Dynamic Voltage Regulator Operation with Demand

Side Management for Voltage Control

A.L.M. Mufaris1 and J. Baba2 S. Yoshizawa and Y. Hayashi
1
Department of Electrical Engineering and Information Department of Electrical Engineering and Bioscience
Systems Waseda University
2
Graduate School of Frontier Science Tokyo, Japan
The University of Tokyo
Tokyo, Japan

Abstract—Widespread interconnection of Photovoltaic (PV) Previously, the authors have proposed local control of
systems in the distribution system may create voltage rise customer controllable load (HPWH) for voltage rise
problem. In this paper, a decentralized voltage control method mitigation [8]. However it is found that the use of HPWHs for
by use of voltage regulators (VRs) and demand side controlling voltage in entire distribution system is not always
management using consumer controllable load has been possible because the number of HPWHs installed in
proposed for voltage rise compensation. The proposed method distribution system is limited and satisfying hot water demand
includes a novel determination method to find dynamic line of consumer should be considered as first priority than
drop compensation (LDC) parameters for a VR that has reverse controlling the voltage. Hence use of voltage regulators (VRs)
power flow using the measured line current through VR and an
along with demand side management for voltage control
optimal method that determines a dead band for a VR in order
to lessens the number of tap operations while minimizing voltage
would ease dependency on the consumer load control. In VR,
deviation and violation. The obtained results show that voltage tap position is adjusted based on line drop compensation
violation is extensively mitigated by the proposed method and no (LDC) technique to maintain voltage within acceptable level.
significant increment in tap operations of VR is encountered. However, conventional LDC based VRs cannot provide
proper voltage with introduction of DGs in a distribution
Index Terms—heat pump water heater (HPWH), LDC, voltage system because voltage regulating point is no longer at the
rise, PV systems same place. In order to alleviate this problem, uniform load
based LDC method [1], optimal LDC control method [2] and
I. INTRODUCTION
MLDC (Multiple Line Drop Compensation) method [3] have
In Japan, large amount of Photovoltaic (PV) systems will been proposed.
be installed in the distribution system. However, the
widespread use of PV systems may create voltage rise However, LDC parameter determination during reverse
problem in future distribution system. Several voltage control power flow at a VR is not discussed and voltage regulation at
schemes have been proposed in the literature to mitigate the low level consumer has not been considered in detail for entire
impacts of distributed generations (DGs) integration. The use residential distribution system. In this paper, we propose a
of voltage regulators for voltage control has been discussed in determination method to find dynamic voltage regulating
many places [1]-[3] in combination with other voltage control point (voltage reference and equivalent line drop
devices such as shunt capacitor (Sh.C) and static var compensation impedance) based on the measured line current
compensator (SVC) [4]-[5]. Besides, utilization of DGs to at the VR including reverse current and an optimal method to
provide reactive power support or active power suppression find dead band that lessens the number of tap operations.
for voltage regulation through the proper control of the power Moreover, a real time decentralized voltage control method
electronics is also discussed. PV system output power is has been proposed by use of VRs and HPWHs.
regulated by power conditioning system (PCS) of the PV II. CONVENTIONAL VOLTAGE CONTROL METHOD
system which has grid over-voltage protection function to
limit voltage rise [6]. In recent years, demand side Voltage control with LDC regulation is shown in Fig. 1.
management becomes popular among researchers for voltage The concept of LDC method is that sending voltage of
control [7]-[8]. Hence consumer load is considered to be regulator ( ) is adjusted to keep the voltage at the regulating
controlled in accordance with DG output. Consumer loads point constant (i.e ) in proportion to current through VR
such as water heaters, air conditioners and electric vehicles are ( ) [2]. In order to attain proper voltage along the
considered as controllable loads due to its energy storage distribution line as in Fig.1, LDC control parameters (
facilities and less effect to consumer comfort by use. and ) for voltage regulators should be determined from
load characteristics so that constant voltage at the regulating

This research was supported by the Japan Science and Technology

Agency (JST), CREST.
point can be attained regardless of the load current. The LDC parameters ( and ) for VRs are
approximated voltage at the regulating point during maximum calculated based on the measured line current at VR and they
and minimum load can be calculated from (1) and (2), where are updated every 15 minutes interval. Based on the dynamic
and are maximum and minimum sending end LDC parameters for measured VR’s current, tap operation of
voltage respectively, and are maximum and VRs are adjusted in order to keep voltage within safe limit by
minimum line current at VR without DGs’ integration. eradicating voltage violation and minimizing the voltage
deviation from nominal value. Meanwhile, the operation of
(1) HPWH is initiated only if the node voltage at a pole
(2) transformer (low voltage distribution system as in Fig.3)
reaches maximum threshold voltage ( ) which means there
IVR is a possibility that voltage at a consumer node would become
VSE VREF larger than the voltage at a pole transformer during when
VR
ZREF Regulating DG’s power output becomes considerably large. Simple flow
LDC Control
Point diagram of proposed method is given in Fig. 4, where k
represents time intervals in a day (k = 1,2,3 …., 96).
Figure 1. Conventional LDC based voltage control method

