Citation for published version:
Gee, AM, Robinson, F & Yuan, W 2017, 'A superconducting magnetic energy storage-emulator/battery
supported dynamic voltage restorer', IEEE Transactions on Energy Conversion, vol. 32, no. 1, 7567599, pp. 55-
64. https://doi.org/10.1109/TEC.2016.2609403
DOI:
10.1109/TEC.2016.2609403
Publication date:
2017
Document Version
Peer reviewed version
Link to publication
© 2017 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other
users, including reprinting/ republishing this material for advertising or promotional purposes, creating new
collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this
work in other works.
University of Bath
Alternative formats
If you require this document in an alternative format, please contact:
openaccess@bath.ac.uk
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners
and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Download date: 26. Aug. 2022
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
1
Abstract— This study examines the use of superconducting
magnetic and battery hybrid energy storage to compensate grid
voltage fluctuations. The superconducting magnetic energy
storage system (SMES) has been emulated by a high current
inductor to investigate a system employing both SMES and
battery energy storage experimentally. The design of the
laboratory prototype is described in detail, which consists of a
series-connected three phase voltage source inverter used to
regulate AC voltage, and two bidirectional DC/DC converters
used to control energy storage system charge and discharge. ‘DC
bus level signaling’ and ‘voltage droop control’ have been used to
automatically control power from the magnetic energy storage
system during short-duration, high power voltage sags, while the
battery is used to provide power during longer-term, low power
under-voltages.
Energy storage system hybridisation is shown to be
advantageous by reducing battery peak power demand compared
with a battery-only system, and by improving long term voltage
support capability compared with a SMES-only system.
Consequently, the SMES/battery hybrid DVR can support both
short term high-power voltage sags and long term undervoltages
with significantly reduced superconducting material cost
compared with a SMES-based system.
Index Terms-- Dynamic Voltage Restorer (DVR), Energy
Storage Integration, Sag, Superconducting Magnetic Energy
Storage, Battery.
I. INTRODUCTION
HE improvement of power quality is an important
objective for electrical utilities and industrial and
commercial consumers. Highly intermittent distributed
generation, rapidly changing loads, and direct-off-line power
electronic systems all contribute to reduced power quality
causing equipment downtime, overload and failure leading to
lost revenue [1].
Voltage disturbance is a common problem and under-
voltage conditions have been seen to occur more frequently
than overvoltage conditions [2]. Short-term under-voltage sags
Manuscript received 02/11/2015. This work was supported in by EPSRC
grant EP/K01496X/1.
Dr. F. Robinson (e-mail: F.V.P.Robinson@bath.ac.uk) and Dr. W. Yuan
(e-mail: W.Yuan@bath.ac.uk) are with the Electronic and Electrical
Engineering Department, University of Bath, Bath BA2 7AY, U.K.
A. Gee (e-mail: ag15969@bristol.ac.uk) was with the Electronic and
Electrical Engineering Department, University of Bath, Bath BA2 7AY, U.K
and is now with Department of Electrical and Electronic Engineering,
University of Bristol, Bristol, BS8 1UB, U.K.
are defined in IEEE Std. 1159-1995 [3] as a decrease to
between 0.1 and 0.9 p.u. (per unit) r.m.s voltage for durations
of 0.5 cycles to 1 min. They occur more frequently than long-
term under-voltages with significant costs to industry [4].
Long-term under-voltage events are defined as a measured
voltage less than 0.8-0.9 p.u. r.m.s voltage, lasting longer than
one minute [3] and can lead to load shedding and potentially
to voltage collapse [5]. The study below presents a means by
which both short-term and long-term voltage fluctuations can
be mitigated at the load using short-term magnetic energy
storage and long-term battery energy storage.
II. LITERATURE REVIEW
Methods to mitigate long-term voltage disturbance, such as
load disconnection [6] or modification of loads for greater
low-voltage ride-through capability may be impractical [7].
