Accelerated Bayesian inference-based history matching of petroleum reservoirs using polynomial chaos expansions

The forecast for oil production from an reservoir is made with the aid of simulations. model parameters in simulations are uncertain whose values estimated by matching simulation predictions history. Bayesian inference (BI) provides a convenient way estimating mathematical model, starting probable distribution parameter and knowing BI techniques history require Markov chain Monte Carlo (MCMC) sampling methods, which involve large number This limits application petroleum engineering, where each can be computationally expensive. To overcome this limitation, we use polynomial chaos expansions (PCEs), represent uncertainty forecasts due to parameters, construct proxy models predictions. As method, present based on black-oil estimate such as porosity, permeability, exponents relative permeability curves. Solutions these problems show that PCE-based method enables accurate estimation two orders magnitude less compared MCMC method.