Integrating tactical and operational berth allocation decisions via simulation-optimization

The berth allocation problem (BAP) arising in maritime container terminals has received great attention in the literature over recent years. It has been largely modeled as an integer mathematical programming formulation to be adopted at a tactical level, where detailed equipment and manpower schedules, as well as real-time operational conditions are not explicitly modeled. In this paper, decision making for the BAP is supported by integrating two separate models into a Simulation-Optimization framework: a mathematical programming model at the tactical level and a simulation model at the operational level. Specifically, the framework uses a beam search heuristics to obtain a weekly plan at the tactical level, followed by a simulated annealing based search process to adjust allocation decisions at the operational level. At this level, randomness in discharge/loading operations is taken into account and modeled by an event-based Monte Carlo simulator. A non-standard ranking and selection procedure is used to compare alternative BAP solutions, within the Simulation-Optimization procedure, in order to reduce the related number of simulation runs required. Numerical experiments performed on real instances show how, under conditions of uncertainty and variability, the tactical solution returned for the BAP requires tuning at the operational level.