It is required to decide an objective to start the process of optimisation. Optimizer quantifies the objective and writes an objective function, which is a common function in the WITNESS MODEL. The possible values or value path are introduced by selecting the command Optimize [on the WITNES Model menu] and choosing the variable factors. It is also possible to introduce restrictions ­i.e. relationships among factors-. Other conditions of the experimentation such as duration of each simulation and number of replications have to be established.

Optimizer makes it possible to establish the number of replications or passing with the same combination of factors but different random seeds; this way it analyses the variability: The more variable the model, (less steady or strong) the more replications would be necessary.

Finally, it is possible to select the method that is going to perform the optimisation process. There are three options:

• experimenting every combination of factors (This is useful for a reduced number of options
• random experimentation (this estimates how the solutions may vary)
• more intelligent and efficient methods such as the above cited heuristic, whose nature we describe below

The power and technique are implemented in the Optimizer heuristic in such a way that it has been proved that it works very well with a quite a low cycle criterion, being often found the optimum after exploring less than 1% of the possible combinations.

if you want more information about this product you can visit the product´s page in Lanner site: