question

The optimizer needs two things: Parameters to vary between the replications and one or more performance measures to indicate how "good" a chosen set of parameter value is.

The performance measure in this case could simply be chosen as the input of the sink (how many items were processed during the replication run).

The parameters have to control the path the items take. In this case, I would add two parameters. The first controls which processor the items go to first (1, 2 or 3) and the second parameter the next one (4, 5 or 6).

There are numerous ways of how to build the model so the parameters decide where the items go. In the attached example I opted to use port-connections between the processors and have the parameters control the output port number in the "Send to Port" option. Thus, both parameters can take values between 1 and 3.

In the experimenter/optimizer window, activate both parameters in the list on the left side of the "Szenarios" tab.

Then, set the objective to maximize the performance measure you created (Throughput).

If you run the optimizer now, the result will tell you that it's best to send the items to the first processor first and then either processor 4 or 6 (as both have a lower or equal process time compared to processor1).

bestroute_fm.fsm