Shale gas reservoir apparent permeability prediction method and related device

Abstract

The invention belongs to the technical field of petroleum engineering, and discloses a shale gas reservoir apparent permeability prediction method and a related device, and the method comprises the steps: obtaining data; judging a gas flow state; establishing a shale gas mass migration model; establishing apparent permeability models of capillary tubes with different sizes; obtaining a shale reservoir apparent permeability model; carrying out transfer learning; according to the method, the influence of the reservoir effective stress and the shale water saturation on the permeability is comprehensively considered, the neural network model is adopted to obtain the effective flow radius, and the prediction accuracy is improved; by introducing the permeability correction coefficient, dynamic correction of the original permeability value is realized, so that the actual permeability can be reflected more accurately; according to the method, by adopting transfer learning, a specific type of shale reservoir apparent permeability model is transferred to other conditions, so that the model can be used for predicting the permeability of different types of shale reservoirs, and the universality is effectively improved.

Description

Technical Field

[0001] The invention belongs to the technical field of petroleum engineering, and relates to a method for predicting the apparent permeability of a shale gas reservoir and a related device. Background Art

[0002] The shale reservoir apparent permeability model can be used to accurately evaluate the reserves, production capacity and development potential of shale gas reservoirs. The model can predict the gas flow characteristics under different conditions, thereby helping decision makers to judge the mining value and economic benefits of gas reservoirs. Therefore, the shale reservoir apparent permeability model has a wide range of application value in petroleum engineering and is of great significance for promoting the rational development and utilization of shale gas resources. The existing shale reservoir apparent permeability model has the following defects when applied: First, reservoir water saturation and stress sensitivity effect are important factors affecting the apparent permeability of shale reservoirs. The existing technology solves the problem of the influence of reservoir water saturation and stress sensitivity effect on the apparent permeability of shale reservoirs by deriving an equivalent formula of the effective flow radius, but simplifying the complex physical process into an equivalent formula of the effective flow radius will ignore the influence on other important parameters such as gas diffusion coefficient and porosity; second, due to the complexity and variability of shale reservoirs, the predicted permeability value cannot accurately reflect the actual permeability; third, different shale reservoirs have different geological characteristics and flow characteristics. The existing model is generally only applicable to specific types of shale reservoirs, and needs to be adjusted or corrected for other situations, so the versatility needs to be improved. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a method and related devices for predicting the apparent permeability of shale gas reservoirs. The present invention adopts transfer learning to migrate the apparent permeability model of a specific type of shale reservoir to other situations, so that the model can be used to predict the permeability of different types of shale reservoirs, effectively improving the versatility.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is as follows: The method for predicting the apparent permeability of shale gas reservoirs includes the following steps: Construct an apparent permeability model for shale gas reservoirs; generating a data set using the shale gas reservoir apparent permeability model; Using the data set to train a deep learning model to obtain a trained deep learning model; The trained deep learning model is used as the source model, and the transfer learning method is adopted to train the migrated deep learning model using data of different types of shale reservoirs, and the parameters of the migrated deep learning model are adjusted to adapt to the characteristics of the target field. After the training is completed, a trained deep learning model for predicting the apparent permeability of different types of shale reservoirs is obtained; The apparent permeability of shale gas reservoirs is predicted using the trained deep learning model of shale reservoir apparent permeability.

[0005] The construction process of the shale gas reservoir apparent permeability model includes: Using the diameter of gas molecules , Effective flow radius corresponding to capillaries of different sizes , reservoir pressure 、Langmuir pressure , Slip coefficient corresponding to effective flow radius of capillaries of different sizes , Contribution coefficients corresponding to effective flow radius of capillaries of different sizes , Knudsen diffusion coefficient corresponding to capillaries of different sizes , gas viscosity , gas molar mass , surface diffusion coefficient , Maximum adsorption concentration of adsorbed gas and the intrinsic permeability of a single capillary The calculated permeability ; Using the actual measured permeability and the intrinsic permeability of a single capillary Calculate the correction factor ; Using correction factor and the calculated permeability Calculate the apparent permeability of capillaries of different sizes ; Using matrix porosity , rock tortuosity , the distribution frequency of capillaries of different sizes and the apparent permeability of capillaries of different sizes Construct an apparent permeability model for shale gas reservoirs.

