How do you validate a simplified source model?

• A. Compare against experimental data or more detailed simulations; perform sensitivity checks and ensure conservative results.
• B. Validate only with expert judgment.
• C. Validation is unnecessary for simple models.
• D. Compare against unrelated datasets.

Answer: (A)

Validating a simplified source model means building confidence in its predictions by checking how well it matches reality and more detailed understanding, and by testing how robust it is to uncertainties. The best approach is to benchmark the simplified model against trusted references: either experimental data from real tests or results from higher-fidelity simulations that are known to be accurate for the scenario you care about. This tells you whether the simplified model captures the essential behavior without overstepping its assumptions.

Next, run sensitivity checks by varying key inputs within realistic ranges. This shows how the outputs respond to changes in assumptions and helps identify any non-physical or unstable behavior. It also highlights which inputs drive the results, guiding where to focus refinement or data collection.

Conservatism is another important facet: ensure the model does not inherently understate risk or bias results toward overly optimistic outcomes. If necessary, apply conservative bounds or quantify uncertainty so decisions remain safe even when inputs are uncertain.

Relying only on expert judgment lacks objective, repeatable validation evidence. Validation against unrelated datasets won’t tell you whether the model behaves correctly in the intended domain, so it’s not appropriate for establishing trust in a simplified model.