A method for epistemic uncertainty quantification and application to uniaxial tension modeling of polymers

Epistemic uncertainty, whether due to lack of knowledge or data, is often prevalent in practical engineering structures. We present an effective uncertainty quantification method based on orthogonal experimental design, range analysis and interval analysis for analyzing and propagating this type of uncertainty. Orthogonal experimental design and range analysis are first used to investigate the effect of epistemic uncertain factors on different metrics (such as minimum computational cost, maximum computational accuracy and a mixture of two). Interval analysis is then employed to propagate the dominant epistemic uncertainty from input to output through a model. The present method is applied to construct a uniaxial tension model of polymer Nafion membrane with four selected epistemic uncertain factors (mesh method, element type, boundary condition and constitutive model). A good understanding of the influence of these factors on the outputs of the constructed model demonstrates the effectiveness of the present method.