energies

Concept Paper
Overview on Grid-Forming Inverter Control Methods
Peter Unruh * , Maria Nuschke *, Philipp Strauß * and Friedrich Welck
Power System Stability and Converter Technology Division, Fraunhofer Institute for Energy Economics and
Energy System Technology IEE, 34119 Kassel, Germany; friedrich.welck@sma.de
* Correspondence: peter.unruh@iee.fraunhofer.de (P.U.); maria.nuschke@iee.fraunhofer.de (M.N.);
philipp.strauss@iee.fraunhofer.de (P.S.)




Received: 29 February 2020; Accepted: 6 May 2020; Published: 20 May 2020


Abstract: In this paper, different control approaches for grid-forming inverters are discussed and
compared with the grid-forming properties of synchronous machines. Grid-forming inverters are
able to operate AC grids with or without rotating machines. In the past, they have been successfully
deployed in inverter dominated island grids or in uninterruptable power supply (UPS) systems.
It is expected that with increasing shares of inverter-based electrical power generation, grid-forming
inverters will also become relevant for interconnected power systems. In contrast to conventional
current-controlled inverters, grid-forming inverters do not immediately follow the grid voltage.
They form voltage phasors that have an inertial behavior. In consequence, they can inherently deliver
momentary reserve and increase power grid resilience.

Keywords: droop control; grid-forming inverter; electrical inertia; inverter-dominated power system;
virtual synchronous machine; voltage-controlled inverter; voltage-forming inverter

1. Introduction
The ongoing transition of power systems from conventional generation, applying mainly
synchronous machines, to inverter-dominated generation is accompanied by the loss of directly
grid-coupled mass inertia. As a result, the need for inertia contribution from inverter-coupled
power units becomes evident. Furthermore, inverters should contribute to grid-voltage forming,
which in the interconnected power systems nowadays is being performed by synchronous generators.
Grid-forming inverters are available products and their applicability for island systems has been
shown [1]. Unlike conventional current-controlled inverters, such grid-forming inverters form a
voltage phasor that has a certain degree of autonomy but still operates synchronously to the grid
voltage. With a suitable control, the inverter can also deliver momentary reserve without any delay by
retarding the voltage phasor. The property of how the voltage phasor is actuated can be used as a key
indicator for the grid-forming attribute of inverters.
The idea of providing inertial response with inverters in cases of frequency deviations in order to
support angle stability or frequency stability in interconnected grids was published in [2]. Meanwhile,
several grid-forming control concepts for inverters have been introduced. Some of them were
specifically developed for island grid applications, but in future, they may also become relevant for
usage in interconnected systems. This paper’s aim is to give an overview of different kinds of control
schemes and to make these different approaches comparable.
In voltage source converters (VSC), the manipulated value is the switched dc-link voltage, which
is realized by means of the switching pattern of the converter legs. Therefore, the converter itself is
assumed to be a controlled voltage source behind a filter inductor. The control schemes discussed
here are related to the short-term voltage source controlling, i.e., how the voltage source is acting
immediately after an event at the point of common coupling (PCC). Additional control loops, e.g., for

Energies 2020, 13, 2589; doi:10.3390/en13102589 www.mdpi.com/journal/energies

Energies 2020, 13, 2589 2 of 21

resonance damping, or overlaid control loops, can be added and are detached from the specific
grid-forming control approach.
Firstly, a definition of the relevant terms inertia, grid-forming and grid-following inverters is
given. After that, the concepts for grid-forming voltage sources are presented. The dynamic behavior
of the provided voltage phasor, for both the voltage amplitude and the voltage angle are analyzed.
Starting with the dynamics of synchronous machines, grid-forming control approaches with their
voltage phasor dynamics are deduced. The output power or an equivalent quantity, such as current,
serves as an input for the deduced dynamical transfer function.

