Not a Problem

Simulation does not obey gate logic

When trying to run a simulation with parallel gate. The simulation treats

the gate as an exclusive gate only allowing one path to be active after the

gate. I have attached an example. When running tokens should be passed to both

tasks after the parallel gate but this does not happen.

This is a big issue for a simulation using two work streams that interact

within the same business process.

Is there something I am doing wrong in the set-up or when defining the parameters

for the simulation.

Files: test.zip
Best Answer
photo

Dear Stephen,

Thank you for your feedback.

We will review this subject and keep you posted.

It would initially seem that the issue lies within the use of Sub-processes directly after the parallel gateway (you may avoid this and verify by using tasks instead).

Best regards,

photo
0

Dear Stephen,

Thank you for your feedback.

We will review this subject and keep you posted.

It would initially seem that the issue lies within the use of Sub-processes directly after the parallel gateway (you may avoid this and verify by using tasks instead).

Best regards,

photo
0

I have tried this workaround but this still does not work. I have also used an inclusive gate and these also do not behave as expected.The fault is within the logic of the gate simulation. the system only produces a percentage split between every path out of the gate.for an exclusive gate the current logic works because the number of tokens passing out of the gate must match those coming in; however for a parallel or inclusive gate this is not the case. In a parallel gate for every token arriving at the gate a token must be pass to every path leaving the gate.In an inclusive gate for every path out a separate percentage is needed to identify the likelihood of it passing a token. I like the idea of being able to simulate a flow, however without being able to use all gate types this is a useless function for me

photo
0

Dear Stephen,

Regarding the posted model, we have reviewed the configuration thoroughly and have seen that you will need to consider the following for the simulation to throw accurate results:

1. Change the terminate end event to a none end event.

lb49

Using the terminate type does not suit for your test. This is so because if you use that one, it will imply that:

Even though tokens are split correctly, once the token completing the "other stuff" task reaches the end, it will immediately kill any other live tokens for that same instance.

And since the "other stuff" task is processed much faster than the "some stuff" Sub-Process (as defined by you in their corresponding wait + processing time), the "secondary token" is recorded as having started a task but it actually never completes it.

2. Specify your available resources accordingly.

Please note that if you define that the "some stuff" Sub-process requires 2 of your Call centre rep resources, then you should have a minimum of 2 of those available in your organization.

To include this definition, use the Resources option available directly for your simulation scenario:

se09

Note too that you should also have at least 1 HR rep to supply the "other stuff" required resource.

Best regards,