Incorporating Input Data Uncertainties in Computer Models of Vehicle Systems using the Polynomial Chaos Quadrature Method

This paper presents a simple method of accounting for input data uncertainties in computer models by propagating these uncertainties to output quantities of interest. Traditional Monte-Carlo methods are too expensive to apply to complex models of vehicle systems since each sample requires significant effort. The proposed method based on the theory of spectral expansions of the random variables requires an order of magnitude less effort. The methodology is applied to simulations of Child Restraint Systems (CRS) where statistics on the output quantities of Head Injury Criteria and strain at selected points in the CRS shell are evaluated under the assumption of uncertain input elastic modulus and friction parameters.