energies

Article
Testing Smart Grid Scenarios with Small Volume Testbed and
Flexible Power Inverter
Milosz Krysik , Krzysztof Piotrowski * and Krzysztof Turchan

IHP—Leibniz Institute for High Performance Microelectronics, 15236 Frankfurt, Germany;

[email protected] (M.K.); [email protected] (K.T.)
* Correspondence: [email protected]

Abstract: The growing penetration of Renewable Energy Sources (RES) due to the transition to future
smart grid requires a huge number of power converters that participate in the power flow. Each of
these devices needs the use of a complex control and communication system, thus a platform for
testing real-life scenarios is necessary. Several test techniques have been so far proposed that are
subject to a trade-off between cost, test coverage, and test fidelity. This paper presents an approach
for testing microgrids, by developing an emulator, with emphasis on the micro-inverter unit and the
possibility of flexible configuration for different grid topologies. In contrast to other approaches, our
testbed is characterized by small volume and significantly scaled-down voltages for safety purposes.
The examination is concentrated specifically on the inverter behavior. The test scenarios include
behaviors in case of load changes, transition between grid-tied and islanded mode, connection and
removal of subsequent inverters, and prioritization of inverters.

Keywords: microgrid (MG); smart grid (SG); power hardware in the loop (PHIL); power inverter;



microgrid emulator


Citation: Krysik, M.; Piotrowski, K.;
Turchan, K. Testing Smart Grid
Scenarios with Small Volume Testbed
1. Introduction
and Flexible Power Inverter. Energies
2022, 15, 428. https://doi.org/ The world undeniably must turn from fossil fuels to a greener course to produce
10.3390/en15020428 energy, for instance, with wind, solar, nuclear or bioenergy sources. An important organi-
zation in the transformation is the European Union (EU). The EU, as part of the European
Academic Editors: Pierluigi Siano,
Green Deal, plans to target by 2030 at least 32.5% improvement in energy efficiency, 32%
Hassan Haes Alhelou and
share for renewable energy, and 40% cuts in greenhouse gas emissions (from 1990 levels) [1].
Muhammad Aziz
The European Council established in 2019 the objective of making the EU climate-neutral
Received: 26 November 2021 by 2050, in line with the Paris Agreement. The smart grid (SG) is a key element in the global
Accepted: 1 January 2022 trend of the future energy system. SG is an intelligent network that combines different
Published: 7 January 2022 entities actively involved in energy exchange [2]. The smart grid differs significantly when
Publisher’s Note: MDPI stays neutral
compared to the traditional grid. The first difference is that it enables bidirectional power
with regard to jurisdictional claims in
flow. The Distributed Energy Resources (DERs) are able to make the transition to smart
published maps and institutional affil- grid possible [3]. Actors that are capable to produce more energy from Renewable Energy
iations. Sources (RES) at a certain time should be able to sell it back to the market. The market
itself has to evolve to a system that will get vast amounts of data as an input that need
to be processed and by that create energy prices in the desired period. These data will be
gathered from smart sensors and meters. It will also be transferred to demand response
Copyright: © 2022 by the authors. systems, whose task is to reduce peak demand. What is more, the SG needs to be resistant
Licensee MDPI, Basel, Switzerland. to cyberattacks. The ability to self-heal after power outages is also highly important. The
This article is an open access article transformation from the traditional grid to the smart grid will last a long time and the
distributed under the terms and path of transition will require a challenging variety of technical, social, and economical
conditions of the Creative Commons innovations [4]. Microgrid (MG) is a concept of coordinated clusters of energy storage,
Attribution (CC BY) license (https://
sources, and loads that interact and cooperate with each other and the power system [5].
creativecommons.org/licenses/by/
The purpose is to maximize the use of Renewable Energy Sources (RES). The social aspects
4.0/).

Energies 2022, 15, 428. https://doi.org/10.3390/en15020428 https://www.mdpi.com/journal/energies

Energies 2022, 15, 428 2 of 20

of microgrids, such as increased awareness of climate catastrophe, job creation, and electrifi-
cation of underdeveloped places, are also worthy of recognition. Numerous administrative
actions to facilitate proper functioning and cooperation of MGs are implemented, e.g.,
tradable green certificates and feed-in tariffs [2]. There are further challenges associated
with MG development, such as uncertainties of demand and supply using RES; design
of appropriate Demand-Side Management (DSM) systems; and economical and reliable
operation of MG with high penetration of RES, plug-and-play feature, voltage and fre-
quency control mechanisms [4]. Due to the complexity of the MG, a flexible test system is
an absolute need to ensure a reliable, secure and cost-effective evaluation environment.
MG can operate as an island or be connected to the grid or to other MGs. Moreover,
multiple microgrids can also be interconnected to support and trade with each other, as
described in [6]. Nowadays, in most countries, there is no possibility, or it is not easy, to
implement for the MG to operate as an island. That trend will change in the following
years. Working as an island means that MG has to produce enough energy by itself to
power all the loads within this MG. The inverter unit can switch to islanding operation
in one of two ways: unintentional and intentional [7]. The first instance, if it is not tested
and handled properly, can pose a massive danger for utility workers, home appliances, or
inverters. According to [8], if the root mean square (RMS) value per unit voltage will be
out of range from 0.50 to 1.20 p.u. for at least 0.16 s, or out of range from 0.88 to 1.10 p.u.
for at least 2 s, or the grid frequency deviation exceeds ±1 Hz, then DER should cease to
energize and trip all units.
The inverters can combine their power and work together in parallel without the main
grid. The possibility of parallel work of inverters in MG is a crucial aspect for the efficiency,
flexibility, and reliability of the distributed generation system. Correct cooperation of
inverter units serves to balance active and reactive power. That balance between real
power (P) and reactive power (Q) must be maintained at an appropriate level. If voltage
amplitude produced by one of the parallel inverters does not match, it creates reactive
power circulations at the Point of Common Coupling (PCC). If additionally, the phase of
the inverters is in mismatch, then both Q and P are circulating. Further, every inverter must
produce voltage with an identical frequency. One of the hardware solutions to restrain
circulating current is the transformer located at the output of the inverter [9]. It provides
additional galvanic isolation and diminishes the DC composite of circulating current. The
current produced by the inverters should be controlled and the circulating current has to
be minimized. There are a lot of control methods, to tackle that issue. Methods are divided
into two main categories: communication-based and methods based on droop control.
A communication-based method like active current sharing or master–slave control is
efficient for load sharing, but the necessity of high-speed communication infrastructure, as
well as low redundancy, are the major drawbacks of this method [10]. Droop control does
not require interconnection between inverters, although, in the most basic form, it suffers
from the slow dynamic response and poor harmonic sharing [4]. Nevertheless, numerous
improvements have been introduced to that method, e.g., angle droop control method,
virtual impedance control method and adaptive voltage droop control method.
The smart grid emulator aims at allowing to model the behavior of the real grid. One
of the main aspects is the distributed energy production in different scenarios. That is
why we need a reliable energy generation block that is configurable and controllable. The
inverter block we present and evaluate in this paper shall provide this functionality.
Nearly all the testbeds existing in the literature work with hazardous voltages. The
paper presents a testbed where hazardous voltages can occur only in the primary substation
and secondary substation emulation does not surpass 230 V AC. However, the input DC
voltage of the inverter is equal to 35 V DC and the output voltage is 24 V AC. Besides the
scaled-down voltages, the system is capable of mimicking grid behaviors with improved
security measures. This approach made it possible to exclude concerns to making mistakes
when studying the effects of individual elements influencing others. One of the key
reasons to build a test bed is not to emulate only the normal work, but rather to test
Energies 2022, 15, 428 3 of 20

