Method and system for optimizing complex fracture networks in shale reservoirs

Abstract

The present invention belongs to the technical field of fracture network optimization in oilfield exploitation. Specifically disclosed are a method and system for optimizing complex fracture networks in shale reservoirs. The method comprises: S1, on the basis of geological-engineering bimodal data fusion, constructing a three-dimensional digital twin of heterogeneous reservoirs; S2, synchronously calculating the interactions among fluid flow, rock fracturing and proppant transport during a hydraulic fracturing process; and S3, automatically and iteratively optimizing a combination of fracturing operation parameters. By means of constructing a closed-loop system consisting of geological modeling, multi-field coupling simulation, intelligent optimization and dynamic correction, efficient design and real-time optimization of complex fracture networks in shale reservoirs are realized.

Description

Method and system for optimizing complex fracture network of shale reservoir TECHNICAL FIELD

[0001] The present application relates to the technical field of oilfield development fracture network optimization, in particular to a method and system for optimizing complex fracture network of shale reservoir. BACKGROUND

[0002] Currently, shale reservoir hydraulic fracturing technology faces multiple challenges, which seriously restricts the efficient construction of complex fracture network and the maximization of reservoir reconstruction effect. First, the traditional geologic modeling method mainly relies on static logging data and seismic interpretation, which is difficult to accurately represent the heterogeneity of the reservoir and the distribution of the natural fracture network, resulting in significant deviation between the fracturing simulation results and the actual fracture network expansion. In addition, the conventional stress field analysis usually ignores the dynamic changes in the fracturing process, making the interaction prediction of artificial fractures and natural fractures distorted;

[0003] Secondly, existing fracturing simulation techniques are mostly based on a single physical field (such as only considering fluid flow or rock rupture), ignoring the multi-field coupling effect of fluid-rock-proppant, resulting in insufficient accuracy of fracture network expansion simulation. At the same time, traditional numerical simulation has low computational efficiency, and it often takes weeks or even months to optimize fracturing parameters, which cannot meet the needs of rapid decision-making on site;

[0004] Finally, existing fracturing operation schemes are mostly static designs, lacking real-time data feedback and dynamic optimization capabilities. Although advanced monitoring technologies such as downhole optical fibers (DAS / DTS) can provide rich data, traditional methods are difficult to efficiently integrate these information for real-time model correction, resulting in increased risk of sand plugging and premature closure of fracture network, affecting the fracturing effect. SUMMARY

[0005] This section aims to outline some aspects of the embodiments of the present application and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract and title of the specification to avoid obscuring the purpose of this section, abstract and title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0006] Therefore, the purpose of the present application is to provide a method and system for optimizing complex fracture network of shale reservoir, which realizes efficient design and real-time optimization of complex fracture network of shale reservoir by constructing a closed-loop system of geologic modeling-multi-field coupling simulation-intelligent optimization-dynamic correction.

[0007] To solve the above technical problems, according to one aspect of the present application, the present application provides the following technical scheme:

[0008] A method for optimizing complex fracture network of shale reservoir, comprising:

[0009] S1. Based on geology-engineering dual-mode data fusion, a three-dimensional heterogeneous reservoir digital twin is constructed, wherein a probabilistic model of natural fracture network and a dynamic evolution algorithm of geostress field are embedded;

[0010] S2. A multi-physics coupled fracture network extension simulator is used to synchronously calculate the interaction of fluid flow-rock failure-proppant migration during hydraulic fracturing, and the simulator is integrated with a machine learning accelerated discrete fracture network generation module;

[0011] S3. Through a reverse reinforcement learning algorithm, a fracture complexity index and a reconstruction volume are used as joint optimization objectives to automatically and iteratively optimize a fracturing operation parameter combination;

[0012] S4. Real-time fiber optic DAS / DTS monitoring data is received, and an online Bayesian updating mechanism is used to dynamically correct the mechanical parameters and fracture network prediction model of the digital twin;

[0013] S5. A three-dimensional fracturing design scheme containing a non-planar fracture network topology is output to guide field construction.

[0014] As a preferred scheme of the method for optimizing a complex fracture network of a shale reservoir according to the present application, in the step S1, the probabilistic model uses an improved Markov random field algorithm, and by introducing a geomechanical similarity constraint term, the confidence difference between outcrop observation data and logging interpretation data is quantified and incorporated into fracture spatial distribution prediction.

