How to Audit an Investment Casting Supplier: 12 Checks
Twelve checkpoints a buyer can verify on site in one day, each with the question to ask and the red flag that means the answer is theatre.
Xinyouyuan Engineering Team
10/22/20266 min read
Why certificates do not tell you what you need to know
Most buyer audits of a casting supplier consist of a document review: ISO 9001 certificate, material certificates, a calibration list, a few signed procedures. All of these can be genuine and the parts can still fail. Documents prove that a system exists. They do not prove that the system is doing anything on the floor this afternoon.
Everything below is chosen on one criterion: a buyer can verify it on site in a single day, with no laboratory and no prior notice. Each item is a physical piece of evidence rather than a claim. For each one there is a question to ask and a red flag that means the answer is theatre.
A. Wax, tooling and pattern control
1. Wax tank levels and drain valves
Look at: the settling and dehydration tanks. Are there sight glasses or drain valves, and is there a schedule posted for drawing off water? Open a drain and see what comes out.
Ask: how often is water removed, and who records it?
Red flag: tanks with no drain point, or a drain that produces a slug of milky liquid. Water in reclaimed wax becomes porosity and surface roughness, and it is invisible in the finished part until it is not.
2. Wax temperature measured at the wax
Look at: ask the operator to take a temperature reading with your own thermometer or theirs, in the holding tank and at the nozzle. Compare with the controller display.
Red flag: a discrepancy of more than a few degrees, or an oven gauge that is the only instrument. Wax at 54 °C and wax at 60 °C produce different shrinkage and different surface finish, and the difference shows up as dimensional drift weeks later.
3. Pattern storage and ageing
Look at: where finished patterns sit before assembly. Are they on flat supports, in a temperature-controlled area, and is there a maximum storage age?
Red flag: patterns stacked in trays, or hanging unsupported. Pattern deformation is a dimensional defect that no downstream inspection catches until the casting is machined.
B. Shell building — where most quality is actually decided
4. The wetting test and the cleaning bath
Look at: ask for the cluster cleaning bath and the wetting check. The standard test is to dip a cleaned cluster into silica sol with about 0.5% wetting agent: complete uniform wetting passes, beading fails.
Ask: when was this bath last changed, and what is the acceptance criterion?
Red flag: no wetting test at all, a bath that has clearly not been changed in months, or an answer that describes the test but cannot produce the record. Failed wetting becomes metal penetration, rough surface and shell inclusion — and nobody traces it back here.
5. Slurry viscosity measured with a calibrated cup
Look at: the Zahn cup in use, its calibration sticker, and the last week of readings. Then ask them to measure the face coat in front of you.
Ask: what is the control range, and what happens when it drifts outside?
Red flag: a cup with no calibration record, readings recorded only once a day on a three-shift operation, or a range so wide it cannot be violated. Viscosity that climbs through a shift thickens the coat, and thick coats crack.
6. Drying room instrumentation: temperature, humidity and air velocity
Look at: the drying area. Are there humidity sensors, and is there any measure of air movement — anemometer readings, fan settings, a recorded velocity?
Ask: what are face coat and back-up coat conditions, and why are they different?
Red flag: temperature control only, or one humidity reading for the whole room. Drying rate depends on humidity and air velocity together; a plant controlling only one of them is controlling drying by accident. Also look for face coats being force-dried — fast surface drying skins the coat and traps water underneath.
7. Stucco grade matched to slurry viscosity
Look at: the stucco actually in the rain sander or fluidised bed, and the bag it came from. Ask which mesh is running on which layer.
Red flag: the same stucco on face and back-up coats, or a viscosity that has been changed without changing mesh. These two variables are a matched pair; treating them independently is one of the most common causes of both rough surface and shell cracking.
C. Dewaxing, burnout and pouring
8. The dewax pressure curve, not just the pressure
Look at: the autoclave chart or data log for the last few cycles. You are looking at the slope of the pressure rise, not the peak.
Ask: how long does it take to reach 0.6 MPa?
Red flag: charts showing a slow ramp, or no chart at all. Slow pressurisation lets wax expand while the shell is cold and brittle; the shell cracks and the crack becomes a fin or a raised ridge on the casting. Fast pressurisation — around 2–3 minutes — is the countermeasure, and it is visible on the chart.
