Fast robust optimization using bias correction applied to the mean model

• DOI: 10.1007/s10596-020-10017-y
• Publication Date: 2020-11-26T01:02:58Z
• Authors: Lingya Wang; Dean S. Oliver
• Keywords: Robust Optimization; Realization (probability)

Abstract Ensemble methods are remarkably powerful for quantifying geological uncertainty. However, the use of ensemble reservoir models robust optimization (RO) can be computationally demanding. The straightforward computation expected net present value (NPV) requires many expensive simulations. To reduce computational burden without sacrificing accuracy, we a fast and effective approach that only simulation mean model with bias correction factor. Information from distinct controls realizations used to estimate different controls. effectiveness various bias-correction linear or quadratic approximation is illustrated by two applications: flow in one-dimensional drilling-order problem synthetic field model. results show NPV significantly improved estimating factor, RO superior both performed using Taylor series representation uncertainty deterministic single realization. Use bias-corrected account allows application fairly general methods. In this paper, apply nonparametric online learning methodology (learned heuristic search) efficiently computing an optimal near-optimal drilling sequence on REEK Field example. This either optimize complete first few wells at reduced cost limiting search depths.