Arc-based modeling, condition monitoring and failure detection of on-load tap-changers

Power transformers are highly valuable units that are extensively used throughout the power systems to change and regulate different voltage levels. The safe, reliable, and continuous operation of power transformers is extremely critical for safeguarding the continuity of power delivery. One of the most expensive and vulnerable parts of power transformer is its on-load tap-changer (OLTC), i.e., an electro-mechanical switching equipment which operates to regulate the output voltage level of power transformer to compensate the over and under voltages resulted from load variations by changing the winding step ratio. An OLTC is also considered as the most vulnerable equipment of power transformers, because it is the only moving part of transformers. Therefore, it is very important to continuously monitor the operating conditions of OLTC. There is a sequence of switching incidents with certain timings during each OLTC operation. These switching events can be compared to the switching incidents of circuit breakers (CBs). CBs are electro-mechanical switching devices which operate widely within the power systems. The principle objective of CBs is to interrupt the fault current in a safe, reliable, and timely manner. As a CB operates through its lifetime, it undergoes severe electrical, mechanical, and thermal stresses that degrade its operating condition. This is the case for OLTCs as well, because they operate in a very similar manner. Thereby, it is highly crucial to continuously monitor the operating conditions of OLTCs for each tap-changing operation. There are three types of condition monitoring (CM) and failure detection (FD) methods applicable for CBs and OLTCs that include off-line, on-line, and real-time. In off-line methods, all measurements are obtained from de-energized equipment during out of service conditions. In on-line techniques, measurements can be taken from equipment under normal operating conditions whenever required, whereas in real-time methods, measurements must be collected continuously utilizing real-time sampling techniques. Considering recent developments in measurement technologies and smart grid plans, the CM and FD techniques based on real-time approaches have become more realistic and justifiable. This thesis presents a new modeling principle for OLTCs based on the CB arc models. The working principle of OLTCs is based on a sequence of switching operations with certain timing intervals. These switching operations are individually modeled and thus a full arc model of OLTC which is capable of performing a full tap-changing operation is derived. Fault location algorithms are influenced by the OLTC operation because it changes the impedance, characteristic impedance, and therefore the traveling wave velocity of the transmission and distribution lines. The proposed OLTC model can be effectively used to further investigate the probable effects of tap-changing operations on different types of fault location algorithms and improve the performace of these techniques. Whenever a CB operates, the medium between its separating contacts becomes highly ionized because of the high temperature originated by the high current flow before its interruption. In this case, the arcing current of CB continues to flow even after the full physical separation of CB's arcing contacts. This phenomenon is called an electric arc which is the key point in CB modeling. Several arc models have been proposed in the literature during the past decades. The combination of two well-known arc models will be used in this thesis. The first model has been introduced by A. M. Cassie in 1939, in which an electric arc channel with fixed values of temperature, electric field intensity, and current density is considered. In this model, the arc conductance is a function of the arc cross section and the arcing energy is obtained by the convection law. It can characterize the arc behavior in high current vicinities. The second arc model has been proposed by O. Mayr in 1943, in which the arc temperature is highly variable but the size and shape of the arc column is fixed. Therefore, the principal mechanism of energy losses is based on the thermal conduction law. It can model the arc characteristics in low current vicinities accurately. As mentioned earlier, the Cassie and Mayr arc models can represent the arc behaviors in high and low current vicinities, respectively. Therefore, the combination of both arc models can simulate the arc behaviors within the entire current spectrum. The difference between the instantaneous power of the input and output terminals of CBs and tap-changing transformers is used to calculate the instantaneous arcing power and its related energy relating to the CB and OLTC operations. The power difference of a CB is always zero as long as a CB is fully closed or open. On the other hand, under normal operating condition, the instantaneous power difference of transformers equals the transformer power loss. However, during