Real-Time Steam Allocation Optimization

Abstract

Traditionally, reservoir model-based open-loop optimization is used to allocate an amount of steam to each injector well. In Steam-Assisted Gravity Drainage (SAGD) recovery, the optimal real-time steam allocation from a shared steam generator to physically connected multi-pads can significantly improve long-term performance goals. However, in real-time optimization (RTO), general-purpose optimization algorithms decide based on short-term responses, unlike long-term optimization processes. Using economic Key Performance Indicators (KPI) such as Net Present Value (NPV) in a single objective, the RTO determines the smallest amount of steam allocation that results in the highest economic returns. Injecting a small amount of steam reduces steam chamber heat loss, growth, and long-term ultimate bitumen recovery. Furthermore, when the oil price is volatile, maximizing steam allocation and non-condensable gas (NCG) at the wind-down stage is essential to ensuring a profit while reducing risk. This research addresses the SAGD RTO workflow limitations of handling oil price volatility and balancing steam chamber development and economics to achieve long-term goals. An adaptive data-driven predictive model developed based on typical Athabasca oil reservoir properties is employed for real-time short-term forecasting of the KPI, reducing the computational cost. A modified version of Modigliani's risk-adjusted performance is proposed and integrated into the workflow as a tradeoff selector of expected returns and risk when handling oil price volatility. Additionally, the workflow is tested on multi-pad steam allocation using an Alternating Direction Method of Multipliers (ADMM) for single-, multi- and many-objective optimization problems. Finally, an alternating set of RTO objectives is proposed to ensure that both short- and long-term KPIs are achieved. The performance of the RTO workflow introduced is tested on single and multi-pad field scale SAGD first principle models. In addition, the cases are designed to mimic SAGD operations steam availability, wind-down, single, multi, and many objective RTO. The impact of the developed workflow is the improved short-term strategies, improved long-term economics, and reduced carbon footprints