the emergency situations that might happened in electrical grid and the algorithms for
adjusting to the disturbances or demand. The testbed with scaled-down voltages can
provide these capabilities without creating extremely dangerous situations. This creates
possibilities to implement more accessible testing environment. Furthermore, more risky
scenarios like variety of grid faults can be simulated. Furthermore, care was taken to select
relatively inexpensive components so that the overall cost of the emulator was significantly
below the average MG emulators. What is more, the emulator uses a non-commercial
inverter board, which has the advantage of being available to quickly exchange or replace
given components like operational amplifiers, transistors, or filters. Moreover, during
the design phase, the principle of using only readily available components was followed.
Most of the MG testbeds require roomy laboratories in order to accommodate high-volume
inverters that are larger due to, for example, larger inductors and radiators necessary in
higher voltage applications. Another bulky and costly element is the grid simulator, which
is also not needed in the presented design.
The primary objective of this paper is to investigate the validity of using the inverter
that can serve as part of the MG emulator. This emulator requires a high degree of flexibility
from the inverter, as the inverter itself can be used in many configurable options. Among
these, the inverter should emulate a collection of several photovoltaic panels or a wind
turbine, as well as small power plant. To meet these requirements, the inverter was
put through a series of tests. Another novelty aspect is represented by the scaled-down
voltages. It can be helpful to test risky scenarios without generating very hazardous
conditions. Moreover, the tests presented in this paper can serve as a foundation for future
multi-domain evaluation considering the energy management algorithms, communication
(data exchange within the system) and simulation of energy market aspects.
The rest of this paper is organized in the following order. Section 2 covers the existing
methods of microgrid testing. Advantages and drawbacks are discussed and the realized
development has been described. In Section 3, the structure and the possible topologies
were defined. Principles of the control system of the inverter as well as the communication
with the master controller were described. Test scenarios and results are presented in
Section 4. Finally, Section 5 provides concluding remarks.

2. Testing in Microgrids
Testing MGs can be done in a variety of ways, among others digital simulation, co-
simulation, Real-Time simulation (RTS) Hardware-In-the-Loop (HIL), Controller-Hardware-
In-the-Loop (CHIL), Power-Hardware-In-the-Loop (PHIL) and emulators. Several reviews
and analyses have been made that compare these testing strategies and present a diversity
of paths to create simulation within each category of test approaches [11,12]. To choose
the desired strategy, the trade-off has to be made between the cost, test fidelity and test
coverage [12].
The digital simulation consists of solving mathematical models with the use of high-
level language code and platforms. It is mostly used for the initial state of the system
development. The benefits are flexibility, rapidity, versatility and no risk of damaging the
equipment [13]. Cost varies depending on license period, but can reach several thousand
EUR by year, in the case of choosing, the most popular solution, e.g., Matlab. Worth
mentioning is that there are free simulation tools available, like Julia or GNU Octave. There
is also a possibility that diverse power system modules can be simulated in different simu-
lation tools; therefore, all simulations need to be merged in a co-simulation environment,
for instance, mosaik.
A real-time system is performed on parallel processors. All parts of the system,
for example, the control, protection and measurements, are modeled on the simulation
platform [14]. The key issue of RTS is to ensure appropriate timescale, including idle time
of processors [11].
When some parts of the system exist physically and the software part of the simulation
is connected with the hardware, then the HIL approach is implemented. It is well known
Energies 2022, 15, 428 4 of 20

that HIL requires more time to build and investment costs are usually higher than in the
previously mentioned approaches. A noticeable benefit is the behavior being closer to
the real behavior of the given module. There are certain issues that ones designing the
HIL system need to take into account, such as stability criteria and also time delay formed
in simulation, which might be reflected in harmonics of output signal [13]. The CHIL
systems add hardware to RT simulator, for instance, a power converter controller and
signals are exchanged between them with the use of Analog-to-Digital or Digital-to-Analog
converters [13]. That approach is widely used, and it gives scientists a valuable, small
volume object to test basic algorithms like voltage control or power-sharing in MGs [15] or
testing data exchange methods [16].
PHIL has a similar architecture to CHIL, but requires an additional power amplifier
stage. This allows the use of devices that are involved in exchanging power such as solar
inverters or generators. The appropriate selection of power amplifier requires a priori deter-
mination of bandwidth, expected accuracy and stability of the desired power amplifier [17].
Voltage and current amplitudes flowing through the PHIL system might be closer to the real
values flowing in the power system, thus PHIL simulation has the ability to test more risky,
complex and demanding scenarios. Examples of possible test cases are fault condition
detection, protection validity or power exchange between units. A common approach
to performing the test is to first simulate the system only in software and then compare
the results in the PHIL environment. The authors of [18] present PHIL with the use of a
commercial micro-inverter and put it through a series of stability tests. A complex test class
was presented in [19], among others, abnormal voltage and frequency conditions, voltage
regulation or unintentional islanding were investigated. Another paper [20] presents a
simulation of a PV inverter, for voltage regulation in distribution networks, with the use of
a variable voltage source. The system features and protection, together with reactive power
behavior in case of fluctuating power capability of the PV array, were well tested. There
are numerous PHIL simulators with the use of inverters. However, most of them use the
commercial inverters that are dealing with dangerous voltages and are hard to configure.
The idea of the MG emulator is to imitate, as much as possible, the real behavior of the
MG. Performing test scenarios on an emulator, instead of the real power system, has several
advantages. For instance, there is no possibility of damaging the real equipment, it has the
possibility of concentration of long tests in a shorter period by changing the timescale, it
consumes less space, costs less and is more flexible [21]. One of the examples of such an
emulator is presented in [5]. The authors present the Generator Emulator Controls that was
designed to reproduce favorable grid-forming characteristic of a synchronous generator.
The paper presents tests of plug-n-play ability in MG with 15-inverters. The testbed used a
commercial inverter in the closed case, and it is slightly limiting the flexibility, although
the authors sized the appropriate filter. In [22], the authors introduced an emulator with
a voltage source inverter. The size of the test system is relatively small and due to the
modular concept of the system, it is flexible to made changes in topology, but no test was
performed to present that feature. In the paper, the authors test parallel operation as well as
response to load step changes, but only for two values of the load and without measuring
the frequency deviations at PCC.

