The mixing equation
A cooling system full of mixed coolant is a single well-stirred volume. Take some out, put the same amount of something else in, and the new concentration is a weighted average. Rearranged to answer the question people actually ask — how much do I take out — it comes to capacity multiplied by the gap between target and current, divided by the gap between what you are adding and the current concentration.
Worked through: a twelve quart system testing at thirty percent, target fifty, refilled with full-strength concentrate. The gap to the target is twenty, the gap between concentrate and current is seventy, so the drain is twelve times twenty over seventy, or about 3.4 quarts. Note what that says: you do not need to drain half the system to move it twenty points, because what goes back in is twice as strong as the target.
The same equation runs backwards. A system that tests at seventy percent, target fifty, refilled with distilled water: twelve times minus twenty over minus seventy, again 3.4 quarts. Over-concentrated systems are more common than people expect, usually from years of topping up with concentrate.
Why one drain rarely does it
The equation assumes you can remove exactly the volume it asks for. Frequently you cannot, in either direction. A radiator petcock on many vehicles drains barely half the system — the block, the heater core and the hoses keep the rest, and it stays there mixing with whatever you pour in. When the drain the arithmetic asks for is larger than what the system will actually give up, you are into repeated partial exchanges.
Each one moves the concentration by the same fraction of the remaining distance. Drain fifty-five percent of a twelve quart system and refill with concentrate, and the first round takes thirty percent up to about sixty-eight. Do it again and it goes higher still. The sequence converges towards the concentration of what you are adding and touches it only in the limit, never in practice. That is a real property of the process, not a limitation of the calculator, and it is why shops use a flush machine that exchanges continuously rather than doing it in rounds.
Freeze protection against concentration
The relationship is not linear and the last twenty percent buys most of the protection. Approximate figures for ethylene glycol, which are close enough for deciding whether you have a problem and not close enough to be a specification:
| Glycol by volume | Freeze protection, roughly |
|---|---|
| 20% | +16 F |
| 30% | +4 F |
| 40% | -12 F |
| 50% | -34 F |
| 60% | -62 F |
Past about sixty percent the curve turns around and the freeze point rises again, while the mixture's ability to move heat keeps falling. Straight concentrate is worse in both respects than a proper mix, which surprises people who reason that if some is good, more is better. Test with a refractometer rather than a floating-ball hydrometer; the cheap ones are temperature sensitive and read differently on a cold morning than on a warm afternoon.
What the calculator does not know about your coolant
It knows volumes and percentages. It does not know what chemistry belongs in your system, and that matters more than the ratio. Inorganic, organic acid and hybrid formulations use different inhibitor packages, they are specified per vehicle, and mixing incompatible types can gel or can strip the protection from aluminium and solder while the freeze point stays perfect. The manufacturer specifies the product. Nothing on this page is a substitute for that specification.
It also does not know why the concentration drifted. A system that keeps needing top-ups is losing coolant somewhere, and coolant that goes missing without a puddle underneath is going somewhere worse. Fix the reason before you fix the ratio. If you are budgeting fluids across a service, the fluid capacity and cost calculator covers the rest of them, and the winter driving checklist covers what else wants checking before the cold arrives.
Questions people ask
Can I just add concentrate to the reservoir until it tests right?
You can move the number that way, slowly and unpredictably. The reservoir is a small part of the system and what you pour in mixes with the rest over some number of heat cycles, so the reading you take an hour later is not the system concentration. It also only works in one direction: you can raise concentration by adding concentrate, but you cannot lower an over-concentrated system without taking fluid out. Draining a measured amount and replacing a measured amount is the only version of this with arithmetic behind it.
Why does my system test lower than the 50/50 I put in?
Several reasons, and they are worth telling apart. Pre-diluted coolant added to a system that still had water in it from a flush gives less than fifty. Topping up with plain water over a few years dilutes steadily. And a system that has been losing coolant and being topped with water drifts down every time. The last one is the one to care about, because it means there is a leak. Test first, then work out which story fits.
Does a partial drain and fill ever fully exchange the coolant?
No. Every partial exchange leaves behind the fraction of the old fluid that did not drain, and mixes it with what you added. Repeat it and the old fluid becomes a smaller and smaller share, approaching zero without reaching it. Three exchanges at fifty-five percent each leave roughly nine percent of the original fluid in the system; four leave about four percent. For a concentration target that is fine. For getting a contaminated or wrong-type coolant out, it is not, and that job needs a proper flush.
Is distilled water actually necessary, or is tap water fine?
Use distilled or deionised. Tap water carries dissolved minerals that deposit inside the radiator and the block as scale, and the harder the local water the faster that happens. Scale insulates exactly the surfaces that are supposed to be transferring heat, and it collects in the narrowest passages, which are the ones that were already the constraint. A few dollars of distilled water against a radiator is not a close call.
How do I know my real system capacity?
From the vehicle documentation, not from measuring what came out. What drains is only the part of the system that drains, which is the whole problem this page exists to work around. If the documented figure is not available, the arithmetic still works with an estimate, but the drain volume it gives you carries the same error your estimate does. Err towards a larger assumed capacity if you are guessing and you want to end up nearer the target from below rather than overshooting.