Author:

Abstract:

I. INTRODUCTION Surface or subsurface explosions cause soil vibrations that can have detrimental effects on structures at large distances. Controlling these effects necessitates an accurate prediction of the soil surface vibrations. The shock non-linearity, high velocities and high pressure levels associated with blast loading considerably complicate all possible prediction strategies. Confronted with the limitations of empirical and analytical modeling, a numerical hydrocode simulation, using explicit time integration, seems the only option to combine a correct approach to the highly non-linear events in the immediate surroundings of an explosion with the site-specific geometry elements. This paper deals with the optimal configuration of an existing hydrocode tool, for blast-induced soil wave propagation simulations. II. NUMERICAL PREDICTION ANSYS-Autodyn offers a wide array of solver techniques, all using explicit time integration, and combines this with a number of tools and adapted material models, destined to enable a correct approach of the highly non-linear propagation of a shockwave and the very large deformations associated with the energies freed by an explosion. A well-considered combination of different solvers can lead to a robust and economic model of the entire non-linear calculation domain, allowing correct material and surface behavior and larger deformations. Up till now, soil blast simulations using this type of tools are limited to small scale models. An intuitive 2D axial-symmetric ANSYS-Autodyn model of a reference surface explosion experiment at the Poelkapelle EOD site was built, using a combination of Euler and Lagrange-type solver techniques. This model delivered promising results but also demonstrated a number of shortcomings, related to solver and software limitations and model configuration restrictions. A number of solutions are proposed to improve the performance, each having their own advantages and disadvantages. Considerable errors are caused by the treatment of model boundaries and solver transitions. These can be minimized by a well thought-out geometry and meshing strategy, which calls for an adaptation of currently available ANSYS-Autodyn model configuration tools. An extended use of the Arbitrary Lagrangian Eulerian solver scheme (ALE) allows combining the advantages of Lagrangian and Eulerian codes. This is obtained by adding an automated mesh rezoning step to a traditional Lagrange scheme, improving material boundary and surface treatment and allowing larger soil deformations. The increase in calculation costs, due to the use of time-consuming explicit hydrocode solvers, can be reduced by limiting their use to the shock wave domain and the hydrodynamic or elasto-plastic soil deformation zone. The vibrations at the borders of this reduced hydrocode model can be used as the input at the boundaries of a model of the linear elastic wave propagation zone, thus replacing the non-linear part of the model by an equivalent elastic source. The linear zone itself can be simulated using Boundary Elements. III. CONCLUSIONS AND FUTURE WORK The prediction of blast-induced soil surface vibrations needs the use of explicit hydrocode models. The different solvers and tools provided by ANSYS-Autodyn can be useful in the prediction process. However, several limitations call for adaptations and extension of these tools, in order to allow a correct approach of the non-linear events near the explosion. Future work includes the development of an adapted linear elastic Boundary Element Model, the determination of the appropriate zoning of both non-linear and linear elastic models and the development of a fitting soil model. Furthermore, a supplementary series of underground blast tests, including short range surface vibration measurements, are currently being planned to be able to validate the hydrocode model and value all abovementioned model optimizations.