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

2017 7th International Conference on Power Systems (ICPS)

College of Engineering Pune, India. Dec 21-23, 2017

Reducing Distribution Losses using Distributed


End-user Reactive Power Support
Gajula Jyothi∗ , Charan Teja S† Student Member , IEEE , Pradeep Kumar Yemula∗∗ Member , IEEE
Department of Electrical Engineering, Indian Institute of Technology Hyderabad, Kandi, Hyderabad, India
Email: ∗ ee14mtech11028@iith.ac.in, † ee14resch01005@iith.ac.in, ∗∗ ypradeep@iith.ac.in

Abstract—Distribution losses are one of the major concern for location every-time when the load or other parameter changes
India. According to central electricity authority(CEA) executive accordingly.
summary report May 2017 transmission and distribution losses
accounts for about 22%. These losses can be reduced by im- II. R ELATED W ORK
proving power factor via local reactive power supply. In order
to address the above issue this paper proposes a method for There are many control actions available to address the
distributed reactive power support using end user equipment. problem of reactive power supply. But most of them are costly
The main contribution of this paper is to find the optimal bus or not practical. Most of the control methods are available at
location for reactive power injection considering the availability. transmission level and substation level. At the transmission
A sensitivity matrix for power loss with reactive power injection
is built for finding the optimal bus location subject to the level actions such as switching of transmission lines [2], using
availability of reactive power support at that location. In order flexible Ac transmission system (FACTS) devices like static
to demonstrate the proposed solution approach, a IEEE 33 bus synchronous compensator (STATCOM), A static VAR com-
radial distribution system is taken. Results of the case study pensator (SVC) and unified power flow controller (or UPFC)
are obtained by using MATLAB programming and the same which are electrical devices for providing fast-acting reactive
are presented. Obtained results show about 30% of reduction in
losses and improvement in the voltage profile. power compensation on high-voltage electricity transmission
Index Terms—Reactive power support, Distribution loss reduc- networks [3]–[6]. At sub station level synchronous condensers
tion, Voltage profile and shunt capacitors are used.
As technology is developing we are able to use more
I. I NTRODUCTION resources to provide control actions. We are able to make a
Reactive power in a power system is an important parameter more comprehensive reactive power control that goes all the
since its unbalance between supply and demand may cause way from the transmission system level to the end-user in
voltage instability, some times even black out of the system smart grids, [1] Sefa et al. [7] and Dominguez-Garcia et al.
[1]. In an electrical system, overheating of generators, motors [8] proposed some control techniques for providing reactive
and other loads are caused due to losses and poor voltage power support. Rogers et al. [9] suggested the idea of using
profile. Proper reactive power support will help in reduction end-user reactive-power capable devices to address low voltage
of losses and improvement of voltage profile. Some type of problems at the transmission system level. These mitigating
loads which include motor, electro-magnetic loads require controls of using end-user reactive-power-capable devices are
reactive power for their operation. For such type of loads made available via smart grid technologies. These methods can
supply of reactive power locally will improve the power factor. be used in the system to maintain a healthy voltage profile.
Traditionally reactive power support is provided at substation Reactive power resources include inverters on solar panels,
level and for industries at their local level using capacitor electric vehicles, uninterrupted power supply systems, inverter-
banks, static var compensator etc. But when the reactive power based home appliances, lighting, and many other distributed
support is provided at substation level, reactive power need to sources [10], [11]. The power buffer concept proposed in [10],
travel for long distances. This results in high consumption of [12], [13] allows the power electronics to take advantage of
reactive power by distribution lines. One of the better solution isolation, and presents a desired behavior to the grid. In the
for reactive power support is to use the end user equipment context of the above papers, the desired behavior is to provide
like inverter, Electric vehicles etc. Also this is a cost effective reactive power injection as needed. Residential-level devices
method as we can use smart devices installed at consumer can be called upon to correct voltage violations in their local
side. With coordination of consumer to supply this reactive area by using secure authenticated messaging to coordinate
power, correction of voltage can be done at distribution side the control. Rogers et al. [9], [14] just considered the reactive
in radial system. To inject the reactive power from consumers power resources of end-users at transmission system level.
into the system, we need best bus location. But best bus In these papers, the control action is centralized and can-
location depends on amount of reactive power available at bus, didate bus placement is done only by use of sensitivity
amount of load, distance from the source. As these parameters analysis. But, they do not consider the fact that end user’s
change with time, it is difficult to have a single solution every reactive power sources change whenever they are on or off.
time. So, the proposed method helps in finding the optimal Aquino-Lugo et al. [15] explained little about the distribution