III. PROPOSED VOLTAGE CONTROL METHOD VR-1 VR - 2

Medium voltage
distribution system
Interconnection of DGs in a distribution system changes
the system voltage profile and power flow direction. When Pole
considerable amount of DGs are introduced, current flows in s Transformer
reverse direction not only at some of distribution line section, Low voltage
n n +1 n+m
specially at line end, but also at VR. In order to understand distribution system

this fact, consider a single line diagram of a distribution line

HPWH
shown in Fig. 2 that has two VRs. During a certain time Consumer PV
interval, lets assume DG’s power output becomes large which Load
makes reverse current at end of line section. Thus current
through the VR close to end of line befalls in two regions; Figure 3. Medium and low voltage distribution system
or 0. Hence minimum loading
current flowing through a VR defined in the conventional Determine optimal dead band
method to determine the LDC parameters are no longer
correct. This means, tap operation of a VR with conventional
Measure and find the average line current at VR over the
LDC parameters for the cases presented in Fig. 2 may not be last 15 minutes interval (IVR(k))
proper to retain voltage within acceptable range. In Japan,
distribution system voltage should be 101± 6 V at end Update VR’s LDC control parameters
customer. VREF (K) and ZREF(K)
IVR-1 0<=IVR-2 < IMIN Reverse power
flow region Adjust the tap of VRs

VR1 VR2 Check all node’s voltage at pole transformer

IVR-2 < 0 Yes Vs,t < VTHR
No
VR1 VR2
Initiate HPWH’s operation
Figure 2. Distribution line with reverse power flow
Figure 4. Proposed voltage control method
In order to overcome the aforementioned problems in
determining LDC parameters and to eradicate voltage A. Dynamic LDC Parameter Determination
deviation from acceptable range, a real time decentralized Since proposed LDC parameter determination method
voltage control method has been proposed by use of VRs and depends on measured current through VR, average current at
HPWHs. VRs are considered as main control agent whereas VR for last 15 minutes ( ) has to be determined for available
HPWHs are regarded as supportive agent operating only VRs in the distribution line. The average current can be
during the emergency condition. The proposed voltage control determined by (3) from the measured current at a VR
method also includes determination of optimal dead band for ( ) for a certain duration, where is total number of
VRs and dynamic LDC parameters, and demand side VRs and it is considered that time interval between the
management by consumer controllable load. measurements is one minute.
The procedure of proposed method is as follows, initially
optimal dead band for VRs are determined with assumption
that load and PV output patterns are known. Then dynamic
 In the proposed method, LDC parameters are determined (9)
in three different ways given by (4) based on averaged
measured line current which even goes below minimum load
(10)
current ( ).
B. Optimal Dead Band Determination
The selection of a dead band ( ) is one of important
(4) parameter in order to reduce total number of tap operations of
VRs and to eradicate voltage violation. Selection of large dead
band lessens tap operation while increases the possibility of
1) Case I: LDC control parameters for this case can be having voltage deviation from acceptable range leads to
determined by solving (1) and (2) and are given by (5-6) voltage violation, whereas small dead band provides higher
which are similar to the conventional method are constant tap operations with proper voltage. Hence optimal dead band
throughout the day in conventional method. should be determined so as to reduce tap operation and to keep
proper voltage at consumer nodes. Following objective
(5) function, , given by (11), (12) is considered to find optimal
dead band. In the objective function, minimization of total tap
(6) operation of all installed VRs, total voltage violation including
voltage rise and drop of all consumer nodes (low voltage
2) Case II: In this case, LDC parameters are determined nodes) and voltage deviation from nominal voltage are
by (5) and (6) in which is replaced by and all considered. In (12), a constant penalty factor (α) has been
other parameters are considered to be same. The detail considered in order to provide major importance on
explanation of this case can be described from Fig. 5, the new maintaining the voltage profile in the objective function, thus
LDC parameters are updated only if line current at a VR voltage violation is increased by a factor (α=2) for voltage rise
becomes less than . Hence LDC parameters for this case and drop, whereas it is zero for proper voltage.
can be determined from (7-8).
(7)

