A Multi objective Optimization of DG using Genetic Algorithm in Distribution systems

Mrs. J.Rajalakshmi Dr.S.Durairaj

Department of EEE Principal, Dhanalakshmi srinivasan college of Engineering

Fatima michael college of Engineering and Technolgy

Abstract: Due to the increased demand in the electrical systems, Distributed Generation (DG) is an important approach to organize performance, operation and control of the distribution system. There are several methods available to perform the operation and control of DG. But still those methods are not able to provide technical and economical benefits. In this proposed system, Genetic Algorithm based Optimal placement of DG has been calculated . In this proposed work, the active and reactive power loss indices, voltage deviation index, reliability index and shift factor indices are considered. A novel MOF has been developed to find optimal sizing and placement of DGs using genetic algorithm approaach. The proposed method is tested on standard 33 bus radial distribution system. The results shows that system losses, energy not supplied, system MVA intakes are reduced, voltage profile are improved for the case with-DGs in the distribution system..

Keywords: Distributed Generation, Genetic Algorithm.

I. INTRODUCTION

Our world concerns about the environment, combined with the progress of technologies to connect renewable energy sources to the grid and deregulation of electric power market have diverted the attention of distribution system planners towards grid connected distributed generation (DG). Distributed Generation (DG) 1 by definition is a small power source (roughly 10 MW or less) connected at the substation, distribution feeder or at the consumer terminals Distributed Generation placement and sizing is an important issue which requires special attention of both planners and system operators.

DG installation at non- optimal places can lead to increase in system losses which imply increase in costs and hence having a negative impact opposite to the desired. The selection of the best location and size in large and complex system is a combinatorial optimization problem. Studies have indicated that inappropriate selection of location and size of DG, may lead to greater system losses than

the losses without DG 2, 3. Researchers have employed various methods in addressing the placement problems. It has been observed that among all the methods reviewed so far analytical method is found to be the most accurate and more practical technique for placement of DG. However, obtaining a t r u l y optimal solution has presented a challenge as some computational methods do not yield global solution as many local solutions exist.

Optimization is a process by which to find out the best solution from set of available parameters. In DG allocation problem, DG locations and sizes need to be optimize in such a way that it give most economical, efficient, technically sound distribution system. In general distribution systems have many nodes and it is very difficult to find out the optimal DG location and size basically. There are numerous optimization techniques used in the literature.

Among the different solution strategies deterministic algorithm such as dynamic programming, mixed integer programming, nonlinear programming have been used. An analytical approach to determine the optimal location of DG is presented 4. Recently heuristic algorithms, such as fuzzy mathematical programming, a genetic algorithm (GA) 5, a Tabu search (TS), evolutionary programming, partial swarm optimization are used. The placement of multiple DGs with artificial intelligence-based optimization methods 6,.The analytical methods for allocation of different types of multi-DGs in primary distribution networks are investigated in 7. In 8, a GA based method has been proposed to find the optimal placement of DG in the compensated distribution network for restoration the system caused by cold load pick up (CLPU) condition and to conserve load diversity for reduction in line losses The MOF formulation for sizing and siting of DG into an existing distribution system is proposed in 9.

Most of the optimal placement techniques to allocate multiple DGs use heuristic approach only, and do not take the advantage of analytical approach. The analytical approaches may not be appropriate for optimal placements of multiple DGs alone. Even though meta-heuristics algorithms presented to obtain results however, convergence is not always guaranteed. A GA approach has been proposed in the present work for optimal placement of multiple DGs of multiple types.

This paper is organised with seven different sections as follows: in Section 1, introduction of the optimal installation of DGs using various techniques are described. Section 2 gives the different type of DG models. The modelling of different type of load models is described in Section 3. Section 4 includes the detailed description of the proposed methodology. The soft computing technique used in this work are explained in Section 5. The results and discussions are reported in Section 6. Finally, conclusions are drawn in Section 7.

II TYPES OF DISTRIBUTED GENERATION:

The electric power generation units placed near to the load and connected directly to the distribution networks is defined as DG. On the basis of the power delivering capability, the classification of DG majorly of four types 10, 11 based on their real and reactive power delivering capabilities are as 12

(i) DG Type 1: The real and reactive power delivered at unity power factor by DG to the system, is known as type-1 DG such as wave energy source wind power source, and tidal energy source etc.

(ii) DG Type 2: Only real power delivered by DG to the system is known as a type-2 DG at 0.8 to 0.99 leading power factors, such as fuel cell, photovoltaic system, and solar power plant etc.

(iii) DG Type 3: Only reactive power support provides by DG to the system is known as type-3 DG at zero power factor, such as synchronous motor in over excited mode, phase modifier circuit or synchronous condenser etc.

(iv) DG Type 4: Provides real power support to the system and absorbs reactive power from the system by DG is known as a type-4 DG at 0.8 to 0.99 lagging power factor, such as DFIG etc.

