Sequential calibration of material constitutive model using mixed-effects calibration

Identifying model parameters is nowadays intrinsically linked with quantifying the associated uncertainties. While classical methods allow to handle some types of uncertainties such as experimental noise, they are not designed take into account variability between different test specimens, significant in particular for composites materials. The estimation impact this intrinsic on material properties can be achieved using population approaches where modeled by a probability distribution (e.g., multivariate Gaussian distribution). objective calibrate (or equivalently its parametric Among found mixed-effects models that characterize each replication decomposed averaged behavior (called fixed-effects) and random-effects). Yet, when number or computational time single run simulations increases (for multiaxial instance), simultaneous, global identification all difficult because unknown quantities estimate required evaluations. Furthermore, do have same influence constitutive depending instance nature load tension, compression). method proposed paper enables multiple experiments. It decomposes overall calibration problem sequence calibrations, subproblem allowing joint subset parameters. process eased reduced compared full problem. applied an orthotropic elastic non linear longitudinal behavior, unidirectional composite ply made carbon fibers epoxy resin. ability sequentially investigated. Its capability ensure consistency throughout also discussed. Results show methodology allows complex framework.