Monitoring Refrigerator Energy Usage

by James Cavallo and James Mapp

The best method for monitoring refrigerator energy usage depends on how precise the measurements need to be.

Figure 1: The electricity usage of a properly functioning refrigerator displays a stable on-off signature with a regular spike, indicating the defrost cycle.

Figure 2: This poorly operating frost-free model was not keeping food at proper temperatures, while using slightly more energy than 1.5 times its rated usage level.

Figure 3: The energy-efficient unit cycles at a lower power level, during the on-off cooling periods, than the unit shown in Figure 1.

Figure 4: After fewer than 200 consecutive ten-minute periods of monitoring, the coefficients of variation consistently stabilized to closely approximate the full sample, indicating that a reasonable estimate of a refrigerator's electricity usage can be obtained in less than two days.

Calculating Annual Usage Obtaining the rated usage for a refrigerator is the first step toward estimating its actual energy usage. To improve the estimate, you need to correct for the difference in temperature between the room the refrigerator is in and the inside of the refrigerator. To figure out how this temperature difference affects the electricity consumption of a given unit, I ran a regression with the ratio of estimated kWh to rated kWh as the dependent variable and the temperature delta as the sole independent variable. The constant of the regression was .23 and the slope was .023. The T-statistic on the slope was 4.08, which shows a high level of significance. Homeowners can use this equation ((Energy usage/Rated usage) = .23 + .023 * temperature difference) to estimate their refrigerator's actual annual usage. To obtain the rated usage for a given refrigerator, the homeowner can look up the model in a database maintained by the Association of Home Appliance Manufacturers (www.aham.org). The average difference between the indoor temperature in their home and the temperature inside the food compartment must also be calculated. Then, the homeowner can plug these two numbers into the equation to arrive at an annual usage for the unit.

Discovering whether an existing refrigerator is operating inefficiently enough to warrant replacing it is an extremely difficult task for a resident who sees only a monthly electric bill. Only by knowing the approximate usage of the existing unit can anyone tell whether it would pay to buy a new, energy-efficient refrigerator.

To estimate the energy usage of a refrigerator, one should start with the rated usage of the model in question--a figure that can be found in a database maintained by the Association of Home Appliance Manufacturers (www.aham.org). However, we have found refrigerators that consume more than 3.25 times the rated usage. Moreover, we found that almost half of the refrigerators at two public housing authorities consumed at least 20% more energy than the rated usage. The only way to acquire reasonably accurate information on individual refrigerators is to monitor them.

In our recent field study at the Chicago Housing Authority (CHA), we explored monitoring approaches that would gather reasonably accurate data quickly and inexpensively. Precise measurement may be required for large shared-savings programs in which an energy services company fronts the cost of the units and accepts a percentage of the annual savings in return. But less precision may be acceptable for projects that are geared toward identifying poorly operating power hogs. Just as one would not use a sledgehammer to crack a nut, one should not use a more sophisticated--and costly--monitoring method than is needed. Less costly methods may be of particular interest to analysts working with public housing programs and those trying to include refrigerators in weatherization programs or other energy efficiency programs.

Replacement Program Reaps 500 kWh Savings

After monitoring more than 150 refrigerators, mostly for periods of from five to seven days, we found that the older refrigerators used at the CHA were consuming on average 980 kWh per year. Comparably sized, energy-efficient units (Magic Chef, model CTN1511AEW) installed under a bulk purchase program created by the Department of Energy and the Consortium for Energy Efficiency consumed an average of 480 kWh per year. When CHA pulled out 10,700 old refrigerators and installed the new units, this annual savings of 500 kWh per unit that was replaced translated into a yearly savings of approximately $570,000 in electricity bills. The unit cost of the new energy-efficient, auto-defrost units was roughly $320. (The bulk purchase cost for this program was approximately the same as the per unit cost, but it will vary by region.) With savings equal to more than $50 per year per unit, the payback period is clearly short enough to justify the expense of the new units.

The per unit savings at CHA are about 10% less than those found at the New York City Housing Authority by researchers at the Pacific Northwest National Laboratory. In that study, the older refrigerators were found to consume an average of 960 kWh per year and the new replacement refrigerators were estimated to consume 420 kWh per year on average. The slightly lower usage estimates could be attributed to small differences in the ages and sizes of the units and to the temperatures of the apartments in the two cities.

In our CHA project, we examined the effect of the difference between the temperature of the kitchen and the temperature of the food compartment within the refrigerator. Not surprisingly, we found that this difference matters a great deal. We found that at normal building temperatures, each 1°F increase in the temperature differential across the refrigerator wall leads, on average, to a 2% increase in electricity consumption of a unit above its rated usage (see "Calculating Annual Usage," p. 35). Indeed, by monitoring units during each season of the year, we found that the temperature differential was enough to explain significant changes in a unit's energy usage through the year.

Our monitoring equipment was simple yet effective. We relied on three data loggers. We logged one week of temperatures inside the food compartment and near the refrigerator in the kitchen, taking care not to place the logger too close to the stove or to any other heat-generating appliance. These measurements were made using Onset's Optic StowAway Temp loggers, which are waterproof and could withstand spills of milk, jelly, or yogurt. The devices cost about $120 apiece and need a special infrared downloading device that costs an additional $130.

We also logged the electricity usage of the refrigerator at ten-minute intervals for the same period of time. For this measurement, we used Brultech's kWh Data Logger. These devices sold at the time for about $250 per unit and required software and cables costing $165.

