EV Heater Core Housing Sheet Metal Work

• 1
  Why an EV Heater Core Housing Is Not Just a Box
• 2
  Where Sheet Metal Work Reaches Its Limit
• 3
  Machined Sheet Metal: Form First, Then Cut the Critical Faces
• 4
  Material Choice for Heater Core Housing Sheet Metal
• 5
  How to Prove the Housing Will Not Leak
• 6
  Which Process Fits Which Housing Feature

A heater core housing sits between the coolant loop and the cabin air path. Coolant at 60–90 °C enters through a port, passes a core or a PTC element, and leaves. The housing carries that fluid, holds the seal, and bolts to the HVAC module. It also has to survive vibration on a chassis mount and thermal expansion every time the cabin calls for heat.

That combination is what makes the part awkward. Sheet metal gives you thin walls, low mass, and a shape that follows the air duct. It does not naturally give you a flat sealing face, a leak-tight corner, or a threaded boss strong enough to torque a fitting. Those features have to come from somewhere else.

The pressure is small. Most coolant loops run at 1–2 bar, and a burst requirement might sit near 3 bar. Leakage is rarely caused by pressure alone. It comes from a seal face that is not flat, a weld that pulls as it cools, or a flange that moves when the bolt torque is applied.

So the real question is not whether sheet metal can make the shape. It is whether the forming sequence can hold the geometry that the seal and the fasteners depend on over the life of the vehicle.

Laser cutting and press braking are fast and cheap per part. A 1.5 mm 5052 or 304 blank cuts in seconds, and a bend takes one hit. For a simple cover that is the right process. The trouble starts at the features that control function.

A bend has a bend radius, and the inside radius has to be at least roughly the material thickness for a clean result. Springback moves the final angle by 1–3° depending on alloy and temper. Hole-to-bend distance below 2.5 times the thickness distorts the hole. These are not defects. They are the physics of the process.

Welding makes it harder. A seam weld on a 1.5 mm wall pulls the flange as it cools. On a 200 mm long seal channel, a 0.3 mm pull is enough to break contact across the gasket. Straightening after welding adds another operation and another chance to move something that was already correct.

We see the same pattern in most projects. The first prototype is welded and hand-fitted, and it seals. Production tooling removes the hand fitting, and the leak rate climbs. The gap between a one-off part and a repeatable part is where the cost lives.

The practical answer is to treat the formed shell as a pre-form, not as a finished part. Bend and weld the shell to a rough envelope, then fixture it and machine the features that must be accurate. The sheet metal carries the shape and the weight. The machining carries the tolerance.

After forming, we locate the shell in soft jaws or a vacuum chuck. From that single setup, a 5-axis machine can face the seal channel, bore the port, drill and tap the mounting holes, and spot-face the boss seats. Datum control stays in one place instead of moving between a brake, a weld table, and a drill press.

This is where the tolerance numbers matter. A gasket needs a flat face, and flatness is what the seal actually reads. We hold ±0.005 mm on machined features and keep seal faces at Ra 0.8–1.6 μm. A rougher Ra 1.6–3.2 μm face is fine for a bracket, but a gasket will not follow the peaks.

There is a cost trade. Every machined feature adds cycle time. The rule we use is simple: machine only what touches a seal, a fastener, or a bearing. Everything else stays as-formed. That keeps the part competitive without giving up the leak-tight face.

Aluminum is the default for mass. 5052 and 5083 form well and resist corrosion, which suits a coolant-wetted housing. 6061-T6 bends less cleanly at tight radii, so we usually specify it for machined flanges or bosses rather than deep formed shells.

Stainless 304 and 316L are the choice when the coolant chemistry is aggressive or when the housing also sees road salt. They weld cleanly and hold a seal face well. The penalty is springback, which is larger than aluminum, and tool wear on the machined features.

Wall thickness is usually set by stiffness, not by pressure. A 1.5 mm wall is common; a 2.0 mm wall appears on larger housings or where the part carries the HVAC module. Going thicker rarely fixes a leak. A flat machined face does.

For prototype volumes we also run machined 6061 or 6082 bodies and, on some programs, die-cast ADC12 with machined sealing faces. If the housing is small and the quantity is low, a fully machined body can beat a welded assembly because it removes the weld seam entirely.

A caliper tells you a length. It does not tell you whether a seal channel is flat, whether two faces are parallel, or whether a port is round. Those are the dimensions that decide whether the part works, so they need the right instrument.

We check flatness on a coordinate measuring machine or with a surface plate and indicator. Port diameter and position come off the CMM report. Surface finish on the seal face gets a profilometer reading, because Ra drives gasket contact. Threads get a go/no-go gauge, and torque is verified on the fixture.

In-process monitoring catches drift before it becomes scrap. If a forming die wears, the flange angle moves slowly. Measuring the pre-form before machining lets us adjust the setup instead of discovering the problem at final inspection.

Every part gets 100% inspection before shipment, and reports are available on request. For a safety-adjacent thermal part, that is the minimum. A sampling plan on a seal face is a bet nobody wants to take.