OPTIMAL PLANNING OF DISTRIBUTED GENERATION SYSTEMS

Abstract

Currently, keen interest in the development and utilization of Distributed Generations (DGs) has been observed worldwide to control the emission of environmentally harmful substances under Kyoto Protocol agreement, to alleviate the energy deficiency in many developing countries, to limit escalation in energy costs associated with the use of conventional energy sources and to encourage the independent power producers for participation in the electricity market systems. A comprehensive study related with different aspects of DG system planning is, therefore, essential before any plan is executed. In this thesis, attempts have been made to explore the potential of integration of renewable energy resources mainly solar, wind, and small hydro based DG, in a planned manner into the Small Autonomous Power Systems (SAPSs) and distribution systems. The developed methodologies will be helpful to the developing countries for integrating DGinto the electrical systems for improvement in the system performance and for mitigation of the power deficiency. In the developing countries, SAPSs are widely used to supplyelectric loads in the rural locations. The capacity planning of SAPSs is based on deterministic criteria, such as percentage reserve margin, loss of largest unit etc. The inability of deterministic criteria to reflect the stochastic behaviour of system components has resulted in the evolution of system well-being approach. This approach incorporates a deterministic criterion in a probabilistic framework. The renewable energy sources have substantial potential to generate clean, safe, and cheap electrical energy, hence, the inclusion of renewable DGs with SAPSs has been paid significant attention during last few decades. References reveal that only Monte-Carlo Simulation (MCS) has been used for system well-being assessment of SAPSs with renewable DG. MCS requires mammoth computational time, large memory size, and massive meteorological data to simulate the random behavior of systems. Naturally, a more efficient approach is the need of the hour. Therefore, an analytical method has been developed in the present thesis for system well-being assessment and for production costing simulation of the SAPSs with renewable energy resources. The developed method is based on complete state enumeration technique to perform the intended task. This method is computationally efficient and requires much less Optimal Planning of Distributed Generation Systems quantum of meteorological data. The correlation between renewable resources and load has been nullified by dividing the study period into several time frames. To handle the uncertainties associated with renewable resources, Beta and Weibull probability density functions have been used to model solar radiation and wind speed, respectively, for each time frame. The forced outages of various generating units have also been included. The severe impact of wind power fluctuation on system stability has been considered by restricting wind power penetration to a certain percentage of system load. The developed analytical method has been applied to different configurations of a SAPS. Comparisons of obtained results with those obtained by MCS validate the use of analytical method for system well-being assessment as well as for production costing simulation of SAPSs with the renewable resources based DGs. Conventional SAPSs are generally fed by diesel generators. In such SAPSs, the type and number of candidate units to be added is determined by the least-cost generation expansion planning. The consideration of photovoltaic and wind energy systems as candidates for generation expansion requires special care for intermittent power output from the renewable DGs. Though considerable amount of work on the sizing of renewable resources based DG has been reported in the literature, the uncertainties associated with the output from renewable DGs have not been given appropriate consideration. The existing works have relied on the measured meteorological data and application of deterministic simulation approaches to optimize the system. Hence, incorporating the uncertainties associated with renewable resources, load, and generating unit availabilities, a mathematical formulation has been developed in this thesis to determine the least-cost expansion scheme for SAPSs with solar arrays and wind turbines. The costs associated with installation and operation of different units has been minimized. The energy not served and environmental benefits have also been taken into account. The constraints on Loss of Load Expectation (LOLE) and Loss of Healthy Expectation (LOHE) ofthe system, on the number ofvarious candidate units under consideration, and on the wind power penetration to load ratio have been incorporated. Various costs have been calculated on the present worth basis. The developed analytical technique has been used to calculate the expected generation from different units, energy not served, and system well-being indices. Employing a Forward Dynamic Programming based technique, the least-cost generation expansion scheme has Abstract been determined for a SAPS in four different cases. The obtained results represent the cost effectiveness and environmental benefits of renewable DG utilization. Optimal placement of DG units in the distribution systems reduces the energy losses, improves the voltage profile, releases the transmission capacity, decreases equipment stress, and defers transmission and distribution upgrades. For even a small distribution network, the selection of the best DG allocation plan among the different possibilities needs computationally arduous efforts. Sensitivity analysis is one of the criteria to select the appropriate buses for DG placement, consequently, to reduce the search space and thus, to save the computational time to attain an optimal solution. As reported in the literature, the existing techniques for sensitivity analysis can be broadly classified into two categories: Jacobian based methods and Adjoint network based methods. Jacobian based methods require the inversion of highly sparse Jacobian matrix associated with the power flow equations of Newton- Raphson method. These methods are widely used for sensitivity analysis of electrical transmission systems. However, these methods may fail or have problems when dealing with distribution networks, mainly because of unbalanced load operation, radial or almost radial topology (weakly meshed), and high R/X ratio of the cables. Adjoint network based methods determine the sensitivity through the analysis of the original network and an adjoint network. These methods require formulation of specific function for each specific sensitivity computation as well as the solution of equivalent adjoint network with the highly sparse adjoint coefficient matrix. The need of an efficient, fast, and simple algorithm, for calculating the sensitivities of active and reactive power flows, active and reactive power losses, and voltage magnitudes with respect to active or reactive power injection at different buses in the distribution system, has been realized. To cater to this requirement, a novel approach has been developed for calculating various sensitivity indices for the radial distribution system. The developed algorithm has been successfully applied to a 69-bus distribution test system. The obtained results have been compared with the