By Richard Bingham
Reprinted from Electrical Contractor Magazine, September 2013
When reviewing data from your power monitoring system, the 30,000-foot approach is a typical starting point. 

Figure 1 shows such a view of the active (or real) power consumption over the last billing period at a light manufacturing facility. This data was compared with the data from the same period the previous year, the daily temperature highs and lows, and the work schedule to look for any deviations, especially unexplained increases, which can result in higher electric bills for demand and energy consumption charges. Note the reduced power for Memorial Day, Mon., May 27. Also, a small crew was working on two of the weekends at this facility, resulting in higher-than-normal weekend power levels. All three phases had similar plots as Phase A. 

However, a closer examination of the weekend data of June 1–2 showed something more than just a small crew working. There were maximums of current, which were 400 amperes above the baseline current levels and on all three phases, occurring at 15-minute intervals. Checking the average 10-minute value of parameters—such as voltage, power factor, harmonic distortion and volt-amperes-reactive (VAR)—indicated that the source was from within the facility. The voltage didn’t change much to correspond with the peaks, the power factor and VARs showed only a small inductive increase, and the current harmonic distortion increased slightly (see Figure 2).

Closer examination of the weekday data showed that the same pattern was there during working hours but was somewhat obscured by the normal changes in current from the manufacturing operation and other loads in the facility. This caught the attention of the building operations manager and the CFO, as it can cause a significant increase in the monthly electric bill. Most nonresidential facilities pay both a demand and energy consumption charge. The latter is the sum of the power consumed over the billing period. The former is a peak of the average demand value of power over the demand interval (typically 15 minutes). It is a “racheted” value in that, once it increases above the previously recorded peak value during one interval, that new value becomes the demand charge for several years, regardless of the actual demand levels during the month. 

It would take a closer look to find the source. The problem actually stopped on a couple of occasions, but no one made a correlation to an operational change. The monitoring system was reprogrammed to capture the waveforms of the voltage and current during these peak current excursions. It showed that the current changed abruptly in one cycle and decreased each subsequent cycle until it was back to the steady-state value within five cycles. Such waveforms, along with the previously mentioned data, were consistent with a three-phase motor startup. It would also be consistent with a heating, ventilating and air conditioning (HVAC) system, but that didn’t make sense since that system was running the same program in the facility for months. Or so they thought.

The missing puzzle piece

The manufacturing operation only occupied 50 percent of the facility. The other half of the facility had been vacant for some time. The landlord had finally secured a tenant for the unused space and was renovating it. While a separate breaker panel was put in for the new tenants, the monitoring system (and utility revenue meter) were on the main service panel at the point of common coupling (PCC). They were reading the electrical power for the entire facility, which years ago was just for one tenant. During the renovation, duct work was disconnected and was being rerouted, but before the work was complete, a service person from the HVAC maintenance company had come in to test the system. It turned out that he reprogrammed those zones as if they were occupied and never shut the system back down.

Fortunately, this potentially costly problem was caught before the summer months and the real heat descended.