When Cooling Fails: A Diagnostic Manual for Secondary Refrigerant Systems

Let's get one thing straight right at the top: when your cooling system stops cooling, the coolant is rarely the problem. It's usually the messenger. Something else changed — a process condition, a mechanical failure, an operating error — and the coolant is just the first thing you noticed because it's the easiest thing to check. Pull a sample, look at it, and if it's not the right color you blame the fluid. I've been in this business long enough to know that's how it goes.

This manual is not a theory document. It's not a product brochure. It's a diagnostic reference built from actual field cases where the cooling failed and someone had to figure out why. Six cases from 2025. Each one looked like a fluid problem. Each one turned out to be something else — or at least something more complicated than "the coolant is bad."

If you're a technician standing in front of a system that won't hit temperature, start here. If you're a plant engineer who's been told "it's probably the fluid," start here. If you're a process manager who needs to explain to your boss why production is down, start here.

Case 1: The Wrong Fluid for the Wrong Job

Chengda Pharma, Jiashan — February 24, 2025

System: Pharmaceutical reactor cooling loop
Fluid in use: LM-1 (customer's existing charge)
Target temperature: -70°C
Symptom: System cannot reach target; stalls at -52°C
Actual cause: LM-1 thermal conductivity too low for the application; LM-11F required

This one is almost embarrassing to include because it's so basic. But it happened. And it keeps happening, at different plants, with different fluids. So here it is.

Chengda Pharma in Jiashan called us in February. Their reactor cooling system — a 400-liter loop feeding a single glass-lined reactor — couldn't reach the process temperature of -70°C. The system was designed for -70°C. The chiller was sized for -70°C. The piping was insulated for -70°C. But the best they could get was -52°C. The operators had been fighting this for two weeks. They'd checked the chiller. They'd checked the insulation. They'd checked the controller. Everything was fine. The system just wouldn't go cold enough.

When I asked what fluid they were running, the plant engineer said "LM-1." I asked why. He said "because that's what we've always used."

LM-1 is a decent general-purpose coolant. But it's not rated for -70°C. Its viscosity at -70°C is astronomical — you'd need a pump the size of a refrigerator to push it through a heat exchanger at any useful flow rate. The thermal conductivity drops off a cliff below -50°C. The system wasn't broken. It was just using the wrong fluid for the job.

We switched them to LM-11F, which is rated for ≥-85°C. Water-free, low viscosity at extreme low temperatures, and the thermal conductivity holds up well below -70°C. The system hit -70°C within 90 minutes of the recharge. The plant engineer was embarrassed. I told him not to be. Fluid selection is deceptively hard. The datasheet gives you a temperature range. It doesn't tell you what happens to the heat transfer coefficient at the bottom of that range.

Verify the fluid's actual low-temperature performance. Don't rely on the rated minimum temperature. At the low end of the range, viscosity and thermal conductivity degrade non-linearly. A fluid rated for -60°C may have only 30% of its room-temperature heat transfer capacity at -55°C. Get the full viscosity and thermal conductivity curves from the manufacturer. If they can't provide them, find a different supplier.

Match the fluid to the duty, not the chiller. The chiller can produce -70°C. The question is whether the fluid can carry that cooling capacity from the chiller to the process. A fluid with high viscosity at low temperature creates a high pressure drop. The pump may not be able to deliver the design flow rate. The result is a system that looks like it's underperforming when it's actually starved for flow.

Check the Reynolds number at the operating temperature. If the flow is laminar (Re < 2,300), the heat transfer coefficient is dramatically lower than in turbulent flow. You may need a different fluid, a bigger pump, or a redesigned heat exchanger. Don't guess. Calculate.

Quick check: If your system can reach the setpoint when the chiller is oversized (e.g., during commissioning with a cold system) but can't maintain it under process load, the problem is almost certainly fluid-side heat transfer, not chiller capacity. The fluid is acting as a thermal bottleneck.

