A S H RA E JOURNAL The following article was published in ASHRAE Journal, December 1997. © Copyright 1997 American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. It is presented for educational purposes only. This article may not be copied and/or distributed electronically or in paper form without permission of ASHRAE. Less Than One Watt per Square Foot Space Cooling Demands From Office Plug Loads By Paul Komor, Ph.D. 1. Using actual information based on expected building use, or 2. Using manufacturer’s published data or information from technical societies, or 3. “Other data based on the designer’s experience of expected loads and occupancy patterns.”2 This first option often isn’t possible for The third option, based on experience and accepted practice, is the most common. Informal discussions with engindersizing space cooling systems neers and HVAC system designers for office buildings can result in reveal that typical assumptions are in the uncomfortable and angry tenants on peak 2 to 5 watts per square foot (21.5 to 53.8 cooling days. However, oversizing W/m2) range—a bit lower than ASHwastes money because more capacity is RAE’s upper boundary. Some are a bit installed than is needed, and oversized higher. In the current $1.2 billion renosystems have a lower vation of the Pentagon energy efficiency which (a 6.5-million ft2 makes operating costs [604,000 m2] governhigher than necessary. ment office building Oversizing can outside Washington, adversely affect comD.C.), initial specififort as well, because cations call for an oversized systems may internal equipment provide poor humidity load assumption of control and large tem5.5 watts per square perature variations. foot (59.2 W/m2) for Correct system sizopen office space.4 ing requires estimating Fortunately, enough building heat loads measurements of plug accurately. Many facloads are now available tors contributing to to replace assumptions building heat loads with measured data. must be considered These data suggest that when sizing cooling nameplate ratings do Table 1: Energy use of office equipment in a typical 50-person office. systems. Space cool- Power consumption numbers are based on measured data for 1995 not indicate actual ing systems must be average stock equipment (not nameplate ratings). power use and that plug sized to remove heat loads are typically and moisture from between 0.4 and 1.1 both external loads (such as solar gain office equipment, because it is usually not watts per square foot (4.3 and 11.8 W/ and outside air) and internal loads— known what types of equipment will be m2)—far below the 2 to 5 watts per square lighting, occupants, and plug loads. used in the space. Even if the building foot (21.5 to 53.8 W/m2) commonly The plug loads category includes any will be owner-occupied, final decisions assumed. electrical equipment that is plugged into about office equipment are almost always Actual Heat Output Not Indicated outlets. These loads typically account made after construction is completed. for about 15 to 20% of total cooling The second option is of limited use by Nameplate Ratings load.1 For office buildings, the plug for office equipment, because until now Several studies have examined the loads that use the most energy are com- there have not been sufficient data avail- relationship between nameplate ratings puters and related equipment such as able for technical societies to provide printers, copiers, and monitors (See guidance. The 1993 ASHRAE HandTable 1). The proliferation of these book—Fundamentals notes that “in About the Author devices in offices has become an offices having computer display termiincreasing concern in recent years. Paul Komor, Ph.D., is research managnals at most desks, heat gains range up to Little guidance is available for esti- 15 Btu/h•ft2 (4.4 watts per square foot er at E Source, Inc., an energy research mating the magnitude of these plug [47.3 W/m2]).”3 This provides a valuand information firm; and is also a facloads. California’s Title 24 standard, for able upper boundary, but does not proulty member at the University of Coloexample, gives three options for calcu- vide the detailed guidance necessary to rado in Boulder. make reasonable assumptions. lating miscellaneous equipment loads: U December 1997 ASHRAE Journal 41 and actual power use.5 These comparisons, summarized in Figure 1, show that actual measured power use is typically 20 to 50% of the nameplate rating. Nameplate ratings are intended for use in sizing electrical wiring. These ratings are not intended for use in calculating actual non-instantaneous power use or heat output. There may be times when nameplate power is drawn for very short periods (for example, when starting equipment). However, these short-term power draws do not produce appreciable amounts of heat, and therefore should not influence cooling system sizing decisions. Actual Power Demand: The Evidence In the last few years, a number of building researchers and engineers have taken real-time measurements of plug load power use. This is difficult because the wiring to outlets is often on the same circuit as wiring to overhead lighting. This makes it challenging to separate out the outlet power. Also, office equipment contains power supplies that can produce third harmonics. These can be difficult to measure but should be included because they contribute to heat output. Despite these challenges, enough high-quality measured data on plug loads exist to substitute real data for informal guidelines. The data agree with each other—but differ from commonly-used assumptions and guidelines. As shown in Figure 2 and Table 2, measured data on actual plug loads range from 0.4 to 1.1 watts per square foot (4.3 to 11.8 W/ m2).6 These data are drawn from 44 buildings, covering a total of 1.3 million ft2 (121,000 m2). The buildings were selected for measurement because they represent “typical” office buildings. Although the results may seem low, consider the type of equipment found in a typical office and the amount of power it actually consumes. For example, assume that an office with a staff of 50 people has a computer with color monitor for each employee, and that there are five networked printers, three copy machines, and three fax machines. In this office there are also two refrigerators, two microwaves, and 10 freestanding incandescent lamps in the office. Total power demand for these devices, as shown in Table 1, adds up to 10.2 kW. (No allowance is made for diversity because all devices are assumed to be always turned on.) Also, assume 200 ft2 (19 m2) of conditioned space per person (this includes meeting rooms, hallways, bathrooms, and other common space). This scenario yields 10,000 ft2 (930 m2), or 1.02 watts per square foot (11.0 W/m2). This is consistent with the measured data shown in Figure 2. Benefits of “Rightsizing” Assuming one watt per square foot for plug loads, as the data suggest is appropriate—rather than the more typical three watts per square foot—can make a significant difference in cooling Table 2: Office equipment plug loads data. Data were compiled from both published and unpublished sources. Unless noted otherwise, all measured watts per square foot data include all plug loads. 42 ASHRAE Journal Fig. 1: Nameplate vs. actual power demand. Each dot indicates a separate study. For example, actual power demand for computers was found to be 14 to 33% of nameplate power demand. Study results differ largely because different brands with different internal equipment were tested. system sizing and costs. For example, assuming three watts per square foot (32.3 W/m2) for plug loads in a 100,000 ft2 (9,300 m2) office building yields a 300 kW office equipment load. Removing the heat from this load would require 85 tons (300 kW) of effective cooling capacity. Assuming one watt per square foot, however, corresponds to about 28 tons (100 kW) of effective cooling capacity. The difference (57 tons [200 kW]) translates into a significant cost savings. Chiller capacity alone costs about $400 per ton ($114/kW),7 so an accurate estimation of plug loads would save about $23,000 in up-front chiller capacity costs. (Cooling system inefficiencies require more than one ton of chiller to remove one ton of load, so actual savings would be a bit higher.) For new construction, additional savings of up to $3,000 per ton ($850/kW) may be possible by equivalent downsizing of ducts, fans, and other cooling-related equipment.8 The operating cost savings of “rightsizing” are significant for packaged December 1997 OFFICE BUILDING rooftop units and chillers. Packaged units are often inefficient during start up, and reach maximum efficiency after about 10 minutes.9 Therefore, an oversized packaged unit that cycles frequently will spend more time operating in its inefficient start-up period. For packaged rooftop units and chillers, there is an efficiency penalty for operating at low load.10 Rightsizing chillers forces the typical operating conditions into the efficient medium to high load range, which can result in operating cost savings. Rightsizing cooling plants has other advantages as well: • A properly-sized cooling plant cycles less frequently, keeping indoor temperature constant and providing better humidity control. • All else being equal, maintenance of smaller units is simpler and cheaper (maintenance contracts are typically charged by the ton). • In retrofits, smaller units free up electrical capacity that can be used for other needs. The data shows that plug loads are considerably lower than commonly thought. It is important, however, to recognize that these are average officewide loads. Certain locations such as kitchens and copy rooms can have much higher plug loads. Although total cooling capacity should be sized to handle average loads, air handling and distribu- tion systems need to be flexible enough to handle higher loads if necessary. The use of diversity factors in load calculation software is a useful method to limit oversizing of cooling plants. Plug Loads Decreasing New HVAC systems in commercial buildings should be sized not only to accommodate the building’s current loads, but to handle expected future loads as well. Future plug loads will depend on equipment density (the number of computers, printers, and other devices per square foot), hours of use per year for each piece of equipment, and energy use for each piece of equipment. Although these variables are uncertain, some general trends are likely. In the United States, the Energy Star program (a U.S. government-sponsored effort to encourage office equipment manufacturers to increase energy efficiency) probably will cause significant reductions in per-device energy use for computers, monitors, printers, copiers, and fax machines sold in the United States. Similar programs are in place or being discussed in Australia, Japan, New Zealand, Sweden, and Switzerland. Forecasts of equipment density for the United States show computer and monitor density continuing to grow, but at a declining rate—reaching about one computer and monitor per person by about 2000.11 (Of course, some offices already may have more than one computer per person, but the average U.S. office currently is at about 0.7 per person).12 Hours of use per year and person density (square feet per person) are unlikely to change significantly. The result of these factors, according to a recent study in the United States,13 