3. Structure of the Proposed Testbed

3.1. Structure and Possible Models to Create
The testbed is a smart grid or an MG Emulator, in which the generation is based on
the inverter module that is presented in this paper. The testbed emulates the power grid,
but also includes a (wireless) sensor network that monitors the power grid. The main
task of the emulator is to mimic a full-scale smart power grid to test and evaluate power
management algorithms. To allow creating different topologies of the grid, it follows the
modular approach. There are currently four classes of blocks that differ in functionality
and can be connected to form the desired topology. These blocks are connected physically
Energies 2022, 15, 428 5 of 20

to define the energy flows, but they are also connected logically to define the hierarchical
relations of the monitoring and control components available in each block.
The monitoring and control module available in each emulator block can be consid-
ered as a smart meter with sensing and control features. In a given emulator deployment,
these devices are configured to be organized in a wireless sensor network. It is a kind
of distributed brain of the whole emulator. Each smart meter consists of three compo-
nents. It is based on the IHPNode [23,24] platform with the main part—the FWNode board
based on the MSP430F5438A microcontroller. This board is responsible for exchanging
measurements and control data between the different emulator blocks. The second com-
ponent is the ADCBoard—an extension board for the FWNode, whose task is to measure
electrical quantities for three independent channels, using its 24-bit ADC converter. The
measured electrical quantities are the active power, frequency, reactive power, apparent
power, current, voltage, and the power factor. The last component is an adapter for the
ADCBoard—the ACFrontEnd board. Its task is to adapt the voltage and current values in
each channel to levels that are measurable by the ADCBoard.
Some of the blocks are further equipped with a management unit—a computing
device with more power than a sensor network node, but it interacts with the emulator
using these nodes and it runs the energy management algorithms. The management units
exchange data between each other using the data exchange middleware [25] that also allows
implementing distributed energy management algorithms according to the framework
defined by the European e-balance project [26].
Primary Substation (PS) is the first block that is a part of the emulator. It is an
equivalent of a large high voltage electrical substation in the full-scale power grid, whose
task is to reduce the voltage level from high voltage (HV) to medium voltage (MV) and to
distribute the energy to the underlying secondary substations. In the emulator the primary
substation voltage ratio is 230 V (HV) to 48 V (MV). It is equipped with both, the smart
meter module that monitors the voltage and current parameters on the MV side of the
transformer, and the management unit.
The Secondary Substation (SS) is the second block in the emulator. Its task is to further
reduce the voltage level from the MV to low voltage (LV) and to distribute the electrical
energy of the area that is connected to it. The LV level in the emulator is defined at 24 V. It
is also equipped with both the smart meter module that monitors the voltage and current
parameters on both sides of the transformer (MV and LV), and the management unit.
The emulator blocks can be connected directly to create the energy flows between
these, but it is also possible to use the Transmission Line (TL) blocks for that. Similar to
full-scale transmission lines, the main purpose of this block is to transmit electrical energy.
The transmission line block has a specific structure that allows to split one input into three
independent outputs or to accumulate up to three independent inputs into one output. The
smart meter module monitors the voltage and current parameters and controls switches
that enable cutting of each of the three outputs/inputs. Furthermore, the transmission line
block has the capability to emulate electricity transmission losses/influences by changing
its parameters using passive electronic components between input and output.
The TL block was primarily designed to test energy management algorithms, such as
the reaction to (emulated) transmission line faults. It can thus be used to detect blackouts
and rearrange the topology to distribute power from another source. A TL block is not
equipped with a management unit and is thus controlled by its respective parent substation.
The link between the TL and the substation is logical on the wireless sensor network
level, connecting the smart meter modules. It gives the opportunity to measure electrical
quantities for the entire sub-grid, but also just for a selected branch of the sub-grid. The
parent substation has also the possibility of cutting off the selected sub-grids in the whole
grid using commands that are dedicated for this purpose.
The last and the most complex block in the emulator is the Prosumer block. This
block is responsible for the emulation of power consumption and production on the LV
or MV level. This is the block that contains the inverter presented in this paper. It can
Energies 2022, 15, 428 6 of 20

emulate a single household, either only consuming energy, but also one with, for instance,
a photovoltaic system. However, it can also emulate a larger energy producing site, like an
independent solar or wind power plant. On the energy grid emulation level, the prosumer
block consists of the inverter module and the energy-consuming module connected in
parallel. Both current flows are monitored using the smart meter module independently.
This allows to emulate scenarios relevant for different regulatory settings in different
countries, i.e., the energy consumption and production can be regarded individually or as
a sum. The prosumer block also includes the management unit to run energy management
algorithms.
All these block classes of the emulator provide the opportunity to flexibly design
multiple topologies and test many smart grid approaches on them. An example topology
to be emulated is presented in Figure 1.

Figure 1. An example energy grid topology.

The main issue within the emulator related to the inverter module is its configurability
and control to prevent hazardous situations. The energy generation in different prosumer
blocks is summarized, i.e., the outputs of inverters are connected almost in parallel, thus a
proper definition of the inverter behavior and reliable control mechanisms to realize that
behavior are a must.

