Verification and validation

Verification and validation (V&V) are two complementary steps for checking a design in product development. Verification is confirmation, through objective evidence, that specified requirements have been fulfilled: does the design output match the input? Validation is confirmation that the requirements for a specific intended use have been fulfilled: does the product work the way the user actually needs? Put simply, verification asks "are we building the product right?" and validation asks "are we building the right product?"

Verification is based on the product requirements specification and includes, for example, drawing checks, calculations, simulations, dimensional measurements and lab tests against specified values. Validation takes place under real or realistically simulated conditions, often with users, and on samples made with the production process or one close to it. A product can pass verification flawlessly and still fail validation if the specification itself was wrong.

In the V-model of development, each level of requirements (the whole product, subassemblies, parts) has a matching level of verification. ISO 9001 also requires design verification and validation, and fields such as medical devices (ISO 13485 and, in the US, FDA design controls), aerospace and automotive demand them with particular rigor.

For simulations, V&V has a specific meaning: verification of the model checks that the problem has been solved correctly in numerical terms (mesh convergence, force equilibrium, comparison with an analytical solution), while validation compares the simulation results with measurements on the real part.

Build the verification and validation plan as soon as the product requirements specification is approved. Assign each requirement a verification method (calculation, simulation, measurement, test, documentation review), an acceptance criterion, the number of samples and the project phase in which it will be verified. Such a traceability matrix shows whether anything is missing and serves as evidence for the customer, an auditor or certification.

Validation is planned on functional and pre-production samples, usually before release to series production. For simulations that important decisions rely on, validate the calculation method against measurements at least once, ideally on a similar part.

The most common mistake is substituting verification for validation: all dimensions are within tolerance and the tests to the specification have passed, but in service the product is awkward to operate, does not fit the installation space or cannot withstand a use nobody anticipated. The second mistake is unverifiable requirements ("durable", "lightweight", "easy to use"). Word every requirement in measurable terms, with a value and test conditions.

Verify on samples that represent the final product. Validation results from a 3D-printed prototype may not hold for the injection-molded production part. For every design change made after verification, assess whether part of the testing has to be repeated, and archive the results in test reports clearly linked to the product revision.