Alternatively, supply voltage can be stabilised by tap changing
transformers, uninterruptable power supplies (UPS), shunt-
connected compensators, or dynamic voltage restorer (DVR)
systems. Tap changing transformers have been shown to suffer
from a slow response time and can only output discrete
voltage levels [8]. UPS systems provide the complete voltage
waveform during a power failure and may prove costly and
unnecessary in the event of partial voltage sags. A DVR is a
series-connected device capable of voltage compensation with
fast response time by injecting a voltage in series with the
supply.
DVR systems can be self-supporting by using power from
the grid to mitigate disturbances [9]. Alternatively, DVR
systems can use energy storage to provide power during
compensation such as capacitors [10] for short-term storage or
batteries [11] for longer-term storage. Nielsen and Blaabjerg
[12] have shown that capacitor-supported DVR systems can
suffer from relatively poor performance for severe and long
duration sags. A recent study has shown that an ultra-capacitor
based DVR [13] can be used to mitigate short-term voltage
sags lasting less than one minute. Wang and Venkataramanan
[14] have shown that flywheels are a viable short-term energy
storage technology for use with voltage restorer systems both
experimentally and by simulation. Kim et al. [15] have
described a 3 MJ/750 kVA SMES-based DVR system and
shown experimental results confirming that SMES is suitable
for the compensation of short-term voltage sags. Shi et al. [16]
have used a system-level simulation to also show that SMES
A Superconducting Magnetic Energy
Storage-Emulator/Battery Supported
Dynamic Voltage Restorer
A. M. Gee, F. Robinson, Member, IEEE and W. Yuan.
T
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
2
energy storage is capable of compensating voltage sags lasting
100ms.
Short-term voltage compensation alone may not be
sufficient to protect a sensitive load as both long-term [5], [17]
and short-term [2-4] voltage stability has been shown to
present a problem for many consumers. For this reason, this
study considers the use of SMES/battery hybrid energy
storage to compensate long and short-term voltage
fluctuations. Woong et al. [18] have also considered a
SMES/battery hybrid and shown it is viable for smoothing of
renewable energy generator output power and can result in
reduced energy storage system capacity and prolonged battery
life. Li et al. [19] have shown that a SMES/battery energy
storage system can improve battery lifetime in electric buses.
Deng et.al. [20] have presented a SMES/battery hybrid system
for reducing peak grid power in an electric vehicle charging
station. Nie et al. have also presented a SMES/battery hybrid
system and shown its feasibility in dealing with long term and
short term charge/discharge events in a wave energy
conversion system [21]. This study extends previous
simulation-based SMES/battery hybrid system studies [18-21]
by considering the hardware implementation of a
SMES/battery energy storage system. The design is shown to
be capable of interfacing SMES and battery energy storage
systems and controlling their power sharing to support a three
phase load, during both long-term and short-term voltage sags.
This has benefits in terms of improved long-term voltage
support capability and reduced costs compared with a purely
SMES-based system. Additional benefits include reduced
battery power rating requirement and an improvement in
expected battery life compared with a battery-only system.
III. METHODOLOGY
Fig. 1 shows the DVR system considered. The SMES has
been emulated by a 15mH, 100A inductor. During a voltage
error a three-phase inverter is used to generate the
compensation voltage at the primary of the injection
transformers (T1-T3) so that the load voltage remains close to
nominal. DC/DC converters are used to interface the battery
and SMES-emulator to the DC bus. An auxiliary supply (Aux.
Supply) is used to support the DC bus during standby
operation and charge the energy storage devices. The auxiliary
supply is disconnected and the energy storage devices provide
the necessary power for the inverter to support the load during
a voltage error.
A. DVR Control
The objective of the DVR control system is to minimise
supply voltage variations at the load terminals. This is
achieved by generating a compensating voltage at the series
injection transformer terminals. The phasor diagram in Fig.
