Experimental Test Bench for Testing DC Microgrid

Control Strategies
Julen Paniagua1 , Eneko Unamuno2 , Student Member, IEEE, and Jon Andoni Barrena3 , Member, IEEE
Electronics and Computing Department
Mondragon Unibertsitatea
Mondragon, Spain
E-mail: 1 [email protected], 2 [email protected], 3 [email protected]

Abstract—During the last decades the number of microgrids grid are interfaced by a power converter, which deteriorates
and their research have increased notably, as they offer a high the dynamic behaviour of the system [2]–[4]. Depending on
versatility for the integration of distributed generation (DG) units the device type connected to these converters and its purpose,
and renewable energies. Among the different types of microgrids,
dc systems are becoming more popular due to their advantages the applied control techniques need to be different.
over conventional ac systems. However, their control techniques
differ from the ones employed at ac networks, and depending on In addition, depending on the type of microgrid, the control
the device type—either generation, storage systems or loads— strategies that need to be employed are also diverse. In the
alternative control techniques need to be developed. Therefore, in literature three types of microgrids can be found based on
this paper an experimental dc platform is developed with the aim their current nature [5], [6]: ac, dc and hybrid ac/dc. Most of
of testing and evaluating these kind of control strategies. This test the current microgrids are purely ac microgrids, making use
bench consists of a bidirectional four switch buck-boost converter
together with a TMS320F28335 DSP for the implementation of the actual grid infrastructure. However, hybrid ac/dc and
of the control strategy on each power converter. The proposed dc systems are causing great interest due to the advantages
platform is suitable for integrating these systems at a 48 V dc bus they offer with respect to purely ac systems. Nowadays,
and facilitates the evaluation of a wide range of control strategies many equipment and devices are based on dc or include
as they are implemented in the Matlab/Simulink R environment. a dc stage, so by employing a dc network the number of
In this case, the validation of the experimental platform has
been carried out by implementing a virtual-impedance-based these intermediate stages could be decreased. Besides, the
control technique. The experimental results included in the paper control becomes simpler because no frequency/phase control is
corroborate the suitability of the platform for the evaluation of necessary and the absence of reactive power circulation helps
control techniques for dc microgrids. to improve the overall efficiency compared to ac systems [7],
Index Terms—distributed generation (DG), energy storage [8].
systems (ESS), microgrids, renewable energy sources (RES),
virtual-impedance, control of power converters, microgrid central The main purpose of this paper is to develop a fast-
controller (MGCC), test bench.
prototyping and flexible test bench to evaluate different types
I. I NTRODUCTION of control techniques for dc-based microgrids. Nowadays most
part of the research regarding microgrid control is carried
The actual environmental and climate situation is asking out by employing hardware in the loop (HIL) or real-time
for a change in terms of energy generation model and types simulators through mathematical models of real systems [9],
of energy sources. In the last decades, mainly due to the [10]. Even though HILs are capable of emulating very fast
increasing penetration of distributed generation (DG) sys- transients, the aim of the platform developed in this paper
tems—usually based on small-scale renewable energy sources is to provide a more realistic environment for testing and
(RESs)—the energy model is shifting from a top-down mainly validating different types of control strategies. This test bench
non-renewable topology to a decentralized system with a consists of a four switch buck-boost dc-dc converter controlled
higher contribution of alternative generation systems. by a TMS320F28335 DSP. This configuration enables a rapid
