Energy Storage Systems – Grid Connection Using Synchronverters

2016 ISCEE International Conference on the Science of Electrical Engineering
978-1-5090-2152-9/16/$31.00 ©2016 IEEE
Energy Storage Systems – Grid Connection Using
Synchronverters
Gal Barzilai Lior Marcus George Weiss
School of Electrical Engineering
Tel Aviv University
Tel Aviv, Israel
Abstract—We investigate the design of future energy storage
systems by exploring one particular solution, in simulations. We
use Lithium-ion batteries for storage, a dual active bridge (DAB)
for DC to DC conversion and a synchronverter to transfer energy
from the DC bus to the utility grid. We show how some recent
improvements to the synchronverter algorithm can be combined
with a battery management algorithm that will charge or
discharge the batteries to achieve a maximal profit for the
operator of the storage station, while at the same time
contributing to the stability of the grid by providing frequency
droop, voltage droop inertia and fault ride-through. We discuss
different open loop and closed loop control algorithms for the
DAB as well as for the overall system that will further increase its
efficiency. We simulate the use of two DABs, one for the positive
DC voltage and one for the negative one, with a 5KW three-level
synchronverter working on the 230V grid.
Keywords—Synchronverter; energy storage system; dual active
bridge; droop control; model predictive control.
I. INTRODUCTION
Renewable energy sources are becoming more and more
popular, for economic and environmental reasons [1]. In recent
years scientific advances have made solar panels and other
renewable energy sources cheaper and more efficient. However
renewable energy sources suffer from intermittency (variable
and partly unpredictable production rate) and lower reliability
in comparison to conventional energy sources, and their
production cannot be tuned to the demand curve. The obvious
solution to address this is the use of energy storage [2] systems
that can inject power to the utility grid according to the demand
at hand. Energy storage systems can also contribute in potential
to the stability of the grid and to the grid quality [3-4,18].
Although today's energy storage systems prices are
prohibitively high, we believe that in the near future their
prices will be reduced, which will enable their widespread use.
Energy storage system injecting energy to the grid using
conventional inverters can damage the stability of the grid,
potentially leading to blackouts that have a huge economic
cost. This problem is associated with the large-scale
penetration of inverters into the power grid, regardless of their
role. A new type of inverters, called synchronverters, have
been developed to address this issue. They mimic synchronous
generators, and crucially they add inertia to the electricity grid,
which helps to stabilize it. Their usage in grid connected
energy storage systems can facilitate the market acceptance of
such storage systems, as well as increasing the grid stability
rather than decreasing it. Typically, a battery based energy
storage system will have a DC bus at an approximately
constant voltage that will not be the same as the battery
voltage. Thus, a typical battery based energy storage system
will contain two converters, one DC/DC converter to interface
between the battery and the DC bus, and the inverter
mentioned earlier that is the interface between the DC bus and
the utility grid. The control system of the storage system has to
oversee both converters, ensure safe operation and fault ride-
through, optimize from an economic point of view (using
market prices), send status reports to the grid operator and
receive demands (such as active and reactive power) both from
the grid operator and from the local operator. This requires a
careful design of the control system.
As already mentioned, due to the high price of batteries,
energy storage systems are still at a research and development
stage, and a myriad of solutions are being explored by large
companies as well as by academia. We contribute to this effort
by designing an experimental energy storage system based on
Lithium-ion batteries, a pair of dual active bridges (DABs)
which we believe will be the dominant technology for bi-
directional DC/DC power conversion with galvanic isolation,
and a synchronverter. Our system is of very small scale (up to
5KW power) but it has all the features that a much larger
system would have.
A DAB converter can be operated with zero voltage
switching (ZVS), thus significantly reducing the power losses
of the DC/DC converter, and of the overall system. One control
algorithm used in such a converter is called phase shifting, also
referred to as single phase shifting (SPS). This algorithm is
simple to implement because we control only one variable, the
phase shift. We will show solutions to both voltage tracking
and to load response. SPS gives good performance when the
ratio between the voltages is close to the transformer windings
ratio. For batteries this is a reasonable assumption but we will
also mention more advanced control algorithms, such as triple-
phase-shifting (TPS).
The rest of this paper is organized as follows: In Section II
the synchronverter algorithm with recent developments is
shortly reviewed. In Section III we examine an optimal energy
management algorithm that uses the variable electricity price as
an input, and is based on model predictive control. In Sections
IV we will survey the architecture and control of DAB, in V
we present simulation results and in VI we present conclusions.

