A Bayesian Optimization Workflow for Integrated Asset Management Under Uncertainty

Abstract

To improve production under existing facility operating constraints, while minimizing emissions and energy consumption, operators today are looking for workflows that integrate surface and subsurface models as a foundational pillar of the digitalization of oilfield operations. In this work, we investigate the investment and operational planning of a multi-field asset development under uncertainty. Our solution is an integrated workflow for scheduling, forecasting, and optimizing the multi-field asset across the entire value chain, from the subsurface reservoirs to delivery to market. Our integrated operation and optimization workflow example consists of one gas condensate and two oil reservoirs models, coupled to a common production surface network model, itself connected to a processing facility model, together with an economic model. Consistent PVT properties have been used throughout the system. The thermo-hydraulic performance of the production network is modelled using a drift flux based mechanistic approach. Operational constraints are honoured on wells, surface network, and processing facility. Our proposed solution is built in a commercial integrated asset management platform that connects the models and orchestrates their execution and optimization. We have successfully integrated the subsurface, gathering systems, and surface facility models. This integration has allowed us to better understand the behaviour of the field and optimize its operation. By combining data and workflows across multiple disciplines, we can make informed decisions more efficiently and more reliably and mitigate risk with confidence. Bayesian optimization offers an alternative solution to traditional ensemble methods by using probabilistic surrogate models that can capture and propagate uncertainties from multiple disciplines without the need for large ensembles. The solution is equipped with unified analytical and diagnostics dashboards for the multidisciplinary technical asset team including management for decision making under uncertainty. This is a truly part of a digital twin of the oilfield with improved asset management under uncertainty in the era of digital transformation.