Question-and-Answer Resource for the Building Energy Modeling Community
 | 1 | initial version |
Impact on simulation results: My guess would be marginal. As you've mentioned, grouping an exterior space with interior spaces would artificially impact the heat balance of zones neighboring the interior spaces, but even here I think that typically that impact would be small compared to the internal loads in the neighboring zone (depending on the building, of course). This is assuming the volume of the combined spaces is calculated accurately (i.e. exterior loads get 'averaged out' over entire zone air volume.
Impact on simulation time: I think the jury is out on this. I found this old report on EnergyPlus runtime, which characterizes a runtime increase by a factor of 2.6 for a tenfold increase in zone windows, and the same factor for a 2x increase in the number of zones (15 to 30, conditioned via VAV) [pg 31]. The applicability of that study to your question (or generally) is suspect, since it looks like it was done back in the FORTRAN days..but maybe relative performance difference didn't change much in the move to C++.
Would you actually do it?: Yes, and I regularly do. Especially for big models, being able to group unconditioned spaces into one zone makes managing input much easier.
I would love to see more empirical responses to these questions, but that's my take.