Case 2: Stuck at -42°C

Chongqing Chemical Plant — May 28, 2025

System: Batch reactor cooling, 600-liter loop
Fluid: LM-8 (water-based, -50~120°C)
Target temperature: -48°C
Symptom: System stalls at -42°C; cannot go lower regardless of chiller output
Actual cause: Ice crystal formation in low-flow zones of heat exchanger

LM-8 is rated for -50°C. The customer was operating at -48°C. That's within spec. Barely. But it should work. The chiller was maintaining -52°C on the primary side. The approach temperature across the heat exchanger was 10°C — way too high. A clean heat exchanger with LM-8 at design flow should have an approach of 3-4°C.

I pulled the heat exchanger plates. The coolant side had a thin layer of what looked like frost. It wasn't frost. It was ice crystals. LM-8 is water-based. The freezing point — with the full inhibitor package — is around -52°C. At -48°C, the bulk fluid is still liquid. But here's the thing: in the low-flow zones of the heat exchanger — the corners, the edges of the gasket grooves, the areas behind the distributor — the fluid velocity drops to near zero. The residence time goes up. The local temperature can drop below the bulk temperature by 2-3°C. That's enough to push it below the freezing point.

The ice crystals formed on the plate surfaces in these low-flow zones. They didn't block the channels. They just added a layer of thermal resistance. Ice is a terrible conductor of heat — about 2.2 W/m·K, compared to 0.4-0.5 W/m·K for LM-8. The ice layer was acting as insulation on the coolant side of the plates. The heat transfer coefficient dropped. The approach temperature went up. The system couldn't reach -48°C.

The fix was to increase the coolant flow rate by 20%. The higher velocity swept the low-flow zones, preventing ice nucleation. The approach temperature dropped to 3.5°C. The system hit -48°C and held it.

Check the approach temperature. Approach = (chiller supply temperature) - (coolant temperature at heat exchanger outlet). If it's more than 5°C, you have a heat transfer problem. The heat exchanger is fouled, the flow rate is too low, or the fluid properties have degraded.

Inspect the heat exchanger surfaces. If you're operating within 5°C of the fluid's freezing point, ice crystal formation on heat exchanger surfaces is a real risk — even if the bulk fluid is above the freezing point. The local temperature at the wall can be 2-3°C below the bulk temperature.

Increase flow rate before you increase chiller capacity. Higher velocity reduces the thermal boundary layer thickness, improves heat transfer, and prevents ice nucleation in low-flow zones. It's the cheapest fix you can try.

Case 3: Stuck at -36°C

Zhejiang Fine Chemical Plant — October 24, 2025

System: Continuous process cooling, 1,200-liter loop
Fluid: LM-8 (water-based, -50~120°C)
Target temperature: -45°C
Symptom: System stalls at -36°C; pump cavitation at lower temperatures
Actual cause: Viscosity increase due to water evaporation over 18 months of service

This case looked similar to the Chongqing one at first glance. LM-8. Not reaching temperature. The difference was the failure mode. In Chongqing, the problem was ice crystals on the heat exchanger surfaces. In Zhejiang, the problem was in the pump.

The system had been in service for 18 months. The LM-8 had never been changed. No makeup fluid had been added. The expansion tank level had been dropping slowly — about 2% per month — which the operators attributed to normal evaporation. They were right about the evaporation. They were wrong about it being normal.

LM-8 is a water-based coolant. When water evaporates from the expansion tank, the coolant concentrate left behind increases in concentration. The viscosity goes up. At -36°C, the viscosity of the aged LM-8 was 3.2 times higher than fresh LM-8. The pump — a centrifugal unit sized for the original viscosity — was cavitating. The NPSH available was below the NPSH required. The pump was essentially choking on its own suction.

The operators had been compensating by reducing the flow rate to avoid cavitation. Lower flow rate meant lower heat transfer. Lower heat transfer meant the system couldn't reach the target temperature. They'd been chasing their tail for weeks.

We drained the system. The drained fluid volume was 920 liters — the original charge was 1,200 liters. The missing 280 liters was water that had evaporated. The concentration of the coolant in the remaining fluid was about 30% higher than specification. We recharged with fresh LM-8 at the correct concentration. The system hit -45°C within an hour.

