Question on Sending by Percentage

Hi

I have a model that has the flowitem that send to two identical processors. The distribution is 60% to the first one and 40% to the other.

I quite surprise that the result obtained from this case is different from switching the ports. That is 40% to the first one and 60% to the second one. I am expecting that the result is the same.

Can anyone explain why?

Regards

David

Sounds like you have something else going on in your model.

Attached is a simple model that shows this pick option working correctly.

Please post a simple example model demonstrating your problem, and we can then figure out why it is behaving that way.

Phil

Thanks for your reply. I have attached my model.

I have run the model for a bit longer time. The throughput are the same, but the waiting time at the queue is different. But what account for the different?

Once again thanks

David

You are passing 60 and -59 as your percentages. (or 40 and -39 for the other scenario.)

You should use 100-getlabelnum(current,"percentage") instead of 1-getlabelnum(current,"percentage")

Note that you will still have different staytimes between these two scenarios. The comparison in the code is checking if the random number is <= the percentage. So you are checking against 40 for one scenario and 60 for the other scenario. So in the 60 scenario, it is the 40% that would go to that port in the other scenario, plus another 20% that will now also go to that port; not the 60% that weren't sent to that port in the first scenario. At different times, you will be sending to ports where there is already an item in the processor. Hence, the difference between scenarios.

Here is a picture that might help explain the difference:



If you want identical staytimes in these scenarios, then you need to base the port on the scenario, not the comparison percentages. See the attached model.