[0015] As a preferred scheme of the method for optimizing a complex fracture network of a shale reservoir according to the present application, in the step S3, the reverse reinforcement learning algorithm specifically includes:

[0016] A reward function of a virtual fracturing engineer is constructed, a parameter space is explored through Monte Carlo tree search, and a double-delay deep deterministic policy gradient algorithm is used to realize continuous optimization of construction parameters.

[0017] As a preferred scheme of the method for optimizing a complex fracture network of a shale reservoir according to the present application, in the step S4, the online Bayesian updating mechanism includes:

[0018] A proxy model based on a variational autoencoder is established to map fiber monitoring data to a latent space;

[0019] A KL divergence driven model parameter automatic adjustment strategy is designed to realize time-varying consistency between the digital twin and the physical reservoir.

[0020] A system for optimizing a complex fracture network of a shale reservoir includes:

[0021] The fusion modeling module comprises a microseismic inversion unit, a core CT scanning analysis unit and a ground stress field reconstruction unit;

[0022] Quantum computing accelerated multi-physics coupling simulator integrated with a crack propagation prediction engine based on a graph neural network;

[0023] Parameter optimization platform configured with a deep deterministic policy gradient (DDPG) algorithm and a Pareto front multi-objective optimizer;

[0024] Assimilation interface supporting real-time stream processing and model updating of distributed optical fiber sensing data;

[0025] Decision terminal that can interactively display the quantitative relationship between the fracture network and the reservoir reconstruction effect.

[0026] As a preferred scheme of the shale reservoir complex fracture network optimization system, the multi-physics coupling simulator adopts a heterogeneous computing architecture:

[0027] The rock fracture process calculation is deployed on a GPU cluster to run parallel calculations based on the phase field method;

[0028] The proppant transport simulation adopts a lattice Boltzmann method accelerated by FPGA hardware;

[0029] The fluid flow solver optimizes the flow channel network parameters through a quantum annealing algorithm.

[0030] As a preferred scheme of the shale reservoir complex fracture network optimization system, the decision terminal is integrated with:

[0031] Fracture network fractal dimension calculation module;

[0032] Dynamic correlation analysis tool for stimulated volume and matrix contact area;

[0033] Wellbore-fracture network spatial relationship perspective function based on augmented reality.

[0034] Compared with the prior art, the present application has the beneficial effects that:

[0035] Fracture network complexity improvement: through geology-engineering bimodal fusion modeling, the natural fracture system is accurately activated to form a three-dimensional non-planar fracture network;

[0036] Design cycle shortened: the quantum accelerated multi-physics coupling simulator compresses the traditional weeks of calculation to hours;

[0037] Dynamic adaptability: real-time assimilation technology of optical fiber data dynamically adjusts the construction scheme in response to the reservoir, reducing the risk of sand plugging and other construction risks;

[0038] Full life cycle management: digital twin closed-loop system from fracturing design to production optimization, reduce the cost of tons of oil. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the present application will be described in detail below with reference to the drawings and detailed embodiments. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor. Among them:

[0040] Figure 1 is a schematic block diagram of the method steps of the present application;

[0041] Figure 2 is a schematic block diagram of the system structure of the present application. DETAILED DESCRIPTION

[0042] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings.

[0043] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0044] Secondly, the present application is described in detail in combination with the schematic diagram, in the detailed description of the embodiments of the present application, in order to facilitate the description, the cross-sectional view of the device structure will be partially enlarged without general proportion, and the schematic diagram is only an example, which should not limit the scope of protection of the present application here. In addition, the three-dimensional spatial dimensions of length, width and depth should be included in actual manufacture.

[0045] In order to make the purposes, technical solutions and advantages of the present application more clear, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0046] The present application provides a method and system for optimizing complex fracture network of shale reservoir, please refer to figures 1-2:

[0047] A method for optimizing complex fracture network of shale reservoir, characterized in that, comprising the following steps:

[0048] S1. Based on the fusion of geological-engineering bimodal data, a three-dimensional heterogeneous reservoir digital twin is constructed, wherein the probability model of natural fracture network and the dynamic evolution algorithm of geostress field are embedded;

[0049] S2. A multi-physics coupled fracture network simulator is used to synchronously calculate the interaction of fluid flow-rock failure-proppant migration during hydraulic fracturing, which integrates a machine learning accelerated discrete fracture network generation module;

[0050] S3. A reverse reinforcement learning algorithm is used to automatically iteratively optimize the combination of fracturing operation parameters with the fracture complexity index and the stimulated volume as the joint optimization objectives;

[0051] S4. Real-time downhole fiber DAS / DTS monitoring data is received to dynamically correct the mechanical parameters and fracture network prediction model of the digital twin through an online Bayesian updating mechanism;

[0052] S5. A three-dimensional fracturing design scheme containing a non-planar fracture network topology is output to guide field construction.

