Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

The 5th International Conference on Electrical Engineering – Boumerdes (ICEE-2017)

October 29-31, 2017, Boumerdes, Algeria.

Optimum Dynamic Distribution Network


Reconfiguration using Minimum Spanning Tree
Algorithm
Mustafa Mosbah1*, Salem Arif1, Ridha Djamel Mohammedi2, Abdelhafid Hellal3
1
LACoSERE Laboratory, Department of Electrical Engineering, Amar Telidji University, Laghouat, Algeria
*
Engineer Operating in Algerian Distribution Electricity and Gas Company
2
Department of Electrical Engineering, University of Djelfa, Djelfa, Algeria
3
Department of Electrical Engineering, National Polytechnic School of Algiers, Algeria
mosbah.mustapha@gmail.com

Abstract—distribution networks reconfiguration is among the dynamic distribution networks reconfiguration is provided.
main tasks in distribution management system. The networks Section III briefs the minimum spanning tree based a Kruskal’s
reconfiguration is determined by opening and closing switches algorithm. Section IV then demonstrates the effectiveness of the
devices. This reconfiguration can be static or dynamic. In static solution algorithm when applied to power distribution networks.
reconfiguration, the loads are assumed to be constant, while in
Section V concludes the paper.
dynamic reconfiguration loads are assumed to vary according to
time. In this work, a dynamic reconfiguration considering the load II. PROBLEM FORMULATION
variation was proposed. A minimum spanning tree (MST) based
on Kruskal’s algorithm has been applied to find the optimal A. Objective Function
reconfiguration under multi times, with minimizing the total
The main objective of dynamic distribution networks
power loss as objective function. This algorithm was tested on
IEEE (33-nodes, 84-nodes) and validated on Algerian distribution reconfiguration is to minimize real power losses by opening and
network (116 node), using MATLAB Software. closing switches devices. To calculate the real power losses, a
power flow analysis is firstly used to compute all of the
Keywords—minimum spanning tree optimization; Kruskal’s unknown parameters (load nodes voltages, power flow through
algorithm; dynamic distribution networks reconfiguration; power lines and reactive power in generation nodes) using Newton
losses. Raphson technique. In radial distribution networks, each
receiving node is powered by a single starting node (Fig.1).
I. INTRODUCTION
Distribution networks reconfiguration is among the main
tasks in distribution management system [1]. The distribution
networks reconfiguration is determined by opening and closing
switches devices [2]. This reconfiguration can be static or
dynamic [3]. In static reconfiguration, the loads are assumed to
be constant, but in dynamic reconfiguration loads are assumed
to vary according to time [4].
The challenge of implementing dynamic distribution Figure 1. Distribution line between node p and node q
networks reconfiguration problem is high number of different
switching combinations possible in a network and the choice of +
best candidates of optimal reconfiguration [5]. To do this, the ( )= = = (1)
distribution networks reconfiguration is known as a
combinatorial constraint, discrete variable, nonlinear and non- where, ( ), is total power loss at time t, / is complex
differentiable [6]. To solve this problem, it is necessary to have voltage and their angle at node p, and are the resistance
a robust optimization technique. In literature, different and reactance of the line between nodes p and q, respectively,
techniques have been used to determine the optimal distribution and are the real and reactive power injections at ith
networks reconfiguration [7, 8, 9, 10, 11, 12, 13 and 14]. node, and is the number of nodes.
In this paper, a minimum spanning tree algorithm based on
Kruskal’s algorithm is applied to determine the optimal dynamic B. Practical Constraints
distribution networks reconfiguration. The objective function is The implementation of distribution networks
to reduce real power losses, considering security, technical and reconfiguration in a large distribution system is thus a complex
topological constraints. The organization of this paper is as combinatorial optimization problem because of the great
follows. In Section II, the mathematical formulation of the number of switches. It may take a long time using combinatorial

978-1-5386-0686-5/17/$31.00 ©2017 IEEE


optimization algorithm to reach a feasible reconfiguration plan
satisfying all constraints (security, technical and topological =[ …. , ... , ... ]
constraints) [15, 16, 17, 18 and 19].

