One division, and then everything else
Face velocity is airflow over media area. A single nominal 20 by 25 measures about 19.5 by 24.5, which is 3.32 square feet rather than the 3.47 the label implies, and 1,200 CFM through it is 362 fpm. Put a second one beside it and the same 1,200 CFM crosses 6.64 square feet at 181 fpm. Nothing else about the filter changed — same media, same pleat, same rating — and the pressure it costs fell by two thirds.
That two-thirds figure is for an exponent of 1.6. Media closer to linear saves less — at an exponent of 1.0, halving the velocity halves the drop and no more. The direction never changes, but the size of it comes off the curve for the filter in your hand.
That is the entire reason filter area is worth arguing about. The media does not know how much air the system moves; it only knows how fast the air is going through it.
A pressure drop with no velocity beside it means nothing
Published clean drops are quoted at a stated face velocity, and the two travel together. A filter listed at 0.25 in wg is listed at 0.25 in wg at, say, 300 fpm. Run it at 362 and the drop climbs — with an exponent of 1.6, by a factor of 1.35, to 0.34. Run the same filter at 181 fpm and it falls to 0.11. Comparing two filters on their published drops without checking the velocity each was quoted at compares nothing at all.
The exponent itself is worth a minute with the manufacturer curve. Take two points off it, divide the drops, divide the velocities, and the exponent is the log of the first over the log of the second. Depth media in laminar flow comes out near 1; a restrictive pleat comes out closer to 2. The default here is a placeholder, not a property of filters.
The projection runs long, on purpose
Two gauge readings and the hours between them give a loading rate, and the page draws a straight line from there to whatever threshold you entered. Real filters do not load in a straight line. The drop rises gently while dust builds in the depth of the media and then accelerates as the surface blinds, so a straight line always predicts a later change-out than the filter reaches. Read the projection as the optimistic end of a range and take a third reading partway through, at which point the curve declares itself.
Blower run hours matter more than the calendar here, and they are the thing nobody records. A system left on continuous fan is accumulating hours around the clock; the same house on auto fan in a mild month might run four hours a day. Two houses with identical filters and identical dust can be a factor of five apart on filter life for that reason alone.
What the page will not tell you
It will not say your face velocity is too high, that your drop is too much, or that the filter is due. Those depend on what the blower can carry, what the equipment manufacturer states for that unit, and what is in the air in that building — a house with a dog and a gravel driveway is not the house next door. The threshold field exists so the number comes from you or from the equipment data rather than from a page that has never seen your system.
One thing worth naming rather than calculating: a filter loading up reduces airflow, and reduced airflow changes conditions across the heat exchanger and the coil. What that means for a particular appliance is in the appliance data. And anything that changes the pressure a mechanical room sits at can change how a naturally drafted appliance vents, which is a safety matter and not a comfort one.
Questions people ask
What is filter face velocity?
The airflow divided by the area of media facing the airstream. A nominal 20 by 25 filter actually measures about 19.5 by 24.5, giving 3.32 square feet, so 1,200 CFM through one is 362 feet per minute. It is the number that governs the pressure the filter costs and how quickly it loads, and it is fixed by the opening rather than by the filter you buy.
How much does a second filter change things?
At the same airflow, two filters side by side halve the face velocity. What that does to the drop depends on the exponent the media follows: at 1.6 the drop goes to about a third, at 2 it goes to a quarter, and at 1.0 it simply halves. The saving is larger than the velocity change whenever the exponent is above 1, which most pleated media is, but it is worth checking against the curve for the filter you are actually buying. The cost is space in the return and one more thing to remember to change.
Why does the calculator ask what velocity the rating was taken at?
Because a published clean pressure drop is meaningless without it. The same filter quoted at 0.25 in wg at 300 fpm reads 0.34 at 362 fpm and 0.11 at 181. Comparing two products on their listed drops without checking the velocities behind them is comparing two different tests.
How do I work out the exponent for my filter?
Take two points off the manufacturer pressure drop curve. Divide the higher drop by the lower, divide the higher velocity by the lower, and the exponent is the natural log of the first ratio over the natural log of the second. Deep media comes out near 1 and a restrictive pleat closer to 2. If you have no curve, run the page at 1.3 and again at 2 and see whether the answer changes anything you would do.
Is a straight-line projection to the change-out pressure reliable?
It is optimistic. Filters load gently at first and then accelerate as the surface blinds over, so a line drawn through two early readings predicts a later change than the filter reaches. Take a third reading partway and the curvature shows. Treat the projected date as the far end of a range rather than a date.