JT
Primary question...is there a document that shows correct BPMN model constructs to use in conjunction with ARIS simulation? Right now I'm using the on-line help information, which is helpful in some ways but, at least for me, is tedious to go through and have some gaps. I'm trying to provide guidance to the folks who are creating BPMN models that facilitate simulating the models.
Right now many of the BPMN models created do not seem to work well with the simulation tool. For example, the following is an santitized excerpt of a model created by one of our BPMN modelers that we are trying to simulate. This represents a parallel split-join. However, under some conditions, the process is terminated if certain condition occurs in one of the parallel task. In looking at the simulation ouptput for Gateways (cumul), I noticed a negative number (-313) for the count of folders waiting for the "Parallel Join" gateway. This does not make sense.
An alternative representation is the following. Using this representation, there are no negative values. The only difference is where the gateway test occurs. Is Version 2 the better ("correct") way to model this type of construct?
Any guidance will be appreciated. Thank you.
In reply to 06/06/12:18:17 pm by schernega
The parallel join expects the same amount of folders produced by the parallel split in one process instance. But it get's from the middle path because of the XOR split. That's more or less a discriminator pattern. I don't know how this is supported by the simulator but have you tried using a complex gateway for the join instead of AND?
No, it is 1.x. I just created a template with the 2.0 values, so that the users don't have to re-learn the models when the company decides to upgrade to BPMN 2. The reasons to go with 1.x were different ones and the simulation capability was a nice extra. Currently you cannot simulate 2.0.
Just to confirm the behavior of the Simulator when you have the AND split and an AND Join with the terminate end event between, the Simulator destroys the remaining tokens which relate to the token hitting the end event, whether these are waiting at the AND Join or being processed in a parallel task.
So changing the process to put the terminate end event after the AND Join will potentially increase the work done by resources at the tasks prior to the terminate end event which is wasted effort and cost for the business process. Obviuously better to 'destroy' the tokens as early as possible.
We did test the Simulator in a stand alone mode to make sure this case was being handled correctly.
Jim,
I didnt actually use a version of ARIS Business Simulator as issued by SAG, we used the core simulation engine we provide and coded a test model by hand. This is often the easiest way for us to test our software component. Not sure if you have provided a sample model to Ralf through ARIS support, I believe thats the best way to ensure it is properly checked out.
Sorry for the inconvenience,
my direct email is ;
Geoff
hi all,
are there news on this topic?
i get the same behaviour in the ARIS PD&A version (with Business Architect 7.2).
I've a "terminate end event" before a "AND join" gateway and this AND gateway seems like it's not informed by the termination: negative process folder waiting numbers are shown.
is it only a visualization bug? or even the process logic is affected?
i notice that also other counters (process folders on functions) in the same process model (BPMN 1.x) seem like if they are "losing" process folders: the total amount of token isn't correct, so i suppose the problem it's not only a visualization bug
any suggestion?
is there a solution?
Best Regards
Antonio
