Please use this identifier to cite or link to this item:
Authors: Kumar, Jagish
Issue Date: 2010
Abstract: Almost all bulk electric power generated and transported is consumed in alternating current (ac) networks. Reactive power is an integral part of the ac power systems because of the nature of loads and transmission media. Consumption of reactive power lowers voltage magnitudes while generation of it increases voltage magnitudes in ac networks. In general, most of the loads connected in power systems are inductive in nature and placed very far from generating stations, thereby drawing reactive power from the source. The flow of reactive power over long distances through ac networks produces voltage drop and transmission line losses; hence efficiency and voltage level decrease at the receiving end. The operation of ac systems at low voltage produces undesirable effects such as degradation of operating performance of electrical equipments, voltage stability problem etc. To ensure reliable and stable operation of power systems, better utilization of the existing ac networks and satisfactory operation of the end-user equipments, proper voltage control of the system through reactive power compensation is required all the time. At the distribution level, the reactive power compensation is also used to improve the supply power factor and at transmission level, it is further used for stabilizing the power systems as well. From the early days of ac power systems, rotating synchronous condensers and mechanically switched capacitor and inductor banks have been in use for this purpose. However, as these are mechanical devices, the main problems associated with these devices are their slow speed of response and mechanical wear and tear. Most significant advances in reactive power compensation took place in 1970s with the advent and use of power electronics based semi conductor devices (thyristors) in place of mechanical switches. In typical applications, thyristor switched capacitors (TSC) provide variable leading reactive power (vars) while thyristor controlled reactors (TCR) provide variable lagging vars, and the parallel combination of these two can be used for continuous control of vars over wide range from leading to lagging. Because of the use of thyristors, the drawbacks of mechanical switched capacitors and inductors were eliminated. However, some issues such as large size of capacitor and inductor banks, dependency of reactive power compensation on operating voltage etc. still remain even though a large number of TCR and TSC units have been successfully installed and operated for many years around the world. Simultaneously, the potential of vars compensators based on static power converters (mainly voltage source converters) have also been recognized and investigated to overcome the above drawbacks. In the early days, the voltage source inverters (VSI) were based on thyristor switches and as a result, they allowed only unidirectional flow of reactive power unless complex force-commutation circuits were used. The invention of self commutating devices such as GTO thyristors in 1990s allows bidirectional flow of reactive power. Consequently, successful operation of VSI based var compensator took place and it was named as static synchronous compensator (STATCOM). The STATCOM is one of the members of Flexible AC Transmission Systems (FACTS) devices. The first prototype GTO-based STATCOM was developed, tested and installed by utilities in 1995. The different topologies of VSI for STATCOM applications can be categorized into two groups: multipulse and multi-level. The early installations of STATCOM are based on multipulse type VSI where basic six-pulse inverter units are interconnected through zigzag transformer for synthesizing high pulse output ac voltage in order to reduce the harmonic distortions in it. The main drawbacks of multipulse type inverter are the need of zigzag transformer. Because of this, the system becomes more complex, more costly and it also requires large space for installation. To overcome the drawbacks of multipulse type inverter, the application of multilevel inverter (MLI) topologies for STATCOM has been proposed and studied extensively in the literature. The MLI synthesizes desired output voltage from several levels of input dc voltages. It has many advantages in comparison with the hard-switched two-level pulse width modulation (PWM) inverter such as capability to operate at high voltage with lower dv/dt per switching, high efficiency and low electromagnetic interference (EMI) etc. By connecting sufficient number of dc levels, a nearly sinusoidal voltage of high magnitude can be produced at fundamental frequency at the output of an MLI. The MLIs are further classified as diode- clamped, flying capacitors and cascade multilevel inverter (CMLI). Among these three, CMLI has a modular structure and requires least number of components as compared to other two topologies and as a result, it is receiving increasing attention for use in many different applications such as electric drives, utility interfacing of renewable energy