3.2. Principles of Operation and Control

The detailed software and hardware architecture is described in master thesis [27]. It
provides a comprehensive explanation of the control mechanism as well as the hardware
design procedure. Nevertheless, the principles of operation and control are presented
in this section. To control voltage and current flow within the emulator, two loops with
proportional-integral (PI) controllers were used. The current control loop is an inner loop.
The operation of this loop starts with the measurements of the instantaneous values of
currents in both leg branches of the inverter bridge. That is performed via the current
Energies 2022, 15, 428 7 of 20

sensing circuit that measures current by calculating the drop on the shunt resistor at the
low side of the inverter branch. Then, the voltage drop is transmitted to the integrated
circuit with an operating amplifier, whose task is to process the signal to the appropriate
level. Then, the signal is transmitted to the Analog-to-Digital Converter (ADC), where
the control unit of this device will process the input signals and generate PWM signals.
By manipulating the PI coefficients, it is possible to shape the characteristics of the PWM
signals. By changing the shunt resistor’s values or components included in the current
sense circuit, it is possible to adjust the resolution and accuracy of the measurements.
Hardware components are easily accessible for replacement, and software coefficients can
be changed via SPI, as described in Section 3.3. The outer voltage control loop senses the
voltage at the output of the inverter and compares it to a desired voltage reference, that is
24 V. The measurement circuits are shown in Figure 2.

Figure 2. Current and voltage sensing loops.

To work in parallel with the grid, or with other inverter units, the synchronization of
phase has to be accomplished. The most widely used, and the most researched method to
ensure synchronizing is the implementation of the phase-locked loop (PLL). This structure
is based on a feedback loop, where a measured output voltage signal is subtracted from a
reference signal, which is the grid voltage. The result of the subtraction is an error signal.
The error signal is minimized by an internal discrete low-pass filter. The discrete oscillators
finally lock the generated output signal with the phase angle of the input signal. The lock
of the signal is proportional to the performance of this system. The structure of this system
contains three basic functional blocks:
• Phase Detector (PD)—the main purpose of this block is to compare the input sinusoidal
signal with the locked sinusoidal signal and to produce error signal out of it.
• Loop filter (LF)—it is a low-pass filter that clears the error signal by damping output
noises and high frequencies from the signal.
• Voltage Controlled Oscillator (VCO)—this block produces the output signal as well as
the locked signal with respect to the nominal frequency.
Energies 2022, 15, 428 8 of 20

Unfortunately, the grid frequency can be close to the cut-off frequency of the Phase-
Locked Loop. Therefore, additional blocks in the PLL system are required. Some ap-
proaches to improve PLL structures are using Adaptive filtering or Second-Order General-
ized Integrator (SOGI) [28]. In both of the methods, the loop filter transfer function must be
implemented and the coefficients of that filter have to be calculated with the discretization
previously performed. The values have to be carefully calculated and tuned. In the project
presented in this paper, the SOGI method was chosen.
When the inverters are working in parallel, one of them, namely, the grid-forming
inverter, must set the reference signals for the rest of the inverters [29]. The other inverters
are following the reference and synchronize to the grid-forming unit. In case of grid
synchronization work, all inverters are grid-following units. When the intentional or
unintentional islanding occurs, then one of the inverters must be transformed into the
grid-forming inverter. This case is covered in the tests. If multiple inverters work in
parallel and the grid-forming inverter fails, another inverter must take its role to ensure the
grid stability.
The activity of the inverter depends on its current state. The state machine is presented
in Figure 3. At first, the inverter is in state 0, where it is inactive. The inverter is waiting
for the command to start operation. If the incoming command arrives, then the inverter
switches to state 1. There, the inverter checks, through the DC voltage sensing circuit, if
there is a DC voltage higher than 34 V DC. If that condition is met, it changes to state 2, and
then, the inverter is sensing the output voltage. In this state, depending on the presence
of the grid voltage, the inverter can go into the synchronization or the islanded mode.
First, if the grid voltage is present, it means that the main grid or another inverters already
power the load. In that case, the inverter control unit needs to make sure that the inverter
is capable to produce enough current according to references that are set to PI controllers.
If this requirement is met, then the transistors are switched on, and the inverter starts to
work in parallel with the other inverter units, on the grid. It remains in that state, until it
will not receive a command from SPI to turn off, or until any of the protections trigger. If
the grid voltage is not present, then it simply goes to islanded mode and powers the loads
by itself. The rules for returning to state 0 are the same as for synchronous operation.

Figure 3. State machine of inverter.

Energies 2022, 15, 428 9 of 20

3.3. Communication between the Inverter and the Master Controller

The communication is done by means of SPI. The SPI is configured with the guidance
from in [30]. The SPI is configured to send a bitstream of length equal to 16 bits. The inverter
unit is always a slave. The four-wire mode is chosen, that is Slave–Output–Master–Input
(SOMI), Slave–Input–Master–Output (SIMO), Slave–Transmit–Enable (STE) and Serial-
Clock (CLK). Data are transmitted at a baud rate of 156.25 KBps. The transmission of
data is performed with the use of the First In, First Out (FIFO) buffer. Each member of
the conversation, namely the transmitter and the receiver, is equipped with the shifted
register and the buffer. In the beginning, the master initializes communication, by setting
the clock. When the data are written to the buffer register, the start of the conversation is
performed. Then, the data are sent, bit by bit, on each rising edge of the clock from the
shifted register. The data are saved on the falling edge of the clock signal. The data from the
Most Significant Bit is transmitted by pin SPISIMO, and the receiver is collecting the data
from the SPISIMO pin. On the receiving side, initially, the data is assigned into the shifted
register, after sending the required numbers of bits, data is passed from shifted register to
buffer. After that operation, the appropriate flag is set and then data is ready to be read.
Afterwards, the interrupt flag is being set, and if the master does not have anything more
to transmit the clock is stopped.
In the proposed MG emulator, the management unit in the prosumer block controls
the inverter in that block. The smart meter module is the interfacing in that case. It is the
master and the inverter is the slave. When the master wants to control the inverter, the
information is transmitted using a 16-bit hexadecimal value that is 0xABBB, where: A is
the ID of the parameter and BBB is the value of the parameter. For instance, the variable
that turns on or off the inverter has an ID equal to 1. Setting logical 1 to that variable turns
on the inverter. Therefore, in order to turn on the inverter, the master controller has to
write a command 0x1001 via SPI. The slave will receive data, encode it, and then change
the parameters.
Further possible commands are:
1. Setting power reference. The inverter can control the current it is producing. When
the current is lowered, then other units should compensate for the voltage drop and
increase current as investigated in Section 4.4.
2. Prioritizing of the inverter. There are 3 levels of priority: low, medium and high. By
default, all units are configured at medium level. When priority is set to high, then
the inverter will try to produce more current, and similarly setting priority to low will
decrease the current.
3. Allowable power limit. The upper and lower ranges of producing power can also be
controlled. If the upper limit is set, and a given inverter is capable of producing
more power, then it will lower the current, to meet the upper boundary. In case the
inverter is not capable of producing more power, than the lower limit indicates, then
the inverter is turned off.
4. Possibility of islanded mode. Master unit can ask for possibility of transition into the
islanded mode. After checking the parameters, the inverter gives an answer about the
possible transition.
5. Request the state. The inverter navigates through the state machine as described in
Section 3.2. After receiving the request, the inverter unit responds in which of the
state it currently operates, among: (i) powered, not active, (ii) islanded mode, (iii)
synchronization mode as a master that is grid-forming control, (iv) synchronization
mode as a slave that is grid-following control.
6. Request the measurement. The microcontroller is equipped with the ADC, thus it
measures electrical quantities within the inverter board like the output AC voltage,
branch current, input DC voltage, frequency. These values can be requested and send
in predefined intervals.
Energies 2022, 15, 428 10 of 20

