APPLYING OPTIMIZATION METHODS AND STOCHASTIC ANALYSIS IN EVALUATING A CASK ACCIDENT

In the following a hypothetical accident scenario is investigated. In this scenario a cask is vertically driven above a structure, elevated above a concrete ground. While driven the cask drops and falls on to the deformable edge of the structure and subsequently into a damped area of the ground. The impact onto the ground will be cushioned by a cavity filled with porous concrete. Should the cask tilt, however, it cannot be excluded that it hits the ground with its top-end outside the damped area made from porous concrete. For this event sequence, the load on the cask needs to be evaluated. To evaluate realistic impact scenarios, the entire complex drop sequence needs to be simulated. Furthermore, the drop sequence is influenced by numerous parameters, so that the worst scenario in terms of loads cannot be determined by theoretical consideration. For this matter, the drop sequence was simulated on the one hand as a simplified and physically feasible FE-model which represented the cask as a rigid body. On the other hand, a sensitivity analysis with the program optiSLang identified those parameters which contribute significantly to a high rigid-body deceleration and therefore cause adverse load situations. OptiSLang offers efficient methods for sampling and stochastic analysis. It eases through the high given number of parameters with the lowest possible number of computation runs to arrive at a resilient statistical evidence and further contributes to the automatization of the computation process. A stress analysis for some selected impact scenarios can be conducted by means of an adequately detailed and sufficiently discretised FE-model. Here, it is possible to position the cask immediately prior to the impact and to initialize the kinematics, which have been determined on the basis of the simplified model, at this point of time. The complete FE-model is composed of individual partial models which need to be adequately realistic. This is especially important for the partial model of the porous concrete. In order to simulate the energy dissipation of this compressible material, a complex material model had to be used. The material characteristics required for this were calculated with optiSLang. This calculation here means an optimization task in order to minimize the deviations of a test and the test recalculation.