The Pain Point: The Leak That Appears at 40,000 Kilometres

A radiator that passes a 2 bar pressure test at the end of the line can still fail in the field within a year. The pattern is familiar: no leak at delivery, a slow weep at a tube-to-header joint by 30,000 km, and a warranty claim that nobody can reproduce on the bench. The part is returned, cut open, and found to have a hairline crack surrounded by a white deposit.

The failure is almost never a single defect. It is the convergence of three things that were each individually acceptable: a brazed joint with slightly excessive filler diffusion, a coolant chemistry that was maintained at the minimum specification, and a thermal cycle count that exceeded the design assumption. Aluminium parts for radiators live in a hostile environment - hot glycol on one side, salt-laden air and stone chips on the other, and continuous vibration from the road - and they are made from walls measured in tenths of a millimetre.

The engineering problem is that all three failure drivers are invisible at the point of manufacture. A supplier can deliver a dimensionally perfect core that will still leak, because the property that matters is metallurgical rather than dimensional.

The Principle: Passive Film, Brazed Joints and Thermal Cycling

The oxide film is both the protection and the obstacle. Aluminium is strongly oxyphilic; a fresh surface forms a dense passive oxide layer within milliseconds, which is why the metal survives years of exposure without corroding. The same film prevents filler metal from wetting the surface during brazing, which is why every aluminium radiator process must either remove the oxide chemically (flux brazing) or break it mechanically and chemically in a vacuum furnace.

Brazing creates a new alloy at the joint. A typical clad fin or tube carries a thin layer of silicon-rich alloy that melts at a lower temperature than the core. During the furnace cycle the cladding flows, dissolves a small amount of the core, and re-solidifies as a silicon-rich fillet. If the furnace profile is too hot or too long, the cladding diffuses into the core and the fillet becomes brittle; if it is too cool, the fillet is thin and incomplete. Both conditions pass a visual inspection and both leak later.

Thermal cycling is a fatigue problem, not a thermal one. A radiator header sees temperatures swinging from ambient to over 100 °C on every drive cycle. The tube, the fin and the header have different stiffnesses and slightly different expansion behaviour, so every cycle applies a small strain to the joint. A brazed fillet that is sound will tolerate hundreds of thousands of cycles; a fillet with a shrinkage void will initiate a crack in a few thousand.

The tube wall is the weakest section. Typical tube walls are 0.20 to 0.35 mm. At that thickness, a 0.05 mm corrosion pit removes fifteen to twenty-five percent of the load-bearing section. This is why internal corrosion control is a structural requirement rather than a maintenance preference.

Equipment Classification: Where Aluminium Radiator Parts Sit

Passenger car engine cooling radiators. Brazed aluminium core with mechanically crimped plastic end tanks or welded aluminium tanks. Highest volume, thinnest walls, most cost pressure.

Heavy-duty and commercial vehicle radiators. Larger cores, thicker tubes, often with reinforced headers and vibration-isolated mounting. Service life targets in the millions of kilometres.

Charge air coolers. Pressure cycling rather than temperature cycling is the dominant load. Bar-and-plate or tube-and-fin construction, with particular sensitivity to internal pressure fatigue at the tank transition.

Oil coolers and transmission coolers. Small stacked-plate cores, frequently exposed to higher internal pressures and to fluid compatibility issues.

Battery and power electronics cold plates. Extruded or machined aluminium with internal channels, sealed by friction stir welding or brazing, where flatness against the cell surface matters as much as leak tightness.

End tanks, headers and covers. The parts where ports, vents, seals and mounting features concentrate. These are the highest-risk components because they combine a pressure boundary with a sealing interface and a stress concentration.

Practical Guide: Designing for a Leak-Free Life

Specify the alloy by function. 3003 and 3005 for brazed tube and header stock where formability and corrosion resistance dominate; 6061 and 6063 for extruded tanks, side plates and machined connectors where strength and machinability matter; 7075 only where high strength justifies its lower corrosion resistance and its unsuitability for brazing.

Control the clad ratio. The cladding thickness determines how much filler is available. Too little gives incomplete fillets; too much gives erosion of the core and blocked coolant passages. State the clad ratio on the material specification, not just the alloy.

Design generous fillets at the tube-header joint. A sharp internal corner concentrates stress and traps flux. A small radius costs nothing and roughly doubles the fatigue life of the joint.

Avoid abrupt section changes in the tank. Pressure cycling cracks initiate where a flat wall meets a thick boss. Taper the transition and add a radius instead of a right angle.

Treat every port as a stress raiser. Inlet and outlet necks, drain ports and vent bosses all interrupt the pressure boundary. Place them away from brazed seams and provide local reinforcement.

Specify the leak test method and the acceptance criterion. A helium mass spectrometer test at 1 × 10⁻⁶ mbar·L/s detects what a bubble test cannot. Define the pressure, the dwell time and the temperature.

Define coolant chemistry in the manual. The radiator cannot control what is poured into it. Specify glycol type, inhibitor package, chloride limit and service interval - and design the internal surfaces on the assumption that the specification will be met only most of the time.

Field Experience: The Core That Passed Every Test

A tier-one supplier delivered a batch of brazed cores for a light commercial vehicle. All units passed the in-line pressure decay test, the burst test and the dimensional audit. Field returns began at about nine months, concentrated in a single climate region with high humidity and heavy road salt.

Sectioning showed the leaks were not in the brazed joints at all. They were in the tube walls, at the point where the tube passed through the header, on the air side. Salt spray had collected in the small crevice between the tube and the header slot, and corrosion had progressed inward until the wall perforated.

The root cause was a design detail, not a process failure: the header slot clearance had been specified at 0.15 mm to ease assembly, which left a capillary gap that held moisture indefinitely. The fix reduced the clearance to 0.05 mm and added a fillet that closed the crevice. In parallel, the coating specification was changed from a full-core paint to a selective coating that covered the header region only, because a full core paint had been blocking fin gaps and reducing heat rejection.

The lesson was that the pressure boundary and the corrosion boundary are not the same surface, and that a test which only checks pressure will never find a crevice.

Summary

Aluminium parts for radiators fail at the intersection of metallurgy, chemistry and fatigue. The passive oxide film that makes aluminium durable in air is exactly what makes brazing difficult; the thin walls that make the core efficient are exactly what makes it vulnerable to pitting; the thermal cycling that the radiator exists to manage is exactly the load that cracks its joints. Designers who treat these as three separate departments - materials, process and validation - will keep finding leaks at 40,000 km. Designers who treat them as one problem will specify clad ratio, fillet radius, crevice clearance and coolant chemistry as a single package.

Outlook

Two shifts will dominate the next decade. Electrification is moving aluminium from engine cooling to battery and power electronics cooling, where cold plates must combine leak tightness with flatness in the tens of micrometres and where the coolant is often a dielectric fluid with very different compatibility. At the same time, brazing furnaces are moving toward tighter profile control and in-line leak detection, because the cost of a field return now exceeds the cost of the core several times over. Suppliers who can document their furnace profile, their clad specification and their leak-test data per unit will increasingly be the only ones qualified for high-volume programmes.