A. Equality Constraints III. MINIMUM SPANNING TREE ALGORITHM

The equality constraints are the power balance constraints, In the mathematical field of graph theory, a spanning tree T
which include two nonlinear recursive power flow equations, for of a connected, undirected graph G is a tree composed of all the
node i, it can be formulated as: vertices and some (or all) of the lines of G. That is, every node
lies in the tree, but no cycles (or loops) are formed [10, 20]. A
spanning tree of a connected graph G can also be defined as a
⎧ ( )= ( )+ ( ) (2) maximal set of lines of G that contains no cycle, or as a minimal
set of lines that connect all the vertices. For a connected graph
⎨ ( )= ( )+ ( ) (3) with V nodes, any spanning tree will have the V-1 lines. Given a
⎩ connected, undirected graph, a spanning tree of that graph is a
subgraph, which is a tree and connects all the vertices together
where ( ( ), ( )), are the total active and reactive power of [10]. A weight is assigned to each line, whose value represents
how unfavorable it is for the considered task. Individual weights
generator at time t, respectively, ( ( ), ( )) total active and
of lines in a spanning tree decide its weight. The total sum of all
reactive power of load at time t, respectively, ( ( ), ( )) is
the weights of the lines in a particular spanning tree is its own
total active and reactive power losses at time t, respectively,
weight.
is number of generator, is number of load and is number
The Minimum Spanning Tree (MST) algorithms are further
of branches.
classified as line based MST algorithms and node based MST
B. Inequality Constraints algorithms. They are Kruskal’s algorithm or Prim’s algorithm
[14]. In this paper, the Kruskal’s algorithm has been applied to
Voltage limit includes the upper and lower voltage
determine the optimal reconfiguration under multi times.
magnitude limits at node i, which can be expressed as:
The Kruskal's algorithm is an algorithm in graph theory
( )≤ ( )≤ ( ) = 1…... (4) presented in 1956 [21, 22] that finds a minimum spanning tree
for a connected weighted graph. This means that it finds a
Power flow through any distribution feeder must comply
subgraph of the lines that form a tree. It includes every node of
with the thermal capacity of the line, which can be expressed as:
the network and the total weight of all the lines in the tree is
( )≤ ( ) = 1……. (5) minimized. To implement the Kruskal’s algorithm, two
conditions have to be satisfied. First, the weight of the lines in a
Real power generation limit includes the upper and lower real graph is arranged in an increasing order. Secondly, an empty
power generation limits of generators at node i, which can be subgraph T (called Traversal matrix) is created [10].
expressed as: Then, the algorithm considers a line, based on the order of
increasing weight. If a line (u, v) (u, v are the starting & ending
( )≤ ( )≤ ( ) = 1….. (6) nodes of a line) does not form a cycle along with the existing
lines in the subgraph (T), then the line (u, v) is added to the
C. Radiality and Connectivity Constraints subgraph (T). Thus the line (u, v) is discarded. If a line is added
The network topology should always be radial; the topology to the subgraph, the counter is incremented. It is checked that the
is radial if it satisfies the two following conditions: counter value is equal to V–1, where V is the number of nodes.
1. No load is left out service. If it is true, the procedure is stopped; otherwise, the process
2. The topology is connected and has no loops continues [22].
The following algorithm represents the necessary steps that
D. Structure of Chromosome for Reconfiguration Problem the Kruskal’s algorithm does, without going into the temporal
The number of genes in each chromosome corresponds to complexity of the UNION-FIND and sorting algorithms.
number of distribution network loops. Each gene shows a binary Sort edges in ascending order
code related to switch number that is going to be open. The Repeat
structure of a chromosome is represented as follows, where Add the minimum weight edge to the available edges.
is the control variables (BL present the number of network x If the added edge forms a cycle, repeat with the next
loops): edge.
x If there are no edges left, the MST is successfully
=[ , , …………. ] completed.
x If all vertices are reached by edges and all vertices are
The state variables vector consisting of, voltage angel in a single disjoint set, the algorithm stops, Carle MST
nodes , load nodes voltages , line power flow . Hence, is terminated.
be expressed as: End repeat.