(8)

Line Voltage
Minimum Load with
Maximum Load DGs (Case III)
VMAX
(12)
VMIN Minimum Load with
VREF DGs (Case II) State variables,
VREF
VREF (13)
VMINline Minimum Load (Case I) Search region,

ZREF ZREF ZREF

[%] (0.1% stride) (14)
Line Length
(Equivalent line impedance) Where, is tap adjustment of a VR at time (in
Figure 5. Dynamic LDC parameter determination
minutes), (=27) is maximum number of tap operations of
a VR in a day, is voltage violation of a node at
3) Case III: Though the sending voltage at a VR has its time , is voltage at a consumer node, N is total number of
maximum and minimum values, voltage at a node in the low voltage nodes, is total duration in a day,
medium voltage distribution line also has allowable its (=1.02p.u.) and (=0.91p.u.) are maximum and
maximum ( ) and minimum ( ) level minimum allowable voltage at consumer node and
(considering that there is a voltage drop from pole transformer (=1p.u.) is nominal voltage. Besides, all voltage values in
to consumer end). When is negative, voltage at the the objective function are considered in p.u and its base
downstream nodes in the medium voltage distribution line voltage is 105V.
should be larger than voltage at VR. Hence allowing the
C. HPWH operation
minimum voltage at VR equal to minimum line voltage
( ) doesn’t violate voltage level in the distribution It is already mentioned that the introduction of demand
system. In order to understand this case, consider the Fig. 5 in side management at consumer premises through consumer
which voltage at a VR is allowed to change from to load control is taken into consideration as supportive control
and new LDC parameters are determined depend on in the proposed method since there is a possibility that voltage
the amount of by (9) and (10). violation can’t be eradicated completely at low voltage
distribution level only with VR operation due to large by LRT (load ratio transformer) and SVR (step voltage
penetration of DGs in future residential distribution system. regulator), where dead band of both VRs are set to 1% in
HPWH is proposed as a customer controllable load due to conventional voltage control method. Almost first half of the
large energy consumption, flexibility in operating time and nodes have 6750/105V tap changing pole transformer and rest
considerable installation in a country like Japan where hot having 6600/105V. PV systems are installed uniformly in the
water consumption per consumer is 30% of the total energy low voltage distribution systems. It is assumed that all
consumption. Though HPWH is usually operated during night, household consumers (total - 479) have HPWH and PV.
it can be operated during daytime without having customer Initially, load flow analysis has been conducted without
inconvenience. A detail HPWH model and its explanation on considering VR’s operation and PV systems installation with
the selection of all model parameters are discussed in [8]. intention of determining the maximum and minimum current
Installed HPWHs under a node as illustrated in Fig. 6 obtains through VRs only with loads so that proposed LDC
the secondary voltage at a pole transformer and its operation parameters can be calculated. Table I shows the specifications
begins based on the voltage or available hot water level. The of distribution system model considered in this paper. For the
detail procedures of the proposed HPWH operation are given simplicity, all consumers’ hot water demand pattern is
as follows in simple steps, where hot water demand patterns of considered to be same and their hot water demand is 12kWh
household consumers are assumed. in terms of thermal energy.
Step 1: Obtains hot water demand for next day at the Check hot
beginning of a particular day water level

Step 2: Measure pole transfomer node voltage ( ), hot

water level in the HPWH’s tank and created hot water for
every minute Less than minimum level Sufficient

Step 3: HPWH operation is initiated when becomes

PHPWH = PR Check Vs,t
equal to or hot water level reaches its minimum
level
Step 4: HPWH operation is continued until hot water
VTHR =< Vs,t < VTHR + 0.2 VTHR + 0.2<=Vs,t < VTHR + 0.5 VTHR + 0.5 <=Vs,t
demand is completed
Pole PHPWH = 0.68PR PHPWH = 0.74PR PHPWH = 0.77PR
Transformer
s Figure 7. Proposed HPWH operation

