This paper investigates the GMRES iterative method performance for solving the sublinear problem associated to the DAEs solution present in power system stability analysis in time domain. It is shown that the unpreconditioned GMRES fails due to a phenomenon known as "stagnation". It was used an incomplete factors preconditioner with threshold - ILUT, unsuccessfully. Through the preconditioner image matrix (original power system Jacobian matrix) condition number, it was checked its bad condition, resulting, therefore, in a low quality preconditioner, justifying the GMRES failure. The image matrix was then preprocessed (scaling, normalizing and reordering) improving its characteristics acting positively over the preconditioner quality. After these actions, the GMRES solved efficiently the linear systems associated to each numerical experiment, all presenting bad conditioned original Jacobian matrices.
BDF; GMRES; Preconditioner; Jacobian Matrix; Power Systems Stability
