Swami,

1. Yes, a flow can be called without a work object. It will need some sort of object on the clipboard that it is working on, but that doesn't need to be something you persist to the database.

2. Creating portals and flows without work objects fails to leverage much of what the Pega Platform offers. As a result most people don't do it, so I don't have a lot of experience with any specific advanatages/disadvantages. The most common case I've seen for flows on data objects are when your work object needs to do something on a list of pages, so you are doing a split for each or spin off, to do that work.

3. If the flow is creating a case, then it is creating it's own workpage. I'm assuming you're letting the process engine create the case with "Creates a new work object" checked on the Process tab. If  you're trying to do it from within an active flow... I've never seen it done, but theoretically you could figure it out. You would still be significantly better off staying within the guardrails and letting the process engine do the work for you.

4. Any object created in a concrete class can be perisisted to the database via an obj-save. As a flow moves from assignment to assignment, provided it isn't a temporary object, it will attempt to perisist itself and the associated assignments to the database. If you are doing a flow on an abstract class, you will probably face errors. You can also create utility shapes that explicitly call the obj-save method on other object than the one the flow is acting on, if you need to save them to the db.

So in short, yes you can do it, by generally you wouldn't want to. At the end of the day, what are your business requirements and how can you best meet them? You may have an edge case where what you suggests makes sesnse, but for the vast majority of the time it won't.

Thanks,

Mike