Quantifying Uncertainty in Reservoir Performance Prediction

EAGE Workshop: Integrating Reservoir Engineering with Geology and Geophysics, 26 May 2002 Summary This paper will describe a technique for generating an ensemble of history matching models using the Neighbourhood Approximation algorithm. Using the ensemble of models generated, we demonstrate how uncertainty in reservoir performance prediction can be quantified by sampling from the posterior distribution. This involves using the neighbourhood approximation algorithm in a Bayesian framework. We validate the technique on the SPE 10 Comparison Solution Project dataset. Fine grid oil rate and average reservoir pressure for 300 days are used in lieu of field data. We generated multiple coarse grid reservoir models and assessed the misfit in oil rate and average field pressure. By running multiple Markov Chain walks on the misfit surface, we are able to quantify the posterior probability of the models in the input ensemble and predict the range of possible fine grid profiles out to 2000 days.