The "Speaking of Simulation" series explores how leading organisations in the NAFEMS community use simulation to create engineering evidence. In this installment, the spotlight falls on Certification by Simulation, or Certification by Analysis (CbA): the point at which a model is no longer only a design aid, but part of the case presented to an independent authority.

That shift changes the question. It is not enough to ask whether a simulation looks plausible or whether it helped an engineer make a decision. The harder question is whether the simulation evidence is credible for a defined context of use, traceable to requirements, supported by validation data, and governed well enough that a regulator, assessor or notified body can trust the route taken to the answer.

This first article looks at where certification by simulation is already working, and why wider acceptance remains difficult. The recently published NAFEMS report [1] shows that the mature examples are rarely pure replacements for testing; they are usually hybrid arguments in which analysis, standards and targeted physical evidence support one another. FDA medical-device submissions, rail vehicle validation, aircraft seat certification and nuclear safety analysis illustrate how simulation can become credible certification evidence when the route to acceptance is well defined [2-5].

The obstacles are just as instructive. Interviewees repeatedly pointed to questions of trust rather than technology alone: inconsistent VVUQ practice, uneven governance, cultural preference for physical tests, differing regulatory regimes, and the absence of agreed metrics for the quality of a certification process. The barriers explain why successful cases often emerge first where testing is impractical, too slow, or unable to cover the full operating envelope.

Testing therefore does not disappear. It becomes more deliberate and coordinated with VVUQ. Physical tests calibrate, challenge and validate simulations, while simulations help choose critical cases, explore envelopes that would be impractical to test exhaustively, and connect sparse evidence into a coherent safety or compliance case.

This article focuses on the first two questions: successful implementations and the obstacles to broader acceptance. The follow-on article, scheduled for the January 2027 issue of Benchmark Magazine, turns to the practical implementation questions: how regulators and industry collaborate, how Verification, Validation and Uncertainty Quantification is built in early, how data is managed, and how AI, competence and credibility frameworks fit into the certification process.

This article appeared in the October 2026 Issue of BENCHMARK.