The choice of the reduced models for large-scale power systems

This thesis is concerned with the application of singular perturbation tech-niques to a class of power system dynamics. Power systems possess the multi-ple time scale property, so singular perturbation techniques are appropriate forstudying power systems characteristics. Using singular perturbation, we obtainedreduced order models for these systems which alleviate the stiffness and dimen-sionality problems of the full order models.In This thesis, second order eigenvalue sensitivities of reduced order modelswith respect to singular perturbation parameter £ have been derived and theirimportance has been illustrated. These sensitivities have been used to obtainbetter estimates for the system modes, especially in case of turning points. Us-ing these sensitivities, reduced order models have been obtained based upon theeigensensitivity concept.An approach based upon the participation principle has been presented to de-fine the separation boundaries between slow and fast subsystems. Also, the validityof the proposed approach and derived formulae have been tested by an applicationto large-scale power system of SCECO-C-SCECO-E system. Finally, the applica-tions showed good agreement with the exact solutions of the eigenproblems.