Switching Secondary Refrigerants: A Hands-On Replacement Guide

Glacier Coolant · Engineering Team · 2025

I've been in plants where the glycol smells like burnt sugar and the pipe threads weep rust-colored drips onto the floor. I've walked into compressor rooms where the maintenance log shows three pump seal replacements in six months and nobody can figure out why. Nine times out of ten, the secondary refrigerant — the coolant circulating through the jacket, the shell-and-tube, the cold room evaporator — hasn't been touched in years. It's just there. Background noise. Until it isn't.

Swapping out a secondary refrigerant sounds simple. Drain the old stuff. Flush. Fill with the new. Right?

Wrong. Or at least, not always. Sometimes it's that simple. And sometimes the new coolant eats a gasket that was perfectly happy with the old one, or the viscosity difference at -30°C is enough to trip a flow switch, or the expansion tank was sized for a fluid with twice the thermal expansion coefficient. We've seen all of it.

This isn't a marketing piece. What follows is four real replacement cases from the past year, with the ugly parts left in. The stuff that worked, the stuff we had to redo, and the decisions that kept people up at night.

What Actually Matters When You Switch

Before jumping into the cases, let's get the obvious out of the way. People ask about freeze point first. Always. It's the wrong question. Freeze point matters, sure, but not as much as people think.

The properties that'll bite you: viscosity at your lowest operating temperature, thermal conductivity, specific heat capacity, material compatibility with every elastomer in the loop, and — this one catches people off guard — whether the new fluid's density means your pump curve shifts enough to matter.

I'll say this once: if you're switching from a water-based secondary refrigerant to a water-free one, or vice versa, the system design implications are significant. It's not plug-and-play. The heat transfer characteristics are different. The pumping requirements are different. The expansion behavior is different. Pretending otherwise is how you end up with a system that limps along at 70% capacity.

Glacier Coolant's product line gives you options across the spectrum. Water-based LM-4 series from -20 to 150°C. Water-based LM-8 from -50 to 120°C with non-flammable formulation. Water-free LM-11D from -60°C. Water-free LM-11F from -85°C. Each one solves a different problem. The trick is knowing which problem you actually have.

Quick reference: LM-4 covers general-purpose anti-corrosion at -20 to 150°C (six sub-models: LM-4, 4D, 4D-YE, 430, 445, 495). LM-8 is water-based, non-flammable, -50 to 120°C. LM-11D is water-free, minimum -60°C. LM-11F is water-free, minimum -85°C. LM-1 was an earlier formulation, now superseded in most applications.

Case 1: Beijing Zhong***— Methanol/Glycol to LM-8
(Note: The full name of the company should be omitted as required by the confidentiality agreement.) 

February 14, 2025. Beijing. A heat source tower system that had been running on a methanol-glycol blend for years.

The operator's complaint was straightforward: the flammability risk of methanol was becoming unacceptable. Insurance was pushing back. The local fire authority had raised questions. And honestly, the maintenance team was tired of working around a fluid that could ignite if someone got careless with a welding torch near a leaking flange.

They came to us asking about LM-8. The spec sheet looked like a fit: water-based, -50 to 120°C, non-flammable. The operating temperature range of their heat source tower fell comfortably within LM-8's working window. On paper, this was a drop-in replacement.

On paper.

The first thing we checked was material compatibility. The existing system had EPDM gaskets throughout. LM-8 is compatible with EPDM, so that box was ticked. The pump seals were mechanical carbon-ceramic — no issue there either. The piping was carbon steel with an existing corrosion inhibitor package in the old glycol, so the metal surfaces were in reasonable shape.

What we didn't anticipate: the old methanol-glycol blend had been masking a slow leak at a heat exchanger flange. The methanol's lower surface tension meant it had been seeping through a gasket interface that LM-8's higher surface tension couldn't penetrate — so the leak didn't start, it stopped. We found it during the pressure test after the flush. Fixed the gasket. Moved on.

The flush itself took two passes. First pass with clean water to push out the bulk of the old glycol. Second pass with a dilute cleaning solution to pull residual organics off the pipe walls. Conductivity monitoring on the flush water told us when we were done: when the reading stabilized at baseline, the system was clean.

They filled with LM-8, circulated for 24 hours, checked the concentration, and went live. The system has been running since February without incident. The insurance carrier is happy. The fire authority is happy. The heat transfer performance is within 3% of the old fluid, which is basically measurement noise.

Total downtime: 36 hours. Planned: 48. That's a win.

Takeaway: Methanol-to-LM-8 swaps are among the most straightforward replacements we do. The key risks are (1) uncovering leaks that the old fluid masked, and (2) incomplete flushing leaving residual methanol that compromises the non-flammable rating. Get the flush right, and the rest follows.

Case 2: Cheng*** — LM-1 to LM-11F
(Note: The full name of the company should be omitted as required by the confidentiality agreement.) 

February 24, 2025. Jiashan. Cheng*** had a low-temperature reactor loop running on LM-1, Glacier's earlier water-free formulation. The required operating temperature had been pushed lower over time — originally -55°C, now they needed -75°C consistently. LM-1 wasn't rated for that. LM-11F is: water-free, minimum -85°C.

This was different from the Beijing case. We weren't changing chemistry families. Both LM-1 and LM-11F are water-free formulations. The question was whether the system could handle the lower-temperature operation, not just whether the fluid could.

Pharma is a different world. The validation requirements alone can make a simple fluid change take weeks. Every material contact surface needs to be documented. Every cleaning step needs to be verified. The production schedule doesn't bend — if the reactor is down, batches are waiting.

We spent three days on compatibility testing before touching the system. LM-11F's formulation is different enough from LM-1 that we wanted to be sure about the pump mechanical seals, the PTFE gaskets, and the stainless steel passivation layer. Everything checked out. The PTFE was fine — PTFE doesn't care about much of anything. The 316L stainless was fine. The pump seals were silicon carbide, which is compatible with the entire LM-11 family.

The flush was the bottleneck. LM-1 and LM-11F are miscible, which helps — you don't get phase separation or gelling at the interface. But the old LM-1 had been in the system for three years, and there was some degradation. Darkening. Slight viscosity increase. Nothing that would cause a failure, but enough that we wanted it out before introducing the new fluid.

We drained hot — got the loop up to about 40°C before draining, which dropped the viscosity of the old LM-1 enough to get better drainage from the low points. Then a nitrogen purge to push the remaining fluid out of the dead legs. Then a fill with LM-11F, circulate, drain again as a sacrificial rinse. Then the final fill.

Performance at -75°C was solid. The viscosity of LM-11F at that temperature is low enough that the pump didn't need to work significantly harder. Chengda has been running on it since February. No flags.

When to use LM-11F vs LM-11D: LM-11F goes to -85°C, LM-11D goes to -60°C. If your process bottoms out at -50°C, LM-11D is fine — and it's less expensive. Don't over-spec. The LM-11F premium is only worth it when you genuinely need sub-minus-60 operation. Chengda needed -75°C, so LM-11F was the only water-free option on the table.

Case 3: Taizhou — LM-4 to LM-11D

October 6, 2025. Taizhou. This one was interesting because it crossed the water-based to water-free boundary.

The existing system was running LM-4, Glacier's water-based general-purpose coolant. LM-4 is rated from -20 to 150°C. The system had been operating at around -10°C for years. No problems. Then the process changed. The customer needed to go to -45°C, and LM-4 at -20°C minimum wasn't going to cut it.

They called us and asked: can we just switch to LM-11D?

Yes. But.

The "but" was the system design. LM-4 is water-based. LM-11D is water-free. The thermal properties are different. Water has a specific heat capacity of roughly 4.2 kJ/kg·K. Water-free organic coolants are typically in the 2.0-2.5 range. That means, for the same cooling duty, you need roughly 1.7 to 2 times the mass flow rate with a water-free coolant. Your pump needs to move more fluid. Your pipe sizing might be marginal. Your heat exchanger might be undersized.

We ran the numbers. The existing pump could handle the flow increase — it had been oversized from the start (the original designer had left margin for a future expansion that never happened). The heat exchanger had enough surface area that the reduced heat transfer coefficient of the water-free fluid was compensated by a larger ΔT across the exchanger. The piping was 2-inch schedule 40 — no issue with the higher flow velocity.

So the system could handle it. But the flush was more involved than a water-to-water swap. LM-4 and LM-11D are not miscible. You can't just drain the LM-4 and fill with LM-11D — the residual water will form a separate phase, freeze at the cold spots, and wreck your pump.

The flush procedure: drain LM-4 hot. Water flush. Water flush again. Nitrogen blow-down of all low points and dead legs. Vacuum dry to remove residual moisture. Then fill with LM-11D, circulate, check for water content. We hit 150 ppm water on the first fill, which is acceptable but not great. Drained, refilled. 80 ppm. Good enough.

The system has been running at -45°C since October. The pump is drawing about 12% more power than it did with LM-4 — expected, and within the motor's service factor. The cooling capacity is actually slightly better than predicted, which we attribute to the heat exchanger being cleaner after the thorough flush.

Water-based to water-free is the hard direction. Going the other way — water-free to water-based — is usually easier because you're going into a fluid with better heat transfer properties. You'll have excess capacity. The risk is residual water-free fluid contaminating the water-based coolant, but a good flush handles that. The Taizhou case is the direction that requires engineering: making sure the system can handle the reduced heat transfer coefficient of the water-free fluid.

Case 4: Jiangsu — LM-8 to LM-11D with Piping Redesign

December 3, 2025. Jiangsu. This was the most involved replacement of the year.

The existing system: LM-8, water-based, operating at -40°C in a pharmaceutical intermediate process. The customer wanted to switch to LM-11D for two reasons: they were planning to push the lower operating limit to -55°C (LM-8 bottoms out at -50°C), and they wanted the corrosion protection advantages of a water-free system for their stainless steel piping.

But here's the thing: their piping was undersized for a water-free coolant.

LM-8, being water-based, has a specific heat capacity around 3.8 kJ/kg·K at operating temperature. LM-11D is around 2.2. The existing piping was 1.5-inch schedule 10 stainless. At the flow rate needed to match the cooling duty with LM-11D, the pressure drop was going to be about 3.5 times higher than the system was designed for. The pump couldn't handle it.

We had two options: a bigger pump, or bigger piping. Bigger pump was cheaper upfront but would increase operating costs permanently. Bigger piping was more expensive now but would pay back in lower pumping costs. The customer chose bigger piping.

The redesign: replaced 1.5-inch lines with 2.5-inch wherever possible, kept 1.5-inch only at the reactor jacket connections where changing the nozzle size wasn't practical. The 1.5-inch sections are short — less than 2 meters total — so the pressure drop impact is manageable. The pump stayed the same. The expansion tank was resized for LM-11D's higher thermal expansion coefficient.

This was a two-week project. Re-piping, flushing, drying, filling. The flush procedure was similar to the Taizhou case: drain, water flush, nitrogen blow-down, vacuum dry, fill. But with the new piping, we had to be extra careful about construction debris — welding slag, cutting oil, metal shavings. The flush included a chemical cleaning pass with a mild citric acid solution to passivate the new stainless welds.

They went live on December 17. The system is running at -55°C with LM-11D. The pressure drop is actually lower than the original system with LM-8, because the new piping is so much larger. The pump is drawing less power than before. The irony: they upgraded the system to handle a worse-cooling fluid, and ended up with better overall performance because the piping was the real bottleneck all along.

Hidden bottleneck: The Jiangsu case is a reminder that the coolant isn't always the limiting factor. Sometimes the system design is the problem, and the coolant swap just exposes it. If they'd simply switched to LM-11D without the piping upgrade, they'd have blamed the coolant for poor performance. The coolant was fine. The pipes were too small.

Comparison: The Four Replacements Side by Side

When Should You NOT Switch?

I get asked this a lot. Someone reads about a coolant that goes to -85°C and thinks, "my system runs at -15°C, but this one's better, right?"

No.

LM-4 is the right coolant for a system running at -15°C. It's water-based, it has excellent heat transfer properties, it's cost-effective, and it has a proven corrosion inhibitor package. Switching to LM-11F would be a downgrade in heat transfer performance, a significant cost increase, and zero benefit. The LM-11F's -85°C capability is irrelevant if you never go below -20°C.

Here's a rule of thumb that's served me well: use the cheapest fluid that meets your temperature requirements and material compatibility constraints. Don't over-spec. The premium products exist for processes that genuinely need them. If your process doesn't, you're just paying for performance you'll never use.

Another situation where you shouldn't switch: when the system has known leak points that the current fluid is tolerating. Water-based coolants with good inhibitor packages can handle minor leaks — the inhibitor passivates the exposed metal at the leak site. Water-free coolants won't cause corrosion at a leak, but they'll leak faster because of lower viscosity. Fix the leaks first. Then switch.

And one more: if your system was designed for a specific fluid and you can't verify compatibility with the replacement, don't do it. I've seen a pump fail in three days because someone switched to a coolant that was incompatible with the Buna-N seals. The coolant was fine. The seal material choice was wrong. But the pump was dead, and that's a very expensive lesson in reading the compatibility chart.

The Flush: Make or Break

If there's one thing I want you to take from this, it's that the flush is everything. A bad flush ruins a good coolant. A good flush makes a mediocre coolant work. The flush is where the replacement succeeds or fails.

For water-to-water swaps (like Beijing — methanol/glycol to LM-8), the flush is forgiving. You're replacing one water-based fluid with another. Residual old fluid at 1-2% concentration won't cause problems. The main concern is removing enough of the old inhibitor package that it doesn't interact with the new one.

