More efficiency is better than less, right? For air conditioning systems under five tons capacity (65,000 Btu/hr, actually) the efficiency figure of merit is the Seasonal Energy Efficiency Ratio, or SEER. If the minimum standard is 10, you want it to be 12, or 13.
But what does that number mean? Is bigger always better? The devil, it turns out, is in the details, and the details are in the method of testing. SEER is supposed to be an estimate of the energy the unit will use over a season of cooling. But wait a minute: If I live in Washington state, that energy is likely to be vastly different than if I live in South Carolina. Plus, the air conditioner has an extremely different function.
In South Carolina, A/C is likely valued more for dehumidification than for anything else. It's that sweltering humidity that drives discomfort. Residents of Washington state know that if humidity ever does appear, it will be way too cool for A/C. Temperatures hot enough to need cooling will come with dry air, thanks to the natural dehumidification of the cold Pacific Ocean. Moreover, Washington state residents will need at most a few hours of cooling at a time rather than the months of heavy use South Carolinians would likely demand. So how does one number for seasonal energy use apply in these cases and their limitless permutations?
It turns out that scientists at the National Institute for Standards and Technology (NIST) given the task to develop the rating in the 1980s took an overall population-weighted average of the temperature across the United States. Big states like Pennsylvania counted for a lot. It might be hotter in west Texas, but there weren't as many people there. Scientists looked at the average cooling temperatures and came up with 82°F. So that's where the SEER test is run.
If you reside in New Mexico, you probably think 82° is a perfectly comfortable temperature. Some New Mexicans don't even turn on the A/C until it's 95° outside. But remember South Carolina. And by the way, Pennsylvania is pretty humid in the summer too. So it averages out. Or does it?
Let's consider the energy used by that unit. Mostly it's consumed by the refrigerant compressor, but then there are those two fans. The condenser fan doesn't use too much juice. It might be rated at one-third horsepower, or one-half horsepower on a big unit. But things start to get tricky with the evaporator fan.
First, the evaporator fan is supplied with the furnace in many units, which often comes from a manufacturer other than that of the cooling system. For this situation, the testing method allows a default value of power to be used for that fan. It's 375 W per 1,000 ft3/min (CFM). Air conditioning units typically want about 350 CFM per ton of cooling, so in round numbers, 1,000 CFM is adequate, barely, for a three-ton (36,000 Btu/hour) system.
The problem is that field tests of real systems indicate that the fan energy is much higher, usually at least 500 W/1,000 CFM. Manufacturers can also test the fans they actually use to get their posted power rating. But why should they, it they can get away with using such a low default value?
Now we come to the second slightly ridiculous assumption in the SEER test: It assumes only a 0.1-in water column of duct and filter static pressure in small systems, and 0.15-in in the large, five ton systems. How realistic is that?
Testing of many systems with real ducts and filters shows external static pressures on the order of 0.6 in. of water column or more, even at the typically low-end air volumes of 300 to 350 CFM per ton of cooling. It's typically worse in the larger systems, as you might expect, which need more air flow.
These extremely low static pressure assumptions get us a fan that's under-designed for its intended application and a great under-emphasis on the amount of energy consumed by that fan/motor combination. Also consider the fact that people are increasingly installing high-efficiency filters which further increase the pressure drop. This exacerbates the situation. If the test accurately reflected field conditions, we'd get more efficient fans, perhaps at the expense of something else. But overall, the system would perform better under actual field conditions.
With today's computers and test equipment, it would be no huge difficulty to estimate energy used at actual temperature profiles of particular geographic locations. It should be possible to even apply a temperature-dependent valuation of electricity which, of course, is much more expensive and scarce when temperatures are high.
But we're also stuck with that 82° simplification and A/C that works great at 82° without any ductwork. It's a little like criticism of the ‘No Child Left Behind’ act in education which among other things, measures student performance on standardized tests. In both cases, you get what you test for, but the answer you get may not be what you really wanted. Go figure.