Track fluid level and concentration separately. A dropping expansion tank level could mean a leak. It could also mean evaporation. If it's evaporation, the fluid left behind is more concentrated. The only way to know is to measure the concentration — refractive index, density, or a simple boiling point test. Don't just top up with water. That fixes the level but dilutes the inhibitor. Top up with pre-mixed coolant at the correct concentration.

Check pump suction conditions at the operating temperature. Viscosity increases as temperature drops. The pump curve you're using was probably measured at 20°C. At -40°C, the same pump may not be able to generate enough suction head. If the pump is cavitating, the problem isn't the fluid — it's the pump sizing. But the fluid viscosity is what determines whether the pump is adequately sized.

Establish a fluid replacement schedule. Water-based coolants don't last forever. Evaporation changes the concentration. Inhibitor depletion changes the chemistry. Thermal cycling changes the physical properties. At minimum, test the fluid annually. Replace it when the concentration deviates more than 10% from the specification.

Warning: A centrifugal pump cavitating at low temperature is a safety hazard. The vapor bubbles collapse with enough force to erode the impeller. Over time, this can cause impeller failure. If you hear a sound like gravel going through the pump, shut it down. Don't try to "ride it out."

Case 4: Stuck at -16°C

Chongqing Jinguan Chemical — November 15, 2025

System: Pilot plant reactor cooling
Fluid: LM-445 (low-temperature water-based)
Target temperature: -30°C
Symptom: System stalls at -16°C; no visible ice, no pump issues
Actual cause: User had diluted LM-445 concentrate with tap water instead of deionized water

Sometimes the problem is so simple you overlook it.

The system was a small pilot plant — 80 liters of LM-445 in a loop cooling a 5-liter jacketed reactor. The target was -30°C. The system had been running fine for about two months. Then, after a scheduled fluid change, it couldn't get below -16°C.

The operators had drained the old fluid. Flushed the system with water. Refilled with fresh LM-445 concentrate, diluted to the recommended concentration. The concentration checked out — 45% by refractometer, right on spec. The pH was normal. The inhibitor level was normal. Everything was normal, except the system wouldn't go below -16°C.

I asked what water they used for dilution. "Tap water," the operator said. Chongqing tap water. Hardness around 180 mg/L as CaCO₃. Total dissolved solids around 300 mg/L. The calcium and magnesium ions in the tap water were depressing the freezing point of the mixture — but not in a useful way. They were forming insoluble salts that precipitated out of solution at low temperatures. The salts weren't visible to the naked eye. They were sub-micron particles that acted as nucleation sites for ice crystal formation. The bulk fluid freezing point, as measured by the refractometer, was still -35°C. But the actual freezing point, in the presence of these nucleation sites, was closer to -18°C.

We drained the system. Flushed it with deionized water. Refilled with LM-445 diluted with deionized water. The system hit -30°C without any issues. Total cost of the fix: about 300 yuan for the deionized water. Total cost of the downtime: roughly 15,000 yuan in lost pilot plant time.

Use deionized or distilled water for dilution. Always. Tap water contains dissolved solids that affect the freezing point, the inhibitor chemistry, and the scaling tendency of the fluid. The hardness ions (calcium, magnesium) will precipitate as the temperature drops, forming nucleation sites that raise the effective freezing point. If you don't have a DI water source, buy distilled water in bottles. It's cheaper than the downtime.

Don't trust the refractometer alone. A refractometer measures the refractive index, which correlates with the total dissolved solids. It doesn't distinguish between the coolant concentrate and the dissolved minerals in the dilution water. A 45% reading on tap water could be 40% coolant + 5% dissolved minerals. The freezing point of that mixture is not the same as 45% coolant in deionized water.

Flush the system with DI water between fluid changes. If the previous fill used tap water, the system walls are coated with precipitated minerals. A new fill of correctly diluted coolant will re-dissolve those minerals over time, degrading the new fluid. Flush until the flush water conductivity is below 10 μS/cm.

