The first assumption is usually that one (or both) of the fans are out of calibration - but that is seldom the case.  Here is how this can happen. Let’s say the perfect after-result is 3500 CFM50. With a little bit of wind, the Baseline can easily vary by 1 Pa.  Subtract gauge and fan error of 1% and 5% respectively, and you get a total of about 10% variation!

This can cause the before-reading to be 3500 + 10% = 3850 CFM50 on the first after-test. Later, for the second after-test when conditions are different, the reading could be 3500 - 10% = 3150 CFM50.

Blower door error

If the gauge and fan calibration were checked, they both maybe approved as "in spec."  So what does this mean?  First, test fans will vary 5% either way when they have just been calibrated. They will fall off about 1% per year unless the tubes get blocked or pinched - then differences can be huge! Factory calibration checks of all blower doors are done under stable lab conditions - where the same fan would typically vary 1 or 2% when a calibration check is repeated.  Fans will typically have errors around 1 or 2% over most of their range, but almost all will approach 5% error in one part of their range.

Wind error

Even in the best situation, using exactly the same fan and the same operator, we can get differences of 5% on a good day, 10% on an average day, and 20% on a really windy day.

Fan set up and testing methods can make a difference, too. 


If the operator pulls the cloth on really tight, when the pressure comes on it may lift the cloth clear of the fan causing the leakage reading to be high.  A quick check to see if the cloth is still in contact with the fan will remove this source of error.  If windows and doors are not tightly closed that will make a difference, but even if one extra interior door is closed, that can change the result by 5 to 10%. The same goes for the damper position for intake air, a small exhaust fan running, or even a hot flue that will pump air out of the house.

Test pressure

If the operator does not achieve the correct test pressure or does not allow the reading to be averaged for at least 30 seconds, we can add another 2 or 3% error.  Operators may not wait for readings to stabilize, simply using the @ extrapolation function to get results at 50 Pascals.

What does this mean in real life testing?

Well, most researchers assume that readings will vary 10% either way unless extreme measures are taken. So, if we want accurate before and after tests to evaluate the effectiveness of airsealing efforts, we must do it with the same fan and under the same conditions.  Even then, expect these readings to vary 5%. The real point is to use the % reduction as a tool for improving airsealing effectiveness. Good crews can get 50% reductions, but only after understanding how the house was put together and where the most likely locations of leaks are.  In many cases a 15% reduction can be achieved by tackling plumbing stacks alone, whereas days of sealing smaller leaks may result in immeasurable differences.

Here are real, random calibrations from two different manufacturers.

X Company Blower Door:

Y Company Blower Door: