Almost any product around us involves some type of assembly. For manual assembly a major goal is to obtain a motion as ergonomic as possible to prevent harm to the operator. Using simulations, it is possible to estimate the comfort of an assembly prior to production. By optimizing the comfort, the risks of occupational injuries can be lowered. It may also save both time and money spent on development. Due to the characteristics of simulation based functions, classical optimization methods cannot be applied and methods involving many function evaluations will be ineffective. Previous work on simulation based optimization shows that most methods are problem specific and not applicable to other problems.
The aim of this thesis is to develop a method that determines which grips to use and when to use them in order to cause as little discomfort as possible for a person carrying out a given task. An optimization model is created in which the objective function is simulation based. Thus, a method requiring few evaluations is desired. To this end, the problem is modelled as a graph in which the edge values are updated using information from previous simulations. A shortest path algorithm is used on the graph to determine which actions to include in the next simulation. The developed method is tested on three test cases, together with three existing black box optimization methods. In terms of evaluations needed, the method developed for this thesis outperforms the black box methods for all cases tested. It also found the best solution in most cases. Even though the developed model still needs further testing, it solves the problem in a reasonable amount of time.