-1-5090-4508-2/17/$31.00
Let us apply the Kruskal’s algorithm to the graph of Fig. 2a
[23]. The edges are considered in the following order: 24, 34,
57, 12, 13, 23, 45, 47, 36, 67, 25, and 46. There are 7 vertices,
so the tree will have 6 edges. 24, 34, 57 and 12 are chosen, 13 is
rejected (it forms a cycle with 12, 24 and 34) - note that we could
have chosen 13 in place of 12 - then 23 is rejected 24 and 34),
then 45, 47 is rejected (it forms a cycle with 45 and 57), 36 is
chosen, again 67 could have been chosen. There are 6 edges, so (a) Weighted 7-node graph (b) MST of 7-node graph
the algorithm ends. The constructed shaft is of weight 42, which
weight can be calculated incrementally as the edges are chosen. Figure 2. The 7-node graph and its MST
We then obtained the tree of Fig. 2b. The flowchart of the Kruskal’s algorithm for dynamic
distribution network reconfiguration is shown in Fig3. Note that
this flowchart must repeat in each one hour. The Fig. 4 present
the Kruskal’salgorithm principle.

Figure 3. Flowchart of Kruskal's algorithm Figure 4. Kruskal’s algorithm principle

IV. SIMULATIONS AND RESULTS As shown in Fig. 5, the total load profiles are extracted from
load measurements of Algerian distribution network on the same
In this section, the proposed algorithm is tested on 33-node, day. Each load shape covers 24 hours and the time interval
and 84-node radial distribution networks. Table I shows the between two points is 60 min. Fig. 6 indicates the total losses
effectiveness of proposed algorithm on system performance in variations during the day for the 116 nodes network. Fig. 7
comparison with base case (before reconfiguration). The presents the Algerian distribution network scheme while Fig. 8
min max
voltage limits are taken as V = 0.95pu and V =1.05pu. illustrates the optimal reconfiguration for radial 116 node
The Kruskal’s algorithm is programmed in MATLAB software. through Kruskal’s algorithm. Table II shows the impact of the
As it can be seen from of column 4 in Table I, the total active change in load on the network reconfiguration. From the results
power losses of all test systems are reduced significantly after of Table II, it can be seen that the reconfiguration changed
reconfiguration. The results obtained using Kruskal’s algorithm twice : one change at 10h00 and another at 02h00. In comparison
is compared with other works existing in literature, these results with other papers results, it is found that the Kruskal’s algorithm
are shown in Table I. is very efficient. Another advantage is a very minimal execution
Applying to the Algerian distribution network (116 node), time (i.e. the network node converges in 8 seconds), which
dynamic network reconfiguration is implemented using the encourages its use in real time.
demonstration system to study the impacts of time-varying loads
on the optimal reconfiguration.

-1-5090-4508-2/17/$31.00
Totale load (kW)

30000
25000
20000
15000
10000
5000
0

Figure5.Total load change day-to-day for Algerian distribution network

TABLE I. COMPARISON WITH PAPERS’ RESULTS IN LITERATURE


Distribution network Applied methods Optimal reconfiguration Real power loss (Kw)
GA [24] 9-28-33-34-36 140.60
33-node HBMO [25] 7-9-14-32-37 139.51
Proposed Kruskal’s 7-10-14-28-32 140.67
HBMO[25] 7-14-34-39-42-55-62-72 83-86-88-90-92 482.14
84-node
Proposed Kruskal’s 7-33-39-42-63-72-82-84-86-88-89-90-92 471.64

