Radiator and heat exchanger control with AI — zero leaks reaching the OEM.

The problem

The leak that matters is not the one you see on the test bench — it is the one the end customer finds under warranty.

Radiators and heat exchangers (engine cooling, HVAC, power electronics in EVs) have a cruel failure profile: the micro-leak that does pass the quick test but fails at 30,000 or 50,000 km. By then it is already in the car of the OEM's customer, and it is a warranty campaign.

1. Assembly — tubes, fins, headers. A deformed fin (fin damage) cuts thermal efficiency; a badly positioned tube complicates brazing. The operator inspects by sampling and the line does not stop.
2. The brazing furnace — the flux melts, the alloy fills, the joint seals. The furnace curve is critical: peak temperature, time in zone, atmosphere. A small drift goes unnoticed until the leak pattern appears.
3. Leak tightness testing — pressure decay to rule out gross leaks, helium for micro-leaks. The result lives inside the test equipment and often is never cross-referenced with the furnace curve of the same batch — two islands holding data that complement each other.
4. The OEM customer — warranty campaigns for cooling or HVAC failure are extremely expensive, damage the contract and bring PPM escalation.

The classic system (operator + checklist + final test) no longer cuts it, because the data that determines reliability at 50,000 km is the furnace data, and that data lives on a different island from the test. The veteran operator senses it, but cannot watch the thermocouples and the helium reading at the same time.

How it fits the IRIS system

iLEAN cross-references the furnace, the test and the line — without asking you to change anything.

Radiator control is not a problem of installing more sensors: it is a problem of cross-referencing the ones you already have. iLEAN is the putty that fills the crack between the brazing furnace curve and the leak test result of the same batch, without asking you to change the furnace, the helium equipment, the MES or the ERP.

Edge sees every assembly before brazing. Connect reads the furnace curve and the leak test at second zero. The agent cross-references with the batch and holds any part in doubt. The person signs — never the other way round.

The three iLEAN pieces applied to radiator and heat exchanger control:

• Edge — a terminal with machine vision (CNN) over the assembly line. It inspects 100% of assemblies before the furnace: fin damage, tube positioning, headers. It ejects the part in milliseconds. It works without a network.
• Connect — captures the brazing furnace curve (thermocouples by zone, atmosphere, conveyor) over OPC-UA or Modbus, the leak test result from the helium or pressure-decay equipment, and the tube and flux delivery notes by email. It unifies everything at second zero.
• Agents — continuously cross-reference the furnace curve with the leak test result in each batch, detect drift before the pattern turns into an OEM escalation, and build the file by batch and by VIN, ready for IATF 16949.

See the full IRIS architecture →

Before and after

Control in islands vs. cross-referenced control with iLEAN

Aspect	Classic control (every machine on its own island)	With iLEAN Edge + Connect + Agents
Inspection before brazing	Visual sampling	100% inspected in line, in milliseconds
Furnace curve × leak test	Never cross-referenced	Cross-referenced continuously, drift caught earlier
Tube / flux batch	Paper delivery note, never reaches the furnace lead	Captured by Connect, linked to the batch
Traceability by VIN	Manual reconstruction when an incident hits	Automatic, continuously maintained file
Operation without a network	n/a	Edge keeps inspecting on panel power
IATF audit / warranty campaign	Defense from a standing start	Data ready, and the conversation with the OEM changes

Impact estimate

Impact estimate for your plant — to be validated with your numbers.

The block below is an estimate to be validated with the specific data of your plant. We put it forward so the committee has an order of magnitude; we refine it during the diagnostic.

• Radiator or heat exchanger plant, one assembly line + brazing furnace + leak test equipment (pressure decay and/or helium), European OEM under IATF 16949.
• Edge pilot before brazing + Connect reading the furnace curve and the test result + an agent that cross-references with the batch and builds the file by VIN. First value expected within a few weeks.
• Indicative payback between 4 and 9 months. The hard lever is a single warranty campaign avoided and the reduction in in-line rework from fin damage.
• Reduction in late-detected micro-leaks against the baseline of typically ≥30% — to be confirmed after measuring the "before".

And the OEM standard

The automotive quality standard for critical components sits on the order of 25 PPM^[1]. Holding that without cross-referencing the furnace with the test is not viable in the medium term — the micro-leak is found by the end customer, not by the quality team. AI in anchored tasks (classifying a furnace curve against the catalog, reading the helium result and comparing it) has error rates below 1.5% in the best models^[2], and even so it does not decide alone — it holds the part and asks for a human signature.

^[1] Symestic — automotive quality standard on the order of 25 PPM for critical components.

^[2] OpenAI paper "Why Language Models Hallucinate", 2025 — on the reliability of AI in anchored tasks.

Related automotive topics: Exhaust pipe and catalytic converter control · Water pump control · Car air conditioning control.