Is this true? Running a semantic check (structure rules for process models) on an EPC produces this rule violation. The model has a loop (a rule branches back to a prior rule).
pseudocode:
Submit Article
Review Article
1a if Article needs revision
1b Revise Article
1c (back to) Submit Article
How then should one express such a typical scenario?
This is true, you cannot get a direct cycle on EPC model. How we usually solve this. We use cycled events – just look at the picture below to understand this approach with your case.
By this approach if you need to use cycle you should make occurrence copy of cycle’s enter event and paste it to the place where cycle ends. After that you should change color for these events – we use yellow (as a standard), and increase border size for b/w printing.
There are some benefits of this approach:
- you can use cycles and all scripts, additional modules like simulation or business optimizer will work.
- it is easy to create and read model – you don’t overload model by additional connections
- you can use ‘cycle in cycle’ and your model will be transparent and obvious.
- you can use some analysis methods like ABC with cycles etc.
Some notes for this approach.
- you should use occurrence copy of events – then all script could evaluate them correctly
- you should use the same color for events. Sometimes it is difficult with ARIS’s color palette so we create user’s symbol for these events.
- as a rule cycles are modeled inside one model – it wouldn’t be correct to use cycles between two connected procedures.
Hi Alexander,
That was so interesting....thanks.
Many thanks for your helpful and insightful approach.
Hi Alexander,
That's very useful. I have used this loop cycles many times in my EPCs. However I haven't looped back to the start event but somewhere in the middle. Symantec check did not throw the error.
So should I use the loop cycles or not?
Also I want to know how did you paste the model in the comment box. I wanted but I am unable to do so. Help!
Thanks,
Deepa Tambe
Hi Deepa!
First moment, you can use cycles at any part of the model. As you can see in example below, cycle is placed in the middle of the model.
About semantic checks. You should check your semantic check profiles, maybe some rules were not checked. In any case, if you show direct connection for cycle then there will be some problems with ARIS tools and scripts.
And about picture insertion – you should choose third command from the right and then upload picture.
.png)
I think it should be considered that this method of modeling violated one of the semantic check rules available in the ARIS repository.
Not checking your models with that rule is not necessarily wrong.
Our semantic checks would allow the loop to be modeled and our reports and queries work right, because they are built according to our conventions which are supported by our selection of rule profiles in our semantic checks.
Personally I like to see the loops as they exemplify 'the hidden factory' of rework.
rgds
donald
I like to support what Donald said. Notations are tools you should leverage so that they generate most value for you. If that means violating the official notation rules, this is ok as long as those violations make sense for you and your colleagues. Of course, there is always a disadvantage by violating the notation like someone familiar with the notation but not with your customisations might not be able to correctly read your models. But if you can deal with that point e.g. through training your co-workers, it is perfectly ok to customize a notation.
Rather than "violating the official notation rules", I consider that ARIS supports different methodologies.
There are two essential reason do not use direct loop. First is that if your model is complex and it has a lot of loops then you should create many crossings of connections or you should draw out connections of loops. In this case it will be more difficult to understand such model, not to mention about printing. Second, as I’ve mentioned, you cannot use tools like simulation with cycles.
Of course, if you use this approach (or another) it should be described in corporate Modeling Guide, it is clear.
I think the problem is that it would be difficult to find some clear information about using of cycles in ARIS Method manual or in materials of IDS trainings. So it is too difficult to say about "violating the official notation rules" because there is nothing to violate :)
P.S. This approach has been working for many years and it is all Ok at many companies. Besides, this approach is very similar to BPMN’s signal events.
Regards