Volt/VAr Optimization of Distribution System with Integrated Distributed Generation

This paper addresses the issues of VVO (Volt/VAr Optimization) such as loss minimization, acceptable voltage profiles and optimized number of switching operations. Basic function of the DMS (Distribution Management System) is to upgrade system intelligence so that it can make dynamic decisions and control the network in realtime. Distributed generators can cause the system to operate above and below the desired limits due to their variable nature. Therefore, devices like SC (Shunt Capacitors) and OLTC (On Load Tap Changers) are used in distribution system as control devices. Main focus of this paper is to inspect effects of DG (Distributed Generation) on switching states of control devices while considering Volt/VAr standards. An optimization search algorithm is employed to search the optimal solution considering the system constraints. The GA (Genetic Algorithm) is used for the optimization process of the system and the simulation is done in MATLAB using IEEE-30 bus system with DG under 24 hour changing load profiles. By setting up constraints of distribution system’s voltage limits, capacitor bank and OLTC, losses are minimized up to 50%. Merits of the proposed optimized method are demonstrated through simulation results .The result achieved from the proposed technique has proven to be beneficial for switching optimization of control devices under variant conditions of loads and distributed generation.

therefore it is necessary to accommodate DGs with maximum profit in system. It is necessary to implement some optimization method for reducing the line losses, optimization using binary particle swarm optimization is proposed in [1][2]. Optimal reactive power dispatch using PSO (Particle Swarm Optimization) was discussed in [3].
A mixed integer linear programming approach has been implemented with embedded generation in [4][5] suggests an approach of dynamic adjustment of OLTC using DG with reactive power support for voltage profile improvement. A couple of methodologies have been explained using dynamic programming and fuzzy logic controllers where the system is divided into two parts such as substation capacitor and feeder capacitors.
Optimal dispatch schedule is calculated by using dynamic programming for both substation and feeder capacitor banks [2]. In [6] some methods are considered using NLP (Nonlinear Programming) to find optimal number of switching operations and applied on PG&E 69 bus system but without considering effects of DG.
Dynamic programming does substation control and feeder level control of devices by using fuzzy logic controller. Both parts of system coordinate but the search space size and computational efficiency lays back.
These techniques are not expedient for large power systems [7][8]. Some techniques consider the communication system layers and large number of sensing and monitoring devices, which are not part of system [9]. Some commercial solutions such as IVVC (Integrated Volt/VAr Control) are implemented in [3]. In some systems decisions are made on comparative cost of switching and line losses [7,4]. Some other techniques of NLP are used such as dynamic programming , supervised learning [5] and other sensitivity based calculations [6]. In these studies, effects such as optimal vary suitable for solving this type of problems [10][11].
An MINLP has the following form of equations: Where f is a scalar function having the nonlinear objective function, which is dependent on the following constraints search methods therefore it is best for optimization issues [13]. GA is proposed in this study rather than any other optimization technique because GA is a direct search method. It provides natural selection solution ,eliminates the weak candidates from solutions through crossover and provides a high quality solution.GA is also better than other techniques such as PSO for combinatorial problems as PSO shows poor performance in combinational scenarios. They are less inclined to getting 'caught' at local optima than gradient search methods.
GA is modeled by using genotype which are a set of potential solutions for which genetic algorithm is searching. These are some basic parameters of GA:

Initial Population
Initial population is randomly created and size of array of chromosome for 24 hour is defined and shown as: There are two main issues while using GA for MINLP:

Mutation
In this process, chromosome bits are randomly changed to make chromosome a best candidate solution.

Elitism
This is an optional parameter of algorithm in which best candidates are copied to next generation as it is.

Crossover
This is a convergence process, which tries to converge system's solution, and alters the chromosome code from one to next generation.

Fitness Function
This step of algorithm helps to evaluate quality of chromosome and evaluates all potential solutions from all proposed solutions. These are some basic parameters of GA and GA optimization problems formulated are according to [16].
Newton-Raphson method is used for formation of objective function and buses with lowest voltages are used to make objective function. To calculate objective function in algorithm we use MATPOWER commands which use Newton's method for running AC power flow [17].

DISTRIBUTION SYSTEM
The test distribution system used to apply proposed algorithm is a 30 bus distribution system. This is composed of one generator at bus 1, 2 photovoltaic DGs and four wind DGs. This system has six capacitors and one tap changer transformer for applying control scheme, shown in Fig. 4   FIG. 6. 24-HOUR SOLAR DATA [19] Wind DG 1

SIMULATION RESULTS
Wind DG 2 Wind DG 3 Wind DG 4 In phases of designing and planning, losses are important discussions in system. In actual scenarios, losses are unavoidable in system but percentages vary in different designs and conditions. Fig. 11 is illustration of fact that approximately 50% of losses have been reduced.     r  u  o  H  1  C  2  C  3  C  4  C  5  C  6  C  p  a  T   1  0  0  6  0  0  6  0  0  0  9  0  0  5   Proposed algorithm is based on an optimization technique, which is a search method but provides efficient solution for real application on distribution system.