EMERGENCY STATE IDENTIFICATION AND OPERATION OF POWER SYSTEMS

Abstract

One of the major responsibilities of the power system operator is to manage his power system network reliably and economically. In the past operators maintained reliable perform ance using their experience and on-the-spot assessment of network conditions. However power networks are now large and more complex, the increased number of possible operating scenarios can lead to problems beyond the operator's analytical ability. Power system operation becomes more complex under emergency conditions due to occurrence of scheduled or random events. These conditions may consist of voltage limit violations or overloading of circuits. This study is directed towards the development of efficient and reliable.methods for indentification of such emergency conditions and alleviation of emergencies in power system operation. Several algorithms, such as automatic contingency selection, evaluation of selected contingencies and corrective measures to alleviate system emergencies help the system operator in analysing the system conditions and in decision-making for efficient and proper operation of power system* Similar types of studies are also required for planning, security studies and reliability calculations of power systems. The automatic contingency selection problem is connected with developing algorithms for quickly identifying these contin gencies which may cause out-of-limit conditions, so as to reduce the number of contingencies that need to be evaluated when assessing the power system's security. In literature, many methods have been reported for ranking and selection of contingencies based on

either MW overload or voltage limit violations. It is observed that (a) entirely different algorithms are in use for contin gency selection and ranking based on MW overload and voltage limit violations, and (b) most algorithms suffer from the wellknown "MASKING EFFECT" i.e. a contingency resulting in a number of lines heavily loaded is ranked equally with one which produces a single overloaded line. Further for each selected contingency full AC load flow is to be carried out to obtain the post contingency operating state. Since, security analysis for overloads as well as voltage problems will normally be done simultaneously, it is desirable to use a single computational algorithm for both. A unified method is developed in this thesis for automatic contingency selection, ranking and outage simulation of the selected conti ngencies. The method developed is capable to achieve the above objective, i.e.. selection and ranking of contingencies simulta neously based on the overloads as well as voltage problems, efficiently and reliably without any significant additional computational burden. In the present practice of selection and ranking of contingencies, the calculation of scalar performance index (PI) is based on overloads or voltage limit violations. It is observed that in the present method of calculating PI for overloads,magnitude of overloads is not taken into account. Hence, two contingencies causing equal percentage (%) of overloads are ranked equally. Worst case is that with slightly higher % of overload in smaller -in capacity line will rank the contingency on the top of the list. The PI based on line loading and normal line load is developed in which due weightage is given to the magnitude of overloads also. In the method developed, first the local solution using Gauss-Seidel method is carried out for few selected buses in the vicinity of the outaged line. The buses for local solution are selected on the basis of the effect of the line outage on the bus voltages, i.e. level of disturbance. A mathematical model is developed for selection of buses for local solution. Thus, buses for local solution are properly selected. The solution so obtained is improved by using the Jacobian at the base case and adjusting the right hand side (RHS) vector for the outaged line admittance. This adjustment in the RHS vector is incorporated by injecting the suita ble power sources at the terminal buses to which the outaged line was connected (Source-Compensation). Performance indices (MW over load and voltage limit violations) are calculated after one itera tion of global solution. If either of the PI is greater than zero then contingency is severe and line outage simulation is needed. In that case a few additional iterations of the global solution for RHS vector adjustments, with fresh estimate of source compensation in each iteration, are carried out till final solution is obtained. An acceleration factor is used to accelerate the process of convergence. The main features of the method developed are, (a) it is free from the masking effect and misrankings, (b) contingencies causing overload as well as voltage problems are simultaneously selected and the post contin gency operating state is obtained for each selected contingency with a very little additional CPU tJ.ne, and (c) the base Jacobian is used and no re-factoriz; cion is required. All these features make this method very att active both in terms of speed and accuracy. The method developed for line outage simulation is further extended to simulate multiple line outages. Methods for line outage simulation reported in literature give approximate solution and hence full AC load flow is required to obtain the exact solution. Accuracy of the results obtained from the method developed is so high that need of full AC load flow is eliminated. After selection and evaluation of the contingencies, oprator's main concern is to take corrective measures so as to alleviate the system emergencies, if any. Emergency conditions of the operator's concern are line overloads or voltage limit violations. In literature, many methods have been reported which are normally of the form of optimal load flow, using optimization techniques. All thses methods are complicated and time consuming from computation point of view, specially for large systems. Under emergency conditions operator has to make quick decisions without caring much for the optimality of the operating point. Hence, direct and efficient methods have been developed in this thesis for alleviation of line overloads and voltage limit viola tions. The line overload problem is defined as alleviation of line overloads by generation rescheduling and load shedding. Taking the advantage of loose coupling which exists between the -Vmagnitude and angle of bus voltages, two algorithms are developed for generation rescheduling and load shedding to alleviate line overloads. First method developed for generation rescheduling and load shedding to alleviate line overloads is based on the sensitiv ity of line overloads to bus power increments. The existing methods on the similar principles called for unnecessary and exces sive load shedding. The method developed has the unique features like, - Increments in the phase angles are calculated taking into account the sensitivity of bus powers with respect to phase angles. Strategy is developed to carry out bus power increments so as to avoid unnecessary load shedding. A concept of local solution is introduced in the'vicinity of the overloaded line to enhance the computational efficiency.

The developed method is superior as a new secure operating point is obtained by significantly smaller load shedding, with little deviation from preadjustment state and at a faster speed. But, like the other existing similar methods, this method also fails to alleviate line overload in the case, where calculated bus power increments cannot be implemented at the terminal buses of the over loaded line, due to operational constraints. Hence, another direct method for generation rescheduling and load shedding to alleviate line overloads is developed, introducing the concept of local optimization. In this method, line overloads are minimized using the conjugate gradient technique of Fletcher and Reeves. Problem variables are decomposed into dependent and independent variables. A few buses in the vicinity of the overloaded line are to be processed for local optimization. Thus, poroblem is reduced to a manageable size. The method developed is capable of alleviating line overloads efficiently even for those cases where earlier methods failed. The concept of local optimization is further extended and a direct method is developed to obtain emergency adjustments to VAR control variables to alleviate the voltage limit viola tions. In this method also, a few buses in the vicinity of the buses having voltage limit violations are processed for local optimization and the same conjugate gradient technique is used for optimization. All the. methods developed above exploit efficiently, the sparcity of network matrix to reduce the computational effort and memory requirements. Various methods developed in this thesis are successully applied to a number of test systems including 6 Bus and 26 Bus models of Saskatchewan Power Corporation (SPC) transmission system, 14 Bus, 24 Bus (RTS), 30 Bus, 57 Bus and 118 Bus IEEE test systems and 64 Bus model of North-west power system of India.