Study on large deformation of soil–rock mixed slope based on GPU accelerated material point method

Abstract

Abstract This study assesses the effect of stone content on the stability of soil–rock mixture slopes and the dynamics of ensuing large displacement landslides using a material point strength reduction method. This method evaluates structural stability by incrementally decreasing material strength parameters. The author created four distinct soil–rock mixture slope models with varying stone contents yet consistent stone size distributions through digital image processing. The initial conditions were established by linearly ramping up the gravity in fixed proportionate steps until the full value was attained. Stability was monitored until a sudden shift in displacement marked the onset of instability. Upon destabilization, the author employed the material point method to reconstruct the landslide dynamics. Due to the substantial computational requirements, the author developed a high-performance GPU-based framework for the material point method, prioritizing the parallelization of the MPM algorithm and the optimization of data structures and memory allocation to exploit GPU parallel processing capabilities. Our results demonstrate a clear positive correlation between stone content and slope stability; increasing stone content from 10 to 20% improved the safety factor from 1.9 to 2.4, and further increments to 30% and 40% ensured comprehensive stability. This study not only sheds light on slope stability and the mechanics of landslides but also underscores the effectiveness of GPU-accelerated methods in handling complex geotechnical simulations.

Keywords

• Material point strength reduction method
• Soil–rock mixture
• Slope stability
• Stone content
• Large deformation
• GPU-based MPM