Development of Surrogate models for CSI probabilistic production forecast of a heavy oil field

The aim of this work is to present the development of a Surrogate Reservoir Model capable of accurately predicting cumulative oil production of a heavy oil field in Mexico considering variables with uncertainty. Five numerical models of wells were considered for the numerical model of the field. Four input uncertain variables (the dead oil viscosity with temperature, the reservoir pressure, the reservoir permeability and oil sand thickness hydraulically connected to the well) were selected as the ones with more impact on the initial hot oil production rate according to an analytical production prediction model. The central composite experimental design technique was selected to capture the maximum amount of information from the model response with a minimum number of reservoir models simulations. Twenty five runs were built to be run with the STARS simulator for each well type on the reservoir model. The results show that Surrogate Reservoir Models are an ideal tool to perform real-time probabilistic production forecasting of the reservoir.