Effective representation of complex three-dimensional simulation results for real-time operations

Patent metadata

Field	Value
Publication number	US-12061980-B2
Application number	US-201716642452-A
Country	US
Kind code	B2
Filing date	Dec 26, 2017
Priority date	Dec 26, 2017
Publication date	Aug 13, 2024
Grant date	Aug 13, 2024

Abstract

System and methods for training neural network models for real-time flow simulations are provided. Input data is acquired. The input data includes values for a plurality of input parameters associated with a multiphase fluid flow. The multiphase fluid flow is simulated using a complex fluid dynamics (CFD) model, based on the acquired input data. The CFD model represents a three-dimensional (3D) domain for the simulation. An area of interest is selected within the 3D domain represented by the CFD model. A two-dimensional (2D) mesh of the selected area of interest is generated. The 2D mesh represents results of the simulation for the selected area of interest. A neural network is then trained based on the simulation results represented by the generated 2D mesh.

First claim

What is claimed is:

1. A computer-implemented method of training neural network models for real-time flow simulations, the method comprising:
acquiring input data including values for a plurality of input parameters associated with a multiphase fluid flow to be simulated;
simulating the multiphase fluid flow using a complex fluid dynamics (CFD) model, based on the acquired input data, the CFD model representing a three-dimensional (3D) domain for the simulation;
selecting an area of interest within the 3D domain represented by the CFD model;
generating a two-dimensional (2D) mesh of the selected area of interest, the 2D mesh representing results of the simulation for the selected area of interest;
training a neural network based on the simulation results represented by the generated 2D mesh; wherein a size of dimensions of arrays in a training dataset for the neural network is dependent on a size of the 3D domain and the number of nodal points used for the training dataset is smaller than a number of nodal points in the 3D domain; and
simulating multiphase fluid flow in real time during a wellsite operation using neural network trained by the simulation results represented by the generated 2D mesh.

2. The method of claim 1 , wherein the neural network is a deep-learning neural network (DNN).

3. The method of claim 1 , wherein the wellsite operation is a stimulation treatment operation performed over different stages along a path of a wellbore through a reservoir formation, and the 3D domain represents the reservoir formation.

4. The method of claim 3 , wherein generating further comprises:
identifying a 2D planar surface within the 3D domain, the 2D planar surface including simulation data corresponding to the selected area of interest; and
applying the simulation data from the identified 2D planar surface to the 2D mesh of the selected area of interest.

5. The method of claim 4 , wherein the selected area of interest includes a fracture network within an area of the reservoir formation surrounding the wellbore.

6. The method of claim 4 , wherein the simulation data is applied to nodal points of the 2D mesh.

7. A system comprising:
at least one processor; and
a memory coupled to the processor, the memory storing instructions, which, when executed by the processor, cause the processor to perform a plurality of functions, including functions to:
acquire input data including values for a plurality of input parameters associated with a multiphase fluid flow to be simulated;
simulate the multiphase fluid flow using a complex fluid dynamics (CFD) model, based on the acquired input data, the CFD model representing a three-dimensional (3D) domain for the simulation;
select an area of interest within the 3D domain represented by the CFD model;
generate a two-dimensional (2D) mesh of the selected area of interest, the 2D mesh representing results of the simulation for the selected area of interest;
train a neural network based on the simulation results represented by the generated 2D mesh; wherein a size of dimensions of arrays in a training dataset for the neural network is dependent on a size of the 3D domain and the number of nodal points used for the training dataset is smaller than a number of nodal points in the 3D domain; and
simulate multiphase fluid flow in real time during a wellsite operation using the neural network trained by the simulation results represented by the generated 2D mesh.

8. The system of claim 7 , wherein the neural network is a deep-learning neural network (DNN).

9. The system of claim 7 , wherein the wellsite operation is a stimulation treatment operation performed over different stages along a path of a wellbore through a reservoir formation, and the 3D domain represents the reservoir formation.

10. The system of claim 9 , wherein the functions performed by the processor further include functions to:
identify a 2D planar surface within the 3D domain, the 2D planar surface including simulation data corresponding to the selected area of interest; and
apply the simulation data from the 2D planar surface to the 2D mesh of the selected area of interest.

11. The system of claim 10 , wherein the selected area of interest includes a fracture network within an area of the reservoir formation surrounding the wellbore.

12. The system of claim 10 , wherein the simulation data is applied to nodal points of the 2D mesh.

13. A non-transitory computer-readable storage medium having instructions stored therein, which when executed by a processor cause the processor to perform a plurality of functions, including functions to:
acquire input data including values for a plurality of input parameters associated with a multiphase fluid flow to be simulated;
simulate the multiphase fluid flow using a complex fluid dynamics (CFD) model, based on the acquired input data, the CFD model representing a three-dimensional (3D) domain for the simulation;
select an area of interest within the 3D domain represented by the CFD model;
generate a two-dimensional (2D) mesh of the selected area of interest, the 2D mesh representing results of the simulation for the selected area of interest;
train a neural network based on the simulation results represented by the generated 2D mesh;
wherein a size of dimensions of arrays in a training dataset for the neural network is dependent on a size of the 3D domain and the number of nodal points used for the training dataset is smaller than a number of nodal points in the 3D domain; and
simulate multiphase fluid flow in real time during a wellsite operation using neural network trained by the simulation results represented by the generated 2D mesh.

14. The non-transitory computer-readable storage medium of claim 13 , wherein the neural network is a deep-learning neural network (DNN).

15. The non-transitory computer-readable storage medium of claim 13 , wherein the wellsite operation is a stimulation treatment operation performed over different stages along a path of a wellbore through a reservoir formation, and the 3D domain represents the reservoir formation.

16. The non-transitory computer-readable storage medium of claim 15 , wherein the functions performed by the processor further include functions to:
identify a 2D planar surface within the 3D domain, the 2D planar surface including simulation data corresponding to the selected area of interest; and
apply the simulation data from the 2D planar surface to nodal points of the 2D mesh of the selected area of interest.

17. The non-transitory computer-readable storage medium of claim 16 , wherein the selected area of interest includes a fracture network within an area of the reservoir formation surrounding the wellbore.

Assignees

• Landmark Graphics Corp

Inventors

• Filippov Andrey
• Lu Jianxin
• Wesley Avinash
• Rangarajan Keshava P
• Madasu Srinath

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06T17/20

  Finite element generation, e.g. wire-frame surface description, {tesselation} · CPC title

• G06F30/27

  using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model · CPC title

• G06F30/28

  using fluid dynamics, e.g. using Navier-Stokes equations or computational fluid dynamics [CFD] · CPC title