Data Physics reservoir modeling and optimization was described in detail in a prior paper (SPE-185507) and can be conceptualized as a physics-based model augmented by machine learning. In brief, the production, injection, temperature, steam quality, completion and other engineering data from an active steamflood are continuously assimilated into the Data Physics model using an Ensemble Kalman Filter (EnKF), which is then used to optimize steam injection rates to maximize/minimize multiple objectives such as net present value (NPV), injection cost etc. using large scale evolutionary optimization algorithms. The solutions are low-order and continuous scale, rather than discretized, therefore modeling, forecasting and optimization are significantly faster than traditional simulation.
The goal of steamflood modeling and optimization is to determine the optimal spatial and temporal distribution of steam injection that will maximize future recovery and/or field economics. Accurately modeling thermodynamic and fluid flow mechanisms in the wellbore, reservoir layers, and overburden can be prohibitively resource-intensive for operators who instead often default to simple decline curve analysis and operational rules of thumb. Data Physics allows operators to leverage readily-available field data to infer reservoir dynamics from first principles.
This paper updates the case study from the previous paper and presents the results of actual implementation of an optimized steam injection plan based on the Data Physics framework. The case study is from a shallow, heavy oil field in the San Joaquin Basin of California, and demonstrates the practical application of Data Physics modeling and the ability to explore future injection plans. The model of the field was fit to historical data in June 2017, after which an optimization was performed and a forward-looking production forecast was established associated with a target plan chosen by the operator. This plan was then implemented in the field over the last year. This paper provides a comparison between the field implementation and the model prediction, which allows for model validation and highlights opportunities for further improvement.
For completeness, this paper includes a summary of the modeling and optimization problem and results from the previous paper.