978-1-5386-1789-2/17/$31.00 ©2017 IEEE 720


systems, they do not explain about the process of acquiring the objective then sensitivity matrix corresponding to voltage
the sensitivity matrix and reactive support grouping in the needs to be considered. Similarly if reducing power loss is the
distribution systems. Also, distributed energy resources of end- objective then sensitivity matrix corresponding to power loss
users have important role in modern systems, it is better to to be considered.
consider other resources such as distributed generation (DG)
and storage with reactive power resources. It is worth noting A. Sensitivity Matrix of Voltage
that there are conventional reactive power control devices such It is the ratio of change in voltage at each bus in the system
as voltage regulators that cannot be eliminated from the future to the amount of reactive power injected at a particular bus
systems, so they should be also considered. Yu et al. [16] used in the system at given time instant. Change in the voltage is
distributed voltage control to improve voltage profile, but it obtained by comparing with base case. [6].
does not consider end-user devices as well as constant voltage
S v(j,i) = ∂V(j,i) /∂Qi (1)
DGs in their work. As mentioned in above papers, the so called
central control manages systems in an integrated form and where:
improves voltage problem better than other control actions. j indicates bus number.
But, the communication burden of this is very high that makes i indicates bus number where reactive power is injected.
it unaccepted in future smart grids. Moreover, the control ∂Qi is amount of reactive power injected at ith bus.
systems will be out of use if control center is broken. On the ∂Vi,j is change in voltage at bus j, with injection of Q at
other hand, in the distributed voltage control, the system is ith bus.
decomposed into some sub-systems that are independent and Sv (j, i) is sensitivity of voltage at bus j w.r.to Qi when Q
have no communication links between them. Communication is injected at ith bus.
burden in this control is low and is suitable for future smart
grids. But, generally voltage mitigation’s quality in this control B. Sensitivity Matrix of Power Loss
action is lower than the centralized control system. It is the ratio of change in power loss in the system to the
In this method, Sensitivity matrix is calculated based on amount of reactive power injected in the system. Change in
real time and variable data. It can be calculated for both power loss is calculated by comparing with base case power
Power variation and voltage variation in the system for change loss of the system. Using this maximum reduction in power
in injected reactive power amount. Based on this sensitivity loss is calculated, which is used for finding optimal location
analysis we can find the behavior of the system for external for injection of reactive power support.
reactive power injection [6]. This method can be used for
S L(i, t) = ∂p loss(t)/∂Q (2)
distribution system, for selection of bus location using end
users as reactive support. In this method time to availability of where:
reactive power from consumers can be checked and accessible i indicates bus number.
reactive power can be used to select the bus location to inject t i indicates time in hours.
that reactive power. ∂Q is amount of reactive power injected at ith bus.
∂ploss(t) is reduction in power loss compared to base case
III. S ELECTION OF B EST B US LOCATION
power loss at time t,with injection of Q at ith bus.
In radial distribution system, there will be many number of SL(i,t) is sensitivity vector of power loss at time t w.r.to Qi
loads. In these loads some may support reactive power like when Q is injected at ith bus.
smart devices and some may not. Those smart devices may
be available at different locations which come under different C. Selection of Best Bus Location
bus. This availability may differ from bus to bus and time Reactive power injection into the system effects voltage and
to time. For each hour it is required to inject reactive power power loss in the system. Change in these two parameters or
collected from sources at one bus. Based on load variation, any one can be taken into consideration to decide best bus
amount of reactive power supplied by each bus at each hour location. Using sensitivity matrix of voltage, find bus with
varies with respect to sensitivities of buses for reactive power maximum change in voltage and select that bus as best bus
injection. Selection of bus is considered such that reactive to inject the reactive power into the system. Using sensitivity
power injection has high impact on voltage variations and matrix of power loss maximum reduction in power loss is
reduction in power loss. This optimal bus location is decided found and it is used for selection of optimal bus for injection
by formulating an objective function. of reactive power. In this paper, since the objective is to reduce
In the given system, to check the effect of reactive power the losses sensitivity matrix corresponding to power loss is
injection from loads and its effects parameters like voltage considered for best bus location. In this study maximum real
variation at the load, voltage drop in the line, losses reduction power loss reduction is considered as objective function.
in the system are considered. Sensitivity matrix provides the Fig.1 shows the flowchart of the proposed method for
information about the voltage changes at each bus because distributed end user reactive power support. Base case load
of reactive power injection at any bus along with the power flow provides real power loss using the branch data and bus
loss information. If improving the voltage profile at time t is data as the inputs. Sensitivity matrix is formed as explained