2. Terminology
Due to the partially misleading or contradicting use of different terms in literature, some relevant
terms are explained in the following.
Electrical inertia, EI, is a property of a power system classically determined by the mechanical
inertia of rotating machines within a synchronous area. The mechanical inertia of a generator results
electrically in the ability to smoothen frequency deviations because of the kinetic energy stored in
the rotor. From a technical point of view, EI refers to the ability to react to a change in the voltage
phase angle at the point of connection with an instantaneous power response, due to the retarded
reaction of the generator’s voltage phasor. Power-electronic devices with suitable control algorithms
(e.g. grid-forming control) can also provide EI. In general, conventionally current-controlled inverters
do not provide EI.
Grid-forming inverter, GFI, denotes an inverter having a control approach with the capability
to control the terminal voltage directly and to form the grid voltage purely by inverters under
consideration of necessary reserve and storage capacity. Such inverters can be able to provide
inherently EI. (Reference [3] was one of the first papers to introduce the term “grid-forming“, and it
has been meanwhile established in academia and industry. Accordingly, it is also used in this paper.
In the opinion of the authors, more suitable terms are “voltage-forming” or “grid-voltage-forming”,
since they state clearly that the grid voltage is the related physical quantity which is formed)
Current-controlled inverter, CCI, or grid-following inverter, in contrast to GFI, denotes an inverter
having a control approach that controls the current injection, e.g., based on terminal voltage
measurement, in order to meet a given power set point. Due to the power set point determination,
the inverter can operate as a grid-feeding inverter, which injects power independent of voltage or
frequency deviations at the terminal. Grid-supporting inverters can be applied to adjust reactive
and active power set points under voltage or frequency deviations. Conventionally, CCIs do not
provide EI, even though they could rapidly adjust their power response to frequency deviations or
even to frequency derivatives. Due to delays in active power set point determination after angle
or frequency changes (e.g., due to frequency measurement) their active power injection is generally
delayed, compared to the active power response of devices with EI capability.

3. Functional Principle of Synchronous Machines

The best-known voltage source responsible for the voltage forming and inertial behavior of today’s
ac power systems is the synchronous machine. The basic concept of a synchronous machine is that
a magnetized rotor induces a voltage at the three-phase stator windings due to the rotational speed.
The rotor can be constructed as permanent-excited or by means of an excitation winding (externally
excited). The latter is assumed here, since it allows a beneficial control action.
Mathematically, the description for the flux linkage between rotor and stator simplifies a lot, when
the variables of both rotor and stator are related to a coordinate system of the same speed as the rotor.
Otherwise, the equations contain inductances, which inconveniently alter with the position of the
rotor [4] (pp. 87–88).
Energies 2020, 13, 2589 3 of 21

This transformation of the coordinate system was first introduced by Park [5], where the x-axis
of the coordinate system was laid on the direct axis of the rotated excitation coil. In literature, it is
referred to as Park transformation or direct-quadrature-zero (dq0) transformation.
Moreover, several simplifications have been proposed, especially for small-signal stability analyses,
which are also applied here. References [6] (pp. 328–347) and [7] give a good overview of the different
models and their effect on the accuracy. Furthermore, detailed explanation concerning backgrounds
and advanced modeling of synchronous machines can be found in [4] (pp. 83–146), [8] (pp. 45–198)
and [9] (pp. 469–496).
Following the model reduction procedure in [10] (pp. 52–54), first the small resistive parts in the
armature are neglected. Besides, another reasonable simplification is achieved by neglecting the very
high-transient behavior of the damper windings. The purpose of the damper windings is to mitigate
rotor oscillations, since a deviation from the synchronous angle speed induces high currents in the
damper windings. As a result, a counter torque proportional to the frequency deviation from the grid
frequency occurs. This effect can be taken into account by adding a damping torque to the swing
equation in relation to the slip of the machine [11] (p. 456), [9] (p. 105). Furthermore, the induced
transformer voltage, which refers to current changing in the coils (e.g. dΨ/dt), is small in comparison to
the rotational electromotive force (emf) and decays quickly, so that this transient effect is neglected too.
On the whole, the mathematical description is reduced to the one-axis 3rd-order model [7].
It consists of three differential Equations (1)–(3). The first is related to the dynamic electromagnetic
linkage between the rotor and stator. The others describe the electromechanical effect in the form of
the swing equation:
1 (xd − xd 0 )
U= 0 Uf − Ir , (1)
Td0 s + 1 Td0 0 s + 1
dω
Pm − Pe = Jω0 , (2)
dt
dϑ
ω= (3)
dt
U denotes the internal voltage source of the synchronous generator. It is composed of one part
related to the excitation voltage Uf and another part related to the reactive current Ir . Both effects
occur with a delay of the open-circuit time constant (Td0 0 ). The first part reflects the reference transfer
behavior of the excitation voltage Uf to the synchronous generated voltage