II. SYNCHRONVERTER ALGORITHM LATEST DEVELOPMENT
A synchronverter is a three phase inverter that behaves
towards the AC grid like a synchronous machine together with
its droop control loop. Due to its control algorithm a
synchronverter exhibits all the merits and flaws associated with
a synchronous machine. The power part of a synchronverter is
just a classical three phase switched inverter, that can be two,
three or multi-level from the DC side, and may or may not
provide a neutral line. If we assume the simplest situation, two
DC levels and no neutral line, IGBTs for switches, and an LCL
filter on each phase of the AC output, then the power part looks
as in Fig. 1.
Since its introduction in [5] several additions were
suggested to the synchronverter algorithm. We have chosen to
add to the initial algorithm the ability of self-synchronization
which was introduced in [6], an alternative to the standard PLL
for the initial synchronization. Instead of using a PLL, this
solution tries to minimize the difference between the grid
voltages on each phase and the corresponding synchronous
internal voltage on that phase (the local average of the voltage
on the IGBTs), which ideally should be zero at the moment of
closing the circuit breaker, so that the switching current in the
circuit breaker is very small. This is done by calculating virtual
currents between the voltages, using the real voltage samples of
the grid, the synchronous internal voltage available from the
algorithm, and a virtual inductive load between the two.
Synchronization is reached when those currents are minimized
below a predefined threshold. One can view this as an
algorithmic implementation to what happened in classical SGs
some decades ago – in many installations there was a light bulb
installed where we use our virtual load. The light bulb turned
off if there was a very small difference in voltage between the
grid and the output terminals of the SG, indicating
synchronism. This way the operator had a physical indication
of the system that helped him decide when to close the circuit
breaker.
We also have chosen to implement the ideas suggested in
[7] and in [8] to get better stability performance. In [8] an anti-
windup mechanism was suggested. We have chosen to
implement one as well for the self-synchronizing stage for the
same reasons as for the on-grid (normal operation) stage. Fig. 4
shows the new control block. In Fig. 3 we denote ga by the
desired average output voltage of the inverter leg on phase a
which is the input voltage of the LCL filter for phase a. This
signal replaces e in Fig. 1. The signal e is computed from the
synchronverter equations as seen in Fig. 2.
Fig. 1. The power part of a synchronverter may be a standard two level
inverter with LCL filter.
Fig. 2. Control scheme of a self-synchronized synchronverter [6]
Fig. 3. A Synchronverter output with virtual inductor and virtual capacitor,
following [8]. Only phase a is shown.
The h signal in Fig. 3 is the voltage after the virtual capacitor
Cvirt, used to stop any unwanted DC current in the output.
Using g enables us to virtually increase the inductance without
the disadvantages of a larger inductor.
ga =
(n − 1)va + ea
n , ha(t) = ea(t) −
1
Cvirt 
0
t
ia(t)dt (1)
III. AN OPTIMAL ENERGY MANAGEMENT FOR GRID STORAGE
SYSTEMS ALGORITHM
In [9] a similar problem of energy management system was
introduced, see Fig. 5. In our problem we don't have to connect
the PV cells, although this is a possibility later on. It was
shown that the PV cells don't change the optimal control and
therefore we can use the same algorithm discussed in [9] on
our system. The algorithm we introduce was examined on a
lead acid battery, a common type of batteries for solar
installations. Today the two major types of batteries in storage
are lead acid batteries and Lithium ion.
The control algorithm used here was suggested in [9],
chapter 4. We solve a near-optimal charge/discharge control
scheme for a discrete system with a battery. The optimization