[0006] Preferably, the correction factor The calculation formula is as follows:

[0007] in, is the permeability actually measured, is the intrinsic permeability of a single capillary; Calculated permeability The calculation formula is as follows:

[0008]

[0009]

[0010] Where: is the gas molecule diameter, in units of ; is the effective flow radius corresponding to capillaries of different sizes, in units of ; is the reservoir pressure, in ; is the Langmuir pressure, in units of ; is the slip coefficient corresponding to the effective flow radius of capillaries of different sizes; is the contribution coefficient corresponding to the effective flow radius of capillaries of different sizes; is the intrinsic permeability of a single capillary tube, in units of ; is the Knudsen diffusion coefficient corresponding to capillaries of different sizes, in units of ; is the gas density in ; is the gas viscosity, in units of ; is the molar mass of the gas in units of ; is the surface diffusion coefficient, in units of ; is the maximum adsorption concentration of adsorbed gas, in units of ; is the average reservoir pressure, in .

[0011] Preferably, the effective flow radius corresponding to capillaries of different sizes is The acquisition process includes: inputting the capillary flow radius, reservoir effective stress and shale water saturation into the neural network model, and the neural network model outputs the effective flow radius; the neural network model is trained with the capillary flow radius, reservoir effective stress and shale water saturation as input and the effective flow radius as output.

[0012] Preferably, the contribution coefficient calculation formula corresponding to the effective flow radius of capillaries of different sizes is as follows:

[0013] in: is a constant, with a value of 1; is the Knudsen coefficient corresponding to capillaries of different sizes; is the Knudsen number at which the transition from continuous flow to quasi-diffusive flow begins; is a constant; Knudsen coefficients for capillaries of different sizes The calculation formula is as follows:

[0014] in: is the Boltzmann constant, in units of ; is the reservoir pressure, in ; is the reservoir temperature, in ; is the collision diameter of gas molecules, in units of ; is the effective flow radius corresponding to capillaries of different sizes, in units of .

[0015] Preferably, the apparent permeability of capillaries of different sizes is The calculation formula is as follows:

[0016] in, is the calculated permeability, is the correction factor.

[0017] Preferably, the calculation formula of the apparent permeability of the shale gas reservoir is as follows:

[0018] Where: is the apparent permeability of the core, in units of ; is the matrix porosity; is the total number of pores; is the rock tortuosity; is the apparent permeability of capillaries of different sizes, in units of ; is the distribution frequency of capillaries of different scales; Rock tortuosity The calculation formula is as follows:

[0019] in: is the rock tortuosity fitting parameter, and its value is 0.77.

[0020] The present invention also provides a shale gas reservoir apparent permeability prediction system, which is used to implement the shale gas reservoir apparent permeability prediction method of the present invention, comprising: Model building unit: used to build the apparent permeability model of shale gas reservoirs; A data set generating unit: used for generating a data set by using the shale gas reservoir apparent permeability model; A first training unit: used to train a deep learning model using the data set to obtain a trained deep learning model; The second training unit is used to use the trained deep learning model as the source model, adopt the transfer learning method, use the data of different types of shale reservoirs to train the migrated deep learning model, and adjust the parameters of the migrated deep learning model to adapt to the characteristics of the target field. After the training is completed, a trained deep learning model for predicting the apparent permeability of different types of shale reservoirs is obtained; Prediction unit: used to predict the apparent permeability of shale gas reservoirs using the trained deep learning model of shale reservoir apparent permeability.

[0021] The present invention also provides an electronic device, comprising: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for predicting the apparent permeability of shale gas reservoirs as described above in the present invention.

[0022] The present invention also provides a storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the method for predicting the apparent permeability of a shale gas reservoir as described above.

[0023] Compared with the prior art, the present invention has the following beneficial effects: The present invention comprehensively considers the influence of reservoir effective stress and shale water saturation on permeability, adopts a neural network model to obtain the effective flow radius, and improves the accuracy of prediction; the present invention introduces a permeability correction coefficient to achieve dynamic correction of the original permeability value, so as to more accurately reflect the actual permeability; the present invention adopts transfer learning to transfer the apparent permeability model from a specific type of shale reservoir to other situations, so that the model can be used to predict the permeability of different types of shale reservoirs, effectively improving the versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a method flow chart of the method for predicting the apparent permeability of shale gas reservoirs in an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention is clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] See also Figure 1 The method for predicting the apparent permeability of shale gas reservoirs in the embodiment of the present invention mainly includes the following steps: Data acquisition: Obtain shale pipe diameter distribution and frequency and basic shale reservoir parameters, including reservoir temperature, gas type, etc.

[0027] Gas flow pattern judgment: The Knudsen coefficient corresponding to different shale capillary diameters in the reservoir environment is used to judge the flow pattern of gas in the capillary.