However, more advanced approaches where these DG sys- prototyping and testing process as the control strategies can
tems can be integrated are necessary, due to their intermittent be developed in the Matlab/Simulink R environment.
and distributed nature. Microgrids, which are low-scale smart
electric grids, are arising as one of the most interesting The rest of the paper is structured as follows. The mi-
alternatives to handle this integration. The advanced control crogrid scenario is explained in Section II. Subsequently, the
strategies they include enable the management of different theoretical analysis and dimensioning of the dc-dc converter
generation, energy storage systems (ESSs) and loads [1]. together with the control strategy is explained in Section III. In
These characteristics make them suitable for a grid-connected Section IV the developed platform is shown, and the performed
or islanded operation. tests as well as the experimental results are included. Finally,
Most of the challenges of these control techniques originate in Section V, the paper concludes with the most important
from the fact that usually the elements connected to a micro- remarks and conclusions of the research.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.
 II. M ICROGRID SCENARIO with different dynamics. Therefore, the aim is to integrate
a hybrid storage system composed by a lithium-ion battery
Fig. 1 shows a concept of the microgrid to be built in the and an ultra-capacitor module. By operating with these two
laboratories of Mondragon Unibertsitatea, where the platform elements, a resulting high energy and high power ESS is
developed in this paper will be integrated. In this section achieved, since the battery can provide power over longer
the main features of this microgrid are introduced, as the periods of time whereas the ultra-capacitors handle higher
experimental platform will be designed accordingly. dynamic power variations.
The microgrid consists of different DG, ESSs and loads, and One of the key components towards the integration of the
the aim is to integrate an ac/dc converter so that the microgrid above mentioned systems is the dc-dc converter (highlighted
can operate both in islanded mode or grid connected (Fig. 1). in green in Fig.1). Each system requires its own converter and
As it can be seen on the figure, different types of energy control so as to provide certain types of services; DG units
generation systems are included in the microgrid. On the one for instance include maximum power point tracking (MPPT)
hand, various solar panel arrays and a mini wind-power system algorithms to extract as much energy as possible from energy
are installed on the roof, connected to the microgrid through sources; ESSs, on the other hand, are responsible for regulating
dc-dc and an ac-dc converter, respectively. On the other hand, the power variations of the microgrid. This paper is hence
two programmable dc sources have been included with the aim focused on the development of a converter platform that can
of programming different generation profiles and emulating the integrate the different generation, storage systems and loads to
behaviour of other DG units. be connected in the microgrid. Moreover, the test bench will
Regarding the loads connected to the system, in this case be responsible for communicating with the microgrid central
different programmable loads are employed. This enables the controller (MGCC)—also known as SCADA in the classical
emulation of different load profiles to test diverse operation grid—to monitor and carry out a high level management of
conditions in the microgrid. all the devices in the future [11].
In order to compensate energy generation and demand
differences, different ESS units are used on the microgrid. III. D EVELOPMENT OF THE EXPERIMENTAL PLATFORM
Taking into account that with a high penetration of renewable This section embraces the analysis, design and integration
energy sources dynamic variations are high, ESSs—and their of the experimental platform used to evaluate control strategies
advanced control strategies—need to be capable of responding for dc microgrids. The platform proposed in this paper, which

