Search

The increasing urbanization of mountainous areas increased the risk imposed on residential buildings and infrastructure. In Switzerland, shallow landslides and hillslope debris flows are responsible every year for high infrastructure damage, blocking of important highways, evacuations and deaths. Up till now, the assessment of these processes has been mainly based on the experience of experts, especially in the assessment of their run-out extent and expected damage. In this research we present a new computationally efficient Discrete Element Model (DEM) which has been developed for the aim of simulating the run-out of hillslope debris flows.YADE-DEM open source code has been extended and an elasto-plastic adhesive contact law have been implemented, which partially account for the presence of the fluid composed of water and find material. This is achieved through the adhesive aspect of the contact law, which would indirectly take the presence of such fluid into account, as this fluid would increase the cohesion of the flowing mass. A parametric study has been carried out to define the most sensitive model parameters, which were found to be the microscopic basal friction angle (Φb) and the ratio between stiffness parameters (loading and unloading) of the flowing particles . Data of full-scale experiments of hillslope debris flows were used to compare the flow kinematics with the model’s prediction. A good agreement between the model and experiments was observed concerning the mean front velocity (average margin of error of 8%) and the maximum applied pressure (average margin of error of 5%), with less agreement of the flow height (average margin of error of 13%). Detailed comparisons of pressure evolution between different selected experiments and simulations revealed the model’s capability of reproducing observed pressure curves, especially during the primary loading phase, leading to maximum pressure.In order to test the model’s prediction of run-out distance of hillslope debris flow, hundreds of past hillslope debris flow events in the Swiss Alps were analyzed and 30 cases were selected representing different situations (i.e. different release volumes, slopes and forest cover). Due to the discrete nature of results in YADE, a GIS algorithm was developed in order to create envelopes representing the temporal evolution of the simulated propagating processes, which were compared to the those of the historical events. Results of the comparison revealed that, with the calibration of the two sensitive parameters in YADE, a fair to very good agreement was observed between the envelopes of the model and those of historical events for 87% of the tested cases. Difficulties in reproducing the envelopes of the rest of the cases are linked to the uncertainties in the mapping of the envelopes of past events, the role of the forest which is not taken into account in the model, and the lack of direct representation of fluid in the model.

This paper presents a Discrete Element-based elasto-plastic-adhesive model which is adapted and tested for producing hillslope debris flows. The numerical model produces three phases of particle contacts: elastic, plastic and adhesion. The model capabilities of simulating different types of cohesive granular flows were tested with different ranges of flow velocities and heights. The basic model parameters, being the basal friction (ϕb) and normal restitution coefficient (ϵn), were calibrated using field experiments of hillslope debris flows impacting two sensors. Simulations of 50 m3 of material were carried out on a channelized surface that is 41 m long and 8 m wide. The calibration process was based on measurements of flow height, flow velocity and the pressure applied to a sensor. Results of the numerical model matched well those of the field data in terms of pressure and flow velocity while less agreement was observed for flow height. Those discrepancies in results were due in part to the deposition of material in the field test which are not reproducible in the model. A parametric study was conducted to further investigate that effect of model parameters and inclination angle on flow height, velocity and pressure. Results of best-fit model parameters against selected experimental tests suggested that a link might exist between the model parameters ϕb and ϵn and the initial conditions of the tested granular material (bulk density and water and fine contents). The good performance of the model against the full-scale field experiments encourages further investigation by conducting lab-scale experiments with detailed variation of water and fine content to better understand their link to the model's parameters.