Submitted by Renata Lopes (renatasil82@gmail.com) on 2016-04-12T19:44:42Z No. of bitstreams: 1 brunomontesanoschettino.pdf: 4192681 bytes, checksum: 0df0705f39a6ff58697ec7dc00759256 (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2016-04-24T03:29:36Z (GMT) No. of bitstreams: 1 brunomontesanoschettino.pdf: 4192681 bytes, checksum: 0df0705f39a6ff58697ec7dc00759256 (MD5) / Made available in DSpace on 2016-04-24T03:29:36Z (GMT). No. of bitstreams: 1 brunomontesanoschettino.pdf: 4192681 bytes, checksum: 0df0705f39a6ff58697ec7dc00759256 (MD5) Previous issue date: 2015-12-08 / Este trabalho propõe um método para detectar a saturação dos núcleos dos transformadores de corrente (TCs) utilizados na proteção de sistemas elétricos de potência (SEPs), além de promover a compensação de seus efeitos através da correção do sinal de corrente secundária distorcido pela saturação. Técnicas de processamento de sinais baseadas no filtro diferenciador de segunda ordem de Savitzky-Golay são utilizadas para localizar os pontos de transição entre partes do sinal de corrente distorcidas e não distorcidas pela saturação. Em seguida, um processo de estimação baseado no critério dos mínimos quadrados que utiliza exclusivamente amostras do sinal contidas nas regiões não distorcidas é efetuado, extraindo os parâmetros necessários à promoção da correção do sinal. As influências do ruído e dos desvios de frequência de operação do SEP foram analisadas, tendo sido desenvolvidos e incorporados meios adaptativos para mitigar seus efeitos. Os algoritmos desenvolvidos foram implementados em MATLAB e a avaliação de desempenho foi realizada utilizando sinais extraídos de simulações de falta ocorridas em um sistema modelado em um simulador digital em tempo real (RTDS). Os resultados indicaram que o método proposto atingiu desempenho satisfatório, independente dos parâmetros do TC e dentro de uma ampla gama de cenários da falta analisados. Além disso, o método mostrou-se robusto em relação ao ruído e eficiente na mitigação dos erros provocados pelos desvios de frequência. Ainda, os recursos técnicos e computacionais necessários para sua execução indicam que o método proposto é passível de implementação nos atuais dispositivos de proteção disponibilizados pela indústria. / This work proposes a method for detecting the saturation of the current-transformer cores used in the protection of electric power systems and promote the compensation for its effects by correcting the secondary current signal distorted due to the saturation. Signal processing techniques based on the second order differentiator Savitzky-Golay filter are used for locating the transition points between distorted and undistorted parts of the current signal. Then, an estimation process based on the least squares criteria that uses exclusively signal samples included in the undistorted regions is performed, extracting the parameters needed for the signal correction. The influences of the noise and the frequency offset were analysed, and adaptive means to mitigate their effects were developed and incorporated. The developed algorithms were implemented in MATLAB and performance evaluation was performed using the signals taken from fault simulations in a system modeled on a real time digital simulator (RTDS). The results indicated that the proposed method reaches a satisfactory performance, regardless of the CT parameters and within a wide range of analysed fault scenarios. Moreover, the method showed to be robust relative to the noise and effective in mitigating the errors due to the frequency offsets.

This thesis investigates the impact of resonant overvoltages and their origin. Series and parallel resonances are present in any electrical grid. The frequency of which this resonance occurs is called resonance frequency. For parallel resonance, which is mainly being studied in this thesis, a high impedance peak can be found at the resonance frequency. This impedance peak in conjunction with a harmonic rich current cause a kind of temporary overvoltages called resonant overvoltages. The harmonic content of the current is high following a fault clearing in the grid, due to transformer saturation. The resonance frequency is heavily dependent on the amount of reactance present in the grid, which entail that a change in reactance causes a change in the resonance frequency. The electromagnetic transient tool PSCAD has been used to investigate resonant overvoltages following transformer energization caused by faults and switching in Stockholm. Secondly, a model was created of a grid connecting off-shore wind power to the mainland via long AC submarine transmission cables. These cables, having a high capacitance, lower the resonance frequency. Faults in this model were simulated to investigate the phenomenon of resonant overvoltages in such a grid. This was especially interesting due to Swedens planned expansion of wind power in the Baltic sea. While resonant overvoltages were found in Stockholm they were not deemed significant due to their low magnitude and longevity. However, severe resonant overvoltages were found in the off-shore wind farm model. The worst resonant overvoltages had a maximum amplitude of the 2nd order harmonic voltage of 130 kV which, while eventually damped, were significant for up to 50 periods. Lastly, the phenomenon of an increased resonance frequency during the saturation of a transformer was studied. The most severe resonant overvoltages occured in a model where the frequency scans showed a resonance frequency of 98 Hz. Indicating, caution needs to be had during EMT-studies of resonant overvoltages while choosing what resonance frequency to study.