Given our unique project conditions we are pursuing this credit via Case 2, Option 2B. This option allows for normalization of historical energy data for things such as, changes to occupancy, operating schedules, space use.
How do we derive the multipliers that should be used for normalizing our data? One reviewer we spoke to indicated that an energy model was not needed and the calculations could be done much more simplistically. Can we use linear relationships such as twice the number of occupants equates to an assumed 2x energy usage? Or, there was an addition to the building, so would a 20% increase in conditioned space volume equate to an assumed 20% increase in energy usage? And what about less quantifiable aspects like space type changes?
Does anyone have experience with this? Any tips or resources for this endeavor is greatly appreciated! We are trying to get this turned around quickly.
Jenny Carney
Vice PresidentWSP
LEEDuser Expert
657 thumbs up
August 6, 2013 - 12:23 pm
Hi Tim,
It's true that an energy model is not needed to further normalize consumption data. But, I would also caution against using unproven generic linear relationships in the normalization, and would recommend starting by identifying variables that have a strong correlation to consumption in this specific building.
EUI on a weather normalized source kBtu / SF will generally be your starting point, and would cover scenarios such as the building addition you mentioned.
Space type changes would be the most difficult to accurately account for, and it would really depend on the specific situation.
For occupancy changes, I would recommend starting by performing correlation tests between occupancy levels and EUI. If there is a strong correlation (high r^2 value), you can use the equation associated with that regression to perform the normalization. If there is not a strong correlation, as would likely be the case in buildings with high process loads, you can use this regression to justify a decision to not do further normalization. This just takes basic statistics skills, and can be performed in an Excel spreadsheet.
If you have multiple variables you believe are influential, the most legitimate approach would be to do a multivariate regression analysis, but this requires more stats know-how and specialized software.