The Geo-Heat Center compiled case studies of about 80 metered residential ground source systems. About half of these were compared to similar houses with conventional systems, and half were compared with simulations of conventional systems. There were also different methods of measuring efficiency. While some case studies monitored energy use of the heat pump systems, others monitored heat pump coefficients of performance (COPs) and estimated COPs of the conventional systems. COP is the ratio of heating energy delivered to electric energy consumed. Electric resistance heat has a COP of 1; a heat pump with a COP of 3 will heat the space three times as much as a resistance heater with the same amount of energy. The Air Conditioning and Refrigeration Institute lists official COP ratings for all ground source heat pumps. Manufacturers often use these ratings to estimate customers' energy savings.

 The Geo-Heat Center found notable energy improvements from ground source heat pumps. Some of the studies found that these systems cost 54% less to run than a combination of electric resistance heat and forced-air AC and 18% less than a natural gas furnace with forced air AC (see Table 1). Many case studies measured the heat pump's COP, with results ranging from 1.8 to 5.7. Most units' measured COPs were between 2 and 3, which is reasonably consistent with the ARI ratings. Simple paybacks from replacing air source heat pumps ranged from 2 to 9 years with an average of 6 years, while paybacks from replacing natural gas/AC systems ranged from 4 to 24 years, with an average of 12 years. 

Table 1. Heat Pump Operating Cost Savings
Operating Savings
Compared to:	Site energy	Dollars	Simple payback
Elec. resistance/AC	57%	54%	4 years
Air source heat pump	31%	31%	6 years
Gas furnace/AC	67%	18%	12 years
Oil furnace/AC	71%	33%	4 years
Other (propane, unspecified)	46%	39%	4 years
Source: Geo-Heat Center

The Geo-Heat Center also calculated the break-even electric rate. If a homeowner is able to finance the system, and the price of electricity is above the break-even value, then the utility bill plus the loan payment will be less than if a less efficient HVAC system were used. The National Rural Electric Cooperative Association and the University of Alabama surveyed 285 companies and individuals associated with ground source heat pumps to find out how much such systems cost. Using this data, the Geo-Heat Center calculated the break-even rate for a well-insulated home with a 2.5 ton ground source, rather than air source, heat pump. With a 30-year fixed-rate mortgage at 8%, the break-even electric rate is about 9¢/kWh in a climate with 4,700 heating degree-days (HDDs), like that of Kansas City. In a climate with 6,500 HDDs, like that of Omaha, the break-even rate is about 6¢/kWh. Vertical loops generally require slightly higher electric rates than horizontal loops in order to be cost-effective. 

A problem with the Geo-Heat Center's report is that it can be misleading to calculate annual energy savings from measured COP. COP measurement equipment and calculation methods were not consistent between case studies. Calculations may or may not include the energy used for electric resistance backup, the desuperheater, the fan, and the loop pump. Furthermore, COP indicates only how much heat is produced per unit of electrical energy input, not how much heat is demanded. When demand increases, savings are lost. Duct losses, for example, make residents demand more heat from forced-air heat pump systems than from electric resistance baseboard systems. Heat pumps are also harder to zone control than baseboard heaters, resulting in greater heating demand. Finally, occupants may erode expected savings by opting for more comfortable summer or winter setpoints than they would with another system-the takeback effect.

 In a separate compilation, researchers in the Building Energy Analysis Group at LBNL attempted to demonstrate the problems with predicting savings based on COP and to characterize actual energy savings. They compiled measured COP and actual energy use data from nine case studies on 70 homes with ground source heat pumps. Only two case studies were used by both LBNL and Geo-Heat Center. For energy data on conventional systems, the studies examined either houses of similar size and construction or the heat pump homes themselves before being retrofitted. In most cases, only whole-house energy use data-utility bills-were available, although a few houses were submetered so that HVAC energy data could be compared. In some cases, total energy savings could not be easily measured, due to changes in wood heating. Table 2 shows what the HVAC and whole house energy savings should have been, based on measured COPs, and what the actual savings were.

The COPs measured in the LBNL and the Geo-Heat Center case studies are similar to the ARI-rated COPs, but the LBNL results show that the actual savings are often considerably less than predicted by COP measurements, and the actual savings can be surprisingly low. For example, a house in Alaska replaced an electric resistance heater with a ground source heat pump. The COP was measured at 4.0 including desuperheater contribution, or 3.3 without desuperheater contribution. Thus, heating plus hot water energy should have declined by about 75%. Considering that heat and hot water accounted for about 50% of the home's annual electric bill, total electric use should have declined by about 40%. However, actual electric bills only went down about 28%. In a Montana case study, 28 homes with heat pumps were found to use about the same heating energy as 40 similarly built homes with electric resistance heat, even though the heat pumps had average measured COPs of 2.5. The actual savings were highly variable; some systems performed as well or better than expected, others were very disappointing.

-Jeff Ross Stein

Jeff Ross Stein is a graduate student research assistant in the Building Energy Analysis Group at Lawrence Berkeley National Laboratory.

Table 2. Heat Pump Savings: Predicted vs. Actual
		Measured COP	HVAC Site Energy Savings		Total Site Energy Savings
Location	Compared to		Predicted[1]	Actual	Predicted	Actual
Ft. Hood, TX	Air conditioner			50% (cooling)
Selfridge, MI	Electric resistance		60%[2]	35% (heating)
Montana	Electric resistance	2.5	60%	0%
Anchorage, AK	Wood, electric	3.89	75%		40%	-27% (electricity only)
Anchorage, AK	Electric resistance	3.3-4.0	75%		40%	28%
Ithaca, NY	Electric resistance	2.61	62%	68%	38%	42%
Oklahoma	Electric furnace					35%
Kentucky	Air source heat pump					32%
Virginia	Air source heat pump (COP=2.5)	3.2-5.0	37%	27% (7% cooling, 34% heating)
Ithaca, NY	Air source heat pump	2.61	44%	48%	21%	24%
Hyde Park, NY	Air source heat pump	2.61	44%	45%	15%	14%
Oklahoma	Air source heat pump					22%
Oklahoma	Air source heat pump					4%
Kentucky	Gas/AC					36%[3]
Alabama	Gas/AC	3.3		30% (including hot water)	20% (rated savings)	18%[4]
1. Predicted based on COP. 2. Manufacturer's prediction. 3. Dollar savings only. Electricity ~4¢/kWh for "all-electric" house with heat pump, ~5¢ for house with gas heat. Gas ~50¢/therm. 4. Dollar savings only. Electricity ~7.3¢/kWh, Gas 77¢/therm. Source: LBNL