation is realized, the following matrix is obtained according to the random variable ξ _i (θ):

The random variable ξ _i (θ) satisfies the standard normal distribution, and satisfies E[ξ _i (θ)]=0; E[ξ _i (θ)ξ _j (θ)]=δ _ij , wherein, δ _ij is the Kronecker-delta function.

Described Abaqus simulator comprises poroelastic finite element simulator; Described import random field in the Abaqus simulator and carry out the concrete content of reservoir ground stress simulation step as follows:

Import the random field of mechanical properties generated by the Karhunen-Loeve expansion method into the Abaqus simulator;

In the Abaqus simulator, a horizontal strain boundary condition and a vertical stress boundary condition from the overlying strata are assigned to the poroelastic mechanics model;

Get the value of the stress tensor distribution with respect to the random field.

Described optimization algorithm comprises genetic algorithm; The concrete steps that described genetic algorithm realizes the fitting of simulation result and actual numerical value within error range are:

Parameter setting: Determine the population size of each generation in the genetic algorithm. The population is 8-15% of the total number of all grids; determine the stop condition of the genetic algorithm. When the type II norm of the difference between two adjacent generations of random numbers is less than 0.001, stop the algorithm; set the genetic algorithm to retain the elite, and the probability of mutation is 0.1 and 0.02;

Coding step: before searching, the solution data (random array) of the solution space is expressed as the genotype string structure data of the genetic space, and different combinations of the string structure data constitute different points;

Fitness evaluation step: use fitness evaluation to indicate the pros and cons of individuals or solutions. For research problems, the fitness function is the difference between the simulated stress results and the actual observations;

Selection step: select excellent individuals from the group as the parent generation for the next generation to reproduce;

Crossover step: Obtain a new generation of individuals that combines the characteristics of the parent individual through crossover;

Mutation step: Randomly select an individual from the population, and randomly change the value of a certain string in the string structure data of the selected individual to realize data mutation.

The principle of selecting excellent individuals in the selection step is to contribute one or more highly adaptable individuals with a high probability of offspring to the next generation.

An automatic fitting system for reservoir geomechanics parameters, which includes a derivation programming module, a random field generation module, a simulation module and an optimal fitting module; the derivation becomes a module to realize the realization process of the derivation Karhunen-Loeve expansion method and according to The realization process carries out corresponding programming; Described random field generation module realizes the mechanical parameter that will obtain by Karhunen-Loeve expansion method and generates the random field about mechanical parameter; Described simulation module realizes that random field is imported in Abaqus simulator and stores Layer stress simulation; the optimization fitting module realizes the fitting between the simulated stress result and the actual observed value within the error range through an optimization algorithm.

The simulation module includes an import unit and a condition assignment unit; the import unit realizes that the mechanical property random field generated by the Karhunen-Loeve expansion method is imported into the Abaqus simulator; the condition assignment unit is implemented in the Abaqus simulator A horizontal strain boundary condition and a vertical stress boundary condition from the overlying strata are given to the random field, and the stress distribution value of the random field is obtained.

The optimal fitting module includes parameter setting, encoding unit, initial population generation unit, evaluation unit, selection crossover unit and variation unit;

The coding unit realizes that the solution data (random array) of the solution space is expressed as the genotype string structure data of the genetic space before the GA searches, and different combinations of the string structure data constitute different points; The fitness evaluation is used to indicate the pros and cons of individuals or solutions. For research problems, the fitness function is the difference between the simulated stress results and the actual observed values; the selection cross unit realizes selecting excellent individuals from the population as The next generation reproduces the parent generation of the offspring, and obtains a new generation of individuals that combines the characteristics of the parents' individuals through crossover; the mutation unit realizes random selection of an individual from the population, and randomly changes the selected individual in the string structure data A string of values to achieve data mutation.

The beneficial effects of the present invention are: a method and system for automatically fitting geomechanical parameters of reservoirs, using the Karhunen-Loeve expansion method to realize the expansion of the random field, using Abaqus software to simulate the stress distribution under formation conditions, and using the optimization algorithm to fit accurately A method for automatic fitting of geomechanical parameters is established, which can automatically fit reservoir geomechanical parameters and obtain a continuously distributed parameter field, which is helpful for engineers to quickly and accurately fit reservoir geomechanical parameters has important guiding significance.

