TY - JOUR
T1 - Fine-Grained Fully Parallel Power Flow Calculation by Incorporating BBDF Method into a Multistep NR Algorithm
AU - Su, Xueneng
AU - Liu, Tianqi
AU - Wu, Lei
N1 - Publisher Copyright:
© 1969-2012 IEEE.
PY - 2018/11
Y1 - 2018/11
N2 - In recognizing urgent needs in fast calculation of AC power flow (PF) problems, PF computation has been explored under different parallel computing platforms. Specifically, a block-bordered-diagonal form (BBDF) method has been widely studied to permute linear equations in PF calculations into a BBDF form for facilitating parallel computation. However, determining an optimal network segmentation scheme that leads to the best speedup ratio of BBDF-based parallel PF is challenging. As a first contribution, this paper proposes a node-tearing-based approach to determine the optimal network segmentation scheme, which leverages sizes of subnetworks and the coordination network to achieve the best speedup ratio of BBDF-based parallel PF calculation. In addition, a fine-grained fully parallel PF approach is proposed to further enhance parallel performance, in which all three key steps of the Newton-Raphson based PF calculation are implemented in parallel. Studies illustrate effectiveness of the proposed network segmentation method and the fully parallel PF approach.
AB - In recognizing urgent needs in fast calculation of AC power flow (PF) problems, PF computation has been explored under different parallel computing platforms. Specifically, a block-bordered-diagonal form (BBDF) method has been widely studied to permute linear equations in PF calculations into a BBDF form for facilitating parallel computation. However, determining an optimal network segmentation scheme that leads to the best speedup ratio of BBDF-based parallel PF is challenging. As a first contribution, this paper proposes a node-tearing-based approach to determine the optimal network segmentation scheme, which leverages sizes of subnetworks and the coordination network to achieve the best speedup ratio of BBDF-based parallel PF calculation. In addition, a fine-grained fully parallel PF approach is proposed to further enhance parallel performance, in which all three key steps of the Newton-Raphson based PF calculation are implemented in parallel. Studies illustrate effectiveness of the proposed network segmentation method and the fully parallel PF approach.
KW - Block bordered diagonal form
KW - Newton-Raphson algorithm
KW - parallel computation
KW - power flow
UR - http://www.scopus.com/inward/record.url?scp=85046812493&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85046812493&partnerID=8YFLogxK
U2 - 10.1109/TPWRS.2018.2834734
DO - 10.1109/TPWRS.2018.2834734
M3 - Article
AN - SCOPUS:85046812493
SN - 0885-8950
VL - 33
SP - 7204
EP - 7214
JO - IEEE Transactions on Power Systems
JF - IEEE Transactions on Power Systems
IS - 6
M1 - 8356661
ER -