Case 5: The Drop-In Replacement That Wasn't

Taizhou Chemical Plant — October 6, 2025

System: Multi-reactor cooling loop, 2,000-liter charge
Old fluid: LM-4 (water-based, -20~150°C)
New fluid: LM-11D (water-free, ≥-60°C)
Target temperature: -40°C (new requirement)
Symptom: After switch to LM-11D, system cannot reach -40°C; pump motor overloads
Actual cause: LM-11D has higher density and viscosity than LM-4; existing pump undersized for new fluid

This is the case that makes me angry every time I think about it. Not at the customer. At the industry.

The customer was running LM-4 in a cooling loop that served three reactors. They needed to go colder — from -15°C to -40°C — for a new product. Their engineering consultant recommended switching to LM-11D. LM-11D is water-free, rated to -60°C, and compatible with the existing piping materials. On paper, it was a drop-in replacement. Drain the LM-4. Flush. Fill with LM-11D. Done.

They drained and flushed the system over a weekend. Filled with LM-11D on Monday morning. Started the pump. The pump motor overloaded within 30 seconds. Thermal overload relay tripped. They reset it. Tried again. Same result. The pump was drawing 18.5 amps — the nameplate rating was 15 amps. The fluid was flowing, but the pump was working far harder than it should have been.

Here's what nobody told them: LM-11D has a density of about 1.05 kg/L at 20°C. LM-4 at the concentration they were using had a density of about 1.03 kg/L. That's only a 2% difference. But LM-11D's viscosity at -40°C is about 28 cP. LM-4's viscosity at -15°C (their old operating temperature) was about 6 cP. The pump was sized for a fluid with a viscosity of 6 cP. It was now being asked to pump a fluid with a viscosity of 28 cP. The power draw of a centrifugal pump is proportional to the product of density and viscosity. The pump was being asked to deliver nearly five times the power it was designed for.

The fix was not a fluid change. The fix was a bigger pump motor. And a VFD to control the speed. And a larger impeller to compensate for the higher viscosity. The "drop-in" replacement ended up costing about 80,000 yuan in pump modifications. The fluid itself was fine. The system design was the problem.

When switching fluid types, re-check the pump sizing. A "drop-in" replacement is rarely drop-in. Different fluids have different densities, viscosities, specific heats, and thermal conductivities. All four affect system performance. The pump is the most sensitive component. A 2% density difference + a 4x viscosity difference = a pump that won't work.

Calculate the system curve with the new fluid properties. Pump power = (flow rate × head × density) / (efficiency). Viscosity affects the pump efficiency curve. The manufacturer's pump curve is usually measured with water. You need to apply viscosity correction factors. The Hydraulic Institute standards (ANSI/HI 9.6.7) provide the correction methodology. If you don't have a pump engineer on staff, hire one.

Check the heat exchanger sizing, too. LM-11D has a lower thermal conductivity than LM-4 — about 0.14 W/m·K vs. 0.45 W/m·K. The heat exchanger that worked with LM-4 may not provide enough heat transfer area with LM-11D. You may need a larger heat exchanger or a higher flow rate. These are not small details. They're the difference between a system that works and a system that doesn't.

Fluid switch checklist: Before changing coolant types, verify (a) pump power draw at design flow and minimum temperature, (b) heat exchanger UA value with the new fluid, (c) piping pressure drop at the new fluid's viscosity, (d) expansion tank sizing for the new fluid's thermal expansion coefficient, (e) material compatibility of all wetted components. If you can't answer all five, don't switch.

Case 6: The Synergist That Backfired

Jinsheng Grain, Linyi — June 19, 2025

System: Grain cooling and storage, 5,000-liter loop
Fluid: LM-4 (water-based, -20~150°C)
Target temperature: 5°C (grain storage cooling)
Symptom: Massive foaming after adding a third-party "synergist" additive; pump loses prime; cooling stops
Actual cause: Incompatible additive chemistry; surfactant in the synergist destabilized the LM-4 inhibitor package

Jinsheng Grain operates a large grain storage facility in Linyi, Shandong. They use LM-4 as a secondary refrigerant to cool the stored grain — keeping it at 5°C to prevent insect infestation and mold growth. The system is a 5,000-liter loop with air-handling units in the storage silos. It's a simple, robust system that had been running for years without issues.

