Data-driven in computational plasticity

Ibáñez, R. ; Abisset-Chavanne, E. ; Cueto, E. (Universidad de Zaragoza) ; Chinesta, F.

Resumen: Computational mechanics is taking an enormous importance in industry nowadays. On one hand, numerical simulations can be seen as a tool that allows the industry to perform fewer experiments, reducing costs. On the other hand, the physical processes that are intended to be simulated are becoming more complex, requiring new constitutive relationships to capture such behaviors. Therefore, when a new material is intended to be classified, an open question still remains: which constitutive equation should be calibrated. In the present work, the use of model order reduction techniques are exploited to identify the plastic behavior of a material, opening an alternative route with respect to traditional calibration methods. Indeed, the main objective is to provide a plastic yield function such that the mismatch between experiments and simulations is minimized. Therefore, once the experimental results just like the parameterization of the plastic yield function are provided, finding the optimal plastic yield function can be seen either as a traditional optimization or interpolation problem. It is important to highlight that the dimensionality of the problem is equal to the number of dimensions related to the parameterization of the yield function. Thus, the use of sparse interpolation techniques seems almost compulsory.

Idioma: Inglés

DOI: 10.1063/1.5034932

Año: 2018

Publicado en: AIP Conference Proceedings 1960 (2018), 090006 [6 pp.]
ISSN: 0094-243X

Factor impacto SCIMAGO: 0.182 - Physics and Astronomy (miscellaneous)

Tipo y forma: Artículo (Versión definitiva)

Área (Departamento): Área Mec.Med.Cont. y Teor.Est. (Dpto. Ingeniería Mecánica)