2.3. Model Status

As previously discussed, LAMÉ is structured to facilitate support of model and application developers as well as analysts. This means that models in various stages of development from initial research to robust analysis tools may be found in LAMÉ. For analysts and end users, this diverse model base could pose a challenge during model selection. To try and aid in this selection, the LAMÉ library has been split into two branches – development and production.

The development branch – as its name implies – is intended for any model that is currently being implemented, improved, or otherwise studied. New models not yet subject to full verification rigors are placed here. As such, the usage of development models in simulations supporting engineering decisions should be carefully considered and vetted. Production models have been subject to strong verification activities and meet further standards to help give credibility and credence to any simulation results.

To establish that a model is at a production state of readiness, the model must meet three criteria. The first is that it must have a minimum of one regression test but more tests are recommended. Given the diversity of model complexities and fidelities arriving at an alternative common baseline requirement is challenging. Second, the model must be sufficiently documented including underlying theory, discussion of numerical implementation, usage guides, and documentation of verification tests. Third, the verification tests and ensuing documentation must be peer-reviewed. This review should consider the appropriateness of the documentation and the verification tests. Importantly, the alignment of verification tests to expected usage and model phenomenology should be assessed to ensure that the test basis is sufficient for model usage. This peer-review should be documented and retained.

Whereas the production status of a model is meant to establish a credibility basis with respect to verification, validation of a model requires consideration of the actual material of interest, the intended usage, and availability of experimental data. Thus, validation must be considered on a case-by-case basis and determinations in that context are out of scope of this manual. Importantly, a model being production status does not make any statement towards validation.