How to Think About Glycol Concentration

Why this matters

Two shops on the same street will charge the same rooftop unit to two different concentrations, and both will describe their number as "what we always use." One of them will be protecting to a temperature the equipment never sees, paying for it in capacity and pump power every hour of the year. The other will be short on the one night a decade that matters, and the failure will not be a slow one: a split coil in an outdoor air stream floods a roof and takes the building down.

The number is not a habit. It comes out of three questions asked in order, and the answers get written down so the next person does not have to guess what you were protecting against.

Before you drain or recharge anything in winter

Draining a loop that serves an outdoor air coil creates, for the duration of the work, the exact failure you are trying to prevent. Schedule the conversion outside the freezing season where you can. Where you cannot, close and physically prove the outdoor air damper closed and lock the supply fan off before the first gallon leaves, so no cold air moves across an empty coil.

Locking that fan off is electrical work. Open the disconnect, lock and tag it, and prove the circuit dead with an instrument checked live before and after on a known source, per NFPA 70E-2021, 120.5. The general industry duty to de-energize and verify for electrical work sits at 29 CFR 1910.333(b)(2); on a construction site the electrical lockout and tagging counterpart is 29 CFR 1926.417.

Read Section 8 of the glycol product's safety data sheet for eye protection and glove class. Ethylene glycol is acutely toxic if swallowed and it tastes sweet, so contain spills where they happen and never leave an open container on a roof or in a plant room.

The drained fluid is waste, and several hundred gallons of it. Arrange the container and the licensed handler before you open the drain, not after. Nothing goes to a storm drain or to ground.

Any temporary connection made to a potable supply to mix or top up is a cross connection, and glycol reaching a drinking water fixture is a public health event. The backflow assembly required is set by the local plumbing code and the water purveyor.

The record

One page per loop, kept with the system. Every field on it is a question somebody will otherwise re-answer from memory in five years.

The same page, filled in for a rooftop unit

Loop and sections. Hydronic heating loop, 600 gallons, serving indoor terminal units plus one rooftop makeup air heating coil and about 60 ft of exposed roof piping.

Exposed components. The makeup air coil, sitting in a 100% outdoor air stream. The roof piping, partly insulated with some sections bare at the supports.

Lowest metal temperature, and how we got it. Not the design outdoor temperature. The winter design temperature is a statistical value that is exceeded by definition, so the starting point is the record low from a recognised climate data source for the location, not the design figure. Then a margin on top of it for two effects the air temperature does not capture: a bare pipe on a clear calm night radiates to the sky and can sit below ambient, and a stagnation event, meaning a pump failure or a control fault with the fan still running, puts the coil at air temperature within minutes rather than degrees above it. With a record low near minus 10 F and a margin for those two, the number on the page is minus 20 F.

Freeze or burst. The coil gets freeze protection. The roof piping, which carries no load and would only need to survive, could take burst protection on its own. Since it is one loop, the higher requirement wins and the whole charge is set by the coil.

Fluid type. Ethylene, because there is no food contact or potable adjacency here, and it carries the smaller penalty in both heat transfer and pumping. Propylene is the choice wherever a leak could reach food, potable water or an occupied space at exposure levels, and it needs a higher percentage for the same protection.

Concentration. Published freeze curves put minus 20 F for ethylene glycol somewhere in the mid forties percent by volume, and propylene noticeably higher; take the number off the chart for the actual inhibited product, because the inhibitor package moves the curve. On this page: 45% by volume.

Constraints checked. At 45% the heat capacity and pumping penalties are real and a sibling article covers how to quantify them; the coil selection is rerun at the actual fluid rather than derated by rule of thumb. Above roughly 60% by volume the freeze point stops falling and begins to rise again, so there is no case for going higher. And there is a floor, covered below.

Why the two protections are not the same number

Freeze point is where the first ice crystals appear. Burst point is the much lower temperature below which the mixture can split a pipe. The gap between them is real and it is often tens of degrees.

The mechanism is worth carrying because it explains the gap. As the first crystals form, they are nearly pure water, so the remaining liquid becomes more concentrated in glycol and its own freezing point drops further. The mixture never goes solid the way plain water does; it becomes a slush that still contains liquid and does not expand hard enough to split the pipe.

