Corrosion Prevention in Secondary Refrigerant Systems: The Shift from Reactive to Predictive
For as long as industrial cooling systems have existed, corrosion has been the enemy you can't see coming. It happens inside pipes, inside heat exchangers, inside reactor jackets — places you can't inspect without shutting down and cutting things open. By the time you notice the symptoms — discolored coolant, rising pH, dropping heat transfer efficiency — the damage is already done. The metal is gone. It's not coming back. The old approach was simple: wait for something to go wrong, then fix it. React. Respond. Replace. That approach is dying. 2025 was the year the data made it impossible to defend.
What changed? Not the technology — the monitoring tools have been available for years. What changed was the accumulation of case evidence. Seven corrosion cases in a single year, each one a miniature disaster that could have been prevented with better monitoring. Seven facilities that spent money on repairs that they could have spent on prevention. The pattern is now so clear that continuing to manage corrosion reactively isn't just inefficient — it's economically irrational. Let me show you why.
The old way: reactive corrosion management
Let me describe what corrosion management looked like ten years ago, because it's still what most facilities do today. You fill your system with a water-based secondary refrigerant — probably something with corrosion inhibitors. You run it. Maybe you check the pH once a quarter if you're diligent. Maybe you send a sample to a lab once a year. Most of the time, you don't think about it at all. The system runs. Temperatures hold. Everything seems fine.
Then one day, your coolant turns black. Or your heat exchanger fails. Or your chiller can't hold temperature. And suddenly corrosion is your number one problem, and you're scrambling. You pull samples. You send them to a lab. You wait for results. Meanwhile, your system is operating in a degraded state, and the corrosion is still happening. Every hour you wait is another hour of metal loss.
This reactive model has a few obvious problems. The biggest one: by the time you detect corrosion, you've already lost metal. That metal is gone forever. You can stop further corrosion by adding inhibitors, adjusting pH, or replacing the coolant, but you can't undo the pitting, the thinning, the stress concentration points that are now part of your system's metallurgy. You're operating on borrowed time after the first corrosion event. The heat exchanger that was rated for 20 years of service might now have 5 years left. The pipe that was supposed to last the life of the facility might need replacement in the next shutdown. And you won't know which component is compromised until you inspect every inch of the system — which nobody does, because it's too expensive and too disruptive.
The second problem: reactive responses are expensive. Emergency shutdowns, rush replacement parts, overtime labor, lost production — these costs dwarf the cost of prevention. The 新华保险 data center in Beijing found this out in April 2025. Their LM-4 coolant turned black and started foaming. They had to pull samples, send them to a lab, and wait for results while their cooling system operated in a degraded state. A data center with degraded cooling is a data center at risk. Every hour of uncertainty was an hour of elevated risk to servers worth millions. The testing eventually revealed what was happening, but the damage had already started. The cost of the testing and the downtime and the elevated risk? Far more than the cost of a monitoring program that would have caught the problem weeks earlier.
The third problem: reactive management doesn't find the root cause. It treats the symptom. The coolant turned black, so you replace the coolant. The heat exchanger failed, so you replace the heat exchanger. But the thing that caused the corrosion — the oxygen ingress, the acid contamination, the galvanic couple — is still there. You've fixed the damage, not the problem. The next failure is already in progress. The China Aerospace case in 2025 is the textbook illustration: the first heat exchanger failed, they replaced it, and four months later the replacement failed too. The root cause — oxygen ingress through the return line — was only identified on the second investigation. The first failure should have triggered a root cause analysis. It didn't. And the cost of that oversight was a second heat exchanger and more downtime.
What 2025 taught us: seven corrosion cases, seven lessons
March 12, 2025
Sanchuan Chemical: Saltwater Meets Ammonia
Saltwater brine entered the ammonia system, triggering corrosion. The root cause was organic acids dropping the pH. The customer wanted a quick fix — maybe an additive, maybe a corrosion inhibitor boost. But the real problem wasn't the coolant chemistry. It was cross-contamination between systems. You can't fix a process design problem with a chemical additive. The organic acids will keep dropping the pH as long as the contamination pathway exists. Until that pathway is closed — until the systems are physically isolated or the leak is repaired — no coolant formulation will save you. This is a process engineering problem masquerading as a coolant problem. The coolant was the messenger. The process design was the culprit.
