Cancel
Showing results for 
Search instead for 
Did you mean: 

Problem with Distributed Simulations in Experiment Manager

Experimenter
Experimenter

Hello everybody,

 

i have a problem with dividing the experiment into two observations and running them simaltaneously with ExperimentManager. It does not write the results in tables for each of the observations, even though the simulation does not show any error. Without Distributed Simulations being enabled the simulation works fine and the results are written inside the tables.

4 REPLIES

Re: Problem with Distributed Simulations in Experiment Manager

Siemens Phenom Siemens Phenom
Siemens Phenom

You talk about observations but most likely you wantetd to mention parallel processes. At least this is what your screenshot shows. In case the number of observation s is set to two too, the ExperimentManager will not be able to calculate the statistical values for your experiments. It is required to have at least three observations per experiment in order to do so.Settings.png

 

Regards


Regards

Always consider using the AttributeExplorer!

Re: Problem with Distributed Simulations in Experiment Manager

Siemens Phenom Siemens Phenom
Siemens Phenom

 

Hi Artemiy,

 

what do you mean by write the results in a table?

Probably you have a tablefile as an output value, how I did it in the example model for Plant Simulation 12.0.

The comparison of the results of the distributed and non-distributed simulation show equal results.

 

Regards,

Peter

Re: Problem with Distributed Simulations in Experiment Manager

Experimenter
Experimenter

Hello @Eventdebugger and @Peter,

 

thank you for your replies! I do not really need statistical comparison, but rather the observations' (in my version - "Beobachtungen" in German) results written in separate tables (see screenshot 'Detailed Results'). It works when I do not apply Distributed Simulations, and for some reason it does not when I do.

I set the amount of parallel processes to 2, the same amount as the amount of observations. As far as I know the ExperimentManager is supposed to divide the calculation of each observation to separate process, or am I wrong here?

 

By the way, @Peter the version of Plant Simulation in my Institute is only 10.1 and, unfortunately, your example model does not work on my workstation. Is there an option for backward compatibility?

 

 

12.PNGSettings13.PNGDetailed Results

 

 

Cheers, 

Artemiy

Re: Problem with Distributed Simulations in Experiment Manager

Siemens Phenom Siemens Phenom
Siemens Phenom

Hi Artemiy,

 

the attached model for Plant Simulation 10.1 also shows that the first and the second observation of the single experiment are equal for a non-distributed and distributed study.

It seems to be that the output tablefile ist not filled at the end of the simulation.

 

Regards,

Peter