Towards reliable simulations of ballistic impact on concrete structures

Protection against weapon effects like ballistic impacts, fragmenting shells and explosions is the core business of the Explosions, Ballistics and Protection department of TNO (The Netherlands). Experimental and numerical research is performed to gain and maintain the knowledge to support the Dutch MoD and the industry to evaluate and develop adequate protection. The TNO high performance cluster (HPC) combined with the Laboratory for Ballistic Research (LBR) located in Ypenburg (NL) provide the essential infrastructure. Hitherto the numerical work was focussed on the response of steel -and RC-structures under blast loading and on ballistics for steel armour. Ballistic protection of ceramic armour concepts and RC panels was hardly studied numerically, while the dynamic response of concrete itself is a major research topic in the collaboration of TNO with the Delft University. It was decided to combine the gained knowledge and extend the numerical research to ballistic impact on “brittle targets”.This paper reports on the first steps of this research. It summarizes some of the difficulties and unknowns that were encountered and two aspects are highlighted.The commercial code (LS-DYNA) is used to simulate a ballistic impact and it is shown that the global variables of the experiments (penetration depth or residual velocity) can be matched quite well. However the damage predictions are questionable and highly dependent on the applied material model, and the available data to calibrate these models. This is illustrated with an example by varying the EOS, which is only one aspect of the material model.The other aspect that the paper addresses in more detail is the localisation problem and the mesh dependency for softening materials like concrete. To represent the dynamic loading conditions for the material, small elements are required to capture the stress gradients. The required element size is far smaller than the width of the fracture process zone. . In combination with the localization that occurs due to the description of concrete as a softening material, the traditional regularization techniques leads to incorrect descriptions of damage during a ballistic impact.A nonlocal method might overcome this limitation in order to attain more accurate damage predictions. A stress based non local approach is proposed and tested using benchmarks to show desired behaviour.