721
ϮϯϮϰϮϱ
Start
ϮϲϮϳϮϴϮϵϯϬϯϭϯϮϯϯ

Input Daily Load curve data at all buses

Running base case load flow ϭϮϯϰϱϲϳϴϵϭϬϭϭϭϮϭϯϭϰϭϱϭϲϭϳϭϴ

bus number b =1
load flow with injecting Q at bus b
ϭϵϮϬϮϭϮϮ
Compare with base case load flow

b=b+1 Finding Sensitivity


Fig. 2. Single line Diagram of IEEE 33 bus distribution network
Yes Is b<= number
t=t+1 of buses
No

Optimum bus location at t

Yes
Is time t<=24

No

Finding Schedule of Q injection

End
Fig. 3. Daily Load Curve Example

Fig. 1. Flow chart of the method

Lt is Load at time t, load may be the real power (P) or


above by comparing the base case load flow. From the obtained Reactive power (Q).
sensitivity matrix optimal bus location is found and reactive Ltb is Base load.
power is injected at that bus to observe the reduced losses. β(min,t) is Minimum percentage variation w.r.to base load
at time t.
IV. CASE STUDY:IEEE 33 BUS RADIAL β(max,t) is Maximum percentage variation w.r.to base load
DISTRIBUTION SYSTEM at time t.
The proposed method is applied to IEEE 33 bus radial Random(β(max,t) , β(min,t) ) is a random function that gen-
network shown in fig.2. It has 33 buses and 37 branches. IEEE erates value between two values β( max, t),β( min, t).
standard 33 bus radial test system data [17] is considered in Load distribution for 33 buses is done by considering the
which total generation is 3715 kW, total reactive power 2300 standard IEEE 33 bus system radial load data [17]. Fig.4 shows
kVA. The system line and load data is taken as main input the load distribution among 33 buses. Proposed method is
to the MATLAB program of proposed method to know the implemented in MATLAB programming and is tested for the
system structure and load distribution. Case Study of proposed following cases.
method has been done by programming in MATLAB. For Case 1:Optimal Bus Location with Injection of Q at Each
solving this case study base values taken as 100 MVA, 12.6 Bus Individually,at Constant Load Assumption made here is
kV. load is constant for 24 hours and Reactive sources are available
A daily load curve shown in fig.3 is obtained by using at each and every bus with same capacity assuming that
the data in [18]. Using the load curve and load variation capacity as Q=0.01 p.u i.e., 1 MVAr. Fig.5 shows the reduction
percentages a load variation calculator is used for creating in losses when reactive power injection is considered at all
some randomness in the data. So, this data is used as input buses.
load data to power flow for analysis. From the results it is observed that
 1) Maximum Reduction in active Power loss
 
Random β(max,t) , β(min,t) (P(loss,max) )=56.8 kW.
Lt = Ltb ∗ 1 + (3) 2) Maximum reduction obtained by injecting 1 MVAR at
100
bus 30
where: 3) Optimum Bus Location =30.