The electricity usage of a properly functioning refrigerator within the CHA stock prior to the replacement program is shown in Figure 1. Notice the stable on-off signature with the regular spike indicating the defrost cycle. A poorly operating frost-free model is shown in Figure 2. This model was not keeping food at proper temperatures, while using slightly more than 1.5 times its rated usage level. The usage pattern of one of the energy-efficient replacement units is shown in Figure 3. This unit has a similar signature to that shown in Figure 1, but it cycles at a lower power level during the on-off cooling periods, the off periods are more pronounced, and the spikes of the defrost cycle are lower.

How Long to Monitor?

We used the ten-minute interval consumption data to answer the question, How long should one monitor to get good results? What constitutes "good results" varies, depending on the objective. To obtain precise results for the operation of a single refrigerator model, one may want to monitor for several weeks under varying temperature conditions. But to find an accurate estimate of the average energy usage of a varied stock of refrigerators in a housing authority's inventory, considerably less monitoring time is required. And it is possible to identify which refrigerators are operating significantly above their rated levels with only a short monitoring period.

We examined the coefficients of variations--the standard deviation of a sample divided by the sample's mean--for the energy usage of the refrigerator shown in Figure 1. We started by taking 100 random samples of two consecutive readings from the week's worth of information collected. Next we calculated the mean and standard deviation of the 100 samples. We then took 100 random samples of three consecutive readings and again calculated the standard deviation and mean. We continued "bootstrapping" random samples with longer and longer consecutive readings until we had 300 consecutive observations.

Figure 4 is a plot of the coefficient of variations for the samples of 2 through 300 consecutive observations. This graph shows that after fewer than 100 consecutive observations, the coefficient of variation stabilizes. The pattern shown in Figure 4 was found repeatedly in analyzing ten-minute interval data for 14 refrigerators. The coefficients of variation consistently stabilized to closely approximate the full sample after fewer than 200 consecutive observations. Since 48 hours would be 288 consecutive ten-minute periods, a reasonable estimate of the electricity usage of a refrigerator can be obtained with less than two days' monitoring.

To be sure, monitoring a unit for only two days could yield a misleading indication of annual energy usage. However, if one is sampling the many refrigerators in a housing authority's inventory, any one outlier will have a limited impact on the mean for the entire sample of refrigerators. This is a straightforward application of the law of large numbers. In our monitoring for the CHA, we sampled 150 units.

Monitoring for Less

Most larger housing authorities are interested in measuring appliance energy usage in order to define a baseline average electricity consumption for a large-scale replacement with energy-efficient units. Robust estimates of the average electricity consumption for a stock of refrigerators can be obtained by sampling two-days' worth of consumption for each unit. One can also use simple watt meters or line loggers to monitor the cumulative usage over the two-day period rather than having to capture an electricity usage profile throughout the sampling period with a more expensive kWh data logger.

Although we often used temperature data loggers to view the profile of the temperature in the food compartment and in the apartment, we also used a $50 digital thermometer. When we used the digital thermometers, we measured the food compartment and apartment temperature at the start and at the conclusion of the monitoring. Many digital thermometers record the minimum and maximum temperatures between resetting. At times we left the thermometers in the refrigerators through the monitoring period to look for substantial temperature swings.

A line logger plugs into an electrical wall outlet, and the refrigerator plugs into it. The line logger accumulates the kWh over the period of monitoring. The time that monitoring begins and the time that it ends must be recorded accurately since many line loggers do not record elapsed time. It is also important to sample a large number of refrigerators. However, with two-day monitoring periods one can move a line logger around to many locations in several weeks. At the Pittsburgh Housing Authority, we used just five line loggers and collected data on 35 refrigerators in four weeks.

Identifying Poor Performers

Cutting down the monitoring period to two days may still be the equivalent of cracking a nut with a sledgehammer if a housing authority only wants to identify and remove refrigerators that perform poorly (see "Other Methods of Shortening Refrigerator Monitoring Time," p. 34). A refrigerator that has lost most of its refrigerant or that has a malfunctioning compressor will run constantly rather than cycle on and off. When a refrigerator's motor never turns off, its hourly energy consumption is a simple multiple of its maximum power usage. For instance, if an apartment-sized unit has a maximum power usage of 250 watts and it uses 0.5 kWh (or 500 watt-hours) during a two-hour test period, or 250 watt-hours per hour (the same as the maximum power usage), this indicates that the refrigerator is running constantly.

If spotting power hogs is the goal, this can be accomplished with a line logger during a typical two-hour residential audit. While two hours would not be adequate to achieve great accuracy in estimating annual consumption, it is sufficient to determine if the refrigerator is operating 100% of the time.

Cincinnati Gas & Electric Company (CG&E ) is using this two-day protocol in its refrigerator replacement initiative for its low-income weatherization program. CG&E is finding that approximately 25%-30% of the refrigerators need to be replaced.

Summing Up

The savings from replacing older refrigerators can be substantial, and collecting the data needed to determine when refrigerators should be replaced is easier and less costly than one might think. In both Chicago and New York City, replacing existing units cut refrigerator electricity usage by more than 50%. Monitoring to develop an average usage for the existing stock of refrigerators is a task that can be completed by maintenance staff in a reasonably short time--and identifying poorly performing units that should be immediately replaced can take just two hours of monitoring.

James Cavallo is manager of the Existing Buildings Efficiency Research program at Argonne National Laboratory, and James Mapp is an energy analyst with the Wisconsin Energy Bureau. The CHA study was supported by the Department of Energy's Rebuild America program.

For more information: Home Energy, January/February 1993 (Special Refrigerator Issue). Pratt, R.G., and J.D. Miller. The New York Power Authority's Energy-Efficient Refrigerator Program for the New York City Housing Authority: 1997 Savings Evaluation, PNNL-11990, September 1998.

Home Energy can be reached at: contact@homeenergy.org
Home Energy magazine -- Please read our Copyright Notice