In this guide: What Level 3 demands from a foundry · The real numbers · 7 fatal mistakes · The one-day audit · Elements that fail most · FAQ

China poured about 51.9 million tons of castings in 2023. That's roughly 47% of the world's foundry output, coming from around 26,000 individual factories. If you're sourcing a casting from China, those numbers sound like infinite choice.

Here's the number nobody puts in the brochure: in 12 years of buying automotive castings as a Tier-1 supplier, I've seen maybe 8 out of 100 Chinese foundries submit a clean PPAP Level 3 casting package on the first attempt. The castings usually pass. The paperwork is what kills it.

This guide is about that gap — and how you stop it from becoming your problem.

First, What "Level 3" Actually Demands from a Casting Foundry

PPAP Level 3 is not a certification. It's a submission level — the default one. When a customer writes "PPAP Level 3" on the RFQ, they mean: deliver the complete 18-element file package per AIAG PPAP 4th Edition, from design records all the way to the signed Part Submission Warrant.

For a casting, that file package is painful to build. Here's what happens when you're on the receiving end:

A foundry with OEM experience nods and says "we know PPAP." Two weeks later, you get a binder with 18 tabs. Tab 7's control plan contradicts the PFMEA in tab 6. The drawing revision on the PSW doesn't match the drawing in tab 1. The dimensional report in tab 9 was measured from a different datum setup than the print calls out, so half the critical dimensions are technically invalid.

I'm not describing a worst-case scenario. I'm describing a typical first attempt from a mid-size Chinese foundry.

A casting is genuinely one of the hardest products to do PPAP on. You're not documenting a machined part with stable tolerances. You're documenting a process with dozens of live variables: melt chemistry, sand properties, core placement, mold hardness, pouring temperature, cooling time, shakeout, trimming, heat treatment. Each variable adds a page to the control plan, a row to the FMEA, and at least one dimensional report. Get one variable wrong, and the whole file falls apart.

If the foundry tells you "PPAP is PPAP, every edition is the same," walk away. It's not, and they haven't been audited recently.

The Real Numbers: How Many Chinese Foundries Can Actually Deliver?

Let's be honest about the funnel — these are planning estimates based on sector data and my own audit experience, not audited statistics. But they've held up across 200+ foundries I've visited:

• ~26,000 foundries in China (China Foundry Association, 2023)
• ~8,000 to 10,000 produce automotive castings — roughly 30% of China's casting tonnage goes into vehicles
• ~2,500 to 3,500 hold IATF 16949 and are serious about automotive compliance
• ~400 to 600 have real, hands-on experience submitting PPAP Level 3 to European or US OEMs
• ~150 can do it cleanly on the first pass without a fire drill

That last number is the one that matters when your program's launch date is fixed.

Where the capable foundries sit

Geography matters in China more than most buyers expect:

• Zhejiang (Ningbo, Wenzhou, Taizhou): the biggest cluster for automotive aluminum die casting and investment casting.
• Jiangsu (Changzhou, Nantong, Suzhou): serious sand-cast ferrous foundries, plus precision machining nearby.
• Shandong and Hebei: the iron casting heartland — cheaper, larger, but often with older process discipline.
• Guangdong (Foshan, Dongguan): high-pressure aluminum die casting, mostly for structural and electronics-heavy parts.

If a foundry claims deep automotive casting export experience and isn't in one of those clusters, ask why. It's not impossible, but it's rare.

Size matters more than you'd think

• Under 50 people: great for prototypes, rarely PPAP-ready. The quality manager is usually the boss's brother, and the CMM is a shared idea.
• 50 to 300 people: the sweet spot for finding a capable partner — but only if they've already been burned by a European customer once. That experience sharpens their documentation discipline.
• 300+ people: likely has a real quality department with a documented system. The trade-off is flexibility on smaller volumes. They may not care about your 2,000-piece pilot order.

So when I hear "I found a casting factory in China that can do PPAP Level 3," my first question is always: what size, where, and how many Level 3 submissions have they actually completed?

