Well pattern optimization using NEWUOA algorithm

• We propose a new construction form and computing solution for well pattern.
• We adopt NEWOUA algorithm to optimize regular well pattern.
• Constrained optimization is used to limit control variables.
• Test cases prove the method correct and applicable for real oil field.

Well pattern design plays a critical role during oil-field development. Inappropriate design would result in huge loss considering that the cost of a well may be millions of dollars. In this paper, our objective is aiming at solving the problem of optimization of regular well pattern for large-scale reservoir development. A new construction form and computing solution is proposed for well pattern. Due to varieties of regular well patterns, it is difficult to give a universal description for the corresponding structures. Hence, based on the definition of well pattern by Onwunalu, 6 variables are developed to describe the well pattern units. These variables are able to realize the rotating, scaling, shifting and shearing regardless of row well pattern, five-spot well pattern, seven-spot pattern or nine-spot pattern. Furthermore, optimal well pattern depends on not only the unit form but also the production control which is reflected in a new variable-total injection-production rate. Therefore, the total number of variables is up to 7 but not too many. And because they are continuous, NEWUOA algorithm can be used for the optimization of regular well pattern. Meanwhile, the constrained problems, such as the boundary restraints of variables and cumulative injection–production rates, can be better considered. Finally, detailed well pattern optimization examples are presented indicating that the well pattern optimization method is quite effective.