Bootstrap Gauss-Seidel load flow

The Gauss-Seidel iterative procedure for load-flow analysis is so well established (and the use of acceleration and extrapolation to hasten convergence so well documented) that the subject might appear to be exhausted. However, a novel variation of the process has now appeared with a convergence rate far superior to conventional accelerated or extrapolated Gauss-Seidel techniques. In this new approach, called `bootstrap Gauss-Seidel¿, a pseudo-, or fictitious, load flow is set up by making simple changes to the power loading at every node in the network. Since these changes are dependent on the mismatch power (or error) at each node, a `bootstrap effect¿ results. Conventionally acceptable solutions are reached considerably faster than normal, since convergence is very rapid. Particular benefit is seen on large networks, which tend to converge at a rate up to three times faster than that obtained using optimum real acceleration. The additional mismatch calculation (not normally required in Gauss-Seidel methods) can be made computationally attractive, and is not a serious limitation of the bootstrap approach. Results obtained for about a dozen a.c. load flows, including standard IEEE test cases, have confirmed the power of this method. Detailed results are included for a number of these cases.