Model Update Under Uncertainty and Error Estimation in Shock Applications

Numerical models are used for computing the shock response in many areas of engineering applications. Current analysis methods do not account for uncertainties in the model parameters. In addition, when numerical models are calibrated based on test data neither the uncertainty which is present in the test data nor the uncertainty in the model are taken into account. In this paper an approach for model update under uncertainty and error estimation for shock applications is presented. Fast running models are developed for the model update based on principal component analysis and surrogate models. Once the numerical model has been updated the fast running models are employed for performing probabilistic analyses and estimate the error in the numerical solution. The new developments are applied for computing the shock response of large scale structures, updating the numerical model based on test data, and estimating the error in the predictions. The initial application of this new capability is in the area of ship response to an underwater explosion. The same methodology can be applied any time that an impact load is applied on a vehicle structure such as vehicle crash analysis, response of a vehicle to a load from a blast, or in modal testing, and when numerical models are used to simulate the vehicle's response.