Stamping, Forging, Extrusion, or CNC? A TRW Guide to Automotive Metal Parts

Every time a drawing lands in my inbox, I do the same thing. I check material, tolerances, annual volume, and critical characteristics. Only then do I start talking about stamping, forging, extrusion, or CNC. The reason is simple: the right process changes with the part.

I'm a quality/compliance manager at an automotive metal forming company that supplies components under the TRW brand. I review around 200 part numbers a year—stamping parts, dies, forging blanks, aluminum extrusions, and CNC-machined components. I've rejected 6% of first-run deliveries in 2025 alone. Not because the parts looked bad. Because they didn't match the drawing.

One caveat before the decision tree: my experience is based on mid-to-high-volume automotive production. If you're sourcing a one-off aerospace bracket, stop reading. This guide won't transfer.

Three questions before the process choice

There is no universal 'best' process. There is only a best match. Ask these three questions, in this order:

• Volume. How many parts per year? 500? 50,000? A million?
• Function. Does it carry load, shield heat, seal fluid, or reduce noise?
• Geometry. Is it a flat blank, a deep draw, a long constant section, or a near-net shape with hidden cavities?

Volume. Function. Geometry. Get those three right, and the process usually points to itself.

Scenario 1: High-volume sheet metal and consistent geometry? Progressive die stamping

You're in this branch if the part is a cover, shield, bracket, or shell made from sheet metal, and the volume is high enough to amortize a die. The classic example is a catalytic converter cover. It's a thin heat shield, usually stainless or aluminized steel, and it's made in huge volumes. A progressive die stamps, pierces, bends, and cuts it in one continuous press run. Per-piece cost is low. Tooling cost is not.

The same thinking applies to a resonator for muffler. Most resonator shells are two stamped halves, formed from sheet stainless steel and welded together. The process decision isn't about strength—it's about controlling gauge and surface finish at scale.

Counterintuitive point: don't automatically 'upgrade' a stamped resonator to a forged one because forging sounds stronger. Forging is for load paths, not noise attenuation. A stamped double-wall resonator does the job better and cheaper.

Before you cut steel, the PPAP Level 3 workbook needs to exist. The first-article report is the cheapest insurance you'll buy. Skip it, and you're betting a week of press time on memory.

Scenario 2: Safety-critical or load-carrying? Forging or heavy stamping

Now we're talking about brake and chassis components. A brake caliper assembly—a TRW caliper is a good example—is not one process. The caliper bracket can be forged to get continuous grain flow; the spring and dust shield are stamped; the piston is machined. When you source metal for a caliper bracket, the load path matters more than the price per kilo.

You're in this branch if the part will carry load, face fatigue, or affect safety. A forged bracket has better fatigue behaviour than a stamped-and-bent bracket made from the same material. That's not marketing. That's the reason the drawing specifies a forging instead of a stamping.

But don't reverse the rule. Forging is not an automatic quality upgrade. If the part is a non-structural cover, a forging just adds draft angles, machining, and cost. I've rejected forged parts that were 'strong enough' but lacked the dimensional consistency the drawing required. Strength without geometry is still a failed part.

Scenario 3: Long, constant cross-section and moderate volume? Aluminum extrusion

Aluminum extrusion is the most underused branch in automotive metal buying. People think extrusion is for window frames and handrails. The reality is that vehicle structural rails, battery tray side members, and heat shield carriers are often extruded profiles. Extrusion gives you a consistent cross-section, good stiffness-to-weight ratio, and a much lower die cost than a progressive stamping die.

You're in this branch if your part is long, has the same cross-section for its entire length, and needs to be light. Instead of stamping and bending a U-channel in several operations, one extruded profile can do the same job with fewer welds.

The catch: extrusion cannot change section along its length. If you need a hole, slot, or machining after 200 mm, those add secondary operations. That's still often cheaper than stamping dies at low volume.

Scenario 4: Low volume, tight tolerances, complex features? CNC machining

CNC machining is the right branch when volume is too low to justify dedicated tooling and tolerances are too tight for forming. A thermostat housing for a replacement application is a good example. Annual volume of 500? Machine it from billet or a near-net extruded blank. Don't design a progressive die for a part that will never see a run rate.

The per-piece price is higher—or rather, the unit cost is higher, but the total cost is lower because you don't amortize a six-figure tool. CNC machining also gives you the luxury of design changes without buying new hard tooling.

But CNC is not the default 'premium' answer. If your part is a simple flat blank at 200,000 parts a year, stamping will beat CNC on both cost and cycle time. Precision is not a workflow; it's a design requirement.

How to tell which branch you're in

Walk this checklist in order:

1. Is the cross-section constant and the length long? Try aluminum extrusion before you design a stamped multipiece assembly.
2. Is the annual volume above 50,000 and the material sheet? Put progressive die stamping at the top of the list.
3. Is the part safety-critical or fatigue-loaded? Add forging or a billet/forged-machined route, and run the FEA.
4. Is the volume low and the tolerance tight? Go to CNC machining and skip the tooling debate.

That's my mental model. It may not match your context if you're working with unusual materials, ultra-low volume, or a customer-mandated process. Use it as a starting point, not a crystal ball.

The source of truth, the TRW login, and the real cost

Before you choose any process, pull the approved drawing. If you're working with an existing part, the supplier portal—call it the TRW login if that's the system you use—will show the current revision level. I've caught two wrong revisions this year by checking the portal before making tooling. The drawing is not a suggestion. It's the contract.

I've also watched teams chase the lowest per-piece quote and ignore the control plan. One quote saved us $0.04 per catalytic converter cover. Then the surface defect appeared in storage. We reworked 8,000 covers and spent $14,000 in freight and labour. That's the 'penny wise, pound foolish' lesson I keep relearning.

Prevention over cure

People think expensive processes deliver better quality. Actually, matching the process to volume, function, and geometry delivers quality. The expensive process is the one that's wrong for the part, because you pay for it every single cycle.

That's why a 12-point checklist matters. Five minutes of verification beats five days of correction. In the IATF 16949 structure, a control plan lists the measurement, method, and frequency for every critical dimension. It exists so the next shift doesn't have to rely on memory. I'd rather see an engineer check the drawing, confirm the revision, and ask one clarifying question before production than hear 'we thought it was close enough' after 8,000 pieces.

Oh, and one more thing: if your search brought you here through 'how do you program a thermostat'—if you mean a home HVAC thermostat, I can't help. If you mean an automotive cooling thermostat, the thermostat itself is a calibrated valve, not a programmed part. The metal housing around it follows the same rule we've been using: volume, function, geometry. Match those, and the process comes into focus.