IMPACT OF DISPERSED GENERATION ON VOLTAGE/REACTIVE POWER CONTROL OF DISTRIBUTION SYSTEMS

Abstract

The ever increasing demand of electric power has significantly and substantially increased the size and complexity of modern day power system. Hence to meet and fulfill the increasing demand, to satisfy the customer needs and to earn the profit out of it, alternate techniques have to be thought of by the power distribution companies. The centralized generation technique by using conventional energy sources for supplying power to the consumers need to build additional or upgrade local distribution lines. In recent days the increasing concerns over environmental issues demand the search for sustainable sources of energy. In this direction distributed generation (DG) which is both environmental friendly and also can solve the problem of serving the increasing demand of power plays a very important role. Therefore, it is a matter of significant concern to consider the integration of distributed generation into the distribution system. Distributed generation can have significant impact on voltages, load demand, power losses, system reliability and economy in distribution networks. Introduction of distributed generation causes reverse power flow and complicated voltage profiles in the distribution systems. Since voltage profiles are directly related to power quality and economy it is of primary concern to consider for this research study. This research work focuses in the area of voltage regulation, finding optimal set points of various control equipments and development of neurocontrollers for the volt/var control in the distribution system with dispersed generation. Once the steady state operating parameters of the distribution system are known through the load flow algorithm, voltage regulation methods are developed to improve the voltage profile. The volt/var control of a distribution system with DG is an optimization problem. Proper optimization methods are needed to optimize the set points of various controls in a distribution system with DG. A suitable voltage or reactive power control technique is developed and by adopting different optimization techniques, optimal set points of various controllers are found. The performance of a conventional controller degrades when the operating point and system configuration changes, whereas neuro-controller shows superior response throughout the system for different operating conditions. Neural- network based controllers are very much required for the volt/var control of a distribution system with DG. Finally, neural network based method has been proposed for the volt/ var control in distribution system with DG. This work shall gain significant importance in the direction towards the distribution system automation which is one of the viable solution for the power system scenario of recent days. The work accomplished in this thesis is given below in a nutshell. In the literature many voltage regulation methods for a distribution system using transformers, voltage regulators, capacitors, line drop compensators etc. have been discussed. Due to the introduction of DG, the conventional voltage regulators cannot handle the reverse power flow and steep voltage fluctuations, which calls for new voltage regulation equipments. The newly introduced equipments along with DG need load flow calculation with proper modeling of these equipments. Voltage regulation is very much necessary for a distribution system connected with DG. Introduction of distributed generation into distribution system makes the voltage profile complicated. Distribution system voltages if within the ANSI C84.1-1995 limits need less attention, but when they exceed the prescribed limits, proper means must be provided to control the voltage so as to bring them within the limits. In the literature many voltage regulation methods of a distribution system are discussed. In the majority of the previous works, constant power load model have been considered, but the actual load of a system is dependant on voltage magnitude. Better and accurate results can be expected by including voltage dependant loads. In this thesis, the voltage regulation by step voltage regulator (SVR) and step voltage regulator with line drop compensation (LDC) or line rise compensation (LRC) function for a radial distribution system with dispersed generation and voltage dependant loads has been addressed. For this purpose, two algorithms have been developed considering SVR and SVR with LRC/LDC for the voltage regulation of a radial distribution system with dispersed generation and voltage dependant loads. With these methods voltage profile of the system is improved and the line losses are reduced. The volt/var control of a distribution system with DG is an optimization problem. The key issues in volt/var control problem of a distribution system with DG include improvement of voltage profile and network loss reduction. Since introduction of DG makes the voltage profile complicated and affects the power quality, deviation of the voltage in each node from a certain reference voltage is considered as an indicative to quantify voltage profile improvement. DG causes reverse power flow in the network and alters the power flows thereby affecting the distribution system operation and economy. The following objective functions and constraints are considered while formulating the problem. 