POWER DISTRIBUTION SYSTEM OPERATION UNDER COLD LOAD PICKUP

Abstract

In the power deficient countries like India, the forced and scheduled power outages are a regular feature. The restoration of the distribution network following an extended outage gives rise to severe problem of overloading, leading to high line-currents with subsequent large voltage drop at the buses of the network. The reason for this condition is loss of diversity among thermostatically controlled electric loads; this condition is called as "Cold Load Pickup" (CLPU). The increased use of scheme of rotating load curtailment owing to limited power availability, high share of thermostatically controlled loads and the extensive use of uninterrupted power supplies (UPS) in total load demand, has made the condition of CLPU as a prevalent feature of distribution system operation. The CLPU problem is very relevant in the context of India, where the average peak power deficiency is nearly 10 percent of peak demand; in provinces like UP, it is more than 20 percent. The forced and planned outages are very common. The restoration of distribution network is performed frequently, or daily in some states, by the utilities. Moreover, due to the changing weather conditions throughout the year, the amount of electric heating or cooling load is very high which causes high magnitude of un-diversified loading during the restoration after the outage. These are the reasons that the CLPU condition is a most severe non-fault condition experienced by the utilities and associated aspects to this condition require proper attention of operation and design personnel. CLPU is an issue that has been recognized for long; at least for more than sixty years. The CLPU problem was first reported for its initial phases caused by very high inrush due to magnetizing currents of transformer/motor cores and motor acceleration. This caused problems with the normal operation of protection equipments. The later phase of enduring current, which is the result of loss of diversity, did not capture the attention of researchers initially; but the increase in thermostatically controlled loads such as electric space heating, and air-conditioning and refrigeration has aggravated the problem and has raised many concerns related to post-outage restoration of the network. The magnitude and the duration of CLPU current depends on various factors such as the outage duration, the types of loads, and the weather conditions. Several models for CLPU have been developed that comprise of the behavior of different types of loads and network components. A number of techniques based on: knowledge-based approaches, heuristics, and computer aided modeling techniques have been employed for mathematical modeling of the CLPU condition. The excessive post outage load during the CLPU condition imposes the operational constraints, and restricts the network restoration process. These constraints include: capacity of substation transformer with the acceptable loss of life, range for distribution voltage, thermal and loading limitations of distribution equipments for CLPU loading, and the desired network configuration. In addition to components capacity limitations, the protective relays also perform undesired tripping operation, by treating the CLPU condition as a fault. The main objective associated with CLPU condition is quick and smooth restoration of the network in the post-outage period. The commonly used techniques include: Reduction in distribution voltage, Section-wise restoration of the network, Optimal planning, Demand side management, and Adaptive relaying. The optimal planning and design of the network for handling the CLPU condition requires the utilization of different optimization techniques that work with the searches to determine the solutions. The sensitivity analysis is one of the criteria that can be used to reduce the search domain; it reduces the size of search space, thereby saving the computational time to reach an optimal solution. In the present work, the different sensitivities evaluation techniques used in power system have been studied, and a simple approach for calculating the voltage magnitude sensitivity, current magnitude sensitivity, and power loss sensitivity with respect to load variation at different buses of the network has been developed. The evaluated sensitivities have been utilized in reduction of search domain for different optimization techniques used for the solution of CLPU problems. The other applications of the evaluated sensitivities such as finding the search directions, determining location-based cost of energy, and deciding the economical locations for additional load increment in a network, have also been demonstrated. The proposed approaches have been illustrated on a 69-bus, 12.66kV radial distribution network (used in literature extensively) and a 41-bus, 33kV real distribution network of a portion of Bhopal city of India. Step-by-step restoration is the commonly adopted process by the utilities for the restoration of a network under CLPU. However, the locations and the sequence of switching the loads in the network decide the optimal network restoration process. The stepwise restoration is performed in such a way that the network is restored at the earliest with the maximum utilization of the existing system strength; this requires an optimal sequence in which the loads are restored. In the proposed scheme of stepwise 11 restoration, the steps are evaluated from the distributed load-points considered on the CLPU load profile. For each of the steps, the optimal locations for load restoration are determined resulting in the optimal switching sequence for entire network restoration. Two alternative approaches have been explored for implementation of the scheme: sensitivity-based approach and Genetic Algorithm-based approach. The first approach works in iteration; it first selects a network element amongst constraints violating elements by making use of designed indices determined using