Application of Assisted History Matching Workflow to Shale Gas Well Using EDFM and Neural Network-Markov Chain Monte Carlo Algorithm

摘要

Production data which is always available at no additional cost can give an invaluable information of fracture geometry and reservoir properties. However, in unconventional reservoirs, it is insufficient to characterize hydraulic fractures geometry and reservoir properties by only one realization because it cannot capture the non-uniqueness of history matching problem and subsurface uncertainties. Therefore, the objective of this study is to obtain multiple realizations in shale reservoirs by adopting Assisted History Matching (AHM). We used Neural Network-Markov Chain Monte Carlo (NN-MCMC) algorithm in the proposed AHM workflow for shale reservoirs. The reason is that MCMC, one of AHM in the Bayesian statistics, has benefits of quantifying uncertainty without bias or being trapped in any local minima. Also, using MCMC with neural network (NN) as a proxy model unlocks the limitation of an infeasible number of simulation runs required by a traditional MCMC algorithm. The proposed AHM workflow also utilized the benefits of Embedded Discrete Fracture Model (EDFM) to model fractures with a higher computational efficiency than a traditional local grid refinement (LGR) method and more accuracy than the continuum approach. We applied the proposed AHM workflow to an actual shale-gas well. We performed history matching on two cases including hydraulic fractures only and hydraulic fractures with natural fractures. The uncertain parameters for history matching consist of fracture geometry, fracture conductivity, matrix permeability, matrix and fracture water saturation, and relative permeability curves. For the case with natural fractures, we included number, length and conductivity of natural fractures as the additional uncertain parameters.