Look at logic for creating internal mass objects from matched spaces with non symmetrical constructions

This grew out of issue #644 where the choice of construction was not consistent from one run to the next. I believe it now uses the one from the space name that is alphabetically first. Dan proposed splitting the area in half and using both constructions.

There are a number of other issues related to internal mass that maybe we should address at the same time as this. I wanted to mention them here.

@Myoldmopar, just thought you may be interested in keeping tabs on this, and of course when we do this we can use your advise.

1. Directly related to this is a question of what is the proper area to use for the internal mass object. For example lets say we have two spaces that share a 10'x10' wall. I think we currently create a 100sqft internal mass object. Brent had mentioned that really only the area shown (100sqft) is exposed the the zone air, and in our use case 200sqft are exposed, so he felt we were not modeling as much internal mass, as is really present. But in doing both I end up with 2x the mass. Not sure it is feasible, but could the wall construction be split in half, with a 100sqft mass object for each of them. Or should we leave things as there are and should EnergyPlus offer a 1 and 2 sided internal mass object, or maybe it can already do something like this?
2. Another thing Brent pointed out, and I observed when modeling cubicles as internal mass objects, that each internal mass object slows down the simulation as much as a heat transfer surface. This is why by default interior partitions are not converted to internal mass objects, because it slowed down OpenStudio generated models too much. Brent's proposal was to combine all internal mass objects of the same construction and in the same zone to a single internal mass object, but with one extra catch. The area of any single internal mass object should not be bigger than the floor area (or more specifically half of the area of all the heat transfer surfaces). So if we implement this we may need to split into 2 or 3 if there are tons surfaces.
3. A long standing user request is to have the placement of internal mass objects affect radiant gain. I don't think there is a good solution for this now. It sounds like some people in EnergyPlus make the surfaces adiabatic heat transfer surfaces instead of internal mass, but from what I heard that comes with it's own problems. I have one kind of odd idea, that I don't think we would want to implement but thought it was worth mentioning. If I modeled a surface as a shading surface, then EnergyPlus can give me timeseries output on solar insolation. Maybe there is a way with EMS to then create a thermal storage object that mimics what really happens? If we did this it could be an EnergyPlus measure.

Start by reading issue #644 and locating the code that creates internal mass objects from matched spaces. The issue names no files or tests and raises several unresolved design questions about construction choice, exposed area, object splitting, and combining objects. Work is only done after the intended behavior is agreed and covered by appropriate tests.