TABLE II. SIMULATION RESULTS OF DYNAMIC RECONFIGURATION FOR ALGERIAN DISTRIBUTION NETWORK
Before Reconfiguration After Reconfiguration
Hours Reconfiguration
Active loss (kW) Minimum voltage (pu) Active loss (kW) Minimum voltage (pu)
03:00 19-42-68-75-99-105-107-118-121 643.8 0.9615 589.4 0.9673
04:00 19-42-68-75-99-105-107-118-121 623.1 0.9625 570.4 0.9678
05:00 19-42-68-75-99-105-107-118-121 708.3 0.9596 648.2 0.9650
06:00 19-42-68-75-99-105-107-118-121 643.8 0.9615 589.4 0.9673
07:00 19-42-68-75-99-105-107-118-121 665 0.9608 608.7 0.9668
08:00 19-42-68-75-99-105-107-118-121 679.5 0.9604 622 0.9664
09:00 19-42-68-75-99-105-107-118-121 753.1 0.9583 689.1 0.9646
10:00 19-42-68-99-105-107-117-118-121 896.1 0.9545 839.7 0.9614
11:00 19-42-68-99-105-107-117-118-121 946.7 0.9532 887 0.9603
12:00 19-42-68-99-105-107-117-118-121 1052.4 0.9506 985.7 0.9582
13:00 19-42-68-99-105-107-117-118-121 1312 0.9448 1228.1 0.9533
14:00 19-42-68-99-105-107-117-118-121 1671.3 0.9375 1563.3 0.9472
15:00 19-42-68-99-105-107-117-118-121 1602.4 0.9388 1499 0.9483
16:00 19-42-68-99-105-107-117-118-121 1502.1 0.9408 1405.5 0.9500
17:00 19-42-68-99-105-107-117-118-121 1437.2 0.9421 1345 0.9511
18:00 19-42-68-99-105-107-117-118-121 1312 0.9447 1228.2 0.9533
19:00 19-42-68-99-105-107-117-118-121 1251.7 0.9461 1171.9 0.9543
20:00 19-42-68-99-105-107-117-118-121 1342.7 0.9441 1256.8 0.9527
21:00 19-42-68-99-105-107-117-118-121 1342.7 0.9441 1256.8 0.9527
22:00 19-42-68-99-105-107-117-118-121 1312 0.9447 1228.2 0.9533
23:00 19-42-68-99-105-107-117-118-121 1222 0.9467 1144.2 0.9549
00:00 19-42-68-99-105-107-117-118-121 1079.7 0.95 1011.3 0.9576
01:00 19-42-68-99-105-107-117-118-121 921.2 0.9538 863.2 0.9609
02:00 19-42-68-75-99-105-107-118-121 799.3 0.957 731.3 0.9636
03:00 19-42-68-75-99-105-107-118-121 643.8 0.9615 589.4 0.9673

-1-5090-4508-2/17/$31.00
loss (kW) before reconfiguration loss (kW) after reconfiguration
1800
1600
1400
1200
1000
800
600
400
200
0

Figure6.Total active loss day-to-day for Algerian distribution network

Figure 7. Algerian 116 node distribution network topology

Figure8. Optimal reconfiguration for Algerian radial distribution network determined by Kruskal’s algorithm