In June 2025, a chemical additive supplier convinced the facility manager to try a "coolant synergist" — a product that supposedly improved heat transfer efficiency and reduced pumping costs. The supplier claimed it was compatible with all water-based coolants. The facility manager added 50 liters of the synergist to the 5,000-liter LM-4 charge. That's 1% by volume. Should have been negligible.

Within 24 hours, the system was producing foam. Not a little foam. The expansion tank was overflowing with foam. The pump was losing prime because the suction line was pulling in a foam-liquid mixture. The cooling stopped. The grain temperature started rising. In a facility holding 80,000 tons of grain worth roughly 160 million yuan, a cooling failure is not an inconvenience. It's an existential threat.

We got the call at 11 p.m. on a Thursday. The facility manager was — understandably — not in a good mood. We had a technical team on site by 8 a.m. Friday. The foam sample was analyzed. The synergist contained a non-ionic surfactant that was supposed to reduce the surface tension of the coolant, improving wetting of the heat exchanger surfaces. What it actually did was destabilize the corrosion inhibitor package in LM-4. The surfactant was competing with the inhibitor molecules for adsorption sites on the metal surfaces. The inhibitor was being displaced. The metal surfaces were being exposed. And the surfactant itself was generating foam — lots of it — because that's what surfactants do.

We drained the entire 5,000 liters. The fluid was not recoverable. The surfactant couldn't be separated from the base fluid. The inhibitor package was destroyed. The entire charge was waste. We flushed the system with water three times to remove residual surfactant. Refilled with fresh LM-4. The system was back online by Saturday evening. Total downtime: 36 hours. Total cost: approximately 120,000 yuan for the fluid replacement, plus the cost of the emergency service call, plus the cost of the grain that was starting to warm up.

The facility manager fired the additive supplier. Well, he canceled the contract. Same thing.

Never add third-party additives to a formulated coolant. Coolants like LM-4 are fully formulated products. The inhibitor package, the pH buffer, the anti-foam agent, the dye — they're all balanced to work together. Adding a third-party chemical is like pouring a random ingredient into a recipe you didn't write. Maybe it works. Probably it doesn't. And when it doesn't, the whole system suffers.

If a supplier claims their additive is "compatible with all coolants," ask for proof. Ask for compatibility test data with your specific coolant. Ask for the chemical composition of the additive. Ask for the name of a customer who has used it successfully for more than a year. If they can't provide all three, walk away.

Foam is a chemical incompatibility signal. Coolants are formulated with anti-foam agents. If the system is foaming, something has overwhelmed the anti-foam. It could be a contaminant. It could be an incompatible additive. It could be a mechanical issue (air entrainment at the pump seal). Whatever it is, the foam is telling you something. Listen.

Warning: Foam in a coolant loop is not just a nuisance. It reduces the bulk density of the fluid. The pump is designed to move liquid, not foam. Pump cavitation, loss of prime, and bearing damage can result. A foaming system is a system that's about to fail mechanically. Shut it down and fix the root cause.

Diagnostic Summary: The Five Questions

If you take nothing else from this manual, take these five questions. Every time a cooling system fails to reach temperature, ask them in order. By the time you get to question five, you'll either have found the problem or you'll know you need to call someone who can.

That's it. Five questions. Five diagnostic paths. They won't solve every problem. But they'll solve most of them. And when they don't, you'll have eliminated the obvious candidates and you'll know what to tell the technical support engineer when you call.

One last thing. Don't guess. I've seen too many systems destroyed because someone guessed that the fluid was fine, guessed that the pump was adequate, guessed that the additive wouldn't cause problems. Cooling systems are chemical, mechanical, and thermal systems all at once. They're not simple. They don't respond well to guessing. Test. Measure. Verify. It's slower than guessing. But it's faster than a shutdown.