Before committing funds to a lighting upgrade, it pays to understand just when projected savings will occur. While it may be easiest to use an average electric rate (i.e., total dollars divided by total kilowatt-hours [kWh]), doing so may obscure the impact of time-variable electric rates. Depending on rate structure, the value of power saved by a given option (e.g., occupancy sensors) may be higher or lower than the average cost for power.

In some parts of the United States (e.g., California and the Northeast), a significant portion (30% to 50%) of an electric bill may be based on monthly peak demand charges. In order for sensors to save power priced at an annual average rate that includes peak demand, they must do so by shutting off lights at that peak time. In many facilities, however, nearly all lights may need to be on during the usual 1 to 3 p.m. peak time or may be off in rarely occupied spaces (e.g., store rooms). In either case, sensors would not be providing savings. On the other hand, they may greatly reduce night time (i.e., off-peak) lighting kWh consumption.

If you are paying a flat rate for power (with no variations for time-of-use and no demand charges), it's safe to use an average \$/kWh: the value of any kWh being saved at any time is the same. If, however, little or no savings occur during the peak time, using an average rate could seriously overestimate the dollar (as vs. kWh) savings that will be seen on the electric bill. If demand charges are (for example) 40% of the total electric bill (and sensors do not cut that peak at the time it occurs), using an average electric rate could overestimate actual dollar savings by 67%.

To properly calculate those savings, check your electric tariff to see exactly how you pay for power. Using our 40% example, an average electric rate of \$.08/kWh would include about \$.032/kWh for the peak demand that might not be saved. As a result, the conservative value for saved electricity would be only \$.048/kWh.

The reverse may be true for lighting upgrades that permanently reduce wattage (e.g., delamping, electronic ballasts). Such options should reduce demand even during the peak period, but (if lights are off at night) will not save the cheapest kWh. As a result, using an average \$/kWh value may underestimate their likely dollar savings.

## Marginal Power Pricing

Note also that lighting savings may occur at the marginal price for power (i.e., the cost for last kWh used). Electric rates using a descending block structure make the last kWh used less expensive than the first (or average) kWh used. As a result, reducing overall usage by a small proportion (e.g., still within the last rate block) may save kWh that are priced below the average consumption-only charge. Where such block structures exist, kWh at the margin may be another 10% to 20+% cheaper than the average rate (without demand). If we knock another 15% off that conservative \$.048/kWh to account for the block differential, actual savings (for sensors) may be only \$.041/kWh, or roughly half the average level (\$.08/kWh).

## Watch Those Burn Hours

Care is also needed regarding the annual burn hours assumed for an upgrade option. If one assumes lights are left on unnecessarily for many hours per year, sensor savings should be significant. To verify that belief before initiating lighting upgrades, install lighting loggers to verify when and how long lights are on while spaces are occupied, and when they are left on unnecessarily.

Just because one drives by a building and sees lights on at 9 p.m. does not mean that those rooms necessarily contain significant wattage, or that they are not later turned off by janitors or security people. Without automatically (and surreptitiously) logging lighting and occupancy hours, any assumption is little more than conjecture. During my own experience in this area (covering over 9 million sq ft and 5,000 automatic lighting controls), I found wasted hours were not as high as expected, and occurred on nights and weekends when demand charges were low or nonexistent.

Fortunately, a variety of low-cost devices exist to log hours of occupancy and hours of illumination in formats that may be downloaded to a spreadsheet for later analysis. Most major sensor manufacturers (and ESCOs) can provide such equipment or services. For a good example, go to the Watt Stopper site at: www.wattstopper.com/products/productline_list.html?id=13&catindx=10. ES