For each one of the resulting scenarios, 30 replicates have been performed (Law and Kelton, 2000). The common features for all the scenarios considered in this work are as follow. The production rate is constant and equal to one unit per period, because the processing times do not influence the performance of the OFS, as noted in Brabazon and MacCarthy (2006). Regarding the product search mechanism, when a product type demanded by the customer
is not available in the FGI, we use the backwards search of an available product inside the production line, recommended by Brabazon and MacCarthy (2004b). In order to capture the real system performance, in all cases, a warm up period with a length of 10 times the pipeline length is considered. The run length of simulations is set in order to allow 20,000 orders entering the system. The simulation has been analysed as a non-terminating system. At the beginning of each simulation the pipeline is initialised with products, prior to serving the first client.