Uncertainty Quantification in Calibration and Simulation of Thermo-Chemical Curing of Epoxy Resins

TL;DR

This paper investigates how uncertainties in material parameters affect the modeling and simulation of epoxy resin curing, using the first-order second-moment method and comparing it to Monte Carlo simulations.

Contribution

It introduces an uncertainty quantification approach for the calibration and simulation of thermo-chemical curing processes in epoxy resins.

Findings

First-order second-moment method provides efficient uncertainty estimates.

The method offers reasonable approximations despite nonlinearity.

Comparison with Monte Carlo validates the approach.

Abstract

Curing of epoxy resins poses a particular challenge in terms of modeling, experimental investigation, and numerical implementation, as it is a thermo-chemo-mechanical process. Several constitutive relations are required to model these processes, yielding numerous material parameters. The calibration of the constitutive relations must be performed using multiple steps, wherein uncertainties unavoidably propagate. In this study, we investigate the propagation of uncertainties during both the multi-step calibration procedure and the numerical simulation of curing processes with the identified parameters. For both, we employ the first-order second-moment method, which is carefully evaluated through coverage tests and by comparing it to the Monte Carlo method as a reference. It is demonstrated that the first-order second-moment method efficiently yields reasonable results, although providing only a first-order approximation of the highly nonlinear stochastic model response.