1. Minimization of sum of the voltage deviation, 2. Minimization of the losses Constraints: Voltage and current in Optimal volt/var control of a radial distribution network with dispersed generation is considered. Volt/var control has been achieved using step voltage regulator (SVR) with line rise compensation (LRC)/line drop compensation (LDC) function. Genetic algorithm based approach has been proposed to find the optimal set points of different volt / var control units. Genetic algorithm approach is used due to its broad applicability, ease of use and high accuracy. In this approach, initially the multi-objective optimization problem is converted into a single objective optimization problem by using weighting factors and Genetic algorithm is used to find the optimal set points of volt/var control units. Set points like resistance of the LRC/LDC, reactance of the LRC/LDC, reactive power output of the DG and tap position of the regulator have been optimized. In the previous approach, the multi-objective volt/var optimization problem has been converted into single objective optimization problem using weighting factors. The values of weighting factors depend on importance of objective functions and values of objective functions and constraints. The values of objective functions and constraints vary from network to network which causes weighting factors to vary from network to network as a result of which for every network weights are to be tuned. Many optimization techniques such as enumeration method, tabu search method etc. have been discussed in the literature to optimize set points of voltage / reactive power control devices. All these methods have some limitations such as, using the enumeration method the optimal setting for multiple control equipment is very difficult since the no. of solutions increase largely with the increase in no. of control equipments and the tabu search suffers from long search cycles. To overcome the above mentioned difficulties, in this work, a method based on Non-Dominated Sorting Genetic Algorithm (NSGA) is proposed for solving the multi-objective volt/var control problem in a radial distribution system with DG. This method does not need IV tuning of weights to find the optimal set points of volt/var control units. The proposed method is simple, easy to apply and is able to give optimal set points of volt/var controls of a distribution system equipped with DG with less number of iterations compared with the previous method. In this work, artificial neural network based method for the volt / var control in a radial power distribution system with dispersed generation (DG) is proposed. Artificial neural networks have been considered due to their ability for real time control, fast acting and adaptability to different operating conditions. The drawback of conventional controllers is that once the operating point or the system configuration changes, the performance of the controller degrades. Neuro-controllers ensure good response throughout the system, for different operating conditions compared to conventional controllers. Considering the limitations of conventional controllers and the merits of neuro-controllers, in this work, neural network based controller to control SVR with LRC/LDC function for the volt/var control in a distribution system with DG has been presented. In this work, neuro-controllers to control DG connected to a radial distribution system in hybrid control mode has been proposed. Many methods have been suggested in the literature to limit the voltage rise caused due to the introduction of DG. It can be accomplished by reducing the output of DG, by lowering the primary substation voltage, by operating DG at automatic power factor control (APFC) mode, by operating DG at automatic voltage control (AVC) mode, automatic voltage/power factor control (AVPFC) mode. AVC method is used to provide system voltage support and the APFC method can be used when the network impedance is low and the machine can export full power without allowing the terminal voltage to vary significantly. AVPFC control takes the advantages of both AVC and APFC so that, steady state voltage profile of the system improves and net energy transfer increases. The hybrid control method or AVPFC method includes automatic voltage control method (AVC) and automatic power factor control method (APFC) which is intelligently selected at appropriate instances. The increasing complexity of modern power grid highlights the need for control techniques for the effective control of DG. Even though at present PID control is still widely used in many control system applications, its ability to cope with some complex situations such as non-linearity and time-varying parameters are known to be very poor. The purpose of adopting neuro-controller is due to its non-linear control, real time control capabilities and its superiority over conventional controllers. Considering the above mentioned widely accepted advantages of hybrid control method and neuro-controllers, in this work, neural network (NN) based controlling of DG in hybrid control mode for the volt/var control in a radial distribution system with DG has been proposed. Neuro-controller has also been proposed to provide reference power factor to operate DG in APFC mode of control. The attempt made in this work to address voltage regulation, finding optimal set points of volt/var control units and neural network based volt/var control shall be a leaping step towards distribution system automation.