A fast algorithm for calculating isothermal phase behavior using machine learning

摘要: Abstract Compositional models are frequently used to describe fluids in petroleum reservoir simulation, particularly for simulations of enhanced oil recovery. While compositional more accurate than black models, they incur a larger computational cost, part, due complex phase-equilibrium calculations and can result longer run times. Here, we develop an algorithm reduce the cost by applying two machine learning techniques: relevance vector machines artificial neural networks. We test on three fluid data sets find speedup over 20% with error 0.01%, 90% maximum 5%. These results suggest that be overall time small impact accuracy.