Assessing Rockfall Barrier Performance Through Block Propagation Codes and Meta-models

The performance of rockfall barriers have been traditionally assessed using standardized impact tests such as SEL and MEL (EOTA 2018); that quantify the barrier service and maximum energy levels, respectively. Unfortunately, these types of tests only provide information for very specific loading conditions, i.e. centered impacts, that fail to consider the way different loading conditions affect the barrier response. It has been shown that, given a certain block kinetic force, changes in the block size (Buzzi and Krummenacher 2014) and impact position (Zhao et al. 2020), can make the difference between withstand and failure. In recent years, numerical approaches such as Finite (FEM, Gentilini et al. 2013) and Discrete (DEM, Bertrand et al. 2012) Element Method have been used to overcome these limitations and investigate more complex loading conditions (Song et al. 2018). However, this type of analysis is still mostly limited to research institutions and universities as it requires a certain degree of computational power and technical expertise, which might not be available to the end user (i.e. engineering firms). Metamodels, also known as surrogate models, are a type of high-level models that aim to approximate the behavior of lower-level, higher fidelity models, reducing the computational burden. They have shown promising results in the scope of rockfall protection, where they have been used to predict failure modes under different loading conditions (Mentani et al. 2016; Toe et al. 2018). In this paper, we combine the efficiency of a meta-model built using a DEM barrier model with the predictive abilities of the block propagation code Hy-Stone (Lanfranconi et al. 2020) to provide site-specific results with increased computational efficiency and ease of use compared to the standard numerical approaches.