Search for: All records

Abstract A hierarchical multiscale modeling framework is proposed to simulate flowslide triggering and runout. It couples a system‐scale sliding‐consolidation model (SCM) resolving hydro‐mechanical feedbacks within a flowslide with a local‐scale solver based on the discrete element method (DEM) replicating the sand deformation response in the liquefied regime. This coupling allows for the simulation of a seamless transition from solid‐ to fluid‐like behavior following liquefaction, which is controlled by the grain‐scale dynamics. To investigate the role of grain‐scale interactions, the DEM simulations replace the constitutive model within the SCM framework, enabling the capture of the emergent rate‐dependent behavior of the sand during the inertial regime of motion. For this purpose, a novel algorithm is proposed to ensure the accurate passage of the strain rate from the global analysis to the local DEM solver under both quasi‐static (pre‐triggering) and dynamic (post‐triggering) regimes of motion. Our findings demonstrate that the specifics of the coupling algorithm do not bear significant consequences to the triggering analysis, in that the grain‐scale dynamics is negligible. By contrast, major differences between the results obtained with traditional algorithms and the proposed algorithm are found for the post‐triggering stage. Specifically, the existing algorithms suffer from loss of convergence and require proper numerical treatment to capture the micro‐inertial effects arising from the post‐liquefaction particle agitation responsible for viscous‐like effects that spontaneously regulate the flowslide velocity. These findings emphasize the important role of rate‐dependent feedback for the analysis of natural hazards involving granular materials, especially for post‐failure propagation analysis. 

Machine-learning (ML) technology has been a key enabler in the push towards realizing ever more sophisticated autonomous-driving features. In deploying such technology, the automotive industry has relied heavily on using "black-box" software and hardware components that were originally intended for non-safety-critical contexts, without a full understanding of their real-time capabilities. A prime example of such a component is CUDA, which is fundamental to the acceleration of ML algorithms using NVIDIA GPUs. In this paper, evidence is presented demonstrating that CUDA can cause unbounded task delays. Such delays are the result of CUDA’s usage of synchronization mechanisms in the POSIX thread (pthread) library, so the latter is implicated as a delay-prone component as well. Such synchronization delays are shown to be the source of a system failure that occurred in an actual autonomous vehicle system during testing at WeRide. Motivated by these findings, a broader experimental study is presented that demonstrates several real-time deficiencies in CUDA, the glibc pthread library, Linux, and the POSIX interface of the safety-certified QNX Operating System for Safety. Partial mitigations for these deficiencies are presented and further actions are proposed for real-time researchers and developers to integrate more complete mitigations.