In the previous post, we looked at how I used benchmarking on a recent project to compare the energy consumption in an Oregon laboratory facility with its peers. While down and dirty, the benchmarks could be interpreted to say that there are potential energy efficiency improvement opportunities on the site and that they probably lie in the area of electrical energy savings.
The next step in my analysis effort was to take a look at average daily energy consumpion. I learned this technique very early in my career from several of my mentors at McClure Engineering, one of whom is now associated with 8760 Engineering. Over the years, I have found that it can provide valuable insights into building energy performance and that it is a good way target building retrocommissioning efforts and monitor their results.
I wrote a paper that goes into the details of the technique that can be downloaded from the California Commissioning Colaborative’s web site if you are interested in the details. In general terms, building energy data is converted to average daily consumption and then normalized to the calendar months. The normalization:
- Correlates the consumption with when it occurred versus when it was billed for. For instance, the electric bill that I just received at home is dated August 11, 2007, but actually reflects the electricity we used here at home in July
- Correlates the consumption data to calendar months. This allows the data to be compared with other data that is typically tracked on a monthly basis like weather, production, occupancy, and other factors that tend to drive energy use.
The impact of the normalization process is illustrated in the following graph, where normalized gas consumption and gross gas consumption for the facility I have been discussing are plotted on the same graph.
Notice that while the general trends for both lines are similar, there are some differences that could be misleading if you were using the data to assess the facility.
For instance, there is a significant difference in the magnitude of the consumption shown for March. This is because the gross consumption is based on a bill that was received in March but actually reflects consumption from February 10th through March 15th. Thus, it reflects what was going on in February as well as March. It also reflects a billing interval of 33 days spanning from February, a month with 28 days into March, a month with 31 days.
Similarly, the gross consumption for November actually reflects 20 days of October consumption and 9 days of November consumption. As a result, the increase in consumption associated with colder weather seems to be shifted to November rather than occurring in October when the weather actually turned colder.
The bottom line is that normalization tends to account for these factors and more accurately reflects the calendar month when contrasted with the raw data for the billing period.
Performing this sort of analysis is not difficult. When I first learned of it, I did the work with a four function calculator, a slide rule, and some graph paper and it didn’t take that long. Computers with spreadsheets make it very easy, especially after you have set up the necessary calculations the first time.
In the next post, we’ll look at the normalized average daily consumption for the lab facility I have been discussing and look at some of the clues it contains regarding retrocommissioning targets and the increase in site energy consumption after the renovation project.
Senior Engineer – Facility Dynamics Engineering