9. Burnout judged by shell colour
Look at: shells coming out of the burnout furnace. They should be white or rose-coloured. Look at a batch that has just been pulled.
Ask: how is the atmosphere controlled, and what is the acceptance criterion for a shell that is grey?
Red flag: grey or patchy shells being poured anyway. Residual carbon produces gas porosity, matte spots and inclusions, and a correct thermocouple reading with a starved atmosphere will still leave carbon behind. Also check the interval between furnace exit and pour — it should be seconds, not minutes.
10. How aluminium is added for final deoxidation
Look at: watch a deoxidation. Is the aluminium scattered on the surface, or wound onto a rod and plunged below it?
Ask: what is the target addition, and is it in one step or split between furnace and ladle?
Red flag: aluminium sprinkled on top. Its density is far below steel, so most of it burns off at the surface; the result is inconsistent deoxidation, gas porosity and residual aluminium variation. Also ask whether deoxidation is two-stage — preliminary Mn-Si followed by final aluminium — or a single addition at the end.
D. Verification and traceability
11. Destructive capability, not just NDT
Look at: what the plant can do in-house versus what it subcontracts. Radiography, penetrant and ultrasonic are useful; sectioning, metallography and a tensile test tell you whether the process is actually controlled.
Ask: for the last three months of rejection data by defect mode, not the overall rate.
Red flag: an overall rejection figure with no breakdown. A plant that cannot tell you whether its rejects are gas, shrinkage, shell or dimensional does not have a corrective action loop; it has a scrap bin. The breakdown is also the fastest way for you to see whether the process is stable.
12. Heat and lot traceability that survives a recall
Look at: pick one finished part at random and ask them to trace it back to the heat number, the wax lot, and the shell batch.
Ask: how long does that take, and where is it recorded?
Red flag: more than an hour, paper only, or a trace that stops at "the heat number is on the box". If a customer reports a problem in the field, this is the difference between a contained issue and a recall of everything you shipped that quarter.
How this explains a 40% price difference
Two quotes for the same part, one 40% higher. Buyers usually assume the difference is labour cost or margin. In practice it is usually a difference in the items above, and it shows up in three places:
Yield. A plant pouring trident gating with dual filters can use nearly twice the metal per good casting compared with a basic system. That metal is in the price.
Inspection cost. Full radiography, penetrant on 100% of parts, and dimensional layout on a CMM are real hours. A cheap quote often means sampling, not 100% inspection.
Rework and concession. Welding, impregnation and straightening are legitimate processes, but a quote that assumes a high rework rate is a quote that expects to ship marginal parts.
The useful question is not "why are you more expensive" but "what is in your price that is not in theirs". A supplier who can answer that specifically is telling you they know their own cost structure. One who cannot is guessing.
A sample approval checklist that actually predicts production
First-article approval is usually a dimensional report plus a material certificate. That combination passes parts that will fail in volume. Add these before you release a part to production:
Section the first article at the gate region and at the last-filled region, not only at the defect you were worried about.
Ask for the casting's weight across ten consecutive pieces. Weight spread is a sensitive proxy for shell and gating stability.
Run the customer's functional test — pressure, leak, flow — on at least five parts, not one.
Require the process parameters used for the first article to be recorded and compared against the parameters used in the first production lot.
Agree in writing what happens when a parameter drifts outside its window: who is notified, and whether the lot is held.
What to do with this list
Use it as a one-day agenda. Items 4, 6, 8 and 9 will tell you more about a foundry than a full document review, because each of them shows you a physical process rather than a policy. A plant that can show you a dewax pressure curve, a wetting test result and a shell that comes out of the furnace white is a plant that is running its process rather than reacting to it.
We are happy to be audited on all twelve. If you are currently qualifying suppliers and want a comparison point, send us the drawing and we will return a quote with the process route, the gating approach and the inspection plan stated explicitly, so you can compare it line by line against the others.
Request a quote with process details or write to [email protected]. Our capability data sheet lists the alloys, sizes and tolerances we hold.
Quality
Precision manufacturing for diverse industrial applications.
Services
Products
+8615912702921