[0028] Establishment of shale gas mass migration model: A shale gas mass migration model is established based on gas flow patterns and migration mechanisms.

[0029] Establishment of apparent permeability model: Combining the unified mass transport equation and the intrinsic permeability of shale measured by indoor experiments, an apparent permeability model of capillaries of different sizes is established.

[0030] Permeability correction: The permeability correction coefficient is introduced to dynamically correct the original permeability value to more accurately reflect the actual permeability.

[0031] Transfer learning application: Use transfer learning technology to migrate the model from one specific type of shale reservoir to other types to improve the versatility of the model.

[0032] The specific method for predicting the apparent permeability of shale gas reservoirs in this embodiment is as follows: The calculation formula of the shale reservoir apparent permeability model in this embodiment is as follows:

[0033] Where: is the core apparent permeability, ; is the matrix porosity; is the total number of pores; is the rock tortuosity; is the apparent permeability of capillaries of different sizes, ; is the distribution frequency of capillaries of different scales; Rock tortuosity The calculation formula is as follows:

[0034] Where: is the rock tortuosity fitting parameter, and its value is 0.77.

[0035] The method for establishing the apparent permeability model of shale reservoir in petroleum engineering of this embodiment includes: step 1, obtaining data; step 2, determining the gas flow state; step 3, establishing a shale gas mass migration model; step 4, establishing the apparent permeability model of capillaries of different sizes; step 5, obtaining the apparent permeability model of shale reservoir; step 6, transfer learning; Among them, in the above step 1, the shale tube diameter distribution and frequency and the basic parameters of the shale reservoir are obtained, and the contribution coefficient and Knudsen coefficient under different capillary tube diameters are calculated. The basic parameters of the shale reservoir include shale reservoir temperature, gas type, gas constant, gas molar mass, gas viscosity, tangential momentum adjustment coefficient, gas molecular density, average pressure, surface maximum concentration, Langmuir pressure and surface diffusion coefficient; the contribution coefficient corresponding to the effective flow radius of capillaries of different sizes The calculation formula is as follows:

[0036] Where: is a constant, with a value of 1; is the Knudsen coefficient corresponding to capillaries of different sizes; is the Knudsen number at the beginning of the transition from continuous flow to quasi-diffusion flow, which is 0.3; is a constant, with a value of 1; Knudsen coefficients for capillaries of different sizes The calculation formula is as follows:

[0037] Where: is the Boltzmann constant, which is ; is the reservoir pressure, ; is the reservoir temperature, ; is a constant, ; is the collision diameter of gas molecules, ; is the effective flow radius, ; Effective flow radius The acquisition method specifically includes: inputting the capillary flow radius, reservoir effective stress and shale water saturation into the neural network model, and the neural network model outputs the effective flow radius; the neural network model is trained with the capillary flow radius, reservoir effective stress and shale water saturation as input and the effective flow radius as output; In the above step 2, the Knudsen coefficient corresponding to different shale capillary diameters under reservoir environment is used to determine the flow state of gas in the capillary; In the above step 3, a shale gas mass migration model is established based on the flow state and migration mechanism of gas in the capillary. The shale gas mass migration model is as follows:

[0038]

[0039]

[0040]

[0041] Where: is the total mass flow rate, ; is the viscous flow mass flow rate, ; is the slip effect mass flow rate, ; is the Knudsen diffusion mass flow rate, ; is the surface diffusion mass flow rate, ; is the gas density, ; is the gas viscosity, ; is the intrinsic permeability of a single capillary, ; is the diameter of gas molecules, ; is the effective flow radius of a single capillary, ; is the reservoir pressure, ; is the Langmuir pressure, ; is the pressure gradient operator symbol; is the slip coefficient; is the Knudsen diffusion coefficient, ; is the molar mass of the gas, ; is the surface diffusion coefficient, ; is the maximum adsorption concentration of adsorbed gas, ; is the gas constant, ; is the average reservoir pressure, ; is the tangential momentum adjustment coefficient, ranging from 0 to 1; is the shale reservoir temperature, ; Among them, in the above step 4, based on the unified mass transfer equation and combined with the inherent permeability of shale measured by indoor experiments, the apparent permeability model of capillaries of different sizes is established. The calculation formula of the apparent permeability of capillaries of different sizes is as follows:

[0042]

[0043] Where: is the calculated permeability, is the permeability actually measured, is the intrinsic permeability of a single capillary, is the correction factor.

