SOME INVESTIGATIONS ON THE OPTIMAL ALLOCATION OF DISTRIBUTED GENERATION RESOURCES IN DISTRIBUTION SYSTEMS

Abstract

Competitive electricity markets, rapid growth of electricity demand, reliability issues of electric power supply, technological advancement in power generation resources, utility supporting devices and increased assimilation of telecommunication and information technology in distribution networks (smart grid) completely transform their behavior and operation. Further, integration of dispatchable or non dispatachable distributed energy resources near the load centres results in enhancement of distribution network performance. The foremost concept behind the smart grid is allocating and operating Distributed Generation (DG) resources in distribution networks tactically by considering various technical, economic and environmental issues. Therefore, the aim of the research work in this thesis is to develop methodologies for the strategic planning and operation of DGs considering various objectives. The primary contribution of the thesis is to address the development and application of meta-heuristic optimization techniques used for optimal allocation of several types of distributed energy resources and / or network reconfiguration under various loading conditions to reduce the power losses, improve network voltage profile, enhance operational cost savings and improve the loadability of the distribution networks. At the outset, a comprehensive teaching-learning based optimization (CTLBO) algorithm has been developed which is almost parameter independent and has the capability to handle mixed integer variables with fast convergence characteristics. Its implementation in standard mathematical benchmark functions demonstrate its x

superiority over many existing optimization algorithms. Thereafter, CTLBO is used for optimal allocation of DGs in different radial distribution networks, considering single as well as multi-objective formulations (based on weighted sum and ε constraints approach). The results again demonstrate the superiority of CTLBO over teaching-learning based optimization (TLBO) and quasi-oppositional teaching learning based optimization (QOTLBO) algorithms. Subsequently, a Hybrid teaching-learning based optimization (HTLBO) algorithm is developed, which can handle a large number of variables. This algorithm is again validated using single as well as multi-objective mathematical benchmark functions. It is observed that the HTLBO algorithm has either equivalent or better performance than many existing algorithms such as TLBO and QOTLBO. Further, this algorithm is used to optimally allocate several DGs for multi-objective problems in several distribution networks. It is found that the ε-constraints based approach gives better solution than weighted sum approach for multi-objective problems and results in significant reduction in losses, improvement in voltage profile and voltage stability index over TLBO and QOTLBO algorithms. Subsequently, an analytical approach based on power loss sensitivity has been developed in this thesis to allocate multiple reactive power compensation devices in different radial distribution networks to reduce power losses and emission, along with improvement in cost savings and various network performance indices. The proposed approach takes very less computational time as compared to immune algorithm (IA), bat algorithm(BA) and bacteria foraging optimization algorithm (BFOA). Subsequently, the CTLBO algorithm is used for optimal allocation of distribution static compensators (DSTATCOMs) and / or network reconfiguration to get multiple operating points (Pareto solution) of the distribution networks. This xi

Pareto solution enables the distributed network operators (DNOs) to maneuver the distribution networks to achieve maximum benefits under different operating conditions. Finally, in this thesis, the CTLBO technique is used to enhance the loadability of distribution networks by adopting a multi-objective approach to address the growing load demand without additional expenditure on the existing networks. Results demonstrate that simultaneous allocation of DGs with varing power factors and network reconfiguration yields the highest loadability enhancement in distribution networks.