The 7 Fatal Mistakes That Kill PPAP Submissions from Chinese Foundries

Here's what I see in almost every failed submission. If you're pre-auditing suppliers, put these on your checklist:

1. Datum mismatch in the dimensional report

The drawing calls out a casting datum based on two locating pads on the pattern side. The foundry's CMM fixture uses different locating logic, and the report doesn't mention it. Every dimension says "pass" — but they were measured against the wrong reference. Ask for the CMM setup sheet and the datum verification summary. If the supplier can't explain how they established the datum, you don't have a valid dimensional report.

2. Documents that don't talk to each other

Control plan says 80 parts per hour. PFMEA says 120. Actual production log says 40. Or the control plan lists "hardness check every 10 pieces" but the MSA study was done on a CMM measuring dimensions, not hardness. This reads exactly as what it is: nobody reviewed the file before sending it. OEMs reject on this instantly. Cross-check PFMEA, control plan, and actual run logs against each other.

3. Material certificates with no heat number traceability

The lab report says tensile 450 MPa. Which heat? Which melt log number? Which day? In many Chinese foundries, the chemistry certificate is generated by the alloy or ingot supplier — not by the foundry itself. The foundry's own melt log is a handwritten notebook in a drawer somewhere. The traceable line must go from the test bar to the heat number to the production batch. If it doesn't, the material test is worthless.

4. Cpk studies done in one afternoon

A valid initial process study needs at least 100 consecutive parts from a production-intent run at the planned production rate — some OEMs demand 300 if you want strong confidence in the process capability. I've seen Chinese foundries run 30 parts over two hours on a cold line and proudly submit Cpk values above 2.0. Nobody believes that. Ask for the production run time log and the subgroup structure.

5. Sub-tier suppliers are invisible

The foundry casts the part. But the core sand comes from a local supplier that's never been qualified. The alloy comes through a trader. The heat treatment is outsourced to a shop with zero PPAP awareness. Under PPAP 4th Edition thinking, all of this belongs in the file: sub-tier quality, incoming material certificates, special process approvals. If the foundry's package is silent on sub-tiers, the package is incomplete.

6. Unqualified or out-of-scope labs

A top Chinese foundry has an ISO/IEC 17025 accredited lab for tensile, hardness, and chemistry. A typical mid-size foundry has a cramped lab room with a tired tensile tester and no accreditation. The report says "our lab," but under audit they can't produce calibration records. Send a verification sample to a third-party lab and watch the results drift. If a supplier says "we'll handle all testing in-house," ask for the accreditation scope and calibration certificates before you believe it.

7. Boilerplate heat treatment and NDT records

High-strength castings need heat treatment — solution treatment, aging, sometimes annealing. The furnace chart must show actual temperature curves for the batch in question. NDT (when customer-specified) needs operator certifications and sensitivity records. Once I received a T6 report where all five batches on the furnace chart had identical temperature curves. Identical. That file went straight back with a rejection note. Verify a few batch records against the actual heat number.

The One-Day PPAP Readiness Audit I Use with Any China Foundry

Before you write a purchase order, spend one day on site. The sequence below has never failed me. It's cheap, it's fast, and it exposes document discipline better than any questionnaire.

Morning: files first, before you touch a part

Ask for a sanitized PPAP package from a previous customer, under NDA. Open the file in front of them. Run a 15-minute consistency check: does the PFMEA match the control plan frequencies? Does the control plan match the actual forms in the back of the binder? Do the dates on the PSW align with the sample production dates? In 15 minutes you'll know if they review files before shipping them — or if the binder is just a collection of unrelated PDFs.

Mid-morning: walk the line, unannounced route

Don't follow the planned tour route. Ask to see the melting area, the sand mixing area, the cleaning room, and the scrap station. Look at the reject pile: how many castings are stacked with internal defect marks? A foundry with a healthy process has a tidy scrap area and knows its top three defect modes. A foundry with no idea will show you a pile and say "these are from a customer project."

Lunchtime: find where the quality manager actually sits

Look around. Is the quality manager's desk in the corner of the shop floor, or crammed into an office with sales? Does their computer have real engineering tools — CATIA, UG/NX, a licensed CMM software package? Or is everything in Excel? The quality manager's toolkit tells you exactly how sophisticated their PPAP work can be.