the CB and OLTC operations, the power difference becomes the arcing power of CBs and tap-changing transformers. For CBs, the instantaneous arcing power is directly available under arcing conditions. However, in case of OLTCs, the instantaneous power loss of power transformers must be estimated and subtracted from the instantaneous power difference in order to obtain the instantaneous arcing power. Then, the cumulative arcing energy and the integration of arcing power are used to propose a novel CM and FD algorithm for CBs and OLTCs. In this thesis, it has been demonstrated that the cumulative arcing energy of CBs and OLTCs is a good indication of the electrical wear of CB's and OLTC's arcing contacts. Therefore, it is used for the CM of CBs and OLTCs. In addition to the CM section, the integration of arcing power for a period of a single opening or tap-changing operation is used to identify any failure in CB or OLTC operation. The moving average filter has been used to smooth the output sequence. In this regard, certain number of points has been taken from the input sequence to produce the output data points. Therefore, more reliable data will be available for the proposed algorithms. The CM threshold quantity which remains constant throughout the implementation of CM algorithm is predetermined for each type of CB and OLTC based on the number of permissible operations at the rated capacity of CB and rated step capacity of OLTC which is defined by manufacturers. The alarm criteria of the proposed CM algorithm is evaluated using the cumulative arcing energy associated to CB or OLTC operations. In this regard, the cumulative integration/summation of the measured arcing powers associated to CB or OLTC operations (i.e., the cumulative arcing energy) is compared to the CM threshold quantity of the proposed CM algorithm following each opening or tap-changing operation. Whenever the accumulated value exceeds the predefined threshold value, an alarm signal is initiated to inform the system operator. This alarm signal indicates that the maximum contact wear of CB or OLTC has been reached and the inspection has become necessary. Otherwise, the algorithm continues to measure and store the arcing energy. On the other hand, the maximum permissible arcing times of CBs are usually defined by manufacturers and are slightly different for various types of CBs. Their values are ranging from 15 ms to 25 ms that indicate the maximum arcing durations that CBs can safely undergo without being defected. Clearly, the measurement of CB arcing times is a highly complicated job. Therefore, the proposed FD algorithm indirectly evaluates the CB arcing times utilizing the integration of arcing powers over time that is actually the arcing energy related to various arcing durations. If the integration of normalized arcing power is greater than the FD threshold quantity, the CB has failed to function properly and therefore a trip signal is initiated to the same CB once again or upstream CBs for fault isolation. Regarding the OLTCs, whenever the estimated arcing energy for a single tap- changing operation reaches the FD trip level, a trip command is initiated to disconnect the power transformer from the power system. Otherwise, the FD algorithm remains stable and evaluates next tap-changing operations. The results of the computer simulation studies prove that the CM algorithm effectively monitors the electrical wear of CB's or OLTC's arcing contacts to evaluate the time of maintenance. On the other hand, the results obtained from the FD algorithm show that the FD algorithm is capable of distinguishing between the successful and faulty opening or tap-changing operations of CBs or OLTCs and effectively detects failure conditions. Chapter 1 includes an introcution and a literature review on CM and FD of OLTCs. Tap-changing effects on fault location algorithms and the motivations of the thesis are explained in Chapter 2. A brief explanation of different arc models of CBs and the implementation of Habedank arc model within electro-magnetic transient program restructured version (EMTP-RV) are discussed in Chapter 3. Chapter 4 contains the existing techniques for CM of CBs and a proposed CM algorithm which uses the measurement of CB power loss caused by arcing condition for this purpose. Chapter 5 discusses an available practice and a new power-based algorithm for FD of CBs. The real-time test studies for the performance analysis of the CM and FD algorithm for CBs are discussed in Chapter 6. An original modeling concept for OLTCs and its implementation details have been introduced in Chapter 7. Chapter 8 contains a modified version of the proposed power-based algorithm and the related computer simulation studies for CM of OLTCs. Another modification of the same power-based algorithm and simulation studies for FD of OLTCs are explained in Chapter 9. The real-time test studies for the performance analysis of the OLTC operation is performed in Chapter 10. The conclusions, recommendations and future research opportunity are discussed in Chapter 11.