Just a thought, as most of you will know, the BPMN model done for implementation (private processes) contain many more objects and detail and does not look the same as the BPMN model the business signed off. Now the question, do you maintain one model and potentially confuse the business or do you keep a logical design and a physical design with common objects where applicable?
In my opinion 2 models would be better and I’m playing with the idea that from your parent process object i.e. value chain object, the logical design would be represented by an assigned BPMN collaboration diagram (private & public) and the physical design would be represented by assigning a BPMN process diagram (private)? Common objects i.e. tasks would appear on both, but if the physical model is enhanced by including extra service objects, they won’t appear on the logical model. This way the logical & physical can be compared and what should be the same on the 2 diagrams should be the:
User tasks – representing consumer processes
Sub processes – representing long running processes
Transactions – representing short running processes
What can differ are automated activities (service or script tasks) that belong to a transaction (short running) or stand alone service or script tasks.
Hope this makes sense?
A little bit offtopic - this is the way the "Model to Execution" (the ARIS <> webMethods integration) is implemented, except that we use EPCs for the businesss view and a logical BPMN model for the technical view.
The EPC is simply much more richer in the usage of different object types, that can be reused in other use cases (e.g. EA) without having to do workarounds. The business model get's transformed into one application of BPMN (that's called "logical BPMN") that meets the wM requirements and pushed into wM Designer to add the execution details. If there is the need to change something due to technical reasons, the logical BPMN model can be pushed back into ARIS. This and the needed semantic checks will be automated by the ARIS Process Governance engine.
GUI-wise you see a new "webMethods" menu in ARIS and from there you can start semantic checks without using the regular "Evaluate > Semantic Check" commands, and it also contains a menu entry that allows you to jump from the EPC to the linked logical BPMN model and back.
I don't want to start a new version of the "perfect flamewar" (see Sebastian's article about this elsewhere), but we see that EPC is easier to understand for our business users and by using different model types and their look-and-feel you can avoid the confusion you describe. As a nice side effect, you are still able to use the regular SAP synchronization capability if you have a mixed environment and want to use ARIS for SAP implementations.