Just before the record string of days above 90°F began, I installed a power monitor at a friend’s residence to determine if the electric revenue meter was reading correctly. We will now look at the data for signatures of various appliances to hopefully determine offenders that resulted in higher energy consumption, provided that it wasn’t an out-of-calibration revenue meter. After three weeks of 10-minute interval data (and some interesting PQ waveforms), it’s time to extract the information from the data and provide answers to the customer.
The electric utility’s revenue meter on the side of the house was read each week, and the data was downloaded from the instrument. The data was reviewed and analyzed in the instrument manufacturer’s software program. I also exported the 10-minute interval kilowatt-hour values into Excel.
From the electric meter data, the residence was found to be consuming 27.4 kilowatt-hours (KWh) per day on average over the 21-day monitoring period. The data from the instrument at the main breaker panel was 27.1 KWh accumulated over the monitoring period, which was close enough to rule out the revenue meter as being the source of the increase in the electric bill. The electric utility’s monthly bill for the residence did state that there was a 13% increase in usage over last year for the same period. So, the hunt began for what was contributing to this.
We started with a top-down approach. The figure shows the electrical energy consumption for the residence over the 21 days. Four distinct patterns appear, which are labeled A, B, C and D on the graph. Pattern A also shows up on day 11, and Pattern C shows up on the last two days of the monitoring period. Pattern D has an almost steady state pattern over the three days. The lowest was at a 0.5 KWh rate, which is less than 40% of the average over the period.
Though not visible in the graph as it is presented here, the step increase in patterns A and C occur at approximately the same time each day during most of the monitoring period, between 7 and 8 p.m. The increase level in A is initially four times that average usage prior to the change but then decays down over the next 10 hours to about twice the baseline daily rate. The amplitude of the step change in kilowatt-hours in Pattern C is nearly double that of the step increase seen in Pattern A. However, this peak value only continues for approximately 4 hours and then decreases to levels seen in Pattern A for the remainder of the night.
Correlating the outside air temperature during the monitoring period made it a relatively easy guess that this was caused by the use of window air conditioners.
Pattern A is caused by two smaller units being turned on at the aforementioned time to cool the second-floor bedrooms when the inside temperature is above 85°F.
When the energy usage levels are higher in Pattern C, it is because another, larger window unit is turned on in the first floor until “it’s time for bed”—Pattern C. Pattern B is a normal day without using any air conditioning units.
Pattern D turned out to be the result of a three-day vacation. There was no cooking, cleaning, television or other typically daily activities. The baseline energy consumption is from two refrigerators (never opened during the time) and all of the electronic devices that consume energy even in sleep mode. This was apparent in the waveform data for the current, which had that typical bump near the peaks of the sine waves, that we have shown in prior articles, caused by the large 3rd, 5th and 7th harmonic current percentages of single phase, full wave rectified power supplies used in most electronic equipment.
Using the Pareto Principle (80% of outcomes result from 20% of causes), it is clear that the AC units are the major source of the kilowatt-hour consumption. While a newer 12,000-Btu unit will consume a bit more than 1 KWh, the older one consumes nearly 2 KWh, although it is only run for several hours versus the 11 hours of the two smaller units. Computing the return on investment shows that the purchase of such during an “end-of-season” sale would be a good move to avoid having next summer be a repeat.
Next month, we will dig deeper into the daily normal day data to see if there are any more opportunities to save money.