Over the years people have often complained about trying to document replacing existing windows with modern windows but maintaining the existing opening. This means swapping windows with families that are the exact same size. This is related to the wall that Revit creates to infill a wall that has a demolished window. You may have encountered this warning message?
Hypothesis A: Within phasing we can replace Existing Window A with New replacement Window B using the exact same sizes (same size opening dimensions) IF they both use an OPENING in the family to cut the host.
Hypothesis B: Within phasing we can replace Existing Window A with New replacement Window B using the exact same sizes (same size opening dimensions) IF they both use a VOID in the family to cut the host.
If either of the above are false then...
Hypothesis C: Within phasing we CANNOT replace Existing Window A with New replacement Window B using the exact same sizes (same size opening dimensions) IF one uses an OPENING and the other uses a VOID, in the family, to cut the host.
Hypothesis A is TRUEHypothesis B is FALSEHypothesis C is neither
I find that I can replace a Void based window with an Opening based window but not the reverse. Also any alterations to the hosting wall in the project; such as length adjustment, or top/bottom offsets, attach/detach will place the void type window at risk of being deleted. Weirder still is that it might not delete all of them, one or more.
It seems that the short answer is: eliminate window families that use a void to cut the hosting wall IF you intend to place identical sized windows in phased projects; to demonstrate existing window replacement without altering the existing openings. Windows created this way will not work for this task. A logical next hypothesis is to anticipate similar behavior in door families.
I speculate that this warning happens because an opening cuts fully through the host while a void (or combination of voids) might cut more and/or less of the host as it travels through it. I suspect that the infill wall can't abide the shape a void might create and that in turn means a void won't be viable as a window to alter the location of the infill wall. I think it's similar to curtain walls only supporting non-rectilinear panels with the system panel families.
I'd love to hear that the developers will test this against their own experience and expectations. I know that a lot of people rely on voids to shape the opening in a host wall to match a variety of actual construction techniques for openings. To eliminate voids in families as an option for this kind of project (replacement windows (and doors?)) is not ideal. They'd probably have to revisit the entire logic of infill elements where demolished hosted elements occur. Perhaps leave it up to us to fill in holes?