Abstract— This paper presents a novel approach for single- stage multi-objective planning of electrical distribution systems using particle swarm optimization. The optimization objectives are: minimization of total installation (and operational) cost and total fault cost. The fault cost is a measure of system reliability. The trade-off analysis of these objectives is performed using Pareto-optimality principle. The particle swarm optimization (PSO) is used as the optimization tool to obtain the Pareto- approximation set solutions, where novel cost-biased particle encoding/decoding and conductor size selection algorithms have been used for simultaneous optimization of network topology and branch conductor sizes. The proposed algorithm is implemented on typical 21 and 100-node distribution systems and performance is assessed by statistical test. Index Terms—Power distribution system planning, Particle swarm optimization, Multi-objective optimization. I. INTRODUCTION The power system deregulation has opened a competitive market for power system utilities. After being unbundled, it is challenging for power companies to hold profit while keeping customers satisfied. This affects more to distribution utilities due to their direct link with end users. Thus, an efficient planning of distribution networks is essential for all utilities. The computerized distribution system planning has a rich literature spanning over last three decades. Any planning model can be broadly classified either as a static or a dynamic model. The static model is just one- step planning of a brand new network; while, in dynamic model, the network is planned for each successive planning year considering an existing system’s load growth and additional load points. The addition of new load nodes to an existing system is known as expansion planning. The dynamic planning can be performed as a single-stage (or multi-stage) planning with total load growth within a time horizon (or with yearly load growth). A review of the existing models can be found in [1]-[3]. The main planning objectives are minimization of installation cost of new facilities, minimization of system operational (maintenance and lost energy) cost, and enhancement of system reliability. In most initial works, the reliability is not considered for simplicity. In present scenario, the reliability is an important factor accounting for customer satisfaction. The reliability is considered in recent works [4-12]. It is mostly optimized by minimization of non- delivered energy due to fault [5-12]. In [4, 7, 9], outage cost due to faults, as seen by utilities, is additionally aggregated with the cost of non-delivered energy to get total fault/failure cost. Among various models, some models [4-7] have used weighted-sum approach. In [8-12], simultaneous optimization of cost and reliability is performed using Pareto -optimality principle [13], which constitutes a set of non- dominated solutions from the perspective of different objectives. In [4-10], deterministic load growth is considered; in [11-12], the load uncertainty is incorporated into the planning model. The distribution system planning problem is a typical nonlinear, non-convex, non-differentiable, constrained optimization problem with integer and continuous decision variables. The problem dimension increases with number of nodes. Normally, numerical optimization tools such as nonlinear programming (NLP) [4] and dynamic programming [5] have been used to solve this problem with lower node systems. In multi-objective problems, there are some specific disadvantages in using these analytical solution strategies, i.e., curse of dimensionality, non-differentiability, discontinuous objective space etc. Moreover, to get a set of solutions (as with Pareto-optimality principle), any numerical method requires several trial runs. In this regard, evolutionary optimization algorithms have distinct advantages, i.e., they can provide multiple solutions in a single run, can handle nonlinear, non-convex problems, and do not require any gradient information. Although the evolutionary algorithms (EAs) cannot guarantee the optimality of the solutions, they are shown to be efficient in providing suboptimal solutions for various problems [13]. All multi-objective distribution system planning models developed so far use EAs, i.e., Genetic Algorithm (GA) [6,8- 10], Tabu search [11], and Artificial Immune System [12]. There is another powerful EA successfully used in many complex problems, including power system optimization problems, is particle swarm optimization (PSO). Its basic principle rests upon the social behavior of a flock of birds, fish school etc [14]. There are many advantages of PSO over other EAs such as GA, i.e., easy implementation, effective memory use, ability to deliver solutions with less number of function evaluations, and efficient maintenance of diversity [15]. Although PSO has been successfully used for many multi-objective problems, it is yet to be evaluated on distribution system planning problem, to the best knowledge of the authors. Multi-Objective Planning of Electrical Distribution Systems using Particle Swarm Optimization S. Ganguly, * N. C. Sahoo, and D. Das Department of Electrical Engineering, Indian Institute of Technology, Kharagpur, India-721302 * Corresponding Author E-mail: ncsahoo@ee.iitkgp.ernet.in Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY KHARAGPUR. Downloaded on April 29,2010 at 05:36:48 UTC from IEEE Xplore. Restrictions apply.