evaluated voltage and current sensitivities, reduces search domain for possible switch locations by comparison of values of the sensitivities, and finally determines the optimal load-locations to be restored at a step. By following the CLPU load-time curve, the load locations for restoration are determined for each of the steps, and are used for stepwise restoration of the complete network. The approach has been demonstrated on the real 33kV, 41-bus radial power distribution system. In the second approach, GA has been utilized for searching the optimal load-locations for network restoration in the step-by-step manner. The approach has been applied to three networks with different topologies and sizes including the same 33kV, 41-bus distribution system, and the results are compared to validate the approaches. All the techniques used for network restoration under CLPU are aimed to reduce the total restoration time. However, in case of outage due to the power shortage, these techniques do not have any plan for the outage period. Most of the adopted approaches are based on step-by-step restoration, which exploit the maximum network capacity, but have no effect on the load-demand during the restoration. Hence, to reduce the demand under CLPU and supplying some part of the network-load during the outage, the power generation devices are proposed to be incorporated in the present scheme in the distribution system. The application of Distributed Generation (DG) is not only useful in the reduction of load demand during the CLPU condition by conserving the load diversity, but also useful in supplying the power to a part of the network during the outage period. This scheme provides single-step restoration of the complete network, which is an advantage over the step-wise restoration method. In the present work, optimally sized and sited Distributed Generator units (DGs) are employed for the singlestep restoration. The placement of DGs has been made in such a way that a part of network could be supplied by the DGs during the outage or other contingencies. Two alternative approaches have been used again for implementation of the scheme: sensitivity-based approach and GA-based approach. The former approach uses voltage ui sensitivities to reduce the search domain for determination of optimal capacity and location of DGs. In the latter approach, GA has been utilized for searching the optimal capacity and locations for the installation of DGs. The effect of in-operation DGs in the network has been taken into account. The required time-duration of operation of the installed DGs has also been evaluated. Before applying the proposed scheme, the reactive power compensation has been carried out by optimal placement of capacitors; this provides reduction in power losses, reduction in reactive power demand, and therefore reduces the required capacity of DG. The complete scheme has been demonstrated on a 33-bus, 12.66kV radial distribution system. The power distribution systems are designed in such a way that the protective relays operate to sense and isolate the faults quickly to limit the extent and duration of an outage. Since, the protection schemes work with adequate sensitivity even for high impedance faults, the pickup values for protective devices are kept low. As the CLPU causes the short-term increase in currents due to loss of load diversity and feeder inrush, high sensitivity of the protective devices gives rise to the chances of undesired operation of the protection devices. The techniques used in the conventional schemes, to overcome this problem, is either disabling the relays during the CLPU condition or raising the pickup settings of the relays. Both the techniques cannot be recommended because of protection and sensitivity concerns. This work proposes two schemes for adaptive relaying, which adapt relay settings according to load behavior under CLPU. Both the schemes have been developed as digital relays and are designed using Electronic Design Automation (EDA) tools. The first scheme makes use of shifting of time-current characteristics to handle high inrush current during the initial phases of CLPU condition whereas in the second scheme, the design has been developed as a 'System on a Programmable Chip' (SOPC). The design implements the standard characteristic for the relaying with the phase-by-phase incorporation of CLPU condition. The proposed schemes have been designed and implemented on 'Field Programmable Gate Arrays' (FPGA). The functionality of the designed relays has been tested and verified using the digital design verification kits. Xilinx SPARTAN II® FPGA consisting of 200k gates and synthesis tool ISE 6.1, and Altera's Cyclone® FPGA with 5980 logic elements and synthesis tool Quartus II have been used for the first and second schemes respectively. In the SOPC based scheme, embedded Nios II soft-core available in Altera as an IP Core has been exploited to develop an efficient adaptive relay system. IV To summarize, the phenomenon of CLPU and various related issues have been studied. The solutions for the related design and operation problems are proposed. The sensitivities of various network parameters have been evaluated and utilized to reduce the search domain for the determination of the optimal solution. Optimal load-shedding and restoration sequence of the loads has been proposed for step-by-step restoration of the network. A scheme for single-step restoration of the network is developed which utilizes the DG that also enables supply of the loads during the outages. Two adaptive relays designed for CLPU condition, have the ability to change their settings according to CLPU phases. The relay-designs are modeled and simulated with the help of EDA tools, and hardware realization has been performed using FPGA. The designs have also been rigorously tested with extensive test-data sets. The work presented here is relevant in the context of countries like India, having power deficiency resulting in extended outages. ACKNOWLEDGEMENTS