Équivalents dynamiques pour l'étude de la stabilité transitoire

Abstract

Today’s transient stability (TS) studies are a necessary tool to predict power systems behavior after the occurring of huge faults (ex: short-circuit, generator’s lost, etc.). These studies allow us to plan the necessary actions to take in order to maintain the required levels of availability and quality of the electricity production. However, because of the fast increase of the size and the complexity of the power systems, these (TS) studies require higher duration and efforts for their calculations to be processed. The problem of this thesis is therefore: How to reduce the (TS) programs duration and calculation efforts while keeping the precision of their results? In order to resolve this problem, an initial literature review was performed. This literature review allowed us to notice that power systems reduction techniques used to decrease duration of (TS) simulation programs were designed long time ago. These techniques have the common goal of reducing the (TS) simulation duration and effort while keeping the same original power system dynamic behavior. This literature review leaded us also to choose a four steps power system reduction method: In the first step, the power system internal and external zones are chosen. In the second step the power system external zone generators able to be aggregated together are identified using the slow coherency method in addition to a time domain (TS) simulation. In the third step, dynamic parameters of the equivalent reduced network machines are derived using the structure preserving method which has the advantage of being useful for both classical and detailed power systems models. In the fourth step, the equivalent power system admittance matrix is processed using Zhukov method. The method proposed to resolve the problem of this thesis was afterwards implemented under Matlab© functions format. These aggregation functions were validated in the case of (TS) and (SIME) programs for the two following test network: « New England » 10 machines 39 buses under classical and detailed models, and the IEEE 50 machines 145 buses under classical model. PSSE© industrial software was also used in order to validate the results. The designed aggregation functions allowed us to have a significant increase of the speed of the (TS) and (SIME) programs while keeping a good accuracy of their outputs.

Date	22 Jan 2014
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Louis-A. Dessaint (Supervisor)