Vs,t n n +1 n+m

LRT SVR

Consumer PV Tap 6750 / 105V Tap 6600 / 105V

HPWH Load
Figure 8. Distribution system model
Figure 6. Demand side management
TABLE I. DISTRIBUTION SYSTEM MODEL SPECIFICATIONS
In step 3, voltage at the pole transformer is considered for
HPWH operation rather considering voltage at its own node Maximum current ( ) 0.439 p.u. 192.1A
( ) because collective HPWH operation would influence Minimum current ( ) 0.171 p.u. 74.8A
much on voltage regulation compared that of individual
HPWH operation. Since consumer comfort is considered as 1.015 p.u. 6700V
major importance in proposed method, HPWH is operated so 1.001 p.u. 6610V
that sufficient hot water is available all the time. The detail
1.015 p.u. 6700V
explanation of step 3 on how HPWH power consumption is
controlled is shown in a flow diagram in Fig. 7 where optimal 0.981 p.u. 6476V
power consumption obtained in [8] for HPWH ( ), Total load capacity 3315 kVA
minimizing voltage violation and maximizing HPWH
operating duration, has been used for voltage control. HPWH Total PV capacity 1622 kVA
is operated with rated power consumption ( = 1kW) for hot
water level less than minimum (20% of the capacity (20kWh,
thermal energy)), whereas power consumption is decided Proposed decentralized voltage control method has been
based on for sufficient hot water level, where is conducted in MATLAB in which simulation interval is
106.0V. considered as one minute. Fig. 9 and 10 show the voltage
profile of all consumer nodes throughout the whole day for
IV. SIMULATION STUDIES AND RESULTS conventional and proposed method respectively for PV output
Distribution system model (6.6kV) considered for this on a sunny day. It is obvious that voltage violation is very
study is illustrated in Fig.8 in which line voltage is controlled significant at most of consumer nodes during daytime with
conventional method as PV output becomes large. Since
HPWH operation for conventional method is assumed to be that makes slope of voltage increment along the line to be
operated on night time, it can be seen that there is a sudden little. Hence voltage regulating point for case III is placed
voltage drop from 18:00 to 22:00 due to HPWH rated power outwards from the regulating points of case I and II which can
consumption. On the other hand, voltage violation is be explained by Fig. 5, making smaller and larger
extensively eradicated by proposed method and most of for case III. Afterwards, LDC parameters again determined
consumers’ node voltage is within acceptable range. using case II and then becomes constant for rest of the period
during which is positive and higher than .
1.06
1.01
1.04 Vref LRT
Vref SVR
1.02 1.005
Node Voltage (p.u.)

1
1

Voltage (p.u.)
0.98
0.995
0.96

0.94 0.99

0.92 0.985
0.9
0 5 10 15 20 0.98
Time (hour) 20 40 60 80
k
Figure 9. All consumer nodes’ voltage profile for conventional method
Figure 11. Proposed dynamic voltage reference ( )

1.06
0.45
1.04 Zref LRT
0.4 Zref SVR
1.02 0.35
Node Voltage (p.u.)

Impedance (p.u.)

1 0.3
0.98 0.25

0.96 0.2

0.94 0.15

0.92 0.1

0.9 0.05
0 5 10 15 20 20 40 60 80
Time (hour) k

Figure 10. All consumer nodes’ voltage profile for proposed method Figure 12. Proposed dymanic line drop compensation impedance ( )

In the proposed voltage control method, VRs’ LDC

0.4
parameters are not constant but changing with flowing current LRT
through VR. Fig. 11 and 12 depict determined voltage SVR
0.3
reference and equivalent line drop compensation impedance
respectively, where calculated LDC parameters approximated
Current(p.u.)

0.2
to 0.005p.u.( ) and 0.01p.u.( ) are illustrated and
considered for analysis for simplicity. Variation in the
0.1
calculated LDC parameters can be understandable from
flowing current at VR which is illustrated in Fig. 13. It is clear
0
that reverse current flows in both VRs due to large penetration
of PVs. Since current through both LRT ( ) and SVR
( ) are higher than for certain duration (for -0.1
0 20 40 60 80
), LDC parameters are constant, determined from case I. k
However, during the daytime for from 31 to about 38, Figure 13. Measured current at both LRT amd SVR
and become less than but higher than zero (case
II) which increases and while decreases . Later on, Fig. 14 illustrates tap operation of VRs for conventional
there is sudden drop in determined but sudden rise in and proposed methods. It can be seen that there are
for (LRT) and (SVR) due to considerable differences in initiation time of tap operation
reverse current at noon time. When comparing the obtained between both methods whereas no significant increment in the
results with the Fig. 5, determined in case III should be tap operation is encountered in proposed method compared
larger than of case II according to Fig. 5, but it is smaller. with conventional method. Fig. 15 depicts the power
The reason is that reverse current flows in both VRs are small consumption pattern of all HPWHs that take part in proposed
method. Most of the HPWHs’ operation is initiated during the TABLE III. OPTIMAL DEAD BAND
daytime with slight differences in their starting time as voltage Optimal dead band (%)
at pole transformer of those HPWHs’ nodes reaches in
1.7
different time, whereas HPWHs at nodes with voltage below
during the whole day are operated during night time. In 1.1
order to find the effectiveness of proposed method, total
voltage violation of all nodes (in voltage seconds) and total
number of tap operations (both LRT and SVR) of V. CONCLUSION
conventional, HPWH control only, combined conventional Dynamic VR operation with demand side management by
and HPWH control, and proposed method for PV output on a use of controllable load (HPWH) has been proposed for
sunny day have been determined, are shown in Table I. It is voltage control for a distribution system with large penetration
clear from the results that voltage violation is extensively of PVs. Proposed dynamic LDC parameter determination
alleviated by proposed method whereas conventional and method adjust the tap operation of VRs very effectively even
HPWH control methods show extensive voltage violation. for a VR with reverse power flow, for which conventional VR
Optimal dead band that mitigates voltage violation and tap operation can’t be able to provide better solution. Besides, the
operations of proposed method are given in Table II. selection of proper dead band and its adjustment based on the
load demand is another important factor that affects voltage
9
LRT controllability and total number of tap operations of VRs
8 Proposed rather having constant dead band value (conventional
Tap position

