Abstract: A model validation approach used to less experimental data is presented. The application process and function of validation experiment are introduced. Model uncertainty is quantified by interval variables and interval arithmetic is used to solve the computational model equation. Based on the area between the measured and calculated system response quantities’ cumulative distribution functions, a validation metric is provided and validation result will be assessed. When the validation requirement is not met, an optimization-based method is proposed to calibrate the model parameter. In this optimization model, the validation requirement is taken as the optimization object and the experimental results are used as the constraint condition. The Sandia static frame problem and Sandia thermal challenge problem are presented to demonstrate the effectiveness of the present method.

Key words: model validation;interval arithmetic;model calibration;Sandia static frame challenge problem;Sandia thermal challenge problem