[0044] Calculated permeability It is expressed as:

[0045]

[0046]

[0047] Where: is the diameter of gas molecules, ; is the effective flow radius corresponding to capillaries of different sizes, ; is the reservoir pressure, ; is the Langmuir pressure, ; is the slip coefficient corresponding to the effective flow radius of capillaries of different sizes; is the contribution coefficient corresponding to the effective flow radius of capillaries of different sizes; is the intrinsic permeability of a single capillary, ; is the Knudsen diffusion coefficient corresponding to capillaries of different sizes, ; is the gas density, ; is the gas viscosity, ; is the molar mass of the gas, ; is the surface diffusion coefficient, ; is the maximum adsorption concentration of adsorbed gas, ; In the above step 5, the apparent permeability of shales containing capillaries of different sizes is superimposed according to the distribution frequencies of capillaries of different sizes, so as to obtain an apparent permeability model of the entire shale reservoir; Step six, using the shale reservoir apparent permeability model obtained in step five to generate a data set, using the data set to train a deep learning model, using the trained model as the source model, using the transfer learning method, using data from different types of shale reservoirs to train the migrated model, adjusting the model parameters to adapt to the characteristics of the target field, and after training, a deep learning model for predicting the apparent permeability of different types of shale reservoirs can be obtained.

[0048] Based on the above, the advantage of the present invention lies in that, when the present invention is used, by comprehensively considering the multiple effects of reservoir effective stress and shale water saturation on permeability, the effective flow radius is accurately calculated using a neural network model, which significantly improves the prediction accuracy compared to the simplified processing using a single equivalent formula; at the same time, the present invention innovatively introduces a permeability correction coefficient to achieve dynamic calibration of the original permeability value, thereby more accurately reflecting the actual permeability conditions; in addition, with the help of transfer learning technology, the present invention successfully transfers the experience of the apparent permeability model of a specific type of shale reservoir to a variety of different scenarios, so that the model can be widely used to predict the permeability of various types of shale reservoirs, significantly enhancing its versatility and practicality.

[0049] The present invention also provides a shale gas reservoir apparent permeability prediction system for implementing the above prediction method, the system comprising: Model building unit: used to build the apparent permeability model of shale gas reservoirs; A data set generating unit: used for generating a data set by using the shale gas reservoir apparent permeability model; A first training unit: used to train a deep learning model using the data set to obtain a trained deep learning model; The second training unit is used to use the trained deep learning model as the source model, adopt the transfer learning method, use the data of different types of shale reservoirs to train the migrated deep learning model, and adjust the parameters of the migrated deep learning model to adapt to the characteristics of the target field. After the training is completed, a trained deep learning model for predicting the apparent permeability of different types of shale reservoirs is obtained; Prediction unit: used to predict the apparent permeability of shale gas reservoirs using the trained deep learning model of shale reservoir apparent permeability.

[0050] The embodiments of the present invention also provide corresponding electronic devices and computer-readable storage media for implementing the solutions provided by the embodiments of the present invention.

[0051] The device includes a memory and a processor, the memory is used to store instructions or codes, and the processor is used to execute the instructions or codes so that the device executes the shale gas reservoir apparent permeability prediction method described in any embodiment of the present application.

[0052] The storage medium stores a computer program, wherein the computer program, when executed by the processor, implements the method for predicting the apparent permeability of a shale gas reservoir described in any embodiment of the present application.

[0053] Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for predicting apparent permeability of shale gas reservoirs, characterized in that: The process includes the following: Construct an apparent permeability model for shale gas reservoirs; generating a data set using the shale gas reservoir apparent permeability model; Using the data set to train a deep learning model to obtain a trained deep learning model; The trained deep learning model is used as the source model, and the transfer learning method is adopted to train the migrated deep learning model using data of different types of shale reservoirs, and the parameters of the migrated deep learning model are adjusted to adapt to the characteristics of the target field. After the training is completed, a trained deep learning model for predicting the apparent permeability of different types of shale reservoirs is obtained; The apparent permeability of shale gas reservoirs is predicted using the trained deep learning model of shale reservoir apparent permeability.

2. The method for predicting apparent permeability of shale gas reservoirs according to claim 1, characterized in that: The construction process of the shale gas reservoir apparent permeability model includes: The permeability is calculated using the gas molecular diameter, the effective flow radius corresponding to capillaries of different sizes, the reservoir pressure, the Langmuir pressure, the slip coefficient corresponding to the effective flow radius of capillaries of different sizes, the contribution coefficient corresponding to the effective flow radius of capillaries of different sizes, the Knudsen diffusion coefficient corresponding to capillaries of different sizes, the gas viscosity, the gas molar mass, the surface diffusion coefficient, the maximum adsorption concentration of adsorbed gas and the intrinsic permeability of a single capillary; The correction factor is calculated using the permeability actually measured and the inherent permeability of a single capillary tube; Using correction factor The apparent permeability of capillaries of different sizes was calculated by comparing the calculated permeability; The apparent permeability model of shale gas reservoirs is constructed using matrix porosity, rock tortuosity, distribution frequency of capillaries of different scales and apparent permeability of capillaries of different sizes.