-1-5090-4508-2/17/$31.00
V. CONCLUSION [16] M. Mosbah, R. D. Mohammedi, S. Arif, A. Hellal, “Optimal of shunt
capacitor placement and size in algerian distribution network using
This paper presented a minimum spanning tree based a particle swarm optimization” ICMIC-2016, IEEE Proceeding, Medea,
Kruskal’s algorithm for dynamic distribution network Algeria, November 15-17, 2016.
reconfiguration. The objective function was the minimization [17] M. Mosbah, A. Hellal, R.D. Mohammedi, S Arif, “Genetic algorithms
based optimal load shedding with transient stability constraints”
of power losses while satisfying the power system constraints. CISTEM-2014, IEEE Proceeding, Tunis, Tunisia, November 3-6 2014.
The proposed algorithm was tested on 33-node, and 84-node [18] M. Mosbah et al, “Optimal sizing and placement of distributed
radial distribution test networks to illustrate the effectiveness generation in transmission systems” ICREGA-2016, Belfort, France,
of the proposed method, then validated on Algerian February 8-10, 2016.
distribution network. According to the simulation results, the [19] M. Mosbah, S Arif, R.D. Mohammedi, “Optimal of shunt capacitor
placement and size in Algeria distribution network” ICPEA-2017,
Kruskal’s algorithm is very efficient in a viewpoint of Djelfa, Algeria, September 16-17, 2017.
optimization quality and execution time, that encourages to [20] N. Deo, “Graph theory with applications to engineering and computer
use in real time. science” Prentice-Hall of India, 2004.
[21] D. J Bertsimas, “The probabilistic minimum spanning tree problem,
REFERENCES Networks” Networks Intr J, Vol. 20, pp. 245-275, May 1990.
[1] S. Civanlar, J. Grainger, H. Yin, S. S. Lee, “Distribution feeder [22] Kruskal J B, “On the shortest spanning subtree of a graph and the
reconfiguration for loss reduction” IEEE Trans on Power Delivery, traveling salesman problem” Proceedings of the American
Vol. 3, pp. 1217-1223, July 1988. Mathematical Society, Vol. 7, pp. 48-50, February 1956.
[2] D. Shirmohammadi, H. W. Hong, “Reconfiguration of electric [23] T. D Sudhakar, “Power restoration in distribution network using MST
distribution networks for resistive line loss reduction” IEEE Trans on algorithms” DOI: 10.5772/35897, 2012.
Power Delivery, Vol. 4, pp. 1492–1498,April 1989. [24] H. Ying-Yi, H. Saw-Yu, “Determination of network configuration
[3] E. L. Ramli, et al, “Ordinal optimization for dynamic network considering multiobjective in distribution systems using genetic
reconfiguration” Electr. Power Compon. Syst, Vol. 39, pp. 1845–1857, algorithms” IEEE Trans on Power Syst, Vol. 20, pp. 1062-1069, May
August 2011. 2005.
[4] Arash Asrari, et al, “Reconfiguration of smart distribution systems [25] T. Niknam, “An efficient multi-objective HBMO algorithm for
with time varying loads using parallel computing” IEEE Trans on distribution feeder reconfiguration” Exp Syst Appl, Vol. 38, pp. 2878-
Smart Grid, Vol. 7, pp. 2713-2723, November 2016. 2887, 2011.
[5] A. M. G. Solo, G. Ramakrishna, R. J. Sarfi, “A knowledge-based
approach for network radiality in distribution system reconfiguration”
IEEE Trans on Power Engineering Society General Meeting, DOI:
10.1109/PES.2006.1709616.
[6] K. Aoki, et al “New approximate optimization method for distribution
system planning” IEEE Trans on Power systems, Vol. 5, pp. 126 –
132,February 1990.
[7] B. Tomoiaga et al, “Optimal reconfiguration of power distribution
systems using a genetics” algorithm based on NSGA-II” Energies, Vol.
6, pp. 1439–55, March 2013.
[8] M. Sedighizadeh, S.Ahmadi, M.Sarvi, “An efficient hybrid big bang–
big crunch algorithm for multi-objective reconfiguration of balanced
and unbalanced distribution systems in fuzzy framework” Electr Power
Compo Syst, Vol. 41, pp. 75–99, December 2013.
[9] A. Abdelaziz, F. Mohamed, S. Mekhamer, M. Badr, “Distribution
system reconfiguration using a modified tabu search algorithm” Electr
Power Syst Res, Vol. 80, pp. 943–53, August 2010.
[10] H. Ahmadi, José R. Marti, “Minimum-loss network reconfiguration: A
minimum spanning tree problem” Sustainable Energy, Grids and
Networks, Vol. 1, pp. 1–9, March 2015.
[11] A. M. Imran, M.Kowsalya, “A new power system reconfiguration
scheme for power loss minimization and voltage profile enhancement
using fireworks algorithm” Electr Power Energy Syst, Vol. 62, pp.
312–22, November 2014.
[12] S. Assis et al, “Switch allocation problems in power distribution
systems” IEEE Trans on Power Syst, Vol. 30, pp. 246–253, January
2015.
[13] H. Falaghi, M. Haghifam, C. Singh, “Ant colony optimization-based
method for placement of sectionalizing switches in distribution
networks using a fuzzy multiobjective approach” IEEE Trans on Power
Delivery, Vol. 24, pp. 268–276, January 2009.
[14] D. Das, “A fuzzy multiobjective approach for network reconfiguration
of distribution systems” IEEE Trans on Power Delivery, Vol. 21, pp.
202–209, January 2006.
[15] R. D Mohammedi, M. Mosbah, A. Hellal, S. Arif, “An efficient BBO
algorithm for optimal allocation and sizing of shunt capacitors in radial
distribution networks” ICEE-2015,IEEE Proceeding, Boumerdes,
December 13-15, 2015.

978-1-5090-4508-2/17/$31.00

You might also like