The temperature gauge is the most trusted instrument on a diesel dash, and for the failure that quietly destroys the most engines it is completely blind. Cylinder liner cavitation erosion happens at normal operating temperature, on a system that is full, with a fan that never needs to cut in. It is not a heat problem at all — it is a chemistry problem, and the only thing standing between it and a hole in a liner is an additive package that depletes with use. Cummins Filtration recommends testing coolant every 500 hours or three months, which is an interval most fleets simply do not have on the schedule. The radiator side matters too, but it fails in ways you can actually see. See a coolant test logged against an engine.
Your Coolant Is a Maintained Chemical, Not a Fluid You Top Up
For fleet maintenance managers, diesel technicians and anyone whose engines keep failing with the gauge reading normal: what cavitation erosion actually is in the engine maker's own words, the four things a coolant test measures, and why topping up with water is worse than being slightly low.
The Failure the Gauge Cannot See
Start with the mechanism, because once you have it the whole maintenance argument follows. This is Cummins describing what happens inside a wet-liner diesel every time it fires.
"Piston-slap causes liner vibration which creates a vacuum and formation of tiny vapour bubbles at the liner surface. As the liner slams back in the coolant the tiny bubbles implode with high pressure."
— Cummins, Coolant and Chemicals Catalogue
Read that again and notice what is missing: temperature. The bubbles form because the liner moves, not because the coolant is hot, and they implode against the metal thousands of times a minute at every engine speed. Each implosion removes a microscopic amount of material. Over enough hours it produces the pinholes that let coolant into a cylinder — and the only defence is a chemical film maintained by the additive in the coolant, which is consumed doing exactly that job.
Cummins publishes a normal operating temperature above 180°F for the X15, with the fan engaging at 210°F and a high-temperature alert at 225°F. The erosion described above proceeds across that entire range. A driver watching the gauge sees a healthy engine, a technician checking the level finds it full, and the liners keep thinning — which is why this failure is almost always found at teardown rather than at inspection. Check the figures for your own engines; these are one manufacturer's for one model.
The Radiator Side, Which Does Give You Warning
Everything above is invisible. The airflow half of a cooling system is the opposite — it fails visibly and gradually, and almost all of it is found by looking. Swipe on mobile.
| What fails | How it presents | What to do |
|---|---|---|
| Core blockage from the outside | Chaff, insects, dust and oil film bridge the fins and stop air passing. The tell is overheating under load while idling stays normal, because at idle the fan is doing the work and airflow demand is low. | Backflush the core with low-pressure water from the engine side out, and never with a pressure washer — bent fins are permanent |
| Internal scaling and deposit | Hard water, over-concentrated additive or a spent inhibitor package leaves deposit on the inside of the tubes. Deposit insulates, so the coolant cannot give up heat even when everything else is right. | Prevented by chemistry rather than cleaned by tools. This is the radiator paying for a coolant programme that was not run |
| Fan clutch not engaging fully | A viscous or air-actuated clutch that slips gives you most of the airflow most of the time, which is enough to look fine and not enough on a grade in summer. | Check engagement under load and listen for the fan coming in; a clutch that never roars is a clutch to test |
| Pressure cap failing | The cap sets system pressure, which sets the boiling point. A weak cap lowers the boiling point and the coolant flashes to steam in the hottest passages while the gauge still reads under the limit. | Cheapest part in the system and the most commonly overlooked; pressure-test it rather than eyeballing the seal |
| Charge air cooler and radiator fighting for air | On most tractors they share the same airflow path, so a blocked CAC raises intake temperature and a blocked radiator raises coolant temperature, and either one loads the other. | Inspect both as one assembly, because cleaning one and not the other only moves the restriction |
| Shutters, seals and shrouding | Missing baffles or perished seals let air bypass the core entirely. The fan still turns, the airflow still exists, and almost none of it goes through the fins. | A visual check that takes a minute and is almost never on a PM sheet |
The pattern worth internalising: a cooling system that overheats under load but not at idle is almost always an airflow or flow-restriction problem, while one that overheats everywhere is more often coolant level, a thermostat or a water pump. Testing at idle in the yard is how airflow faults get signed off as fine.
What a Coolant Test Actually Measures
Four parameters, and the first one is the whole reason this is a scheduled task rather than a visual check. Swipe on mobile.
| Parameter | Why it matters | How it is measured |
|---|---|---|
| Nitrite or SCA level | The additive that protects the liner. Depleted additive is the actual failure, and it happens silently on a system that is otherwise full and cool. | Test strip, against the range for your coolant |
| pH | Coolant that has gone acidic attacks everything it touches — solder, gaskets, aluminium. A reading outside range means the inhibitor package is spent. | Test strip or lab sample |
| Freeze point / glycol concentration | Too little glycol loses freeze and boil protection; too much loses heat transfer, which is the counter-intuitive one people get wrong when topping up. | Refractometer, not a floating-ball tester |
| Contamination | Oil, fuel or combustion gas in the coolant changes the problem entirely and means something upstream has failed. | Visual, plus a sniff test and a lab sample if suspected |
Cummins Filtration recommends testing "at least twice per year" and states a preference for "every 500 hours or 3 months." For supplemental coolant additive it gives an ideal target of 0.7 units per litre. Ranges vary by coolant technology, so read the acceptable window off the test kit for the product actually in the engine rather than working from a number you remember.
Four Things Not to Do to a Cooling System
Every one of these is common, well-intentioned, and makes the situation worse than leaving it alone would have.
- Do not top up with waterCummins is blunt about it: “DO NOT ADD WATER as this will dilute the chemical balance.” A litre of water in a protected system is a litre of unprotected coolant, and it does not mix back to spec.