722
Load on the IEEE 33 bus System
600
Load on bus P(kW))

400

200

0
0 5 10 15 20 25 30 35
Bus Number

600
Load on bus Q(kVAr)

400
Fig. 6. Reduction in active power loss(kW) at each bus for case2
200

0
0 5 10 15 20 25 30 35
Bus Number

Fig. 4. load distribution

220

Ploss with Q injection


Ploss in base case

200
Reduction in Active power loss in kw

180

Fig. 7. Reduction in active power loss(kW) at each bus for case3


160

140
injecting 1 MVAR at bus loation 31
Best Bus Location is bus number 30 3) Optimum Bus Location =31.
120
Case 3 :Finding Best Combination of Bus Locations when
Q is available at Few Buses with Constant Load: Assumption
100
0 5 10 15 20
bus number where Q is injected
25 30 35 made here is load is constant for 24 hours and reactive sources
are available at few buses (assuming number buses where Q is
Fig. 5. Reduction in active power loss(kW) at each bus for case1: available is m) with same capacity. Injection of reactive power
into the system from more than one location is considered.
Those number of locations(taking as n and m<n) may vary.
From the above results it is observed that reactive power In this case many number of combinations of buses are
injection will effect power loss in the system. For the same obtained, reactive power injected from the buses of particular
amount of reactive power injected at all buses, loss varies from combination at a time. After finding and comparing the effect
bus to bus. The power loss may increase with reactive power of all combinations, best combination can be found. Taking
injection as distance from the source changes. For example at reactive sources are available at buses 6,13,23,31, Q=0.01 and
bus 22 it is observed that more power loss compared other number of locations to inject reactive =2. With these inputs
buses since it is end bus of branches. the following fig.7 shows the variation of power loss with
Case 2 :Finding Optimal Bus Location where Q is Available combination of buses.
at Few Buses only at Constant Load: Assumption made From the results it is observed that
here is load is constant for 24 hours and reactive sources 1) Maximum Reduction in active Power loss
are available at few buses with same capacity assuming that (P(loss,max) )=61.6 kW.
capacity as Q=0.01 p.u i.e., 1 MVAr. This Q injected at each 2) Maximum reduction in active power loss obtained by
bus individually. Q amount will be given as input to program. injecting 1 MVAR for combination 6,31.
Taking reactive sources are available at buses 6,13,23,31. On 3) Best combination of bus locations = 6,31.
execution of the program we get results as shown in fig.6. Case 4 :Optimal Bus Location with Injection of Q at Each
From the results it is observed that Bus Individually at Varying Load: Assumption made here is
1) Maximum Reduction in active Power loss load is varying for 24 hours and reactive sources are available
(P(loss,max) )=54.0 kW. at each and every bus with same capacity assuming that
2) Maximum reduction in active power loss obtained by capacity as Q=0.01 p.u i.e., 1 MVAr. Load variation calculator

723
TABLE I TABLE II
O PTIMAL B US L OCATION FOR 24 HOURS FOR CASE 4 O PTIMAL B US L OCATION FOR 24 HOURS FOR CASE 5

Hour Best Bus Location Reduction in Power Loss(kW) Hour Best Bus Location Reduction in Power Loss(kW)
1 30 83.31 1 31 81.66
2 30 86.94 2 31 85.37
3 30 83.16 3 31 81.46
4 30 81.13 4 31 79.32
5 30 83.49 5 31 81.77
6 30 84.80 6 31 83.20
7 30 93.45 7 31 92.27
8 30 90.94 8 31 89.59
9 30 105.99 9 31 105.44
10 30 110.95 10 31 110.59
11 30 115.75 11 31 115.68
12 30 114.66 12 31 114.62
13 30 110.16 13 31 109.75
14 30 102.89 14 31 102.16
15 30 106.91 15 31 106.26
16 30 111.03 16 31 110.72
17 30 115.56 17 31 115.46
18 31 83.31 18 31 121.42
19 31 86.94 19 31 122.23
20 30 83.16 20 31 110.70
21 30 81.13 21 31 99.60
22 30 83.49 22 31 91.47
23 30 84.80 23 31 84.39
24 30 93.45 24 31 81.78