Description of drawings

Fig. 1 is a flowchart of the present invention;

Fig. 2 is the accurate Young's modulus parameter schematic diagram;

Figure 3 is a schematic diagram of the first-generation Young's modulus parameter field generated by the Karhunen-Loeve expansion method;

Figure 4 is a schematic diagram of the 25th generation Young's modulus parameter field generated by the Karhunen-Loeve expansion method;

Figure 5 is a schematic diagram of drawing accurate stress distribution using Abaqus software;

Figure 6 is a schematic diagram of the stress distribution of the first generation model simulated by Abaqus software;

Figure 7 shows the stress distribution in the 25th generation model simulated by Abaqus software.

Detailed ways

The present invention will be further described below in conjunction with the accompanying drawings, but the protection scope of the present invention is not limited to the following description.

As shown in Figure 1, an automatic fitting method for reservoir geomechanics parameters, which includes the following steps:

S1, generate a random field about the mechanical parameters from the obtained mechanical parameters through the Karhunen-Loeve expansion method;

S2. Import the random field into the Abaqus simulator to simulate the stress of the reservoir;

S3. Realize the fitting between the simulation result and the actual value within the allowable error range through the optimization algorithm.

The step of derivation of the Karhunen-Loeve expansion method needs to be completed before performing the step of generating a random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method.

After completing the derivation process steps of the Karhunen-Loeve expansion method and before performing the step of generating a random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method, it is also necessary to complete the Karhunen-Loeve according to the derivation implementation process Related procedures of the expansion method.

The specific content of the step of generating the random field about the mechanical parameters from the mechanical parameters obtained by the Karhunen-Loeve expansion method is as follows:

S11. Generate N mutually independent Gaussian random variables ξ _i (θ);

S12, calculating the integral equation of the covariance matrix to obtain the eigenvalue λ _n and the eigenfunction f _n ;

S13, eigenvalue λ _n and eigenfunction f _n and Gaussian random variable ξ _i (θ) are substituted into Take N-order truncation to realize the generation of a random field;

S14. Repeat the third step above for P times to realize the generation of random fields for P times;

S15. After P times of random field generation is realized, the following matrix is obtained according to the random variable ξ _i (θ):

The random variable ξ _i (θ) satisfies the standard normal distribution, and satisfies E[ξ _i (θ)]=0; E[ξ _i (θ)ξ _j (θ)]=δ _ij , wherein, δ _ij is the Kronecker-delta function.

Described Abaqus simulator comprises poroelastic finite element simulator; Described import random field in the Abaqus simulator and carry out the concrete content of reservoir ground stress simulation step as follows:

S21. Import the random field of mechanical properties generated by the Karhunen-Loeve expansion method into the Abaqus simulator;

S22, assign boundary conditions and initial stress conditions to the random field in the Abaqus simulator;

S23. Obtain the stress distribution value of the random field.

Described optimization algorithm comprises genetic algorithm; The concrete steps that described genetic algorithm realizes the fitting of simulation result and actual numerical value within error range are:

Coding step: GA expresses the solution data in the solution space as genotype string structure data in the genetic space before searching, and different combinations of the string structure data constitute different points;

Step of generating an initial group: Randomly generate M initial string structure data, each string structure data is called an individual, and M individuals form a group; wherein, the GA starts with the M string structure data as an initial point evolution;

Fitness evaluation step: use fitness evaluation to indicate the pros and cons of individuals or solutions, and implement different definitions of fitness functions for different problems;

Selection step: select excellent individuals from the group as the parent generation for the next generation to reproduce;

Crossover step: Obtain a new generation of individuals that combines the characteristics of the parent individual through crossover;

Mutation step: Randomly select an individual from the population, and randomly change the value of a certain string in the string structure data of the selected individual to realize data mutation.

Among them, the probability of mutation in GA is very low, so the value is usually very small.

The principle of selecting excellent individuals in the selection step is to contribute one or more highly adaptable individuals with a high probability of offspring (maximum entry probability) to the next generation.

An automatic fitting system for reservoir geomechanics parameters, which includes a derivation programming module, a random field generation module, a simulation module and an optimal fitting module; the derivation becomes a module to realize the realization process of the derivation Karhunen-Loeve expansion method and according to The realization process carries out corresponding programming; Described random field generation module realizes the mechanical parameter that will obtain by Karhunen-Loeve expansion method and generates the random field about mechanical parameter; Described simulation module realizes that random field is imported in Abaqus simulator and stores Layer ground stress simulation; the optimization fitting module realizes the fitting between the simulation result and the actual value within the error range through an optimization algorithm.

The simulation module includes an import unit and a condition assignment unit; the import unit realizes that the mechanical property random field generated by the Karhunen-Loeve expansion method is imported into the Abaqus simulator; the condition assignment unit is implemented in the Abaqus simulator Boundary conditions and initial stress conditions are given to the random field, and the stress distribution value of the random field is obtained.