Conventional
7
6
method). It is confirmed that operation of consumer
5 controllable load is necessary during daytime, thus VR can’t
4 always be able to control the voltage successfully for entire
3
0 5 10 15 20 distribution system. It is apparent that proposed method
SVR mitigates voltage violation extensively while having
9 insignificant increment in tap operations compared that of
8
Tap position

7 conventional method. Hence dynamic VR operation can be

6
5
considered as first step and demand side management as
4 second step to provide safe voltage in the future distribution
3
0 5 10 15 20 system.
Time (hour)
REFERENCES
Figure 14. Tap operation of VRs
[1] K.M. Muttaqi, A.D.T. Le, M. Negnevitsky, G. Ledwich, “A Coordinated
voltage control approach for coordination of OLTC, voltage regulator and
DG to regulate voltage in a distribution feeder”, Industry Applications
Society Annual Meeting, 2013, Lake Buena Vista, FL
1
[2] Y. Hayashi, Y. Hanai,J. Matsuki, Y. Fuwa and K. Mori, “Determination of
Power consumption (kW)

optimal control parameters of line drop compensator in distribution

0.8
network with distributed generators”, 16th PSCC, Glasgow, Scotland, July
14-18, 2008
0.6
[3] J. Choi and J. Kim, “Advanced voltage regulation method at the power
0.4 distribution systems interconnected with dispersed storage and generation
systems”, IEEE Trans. Power Delivery, Vol.15, No. 2, April 2000
0.2 [4] N. Daratha, B. Das and J. Sharma, "Coordination Between OLTC and
SVC for Voltage Regulation in Unbalanced Distribution System
0 Distributed Generation", IEEE Trans. Power Systems, Vol. 29, No. 1,
0 5 10 15 20
Time (hour)
January 2014
[5] Y. Kim, S. Ahn, P. Hwang, G. Pyo and S. Moon, "Coordinated Control of
Figure 15. Power consumption pattern of HPWHs
a DG and Voltage Control Devices Using a Dynamic Programming
Algorithm", IEEE Trans. Power Systems, Vol. 28, No. 1, February 2013
TABLE II. SIMULATION RESULTS COMPARISON
[6]Y. Miyamoto and Y. Hayashi, “Evaluating Improved Generation
Total tap Efficiency: One Year Using Residential PV Voltage Control with a
Voltage violation operations Clustered Residential Grid- Interconnected PV”, 3rd IEEE PES Innovative
(V s) Smart Grid Technologies Europe (ISGT Europe), Berlin, 2012
LRT SVR
6 [7] J. Yi, P. Wang, P. C. Taylor, P. J. Davison, P. F. Lyons, D. Liang, S.
Conventional method 2.1 x 10 1 6 Brown and D. Roberts, “ Distribution network voltage control using
HPWH control 1.19 x 10
6
- - energy storage and demand side response”, 2012 3rd IEEE PES Innovative
Smart Grid Technologies Europe (ISGT Europe), Berlin
Conventional method 5
6.17 x 10 1 6 [8] A.L.M.Mufrais and J.Baba, “Local control of heat pump water heaters for
and HPWH control
3 voltage control with high penetration of residential PV systems”, Eighth
Proposed method 2.89x10 2 7 International Conference on Industrial and Information Systems (ICIIS
2013), 17-20 Dec 2013, Kandy, Sri Lanka