48 V Grid 380 V
Outside

Mains grid

Ethernet

48 V 380 V
MGCC
Microgrid lab

Prog. supply
Prog. supply

Prog. load Prog. load

Energy storage Energy storage

system system
Garage

Fig. 1. Concept of the future dc microgrid experimental platform at the laboratories of Mondragon Unibertsitatea. The converters highlighted in green are
developed in the present paper.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.
can be observed in Fig. 2, consists of a dc-dc converter with a By applying the volt-second balance, the converter conver-
DSP that enables the fast integration of different technologies sion ratio can be deduced, which depends on the value of the
of energy generation and storage systems at a 48 V microgrid. converter duty cycle (D) as follows:
Vbb D
A. Converter topology = (1)
Vi 1−D
Among the different converter topologies that are available
Assuming an ideal converter with no losses, the inductor
in the industry for this type of applications [12], a four
average current equation can be obtained:
switch buck-boost converter has been chosen (Fig. 3). This
topology provides a high flexibility as it is capable of operating Vbb2
with voltage levels above and below 48 V in the input and ILAV = (2)
Vi DR
output, and enables the power transmission in both directions.
Therefore, this type of converter is an interesting alternative where R is the output resistive load of the converter.
to handle the varying behaviour of generation and storage From this equation, the minimum value of the inductor can
systems. On the one hand, the converter must manage the be deduced so that the converter operates in the continuous
intermittent nature of renewable-energy-based DG units and conduction mode:
this device should be controlled so as to extract as much energy 2
R(1 − D)
as possible from the energy source. On the other hand, the Lminccm = (3)
voltage of ESSs is not constant and varies depending on the 2fsw
state of charge (SOC), so the converter must be capable of where fsw is the switching frequency of the buck-boost
operating at certain input voltage range. converter.
Similarly, the minimum value of the inductor must be
B. Converter design calculated to ensure a maximum current ripple in it:
Based on the topology shown in Fig. 3, it can be noticed Vi D
Lmin∆I = (4)
that the four switch buck-boost converter can operate differ- ∆IL fsw
ently depending on the position of its switches. Three main
operation modes are distinguished in the literature, namely
buck, boost or buck-boost mode. In this case, as the input and 1 DTs L
output voltages are very close to each other, the converter will
ii ibb
operate entirely in the buck-boost mode. The switching states S1 iL S4
for this operation mode, taking into account that the power vi C i S2 S3 Co vbb
flows from left to right, can be observed in Fig. 4.
From these switching states the waveforms of the voltage
and current of the inductor can be drawn (Fig. 5).
2 D'Ts L
vg
ii ibb
DSP S1 iL S4
ig vi C i S2 S3 Co vbb
ii
ibb Lf vg
vi
vbb
Fig. 4. Switching states of the four switch buck-boost converter.
ig R
DC-DC Cf
vL iL
S1S3 S2S4
Fig. 2. Experimental test bench diagram. Vi
iLmax
ΔiL
L iLAV
iLmin
S1 S4
iL
vi C i S2 S3 Co vbb DTs D'Ts t
-Vbb

Fig. 5. Inductance voltage and current waveforms of the buck-boost dc-dc

Fig. 3. Four switch buck-boost topology. converter operating in buck-boost mode.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.
 A similar procedure can be applied to deduce the minimum C. Control platform
capacitor at the output of the converter to ensure a certain The aim of the experimental platform is to enable a fast yet
voltage ripple: reliable way of testing different control strategies that might
be integrated in a dc microgrid. Therefore, a TMS320F28335
Vbb D DSP from Texas Instruments has been chosen. This DSP pro-
Comin = (5)
∆Vo Rfsw vides a high flexibility as the control strategies and measure-
ments can be easily programmed in the Matlab/Simulink envi-
As the converter is symmetrical, in this case the same
ronment. The control platform includes several inputs/outputs
capacitance value will be established at the output and the
and peripherals so the communication of variables can be set
input of the converter.
up rapidly.
Fig. 6 is a diagram showing the variable exchange between
For the parameters defined in Table I, the following values the converter, the control platform, the connected device and
have been obtained for the previous variables: the MGCC.
As it can be noticed, the local control of the converter
Lminccm = 293.88 µH is included in the DSP, while control references come from
Lmin∆i = 342.86 µH (6) global control techniques integrated at the MGCC. In this
case, the control is configured so that the converter operates
Comin = Cimin = 57 µF
autonomously with locally measured variables. Therefore, the
MGCC in this case is only employed for monitoring critical
Taking into account the most restrictive ratings for the
values of the converter—currents, voltages, etc.
parameters, the values finally chosen for the inductor and the
Apart from these connections, the platform includes a CAN
capacitors of the converter are:
transceiver in case a commercial battery pack is connected to
the converter via a battery management system (BMS).
L = 350 µH Regarding the control strategies, in this case the virtual-
(7)
Co = Ci = 60 µF impedance technique developed in [3] has been employed in
order to evaluate the experimental test bench. This virtual-
As can be seen on Fig. 2, an LC filter has been included impedance algorithm is a primary control strategy that enables
at the output of the converter to interface it to the 48 V. The to adjust separately the dynamics of the transient as well as
values chosen for this filter are: steady-state response of the converter by adapting different
control parameters. This technique is suitable for controlling
Lf = 100 µH parallel-connected converters regulating the bus voltage with
(8) the aim of sharing power variations and improving the stability
Cf = 30 µF of the microgrid.
The experimental test bench resulting from the previous
development can be observed in Fig. 2. In this case, the
TABLE I component ratings have been calculated to design a converter
F OUR - SWITCH BUCK - BOOST CONVERTER REQUIREMENTS
that is capable of handling approximately 300 W.
Parameter Value IV. E XPERIMENTAL RESULTS
vi [40 − 60] V
An experimental configuration has been set up in order
vo [30 − 80] V
to evaluate the behaviour of the developed experimental test
fsw 100 kHz
bench. In this context, a programmable voltage source has
∆iL 1A
been connected to the input of the buck-boost converter to
∆vo 0.5 V
emulate the voltage of a battery pack (Fig. 7. Similarly,

