Why Glycol Degrades and What It Turns Into

Why this matters

Glycol is not a stable working fluid that eventually gets dirty. It is an organic compound that oxidises to acids, and the inhibited package it ships in contains a finite quantity of alkaline buffer whose entire job is to absorb those acids until it cannot. While the buffer holds, pH barely moves. That is the trap: the number every tech checks is the number that stays quiet longest, and by the time it starts falling the fluid is nearly through.

A loop can be four years into a five year decline with a pH reading that looks fine on the ticket.

Before you drain, clean, or touch the heat source

Read Section 8 of the safety data sheet for the fluid and separately for any cleaner, and use the glove class and eye protection each one names. They are different products in different hazard classes. 29 CFR 1910.1200 requires those sheets be available to you.

Never bring an acid cleaner and a hypochlorite product together, in the loop, in a container, or in sequence without a full flush between. That combination releases chlorine gas within seconds, and in a plant room that is an inhalation event rather than a splash. Separation and flushing are the control; a respirator is only ever part of an answer under a written program with fit testing per 29 CFR 1910.134.

If the heat source is electric and you intend to inspect or replace an element, 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 is 29 CFR 1910.333(b)(2); the construction counterpart for electrical lockout and tagging is 29 CFR 1926.417.

Read loop temperature before opening a drain. Above roughly 140 F the fluid flashes at the opening and scalds, so isolate and let it cool, and run the discharge into a closed container placed where the stream cannot reach anybody.

Degraded glycol carries dissolved metals it picked up on the way down. It is not a drain disposal. Contain it, and route it through the fluid supplier or a licensed waste handler under the local authority's rules.

The call that came back three times

A glycol heating loop serving a process load through a plate heat exchanger. Complaint: the strainer blocked with a sticky brown deposit and the exchanger's approach temperature drifted. The shop cleaned the strainer and the plates, and it came back. They cleaned it again, and it came back sooner. On the third visit somebody pulled the annual fluid analyses instead of the strainer.

Five sheets in a folder, and the only column anybody had ever read out loud was pH.

Why pH was quiet for three years

Across the three years from the first sample to the fourth, pH fell from 9.4 to 8.8. That is 0.6 of a unit over three years and any tech reading it annually would describe the fluid as stable, because it is a small movement in a range that is still comfortably alkaline.

The reserve alkalinity column over those same three years fell from 14.0 to 2.5, which is 82% of the buffer gone. That is not a stable fluid. That is a fluid whose entire acid absorbing capacity has been consumed while the one indicator being watched barely moved.

Then in the single year from the fourth sample to the fifth, pH fell 1.9 units and dissolved iron went from 1.4 to 9.8 mg/L, a factor of seven. That shape is not a coincidence and it is not specific to glycol. A buffered solution resists pH change while buffer remains and then falls steeply once it is exhausted, which is the same curve as any titration. Reserve alkalinity is the quantity being spent; pH is what you see after it runs out.

The supplier's own action level was already breached two years before the failure. The product's guidance called for action once reserve alkalinity fell below half its commissioning value, and half of 14.0 is 7.0. The Year 3 sample read 6.5, which is under that limit. The fluid was formally out of specification at the Year 3 visit, on a page that also showed pH 9.1 and a nearly clear sample, and nobody acted because nothing looked wrong.

What the acids are and what they attack

Glycol oxidises through an aldehyde intermediate to organic acids. Ethylene glycol gives glycolic, oxalic and formic acids. Propylene glycol gives lactic, pyruvic and acetic acids. The specific acids differ; the consequence does not.

Those acids do four things, and only the first is obvious.

• They consume the buffer, which is what the reserve alkalinity column is measuring, and which is why that column is the early warning.
• Once the buffer is gone they drop pH, and low pH attacks the things that low pH always attacks: solder joints, zinc coatings on galvanised components, and aluminium, which is unstable at both ends of the pH range rather than only in acid.
• The dissolved metals they release then catalyse further degradation. Copper in particular accelerates glycol oxidation, so a loop that has been quietly losing copper is also shortening its own fluid life. That is one reason the copper specific inhibitors matter more than their small dose suggests.
• The heavier products are tacky and they foul heat transfer surfaces, and as they accumulate the fluid's viscosity rises, which reduces flow at a fixed pump duty.

