|Metrix » Tuning|
Other knowledgeable analysts who have tried regressions have found little change in significance between regressing demand vs. average ΔT vs. max ΔT (we do the former). While we have not verified this to be true, we have no trouble believing it, since daily average and max. temperatures are highly correlated within one and the same weather station.
The fact that demand does not correlate well with outside temperature is not surprising. All that demand usually means is highest electrical power over any 15-minute interval. Even if we assume that the demand-triggering device(s) are temperature sensitive (say, some chillers and air-handler fans), it is fairly easy to see that they will "max out" fairly soon, i.e. while their daily kWh consumption will increase with daily degree-days (or average ΔT), demand will not, as several 15-minute intervals at full power will register the same as a single 15-minute interval. At best, I would expect such temperature-sensitive demand to have two plateaus, a winter and a summer plateau, with the only temperature sensitivity in a few swing months, too little to do any serious regressions on.
By the way, the same "knowledgeable sources" above confirmed then and now that demand correlates poorly with just about anything.
All this being said, it would be technically not difficult to give users an option what to regress against (and let them see for themselves how little difference it makes). What *would* be messy is how to document and report all this. E.g. how to show in a verification or contract report that the "DD" regression coefficients are vs. avg. temp, not max. temp. Should we have it a separate choice for cooling and heating DD? If we simply made an arbitrary decision to correlate energy vs. average T and demand vs. max. T, what would we do with historical projects regressed against average T?
A long way of saying that if we thought this would bring a noticeable increase in quality of analysis, we would be all for it.
(* Δ = Delta)
Otherwise feel free to contact our Tech Support staff at (805) 329-6565, or via email at email@example.com.