Optimal investment planning for production networks with fixed production profiles

In this paper, we consider an oilfield planning problem with decisions about where and when to invest in wells and facilities to maximize profit. The model, in the form of a mixed-integer linear program, includes an option to expand capacity for existing facilities, annual budget constraints, well closing decisions, and fixed production profiles once wells are opened. While fixed profiles area novel and important feature, they add another set of time-indexed binary variables that makes the problem difficult to solve. To find solutions, we develop a three-phase sequential algorithm that includes (1) ranking, (2) branching, and (3) refinement. Phases 1 and 2 determine which facilities and wells to open, along with well-facility assignments. Phase 3 ensures feasibility with respect to budget constraints and adjusts construction times and facility capacities to increase profit. We first demonstrate how our algorithm navigates the problem's complex features by applying it to a case study parameterized with realistic production profiles. Then, we perform computational experiments on small instances and show that our algorithm generally achieves the same objective function values as CPLEX but in much less time. Lastly, we solve larger instances using our three-phase algorithm and several variations to demonstrate its scalability and to highlight the roles of specific algorithmic components.