Accuracy and Tolerances in Humidity Measurement in a Drying Kiln

Super precision is rarely critical in wood drying, but reasonable accuracy is helpful. Here's a discussion of RH measuring precision, and its importance for maintaining proper real-world kiln conditions. July 29, 2012

From Professor Gene Wengert, forum technical advisor:
Letís assume that the 100% perfect temperatures in a kiln are 110.0 and 106.0 F. This is a 4.0 degree depression and so the RH = 87% .

Now assume the temperatures change to 110.4 and 105.6 F. On a 1 degree resolution thermometer instrument, you will still get a reading of 110 and 106, so the RH and EMC you measure will be the same at 110.4 and 105.6 as at 110.0 and 106.0. However, with a more accurate instrument, you will find that the new conditions are actually RH = 84%. (Stated another way, a DB = 110 means that the actual temperature could be anywhere between 109.5 to 110.5 F.)

We can repeat this same idea when the actual temperatures are 109.6 and 106.4. The measured RH with the 1 degree instrument that still shows 110 and 106 is 87% RH, but the true value is 90% RH.

So assuming that the thermometers are perfectly matched to each other (but this is not possible as one can only read them to 1 degree, or maybe 1/2 degree; a matched pair of thermometers that are used by the weather bureau for observations and will also be calibrated to a secondary standard will run over $500 today... closer to $750), the accuracy of a DB WB hygrometer instrument is plus or minus 3% RH.

Of course, there also are air flow issues (600 fpm across the WB) and distilled water issues and wick issues that all affect the accuracy of the WB reading itself.

Bottom line: Any instrument is fine, but appreciate the uncertainty that it has to have.

A manufacturer of a dry kiln controlled, if they want, can calibrate the instrument to read much more accurately that 1 F and measure RH closer than 2% RH. The better it is, the more expensive it is.

Forum Responses
(Sawing and Drying Forum)
From contributor K:
Temperature calibration is incredibly picky at best. The equipment required is expensive. Mercury bulb types are good to the nearest degree unless it is an extended range and those can be a couple of feet long.

A very good self-calibration check is to make up a distilled water ice bath (shaved ice and water to make a slush from top to bottom in an insulated jug). There you will have 0 degrees C or 32 degrees F. Second point would be a boiling distilled water bath at sea level; you would have 100 degrees C of 212 degrees F. Above sea level would have to be compensated for barometric pressure. In between is as good as the scale on the thermometer.

NIST will give you a calibration curve good for + - 0.3 degrees C, but it comes with a cost that will knock your socks off - around $400 per calibration point.

From Professor Gene Wengert, forum technical advisor:
Close and accurate measurement of db/wb or RH is necessary when drying a few species, like oak, green from the saw in a kiln. As the RH drops with lower MCs, the possible errors in RH are not as serious.

Perhaps the following will help in understanding. Let's assume that you are watching the Biggest Loser on TV tonight. A contestant steps on the scale and the weight given is 245 pounds. Because they are measuring to the closest one pound, a reading of 245 pounds actually means that the weight is between 244.5 and 245.5 pounds. Without more resolution, we just do not know.

Next week the contestant weighs 240 pounds. Again, the weight could be 239.5 to 240.5 pounds. The TV show calculates the weight loss as 5 pounds, but in truth, the weight loss could be anywhere between 4.0 pounds and 6.0 pounds (244.5 to 240.5 or 245.5 to 239.5). We just do not know and within this two pound range, one weight is just as likely as another weight.

Note that because we use the same scale both weeks and the chance that the calibration of the scale changes over a week's time is very unlikely, we do not have to worry about calibration. (Similarly, if we have a matched DB and WB, if both are off the same amount, it is really not an issue for us.)

Note: This is also the reason we want to have the MC of samples calculated to the 1/10 of a percent and then those values used to calculate the daily loss.

Now if you watch the TV show, you know that they calculate the percent weight loss. So, when we calculate all the possible percentage weight losses, we find that (because the scale is only a one pound resolution) the percentage weight loss in the example above could be anywhere between 1.6% to 2.4%.

From Professor Gene Wengert, forum technical advisor:
I should have added: This example also shows why we need to measure our moisture sections to at least 1/10 gram (four digits) and our kiln samples to at least four digits (unless precise MC is not necessary).

From contributor K:
Geneís examples are a very good lesson in tolerances and uncertainty canít argue with any of it. Cost is a big factor.

Example 1. You are going to do a calibration check on the two thermometers in your shop. You go to the drug store and buy a gallon of distilled water and proceed to do a check like I described before. Your cost, maybe less than ten dollars and a couple of hours.

Example 2. Send your 2 thermometers off to NIST for the same 2 points. 2 thermocouples, 2 points each, $400 a point, total $1600. Are the results any better? Not really, but now you have proof in the paperwork that the temperatures were what they were when the technician at NIST read them.

The point is you can drive yourself nuts worrying about details. A man with 1 watch always knows what time it is. A man with 2 is never sure.

From contributor D:
While I understand your point on the accuracy of measuring so fine, I question the reality of it under real kiln conditions. I doubt anyone can hold that fine of an accuracy through the entire kiln. Anyone who dries lumber for a living has the reality of averaging things out.

I can measure wood to ten thousandths of an inch, but that doesn't mean my moulder will produce or hold that kind of accuracy over a 10,000 ft run, any more than expect all lumber out of my kiln to be exactly 6 percent.

From Professor Gene Wengert, forum technical advisor:
As mentioned above, close tolerances are needed in only a few cases. On the other hand, if we are not fairly accurate or close in DB and WB, and we are not close in weighing and so on, eventually we will be sloppy enough and do a poorer job than reasonable or desired. So let's be as close as reasonably possible in all measurements and let's recognize the uncertainty in some measurements.

Incidentally, wood quality has changed over the years, as we are drying more lower grade (more prone to warp, checks, etc.) and kilns are more powerful today, so I believe more accurate or precise measurements are becoming increasingly important.