Afternoon: run a document pop quiz

Point at a batch of castings currently in production. Say: "Show me the heat number. Show me the melt log. Show me where it's recorded in the control plan." That's a 30-minute test. If they can't produce all three within 30 minutes, the documentation system is decorative, not functional.

Closing: ask about their last rejection

Chinese foundries love talking about success. They hate talking about failures. So ask directly: "What did your last export customer reject you for, and what did you change?" A disciplined foundry answers immediately with specifics. A weak one gets vague: "small problem, customer too strict." That vagueness is your answer.

One cultural note: when a Chinese supplier says "no problem," don't treat it as a commitment. "No problem" is polite closure — it sometimes means "I heard you," and occasionally means "please stop pushing." A foundry that asks zero clarifying questions about your drawing, your volumes, and your special characteristics is a red flag. The ones who actually know PPAP Level 3 ask questions. Questions are a good sign. Silence is not.

Casting PPAP Level 3: The 18 Elements, and the 8 That Fail Most

The complete package has 18 elements. In my experience, these eight are where Chinese foundry submissions break. If these are solid, the rest usually follows.

For the full 18-element breakdown, including how the other ten elements apply to foundries specifically, see our casting supplier audit checklist. It's the exact spreadsheet I use when reviewing PPAP packages from Chinese foundries.

FAQ: PPAP Level 3 and China Foundries

Can Chinese foundries really deliver PPAP Level 3?

Yes. But the group that can do it cleanly is small — a few hundred, concentrated in Zhejiang, Jiangsu, Shandong, and Guangdong. Most Chinese foundries can produce the parts. What separates the capable few is document discipline: traceable material certs, consistent control plans, valid dimensional reports. Screen for that, not just casting capability.

How long does a PPAP Level 3 submission from a Chinese foundry take?

First submission typically runs 6 to 10 weeks after sample production. That includes lab testing, CMM programming, and file assembly. With a pre-submission audit, plan for 8 to 12 weeks. Correction loops add another 2 to 4 weeks per round. To cut loops, ask for raw CMM data and the melt log on the very first submission.

Do I need CQI-27 for a casting supplier?

If your customer's special requirements reference AIAG CQI-27 (Casting System Assessment), then yes. Many European OEMs now require CQI-27 self-assessment or customer assessment for safety-critical castings alongside the PPAP file. It's not one of the 18 PPAP elements, but a foundry that has been through CQI-27 is usually far more disciplined about process documentation.

What if I suspect a Chinese foundry is faking PPAP data?

I've seen it happen. The classic fake is a dimensional report with copied dates, or a material certificate that doesn't match the heat log. Ask for three things: the CMM raw program file, the furnace melt log for that heat, and a third-party lab check on your own sample. If the price is impossibly low and the PPAP file is impossibly clean on the first attempt, be more suspicious, not less.

PPAP Level 3 vs Level 2 — which do I need for castings?

For safety-critical castings — steering knuckles, brake components, engine structural parts — always Level 3 unless your customer explicitly writes otherwise. Level 3 is the full 18-element package, which is the practical default in automotive. If a foundry pushes you toward Level 2 to reduce their workload, treat it as a red flag about their actual process control.

Bottom Line: What to Do Next

If you're leaning toward a Chinese foundry for a casting that requires PPAP Level 3, the smart money is on screening for documentation discipline before you negotiate price. Three actions to take this week:

1. Get a sanitized PPAP file from every finalist. Run the consistency check: PFMEA vs control plan vs PSW vs actual dates. Fifteen minutes per file.
2. Do a one-day audit. Use the sequence above — files in the morning, line walk mid-day, document pop quiz in the afternoon. It costs you one flight and saves you six months.
3. Plan a pilot run with a full document gate. Sample approval is not enough. The full PPAP package must be reviewed, corrected, and approved before you scale production.

For more context on choosing a qualified partner, read our guide to China casting regions and where the capable foundries actually sit. And if you're mid-process with a Chinese foundry and the file package is already a mess, send us your drawing and the current PSW — we'll run a 48-hour PPAP readiness gap check and hand you a list of what's missing, what's inconsistent, and what your OEM will reject. No sales pitch. Just the gap list.

The casting you need is already being made in China. The question is whether the paper around it looks like it.