Illustration: batches A and B linked to individual sample records. Concept only, not customer data or an equipment photograph.
Record provenance, not just a good weld
Use a separate record for each sample. Our suggested worksheet supports purchasing discussions; it is not a statistical sampling plan, a qualification procedure or a required sample count.
Sample record
- Sample ID; batch/date; drawing revision; observed difference; agreed check; result reference; follow-up owner.
Select samples with the trial provider
Discuss differences in preparation, fit-up and loading that are relevant to the intended evaluation. Distinguish observations from measured values. Unknown variation should remain unknown, not become an invented tolerance. Agree selection and acceptance criteria before drawing conclusions.
Keep conclusions within the tested scope
Record the setup and any manual interventions alongside the sample result. If the intended production conditions differ, discuss further evaluation rather than assuming that a successful demonstration covers all batches. One record does not prove repeatability or factory-wide output.
Does this replace procedure documentation?
No. TWI explains that a welding procedure specification records welding variables and is supported by separate records. This sample worksheet does not replace those documents or applicable validation.
Source: TWI, Welding procedure — Job Knowledge 39, accessed 14 September 2026. Supports the documentation distinction only; the worksheet is CBOXTEC purchasing guidance. TWI.