sources, STATCOM etc. For STATCOM applications, the separate dc sources are replaced by dc capacitors. Now, for proper integration of CMLL STATCOM at high voltage high power level, the output ac voltage generated by CMLI must meet limitations on individual harmonic components as well as on total harmonic distortion (THD) as specified in various standards like IEEE-519, IEC 61000-2-2, EN 50160 etc. to minimize the undesirable effects of harmonics in the power systems. Although several switching strategies for CMLI has been proposed in the literature, none of these strategies has addressed the ii issues of limitations on individual harmonic component as well as on THD as imposed by different standards. Further, in the literature, CMLI based STATCOM has primarily been used for compensation of load reactive power, not for controlling the load voltage directly. As it is not necessary for the load voltage to be maintained at the rated value even in the event of 100% compensation of its reactive power by the STATCOM, the issue of controlling the load voltage directly by the CMLI STATCOM is still largely unexplored in the literature. For satisfactory load voltage control, a proper control system for the CMLI STATCOM needs to be designed, which, in turn, requires a suitable model/transfer function of the STATCOM. However, model development of a CMLI is not a trivial task because of the following issues: the dynamic properties of individual H-bridges are different, the H-bridges have individual control capabilities, and longer conduction periods imply that the dc voltage dynamics between consecutive switching can not be neglected. Therefore, a suitable technique still needs to be devised for development of an accurate model of a CMLI STATCOM. In this thesis an attempt has been made to address the issues described above. To begin with, the problem of determining a suitable switching strategy for CMLI STATCOM has been addressed. Now, the most common method reported in the literature for switching a CMLI is selective harmonic elimination (SHE) method. The problem with SHE technique is that the switching angles can not be computed for whole range of modulation index (m) from 0 to 1. To overcome this drawback, an optimization based technique which minimizes harmonics up to 13th order (non-triplen odd harmonic components only) is proposed to compute switching angles for complete range of m (from 0 to 1) which is required for continuous and smooth variation of ac output voltage of the inverter. The switching angles computed using SHE technique and optimization method for those values of m for which solutions of SHE method exist are found to be same. After obtaining the solutions for the switching angles, a detailed study has been carried out analytically and through MATLAB simulations regarding the magnitude of individual harmonic components, total harmonic distortion (THD), the weighted total harmonic distortion (WTHD) and different harmonic losses produced by the harmonic components present in ac output voltage of CMLI. It has been found that the magnitudes of some individual harmonic components and THD exceeds the limits set by different standards on harmonics such as IEEE-519, IEC 61000-2-2, EN 50160 etc. for most of the values of m. Also the harmonic losses have been found to be high. To overcome this problem, switching angles have been calculated.using the above optimization technique which minimizes harmonics up to 49th order (non-triplen odd harmonic only) and same study as discussed above has been carried out. It has been found that the individual harmonic components as well as the THD are well below the limits set by different standards for iii most of values of m and also harmonic losses are less. Analytical and simulated results are verified experimentally on a laboratory prototype, 11-level CMLI set-up and it has been found that the experimental results are in close agreement with analytical and simulated results. Therefore, the second switching strategy (in which the harmonics up to 49th order are minimized) can be considered to be superior to the first strategy (in which harmonics up to 13th order only are minimized) and therefore, is the final switching strategy proposed in this thesis for CMLI. To extend the application of CMLI based STATCOM for controlling the bus voltage of a distribution system directly; both indirect and direct control philosophies (which have been suggested in the literature for controlling any general STATCOM) have been explored in this thesis. In the indirect control philosophy, the bus voltage is controlled by varying the angle between the bus voltage and the STATCOM output voltage (henceforth termed as load angle) while keeping the modulation index fixed. On the other hand, in the direct control scheme, the bus voltage is controlled by varying 'm' while the dc capacitor voltages are maintained constant with the help of the load angle. Now, for both these control philosophies, appropriate transfer functions are required for designing the appropriate controller. To obtain these transfer functions while taking