Implementation of Two Different Moving Window FDTD Methods to Simulate the Electromagnetic Propagation in Tunnel with Parallel Computing
Yunfei Mao, Jiahong Chen, Liang Pan, Xiaoming Wang
Abstract
Yunfei Mao, Jiahong Chen, Liang Pan, Xiaoming Wang
Abstract
Tunnel engineering is electrically large compared with the GHz electromagnetic pulse (EMP), it is difficult to simulate the EMP propagation in large-scale and long-distances vaulted tunnel by using the conventional finite- difference time-domain (FDTD) method. In this work, based on the parallel computing, two kinds of moving window FDTD (MW-FDTD) methods are presented to simulate the EMP propagation in tunnel, the results are validated by comparing with the results of the conventional parallel FDTD method. The convolution PML (CPML) is adopted to truncate the computation domain, which reduces the reflection error greatly. The accuracy and the efficiency of the proposed method are proposed by comparing with the conventional method. Results show that the relative errors for the Alternate MW- FDTD (AMW-FDTD) and the Chain MW-FDTD (CMW-FDTD) are 0.11% and 0.43%, respectively. The CPU time for the AMW-FDTD method can be reduced to about 45% of the conventional FDTD method, while the CMW-FDTD method can be reduced to about 35%.
A significance statement is not available in the OpenAlex record.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
Tunnel engineering is electrically large compared with the GHz electromagnetic pulse (EMP), it is difficult to simulate the EMP propagation in large-scale and long-distances vaulted tunnel by using the conventional finite- difference time-domain (FDTD) method. In this work, based on the parallel computing, two kinds of moving window FDTD (MW-FDTD) methods are presented to simulate the EMP propagation in tunnel, the results are validated by comparing with the results of the conventional parallel FDTD method. The convolution PML (CPML) is adopted to truncate the computation domain, which reduces the reflection error greatly. The accuracy and the efficiency of the proposed method are proposed by comparing with the conventional method. Results show that the relative errors for the Alternate MW- FDTD (AMW-FDTD) and the Chain MW-FDTD (CMW-FDTD) are 0.11% and 0.43%, respectively. The CPU time for the AMW-FDTD method can be reduced to about 45% of the conventional FDTD method, while the CMW-FDTD method can be reduced to about 35%.
Key concepts: Finite-difference time-domain method, Electromagnetic pulse, Computation, Computer science, Convolution (computer science), Finite difference method, Computational electromagnetics, Computational science