I have a business process where the main process needs to split out into a number of exactly alike subprocesses, and then join back up again into the main process once each of those subprocesses have finished running their course. The problem is that the number of sub-processes that need to be spawned is not consistent, and is actually input by the end user at runtime. I have no idea how to model this behaviour.
An analogy is the process "Publish a book", where the sub-processes in question would be "Edit a chapter". Each chapter will go through the same steps as a sub-process, but the number of chapters (and hence the number of sub-processes required) is only known once the "Publish a book" main process is in course.
you have to write a custom node for this (e.g. a custom fork). There have been some examples published, either on the forum or on the wiki. Please see if you can find and use them