April 22, 2025
XX Data Center: LM-4 Turns Black, Foams
Data center cooling is supposed to be boring. Predictable. When the coolant in a data center starts changing color and foaming, something has gone seriously wrong. The LM-4 was sampled and sent for analysis. The discoloration and foaming pointed to contamination — possibly biological, possibly chemical, possibly both. In a data center, you can't afford uncertainty. Every hour of degraded cooling is an hour of elevated risk to servers worth millions of yuan. The investigation revealed that the coolant's inhibitor package had been depleted, probably by a combination of normal aging and contamination. The fix was a system flush and refill with fresh LM-4. The prevention — the thing that would have caught this before the coolant turned black — was a simple pH and conductivity monitoring program that didn't exist.
April 26, 2025
Xinjiang Cold Storage: Copper Corrosion, Ammonia Retrofit
Cold storage facility in Xinjiang. LM-4 coolant. Copper corrosion discovered. The facility had been retrofitted from an ammonia system, and the legacy copper components were reacting with something in the coolant loop. Replacement cost was cited as high — that's the recurring theme with corrosion. Once it starts, fixing it is never cheap. The coolant was sampled for analysis, but the economic reality was already clear: replacing the damaged copper components would cost more than the coolant itself by an order of magnitude. The lesson: when you retrofit a system, you inherit the corrosion history of the old system. The old ammonia system might have left residues, surface pitting, or galvanic couples that the new coolant inherits. A thorough system inspection and cleaning during the retrofit could have caught this. It didn't happen, and the cost was significant.
April 28, 2025
Shandong Huatai: Hydrochloric Acid Enters LM-4 System
Hydrochloric acid leaked into the LM-4 cooling loop. This is about as bad as it gets for a water-based coolant. HCl is fully dissociated in water — every molecule that enters the system immediately releases a hydrogen ion that attacks any metal it can find. The pH drops. The inhibitors are consumed. The corrosion rate skyrockets. The response was to add a synergist and caustic to raise the pH, then monitor pH continuously. This is firefighting. You're neutralizing acid as fast as it enters, hoping the inhibitors can keep up. The fundamental problem — the HCl leak — has to be fixed at the source. Everything else is buying time. The facility was lucky: they caught the pH drop before the metal loss became catastrophic. But "lucky" is not a corrosion management strategy.
May 26, 2025
Fujian Cold Storage: Aluminum Tube Corrosion, LM-4 Leak
Aluminum evaporator tubes in a cold storage facility were leaking. The coolant was LM-4. Initial suspicion fell on aluminum corrosion caused by the coolant — aluminum is amphoteric, meaning it corrodes at both high and low pH, so it's a sensitive material in any aqueous environment. Samples were pulled and tested. The results showed something unexpected: no aluminum corrosion products in the coolant. The corrosion — if it was corrosion — wasn't coming from the aluminum tubes. Or it was a mechanical failure, not a chemical one. Plans were made to send samples to Dalian University of Technology for advanced analysis. The investigation was ongoing at year-end. This case is a reminder that not every leak is corrosion, and not every corrosion case is what it first appears to be. Sometimes you need a research lab to figure out what's actually happening.
June 10, 2025
China Aerospace: Heat Exchanger Failure — Twice in One Year
A heat exchanger failed after one year of service. It was replaced. The replacement failed four months later. That's two failures in sixteen months. The root cause was identified as oxygen ingress through the return line spray — the coolant was picking up oxygen from the air as it returned to the tank, and that dissolved oxygen was driving corrosion in the heat exchanger. The short-term fix was to seal the return line — eliminate the spray, eliminate the oxygen pickup. The long-term recommendation was to switch from LM-8 to LM-4, which has better corrosion inhibition for systems with oxygen exposure. This case is a perfect example of why reactive approaches fail: the first failure should have triggered a root cause investigation, not just a parts swap. The cost of that missed investigation was a second heat exchanger, more downtime, and a facility that lost confidence in its cooling system.
June 18, 2025
Wanwei Group: Glycol Corrosion, PV Material Contamination
An ethylene glycol-based coolant was corroding system components and carrying polyvinyl (PV) material contamination through the entire loop. The glycol itself was chemically attacking the PV components, and the resulting corrosion products were circulating everywhere — through the pump, through the heat exchanger, through the process equipment. The recommendation was to switch to LM-4, which offers better material compatibility with the polymers and elastomers commonly found in chemical processing systems. This case highlights a subtlety that's easy to miss: the coolant that works for one system might be completely wrong for another, not because of temperature or viscosity, but because of what it does to the materials it touches. Material compatibility is not a secondary consideration. It's a primary one.
