Blog
Why Sizing Dehumidifiers is Hard – GreenBuildingAdvisor
I’ve been thinking a lot about humidity. I live in a very humid place, Washington, DC. It’s very common for people here to need a dehumidifier in addition to their air conditioning. I’ve been thinking about the question of how one goes about sizing a dehumidifier.
Modeling a house’s heating and cooling loads through the Manual J process has become routine. It would be natural to wonder if an analogous process could be used for sizing a dehumidifier. It turns out that the answer is probably no, and it has to do with the nature of modeling. In economics, they like to say, “all models are wrong, some are useful.” One of the things that makes a model useful is if it is relatively invariant to the inputs, which is a fancy way of saying that small variations in the numbers you plug into the model don’t change the output that much. This will mean that errors in measurement won’t change the final conclusion. The problem with modeling humidity removal in the same way that heating and cooling are modeled is that you end up with outputs that are overly sensitive to your assumptions.
To size an air conditioner you use a process called Manual J, where you estimate the cooling load based on the construction of a building and the local climate. You look at weather records (available at https://ashrae-meteo.info/v3.0/index.php) and calculate the 99th percentile of hourly temperature records, and then you estimate the cooling load at that temperature. When calculating cooling loads, you look at everything that would be putting heat into the building: solar gain, conduction of heat through the walls, occupant activity, air infiltration, and any appliances and equipment being operated. These are called sensible loads. You also look at sources of humidity, which will have to be removed, which requires cooling as well. The sources of humidity are occupant behavior like bathing, cooking, breathing and sweating, and infiltration of humid outdoor air.
Just to give an idea of the magnitude of these components, here is the summary of a Manual J for a 4-bedroom house in Washington, DC:
(in BTU/hr)
Sensible gains
Solar gains 13,939
Conductive gains 10,448
Occupant activity (5 people) 1,150
Infiltration, 20 CFM 367
Internal 2,400
Latent gains:
Occupant activity 1,000
Infiltration 531
total Sensible 28,304
Total Latent 1,531
Grand total 29,835
Sensible heat ratio: 95%
The “sensible heat ratio” or SHR is how much of the total cooling load comes from sensible loads as opposed to dehumidification. The lower that number, the more dehumidification you need, a SHR of 100% means you don’t need any dehumidification at all.
Now to look at a similar calculation for dehumidification. The ASHRAE database also has 99th percentile information for humidity. In DC the cooling design day is 92F with 45% relative humidity, but the humidity design day is 82F with 77% relative humidity, a dew point of 74F.
In general, the more sensible gain you have the better job the air conditioner does at removing humidity, so let’s assume that the humidity design day is going to be a cloudy day with no solar gain.
Running those numbers for the humidity design day with the same assumptions, and the conductive gains adjusted for the lower outdoor temperature:
Sensible loads:
Solar gains –
Conductive gains 4,302
Occupant activity 1,150
infiltration 151
Internal gains 2,400
Latent loads:
occupant 1,000
infiltration 855
total sensible 8,003
Total Latent 1,855
Grand total 9,859
SHR 81%
At first glance it would seem that modeling humidifier would be a simple matter of figuring out what SHR your air conditioner is capable of, and then adding a dehumidifier that can make up any difference. Air conditioners have a SHR rating, right? But this is where the instability in the model starts being a problem.
The first source of instability in this model is the amount of infiltration, which is at best an estimate informed by blower door testing and at worst just an outright guess. In the cooling load model, changing the infiltration from 20 CFM to, say, 50 CFM, doesn’t really change the picture. The total cooling load goes from 29,835 to 31,182 and the SHR goes from 95% to 91%. But in the dehumidification model you see much bigger proportional changes: the total load goes from 9,859 to 11,369 and the SHR goes from 81% to 72%.
To put those numbers into context, at 20CFM the latent load is 1,855 BTU/hr. Dehumidifiers are commonly measured in pints per day, a pint of dehumidification is about 1000 BTU of cooling. So that’s 44,500 BTU/day, or 44.5 pints per day, or about 5.6 gallons per day. At 50CFM it’s 9.4 gallons per day. (Did I say it’s really humid here?) But the point is a rather miniscule change in assumptions – or occupant behavior – brings in gallons a day of moisture that needs to be removed.
The other input that is very sensitive to assumptions is the amount of dehumidification provided by the air conditioning. Yes, it’s true that air conditioners have SHR ratings, but they are only useful for comparing equipment, they have no predictive power. What does have predictive power is the outlet temperature of the air conditioner. If air is cooled below its dew point, moisture will precipitate out. If you know the outlet temperature of an air conditioner, you can predict how much dehumidification it can do. Indoor air at 75F and 50% RH is at a dew point of just under 55F. If the output of the AC is at 55F, no dehumidification happens, the SHR is 100%. Lowering the temperature by one degree, at 54F the SHR is 90%, and lowering by another degree, at 53F it’s 86%.
So tiny changes in the assumptions have a big impact on the output. The difference between 55F and 53F coming out of an air conditioner is literally the difference between a clean filter and a dirty filter.
So what’s the answer? I don’t know. I’m curious to hear everyone’s thoughts.