期刊
COMPUTERS AND GEOTECHNICS
卷 118, 期 -, 页码 -出版社
ELSEVIER SCI LTD
DOI: 10.1016/j.compgeo.2019.103351
关键词
GPU-based parallel computation; CUDA; Numerical manifold method; Excavation damaged zone; Deeply buried rock tunnel
类别
资金
- National Natural Science Foundation of China [41772309]
- National Key Research and Development Program of China [2017YFC1501302]
- Major Program of Technological Innovation of Hubei Province, China [2017ACA102]
In this study, combined with the zero-thickness cohesive element (ZE) model and explicit integration method, a parallelization technique based on graphics processing units (GPU) is proposed to accelerate the computational efficiency of the NMM for modeling the formation of the excavation damaged zone in deep rock tunnels. To optimize the performance of the original NMM when simulating rock masses with fine grains, a ZE model is adopted and a speedup ratio of 41 compared with the original NMM is achieved. To simulate the behavior of the failed rock masses more efficiently and to achieve a higher speedup ratio after parallelization, the explicit integration scheme is introduced to avoid the assembly and solving of linear equations. To further improve the computation efficiency of hardware-based approaches, the GPU-based parallel technique, including a hybrid CPU GPU framework, a 'single instruction on multiple data' (SIMD) model and a rainbow coloring strategy, is adopted and a speedup ratio of up to 34.7x is achieved. Finally, a series of excavation models are simulated, and the predicted results validate the efficiency and capability of this developed method for modeling the formation of the excavation damaged zone in deeply buried rock tunnels.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据