[0053] The probability model in step S1 uses an improved Markov random field algorithm, which quantifies the confidence difference between outcrop observation data and logging interpretation data into fracture spatial distribution prediction by introducing a geomechanical similarity constraint term.

[0054] The reverse reinforcement learning algorithm in step S3 specifically includes:

[0055] A reward function of a virtual fracturing engineer is constructed, the parameter space is explored through Monte Carlo tree search, and the continuous optimization of operation parameters is realized by using a double-delay deep deterministic policy gradient algorithm.

[0056] The online Bayesian updating mechanism in step S4 includes:

[0057] A proxy model based on a variational autoencoder is established to map fiber monitoring data to a latent space;

[0058] A model parameter automatic adjustment strategy driven by KL divergence is designed to realize the time-varying consistency of the digital twin and the physical reservoir.

[0059] A shale reservoir complex fracture network optimization system, characterized by comprising:

[0060] A fusion modeling module including a microseismic inversion unit, a core CT scanning analysis unit, and a geostress field reconstruction unit;

[0061] A quantum computing accelerated multi-physics coupled simulator integrating a fracture expansion prediction engine based on a graph neural network;

[0062] A parameter optimization platform configured with a deep deterministic policy gradient (DDPG) algorithm and a Pareto front multi-objective optimizer;

[0063] An assimilation interface supporting real-time stream processing and model updating of distributed fiber sensing data;

[0064] Decision terminal, which can interactively display the quantitative relationship between fracture network and reservoir reconstruction effect.

[0065] The multi-physics coupling simulator adopts a heterogeneous computing architecture, wherein:

[0066] The rock fracture process calculation is deployed on a GPU cluster to run parallel calculations based on the phase field method;

[0067] The proppant transport simulation adopts a lattice Boltzmann method accelerated by FPGA hardware;

[0068] The fluid flow solver optimizes the flow channel network parameters through a quantum annealing algorithm.

[0069] The decision terminal is integrated with:

[0070] A fracture network fractal dimension calculation module;

[0071] A dynamic correlation analysis tool for the contact area between the reconstruction volume and the matrix;

[0072] An augmented reality-based wellbore-fracture network spatial relationship perspective function;

[0073] Working principle:

[0074] Geology-engineering bimodal fusion modeling:

[0075] Integrating logging, core CT scanning, microseismic, and outcrop data, an improved Markov random field algorithm is used to construct a three-dimensional heterogeneous reservoir digital twin;

[0076] A geostress dynamic evolution model is introduced, combined with the probability distribution of natural fractures, to accurately characterize the mechanical properties of the reservoir and the spatial distribution of the fracture network;

[0077] Multi-physics coupling simulation (quantum computing acceleration):

[0078] A GPU+FPGA+quantum computing hybrid architecture is used to solve the coupled processes of fluid flow-rock fracture-proppant transport in parallel:

[0079] Fluid flow: based on the lattice Boltzmann method, the non-Newtonian fluid behavior of fracturing fluid in complex fracture networks is simulated;

[0080] Rock fracture: the phase field method (Phase-Field) is used to simulate the dynamic expansion of hydraulic fractures and their interaction with natural fractures;

[0081] Proppant transport: combined with machine learning, the settling and placement rules of proppants of different particle sizes in the fracture network are predicted;

[0082] Inverse reinforcement learning (IRL) driven parameter optimization:

[0083] Construct a virtual fracturing engineer reward function to optimize fracture complexity index (FCI) and stimulated reservoir volume (SRV);

[0084] Use Monte Carlo tree search (MCTS) + TD3 reinforcement learning algorithm to automatically explore the optimal combination of construction parameters such as displacement, sand ratio, and liquid viscosity;

[0085] Real-time data assimilation and dynamic correction:

[0086] Real-time monitoring of temperature, strain, and acoustic signals through downhole optical fibers (DAS / DTS) and feedback to the digital twin;

[0087] Use Bayesian update + variational autoencoder (VAE) to dynamically adjust model parameters to ensure that the simulation results are consistent with the actual reservoir response;

[0088] Three-dimensional visual decision support:

[0089] Output non-planar fracture network topology and combine with augmented reality (AR) technology to visually display the spatial relationship between fractures and wellbores;

[0090] Generate risk thermodynamic maps (such as sand plug early warning, fracture network interference analysis) to guide real-time adjustment of construction plans.