3.4. Testing Circuit

To test the behavior of microgrids with different number of inverters with different
configurations, several test scenarios were defined. In order to simplify the test procedure,
test schemes were defined using the inverters only.
Figure 4 presents the default circuit for all testing scenarios unless otherwise stated. In
the preface of every test scenario in Section 4, the number of inverters n participating in the
test scenario as well as the preset connection of relays SWn1 and SWn2 are defined. The
physical test circuit is presented in Figure 5.

Figure 4. Block diagram of testing circuit.

Figure 5. Physical testing circuit.

3.5. Measurements
The voltage and current measurement samples are captured by the RIGOL MSO5104
oscilloscope. The data are saved and eventually processed for RMS values or frequency.
Energies 2022, 15, 428 11 of 20

The moving RMS values are computed in a window of five cycles of the 50 Hz sinusoidal
waveform that is recommended by the standards for transient measurements [8]. The
sample rate of oscilloscope measurement varies between test cases. In load transient
changes, the sample rate was equal 500 kSa/s, in connection and removal, the sample
rate was within range (depending upon sub-case) from 50 kSa/s to 5 kSa/s. The voltage
is measured directly at PCC and currents are measured at the output of the transformer
via a current sensor circuit that contains a shunt resistor and Current-Sense Amplifier
INA240 [31].

4. Test Scenarios and Results

This section presents the realized test scenarios and obtained results.

4.1. Connection and Removal of Inverters in Parallel

To assure efficient operation of inverters, the droop control is used. The main principle
of a droop control method is that the voltage amplitude and the phase follow the reference
when droop of load current changes. Active and reactive power flowing through parallel
inverter system can be expressed as on Equations (1) and (2), respectively.

EV
P= sin φ (1)
X

EV cos φ − V 2
Q= (2)
X
where X is the output impedance of the inverter, φ is the phase angle, E is the amplitude of
output voltage produced by the inverter and V is the load voltage.
Therefore, P is mainly dependent on φ, while Q depends on E. Accordingly, the
voltage droop is introduced as Equation (3) and the phase drop as Equation (4), on which
E∗ and f ∗ are the open circuit equivalents, and m and n are the droop coefficients. P0i and
Q0i are ratings of P and Q, while Pi and Qi are real-time values P and Q.

E = E∗ − n( Q0i − Qi ) (3)
f = f ∗ − m( P0i − Pi ) (4)
The ability of parallel work of inverters is one of the key aspects of MGs. The presented
paper includes a test with up to n = 6 inverter units working in parallel. The inverters
work without the main grid that is, switches SWn3 were disconnected, SWn1 and SWn2
were connected. The distortion of sinusoidal waveform of output voltage in the case
of six inverters working in parallel does not differ significantly from the single inverter
operation as presented in Figure 6. A slight decrease in RMS voltage may be observed.
Nevertheless, the value is not meaningful and the capacity of the grid in that configuration
is not weakened and, there is still possibility to attach other units.
The test case is divided into three subcases. The first one involves the connection of six
inverters sequentially, and then the removal of them in the same manner. The total time of
the test takes 200 s. Each new inverter was connected every 20 s. That time interval allows
to equalize the voltage and move the system back to steady state. The second subcase was
testing the connection and further removal of two inverters at the same time. It allows to
test a possibility of connection more than one unit at the same time if the system urgently
requires lots of energy. What is more, it allows to analyze the merge of two unconnected
MGs. That can happen when one MG cannot produce enough energy by itself to satisfy
the demand and other MG that can produce a surplus of the energy can fill the energy gap.
The removal can symbolize the fault of the given unit, group of inverters, or transmission
line fault. The third subcase, as the previous one, checks the possibility to connect multiple
units. In this case to a MG with two inverters the connection and removal of another four
inverters at the same time was analyzed. The results are presented in figures, where the top
Energies 2022, 15, 428 12 of 20

subplot presents the voltage RMS per unit value, the middle subplot presents the number
of inverters working in parallel, and on the bottom subplot, the frequency is presented.
Note, that for a given subcase there might be different ranges on the y-axes of the plot and
also every subcase takes a different time.

Figure 6. Comparison of the output voltage of 1, 3 and 6 inverters working in parallel.

The guideline of allowable voltage levels and time of exceeding these levels are taken
from IEEE standards [8]. These limits are presented in Figure 7. After connection or
removal, the voltage should not exceed boundaries from 0.50 to 1.20 voltage (p.u.) in a
period of maximum 0.16 s.

Figure 7. Allowed voltage levels, adapted from [8].

