ok, so I would say that this is a major/core feature of Capella (and Sirius): diagrams are views on your semantic models. If your semantic model changes, and if some elements that changed are displayed in your diagram, then your diagram is updated. Otherwise, this will be very misleading for a user if a diagram could show information that is not the actual one on your model. Users will quickly lose confidence in the tool.
Now I think I understand what you are trying to do. What you are calling a use case may be viewed as an “alternative”: You have 2 alternatives, you want to model them and evaluate them.
Cloning a diagram does not mean that you are creating another alternative in your model. It means that you are just cloning a view of your model. With your cloned view, you are still modifying the same model.
So there may be different ways in achieving what you are trying to achieve, but then you would have to be more specific about your context, and the modeling elements you are using. It could range from things like:
- generate an image of your first SAB, it won’t change when you modify your model…
- create new elements (functions/exchange) in your model that are separate from your first alternative
- duplicate your Capella project to study the 2 alternatives
- use the system to subsystem transition and/or the filtering add-on to manage your alternatives
- use REC/RPL and libraries
And maybe other ways. but again, this will depend on understanding what you are trying to do and finding the best way to do it with Capella.