Metodologias para representação do suporte de frequência por elos HVDC e avaliação de desempenho

Abstract

In its first part, this work presents an alternative way to evaluate convergence problems using the principal component analysis (PCA) mathematical technique. This methodology aims to identify failures in data entry or interactions between control devices in the power flow. For comparison purposes, this technique is confronted with the modal analysis by eigenvalues and eigenvectors. In the second part, a methodology is proposed to include the participation of direct current links in the primary frequency control. The different ways to vary the link power with frequency are discussed, with three models being presented and integrated into the power flow: the conventional one, the Stab50 (used in Itaipu) and an alternative model used to ensure voltage fluctuation suppression (VFS) on adjacent AC buses. In this dissertation, the models and methodologies were developed in MATLAB. The software Anatem, developed by CEPEL, was used to validate some results. The results obtained through a case of the Brazilian South/Southeast system demonstrated the ability of the ACP to identify conflicts in large systems. In addition, tests with a 3-area tutorial system showed the ability of the ACP to correctly identify mutual conflicts between controls. In the case of frequency control, the results verified the accuracy of the proposed models and pointed to the technique with SFT and variable frequency gain as promising, since it was able to keep the voltage stable in the adjacent bars of the link during the active power variation. Furthermore, this control model allowed precise frequency support between power systems without the need for additional communication. The results obtained through the test systems demonstrated ACP’s ability to identify mutual conflicts between controls correctly. In the case of frequency control, the results verified the accuracy of the proposed models and indicated the technique with VFS as promising.