Energies 2022, 15, 428 13 of 20

The first sub-case results are presented in Figure 8. From the results, it can be concluded
that the voltage spikes have the biggest amplitudes on the connection and removal of the
second and third inverter. Then, the inverters working together, starting to resemble a
stronger grid. Each subsequent inverter is loaded with a smaller value of current, and
the voltage spikes can be absorbed by higher numbers of output filters. The acceptable
limits are not violated in any connection or removal. Each change of inverter numbers
creates frequency oscillations. In every case, the period of oscillations never exceeds 0.20 s.
The frequency spikes were significantly higher when a new inverter was connected than
removed. The removal of the inverter results in smaller voltage spikes, and that is the
reason why the connection is riskier for the system. The moment of connection and removal
of the same inverter is presented in Figure 9. After the removal, the inverter comes back to
steady-state longer than in the case of connection.

Figure 8. The sequence of connection and removal following inverters.

Figure 9. Connection (a) and removal (b) of the inverter.

Energies 2022, 15, 428 14 of 20

The second sub-case presents the connection and removal of two inverters. The voltage
spikes during the event of connection sometimes exceed the upper level and reach a value
of approximately 1.45 volts per unit as presented in Figure 10. The tests were repeated
multiple times and in some cases, the first group of two inverters and in other repetition,
the next group of inverters generates bigger spikes. The random nature of the voltage spike
amplitude depends upon the phase of voltage at PCC at the time of connection. If the phase
difference is further away from 0 degrees, then the connection might be more difficult and
creates more spikes. The spike lasts less than 0.1 s, thus it falls in “may ride-through or may
trip” zone.

Figure 10. The sequence of connection and removal, two inverters at the same time.

The connection and removal of 4 inverters into 2 inverters MG has violated levels
defined in standards (see Figure 11). In that sub-case, the emulator fails most of the
repetition. If the connection of MGs with that much indifferent power level is needed,
then voltage spikes protection devices alike varistor, surge protectors, or special protection
circuits are necessary to implement.
Energies 2022, 15, 428 15 of 20

Figure 11. Connection and removal of neighbor MG with twice the power of the first MG.

4.2. Adaptation to Load Changes

The circuit used for testing is shown in Figure 4 for n = 6 inverters. The inverters work
without the main grid, that is, switches SWn3 were disconnected, and SWn1 and SWn2
were connected. The purpose of this test was to verify that the voltage would stabilize
during the load changes and the possible overshoots or undershoots of load changes should
not exceed standards presented in Section 4.1. The load was switched between 250 Ω to
16 kΩ. With nominal RMS voltage equal to 24 V, the following load values were used:
2.30 W, 1.15 W, 0.60 W, 0.30 W, 0.15 W, 0.05 W. According to standards, the voltage should
not exceed ±10% of nominal value. Therefore, these levels are marked with green and
orange dashed lines. The results are presented in Figure 12.
At the top subplot, the RMS voltage per unit is presented. The green dashed line
stands for the upper allowable voltage limit and the yellow dashed line stands for the
lower limit. Initially, six inverters start to work in parallel and achieve a steady state with a
maximum load within 2 s. The acceptable voltage and frequency ranges are presented in
Figure 7. The series of events was triggered to change the load to smaller values. After that,
maximum step change was done and load from 0.05 W was changed in t = 54 s to 2.30 W
and then back to the minimum value. The particular times of switching the variable loads
are presented in Table 1. First, the load from 2.30 W was changed to half of its value. The
upper voltage was violated only for around 0.03 voltage p.u. The duration of exceeding
boundaries was less than 1 s, thus no protection devices should trip. Each subsequent
change of the load has a smaller overshoot, and shorter settling time, until changes between
maximal values. The heavy load changes may lead to huge voltage drop resulting in
cascading inverter fault and finally to black-out [9]. In t = 54 s, the change from minimal
tested value to maximum tested value was performed. The undershoot of RMS voltage
was equal to 0.11 p.u. below the lower boundary, and subsequent overshoot on change
from maximal to minimal load was 0.16 p.u. above the upper level. The detailed results are
presented in Table 1. The system frequency is resistant to load changes. During the testing,
multiple test runs were performed in order to ensure the repeatability of this test setup.
Energies 2022, 15, 428 16 of 20

Figure 12. Electrical quantities of switching load test.

Table 1. Resistance values of variable load at the time.

Settling Time Overshoot Undershoot

Time [s] Load [W]
[s] [V p.u.] [V p.u.]
0–21 2.30 2.6 0.11 -
21–28 1.15 1.4 0.08 -
28–34 0.60 0.6 0.06 -
34–41 0.30 0.2 0.04 -
41–48 0.15 0.1 0.03 -
48–54 0.05 0.05 0.02 -
54–62 2.30 7.2 - 0.21
62–68 0.05 5.1 0.26 -
68–78 random - 0.05 0.08

Afterward, in t = 68 s, when the system returns to a steady-state, the load started to

change randomly in a dynamic way in a short interval of hundreds of milliseconds within
ten seconds. That has caused only small voltage disturbances and the system responds
quickly and boundaries of applicably voltage levels were not violated.

4.3. Detection of the Anti-Islanding

In the proposed emulator, a combination of two local, passive methods is used:
Over/Undervoltage Protection and Over/Underfrequency Protection [32]. The standards
are described in Section 1. The following test is designed to prove the appropriate operation
of unintentional transition from the grid-connected mode to the islanded mode. The test
circuit is presented in Figure 4 for n = 1 and switches SW11 and SW12 are closed. The
DUT is an inverter that is synchronized with the main grid. In steady-state operation, the
grid fault was simulated via disconnection of the main grid from the inverter unit. In that
occurrence, the flag on the inverter was raised that indicates that the grid voltage is below
the applicable level. The task of the DUT is to transition from the grid-following mode to
the grid-forming mode.
Energies 2022, 15, 428 17 of 20

The moment of transition is captured on the oscillogram and presented in Figure 13.
The oscillogram shows that the inverter requires two periods that is 40 ms to change
between the two previously mentioned modes. Note that during the transition, no voltage
spikes occurred, thus there is no possibility to damage the Device Under Test (DUT) or
any home appliances. The transition was repeated multiple times in order to consolidate
results. In Figure 13, the transition is made instantaneous. Generally, a delay is worth
implementing once the occurrence of under voltage has been detected. This is relevant to
not force the islanded operation in case of temporal voltage disturbances. That ability is
also present in the emulator, and the delay can be freely adjusted to current standards. For
the test, 50 ms delay was selected. To delay duration, a constant of 40 ms should be added,
for the time required for transition. The result is presented in Figure 14.

