Is Experimenter facing a problem in v2016?

[lucas_antonio_risso]

[ FlexSim 16.0.1 ]

I am running the Experimenter with 100 replications for 48 performance indicators, and the results are not being showed after the running, like showed in the image below. However, if I reduce the number of KPI's or the number of replications, it runs normally.

Has anyone already faced a similar situation? By any chance, is there a limit for the settings?

Please, I appreciate if you could check my model. experimenter-blank-results.fsm

Thank you for your help.

[marco_baccalaro]

Maybe you could have some error on KPIs code like a division by zero.

[lucas_antonio_risso]

I don't think so, because they are all standard measures, and when I set 25 replications, it runs perfectly.

[michael_machado]

Lucas, try with a general or demo model and see if you will still face the same problem. If so, share the demo model with us here. Thanks,

[marco_baccalaro]

Right. Another idea: it happened to me that the system console was growing up too much and so I had problems on big experimenters. Maybe take a look at it.

[lucas_antonio_risso]

This is probably the case... How can I check that? What should I do if the experimenter is very big? Thanks!

[JordanLJohnson]

We have seen a similar issue that happens when you save the model with the experimenter window open. Does that sound like it applies to you?

[lucas_antonio_risso]

Yes, it can be the case. Maybe I have saved the model with the window open. So how do I fix this? What did you do? Thanks.

[JordanLJohnson]

You can try closing and reopening the results window, along with the experimenter window. The results from each experiment are stored in the tree, even after you hit the "Reset Experiment" button. If you run a new experiment, those results will be replaced with results from the current experiment.

That means that you should be able to open the results window (using the "View Results" button) any time after the experiment has been run. If that still doesn't work, consider inspecting the experimenter tree (MODEL:/Tools/Experimenter), looking for anything out of the ordinary. Without a model, though, it is difficult to understand the real issue.

[lucas_antonio_risso]

Hello everyone, this model is presenting the problem that I described above . Thank you for any help. experimenter-blank-results.fsm

[matt_long]

Looks like it's being finicky, though I'm not sure why. The first time I ran the experiment the results didn't display. The next time I did it seemed to work just fine. Try running the experiment and hit the View Results while the experiment is still running (and only a few replications have finished). Seems like when I did that and results displayed, the continued displaying even after I hit 100 replications.

[michael_machado]

Please, @phil.bobo May you look into @Lucas Antonio Risso's model to see if that is a problem in the Experimenter? We've never saw something like that here. Thanks,

[philboboADSK · Accepted solution]

Replication 35 of KPI "Products lead time - LT - CartonPACK (8) Average" has an invalid return value.

Replication 92 of KPI "Products lead time - LT - LILA (7) Average" also has an invalid value.

[lucas_antonio_risso]

Thanks @michael.machado, and @phil.bobo... Phil, can you tell me how to fix the mistake that is causing the return of an invalid value?

[philboboADSK]

I'll look into it further and let you know.

[lucas_antonio_risso]

Thank you so much... The idea for this KPI is to measure the average staytime of all the flowitems that pass by the correspondent queue, but "excluding" the staytime of the last ones that didn't leave the queue by the end of the run time... So, I used a tracked variable. If there is another way to accomplish it that is easier than fixing this logic, I would appreciate to learn. Once again thank you so much.

[philboboADSK]

The problem is that nothing has left the queue so the tracked variable has never been set.

On the Advanced tab of the Experimenter, you can set it to repeat the random streams of a certain replication. If you set that to 35 and then reset and run your model, you will see that nothing has exited the LT Eight queue and so the LT (8) tracked variable has never been set.

You have the same problem on LT Seven in replication 92.

[lucas_antonio_risso]

If nothing leave the queue, the KPI should assume the value 0, how can I set this option? Would an "if" condition on the Trigger work?

Thank you for showing this gap.

[philboboADSK]

Set the tracked variables to 0 in a reset trigger.

[lucas_antonio_risso]

I set the tracked variables to 0 on a reset trigger, but it is still not working.

Maybe it is needed a way to attend to the possibility of the item not leave the queue.