1
G(s) = 0 , (4)
Td0 s + 1

whereas the second term belongs to the disturbance transfer function and contributes to the internal
reactance [12] (p. 6):
(xd − xd 0 )
xd ( s ) = + xd 0 . (5)
Td0 0 s + 1
The interpretation is, that in the first moment after a current step, only the transient reactance
xd 0 is active. During the process, the reactance decays to the synchronous value xd . This transition is
illustrated in Figure 1.
The angle of the synchronously generated voltage U is determined by the swing
Equations (2) and (3). Together with the electromagnetic Equations (1), the block diagram of the
synchronous machine, shown in Figure 2, is established. In summary, it results in a voltage phasor
that retardedly follows the grid voltage phasor in relation to both the amplitude (according the coil
excitation time constant) and the phase angle (according to the mass inertia time constant of the rotor).
Energies 2020, 13, x FOR PEER REVIEW 4 of 22

Energies
Energies2020,
2020,13,
13,x xFOR
FORPEER
PEERREVIEW
REVIEW 4 4ofof2222
Energies 2020, 13, 2589 4 of 21

0
x0d0′
xdx′d′

xd
Td ‘ t
xdxd
TT
d d‘ ‘
Figure 1. Transient evolution of the impedance t
after a current step (derived
t
from [13] (p. 244)).
Figure
Figure Transient
1.1.Transient
Figure1. evolution
Transientevolution ofofthe
evolutionof the impedance
impedanceafter
theimpedance afteraa acurrent
after currentstep
current step(derived
step (derivedfrom
(derived from[13]
from [13](p.
[13] (p.244)).
(p. 244)).
244)).
U

I U
+U U
- 1
xd − xd ′
II 1 + Td0 ′s UU
- - ++ 11
0 xdxd−−xdx′d ′
11++Td0 ′s′s
Td0
00
Pe
Pm xdPP
+ e e- + ϑ
1 +
PP
mm xd+x- d - -
+ Jω0 s ++ s ϑϑ
11 ++
-- JωJω
00 ss ss t
tt

Figure 2. Block diagram of the synchronous machine (3rd-order model).

Figure 2. Block diagram of the synchronous machine (3rd-order model).
Figure
Figure2.2.Block
Blockdiagram
diagramofofthe
thesynchronous
synchronousmachine
machine(3rd-order
(3rd-ordermodel).
model).
This block representation of the synchronous machine is taken as the basis for subsequent
This block representation of the synchronous machine is taken as the basis for subsequent
comparisons to grid-forming inverters. Therein, ∆f represents the frequency deviation from the
This
Thisblock
comparisons to representation
block representation
grid-forming ofofthe
inverters. Therein, ∆fmachine
thesynchronous
synchronous machine isistaken
represents taken
the asasthe
frequency thebasis
basisfor
forsubsequent
deviation subsequent
from the
nominal value f0 and ϑ is the rotation angle. The damping parameter D represents the mechanical
comparisons
comparisons toto grid-forming
grid-forming inverters.
inverters. Therein,
Therein, ∆f
∆f represents
represents the
the
nominal value f0 and ϑ is the rotation angle. The damping parameter D represents the mechanical frequency
frequency deviation
deviation from
fromthethe
damping according to friction and windage.
nominal
nominalvalue
damping valuef0f0and
according toϑfriction
and ϑisisthe rotation
theand
rotation angle.
windage.angle.The
Thedamping
dampingparameter
parameterDDrepresents representsthe themechanical
mechanical
The equivalent circuit is depicted in Figure 3. The two circuits of the direct and quadrature axis
damping
damping according
according to
tofriction
friction and
and windage.
windage.
The equivalent circuit is depicted in Figure 3. The two circuits of the direct and quadrature axis are
are condensed to one complex circuit, which also serves as basis for the following elaborations. It
The
Theequivalent
condensed equivalent
to one complexcircuit isisdepicted
circuitcircuit,
depicted
which ininFigure
Figure
also 3.3.The
serves The two
twocircuits
as basis circuits
for ofofthe
the following thedirect
direct and
andquadrature
quadrature
elaborations. axis
axis
It consists
consists of an adjusted voltage source U behind an internal inductance xi. For the whole paper, the
are
ofare
ancondensed
condensed totoone
adjusted voltage onecomplex
complex
source U circuit,
circuit,which
behind anwhich also serves
servesasasbasis
alsoinductance
internal basis
xi . Forfor
for the
the the following
whole following
paper,elaborations.
elaborations.
the generator ItIt
generator convention is adopted [14] (pp. 20–27) as commonly used for power sources.
consists
consists of
ofan
an adjusted
adjusted voltage
voltage source
source U U behind
behind ananinternal
internal
convention is adopted [14] (pp. 20–27) as commonly used for power sources. inductance
inductance x x
i. i.For
Forthe
thewhole
whole paper,
paper, the
the
generator
generatorconvention
conventionisisadoptedadopted[14] [14](pp.
(pp.20–27)
20–27)asascommonly
commonlyused usedfor forpower
powersources.
sources.
xd(s) xq x1i = xd I
Ir Ia
xdx(s)
d(s) xqxq x1ix1=
i =xdxd I I
IrIr IaIa

