Comparison of depth-averaged and 3D models for dense granular flows

Debris flows are one of the major threats to mountain communities. They consist of the downslope flow of fine and coarse material, saturated with water, along channelized paths. Due to their high velocity and unpredictability, the evacuation of hit areas may be difficult to execute. To avoid casualties and economic losses, mitigation structures, like filter barriers, are therefore usually adopted. Their primary task is to reduce the flow energy and to retain larger boulders. However, considerable room to improve the design of these structures still exists. In particular, gaining a better understanding of debris flows dynamics is a necessary step to improve the design of barriers. Numerical modelling can contribute to its understanding, and in an effective simulation of the flowing mass dynamics and impact against mitigation barriers. In this frame, the continuum-based Depth-Averaged Modelling (DAM) has been widely used since the 90s. In spite of the good results of this approach, together with the low computational time, the averaging procedure of velocity and pressure along the flow depth causes the loss of crucial information, which is important for correctly simulating the interaction with mitigation structures. A full 3D modelling can overcome this shortcoming by allowing a more complete flow representation, and a more accurate computation of impact forces. However, since debris flow may run for long distances, 3D models would require a large computational time. In this work we aim to study both the shortcomings and the advantages of the DAMs and 3D models. In particular, The DAM model used is DAN-W, while the 3D model is based on the lattice-Boltzmann method. To compare the results from numerical modelling, we use the experimental work performed by Moriguchi et al. (2009) in which a mass of dry sand flows on a steep chute