I spent many years modeling energy use in buildings using DOE2 and have modeled millions of square feet of commercial buildings and am quite familiar with that program's limitations. I'll admit up front that I am not a fan of DOE2. I much prefer continuous models that are object oriented. I do not like models that are based on discrete hourly approximations, although I understand why we sometimes need them in the short term.
Since coming to work for an electric electric utility 12 years ago, I have been analyzing commercial and residential building energy consumption by plotting average Watts/Ft2 (or Watts/m2) verses average monthly temperature that I obtain from monthly utility billing data. The resulting graphs are much more useful than the single number that is produced by EnergySTAR's Portfolio manager. The graph readily illustrates energy conservation potential, indicates which system is using the most energy (heating, building envelope, cooling, reheat or baseload), and is able to verify and quantifty improvements in energy performance after only one month of improvements being made to a facility. I'm happy to share this approach with any other building scientists that are interested.
I have also been using Matlab and Simulink to create a dynamic models of residential building's heating and cooling system. The results are quite interesting. My current model includes the effects of temperature setback on interior building materials and air temperatures. I will also be adding electric heat pumps to see how their auxilliary electric heaters respond to agressive temperature setbacks and to variations in outside air temperatures. I am trying to add mass effect of exterior surfaces to mimic those systems as well.