Avoiding Misleading Performance Predictions for a Reservoir Systemwith Limited Production Data by the Application of ExperimentalDesign Technology

摘要

The general aim of reservoir modelling is to provide atool capable of generating reliable ranges ofperformance predictions on which to makedevelopment and investment decisions in a way thatmanages the recognised uncertainties associated withthe development. The standard approach is togenerate a range of static and dynamic models andparameters that capture the uncertainties, based onavailable sub-surface and surface information,correlations or engineering judgement.Experimental design (ED) techniques are beingincreasing adopted as a method to consistentlycombine the uncertainty parameters and to construct arange of reservoir models. The standard ED approachis to adopt global changes to the range of inputparameters and seek combinations that best reproducethe historical data. Calibrating the reservoir models toavailable historical performance data is a vital step inproviding confidence on the predicted results.However, in situations with limited performance data,great care is needed to understand the implications ofmodel calibration so as to avoid misleadingprediction results.