The relation, and where it comes from
Air removes heat by getting warmer and leaving. How much heat a given stream of air can carry depends on how much of it there is and how much warmer it is allowed to get, and the constant that links them is the heat capacity of air. In the imperial units fans are sold in, the sensible-heat relation is that heat in BTU per hour equals 1.08 multiplied by airflow in CFM multiplied by the temperature rise in degrees Fahrenheit. Since a watt is 3.412 BTU per hour, the version that matters here is that watts equal CFM times the rise in degrees Fahrenheit times 1.08 divided by 3.412 — about a third of a watt per CFM per degree.
Worked through: 450 watts held to a 15 degree Fahrenheit rise needs 450 times 3.412, divided by 1.08 times 15, which is about 95 CFM of air genuinely moving through the case. In metric the same relation is roughly 0.33 watts per cubic metre per hour per degree Celsius, which is the figure to use if your fans are rated in cubic metres per hour rather than cubic feet per minute.
Two things fall out of this immediately. Airflow and temperature rise trade against each other exactly — halving the allowed rise doubles the air required. And nothing in the relation cares how many fans there are, only how much air actually passes through the enclosure.
Rated airflow and delivered airflow are different numbers
A fan's CFM rating is measured in free air, spinning in open space with nothing in front of it or behind it. That number is real and it is also the maximum the fan will ever achieve. Put it behind a dust filter, in front of a mesh panel, with a radiator and a bundle of cables and a drive cage in the path, and the fan is now working against static pressure. Its output falls along its pressure curve, sometimes to half the rating or less.
| What is in the path | Effect on delivered flow |
|---|---|
| Open panel, no filter | Closest to the rating, and the dustiest option |
| Fine dust filter, clean | Noticeable loss even when new |
| Fine dust filter, six months uncleaned | Large and progressive, and invisible until temperatures climb |
| Radiator in the path | Substantial, and dependent on fin density |
| Cables and drive cages in the flow | Modest individually, cumulative in a crowded case |
This is why the calculator asks for a delivered share rather than assuming one. A clean, open, well-routed case might genuinely see seventy percent of rated flow; a filtered case with a thick radiator and a neglected filter can be under forty. If you have no idea, fifty-five is a reasonable middle assumption and the honest thing to do is treat the result as a range rather than a figure.
Through-flow, and why the smaller side wins
Air that goes in has to come out. If you have 180 CFM of intake and 60 CFM of exhaust, you do not have 180 CFM of cooling — you have whatever the enclosure can actually pass, and the restricted side dominates. The calculator takes the smaller of the two delivered figures as the through-flow, which is a simplification but errs in the right direction and makes the point that matters: adding fans to the side that is already generous buys very little.
The difference between the two sides is the pressure balance, and it decides where dust comes from rather than how well the case cools. With more intake than exhaust the case runs slightly positive, air escapes through every gap, and dust arrives mainly through the filters you can pull out and wash. With more exhaust than intake the case runs negative and pulls air in through every unfiltered seam, which is why negative-pressure builds accumulate dust in places you cannot reach. Neither arrangement is meaningfully better at cooling, which surprises people who have been told otherwise.
Air temperature is not component temperature
What this calculator models is the temperature of the air passing through the case. It says nothing directly about how hot any particular part runs, because heat from a component has to cross the die, the interface material, the heatsink and a boundary layer of slow-moving air before it reaches the stream at all. A part can sit forty or fifty degrees above the air around it and be entirely within specification.
What good case airflow does is set the starting point. Every degree the internal air sits above the room is a degree added to every component temperature in the machine, so a case that is 25 degrees hotter inside than out has handicapped every heatsink in it before they start. Getting the air right is a precondition for component cooling rather than a replacement for it — sizing an individual heatsink for a specific dissipation is a different calculation, handled by the heat sink calculator.
Fan speed, noise, and the trade worth making
Airflow scales roughly in proportion to fan speed, so running a fan at 70 percent gives about 70 percent of the flow. Noise does not behave anything like that — it rises far faster than speed, which means the last 30 percent of airflow costs a disproportionate share of the noise. This is the whole argument for more fans turning slowly rather than fewer turning fast: the same total flow at a fraction of the noise, at the cost of fan mounts and money.
The other half of the trade is that the target temperature rise is yours to set. Allowing a 20 degree rise instead of a 10 degree one halves the air required and takes the fans well down their noise curve, and whether that matters depends entirely on whether anything in the machine is close to its limits. Work out the heat load first with the power supply wattage calculator, since that number is the input to everything here. If the enclosure in question is a closed AV cabinet rather than a computer case, the AV rack power and heat calculator handles the same physics with different assumptions, and for whole-room ventilation rather than a box, the air changes calculator is the right tool.
Questions people ask
How many case fans do I actually need?
The question the physics answers is how much airflow, not how many fans. For a 450 watt load held to a 15 degree Fahrenheit rise, that is around 95 CFM of real through-flow, and how many fans it takes depends entirely on what each one delivers behind your particular filters and panels. Three fans rated at 60 CFM each sounds like 180 and might deliver 95 after restrictions, which is why the delivered share matters more than the fan count. Two well-placed fans in an open case can beat six in a restricted one.
Is positive or negative pressure better for cooling?
Neither, to a first approximation. Total through-flow determines cooling, and the balance between intake and exhaust determines where the air that leaks comes from. Positive pressure means air escapes through gaps and dust enters mainly through filters you can clean, which is the practical argument for it. Negative pressure pulls unfiltered air in through every seam. Some people report marginally better results from a slight negative balance in specific cases, and the effect is small enough to be dominated by whether the filters are clean.
Do I add the power supply wattage to the heat load?
Only if the supply draws its air from inside the case. Most modern enclosures put the supply in its own chamber with its own intake and exhaust, so its losses leave without passing through the main compartment and should not be counted. If the supply is in the main compartment and vents through it, its losses do heat the case air and belong in the total. Either way it is the losses that matter, not the supply rating, and the losses are the wall draw minus the DC draw.
Will a bigger fan always move more air?
For the same rotational speed, a larger fan moves more air and makes less noise doing it, which is why case fans have grown over time. But the rating on the box is free-air flow, and a large fan with a low static pressure rating can be outperformed behind a dense radiator by a smaller one designed for pressure. Match the fan to the restriction: open panels reward high-flow designs, radiators and thick filters reward high-pressure ones. A fan chosen on its CFM number alone will disappoint behind anything restrictive.
Does altitude really matter for case cooling?
It does, and it is one of the few environmental factors people never consider. Air density falls with altitude, so the same volume of air carries proportionally less heat. At around 5,000 feet the air is roughly 85 percent as dense as at sea level, so a given CFM removes about 85 percent as much heat and the temperature rise goes up accordingly. It is a modest effect compared with a dirty filter, but it is real, it is permanent, and it stacks with everything else rather than replacing it.