Common build failures: what actually goes wrong on the shop floor

Two related failures share a root cause in the heat-treat cycle. Both pass the hardness certificate. Both fail in service.

Under-tempered die that cracks on hit #50. A fresh H13 die cracks in the first few hundred cycles, typically through a parting line, a corner, or a feature that sees impact loading. The hardness certificate reads in spec. The fracture surface shows brittle intergranular failure, not fatigue striations. The root cause is a single temper on a 5 percent Cr steel. As-quenched H13 holds 15-25 percent retained austenite. One temper at the secondary-hardening peak (540-560°C) transforms some austenite to fresh martensite, but the fresh martensite is itself untempered and brittle. The second temper transforms and tempers it. Skipping the second temper leaves a structure that holds the hardness target on a Rockwell tester but cracks under the first real load.

Retained austenite that grows the die in service. A die that was on dimension at ship date measures out of tolerance after 5,000 hits. Critical cavity dimensions have grown 0.001-0.003 inch. Hardness has dropped slightly. Surface treatments look intact. The die is dimensionally scrap. The root cause is retained austenite in the bulk, slowly transforming to martensite under service thermal cycling. Austenite is denser than martensite. Every percent of austenite that transforms in service adds about 0.0001 inch per inch to the part. A die with 12 percent retained austenite at ship date can grow a measurable amount as the structure stabilizes over the first few thousand cycles.

The catching inspection is the same in both cases. Retained austenite measurement on the witness coupon post-HT. X-ray diffraction is the reference method. Magnetic saturation is acceptable and faster. Target below 5 percent for forge service, below 3 percent for precision work. NADCA #207 calls for double temper minimum, triple temper for critical work. Verify the temper count on the heat-treat traveler before signing off and require the cycle trace. The same inspection catches both screwups, which is why adding it to the standard HT acceptance protocol is the single highest-value step in incoming inspection on a forge die.