One of the benefits of an accurate energy model is the ability to separate the signal from the noise of your building. Normally the link between operational management and energy use is obscured by weather, so small changes only show up in the long-term energy profile, if at all. Snapmeter provides a weather-normalized snapshot of building behavior, allowing you to run quick, easy energy experiments.
When conducted properly, energy experiments provide a powerful and inexpensive way to test different operating procedures. In fact, experiments are often the only reliable way to improve operations, because building complexity overwhelms other management methods. Buildings are systems, and the interactions between setpoints, fans, pumps, compressors and occupants, among other factors, determine how much energy you use. You could measure and monitor every point, but doing so is cost-prohibitive, the resulting dataset would be unwieldy, and you would still need to weather-normalize to separate the signal from the noise. (Also, occupants tend to object to the electrodes.)
Instead, focus on your whole-building meter data and compare what you actually did with what you were expected to do. Two examples help illustrate.
The first is a simple experiment to determine how much weekend HVAC affects spend. On a relatively mild weekend, building staff disabled HVAC during the morning hours, letting it catch up after 10am. You can see the effects clearly in the customer’s Snapmeter report:
The weather-normalized results show clearly that this change brought about a net savings. Savings are 3% of Saturday spend, or about $13 ($700 over the course of a year). Peak demand is a bit higher than expected on Saturday, which also makes sense, because the cooling system has to work to catch up. This little bit of excess demand is costless, because it is nowhere near the peak for the billing cycle. As always, note that Snapmeter only reports variances if they are statistically significant.
The second is a lighting control experiment. During off-hours, many buildings can either allow for manual override of lighting shutdown, or use sensors to control lights. Sensors are more convenient for occupants, but some draw power from the ballast and all can be falsely triggered. In this experiment, the chief alternated lighting strategies for two hours on consecutive Saturdays. Here is the result:
The hard override saved $0.42 an hour per floor compared to sensors. This may sound small, but it adds up to $44,000 of savings over the course of a year. Not bad for a quick little experiment.