2(a) shows various DVR voltage control techniques. ‘In phase
compensation’ causes the compensating voltage to be in phase
with the incoming supply voltage and has been shown to result
in the lowest DVR power rating [22]. ‘Pre-sag compensation’
preserves the phase of the incoming supply at the time a sag
occurs which can be beneficial in protecting loads that are
sensitive to phase disturbances. ‘Energy optimal’ control is
used to minimise DVR energy storage requirement by
injecting a voltage in quadrature to the load current. ‘In phase
compensation’ and ‘Pre-sag compensation’ have been
considered in this study. The control scheme was
implemented in the synchronous reference frame as shown in
Fig. 2(b) by converting three phase AC quantities to
equivalent two phase quantities:
sc
sb
sa
s
s
V
V
V
V
V
2
3
2/1
2
3
2/1
0
1
3
2
(1)
Figure 1. Hybrid energy storage DVR system configuration.
Figure 2. Vector diagram of DVR control strategies [1] U
dvr1
: ‘In
phase compensation’. U
dvr2
‘Pre-sag compensation’. U
dvr3
: ‘Energy
optimal control’. (b) DVR control system.
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
3
s
s
pllpll
pllpll
sq
sd
V
V
V
V
cossin
sincos
(2)
where V
sa,b,c
, V
s
a
,
b
, V
sd,q
are the supply voltages and θ
pll
is the
estimated supply phase angle.
A phase-locked loop (PLL) was used to determine the phase
angle θ
pll
of the incoming supply based on an algorithm which
is robust in the presence of harmonics, non-symmetry and
transients [23]. Fig. 3 shows a Simulink implementation of the
PLL algorithm. This algorithm minimises the sine of the phase
error term causing the control system to be non-linear. For this
reason, the PI controller gains were tuned empirically. The
PLL algorithm requires the cosine and sine of the incoming
supply angle as inputs which can be obtained geometrically
using the orthogonal reference frame voltages from (1) as:
22
)cos(
ss
s
VV
V
(3)
22
)sin(
ss
s
VV
V
(4)
The PLL controller can be tuned to preserve the phase of
the incoming supply before a sag event with phase jump or,
alternatively, the PLL can be made to track the phase of the
incoming supply during a sag with phase-jump. Consequently,
by changing the PLL gains the system can be controlled to
provide ‘in phase’ or ‘pre-sag’ compensation. Fig. 4 illustrates
the results of tuning the PI controller in this way.
To detect the presence of a voltage error, the following
inequality was used [24], [25]:
thresholdqsqs,dd
VVVVV
2
,
2
**
(5)
where V
s,d,q
is the measured load voltage and V*
d,q
is the
desired nominal voltage in the synchronous reference frame.
Inequality (5) was also used to trigger the disconnection of the
DC bus auxilliary power supply (see Fig. 1).
The compensation voltage V
refd,q
is determined, based on the
error between the desired nominal voltage and the supply
voltage:
qdqdqrefd
VVV
,,,
*
(6)
The PWM phase reference voltages V
refa,b,c
were generated by
transforming the required compensation voltage to the rotating
three phase reference frame:
refq
refd
pllpll
pllpll
ref
ref
V
V
V
V
cossin
sincos
(7)
ref
ref
refc
refb
refa
V
V
k
V
V
V
31
31
0
31
31
32
2
3
(8)
The injection voltages, V
ref_a,b,c
, were multiplied by a feed-
forward constant, k to compensate for losses within the power
stage.
Sine-wave pulse width modulation (SPWM) or space vector
modulation (SVM) were considered for generating the inverter
output voltage. SVM is advantageous due to better utilisation
of the DC bus voltage and which allows deeper sag
compensation. However, SPWM allows the possibility to
mitigiate unbalanced faults so this technique was implemented
in this study. The inverter control was implemented using a
Texas Instruments F28069 32-bit micro-controller by
discretisation of the control and PLL algorithms. The inverter
system parameters are listed in the Appendix, Table AI.
Figure 3. Simulink implementation of PLL algorithm where ω0 is the
fundamental output frequency in rad/s.
Figure 4. Simulated PLL Algorithm results: (a) Simulated voltage sag
with phase jump (b) Phase jump angle (c) Blue trace: supply phase
angle. Red trace: PLL output: ‘Pre-sag compensation’ with controller
gains: k
p
= 0.5, k
i
= 5, (d) Blue trace: supply phase angle. Red trace: PLL
output: ‘In phase compensation’ with controller gains k
p
= 200, k
i
= 50.
