A critical review on intelligent optimization algorithms and surrogate models for conventional and unconventional reservoir production optimization

Aiming to find the most suitable development schemes of conventional and unconventional reservoirs for maximum energy supply or economic benefits, reservoir production optimization is one of the most essential challenges in closed-loop reservoir management. With the developments of artificial intelligence technologies during the past decades, both intelligent optimization algorithms and surrogate models have been adopted to solve reservoir production optimization problems for improved efficiency and/or accuracy in the final optimization results. In this paper, a critical review of intelligent optimization algorithms and surrogate models applied to production optimization problems in conventional and unconventional reservoirs is conducted. It covers a few different topics within the target research area, ranging from the basic elements (optimization variables, objective function and constraints) that constitute a reservoir production optimization problem, to various intelligent optimization algorithms developed from different perspectives and for different types of optimization problems (e.g., with single or multiple objective functions), and intelligent surrogate models that are built based on different artificial intelligence technologies and for different application purposes. The particular issues of production optimization in unconventional reservoirs are highlighted, and future challenges and prospects within the area of reservoir production optimization are also discussed. It is our hope that this critical review may help attract more attention to intelligent optimization algorithms and surrogate models applied to production optimization problems in conventional and unconventional reservoirs, and promote research and development activities within this area in the future.