(xd − xd ′)
Uf ·G(s) 1 U = G(s)Uf − Ir ∙ ejϑ
Td0 ′
(x(x −−xdx′)
d ′)
11 UU== G(s)U Udd I I ∙ e∙ jϑ
ejϑ
UU
f·
fG(s)
·G(s) G(s)Uf −
f −
Td0′ ′ r r
Td0
UU

real equivalent circuits for direct complex equivalent circuit

and quadrature axis
real
realequivalent
equivalentcircuits
circuitsfor
fordirect
direct complex
complexequivalent
equivalentcircuit
circuit
and Figure
andquadrature
quadrature axis3. Equivalent circuit of the synchronous machine.
axis
Figure 3. Equivalent circuit of the synchronous machine.

Figure
Figure3.3.Equivalent
Equivalentcircuit
circuitofofthe
thesynchronous
synchronousmachine.
machine.
 Energies 2020, 13, x FOR PEER REVIEW 5 of 22

Energies 2020, 13, x FOR PEER REVIEW 5 of 22

4. Grid-Forming Inverter Control Methods
Energies 2020, 13, 2589 5 of 21
4. Grid-Forming Inverter Control Methods
4.1. Droop-Based Grid-Forming Control Methods (e.g. Selfsync)
4. Grid-Forming
4.1. Droop-Based Inverter Control
The idea ofGrid-Forming
synchronizing ControlMethods
Methods
parallel static(e.g. Selfsync)
inverters by means of power droops reaches back to
[15] (see [16]). Following the operational principle of conventional power plants, f(P)- and U(Q)-
4.1.The idea of synchronizing
Droop-Based parallelMethods
Grid-Forming Control static inverters by means of power droops reaches back to
(e.g. Selfsync)
droops with reversed relation were exploited (see Figure 4). Therein, quantities f0 and U0 denote the
[15] (see [16]). Following the operational principle of conventional power plants, f(P)- and U(Q)-
nominalThe values.
idea of synchronizing parallel
In a similar way, static inverters
reference by means
[17] deals of power
with the droops operation
stand-alone reaches back
of to [15]
static
droops with reversed relation were exploited (see Figure 4). Therein, quantities f0 and U0 denote the
(see [16]). Following the operational principle of conventional power plants, f(P)-
inverters and suggests a method in which a voltage and frequency drop behavior in relation to the and U(Q)-droops
nominal values. In a similar way, reference [17] deals with the stand-alone operation of static
with reversed
output reactiverelation werepower
and active exploited (see Figure
is imposed. The4).purpose
Therein,isquantities
to achievef0load
and sharing
U0 denote the nominal
among power
inverters and suggests a method in which a voltage and frequency drop behavior in relation to the
values. without
sources In a similar way,communication.
explicit reference [17] deals with in
Likewise, the[18]
stand-alone
droops are operation
deployedofinstatic inverterswith
combination and
output reactive and active power is imposed. The purpose is to achieve load sharing among power
suggests a method in which a voltage and frequency drop behavior in relation to the
a virtual flux control. In the meantime, a multitude of droop-based approaches have been published output reactive
sources without explicit communication. Likewise, in [18] droops are deployed in combination with
and active power is imposed. The purpose is to achieve load sharing among power sources without
[19].
a virtual flux control. In the meantime, a multitude of droop-based approaches have been published
explicit communication. Likewise, in [18] droops are deployed in combination with a virtual flux
[19].
control. In the meantime, a multitude of droop-based approaches have been published [19].
U f

