Common Mistakes

Costly Mistakes in Nuclear and Critical Infrastructure Manufacturing

The recurring, expensive mistakes in nuclear and critical infrastructure manufacturing, each with its symptom, root cause, and a numeric fix.

The first mistake is underscoping documentation early in the quote. Symptom: a job budgeted for 40 hours of paperwork balloons to 180 by shipment. Root cause is treating a safety-class (Class 1 or 1E) part like a commercial one, when 10 CFR 50 Appendix B and NQA-1 add certified material test reports, weld maps, and a data package for every serial. Fix: run the Nuclear Documentation Burden calculator before pricing. A single ASME Section III pressure part routinely carries 12 to 20 documents per serial versus 2 to 3 commercial, and that 6x factor must sit in the bid, not the variance report.

Missed or resequenced hold points is the second killer. Symptom: an inspector arrives and the weld is already covered by a subsequent pass, forcing rework or a use-as-is dispositioned nonconformance. Root cause is treating witness (W) and hold (H) points as advisory. An ITP hold point legally stops work until the ANI or customer signs. Fix: schedule with the Quality Hold Point Delay tool. Each hold typically adds 1 to 4 shop days of wait for a traveling inspector, so a router with 8 hold points can hide 15 to 25 idle days that never appear on the machining estimate.

Broken serialized genealogy is the third. Symptom: during a spare order five years later you cannot prove which heat lot fed which serial, and the whole batch becomes suspect. Root cause is capturing traceability at the lot level instead of the individual serial, or hand-keying heat numbers into a spreadsheet where a transposed digit (say 8472 versus 8742) severs the chain. Fix: model the real linkage with the Serialized Component Genealogy Cost tool. Full serial-to-heat-to-CMTR genealogy runs 20 to 45 minutes of record work per component, and skipping it now costs 10x to reconstruct under audit later.

Unit and rev-level errors on inspection scope quietly inflate cost. Symptom: your inspection estimate is 2x the shop average with no obvious reason. Root cause is mixing inspection frequency bases, applying a 100 percent dimensional plan (every serial, every feature) where the code allows AQL sampling, or costing to a superseded drawing revision. Fix: separate safety-class from commercial features and price them apart with the Safety-Class Inspection Cost calculator. Safety-class features may demand 100 percent verification and dual signoff at 3 to 5 minutes each, while commercial features sample at C=0 on a 32-piece lot, an order of magnitude cheaper.

Assuming an existing supplier is still qualified is a silent process failure. Symptom: a receiving inspection rejects material and the paperwork trail reveals the vendor's ASME certificate or NUPIC audit lapsed 8 months ago. Root cause is no tickler on qualification expiry and no triennial reaudit budget. Fix: track the audit cycle with the Qualified Supplier Audit Cost tool. Commercial-grade dedication (CGD) or a full 10 CFR 50 App B supplier audit runs 3 to 6 auditor-days, and NQA-1 requires reaudit on roughly a 3-year cycle, so budget one audit per qualified source every 36 months, not once at onboarding.

Slow nonconformance disposition strangles throughput. Symptom: work-in-process piles up at a quarantine cage and on-time delivery drops even though machines sit idle. Root cause is routing every NCR through a full Material Review Board when many are minor and dispositionable in hours. Fix: triage using the Nonconformance Disposition Time tool. A minor rework NCR should close in 2 to 8 hours, but a repair or use-as-is on a safety-class item needs engineering plus ANI concurrence and legitimately takes 5 to 15 business days. Mixing those two populations in one average hides the real bottleneck.

Ignoring obsolescence on long-lived assets bites decades out. Symptom: a plant needs a replacement I&C card or valve internal and the sole qualified supplier discontinued it 6 years ago, triggering an emergency dedication project. Root cause is no lifecycle scan against 40 to 60 year plant design lives. Fix: score every serialized part with the Spare Part Obsolescence Risk tool. Electronic and instrumentation components carry 5 to 15 year commercial availability against a 40-plus year demand, so any part with a supply-to-demand ratio under about 0.3 needs a last-time-buy or a dedication plan on the books now, not at failure.

Finally, underestimating regulatory package review load derails ship dates. Symptom: the hardware passed final inspection two weeks ago but is still sitting because the data package is in review. Root cause is planning zero float for document review and signature routing. Fix: size it with the Regulatory Package Review Load and Traceability Record Workload tools. A complete safety-class data package for a moderately complex assembly commonly runs 150 to 400 pages and takes a reviewer 8 to 20 hours plus 3 to 5 days of routing, so add 1 to 2 weeks of package float after the last physical operation on every schedule.

Published 2026-07-01.