That last pairing is where a degrading loop starts driving itself.

The temperature that was never measured

Oxidation rate rises steeply with temperature, and the temperature that governs it is the one at the heat transfer surface, not the one on the loop gauge. Fluid manufacturers publish a maximum bulk temperature and a maximum film temperature, and the film limit is the one that gets exceeded because nothing on the system measures it.

This loop ran 180 F bulk, comfortably inside the bulk limit, with an electric element as its heat source. An element transfers all its output through a small surface area, so the fluid touching it runs far hotter than the loop, and how much hotter depends entirely on how fast fluid moves past it. Reduce the flow and the film temperature climbs.

The direction is worth being able to state at both ends, because it decides whether this is your failure mode at all. The same inhibited fluid in a loop running 120 F bulk with a modulating source and generous velocity across the heat transfer surface will outlive this one by years. In a chilled water loop at 45 F, thermal oxidation is essentially not the failure mode; those loops lose their fluid to biological attack and oxygen ingress instead, and a chilled glycol charge that has gone bad will show a different set of columns moving.

Why cleaning it kept making it worse

Put the two together and the reinforcing path is obvious in hindsight.

Degradation products foul the strainer and the plates. A partly blocked strainer reduces flow past the element. Reduced flow raises the film temperature at the element surface. Higher film temperature accelerates oxidation. More oxidation produces more deposit, and the interval between blockages shortens, which is exactly what the shop observed across the three visits.

Every cleaning visit restored flow and bought time, and every one of them left the fluid a little further along. The cleaning was not wrong; it was necessary and insufficient, and treating it as the repair is what turned a fluid replacement into three visits plus a fluid replacement.

There is a related trap on the same job. If a flow proving device or a high limit on the element had ever been bypassed to keep the system in service, that device was reporting the actual condition and the bypass removed the last thing standing between the loop and this outcome. Where a protective device is behaving as the obstacle, establish why it is opening before treating it as the fault.

What actually fixes it

Not a top up of buffer. Adding alkalinity to acidic degraded fluid neutralises what is already there and does nothing about the intermediates, the catalysing metals, or the deposits, all of which are still in the loop generating more acid. It buys months and it makes the next analysis harder to read.

The fluid gets replaced, and the replacement is a sequence rather than a drain and refill:

1. Drain fully, including low points and any section that does not gravity drain, because whatever is left behind seeds the new charge with catalysing metals and degradation products.
2. Clean the deposits with a product matched to what they are, which comes from the fluid supplier rather than from the shelf. Sticky organic degradation product and mineral scale are different problems and the wrong cleaner does nothing.
3. Flush until the rinse water matches the fill water on conductivity and pH. That is the objective test; a rinse that looks clear is not evidence.
4. Recharge to the concentration on the system's record, calculated on the measured loop volume, and measure the result rather than trusting the drum quantities.
5. Fix the film temperature cause before returning to service. Restore full flow across the heat transfer surface, prove that flow with the system's own proving device rather than by observation, restore any protective device that had been bypassed, and only then energize the heat source. Energizing an element into inadequate flow is the mechanism that produced all of this.

Then reset the monitoring so it watches the right column. An annual analysis that reports pH and glycol percentage will miss the next decline exactly as it missed this one. Reserve alkalinity and dissolved iron are the two that move early, and the action level is a fraction of the commissioning value, which means the commissioning value has to be on the record for the fraction to mean anything.

References

• 29 CFR 1910.1200, Hazard Communication, for the safety data sheets whose Section 8 sets glove class and eye protection for the fluid and for any cleaner used
• 29 CFR 1910.134, Respiratory Protection, for the written program and fit testing required before a respirator is part of a chemical control
• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and locking out an electric heating element before inspection
• NFPA 70E-2021, 120.5, for the live-dead-live instrument verification sequence
• Fluid manufacturer documentation for maximum bulk and film temperatures, reserve alkalinity action levels, and approved cleaning products