Much research in the past has focussed on how efficiently air conditioners produce cooling. However, most homeowners care about comfort and high bills, not about efficiency, duct leakage, infiltration and all of the other things that scientists usually talk about. This has prompted the Energy Performance of Buildings group to broaden our study of air conditioning and try and look at air conditioning from the perspective of the homeowner. The central question we are trying to address is how much useful cooling will an air conditioner provide.
There is much confusion about air conditioner capacity and duct efficiency
because there are many different terms that mean different things to different
people. There are two basic kinds of measures of air conditioner performance:
| Air Conditioner Capacity | These measures are intended to describe how much cooling an air conditioner provides. Some measures of this type account for an air conditioner that is not performing well because of problems such as low air handler flow or refrigerant charge. | |
| Duct efficiency | Duct efficiency measures describe how much of the cooling created by the air conditioner is lost through duct leakage and conduction losses. |
These two parameters can often combine to greatly diminish the useful
cooling capacity of an air conditioner. For example, a three ton air conditioner,
might only really be a two ton air conditioner because the nameplate capacity
is typically a gross overstatement of capacity and it may have inadequate
refrigerant charge and flow. Once it is connected to a duct system of 75%
efficiency, it effectively becomes a one and a half ton air conditioner
- even though the homeowner has paid for three tons of cooling. For this
reason, we have come up with a third type of measure of air conditioner
performance.
| Whole Cooling System | These measures combine duct efficiency and air conditioner capacity measures to get an idea of who much cooling actually goes to the space. |
The following table shows 7 common cooling performance measures. Air
conditioner capacity measures are highlighted in blue, duct efficiency
in green, and whole cooling system performance in red.
| Measure | Units | Meaning |
| Nameplate or Raw Capacity | Tons(1 Ton = 12000 BTU/hr = cooling capacity of 1 ton of ice) | This is what is stamped on the air conditioner. It is a considerable overestimate of how much cooling the air conditioner will actually provide. |
| ARI Capacity | kBTU/hr | This is the Air-Conditioning and Refrigeration Institute tested performance of the air conditioner at a particular set of temperature conditions. It assumes perfect refrigerant charge and airflow and an ideal fan, not the real air handler fan. |
| Energy Efficiency Ratio | kW-hour/kBTU | Tested/Simulated coefficient of performance of air conditioner. Higher is better. It has the same assumptions, and therefore limitations, as the ARI capacity. |
| Duct System Efficiency | % | Fraction of cooling energy created by the air conditioner actually delivered to the space (including cooling that is lost from the ducts but ultimately enters the space). |
| Duct Delivery Effectiveness | % | Fraction of cooling energy delivered to the space through the registers directly. Duct efficiency measures are described more completely at the EPB's ducts web page. |
| Tons at the Register (TAR) | Tons | Actual cooling delivered to the space through the registers. In houses with significant regain of duct losses (such as those with unvented attics) it is an underestimate of actual useful cooling energy. Future work will improve TAR to take this into account. |
| Pulldown Time | Hours | How long it takes for a hot house to cool down to a comfortable temperature.
Homeowners cite a fast pulldown time as an important aspect of air conditioner
performance.
![]() |
The Energy Performance of Buildings group has chosen to focus on the
cooling system performance measures. Combining information about Tons at
the Register and Pulldown Time with energy use and peak power demand allows
for a complete picture of cooling performance.
Last Modified: June 27, 2000