- Do not mix coolant technologiesConventional, extended-life and nitrite-free chemistries protect by different mechanisms. Blending them can produce a mixture that tests inside range on one parameter and protects against nothing.
- Do not use automotive coolantLight-vehicle coolant is not formulated for a wet-liner diesel, because a car engine does not have the cavitation problem this page is about. It is the cheapest way to destroy an expensive engine slowly.
- Do not add SCA to extended-life coolantUnless the manufacturer specifically calls for it. Over-concentrated additive drops out of solution and becomes a deposit, which insulates the surfaces it settles on.
Does your PM schedule have a coolant test on it at all?
Most do not — there is a level check, and a flush interval somebody half-remembers. In thirty minutes we will put the 500-hour and three-month clocks on one engine, with nitrite, pH and freeze point stored as values so the second test means something.
How FleetRabbit Handles This
Four things, and the third is the one that catches an internal leak months before it becomes a teardown.
Hours and months tracked in parallel per engine, so a coolant test comes due on its own rather than when somebody remembers there is such a thing.
Nitrite, pH and freeze point held as numbers per unit. Three tests make a trend; three ticks make nothing.
A litre added here and there is invisible until you can see it totalled. A unit quietly consuming coolant is a diagnosis waiting to be read.
Service life is measured in years and hours as well as miles, so the replacement interval is tracked against whichever clock arrives first.
Symptom to Cause
Six findings, what usually sits behind each, and where to start. The last one is not a diagnostic exercise.
Head gasket, liner seal, or an internal leak into the oil or a cylinder
Pressure-test cold, then check the oil and the exhaust before chasing hosesBlocked radiator core, a fan clutch not engaging, or a partially collapsed hose
Load-test rather than idle-test — an idling engine hides all threeThermostat sticking, or low coolant letting the sensor see air intermittently
Level and cap first, thermostat secondThermostat stuck open, which is not harmless — a cool diesel makes more soot and dilutes its oil
Fit the correct-temperature thermostat, not a cooler oneAdditive package spent, or an oil cooler leaking
Test before flushing, because the flush destroys the evidenceCoolant reaching a combustion chamber
Stop running it. This is the failure the chemistry was protecting againstNote the recurring instruction to test or measure before flushing or replacing. A flush removes the evidence along with the contamination, and a fleet that flushes on suspicion never finds out what was actually wrong — which is how the same engine comes back.
What Goes on the Schedule
Five tiers, and the third is the one most fleets are missing entirely.
- Daily, by the driverLevel at the sight glass or tank, any visible leak, and the temperature behaving normally through the shift — including whether it reaches temperature at all.
- Every PMCap pressure-tested, core inspected front and back for blockage and fin damage, hoses squeezed for softness and checked at the clamps, belt condition and tension, fan and shroud intact.
- Every 500 hours or 3 monthsThe coolant test: additive or nitrite level, pH, freeze point, and a look for oil or contamination. Recorded as values rather than as a tick.
- SeasonallyBefore summer, airflow — core, fan clutch, shutters and seals, with the charge air cooler inspected as part of the same assembly. Before winter, freeze point confirmed by refractometer.
- At the coolant's service lifeDrain, flush and refill with one known product, against whichever of the hours, years or distance limits arrives first rather than whichever is most convenient.
Questions Diesel Fleets Ask
How often should diesel coolant be tested?
Cummins Filtration recommends at least twice a year, and states a preference for every 500 hours or three months. The hours clock is the one that matters on equipment that idles or works short runs, because it arrives far sooner than any mileage-based interval would. Confirm against your engine maker's own guidance. Track hours and months free.
Can I top up with water in an emergency?
To get a truck off the road safely, yes — but treat it as a fault that now needs correcting, not a job finished. Cummins is explicit that adding water dilutes the chemical balance, and dilution does not correct itself. Record how much went in and get the system tested and brought back to specification. Ask how we log top-ups.
What is cylinder liner pitting?
Cavitation erosion. Piston slap vibrates the liner, which forms vapour bubbles in the coolant that then implode against the metal with high pressure, removing a little material each time. It happens at normal operating temperature and is prevented chemically rather than thermally, which is why coolant additive level is an engine-protection measurement and not a housekeeping one. Log your first test free.
Is more antifreeze better?
No, and this is the one that catches people topping up. Too little glycol loses freeze and boil protection, but too much reduces heat transfer, so an over-concentrated system runs hotter than a correct one. Measure with a refractometer and target the specified ratio rather than assuming more is safer. Review your mixing practice.
Can I mix extended-life and conventional coolant?
Avoid it. They protect by different mechanisms, and a blend can read acceptably on one test parameter while protecting properly against nothing. If a system has been mixed, the honest answer is usually to drain, flush and refill with one known product rather than to test your way to confidence. Standardise across the fleet.
How long does extended-life coolant last?
Cummins quotes up to six years, 1.5 million kilometres or 36,000 hours for its longest-life product, and five years, one million kilometres or 20,000 hours for another — note that all three limits are stated, and the first one reached is the one that counts. Products differ substantially, so use the figures for what is actually in your engines. Match intervals to your product.
What does cooling system maintenance cost?
A test strip and a few minutes, several times a year, plus coolant at its replacement interval. Set against that: a liner and a rebuild. The reason the comparison is so lopsided is that nothing warns you which side you are on, which makes this one of the clearest cases in the whole maintenance programme for spending on schedule rather than on symptoms. Put it on a schedule free.
Test the Coolant While the Gauge Still Reads Normal
FleetRabbit runs the 500-hour and three-month clocks per engine, stores nitrite, pH and freeze point as values you can trend, totals the coolant top-ups nobody raises a job for, and keeps the replacement interval against hours, years and distance at once.
Free for 3 vehicles · no credit card required