n
Global control Local control 1
Si 4
MGCC TI DSP ii,ibb,io 3 BB4S BMS
CAN 2
vi,vbb,vo 3
CAN 2

Ethernet

Fig. 6. Variable exchange between the DSP, the converter and the MGCC.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.
 Buck-boost converter TMS320F28335 DSP step-shaped load current profile io , making it to transfer power
Prog. supply Prog. supply
from the battery to the load. Figs. 8 and 9 illustrate the
voltages, currents and powers at the input and output of the
four-switch buck-boost converter. In this case, the converter
supplies the load while regulating the output voltage vo
according to the virtual-impedance ZV control strategy (refer
to [3] for more details).
Even though the input voltage vi is constant and the
output current io is defined by the programmed profile, an
important difference can be observed between two figures in
the output voltage of the converter. This is due to the fact
that in each case, the implemented virtual-capacitor technique
is configured differently. Although in both cases the low-pass
filter gain KLPF is set to 40 and the time constant τLPF to
0.01 s, the virtual-impedance ZV is configured to 10 Ω and
4 Ω, respectively.
Fig. 7. Experimental test bench with a programmable power supply, converter As it can be appreciated, Figs. 8 and 9 show that by varying
and load.
the virtual-impedance value, the output voltage vo steady-state
response is modified. A higher ZV value means a higher
a programmable load has been connected to the output of voltage drop at the output for the same current variation.
the converter, as the purpose is to evaluate the response Therefore, it can be noticed that ZV would be equivalent to a
of the converter and the control platform for different load classical droop controller.
variations). The experimental results demonstrate that the test bench
In the experiment, the converter has been submitted to a proposed in this paper is a flexible and useful tool to test differ-

50 50
vi (V)

vi (V)

48 48

46 46
2 2
i i (A)

i i (A)

1 1

0 0

50 50
vo (V)

vo (V)

40 40

30 30

2 2
i o (A)

i o (A)

1 1

0 0
100 100
Pi Pi
p (W)

p (W)

50 50
Po Po

0 0
0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400
t (s) t (s)

