Automated multivariate modeling of field development using advanced programming features of dynamic simulator

Abstract

Introduction. Modern business is facing new challenges and increasing requirements for the efficiency of field development management, including the solution of complex, integrated and massive problems under uncertainty of external conditions. The evolution of information technology (IT) is opening up new opportunities for scaling high-performance computing and implementing digital solutions, which is particularly important for multivariate field development prediction.
Aim. Improvement of methods for solving multivariate problems on the basis of changing approaches to formulating the model description and managing the computation workflow using the developed dynamic simulator. The main tasks are to search for approaches to effectively solve multivariate problems, to implement these approaches, and to verify their performance and prospectivity.
Methods and approaches. To automate the solution of multivariate problems based on the author’s simulator, the integration of the Lua scripting language into the simulator kernel, the creation of an alternative input data format and the extraction of the internal computational workflow into the script space are proposed. In order to maintain high performance of calculations, the optimal division of functionality between the interpreted and compiled parts of the simulator is organized.
Results. A hybrid simulator with an integrated scripting environment was developed; its efficiency was demonstrated by using an alternative data input format, extending the functionality to model complex reservoirs, creating fully “random” models, and describing a set of stochastic models in a single input file. It has been observed that in auto-history-matching and optimization problems, there is a simplification in the construction of multi-realization scenarios, both for external launcher programs and for the implementation of digital development strategies.
Conclusion. Successful integration of the high-level programming language with the author’s simulator demonstrated a new level of simulator control, creation of flexible and dynamic input data format, automation of multivariate task preparation, possibility of creating digital strategies for field development.