Pattern recognition: what the seven cases have in common
Look at those seven cases together and a pattern emerges that's almost too obvious to state: in every single case, the corrosion was detected after it had already caused damage. Not before. The coolant changed color, the heat exchanger failed, the pH dropped, the system leaked — and that's what triggered the investigation. Nobody caught the corrosion early. Nobody prevented it. The damage was done, and then the response began.
The second pattern: contamination was the trigger in the majority of cases. HCl ingress at Shandong Huatai. Ammonia cross-contamination at Sanchuan. Oxygen ingress at China Aerospace. Glycol-material interaction at Wanwei. The coolant formulations themselves weren't failing — the systems were being contaminated by external factors, and the coolant was the canary in the coal mine. If you're blaming the coolant for a corrosion problem that was caused by contamination, you're blaming the canary for the gas leak.
The third pattern: the cost of corrosion always exceeds the cost of prevention. Always. The Xinjiang cold storage facility faced high replacement costs. The China Aerospace facility lost two heat exchangers in sixteen months. The XX data center operated with degraded cooling while waiting for lab results. Add up the downtime, the replacement parts, the labor, the lost production, and the elevated risk — and compare it to the cost of a proper monitoring program. It's not close. The monitoring program costs a fraction of a single failure. The math is so one-sided that the only explanation for not doing it is that nobody has done the math.
Key Insight from 2025
Corrosion in secondary refrigerant systems is almost never caused by the coolant formulation itself. It's caused by what gets into the coolant — oxygen, acids, cross-contamination, incompatible materials. The coolant is the messenger. Stop blaming the messenger. Start controlling what enters the system.
The new approach: predictive corrosion management
So what does the alternative look like? It looks like monitoring. Not quarterly checks. Not annual lab samples. Continuous or near-continuous monitoring of the parameters that predict corrosion before it happens. The shift from reactive to predictive is not a technology problem anymore. The sensors exist. The data loggers exist. The analytics exist. The barrier is organizational — someone has to decide to do it, and someone has to pay for it. But the payback is so fast that the decision should be easy.
The most important parameter is pH. In a water-based secondary refrigerant like LM-4 or LM-8, the pH is buffered by the inhibitor package. A drop in pH means the inhibitors are being consumed — either by normal aging or by acid contamination. If you catch the pH drop early, you can add inhibitors or address the contamination before the metal starts dissolving. If you wait until the coolant changes color, you've already lost that battle. The pH drop is the early warning. The color change is the obituary.
Conductivity is another early warning. A sudden rise in conductivity often means dissolved metal ions — corrosion products — are building up in the coolant. It can also mean contamination from an external source, like a process leak. Either way, it's a signal that something has changed, and that something is probably not good. A conductivity trend that's been flat for six months and then starts climbing is a red flag. Ignore it at your peril.
Dissolved oxygen is harder to monitor continuously but worth measuring regularly. The China Aerospace case is a textbook example: oxygen dissolved in the coolant from an open return line, and that oxygen drove corrosion in the heat exchanger. A dissolved oxygen measurement would have caught the problem before the first heat exchanger failed. Not after the second one. Dissolved oxygen sensors have become more reliable and more affordable in recent years. There's no technical reason not to use them.
For water-free systems, the parameter to watch is water content. The ice blockage cases showed that even 47 ppm of water in LM-11D can cause problems. But water in a water-free system also creates corrosion risk — water plus whatever contaminants are present can create localized corrosion cells even in a predominantly non-aqueous environment. Monitor water content. Stay below 50 ppm if you can. Below 100 ppm at the absolute worst. And if you see the water content trending upward, find the source. Don't wait for the ice blockage to tell you there's a problem.
The monitoring doesn't have to be expensive. A basic pH and conductivity monitoring system for a water-based loop might cost a few thousand yuan. Compare that to the cost of a heat exchanger replacement, which might be tens of thousands or more, plus downtime. The return on investment is measured in weeks, not years. The only question is whether anyone bothers to do the calculation.