The optimal fitting module includes a coding unit, an initial population generation unit, an evaluation unit, a selection crossover unit and a variation unit;

The coding unit realizes that GA expresses the solution data of the solution space as genotype string structure data of the genetic space before searching, and different combinations of the string structure data constitute different points; the initial population The generating unit realizes random generation of M initial string structure data, each string structure data is called an individual, and M individuals form a group, and the GA begins to evolve with the M string structure data as an initial point; The evaluation unit realizes the fitness evaluation to indicate the pros and cons of the individual or the solution. For different problems, different definition methods of the fitness function are realized; the selection intersection unit realizes the selection of excellent individuals from the group as the following A generation of parents who breed offspring, and obtain a new generation of individuals that combines the characteristics of the parents' individuals through crossover; the mutation unit realizes random selection of an individual from the population, and randomly changes the selected individual to a certain value in the string structure data. A string of values to mutate the data.

Preferably, as shown in FIGS. 2-7 , the method for automatic fitting of reservoir geomechanics parameters described in the present invention is introduced by taking the distribution of Young's modulus of a small-scale theoretical model as an example. The basic parameters are as follows:

The mean value in the random field of Young's modulus parameter is 5; the variance is 1; the theoretical model size: 13*13*3; the truncation item in the Karhunen-Loeve expansion method is N=20; the number of individuals included in each generation population is M=20; A total of 25 generations of individuals were generated.

As the reference field, the accurate Young's modulus parameter field is used to evaluate the fitting effect of the present invention. As can be seen from Fig. 2 and Fig. 3, the first generation Young's modulus random field and the reference field Young's modulus distribution There is a large deviation, and the randomness has been fully considered; from Figure 4, it can be found that Figure 2 and Figure 4 have a high degree of similarity, indicating that after 25 generations of genetic algorithm optimization selection, the automatically fitted poplar The modulus parameter field can characterize the reference field more accurately. Further observing the stress distribution of the model, it can be found that there is a large deviation between the stress distribution of the first generation model simulated by Abaqus software in Figure 6 and the accurate stress distribution in Figure 5 of the model, and after 25 generations of automatic fitting, the obtained model The stress distribution schematic diagram 7 is highly similar to the accurate stress distribution diagram 5, indicating that the present invention can accurately and quickly realize the automatic fitting of the model parameter field. Further, for reservoir geological simulation, it can accurately and automatically fit the reservoir geomechanical parameter field, with simple operation, fast implementation speed and high fitting accuracy.

The above description does not limit the present invention in any form. Although the present invention has been disclosed by the above-mentioned embodiments, it is not intended to limit the present invention. When the technical content disclosed above can be used to make some changes or be modified into equivalent embodiments with equivalent changes, but if they do not deviate from the content of the technical solution of the present invention, any simple modifications made to the above embodiments according to the technical essence of the present invention, are equivalent to Changes and modifications all still belong to the scope of the technical solution of the present invention.

Claims (9)

1. A reservoir geomechanical parameter automatic fitting method is characterized in that: it comprises the following steps:

Generating a random field about the mechanical parameters by a Karhunen-Loeve expansion method;

Guiding the random field into an Abaqus simulator to simulate the reservoir crustal stress;

fitting the simulation result and the actual numerical value within an allowable error range through an optimization algorithm;

the Abaqus simulator comprises a pore elastic finite element simulator; the specific contents of the step of guiding the random field into an Abaqus simulator to simulate the reservoir ground stress are as follows:

Introducing a mechanical property random field generated by a Karhunen-Loeve expansion method into an Abaqus simulator;

endowing a strain boundary condition in the horizontal direction and a vertical stress boundary condition from an overburden layer to the pore elastic mechanics model in an Abaqus simulator;

The stress tensor distribution values for the random field are obtained.

2. The method for automatically fitting reservoir geomechanical parameters of claim 1, wherein: the derivation of the Karhunen-Loeve expansion is also required before the step of generating random fields for mechanical parameters by Karhunen-Loeve expansion is performed.

3. the method for automatically fitting reservoir geomechanical parameters of claim 2, wherein: after completing the Karhunen-

After the derivation procedure of the Loeve expansion method and before the step of generating the random field about the mechanical parameters from the obtained mechanical parameters by using the Karhunen-Loeve expansion method, it is necessary to complete the related procedures of writing the Karhunen-Loeve expansion method according to the derivation implementation procedure.

