Two-phase Inflow Performance Relationship Prediction Using Two Artificial Intelligence Techniques: Multi-layer Perceptron Versus Genetic Programming

Abstract

A genetic programming model has been compared with multi-layer perceptron (MLP) and empirical correlations to predict the inflow performance of vertical oil wells experiencing two-phase flow. The genetic programming under discussion in this work relies on tree-like building blocks, and thus supports process modeling with varying structure. The necessary training data have been obtained from 16 different simulated reservoir models, covering a wide range of fluid properties and relative permeabilities. The results show that the fitted genetic programming model gives the smallest error for unseen data, when compared with MLP and empirical correlations.