A new adaptive dynamic reduction method for power system transient stability problems

•The proposed method determines weight of each generator based on its contribution to error of the reduced-order system.•It is shown that the resulting reduced-order system is as accurate as full-order system as long as all generators in a group are tightly coherent.•A new methodology is presented to identify non-coherent generators and improving the accuracy of the reduced-order system.•IEEE test system with 50 generators and 145 buses and the commercial-grade DSATools are used to validate the proposed method.•The proposed method is compared with the inertial and slow coherency aggregation method and it is demonstrated to be more accurate.

Aggregating coherent generators is an important step in dynamic equivalencing for creating reduced-order systems. Most traditional/available methods tend to keep the dynamics of large generators since these generators provide large inertia and potentially a significant amount of power to the system. However, this methodology loses its accuracy if, for example, a large generator is not able to provide enough power in the post-fault state of the system. The typical methods may also fail to create an accurate reduced-order system when the generators in a group are not tightly coherent. This manuscript presents a new adaptive aggregation method, wherein large weights are assigned to the generators which are capable of introducing large errors in the solution of reduced-order system. The parameters of generators as well as structure of transmission network are taken into account to determine weights of generators, which help to improve the accuracy of the reduced-order system. Moreover, a new criterion is defined to evaluate the accuracy of the reduced-order system. Using the proposed criterion, the non-coherent generators can be identified and separated from the corresponding group. The parameters of the reduced-order system are subsequently updated to preserve the accuracy of the solution. The effectiveness of the proposed method is demonstrated using a IEEE 50-gen test system and the results are compared with the Powertech Labs’ DSATools and inertial aggregation algorithm. It is shown that the reduced-order system created by the proposed aggregation method is more accurate than traditional aggregation methods. In addition, the proposed criterion is able to identify and isolate non-coherent generators, which further improves the accuracy of the reduced-order system.