into account the various issues of CMLI as described earlier faithfully, system identification technique has been used in this thesis. Towards this goal, a basic configuration of the distribution system has been considered in which a load bus is supplied from a substation through a feeder. In order to maintain the rated voltage at load bus, a CMLI based STATCOM is connected at the same bus. For designing an appropriate controller corresponding to the indirect control scheme, a transfer function relating the load bus voltage and load angle is required. To obtain this transfer function, a pseudo random binary signal (PRBS) is applied at the input (load angle) and corresponding load voltage variations is recorded. Subsequently, a system identification (SI) technique based on prediction error method (PEM) has been applied using the obtained input-output data set for the estimation of system transfer function. For the same configuration, a fundamental frequency based model has also been developed analytically. To compare the accuracies of these two models (one obtained analytically and another by using SI technique), a common PRBS has been applied to these two models. The resulting variations of the bus voltage obtained from these two models have been compared with the time-domain bus voltage variation obtained with MATLAB/SIMULINK simulation study for the same PRBS input. It has been observed that the time domain output is closer to the output of the identified model than that of the analytical model. Thus, the identified model is more accurate representation of the system under study and hence it has been used to design the voltage controller. A PI iv controller has been chosen for this purpose and its parameters have been found out through Ziegler-Nichols (Z-N) technique. The performance of the PI controller has been evaluated for load bus voltage regulation under varying loading condition through simulation studies, and it has been found to be satisfactory. A rotating switching scheme has been implemented in order to maintain equal voltage across each capacitor connected to the individual H-bridges of CMLI. The system identification technique has further been applied for determining the transfer functions corresponding to the direct control scheme also. The same configuration of distribution system as that chosen for indirect control scheme has been used again. In case of direct control scheme, two transfer functions have been determined; one relating load bus voltage and modulation index (m), and the second one relates the dc capacitors voltages and the load angle. These two transfer functions have also been derived through analytical technique. By following the same procedure as discussed in the indirect control scheme, the identified models have been found to be more accurate representation of the study system. Subsequently, two PI controllers have been designed using Z-N technique corresponding to these two identified transfer functions. The rotating switching scheme for capacitors charge balancing has been implemented for direct control scheme also. The performances of the controllers for load bus voltage regulation and capacitors charge balancing under varying loading condition have been investigated using simulation studies. It has been observed that load bus voltage regulation is very fast in case of direct control as compared to the indirect control scheme. For the verification of the simulated results, two laboratory prototypes of CMLI based STATCOM have been developed. These two prototypes are; i) single phase, 11- level and ii) three phase, five level. Both these prototypes have a rating of 1 KVAR/240 V (peak line- to-line voltage). The switching device selected is MOSFET IRF 740. Different hardware components as required for the operation of the experimental set-up such as delay circuit, pulse amplification and isolation circuit, voltage and current sensors circuit etc. have been designed and developed. The firing pulses have been generated using real-time simulation with the help of DSP of dSPACE 1104 R&D controller board with fixed-point calculations. The real-time workshop (RTW) of MATLAB and real-time interface (RTI) feature of dSPACE result in the real-time simulation of the SIMULINK model of controller which results in the generation of control pulses. These generated control pulses.are given to semiconductor devices of each H-bridge through delay and isolation and amplification circuit in real time. Load bus voltage and dc capacitors voltage have been sensed and are given to ADC channels of dSPACE for comparison with reference values in order to generate error signals for PI controllers. Experimental results have been obtained for the same configuration corresponding to MATLAB/SIMULINK based simulation studies for both indirect and direct control schemes. The experimental results corroborate the simulation results quite reasonably well.
Other Identifiers: Ph.D
Appears in Collections:DOCTORAL THESES (Electrical Engg)

Files in This Item:
File Description SizeFormat 

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.