SYSE.19:4.1.4 - Construct the missing revalidation from its required result
Start with the conclusion now needed and work backward to the missing support. Then follow a proposed way forward: what input the engineer obtains, what operation transforms it, which result comes back, and why that result would repair the specific inference. Keep a qualified available operation intact. Construct or obtain only the operation or connection that is missing.
| What prevents the receiving conclusion | Candidate way to obtain the needed contribution |
|---|---|
| A parameter changed but the model and its use remain applicable | Recompute the affected output or sensitivity bound with the new parameter; compare it with the same required result. |
| Existing observations are valid but the old interpretation changed | Reanalyse the original observations under the current interpretation, if they contain the required quantities and conditions. A missing measurement cannot be supplied by reinterpretation. |
| Support is absent for a particular unit or operating range | Obtain a suitable calibration, experiment or trial on those units and conditions; establish how its observations support the intended range and duration. |
| The relied-on Method or model is unsuitable | Use SYSE.15 for a changed Method choice or SYSE.7 for a changed decision-usable description. Obtain any missing professional Method-development result before relying on it, then qualify its evidence through SYSE.10. Repeating the old operation does not fix its failed premise. |
| Applicability, obligation or permission is unsettled | Obtain the relevant source or decision from its responsible holder. A successful engineering test supplies neither that interpretation nor that authority. |
For a test proposal, derive the discriminating observation from the failed claim. Name the changed condition to exercise, the unchanged controls needed for attribution, and the result that would support, contradict or leave the claim unresolved. Do not select a familiar test merely because it is available. If a bound is claimed over an interval, show why the sampled observations cover that interval; add a justified model or uncertainty bound, change the test, or limit the conclusion to the observed points. Similarly, one calibrated unit does not qualify its neighbours without a justified population inference.
This construction may return several adequate ways: reuse qualified observations, obtain a supplier’s
applicable result, perform a targeted experiment, change the engineering solution, or restrict its use. It
may also expose that the required evidence cannot be obtained. Compare these whole continuations, including
the resulting usefulness, through C.11.DUA:4; use SYSE.6 when they become substantial design alternatives.
There is no fixed requirement to run a model, a calibration and a chamber test on every source change.