Recently, I was working on an energy model for a high-performance housing prototype. Mostly, I apply energy modeling to non-residential structures. I did not have a coefficient of performance (COP) at my fingertips for the air-source heat pump (ASHP) used in my baseline model.

In this context, the COP is the ratio of the amount of thermal energy (heat) delivered by a heat pump to the amount of energy expended. COP = heat delivered / energy expended.

The COP is kind of like gauging the efficiency of a piece of equipment. However, heat pumps can be over 100 percent. It is possible for one unit of energy input to result in over one unit of heat being “pumped” from one environment to another.

The reconnaissance mission reminded me of an important performance consideration regarding ASHPs that a significant amount of the U.S. is experiencing during this arctic February: the COP decreases as the outdoor air temperature drops. For residences relying on ASHPs for heating—as the outdoor air temperature drops well below freezing—they will start to see the dreaded “auxiliary” heat kicking on (indicated by AUX on many thermostats).

The ASHP is still “pumping” heat out of the frigid outdoor air. It might seem counterintuitive, but there is still heat in sub-freezing air. However, the ASHP’s capacity is quite low at such cold temperatures. At some point (usually near the freezing temperature), the heat pump’s output is eclipsed by the home’s heating requirement. For many residences, an inefficient auxiliary heat coil is activated—which has a COP of approximately 1.0. This is not good, considering the ASHP has a COP around 2.7 near the balance temperature.

The complication for energy modeling is that the sliding COP of the air-source heat pumps, contingent on the outdoor air temperature, should be taken into consideration for the most accurate results. If only a single COP value can be used, assume an efficient electric air-source heat pump will have a COP between 2.3 and 2.7.

By comparison, the COP of a ground-source heat pump (GSHP)—which our team is considering for the proposed design—may range between 2.7 and 3.5.