Quantifying Uncertainty in Reservoir Performance Prediction

Monday, January 7, 2002 - 9:30am - 10:30am
Keller 3-180
Michael Christie (Heriot-Watt University)
Predicting the performance of oil reservoirs is inherently uncertain: data constraining the rock and rock-fluid properties is available at only a small number of spatial locations, and other measurements are integrated responses providing limited constraints on model properties. Calibrating a reservoir model to observed data is time consuming, and it is rare for multiple models to be 'history matched'. Uncertainty quantification usually consists of identifying high-side and low-side adjustments to the base case.

This paper will describe a technique for quantifying uncertainty in reservoir performance prediction. The method, known as the Neighbourhood Algorithm, is a stochastic sampling algorithm developed for earthquake seismology. It works by adaptively sampling in parameter space using geometrical properties of Voronoi cells to bias the sampling to regions of good fit to data. The algorithm evaluates the high dimensional integrals needed for quantifying the posterior probability distribution using Markov Chain Monte Carlo run on the misfit surface defined on the Voronoi cells.

We demonstrate the performance of the algorithm on a synthetic case originally developed for use the in the SPE Comparative Solution Project. Reservoir oil and water rates, and average reservoir pressure are computed from the fine grid solution and the reservoir performance data for the first 300 days is used as input. We generated multiple coarse grid reservoir models and assessed the misfit in oil rate and pressure. We then use the Neighbourhood Algorithm to generate multiple models that match observed history data and predict the range of possible reservoir rates out to 2000 days.

The results presented will show both the accuracy of the maximum likelihood model fit to the data and the ability of the method to sample effectively from the posterior distribution.