Many heavy oil fields have recently seen exponentially higher volumes of data being made available from omnipresent connectivity. Existing data platforms have traditionally focused on solving the problem of data storage and access. The more complex problem of true knowledge discovery and systematic value creation from all this massive amount of data is less frequently addressed. There is a new, novel workflow that tackles the problem of building intelligent data analytics in heavy oil fields. This multidisciplinary solution builds from Augmented AI, i.e., a combination of artificial intelligence/machine learning and domain knowledge or expertise.
Our solution has been proven effective in a Latin American heavy oil field and is presented here. We built an integrated view of the reservoir through a series of smart metrics and KPIs. This was accomplished by integrating expert-based knowledge with the analysis of geological data, reservoir behavior and production and operational performance. Our analytics-based solutions were designed from the perspective of reservoir management, and consequently, they could assimilate production and cost/economic analysis with geological information (e.g., well logs and/or existing geological models) and reservoir performance (metrics for pressure, voidage, fractional flow, reservoir contact, etc.). From here, KROs (key recovery obstacles) were identified for this heavy oil field, and robust field development opportunities (i.e., behind-pipe opportunities and/or new well targets) were methodically proposed based on an innovative saturation mapping approach. Our solution exploited existing cloud economics at scale, sustained advances in hardware capabilities (including GPUs running machine learning workloads) and iterative improvements in algorithmic learning techniques.
Ultimately, we will demonstrate a smart analytics-based solution that can be systematically implemented in heavy oil reservoirs for robust diagnostics and comprehensive field development opportunity identification. The proposed methodology intelligently combines both field experience with data-driven and machine learning methods.