Combining Optimisation and Simulation Using Logic Based Benders Decomposition

Operations Research practitioners often want to model complicated functions that are difficult encode in their underlying optimisation framework. A common approach is solve an approximate model, and then use a simulation evaluate the true objective value of one or more solutions. We propose new integrating into itself. The idea run at each incumbent solution master problem. results used guide trajectory itself using logic-based Benders cuts. test on class stochastic resource allocation problems with monotonic performance measures. derive strong novel cuts provably valid for all given form. consider two concrete examples: nursing home shift scheduling problem, airport check counter While previous papers these applications could only approximately realistic instances, we able them exactly within reasonable amount time. Moreover, while those account inherent variance problem by including estimates random variables as parameters, compute sample-average approximations optimality up 100 scenarios.