[0091] Although the present application has been described with reference to the embodiments above, various improvements can be made thereto and equivalents can be substituted therefor without departing from the scope of the present application. In particular, features of the disclosed embodiments can be combined together in any manner as long as there is no structural conflict. The combinations are not exhaustively described in the specification only for the purpose of saving space and resources. Therefore, the present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for optimizing complex fracture networks in shale reservoirs, characterized by, Comprising the following steps: S1. Constructing a three-dimensional heterogeneous reservoir digital twin based on geology-engineering dual-modal data fusion, wherein a probabilistic model of natural fracture network and a dynamic evolution algorithm of geostress field are embedded; S2. Simultaneously calculating the interaction of fluid flow-rock failure-proppant migration during hydraulic fracturing process by using a multi-physics coupled fracture network propagator, which integrates a machine learning accelerated discrete fracture network generation module; S3. Automatically and iteratively optimizing the combination of fracturing operation parameters by using a reverse reinforcement learning algorithm, with fracture complexity index and stimulated volume as joint optimization objectives; S4. Real-time receiving downhole fiber DAS / DTS monitoring data, and dynamically correcting the mechanical parameters and fracture network prediction model of the digital twin through an online Bayesian updating mechanism; S5. Outputting a three-dimensional fracturing design scheme containing non-planar fracture network topology, to guide field operation.

2. The method for optimizing complex fracture network in shale reservoirs of claim 1, wherein, The probabilistic model in step S1 adopts an improved Markov random field algorithm, which quantifies the confidence difference between outcrop observation data and logging interpretation data into fracture spatial distribution prediction by introducing a geomechanical similarity constraint term.

3. The method for optimizing complex fracture network in shale reservoirs of claim 1, wherein, The reverse reinforcement learning algorithm in step S3 specifically includes: Constructing a reward function of virtual fracturing engineer, exploring parameter space through Monte Carlo tree search, and realizing continuous optimization of operation parameters by using a double-delay deep deterministic policy gradient algorithm.

4. The method for optimizing complex fracture network in shale reservoirs of claim 1, wherein, The online Bayesian updating mechanism in step S4 includes: Establishing a proxy model based on variational autoencoder to map fiber monitoring data to latent space; Designing a KL divergence driven model parameter automatic adjustment strategy to realize time-varying consistency between digital twin and physical reservoir.

5. A shale reservoir complex fracture network optimization system, characterized in that, Comprise: Fusion modeling module, including microseismic inversion unit, core CT scanning analysis unit and geostress field reconstruction unit; Quantum computing accelerated multi-physics coupled simulator, integrated with a fracture propagation prediction engine based on graph neural network; Parameter optimization platform, configured with deep deterministic policy gradient (DDPG) algorithm and Pareto front multi-objective optimizer; Assimilation interface, supporting real-time stream processing and model updating of distributed fiber sensing data; Decision terminal, which can interactively display the quantitative relationship between fracture network and reservoir reconstruction effect.

6. A system for optimizing complex fracture network optimization in shale reservoirs according to claim 5, wherein, The multi-physics coupled simulator adopts a heterogeneous computing architecture: Rock failure process calculation is deployed on a GPU cluster to run parallel computation based on phase field method; Proppant migration simulation uses FPGA hardware accelerated lattice Boltzmann method; Fluid flow solver optimizes flow channel network parameters through quantum annealing algorithm.

7. A system for optimizing complex fracture network optimization in shale reservoirs according to claim 5, wherein, The decision terminal integrates: Fractal dimension calculation module of fracture network; Dynamic correlation analysis tool of stimulated volume and matrix contact area; Wellbore-fracture network spatial relationship perspective function based on augmented reality.

Images