Discussion of model calibration and validation for transient dynamics simulation.
- François M.
- Scott W.
- Amanda C.
Model calibration refers to a family of inverse problem-solving numerical techniques used to infer the value of parameters from test data sets. The purpose of model calibration is to optimize parametric or non-parametric models in such a way that their predictions match reality. In structural dynamics an example of calibration is the finite element model updating technology. Our purpose is essentially to discuss calibration in the broader context of model validation. Formal definitions are proposed and the notions of calibration and validation are illustrated using an example of transient structural dynamics that deals with the propagation of a shock wave through a hyper-foam pad. An important distinction that has not been made in finite element model updating and that is introduced here is that parameters of the numerical models or physical tests are categorized into input parameters, calibration variables, controllable and uncontrollable variables. Such classification helps to define model validation goals. Finally a path forward for validating numerical model is discussed and the relationship with uncertainty assessment is stressed.
- Research Organization:
- Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
- Sponsoring Organization:
- USDOE
- OSTI ID:
- 975888
- Report Number(s):
- LA-UR-01-6574; TRN: US201018%%974
- Resource Relation:
- Conference: Submitted to: Proceedings of IMAC-XX, 20th International Modal Analysis Conference, February 4-7, 2002, Los Angeles, CA
- Country of Publication:
- United States
- Language:
- English
Similar Records
Validation of the Community Land Model Version 5 over the Contiguous United States (CONUS) using in-situ and remote sensing datasets
Validation of the Community Land Model Version 5 over the Contiguous United States (CONUS) using in situ and remote sensing data sets