Figure 13. The transition from grid synchronization mode to islanded mode.

Figure 14. The transition from grid synchronization mode to islanded mode with delay.
Energies 2022, 15, 428 18 of 20

4.4. Prioritization of the Power Produced by the Inverters

The purpose of this test is to introduce prioritization of the inverters in parallel work.
The main idea behind it is to control the current that is produced by a given inverter to
maximize benefits from the current condition of the system that is energy price or the
possible amount of energy to be produced from RES. The results of the test are presented
in Figure 15. The middle subplot presents changes in current produced by the inverter in
grid-following mode. The current reference was changed in steps. The changes in current
did not disturb the voltage, neither the frequency.

Figure 15. The change of current reference of controlled inverter.

5. Conclusions
The paper proposed an inverter-based microgrid energy production emulation for
smart grid testing. The focus was on the challenges of parallel energy production under
diversity of scenarios in microgrid emulation.
The effectiveness of the response to load changes was investigated. The settling time
of DUT was appropriately fast and over- or undershoot was within desired ranges. The
ability of connection and removal of a specified number of inverters was tested. Satisfactory
results were obtained when the number of parallel working units was changed by one,
sufficient results were achieved when the number was changed by two. When the working
microgrid consisting of two inverters and another microgrid consisting of four inverters
were turned on, the voltage spike ranges defined by standards were violated, thus other
supplementary devices should be built-in. The method used to sense grid fault detection
and afterward transition to islanding operation works correctly. After about two periods
of 50 Hz the sinusoidal waveforms of output voltage, the inverter remodels the operating
state from grid-tied to islanding without voltage spikes or overshoot. Moreover, delay
extending the beginning of transition was implemented for safety purposes. The single
inverter of the emulator has the ability to control the current that it is producing when the
MG consists of more than one inverter.
Energies 2022, 15, 428 19 of 20

The test results for the scenarios evaluated in the paper confirmed the validity of
using an inverter in the emulator. An in-depth test procedure revealed the capabilities and
possible limitations of using this inverter. Having a good representation of the behavior of
the power electronic devices in cooperation with the rest of the emulator will allow in the
future to implement additional layers of microgrid management.

6. Future Work
Future research will be devoted to the developments towards testing the emulator
as a whole, considering all the levels of the electrical system from primary substation
through secondary substations down to the prosumer blocks. Further developments and
tests are planned to support the grid resilience aspects, like active and reactive power
control or voltage and frequency ride-through capabilities. On top of the emulator energy
management algorithms will be deployed and evaluated. Another goal of the emulator
is to investigate the accurate behavior of energy end users and the information exchange
between different actors in the grid.

Author Contributions: Conceptualization, M.K. and K.P.; methodology, M.K. and K.P.; software,
M.K.; validation, M.K., K.P. and K.T.; formal analysis, M.K., K.P. and K.T.; investigation, M.K.;
resources, M.K., K.P. and K.T.; data curation, M.K.; writing—original draft preparation, M.K., K.P.
and K.T.; writing—review and editing, M.K. and K.P.; visualization, M.K. and K.P.; supervision, K.P.;
project administration, M.K. and K.P.; funding acquisition, K.P. All authors have read and agreed to
the published version of the manuscript.
Funding: This research was funded by EBALANCE-PLUS grant number 864283.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Deppermann, A.; Paroussos, L.e.a. EU Reference Scenario 2020; Technical Report; European Commission: Brussels, Belgium, 2021.
2. Buchholz, B.; Styczynski, Z. Smart Grids: Fundamentals and Technologies in Electric Power Systems of the Future; Springer: Berlin,
Germany, 2020. [CrossRef]
3. Vadari, M. Smart Grid Redefined: Transformation of the Electric Utility; Artech House Titles in Power Engineering; Artech House:
Boston, MA, USA, 2018.
4. Olivares, D.E.; Mehrizi-Sani, A.; Etemadi, A.H.; Cañizares, C.A.; Iravani, R.; Kazerani, M.; Hajimiragha, A.H.; Gomis-Bellmunt,
O.; Saeedifard, M.; Palma-Behnke, R.; et al. Trends in Microgrid Control. IEEE Trans. Smart Grid 2014, 5, 1905–1919. [CrossRef]
5. Alatrash, H.; Mensah, A.; Mark, E.; Amarin, R.; Enslin, J. Generator Emulation Controls for photovoltaic inverters. In Proceedings
of the 2011 14th European Conference on Power Electronics and Applications, Birmingham, UK, 30 August–1 September 2011;
pp. 1–10. [CrossRef]
6. Wei, C.; Shen, Z.; Xiao, D.; Wang, L.; Bai, X.; Chen, H. An optimal scheduling strategy for peer-to-peer trading in interconnected
microgrids based on RO and Nash bargaining. Appl. Energy 2021, 295, 117024. [CrossRef]
7. Hosseinzadeh, M.; Rajaei Salmasi, F. Islanding Fault Detection in Microgrids—A Survey. Energies 2020, 13, 3479. [CrossRef]
8. IEEE. IEEE Standard for Interconnection and Interoperability of Distributed Energy Resources with Associated Electric Power
Systems Interfaces. In IEEE Std 1547-2018 (Revision of IEEE Std 1547-2003); IEEE: New York, NY, USA, 2018; pp. 1–138. [CrossRef]
9. Khadem, M.S.K.; Basu, M.; Conlon, M. Parallel operation of inverters and active power filters in distributed generationsystem—A
review. Renew. Sustain. Energy Rev. 2011, 15, 5155–5168. [CrossRef]
10. Han, H.; Hou, X.; Yang, J.; Wu, J.; Su, M.; Guerrero, J.M. Review of Power Sharing Control Strategies for Islanding Operation of
AC Microgrids. IEEE Trans. Smart Grid 2016, 7, 200–215. [CrossRef]
11. Vijay, A.S.; Doolla, S.; Chandorkar, M.C. Real-Time Testing Approaches for Microgrids. IEEE J. Emerg. Sel. Top. Power Electron.
2017, 5, 1356–1376. [CrossRef]
12. García-Martínez, E.; Sanz, J.F.; Muñoz-Cruzado, J.; Perié, J.M. A Review of PHIL Testing for Smart Grids—Selection Guide,
Classification and Online Database Analysis. Electronics 2020, 9, 382. [CrossRef]
13. Strasser, T.; de Jong, E.; Sosnina, M. European Guide to Power System Testing: The ERIGrid Holistic Approach for Evaluating Complex
Smart Grid Configurations; Springer: Cham, Switzerland, 2020. [CrossRef]
Energies 2022, 15, 428 20 of 20

