Abstract
A comprehensive intelligent decision support system (IDSS) for unconventional field development design is presented in this paper. The proposed IDSS combines data-driven models with physics-based reservoir engineering methods and it is a stack of three AI layers: Predictive, Prescriptive and Cognitive. The predictive unit receives physical reservoir parameters as input and predicts the outcome for thousands of possible designs under different subsurface scenarios by using advanced machine learning and deep learning methods. The prescriptive unit searches between these outcomes and finds the most optimum solutions based on the project goals and risks using optimization techniques. Finally, the cognitive unit tries to understand the utility function of expert decision makers and finds the best solution from the optimum subset. For this paper, we focus on the predictive unit results.
Because of the time it takes to setup reservoir simulations, reservoir engineers can only test tens of well designs amongst thousands of possibilities for a particular project, which results in suboptimal outcome. Using the IDSS technology, we enable analyzing thousands of options in less than a week and consequently the optimum field development design can be achieved, which can decrease average cost per barrel of production by 15%.
This is the first study on a cognition-driven decision support system in the upstream oil and gas industry. Unlike conventional field development, for unconventional oil and gas fields the development process is a high-dimensional decision-making problem. In such problems, a cognition-driven DSS is a necessary tool for mitigating human error. The success of cognitive DSS, especially for a complex problem such as unconventional field development, paves the way towards wider usage of this technology in the oil and gas industry.
Introduction
A Decision Support System (DSS) is an information system used to support organizations and businesses to make informed decisions. If this system uses artificial intelligence (AI) techniques extensively, it is called Artificial Intelligence DSS (AIDSS) or Intelligence DSS (IDSS). Nowadays, DSS and IDSS systems are widely used in healthcare, finance, environment, security among many others. The oil and gas industry also utilizes DSS and IDSS for different purposes and business problems. Korovin and Kalayev (2015) listed some of the DSS systems in the oil and gas industry.