TABLE III
is used here. On execution of above program the best bus O PTIMAL B US L OCATION FOR 24 HOURS FOR CASE 6
location at each hour for given load and given amount of Hour Best Combination of Bus Locations)
reactive power as shown in table I. 1 6,31
From the results it is observed that as load changing at each 2 6,31
3 6,31
hour, for the given reactive power support, loss in the system 4 6,31
and best bus location are varying. 5 6,31
Case 5 :Finding Optimal Bus Location when Q is Available 6 6,31
7 13,31
at Few Buses with Varying Load: Assumption made here is 8 6,31
load is varying for 24 hours and reactive sources are available 9 6,31
at few buses with same capacity assuming that capacity as 10 13,31
11 13,31
Q=0.01 p.u i.e., 1 MVAr. Taking reactive sources are available 12 13,31
at buses 6,13,23,31. On execution of above program the best 13 13,31
bus location at each hour for given load and given amount of 14 13,31
15 13,31
reactive power as shown in table II. 16 13,31
From the results it is observe that as load changing at each 17 13,31
18 13,31
hour, for the given reactive power support and given load, 19 13,31
among the bus locations 6,13,23,31, the best bus location is 20 13,31
31. 21 13,31
22 13,31
Case 6 :Finding Best Combination of Bus Locations when Q 23 6,31
is Available at Few Buses at Varying Load: Assumption made 24 6,31
here is load is varying for 24 hours and Reactive sources are
available at few buses (assuming number buses where Q is
available is m) with same capacity assuming that capacity. in table III.
Injection of reactive power into the system can be from more From the results it is observed that as load changing at each
than one location. Those number of location(taking as n and hour, for the given reactive power support, loss in the system
m<n) may vary. In this case many number of combinations of and best combination of bus locations are varying.
buses are considered, reactive power injected from the buses
with particular combination at a time. Taking reactive sources V. C ONCLUSION
are available at buses 6,13,23,31, Q=0.01 and number of Reactive Power plays an important role in maintaining
locations to inject reactive =2. On execution of above program stability of power system. Difference in supply and demand,
the obtained combinations of best bus location at each hour reactive power in the system will cause voltage fluctuations.
for given load and given amount of reactive power as shown To decrease these voltage fluctuations and power loss in the

