Automated uncertainty quantification of geological model using Bayesian Evidential Learning

David Zhen Yin
Uncertainty quantification is at the heart of decision making during reservoir appraisals. Based on Bayesian Evidential Learning (BEL), this project aims to develop an automated framework for uncertainty quantification of geological model for the reservoir appraisal. Under this framework, when new wells are drilled, multiple components of geological model need to be updated jointly and automatically by means of a sequential decomposition following geological rules. During the updating, we extend the direct forecasting to perform joint model uncertainty reduction using new well observations, thus leading to the automation. This enables updating the geological model uncertainty without conventional model rebuilding, which significantly reduces the time-consumption.