Material review: deciding what happens to parts that did not meet the drawing

The outcomes are few and they are not equivalent

Use as it stands accepts the deviation and keeps the value. Rework brings the part back to the drawing and consumes hours. Repair returns it to a usable but different condition, which is a permanent change requiring agreement. Downgrade sends it to a less demanding application. Return recovers cost from the supplier. Scrap ends the argument at maximum expense. Each carries a different requirement for evidence and authority, and confusing rework with repair is the mistake with the longest consequences, because a repaired part remains different from the drawing forever.

Who has to be present for the decision to mean anything

At minimum, somebody who understands the function of the affected feature, somebody who knows what the process can and cannot recover, and somebody holding quality authority who is able to refuse. For customer-owned designs the customer's engineer is the only party who can properly judge fitness for purpose, which places a boundary on what an internal group may decide alone. Where the board is a single quality engineer signing forms, the mechanism has the appearance of control without the substance, and its decisions will not survive scrutiny after a field failure.

Concessions belong to the customer, not to your engineers

Shipping something that departs from the agreed specification is a request, not a decision. It needs to be asked in writing, for a stated quantity, for a stated deviation, with the reason and the affected batch identified, and it should carry an expiry so a temporary allowance does not become a permanent understanding. Concessions granted verbally have a habit of being remembered differently when a problem appears, and open-ended concessions accumulate until nobody can say what the current specification actually is. Marking the affected consignments is part of the request, not an optional courtesy.

Rework has to be engineered, not improvised

A rework decision is really an instruction to perform an unplanned operation, and it deserves the same treatment as a planned one: a written method, defined equipment, someone trained to do it, and a check afterwards proving the result. Improvised recovery is where secondary damage is introduced, heat is applied where it should not be, coatings are compromised, and dimensions previously conforming are pushed out. The other trap is cost. Recovery labour is rarely captured against the original problem, so rework looks free while consuming capacity the plant was short of anyway.

The drift towards accepting everything

Use as it stands costs least on the day and the most damaging as a habit. Each individual decision looks defensible; the pattern is that a tolerance has effectively been widened without anybody agreeing to it, and eventually a part is accepted that genuinely matters. Two controls slow the drift. Count the dispositions by outcome and review the trend rather than the individual cases. And where the same deviation is accepted repeatedly, force the question upward: either the drawing is wrong and should be changed, or the process is wrong and should be fixed.

Frequently asked questions

What is the difference between rework and repair?

Rework restores the part to full conformity with the drawing, so afterwards there is nothing to declare. Repair produces something acceptable for use but permanently different from the specification, which is why it needs the design authority's agreement and a record that travels with the part. Treating a repair as rework hides a permanent deviation in the plant's own records, and it is the version of this mistake that causes trouble years later during a failure investigation.

Can we accept nonconforming material without telling the customer?

Only where the requirement missed is genuinely internal to your process and the product still meets everything the customer specified. If the deviation touches anything they defined, the decision is theirs and the request should be in writing with the affected quantity identified. Silent acceptance is a poor bet: it is usually discovered eventually, and the discovery converts a routine technical conversation into a question about whether your records can be trusted at all.

How do we stop use-as-is becoming the standard answer?

Measure it. Report dispositions by outcome each month and watch the proportion accepted without correction, by part and by defect type. Set a rule that the same deviation accepted repeatedly triggers either a drawing change request or a process investigation, so the third occurrence forces a decision that the first two avoided. Also make sure the board includes somebody with no delivery responsibility, since the pressure to accept is strongest exactly where the schedule is owned.