The consequences are different in a way that decides the choice. Freeze protection means the fluid keeps circulating and the equipment keeps working. Burst protection means the pipe survives and the loop is blocked. On a standby line or an idle outdoor run, blocked and intact is a perfectly good outcome. On a coil in a moving outdoor air stream it is not, because a blocked coil stops transferring heat while cold air keeps crossing it, and the section that fails next is whatever is downstream. Choosing burst protection for a working coil buys an intact coil inside a failed building.

Why you cannot get there by topping up

The loop above already held a 20% mix from an earlier attempt. The instinct is to drain 270 gallons and add 270 gallons of glycol, since 45% of 600 is 270.

Work it through. Draining a mixture removes glycol along with the water, proportionally. Taking 270 gallons out of a 20% mix removes 54 gallons of glycol and 216 of water, leaving 330 gallons in the loop made up of 66 glycol and 264 water. Add 270 gallons of neat glycol to that and you have 336 gallons of glycol in 600, which is 56%. Eleven percentage points past target, in the region where extra glycol buys almost no additional freeze depression and costs real capacity and pump power every hour for the rest of the system's life.

The correct charge starts from the target quantities. Six hundred gallons at 45% means 270 gallons of glycol and 330 of water. You already have 66 and 264. So you add 204 gallons of neat glycol and 66 gallons of water, which together are the 270 gallons of space you made. Take the dilution water quality from the fluid manufacturer rather than from the tap: inhibited packages set limits on chloride, sulfate and hardness and many specify deionised water, because hard or chloride-bearing makeup drops the inhibitor package out and spends the reserve alkalinity before the loop has run a season. Check it: 270 in 600 is 45%.

That arithmetic is the whole reason the volume field is on the page. Without a real system volume, none of it can be done, and topping up by feel on a partially charged loop is how a system ends up at 56% or at 28% with nobody able to say which.

The floor nobody expects

Concentration has a minimum as well as a maximum, and getting this wrong runs the opposite way to intuition. Inhibited glycol products carry a supplier stated minimum concentration, commonly somewhere around a quarter to a third by volume, below which the inhibitor package cannot hold its protection. Below that floor the fluid is not merely less protected than a proper mix; it can be more corrosive than plain treated water, because dilute glycol still oxidises to organic acids while there is no longer enough buffer in the charge to absorb them.

So a loop diluted by years of water top ups to 15% has bought itself almost no freeze protection and has acquired a corrosion problem it did not previously have. That is the reading that makes a low concentration result urgent rather than routine, and it is why a measured concentration well under the product's floor gets corrected or the fluid gets replaced, never left to drift further.

Verifying the page is true

Measure after charging and after circulation, not from the drum quantities. Quantities on the page are what you intended to add; a measurement is what is in the loop, and the two disagree often enough that the difference is the point of measuring. Sampling and instrument technique are a subject of their own and a sibling article covers them.

Then compare the measured concentration against the protection temperature on the page using the same chart you selected from. If the measurement lands at 41% against the 45% this page requires, look up what 41% actually protects to before deciding anything: on published ethylene curves that is somewhere near minus 14 F against a page that requires minus 20 F, so the charge is short and gets corrected rather than noted. on the steep part of the freeze curve a few percentage points move the protection temperature a long way, and on the flat part near the minimum they move it hardly at all. Write both the measured percentage and the temperature it corresponds to on the page, because the temperature is the thing anybody actually cares about and the percentage is only how you got there.

References

• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and locking out the fan circuit before draining a coil in an outdoor air stream
• NFPA 70E-2021, 120.5, for the live-dead-live instrument verification sequence used to prove that circuit dead
• 29 CFR 1910.1200, Hazard Communication, for the safety data sheet whose Section 8 sets glove class and eye protection for the glycol package
• Fluid manufacturer freeze and burst protection charts and the stated minimum concentration for the specific inhibited product
• See related: What Glycol Costs You in Heat Transfer and Pumping, which owns the capacity and pumping penalties referenced here