14. Vijay, A.S.; Doolla, S.; Chandorkar, M.C. A Real-time de-risked Emulation based Testing Platform for AC Microgrids. In
Proceedings of the 2020 IEEE Energy Conversion Congress and Exposition (ECCE), Detroit, MI, USA, 11–15 October 2020;
pp. 2480–2485. [CrossRef]
15. Akdogan, M.E.; Ahmed, S. Control Hardware-in-the-loop for Voltage Controlled Inverters with Unbalanced and Non-linear
Loads in Stand-alone Photovoltaic (PV) Islanded Microgrids. In Proceedings of the 2020 IEEE Energy Conversion Congress and
Exposition (ECCE), Detroit, MI, USA, 11–15 October 2020; pp. 2431–2438. [CrossRef]
16. Prabakar, K.; Valibeygi, A.; Konakalla, S.A.R.; Miller, B.; De Callafon, R.A.; Pratt, A.; Symko-Davies, M.; Bialek, T. Remote
Hardware-in-the-Loop Approach for Microgrid Controller Evaluation. In Proceedings of the 2020 Clemson University Power
Systems Conference (PSC), Clemson, SC, USA, 10–13 March 2020; pp. 1–8. [CrossRef]
17. Lauss, G.F.; Faruque, M.O.; Schoder, K.; Dufour, C.; Viehweider, A.; Langston, J. Characteristics and Design of Power Hardware-
in-the-Loop Simulations for Electrical Power Systems. IEEE Trans. Ind. Electron. 2016, 63, 406–417. [CrossRef]
18. Nzimako, O.; Wierckx, R. Stability and accuracy evaluation of a power hardware in the loop (PHIL) interface with a photovoltaic
micro-inverter. In Proceedings of the IECON 2015—41st Annual Conference of the IEEE Industrial Electronics Society, Yokohama,
Japan, 9–12 November 2015; pp. 5285–5291. [CrossRef]
19. Davidson, M.; Abbood, H.D.; Benigni, A. Power Hardware in the Loop testing of a PV micro-inverter. In Proceedings of the 2017
6th International Conference on Clean Electrical Power (ICCEP), Santa Margherita Ligure, Italy, 27–29 June 2017; pp. 145–151.
[CrossRef]
20. Wang, Y.; Syed, M.H.; Guillo-Sansano, E.; Xu, Y.; Burt, G.M. Inverter-Based Voltage Control of Distribution Networks: A
Three-Level Coordinated Method and Power Hardware-in-the-Loop Validation. IEEE Trans. Sustain. Energy 2020, 11, 2380–2391.
[CrossRef]
21. Prieto-Araujo, E.; Olivella-Rosell, P.; Cheah-Mañe, M.; Villafafila-Robles, R.; Gomis-Bellmunt, O. Renewable energy emulation
concepts for microgrids. Renew. Sustain. Energy Rev. 2015, 50, 325–345. [CrossRef]
22. Akbarian, H.; Pillay, P.; Lopes, L. Design of a power electronic emulator for parallel operation of renewable energy resources in
microgrids. In Proceedings of the 2015 IEEE International Electric Machines Drives Conference (IEMDC), Coeur d’Alene, ID,
USA, 10–13 May 2015; pp. 1532–1537. [CrossRef]
23. Piotrowski, K.; Sojka-Piotrowska, A.; Stamenkovic, Z.; Kraemer, R. IHP Node–The Experimental Platform for Wireless Sensor
Networks and Internet of Things . In Proceedings of the 24th Telecommunications Forum (TELFOR), Belgrade, Serbia, 22–23
November 2016. [CrossRef]
24. Piotrowski, K.; Sojka-Piotrowska, A. IHP Node Platform as a Base for Precision Farming and Remote Diagnosis in Agriculture.
In Proceedings of the 11th Scientific Conference on Measurement Systems in Research and in Industry, Beijing, China, 20–22
April 2016.
25. Chen, J.; Cañete-Carmona, E.; Garrido, D.; Díaz, M.; Piotrowski, K. PICO: A platform independent communications middleware
for heterogeneous devices in smart grids. Comput. Stand. Interfaces 2019, 65. [CrossRef]
26. Piotrowski, K.; Peralta, J.J.; Jiménez-Redondo, N.; Matusiak, B.E.; Zieliński, J.; Casaca, A.; Ciemniewski, W.; Krejtz, K.; Kowalski, J.
How to Balance the Energy Production and Consumption in Energy Efficient Smart Neighbourhood; MedPower: Athens, Greece, 2014.
[CrossRef]
27. Krysik, M. Control System of the Inverter Model with Regulated Maximum Power and Synchronization with the Power Grid.
Master’s Thesis, University of Zielona Góra: Zielona Góra, Poland, 2020.
28. Teodorescu, R.; Liserre, M.; Rodriguez, P. Frontmatter. In Grid Converters for Photovoltaic and Wind Power Systems; John Wiley &
Sons: Hoboken, NJ, USA, 2007; pp. i–xvii. [CrossRef]
29. Pattabiraman, D.; Lasseter, R.H.; Jahns, T.M. Comparison of Grid Following and Grid Forming Control for a High Inverter
Penetration Power System. In Proceedings of the 2018 IEEE Power Energy Society General Meeting (PESGM), Portland, OR,
USA, 5–9 August 2018; pp. 1–5. [CrossRef]
30. Texas Instruments. TMS320F2803x Piccolo, Technical Reference Manual; Texas Instruments Incorporated : Dallas, TX, USA, 2018.
31. Texas Instruments. INA240 High- and Low-Side, Bidirectional, Zero-Drift, Current-Sense Amplifier With Enhanced PWM Rejection;
Texas Instruments Incorporated : Dallas, TX, USA, 2018.
32. Kim, M.S.; Haider, R.; Cho, G.J.; Kim, C.H.; Won, C.Y.; Chai, J.S. Comprehensive Review of Islanding Detection Methods for
Distributed Generation Systems. Energies 2019, 12, 837. [CrossRef]