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) <
4
B. Energy Storage top-Level Control
The objective of the top-level energy storage control
strategy was to control the charge/discharge of each energy
storage device. A current vs. voltage active droop
characteristic was chosen as this strategy has been shown to
provide good stability and active power sharing [26]. The
converter reference currents are based solely on the level of
the DC bus voltage which is advantageous as high-bandwidth
communication between the three different converters is not
necessary. The charge or discharge priority of an energy
storage device can be adjusted raising or lowering its nominal
DC bus voltage using a technique known as “DC bus
signaling” [27]. The system was configured to prioritise the
SMES-emulator to charge/discharge before the battery. By
always prioritising the short term energy storage, battery
power cycling is reduced which can improve battery lifetime
[19].
The current droop characteristic for each device is shown in
Fig. 5, and is made up of three regions of operation. When the
DC voltage is above voltage level V
h(x)
(where x refers to
energy storage system 1 or 2) or below below V
l(x)
,, the
converter current is limited to I
max(x)
or -I
max(x)
. In between V
h(x)
and V
l(x)
, current is controlled based on the linear current vs.
voltage relationship:
xbusnomx
kVVI
(9)
where k
x
is a droop coefficient (A/V) and I
x
is the energy
storage converter reference current.
C. SMES-Emulator System
To reduce costs and be able to evaluate a SMES-battery
hybrid DVR control platform, the SMES device was emulated
by using a 15mH iron-core inductor in this study. The SMES
converter was based on the asymmetric H-bridge
configuration shown in Fig. 6. This converter was rated at up
to 220A continuous current using forced air cooling. During
charge, Q2 is held ON and Q1 is modulated whereas during
discharge Q1 is held off and Q2 is modulated. The
relationship between DC bus current, inductor current and
duty ratio, is given by (12) and (13) for charge and discharge,
respectively [16].
1
DII
smescsmes
(10)
)1(
2
DII
smescsmes
(11)
For active current droop control according to (9), the desired
converter output current Icsmes is given by (12).
smessmesnombuscsmes
kVVI
_
(12)
where k
smes
is the gradient of the droop controller and V
nom_smes
is the nominal voltage of the droop controller.
The required duty ratio can be determined by substituting
Eq. (10) during charge or (11) during discharge into (12) and
solving for duty ratio in real time. This allows the SMES-
inductor to be charged or discharged by simply raising or
lowering the DC bus voltage relative to the nominal V
nom_smes
.
The controller was implemented using a 16-bit microcontroller
(PIC24HJ128GP502) with a switching frequency of 500Hz.
This low switching frequency was chosen to allow for extreme
(>>90% and <<10%) duty ratio operation without causing
overly narrow gate pulses. When operating at 100A nominal
current, conduction losses could be reduced significantly by
the use of low on-state resistance MOSFETs as opposed to
IGBTs in this system. This is expected not to be the case for
larger systems, operating at significantly higher nominal
currents and voltage ratings.
D. Battery System
A bidirectional synchronous-buck converter rated at 40A
output current with hysteresis current control was used to
control battery current. This topology has been previously
reported for use with interfacing an ultra-capacitor energy
storage system to the DC bus in DVR application [13].
However, the proposed system differs from previous studies
[13] in that a variable-frequency hysteresis current control has
been used. This is advantageous as it features cycle-by-cycle
current limiting, making it tolerant to short circuit faults. Also,
the proposed technique is shown below to be globally stable
over the operating range whereas typical current mode control
techniques described previously [13] require slope
compensation to ensure global stability [28]. Further
advantages include good dynamic current-tracking capability,
and robust performance despite variation and uncertainty in
operating conditions [29].
Assuming lossless, ideal components, the inductor current
in Fig. 7 is given by (13).
Figure 6. SMES DC/DC converter.
Figure 5. Energy storage systems active current droop control.