A Meta-Model-Based Procedure for Quantifying the On-Site Efficiency of Rockfall Barriers

This article proposes a procedure for developing tools to quantify the on-site efficiency of any rockfall barrier. This procedure relies on meta-modeling techniques to predict the barrier ability in arresting rock blocks, whatever their trajectory. For demonstration purpose, a specific low-energy barrier for which a finite element model was available is considered. The barrier response is simulated varying six parameters describing the rock block kinematics. Six different methods are used to create meta-models predicting the simulated barrier response. The ability of each method in creating meta-models with good prediction capacities is evaluated. Meta-models created utilizing the best methods are then used to quantify the efficiency of the barrier in arresting rock blocks in two real situations. These situations exhibit very similar 95% percentiles of the block passing height and kinetic energy but very different distributions for the other parameters describing the kinematics of the rock blocks. The predictions reveal that the barrier efficiency is extremely site-dependant. The discussion addresses the meta-models performance and highlights the benefits in using such meta-models for quantifying the barrier efficiency, in particular with respect to more classical barrier design approaches. Last, the proposed eight-step procedure for generating meta-models to be used in operational contexts is described.