Preconditioning Markov Chain Monte Carlo Method for Geomechanical Subsidence using multiscale method and machine learning technique

TL;DR

This paper introduces a two-stage MCMC method for geomechanical subsidence that employs multiscale modeling and machine learning for efficient preconditioning, significantly reducing computational costs in stochastic poroelasticity problems.

Contribution

It develops a novel two-stage MCMC approach with multiscale and neural network preconditioning techniques for faster sampling in stochastic geomechanical simulations.

Findings

Multiscale basis functions effectively reduce model complexity.

Neural network preconditioning accelerates sampling significantly.

Numerical results demonstrate efficiency in 2D and 3D cases.

Abstract

In this paper, we consider the numerical solution of the poroelasticity problem with stochastic properties. We present a Two-stage Markov Chain Monte Carlo method for geomechanical subsidence. In this work, we study two techniques of preconditioning: (MS) multiscale method for model order reduction and (ML) machine learning technique. The purpose of preconditioning is the fast sampling, where a new proposal is first testes by a cheap multiscale solver or using fast prediction of the neural network and the full fine grid computations will be conducted only if the proposal passes the first step. To construct a reduced order model, we use the Generalized Multiscale Finite Element Method and present construction of the multiscale basis functions for pressure and displacements in stochastic fields. In order to construct a machine learning based preconditioning, we generate a dataset using a multiscale solver and use it to train neural networks. The Karhunen-Loeve expansion is used to represent the realization of the stochastic field. Numerical results are presented for two- and three-dimensional model examples.