III ISSUES WITH DGS DURING OPERATION AND PLANNING

In general Distribution planning accounts that the investigation of future power delivery needs and options, with an objective of creating a accurate action to the networks required to attain agreeable levels of service at a minimum overall cost . Again, on the other hand, installing DG in the distribution networks can also increase the complexity of networks planning. With the existing protective devices and schemes DG must be satisfactorily introduced and facilitated. With higher penetration levels of DG may cause conventional power flows to alter (reverse direction), since with generation from DG units, power may be injected at any point on the feeder. New planning systems should guarantee that feeders can suit changes in load configuration. Limitations and problems must be solved before picking DG as a planning alternative.

3.1 DG Sitting

For location of DG units there are no limitations in the distribution system, as there are no geological limitations as on account of substations. In the event that the choice of DG location is focused around a few electrical factors, for example:

• Providing the required extra load demand

• Reducing networks losses

• Enhancing networks voltage profile and expanding substations capacities

Moreover, DG units must be put on feeders that don’t affect the existing protective device coordination and ratings.

3.2 DG Sizing

To enhance the networks voltage profile and reduce power losses, it is sufficient to utilize DG units of aggregate capacity in the reach of 10-20% of the aggregate feeder demand . The DG unit size can affect networks protection coordination schemes and devices as it affects the value of the short circuit current during fault. Hence, as the DG size increases, the protection devices, fuses, re-closers and relays settings have to be readjusted and/or overhauled .

The impact of the DG unit on distribution network may be positive or negative, depending upon the size and operating condition of the DG unit. Usually, the DG unit size should be consumable within the distribution substation boundary. Further increase of the DG size can cause reverse flow of power through DG buses and then high system losses.

IV PROPOSED METHODOLOGY

In this distribution system planning problem optimization is done by genetic algorithm to determine an economical yet reliable network with better technical features, such as lower power loss, better node voltage profile, and better maximizing DG power in order to reduce the stress of excessive active and reactive power demand from transmission networks

4.1 Problem formulation

The MOF-based problem formulation are introduced for optimal placement and sizing of multi-DG uses GA techniques in different test system are explained as

?Y_MOF=X?_1 PLI+X_2 QLI+X_3 VDI+X_4 RI+X_5 SFI

(1)

where X1, X2, X3, X4 and X5 are the indices weight factors and PLI, QLI, VDI, RI and SFI are indexes of active power loss, reactive power loss, voltage deviation, reliability and sensitivity or shift factor respectively of the test system.. The installation of three-DG in the test systems are considered for the multi-DG approach. The detailed concepts for selecting the weight factor of the indices are given in 13-15. All these weight factors are decided on the basis of the individual impacts and importance of the particular index while installing the DG. The main aim is to minimize the overall power losses of the system, so the active PLI gets the highest weight of 0.38, after that QLI gets second highest weight of 0.25. The VDI gets a weight of 0.15, to maintain the power quality and voltage profile of the system. The RI indicates the reliability of the system hence it gets a weight of 0.12. The SFI decides the change in power at other buses due to particular injection of the DG size at the bus; hence it gets a weight of 0.10.

The formulae for system parameter calculations are taken from 16.

Active power losses (PL)

PL =?_(K=1)^(N_br)??|I_k |^2 X R_k ?

(2)

Reactive power losses (QL)

QL=?_(K=1)^(N_br)??|I_br |^2 X X_K ?

(3)

where Ik is the branch current, R_k is the resistance, X_k is the reactance of kth branch or line.

4.2 Evaluation of system performance indices

The active power loss index (PLI) can be expressed by the following equation

PLI=?PL?_DG/?PL?_(NO-DG)

(4)

The reactive power loss index (QLI) given by the following

equation

QLI=?QL?_DG/?QL?_(NO-DG)

(5)

This voltage deviation index (VDI) is given as

VDI=?(n@max@j=2)((V_reff-V_DGj)/V_reff )

(6)

where n is the total no. of buses. The V_reff and V_DGj are the reference voltage and the system voltage value with DG, respectively.

The installation of DG with particular size will inject some power say at bus and due to this injection the change in power is ?x, then SFI can be given as

SFI=?(j=1@j?slack and@pv bus)^?(n-1@max) |(?x_j)/x_(inj,j) |

(7)

The RI of the system is given by the following equation

RI=?ENS?_DG/?ENS?_NODG

(8)

where?ENS?_DGis energy not supplied (ENS) with-DG and ENSNo?DG is ENS without-DG condition.

V GENETIC ALGORITHM

5.1 Genetic algorithm

Genetic Algorithms is a ‘one size fits all’ solution to problem solving involving search. GA’s can be applied to most of the problems to optimize and a good choice of representation and interpretation unlike other conventional search alternatives, with a good function specification. A genetic algorithm is a heuristically guided random search technique that concurrently evaluates thousands of postulated solutions. The GA has the capacity to develop an initial population associated with probable optimum solutions. After that, GA recombines these individuals in such a way to steer their particular search towards the most favorable results in search space The coding and manipulation of search data is based upon the operation of genetic DNA and the selection process is derived from Darwin’s survival of the fittest’. Search data are usually coded as binary strings called chromosomes, which collectively form populations. Evaluation is carried out over the whole population and involves the application of, often complex ‘fitness’ functions to the string of values (genes) within each chromosome