simulated results and with the adjoint network approach results in terms of accuracy and computational time required. The proposed method is able to calculate various sensitivity indices for the distribution systems with single as well as multiple DG units. The developed algorithm computes various sensitivity indices using only the base case load flow solution and in much less time as compared to the simulated method. It also obviates Optimal Planning of Distributed Generation Systems the formulation of different network functions for calculating various sensitivities as required by the adjoint network based methods. The proper allocation of DG units in distribution system plays a decisive role in achieving economical, technical, and qualitative benefits. Depending on their location, DG units may improve or worsen the system performance. The reduction of real power losses, improvement in voltage profile, diminution of harmonic pollution, enhancement in reliability, and deferral of network upgrade have been reported as the primary aims for DG placement in the literature. Most of these DG allocation techniques are well suited to allocate DGs with controllable and dispatchable power output. Since these existing techniques do not incorporate the uncertainties associated with intermittent outputs from renewable resources based DGs, these techniques may not be very appropriate for allocation of such DGs. To fill this void, an Evolutionary Programming (EP) based approach has been developed in this thesis for allocating renewable DG units in the distribution network/small-autonomous power system. The objective function, considering the costs of system losses and costs of energy from different energy sources, has been minimized. The constraints on bus voltages, on line loadings, on numbers of photovoltaic arrays and wind turbines to be located, and on wind power dispatch to load ratio have been considered. To consider the adverse impact of wind power fluctuation on system stability, the wind power penetration has been limited to a specified percentage of system load either by turning off the wind turbine generators or by clipping the power output from wind turbine generators. The uncertainties associated with system load and renewable resources have been modelled by exploiting suitable probabilistic techniques and the expected values of various associated costs have been calculated. The developed sensitivity analysis algorithm has been used to select the candidate buses for DG allocation for accelerating the search procedure for the optimal solution. The developed technique has been applied to a small autonomous power system, to a 33-bus distribution test system, and also to a 69-bus distribution test system. The placement of DG units, by proposed method, results in reduced active and reactive system losses, relieved source capacity as well as improved voltage profile in the system. Nowadays, the use of renewable energy sources (mainly wind energy sources) with hydro and/or pumped storage plant is attracting substantial attention, as energystorages significantly alter the penetration and economic use of renewable IV Abstract energy sources. Optimal scheduling, technical and economic analysis, parametric analysis, relaxing the confined penetration limit on wind power, smoothing of intermittent wind power etc. have been reported as the various attained objectives for Wind-Pump Storage Hydro (WPSH) systems in the literature. For performing daily scheduling, the WPSH facilities have been considered as the participants in electricity markets in most of the reported works. In such scheduling strategies, the operational costs of WPSH facilities have not been included. Moreover, a suitable operational strategy is lacking for optimal energy management from the wind and hydro energy sources in an autonomous system. Therefore, in this work suitable mathematical formulations have been developed to identify the optimum daily operational strategies to be followed by the hybrid WPSH facilities. For each formulation, two cases have been considered depending on the systems benefited by WPSH facility. In the first case, WPSH system has been considered as participant in a day-ahead electricity market, while, in the second case, WPSH system has been assumed to serve an autonomous system. The income from the power sold to electricity market/autonomous system has been maximized, while the expenses made towards the operation of WPSH facility have been minimized. The considerations of various electrical and hydraulic operational constraints have been included in the developed formulations. Representing hourly wind speed bya Weibull probability density function, 250 wind speed scenarios have been generated through MCS to incorporate the uncertainties associated with wind speed. For each of 250 wind speed scenarios, the formulated problems have been solved by Linear Programming technique. The obtained results include lower bound, upper bound, and average values of different variables for the WPSH facility. The existing literature reveals that, for optimal sizing of the WPSH systems, the systems have been regarded as autonomous systems and the uncertainties associated with the wind power have not been given due consideration in the techniques reported. Hence, a suitable methodology is required for optimal sizing of WPSH system supplying energy to an autonomous system. The increasing deregulation trend is motivating the independent renewable energy producers for the participation in electricity markets. For the independent power producers, it will be difficult to fulfill the contract agreement with stochastic renewable energy generation without employment of suitable energy storages. Hence, a suitable technique is also needed to facilitate such generation companies in selection of optimal pump/hydro Optimal Planning of Distributed Generation Systems unit rating and reservoir capacity. Therefore, mathematical formulations for optimal sizing of the pump/hydro plant and reservoir have also been developed in this thesis for two cases as considered for daily scheduling problems. The reservoir energy capacity has been expressed as a function of the autonomy hours and of the capacity of pump storage hydro unit. The developed formulations are mixed-integer non-linear optimization problems and their solutions have been obtained by a hybrid technique based on Evolutionary Programming and Linear Programming. The various contributions made through this thesis are summarized as follows: > An analytical method has been presented for system well-being assessment and production costing simulations of the SAPSs with renewable resources based DGs, > A Forward Dynamic Programming based approach has been developed for optimal generation expansion planning of the SAPSs with solar arrays and wind turbines, > A fast, efficient, and simple algorithm has been proposed for calculating sensitivities of active and reactive power flows, active and reactive power losses, and voltage magnitudes with respect to active or reactive power injection at different buses for the radial distribution systems, > An Evolutionary Programming based approach has been developed for optimal allocation of the renewable resources based DG units in the distribution networks, > Methods have been developed to identify the optimum daily operational strategy and to determine the optimal sizing of pump storage hydro unit and reservoir for a hybridWPSH facility.