Fig. 4. The proposed control scheme for a synchronverter, based on [6], [7] and [8].
criterion is as in (2)-(4), Where Jd2 is our price function
representing the actual electricity cost of our system, which we
want to minimize. a is the usage aging rate of the storage
device. Pe is the price of energy function. ξ accounts for the
efficiency of the DC/DC converter. un is the control function
corresponded to the system current, and δ is the discreet time
step. x1 and x2 represent the remaining charge carriers in the
charge wells. The expression vminδ – Runδ + αI2 is the
discretization of the battery state space module output voltage.
Here, R is the internal resistance, and α represents the
capacitance of the battery. However the contribution to the
voltage is from two sources of charges derived from a two state
approximation of the diffusion model, (2) and (3) below are the
discretization of those two states:
Jd2 = 
n=1
N−1
{a|un|δ − Peξ(un)un(vminδ − Runδ + αI2)} , (2)
I2 =
tn
tn+1
x1(t)dt = x1n I1+
(x1n + x2n)k'c − un
k' (δ – I1) −
1
2 cunδ2
+
cun
k' (I1 – 1) (3)
I1 = 
tn
tn+1
e
−k'(t – tn)
dt = −
1
k' (e−δk'
− 1) (4)
It is important for the next section that we will take a look
at ξ(un) . The efficiency of the DC/DC converter has an impact
on the control system. Therefore a very high utilization is
desired here, and if it could be a high one which can be
simplified into a constant this will simplify our calculation. The
control itself is based on [10] and is based on the quantization
of the state space. Further details on that can be found in [9].
IV. DUAL ACTIVE BRIDGE
There are numerous types of DC/DC converters. We are
using a DAB because of its advantages. The DAB allows bi-
directional power flow. The high number of switches in
comparison to other converters enables it to endure high power.
The transformer gives us galvanic isolation between the low
voltage side and the high voltage side. Article [11] reviews
different control schemes for the DAB from Fig. 6. The most
well-known control strategy is the SPS (single phase shifting),
in this scheme the duty cycles of all the switches is set on 0.5,
the upper switches are complementary to the lower ones in
their state. There is 180° shift between the left pair to the right
pair in each bridge. The control is on the phase shift φ between
the on states of each H-bridge from S1 to S5, from S3 to S7, etc.
More advanced control algorithms were made such as
extended-phase-shift (EPS) and dual-phase-shift (DPS) control,
and triple-phase-shift (TPS) control in which we control ,
(the phase shift between the left side and the right side of each
bridge in Fig. 6.
Fig. 5. PV system with storage through DC/DC converter.