724
system reactive power sources are required to be installed at [7] I. Sefa, N. Altin, and S. Ozdemir, “An implementation of grid interactive
distribution location. For this prosumer technology is used inverter with reactive power support capability for renewable energy
sources,” in Power Engineering, Energy and Electrical Drives (POW-
where reactive power is injected using end user equipment. ERENG), 2011 International Conference on. IEEE, 2011, pp. 1–6.
Coordinating with consumers, using consumers smart devices [8] A. D. Domı́nguez-Garcı́a, C. N. Hadjicostis, P. T. Krein, and S. T.
which have capability of providing reactive power support Cady, “Small inverter-interfaced distributed energy resources for reactive
power support,” in Applied Power Electronics Conference and Exposi-
as sources, reactive power is injected into the system. These tion (APEC), 2011 Twenty-Sixth Annual IEEE. IEEE, 2011, pp. 1616–
sources may be available at all buses or few buses in particular 1621.
system. For all the cases of availability of sources, one or [9] K. M. Rogers, R. Klump, H. Khurana, A. A. Aquino-Lugo, and T. J.
Overbye, “An authenticated control framework for distributed voltage
more buses must be chosen at each time to inject the reactive support on the smart grid,” IEEE Transactions on Smart Grid, vol. 1,
power from consumer end into the system. Using the proposed no. 1, pp. 40–47, 2010.
method best bus location is chosen from the buses where [10] D. L. Logue and P. T. Krein, “Utility distributed reactive power control
using correlation techniques,” in Applied Power Electronics Conference
reactive sources are available so that power loss in the system and Exposition, 2001. APEC 2001. Sixteenth Annual IEEE, vol. 2.
reduces. Even if load varies on the system, best bus location for IEEE, 2001, pp. 1294–1300.
scheduling of reactive power from consumers into the system [11] M. Begović, A. Pregelj, A. Rohatgi, and D. Novosel, “Impact of
renewable distributed generation on power systems,” in Proceedings of
can be find out. Further improvements in the method can be the 34th Hawaii international conference on system sciences, 2001, pp.
done by considering variable reactive sources and variable 2001–2008.
amount of reactive power. Mathematical modeling of reactive [12] D. Logue and P. T. Krein, “The power buffer concept for utility load
decoupling,” in Power Electronics Specialists Conference, 2000. PESC
sources can be considered for more practical insight of the 00. 2000 IEEE 31st Annual, vol. 2. IEEE, 2000, pp. 973–978.
method. [13] W. W. Weaver and P. T. Krein, “Mitigation of power system collapse
through active dynamic buffers,” in Power Electronics Specialists Con-
R EFERENCES ference, 2004. PESC 04. 2004 IEEE 35th Annual, vol. 2. IEEE, 2004,
pp. 1080–1084.
[1] C. Recio, “The utilization of power converters in consumer products for [14] K. M. Rogers, R. Klump, H. Khurana, and T. J. Overbye, “Smart-
distributed reactive power support,” 2012. grid-enabled load and distributed generation as a reactive resource,” in
[2] K. W. Hedman, R. P. O’Neill, E. B. Fisher, and S. S. Oren, “Optimal Innovative Smart Grid Technologies (ISGT), 2010. IEEE, 2010, pp.
transmission switching with contingency analysis,” IEEE Transactions 1–8.
on Power Systems, vol. 24, no. 3, pp. 1577–1586, 2009. [15] A. A. Aquino-Lugo, R. Klump, and T. J. Overbye, “A control framework
[3] A. Mazi, B. Wollenberg, and M. Hesse, “Corrective control of power for the smart grid for voltage support using agent-based technologies,”
system flows by line and bus-bar switching,” IEEE Transactions on IEEE Transactions on Smart Grid, vol. 2, no. 1, pp. 173–180, 2011.
Power Systems, vol. 1, no. 3, pp. 258–264, 1986. [16] L. Yu, D. Czarkowski, and F. De León, “Optimal distributed voltage
[4] K. Ma and J. Mutale, “Risk and reliability worth assessment of power regulation for secondary networks with dgs,” IEEE Transactions on
systems under corrective control,” in North American Power Symposium Smart Grid, vol. 3, no. 2, pp. 959–967, 2012.
(NAPS), 2009. IEEE, 2009, pp. 1–6. [17] R. D. Zimmerman, C. E. Murillo-Sánchez, and R. J. Thomas, “Mat-
[5] W. Shao and V. Vittal, “Lp-based opf for corrective facts control to power: Steady-state operations, planning, and analysis tools for power
relieve overloads and voltage violations,” IEEE Transactions on Power systems research and education,” IEEE Transactions on power systems,
Systems, vol. 21, no. 4, pp. 1832–1839, 2006. vol. 26, no. 1, pp. 12–19, 2011.
[6] A. Abessi, V. Vahidinasab, and M. S. Ghazizadeh, “Centralized support [18] S. Charan Teja and Y. Pradeep Kumar, “Energy management of grid
distributed voltage control by using end-users as reactive power support,” connected rooftop solar system with battery storage,” in Innovative
IEEE Transactions on Smart Grid, vol. 7, no. 1, pp. 178–188, 2016. Smart Grid Technologies-Asia (ISGT-Asia), 2016 IEEE. IEEE, 2016,
pp. 1195–1200.

725

You might also like