An approach based on simulation optimization and AHP to support collaborative design: With an application to supply chains

In certain design problems, the solution can have collective implications that are experienced by a number of different people with different responsibilities — a team of decision-makers. In such cases, the design problem should be addressed in a collective manner, so that everyone's considerations are taken into account. Unfortunately, even though there is a vast body of literature on simulation optimization, which is widely used to solve the design problems encountered in practice, the existing research generally concentrates on providing a single solution that is optimized according to one or more performance measures. In this paper, we consider the problem of determining the values of several decision variables of a design problem where several decision-makers are involved, who have different preferences for the final solution. The different designers' considerations may not be all known in advance or may not be included in the simulation model, but can only be examined once a candidate solution is proposed. To cope with such difficulties, we propose a two-stage approach. It is first necessary to find a set of different enough designs that can be considered efficient in terms of performance. The solutions can afterwards be passed on to the decision-makers and the most appropriate one can be decided on according to their preferences. We use the crowding clustering genetic algorithm (CCGA) to solve the first sub-problem, where the performances of the candidate designs are evaluated using simulation. We address the second sub-problem with a multiplicative variant of the popular analytic hierarchy process (AHP), which does not suffer from the dependence on irrelevant alternatives as the original version. We illustrate the benefits of the proposed two-stage approach on a supply chain design problem.