3. The method for predicting apparent permeability of shale gas reservoirs according to claim 2, characterized in that: Correction factor The calculation formula is as follows: in, is the permeability actually measured, is the intrinsic permeability of a single capillary; Calculated permeability The calculation formula is as follows: Where: is the gas molecule diameter, in units of ; is the effective flow radius corresponding to capillaries of different sizes, in units of ; is the reservoir pressure, in ; is the Langmuir pressure, in units of ; is the slip coefficient corresponding to the effective flow radius of capillaries of different sizes; is the contribution coefficient corresponding to the effective flow radius of capillaries of different sizes; is the intrinsic permeability of a single capillary tube, in units of ; is the Knudsen diffusion coefficient corresponding to capillaries of different sizes, in units of ; is the gas density in ; is the gas viscosity, in units of ; is the molar mass of the gas, in units of ; is the surface diffusion coefficient, in units of ; is the maximum adsorption concentration of adsorbed gas, in units of ; is the average reservoir pressure, in .

4. The method for predicting apparent permeability of shale gas reservoirs according to claim 3, characterized in that: The process of obtaining the effective flow radius corresponding to capillaries of different sizes includes: inputting the flow radius of the capillary, the effective stress of the reservoir and the water saturation of the shale into the neural network model, and the neural network model outputs the effective flow radius; the neural network model is trained with the flow radius of the capillary, the effective stress of the reservoir and the water saturation of the shale as input and the effective flow radius as output.

5. The method for predicting apparent permeability of shale gas reservoirs according to claim 3, characterized in that: Contribution coefficients corresponding to effective flow radius of capillaries of different sizes The calculation formula is as follows: in: is a constant; is the Knudsen coefficient corresponding to capillaries of different sizes; is the Knudsen number at which the transition from continuous flow to quasi-diffusive flow begins; is a constant; Knudsen coefficients for capillaries of different sizes The calculation formula is as follows: in: is the Boltzmann constant, in units of ; is the reservoir pressure, in ; is the reservoir temperature, in ; is the collision diameter of gas molecules, in units of ; is the effective flow radius corresponding to capillaries of different sizes, in units of .

6. The method for predicting apparent permeability of shale gas reservoirs according to claim 2, characterized in that: Apparent permeability of capillaries of different sizes The calculation formula is as follows: in, is the calculated permeability, is the correction factor.

7. The method for predicting apparent permeability of shale gas reservoirs according to claim 2, characterized in that: The calculation formula of apparent permeability of shale gas reservoir is as follows: Where: is the apparent permeability of the core, in units of ; is the matrix porosity; is the total number of pores; is the rock tortuosity; is the apparent permeability of capillaries of different sizes, in units of ; is the distribution frequency of capillaries of different scales; Rock tortuosity The calculation formula is as follows: in: is the rock tortuosity fitting parameter, and its value is 0.

77.

8. Shale gas reservoir apparent permeability prediction system, characterized in that: include: Model building unit: used to build the apparent permeability model of shale gas reservoirs; A data set generating unit: used for generating a data set by using the shale gas reservoir apparent permeability model; A first training unit: used to train a deep learning model using the data set to obtain a trained deep learning model; The second training unit is used to use the trained deep learning model as the source model, adopt the transfer learning method, use the data of different types of shale reservoirs to train the migrated deep learning model, and adjust the parameters of the migrated deep learning model to adapt to the characteristics of the target field. After the training is completed, a trained deep learning model for predicting the apparent permeability of different types of shale reservoirs is obtained; Prediction unit: used to predict the apparent permeability of shale gas reservoirs using the trained deep learning model of shale reservoir apparent permeability.

9. An electronic device, characterized in that: include: one or more processors; a storage device having one or more programs stored thereon; When the one or more programs are executed by the one or more processors, the one or more processors implement the method for predicting the apparent permeability of shale gas reservoirs as described in any one of claims 1 to 7.

10. A storage medium, characterized in that: A computer program is stored thereon, wherein when the computer program is executed by a processor, the method for predicting the apparent permeability of a shale gas reservoir as claimed in any one of claims 1 to 7 is implemented.

Images