Fig. 6. General structure of the DAB
We remark that SPS is a special case of TPS in which the inter
phase shifts are of 180°.
A closed form solution designed to minimize the
conduction losses both on the transistors and on the
transformer was derived in [12]. This solution is different in a
sense that instead of controlling the three phase shifts
mentioned above, it controls the phase shift between the
bridges and the duty cycle of each bridge. The minimization is
of because these losses are proportional to . This is a
complex solution that requires a lot of computing power.
Another approach is a Fourier based modeling and control of
Active-Bridge converters as suggested in [13]. This approach
tries to simplify the analysis in order to increase the ability to
control a number of active bridges for other topologies such as
Quad-Active-Bridge (QAB), by looking at the first harmonic of
the waveform and then trying to optimize the problem using
optimal control. This method could be extended to a larger
number of harmonics, at the expense of the computational
effort.
A recent article [14] suggests a unified standard for TPS,
while optimizing it and showing that the optimal solution also
provides ZVS for practical implementation. A closed loop
solution was also suggested applying three PI Loops. A
detailed description for closed loops implementation for DAB
can be found on [15-17].
Due to the relatively small differences of the possible input
voltage (small deviation from nominal value) we decided to
demonstrate the use of synchronverters for grid stabilization
via a closed loop based SPS algorithm, as seen in Fig. 7, in
which we used a discrete PI to control the phase for a desired
Vref value which was chosen as 400V. We used the analysis
suggested in [15-16] in order to both include a solution to deal
with variation in the load (modeled as a disturbance with a
feed-forward disturbance compensation as way to mitigate the
load effect). This was done by calculating steady state DAB
power transfer function for all possible angles between −π to π.
The interval is of course continuous but we divide it in to many
discrete intervals. How many is an implementation issue, as
well as how many harmonics to consider in (6) below. For
further details please refer to [15-16].
Iload
Vin
=
8
π2
Np
Ns n=0
N
1
[2n+1]3
sin([2n+1]δ)
ωsL (6)
Fig. 7. Closed loop control for a DAB.
V. SIMULATION RESULTS
We have conducted our simulations using Matlab 2016b.
We have chosen to simulate the batteries using the
specifications of an LG Chem’s RESU 6.4 Ex. We have used
two batteries of this kind, and simulated them by entering their
specifications into a Simpower battery model. The module
chosen was a Lithium-ion model.
We have used Simpower to simulate the DAB and its
control. The DAB was designed with a nominal low input
voltage of 50V and a high output voltage of 400V, using a
transformer winding ratio of 1 to 8. In order to emulate the
closest conditions to the experimental tests we will run, we
used two units of DAB. One was connect as the 400V DC
source of the synchronverter and the other was connected as
the -400V DC source. Fig. 8 below shows that the DAB unit
responds well to a change from no load to the maximum load
of 2.5KW. The effect is a small and short dip of the voltage
We used the synchronverter simulations designed in [7] and
updated its control according to all the modifications listed in
Section II. Finally, we connected the synchronverter to the
grid. We took the simple and yet important case of an ideal
grid. Our simulations include Sections II and IV, the
combination of the system described in Section III is for future
research. The inverter is connected to an ideal grid through an
LCL filter, and we simulated both with a 800[V] voltage
source vs the use of two DAB, each supplying 400[V].
Fig. 8. The controlled voltage when, at 0.5s the maximum load of 2.5KW is
introduced into the circuit
Vout[V]

