Discussion of “ A New Preconditioned Conjugate Gradient Power Flow ”

The authors [1] should be commended for their interesting contribution to the power flow solution via iterative methods. The availability of a good preconditioner is perhaps the decisive factor influencing whether or not iterative solvers would have an impact in the future of power flow calculations in power systems. The paper contributes in this direction. I would like to bring to the attention of the readers a recent publication on the subject [2]. In [2], we have preferred the use of the (full) Newton power flow (NPF) method over the fast decoupled (FD) approach as the basis for the implementation of iterative solvers for the following two reasons. 1) In both the NPF and the FD using the traditional LU decomposition and substitutions, the greatest amount of computing time is spent in the factorization of the large matrices. Among other reasons, the use of FD is so widespread because only one LU factorization is needed. However, when employing iterative solvers, there is no factorization involved and we should aim to minimize the number of linear set of equations to be solved. In the FD approach, as the size of the system increases, the number of (external) iterations also increases (see Table I of the paper). The same phenomenon does not occur in NPF; see [3] for example. 2) There are situations where the FD fails to converge: under high loading conditions and/or when the transmission systems have large R/X ratios. This is typical of the radial systems frequently found in developing countries. Iterative solvers would find their applications in the solution of power flows for (very) large systems. Therefore, I feel that future efforts in developing preconditioners should focus on the NPF method rather than on the FD approach. The authors’ comments would be very much appreciated.