A hybrid multi-objective genetic algorithm for scheduling heterogeneous workover rigs on onshore oil fields

Abstract

Onshore oil fields are composed of a set of geographically distributed wells that, after some time of operation, might present some malfunction and have their production interrupted. When the oil production of some wells is interrupted, specially-equipped vehicles, also called workover rigs, are deployed to service the wells and guarantee that their activity is restored. Given the limited number of workover rigs and the large number of wells around the oil field, the workover rig problem consists of finding the best scheduling for the workover rigs, so the total production loss of wells is minimized. The scheduling considers some factors such as the production loss rate of each well, the service level required and the planning time horizon in which the schedule must be executed. This research presents a hybrid genetic algorithm to solve the multi-objective workover rig problem with a heterogeneous fleet and a finite time horizon. The hybrid genetic algorithm incorporates a variable neighborhood descent heuristic as a local search procedure to increase the convergence speed of the set of solutions. Both objectives of minimization of the production loss and fleet cost associated with the rent of workover rigs are taken into consideration. The fleet is held variable, so a workover rig depot is included at a strategic position on the oil field to guarantee that new workover rigs, besides the already existent ones on the oil field, might be included in the scheduling when required. The genetic algorithm was tested on a set of practical-sized instances up to 200 wells, 10 workover rigs and 300 as time horizon. Results show a high conflict between the objectives of minimizing the production loss and fleet cost for the workover rig problem, besides important aspects of the solutions obtained by the proposed algorithm.