For water-free to water-free swaps (like Chengda — LM-1 to LM-11F), the flush is moderate. The fluids are miscible. Residual old fluid is a contaminant, not a disaster. The concern is degradation products from the old fluid that might affect the new fluid's properties.

For water-based to water-free swaps (like Taizhou and Jiangsu), the flush is unforgiving. Residual water in a water-free system is a serious problem. At low temperatures, that water freezes. Ice crystals wreck pump seals. Ice blocks flow through small passages. Water in a water-free coolant also promotes corrosion by creating localized aqueous environments at metal surfaces.

Our standard water-to-water-free flush protocol: drain hot, water flush twice, nitrogen blow-down of all low points, vacuum dry to below 500 ppm moisture, fill, circulate, sample, and only proceed if water content is below 200 ppm. It takes time. It takes patience. Skip a step and you'll pay for it later.

Material Compatibility: The Chart Nobody Reads

Every coolant has a material compatibility chart. Almost nobody reads it. Then a gasket swells or a seal fails and suddenly everyone's very interested in the chart.

The LM-4 series (water-based) is compatible with most common elastomers: EPDM, nitrile, Viton, PTFE. The corrosion inhibitor package is designed for mixed-metal systems — carbon steel, copper, brass, cast iron, aluminum. The LM-4D variant is specifically formulated without nitrite for copper-sensitive applications. LM-4D-YE has ultra-low conductivity (below 800 μS/cm) for applications where electrical conductivity matters, like direct-cooled power electronics.

The LM-8 (water-based, non-flammable) is similarly broad in compatibility. The non-flammable formulation doesn't change the material compatibility profile significantly — it's still water-based with an inhibitor package. The main difference is the absence of flammable glycols or alcohols.

The LM-11 series (water-free) is compatible with most metals but requires attention to elastomers. Some EPDM formulations swell slightly in water-free organic coolants. Most nitrile compounds are fine. PTFE is universally fine. Always check. Always.

The LM-14 series (fluorinated fluids, sub-minus-100°C) is a different world entirely. These are perfluorinated compounds. They're essentially inert. They don't swell anything. They don't corrode anything. They also have very low heat transfer coefficients and very high cost. You use LM-14 when nothing else will work at the temperature you need.

The LM-15 series (heat transfer oil, -15 to +160°C) is for applications above the practical range of water-based coolants. These are synthetic organic heat transfer fluids. Material compatibility is generally good with metals, but elastomers need checking — some silicone compounds don't hold up well in hot organic fluids.

Quick Selection Guide

Things That Went Wrong (So You Don't Have To)

I've been doing this long enough to have a collection of failures. They're more instructive than the successes.

We had a customer switch from a competitor's glycol to LM-4 without checking the expansion tank sizing. The old glycol had a higher thermal expansion coefficient. The new LM-4 expanded less. The expansion tank was suddenly too large — not a safety issue, but the system pressure was fluctuating more than expected because the tank's gas cushion was oversized for the reduced expansion volume. Took us two days to figure out why the pressure relief valve was chattering on startup. The fix was simple: adjust the pre-charge pressure on the expansion tank. The diagnosis was not.

Another one: a customer drained their old coolant, flushed with city water, and filled with LM-8. What they didn't know was that their city water had 350 ppm chloride. The residual chloride in the system after flushing was enough to stress the stainless steel heat exchanger. LM-8's inhibitor package handled it, but the safety margin was thinner than it should have been. We now recommend deionized water for the final flush pass in any system with stainless steel components.

And my favorite: a customer in the food industry drained their coolant, flushed, and filled with LM-4. All good. Except they forgot to re-tighten the drain valve after the flush. They filled 200 liters of LM-4 straight onto the floor before anyone noticed. The coolant was fine. The floor was very clean. The drain valve is now on the startup checklist.

Final Thoughts

Switching secondary refrigerants isn't rocket science. It's plumbing, chemistry, and a bit of engineering judgment. The four cases here — Beijing, Chengda, Taizhou, Jiangsu — cover the main scenarios: water-to-water, water-free-to-water-free, water-to-water-free, and water-to-water-free with system modifications. Each one had its own wrinkles. Each one worked.

The common thread: planning. Every one of these replacements succeeded because someone spent the time to check compatibility, size the flush, verify the system design, and think through the failure modes before touching a valve. The ones that fail — and they do fail, I've seen them — fail because someone assumed it would be fine and didn't check.

Don't be that person. Check the compatibility chart. Size the flush properly. Verify the pump curve. Monitor the water content. And if you're not sure, call someone who's done it before. We answer the phone.