will be a decrease in office equipment energy use intensity (kWh per square foot per year) from now until about 2002 (see Figure 3). Forecasting is an uncertain business, but there is little evidence that office equipment power densities (watts per square foot) will increase significantly in the next decade. Rather, they will likely decrease, partly because of technical advances promoted by the Energy Star and other similar programs. Conclusions Plug loads require significant cooling capacity. Moderate oversizing for plug loads (in case tenants are unusually equipment-intensive) is reasonable. A useful guide for such oversizing is considering how outdoor design conditions are determined: a typical assumption is to use 2.5% conditions, meaning that the cooling system capacity is designed to provide comfort in all but the hottest and/or most humid 2.5 hours out of 100.14 By analogy, the data shown in Figure 2 have a mean (µ) of 0.83 watts per square foot (8.9 W/m2) and a sample Fig. 2: Office equipment plug loads. This figure shows measured office equipment power use (in watts per square foot) for 44 buildings comprising 1.3 million ft2 (121 000 m2). The simple average is 0.83 watts per square foot (8.9 W/m2). December 1997 ASHRAE Journal 43 standard deviation (σ) of 0.21 watts per square foot (2.3 W/ m2). Assuming the data are normally distributed, therefore, suggests that only 2.5% of the data will have a value greater than µ+2σ, or 1.25 watts per square foot (13.4 W/m2). When building new facilities or replacing existing cooling systems, energy users should carefully examine assumptions for office equipment energy use. If plug loads greater than 1.25 watts per square foot (13.4 W/m2) are specified, there should be a compelling reason for doing so—such as clear evidence that actual loads in the space will be much higher than typical loads. In such cases, these loads should not be overestimated. Even in very equipment-intensive settings such as commodity trading floors, plug loads have been measured at three watts McGaffin, “Measuring Computer Equipment Loads in Office Buildings,” ASHRAE Journal (August 1994). 6. Dubin-Bloome Associates, P.C., “Conceptual Design Report for Pacific Gas and Electric Customer Technology Test (ACT 2 ) For Energy Efficiency,” prepared for PG&E (January 1991); J. Farley, personal communication (March 1996), Center for Electric End-use Data (CEED), EPS Solutions, 1 Richmond Square, Suite 122C, Providence, RI 02906; Larry Lister (March 1996); E. Martin, “Energy Efficiency of Electronic Office Equipment: Case Study for a Building Retrofit,” Proceedings of the 1992 ACEEE Summer Study on Energy Efficiency in Buildings, ACEEE, Washington, DC, v. 1, pp. 167 Load Equipment Energy Use in the Richard Blanshard Building, 1992 and 1994,” prepared for British Columbia Buildings Corporation (May 1995); and Wilkins and McGaffin (1994). 7. Author’s estimate based in part on Means Mechanical Cost Data, 19th Edition (1995), p. 281. 8. Martin Kiley and William Moselle, eds., National Construction Estimator, 40th Edition (Carlsbad, CA: Craftsman Book Company, 1992), p. 476. 9. J. Proctor, Z. Katsnelson, and B. Wilson, “Bigger Is Not Better: Sizing Air Conditioners Properly,” Refrigeration Services and Contracting, v. 64, no. 4 (April 1996), p. 24. Fig. 3: Forecast of office equipment energy intensity. Energy use intensity is expected to decrease until about 2002, and to rise slowly thereafter. per square foot (32.3 W/m2), rather than the 8 to 10 watts per square foot (86.1 to 107.6 W/m2) specified.15 10. S. Silver, P. Fine, and F. Rose, “Performance Monitoring of DX Rooftop Cooling Equipment,” Energy Engineering Vol. 87 No. 5, 1990, pp. 32-41; L. Fryer, “Electric Chiller Buyer’s Guide,” E Source Tech Update TU-95-1, February 1995, p. 17. 11. J. Koomey, M. Cramer, M. Piette, and J. Eto, “Efficiency Improvements in U.S. Office Equipment: Expected Policy Impacts and Uncertainties,” LBNL, Berkeley, CA (December 1995), p. 10. References 12. Koomey et al. 1995, p. 10. 1. See, for example, M. Shepard et al., Commercial Space Cooling And Air Handling Technology Atlas, June 1995, E Source, Boulder, CO, p. 3.1. 13. Koomey et al. 1995. 2. California Code of Regulations, Title 24, Part 6, Subchapter 5, Section 144(a)9. This document is available at www.energy.ca.gov/ reports/title24. 3. 1993 ASHRAE Handbook—Fundamentals, p. 26.11. 4. Larry Lister, personal communication (May 1996), Construction Engineering Research Laboratories (CERL), U.S. Army Corps of Engineers, P.O. Box 9005, Champaign, IL 61821-9005. 5. G. Newsham and D. Tiller, “The Energy Consumption of Desktop Computers: Measurement and Savings Potential,” IEEE Transactions on Industry Applications, v. 30, no. 4 (July/August 1994), pp. 1065-1072; M. Piette, J. Eto, and J. Harris “Office Equipment Energy Use and Trends,” Lawrence Berkeley Laboratory, LBL-31308 (September 1991); R. Szydlowski and W. Chvála, “Energy Consumption of Personal Computer Workstations,” Battelle/Pacific Northwest Laboratory, PNL-9061 (February 1994; and C. Wilkins and N. 44 ASHRAE Journal 14. For more information, see 1993 ASHRAE Handbook—Fundamentals, chapter 24. 15. Scott Frank, personal communication (June 1996), Jaros Baum and Bolles Consulting Engineers, 345 Park Avenue, New York, NY 10154-0002. Please circle the appropriate number on the Reader Service Card at the back of the publication. Extremely Helpful ........................................................454 Helpful .......................................................................455 Somewhat Helpful .......................................................456 Not Helpful.................................................................457 December 1997