4. The method for automatically fitting reservoir geomechanical parameters of claim 2, wherein: the specific content of the step of generating the random field related to the mechanical parameters by using the Karhunen-Loeve expansion method is as follows:

Calculating an integral equation of the covariance matrix to obtain an eigenvalue lambda_nand a characteristic function f_n；

Generating M independent Gaussian random variables epsilon_i(θ), the length M is determined by: firstly, all characteristic values are arranged in a descending order; making the sum of the first M eigenvalues greater than 80% of the sum of all eigenvalues;

the characteristic value lambda is measured_nAnd a characteristic function f_nand a Gaussian random variable ε_isubstitution of (theta)

taking M-order truncation to realize generation of a primary random field;

Repeating the third step for P times to realize the generation of the random field for P times;

After P times of generation of random field is realized, according to random variable xi_i(θ) the following matrix is obtained:

5. the method for automatically fitting reservoir geomechanical parameters of claim 4, wherein: the random variable epsilon_i(theta) satisfies the standard normal distribution and satisfies E [ epsilon ]_i(θ)]＝0；E[ε_i(θ)ε_j(θ)]＝δ_ijwherein, delta_ijIs a Kronecker-delta function.

6. the method for automatically fitting reservoir geomechanical parameters of claim 1, wherein: the optimization algorithm comprises a genetic algorithm; the specific steps of the genetic algorithm for realizing the fitting of the simulation result and the actual observation value in the error range and updating the random number are as follows: setting parameters: determining the population number of each generation in the genetic algorithm, wherein the population number is 8-15% of the total number of all grids; determining a stopping condition of the genetic algorithm, and stopping the algorithm when the II-type norm of the difference value of two adjacent generations of random numbers is less than 0.001; setting the retention of elite in the genetic algorithm and the mutation probability to be 0.1 and 0.02;

and (3) encoding: before searching, expressing solution data of a solution space into genotype string structure data of a genetic space, wherein different combinations of the string structure data form different points;

And a fitness evaluation step: the fitness evaluation is used for indicating the advantages and disadvantages of the individual or the solution, and for the research problem, a fitness function is the difference between the simulated stress result and the actual observed value;

selecting: selecting excellent individuals from the group as parents for breeding descendants for the next generation;

A crossing step: obtaining a new generation of individuals with combined characteristics of the parents through crossing;

A mutation step: randomly selecting an individual from the group, and randomly changing the value of a certain string in the string structure data by the selected individual to realize the data mutation.

7. a reservoir geomechanical parameters auto-fitting system according to any one of claims 1 to 6, wherein: the device comprises a derivation programming module, a random field generation module, a simulation module and an optimization fitting module; the derivation programming module realizes the implementation process of deriving the Karhunen-Loeve expansion method and carries out corresponding programming according to the implementation process; the random field generation module generates a random field related to mechanical parameters by a Karhunen-Loeve expansion method; the simulation module is used for guiding a random field into an Abaqus simulator to simulate the reservoir ground stress; the optimization fitting module realizes fitting of a simulation result and an actual numerical value within an error range through an optimization algorithm, so that an optimal geomechanical parameter field is obtained.

8. A reservoir geomechanical parameters auto-fitting system of claim 7, wherein: the simulation module comprises an importing unit and a condition endowing unit; the leading-in unit leads the mechanical property random field generated by the Karhunen-Loeve expansion method into an Abaqus simulator; the condition endowing unit endows the random field with horizontal strain and vertical stress boundary conditions in an Abaqus simulator and obtains a stress distribution value related to the random field.

9. A reservoir geomechanical parameters auto-fitting system of claim 7, wherein: the optimization fitting module comprises an initial parameter setting unit, a coding unit, an elite unit, an evaluation unit, a cross selection unit and a variation unit;

setting the initial parameters to determine the population of each generation in the genetic algorithm, wherein the population is 8-15% of the total number of all grids; determining a stopping condition of the genetic algorithm, and stopping the algorithm when the II-type norm of the difference value of two adjacent generations of random numbers is less than 0.001; setting the retention of elite in the genetic algorithm and the mutation probability to be 0.1 and 0.02;

The encoding unit realizes that the solution data of the solution space is expressed into genotype string structure data of a genetic space before the GA searches, and different combinations of the string structure data form different points; the fitness evaluation is used for indicating the advantages and disadvantages of the individual or the solution, and for the research problem, a fitness function is the difference between the simulated stress result and the actual observed value; the selective crossing unit is used for selecting excellent individuals from the group as parents for breeding descendants for the next generation, and obtaining a new generation of individuals combined with the characteristics of the parents through crossing; the mutation unit randomly selects an individual from the group, and randomly changes the value of a certain string in the string structure data of the selected individual to realize the mutation of the data.