Fig. 9. Active power sent to the grid (after LCL filter).
Fig. 10. Reactive power sent to the grid (after LCL filter).
We started with Pset=0[W], Qset=0[VAr]. We changed Pset
to 4500[W] at 6.5[s]. As can be observed the active power
adapted very well to the change (and we can observe the effect
of the LPF on Pset). As we can see there was a temporary
change at Q, which is expected. The rest of the Pset and Qset
changes are as follows: t=9[s]: Qset =500[VAr]; t=13[s]:
Pset =-4500[W]; t=18[s]: Qset =-500[VAr]. As we can see both
models were very close with their results. One point worth
mentioning is that a small overshoot was detected on P, the
reasons for which are explained in [8].
VI. CONCLUSION AND DISCUSSIONS
We have investigated how a DAB can be used to connect
Lithium-ion batteries as storage to the power grid via
synchronverters. The voltage created by the DAB is very close
to the one expected to be received by the synchronverter
(400[V]). We are in the process of building a prototype to
demonstrate it as one can see in Fig. 11:
We also suggest simulating and testing the ideas suggested in
Section III.
We believe that the suggested system in this paper can help
improve the stability of the grid as well as supplying power in
times of high demand while taking into account both technical
considerations as well as economical ones.
Fig. 11. Synchronverter storage system for testing purposes
REFERENCES
[1] O. Ellabban, H. Abu-Rub, and F. Blaabjerg, “Renewable energy
resources: Current status, future prospects and their enabling
technology”, Renewable and Sustainable Energy Reviews, vol. 39, pp.
748–764, 2014.
[2] J.P. Barton and D.G. Infield, “Energy storage and its use with
intermittent renewable energy”, IEEE Trans. Energy Convers., vol. 19,
no. 2, pp. 441–448, Jun. 2004.
[3] A. Mohd, E. Ortjohann, A. Schmelter, N. Hamsic, and D. Morton,
“Challenges in integrating distributed energy storage systems into future
smart grid”, in Proc. IEEE ISIE, Cambridge, U.K., Jun. 30–Jul. 2, 2008,
pp. 1627–1632.
[4] S. Vazquez, S.M. Lukic, E. Galvan, L. G. Franquelo, and J. M. Carrasco,
“Energy storage systems for transport and grid applications”, IEEE
Trans. Ind. Electron., vol. 57, no. 12, pp. 3881–3895, 2010.
[5] Q.-C. Zhong and G. Weiss, "Synchronverters: Inverters that mimic
synchronous generators", IEEE Trans. Industr. Electronics, vol. 58,
2011, pp. 1259-1267.
[6] Q.-C. Zhong, P.-L. Nguyen, Z. Ma and W. Sheng, "Self-synchronized
Synchronverters: Inverters without a dedicated synchronization units",
IEEE Trans. Power Electronics, vol. 29, 2014, pp. 617-630.
[7] E. Brown, A Study of the Use of Synchronverters for Grid Stabilization
Using Simulations in SimPower, MSc thesis, Tel Aviv University, 2015.
[8] V. Natarajan and G. Weiss, "Synchronverters with better stability due to
virtual inductors, virtual capacitors and anti-windup", submitted in 2016.
[9] D. Lifshitz and G. Weiss, "Optimal energy management for grid-
connected storage systems", Optim. Control Applications and Methods,
vol. 36, 2015, pp. 447-462.
[10] Tazaki Y, Imura J. Finite abstractions of discrete-time linear systems
and its application to optimal control, Proceedings of the 17th IFAC
World Congress 2008, 10201–10206.
[11] B. Zhao, Q. Song, W. Liu and Y. Sun, “Overview of dual-active-bridge
isolated bidirectional dc dc converter for high-frequency-link power-
conversion system”, IEEE Transactions on Power Electronics, vol. 29,
no. 8, pp. 4091–4106, 2014.
[12] F. Krismer and J.W. Kolar, “Closed form solution for minimum
conduction loss modulation of DAB converters”, IEEE Trans. Power
Electronics, vol. 27, no. 1, pp. 174–188, 2012.
[13] B.J.D. Vermulst, J.L. Duarte, C.G.E. Wijnands, E.A Lomonova,
"Introducing Fourier-Based modeling and control of Active-Bridge
Converters", 2016 IEEE Applied Power Electronics Conference and
Exposition (APEC), pp. 3158 – 3164.
[14] J. Huang, Y. Wang, Z. Li, W. Lei “Unified Triple-Phase-Shift Control to
Minimize Current Stress and Achieve Full Soft-Switching of Isolated
Bidirectional DC–DC Converter”, IEEE Transactions on Industrial
Electronics, vol. 63, no.7, pp. 4169-4179, 2016
[15] D.S. Segaran, Dynamic Modelling and Control of Dual Active Bridge
Bi-directional DC-DC Converters for Smart Grid Applications, PhD
Thesis, Royal Melbourne Inst. of Technology University, 2013.
[16] D. Segaran, D.G. Holmes, and B.P. McGrath, “Enhanced load step
response for a bidirectional dc–dc converter”, IEEE Trans. Power
Electronics, vol. 28, no. 1, pp. 371–379, 2013.
[17] D. Segaran, B. McGrath, and D.G. Holmes, “Adaptive dynamic control
of a bi-directional DC-DC converter”, in Proc. IEEE Energy Convers.
Congr. and Exposition Atlanta, Sept. 2010, pp. 1442–1449
[18] J.M. Guerrero, P.C. Loh, T.L. Lee and M. Chandorkar, "Advanced
control architectures for intelligent microgrids—Part II: power quality,
energy storage, and AC/DC microgrids", in IEEE Transactions on
Industrial Electronics, vol. 60, no. 4, pp. 1263-1270, April 2013.
P(Watt)Q(VAR)

Energy Storage Systems – Grid Connection Using Synchronverters

More Related Content

What's hot (20)

Viewers also liked (9)

Similar to Energy Storage Systems – Grid Connection Using Synchronverters (20)

Energy Storage Systems – Grid Connection Using Synchronverters