Selecting the Exact Location of a Substation Using Knowledge-Based Expert Systems

Abstract

This paper reports a method for substation planning - the exact location for the substation is identified from a given set of locations. The load points are connected to the substation in the optimum route using knowledge-based expert systems to avoid any obstruction between any two load points. This also reduces the overall feeder length. The optimum branch conductors are selected using the method proposed by Tram and Wall (1988) to reduce further costs. The effectiveness of the proposed method is demonstrated with the help of an example.

It is important to have effective planning of distribution substations in order to get high standards of power reliability, security and quantity, and also to ensure maximum utilization of optimal investment. Although the optimum location of a substation reduces the system cost and system loss, sometimes it is not possible to use the exact optimal location of substation either due to geographical or social causes. The near optimal position is used in such cases. If a number of locations for the substation are provided, the location that gives minimum loss and cost is selected.

Literature survey shows that a good amount of research work has been carried out for distribution system planning. Adams and Laughton (1974) proposed the fixed and variable costs of power flow for a particular connection. Crawford and Halt (1975) suggested a procedure that utilizes the transportation model framework to find the optimal substation service boundary. This method was based on the analysis of loads and feeder on a grid basis, which can be used empirically to identify the desirable substation locations and their sizes. Wall et al. (1979) proposed a method for distribution system planning using a linear transshipment model, which, however, may not be feasible due to the assumption of linearity. Thompson and Wall (1981); Sun et al. (1982); and Fawzi and El-Sobki (1983) proposed methods for distribution system planning using branch and bound criteria that fathom some f the possible combinations prior to their solutions. El-Kady (1984) suggested a method, which explicitly includes time-dependent fixed and variable charges, as well as time dependent cost of losses. Nara et al. (1992) suggested a multi-term distribution expansion-planning model while proposing a new decomposition algorithm based on the branch exchange method to solve large-scale problems. Goswami (1997) has proposed an algorithm for the radial system using the branch exchange technique, which is highly time consuming due to the requirement of a complete power flow after each branch exchange.