Fig. 8. Experimental results for a virtual impedance value of Zv 10, operating Fig. 9. Experimental results for a virtual impedance value of Zv 4, operating
with a constant current profile. with a constant current profile.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.
ent converter control strategies for dc microgrids. Even though [6] E. Unamuno and J. A. Barrena, “Hybrid ac/dc microgrids—Part I:
the control technique employed in this case is autonomous Review and classification of topologies,” Renewable and Sustainable
Energy Reviews, vol. 52, pp. 1251–1259, dec 2015.
and relies on local measurements, the platform is designed to [7] R. Sirsi, S. Prasad, A. Sonawane, and A. Lokhande, “Efficiency compar-
enable the communication with upper-level regulators such as ison of AC distribution system and DC distribution system in microgrid,”
secondary or tertiary controllers. This feature opens the path in 2016 International Conference on Energy Efficient Technologies for
Sustainability (ICEETS). IEEE, apr 2016, pp. 325–329.
to the development of a diverse amount of control techniques. [8] U. Manandhar, A. Ukil, and T. K. K. Jonathan, “Efficiency comparison
of DC and AC microgrid,” in 2015 IEEE Innovative Smart Grid
V. C ONCLUSIONS Technologies - Asia (ISGT ASIA). IEEE, nov 2015, pp. 1–6.
[9] C. Patrascu, N. Muntean, O. Cornea, and A. Hedes, “Microgrid lab-
In the last decades dc grids have gained a lot of interest oratory for educational and research purposes,” in 2016 IEEE 16th
and therefore they are being widely researched. These systems International Conference on Environment and Electrical Engineering
require more advanced control strategies for the integration of (EEEIC). IEEE, jun 2016, pp. 1–6.
[10] A. Yamane and S. Abourida, “Real-time simulation of distributed energy
distributed generation and storage systems. In this context, dif- systems and microgrids,” in 2015 International Conference on Sus-
ferent tools or platforms are being developed for the evaluation tainable Mobility Applications, Renewables and Technology (SMART).
of these control techniques in realistic environments. IEEE, nov 2015, pp. 1–6.
[11] E. Rodriguez-Diaz, E. J. Palacios-Garcia, M. Savaghebi, J. C. Vasquez,
In this paper an experimental test bench suitable for testing and J. M. Guerrero, “Development and integration of a HEMS with
and validating dc microgrid control strategies is proposed. The an advanced smart metering infrastructure,” 2016 IEEE International
high versatility makes the platform ideal for research purposes: Conference on Consumer Electronics, ICCE 2016, pp. 544–545, 2016.
[12] A. Kwasinski, “Quantitative evaluation of DC microgrids availability:
fast prototyping, flexibility to integrate different devices, ca- Effects of system architecture and converter topology design choices,”
pability to communicate with upper-level regulators, etc. IEEE Transactions on Power Electronics, vol. 26, no. 3, pp. 835–851,
This test bench is composed by a four-switch buck-boost 2011.
converter controlled by a DSP that can be programmed in
the Matlab/Simulink environment. The paper has covered
the design of this converter and the configuration of the
communication between the different agents that are integrated
in the platform.
The validation of the test bench has been carried out by
employing a programmable voltage source and an electronic
load. The voltage source is used to emulate the constant
voltage of a battery pack, whereas with the programmable
load a varying demand profile is reproduced. The experimental
results demonstrate that the platform can be easily updated to
integrate different types of controllers and control parameters.
Moreover, the configuration of the communication network
enables the integration of global controllers such as secondary
or tertiary layers, and the interconnection to other converters
connected in parallel. This confirms the suitability of the
platform to develop diverse dc microgrid control algorithms
in a realistic environment.
VI. ACKNOWLEDGEMENTS
This work has been partially funded by a predoctoral grant
of the Basque Government (PRE 2016 2 0241).
R EFERENCES
[1] A. Kwasinski and C. N. Onwuchekwa, “Dynamic Behavior and Stabi-
lization of DC Microgrids With Instantaneous Constant-Power Loads,”
IEEE Transactions on Power Electronics, vol. 26, no. 3, pp. 822–834,
mar 2011.
[2] E. Unamuno and J. A. Barrena, “Primary control operation modes
in islanded hybrid ac/dc microgrids,” in IEEE EUROCON 2015 -
International Conference on Computer as a Tool (EUROCON). IEEE,
sep 2015, pp. 1–6.
[3] ——, “Equivalence of Primary Control Strategies for AC and DC
Microgrids,” Energies, vol. 10, no. 1, p. 91, jan 2017.
[4] M. Kumar, S. C. Srivastava, and S. N. Singh, “Control Strategies of
a DC Microgrid for Grid Connected and Islanded Operations,” IEEE
Transactions on Smart Grid, vol. 6, no. 4, pp. 1588–1601, jul 2015.
[5] E. Planas, J. Andreu, J. I. Gárate, I. Martı́nez de Alegrı́a, and E. Ibarra,
“AC and DC technology in microgrids: A review,” Renewable and
Sustainable Energy Reviews, vol. 43, pp. 726–749, mar 2015.

Authorized licensed use limited to: Qatar University. Downloaded on September 18,2024 at 13:11:38 UTC from IEEE Xplore. Restrictions apply.