Technology enablers: what makes predictive possible now
Ten years ago, continuous monitoring of coolant chemistry was expensive and unreliable. pH probes drifted. Conductivity sensors fouled. Dissolved oxygen meters required frequent calibration. The data was noisy, the maintenance was high, and the confidence in the readings was low. It wasn't practical for most industrial facilities. The juice wasn't worth the squeeze.
What changed? A few things. Online sensors got better and cheaper. Solid-state pH probes with longer life and less drift became available. Conductivity sensors with self-cleaning capabilities reduced the fouling problem. Automated sampling systems became available that pull a small coolant sample, measure it, and return it to the system — no waste, no manual handling, no exposure risk. Data logging and cloud connectivity mean you can trend your coolant parameters over months and years, not just spot-check them. You can set alerts. You can see patterns. You can catch problems before they become failures.
The technology isn't exotic anymore. It's commodity hardware with good software behind it. The sensors are the same ones used in water treatment plants and boiler systems — mature technology, proven in the field, available from multiple suppliers. The data loggers are off-the-shelf industrial controllers. The cloud platforms are standard IoT infrastructure. Nothing about this is experimental or cutting-edge. It's just not yet standard practice in the secondary refrigerant industry. But it will be.
The other enabler is simply awareness. The seven corrosion cases from 2025 — and the earlier cases from previous years — have made it clear that reactive corrosion management is economically irrational. You can spend a small amount on monitoring or a large amount on repairs. Those are the options. There is no third option where you spend nothing and nothing bad happens. The bad things are going to happen. The question is whether you find out about them early, when you can do something about them, or late, when the damage is done.
What's next: 2026 and beyond
I think we're going to see a few things happen in the next year or two. First, more facilities will adopt basic continuous monitoring — at minimum, pH and conductivity on water-based systems, and water content on water-free systems. The cost has dropped to the point where it's hard to justify not doing it. The economics are too compelling to ignore. A facility that spends 50,000 yuan on a heat exchanger repair but won't spend 5,000 yuan on a monitoring system is making a decision that's hard to defend in a budget meeting.
Second, I expect more attention to system design for corrosion prevention. The China Aerospace case showed that a simple open return line can destroy a heat exchanger in months. Nitrogen blanketing on expansion tanks — already recommended for water-free systems — will become standard practice for water-based systems too, especially in facilities where oxygen ingress is a risk. The cost of a nitrogen blanket system is negligible compared to the cost of the corrosion it prevents. The engineering is simple. The benefits are proven. The only barrier is awareness.
Third, the coolant formulations themselves will continue to evolve. Glacier Coolant's LM-4 series already has strong corrosion inhibition. The LM-4D variant is specifically formulated with low conductivity and no nitrite, making it suitable for systems with copper components — copper is sensitive to nitrite-based inhibitors, and the LM-4D formulation avoids that problem. The LM-4D-YE variant pushes conductivity below 800 μS/cm — that's useful for applications where ionic contamination is a concern, such as electronics cooling or systems with sensitive instrumentation. These aren't random product variations. They're responses to specific corrosion challenges that customers have encountered in the field. The product line evolves because the problems evolve.
Fourth, I think we'll see more integration between coolant chemistry data and overall facility management systems. If your coolant pH starts trending down, your building management system should alert you — not just log the data point and wait for someone to notice it on a quarterly report. The technology exists. The protocols exist. The integration just needs to be done. A facility that runs a sophisticated BMS for HVAC but doesn't monitor its process cooling loops is monitoring the wrong things. The process cooling loop is where the expensive failures happen.
Fifth, and this is more speculative, I think we'll start seeing predictive analytics applied to coolant chemistry data. Instead of just alerting when pH drops below a threshold, the system will predict when the pH will drop below the threshold based on the trend. Instead of just reporting conductivity, the system will correlate conductivity changes with other parameters — temperature, flow rate, production schedule — to identify the likely source of contamination. This is what predictive maintenance looks like for rotating equipment, and it's coming to cooling systems. The data is there. The algorithms exist. The integration is the next step.
The bottom line: corrosion is not inevitable. It's predictable. It's preventable. The seven cases from 2025 all could have been caught earlier. Most of them could have been prevented entirely with better system design and monitoring. The industry is moving in that direction. The question is whether your facility will be ahead of the curve or behind it. The cost of being behind is measured in failed heat exchangers, lost production, and emergency repairs. The cost of being ahead is measured in sensors and software. The math is not complicated. It just needs someone to do it.