U f

Q P
U = U0 − k q (Q − Q0 ) Q = 0 − k p (P − P0 ) P
U = U0 − k q (Q − Q0 ) = 0 − k p (P − P0 )
Figure 4. Droop characteristics.
Droopcharacteristics.
Figure4.4.Droop
Figure characteristics.
Mostly, the output power is measured and low-pass filtered with the time constant (Te, Tm), in
orderMostly, the output
to eliminate powerdue
oscillations, is measured
to harmonics,andand
low-pass filtered
to obtain with the time
a more-smooth constant
reaction. (T , Tm ),
Afterwards,
Mostly, the output power is measured and low-pass filtered with the time constant (Te, Tme ), in
in order to eliminate oscillations, due to harmonics, and to obtain a more-smooth
the value is fed back to generate the internal inverter voltage. A reasonable improvement comparedreaction. Afterwards,
order to eliminate oscillations, due to harmonics, and to obtain a more-smooth reaction. Afterwards,
thethe
to value is fed
normal back control
droop to generate thegained
can be internalbyinverter
an anglevoltage. A reasonable
feedforward improvement
proportional compared
to the active powerto
the value is fed back to generate the internal inverter voltage. A reasonable improvement compared
the normal
flow, which droop control
was first can be in
introduced gained
[20] orby[21]
an and
angleis feedforward
referenced asproportional
Selfsync. We to the activeon
concentrate power
this
to the normal droop control can be gained by an angle feedforward proportional to the active power
flow, which
approach wasfollowing
in the first introduced in [20] or [21] and is referenced as Selfsync. We concentrate on this
elaboration.
flow, which was first introduced in [20] or [21] and is referenced as Selfsync. We concentrate on this
approach in the following
The related block diagramelaboration.
is illustrated in Figure 5. In the next step, this structure is rearranged
approach in the following elaboration.
The related
as depicted block diagram
in Figure is illustrated
6. It becomes apparent in Figure
that the5. In the next step,
functional this structure
principle is rearranged
is equivalent as
to that in
The related block diagram is illustrated in Figure 5. In the next step, this structure is rearranged
depicted
Figure 2. in Figure 6. It becomes apparent that the functional principle is equivalent to that in Figure 2.
as depicted in Figure 6. It becomes apparent that the functional principle is equivalent to that in
Figure 2.
f0
f0
P 1 - 2
kp
- Tm s + 1 s - ϑ
P 1 - 2
kp
- P0 Tm s + 1 s - ϑ
P0 kp'
kp'
Q 1 -
kq
- Te s + 1 - U
Q 1
kq
- Q0 Te s + 1 U
U0
Q0
Figure 5.
Figure Blockdiagram
5. Block diagramof
ofthe Selfsync
theSelfsync
U0 control algorithm.
control algorithm.

Figure 5. Block diagram of the Selfsync control algorithm.

Energies 2020, 13, 2589 6 of 21
Energies 2020, 13, x FOR PEER REVIEW 6 of 22

f0
P0

P - kp
- Tm s s - ϑ

1/kp kp'/kp

Q 1
kq Te s + 1
- U
kqQ0
U0
Figure 6. Rearranged block diagram of Selfsync control algorithm.
Figure 6. Rearranged block diagram of Selfsync control algorithm.

Likewise, a reactive power value is fed back to adjust the internal voltage (electro-magnetic force),
Likewise, a reactive power value is fed back to adjust the internal voltage (electro-magnetic
so that the droop parameter kq corresponds to (xd − xd 0 ). Consequently, the reactive power droop
force), so that the droop parameter kq corresponds to (xd − xd’). Consequently, the reactive power
can be interpreted as impedance evolution similar to that of the synchronous machine (see Figure 1):
droop can be interpreted as impedance evolution similar to that of the synchronous machine (see
In the first moment only the filter impedance of the inverter is active and due to the imposed droop,
Figure 1): In the first moment only the filter impedance of the inverter is active and due to the
the impedance in the reactive current path is adjusted to a higher stationary (synchronous) value.
imposed droop, the impedance in the reactive current path is adjusted to a higher stationary
The time constant Te is in accordance to the open-circuit time constant Td0 0 of the synchronous machine.
(synchronous) value. The time constant Te is in accordance to the open-circuit time constant Td0’ of
It should be noted that the reactive power is used here. Strictly speaking, the relation between the
the synchronous machine. It should be noted that the reactive power is used here. Strictly speaking,
reactive current and the voltage amplitude is not precisely linear here, due to voltage variations.
the relation between the reactive current and the voltage amplitude is not precisely linear here, due
Therefore, these conclusions are only valid around the nominal operating voltage. On the other hand,
to voltage variations. Therefore, these conclusions are only valid around the nominal operating
it is possible to normalize the droop parameter by the output voltage. By doing so, the reactive power
voltage. On the other hand, it is possible to normalize the droop parameter by the output voltage. By
droop changes to a reactive current droop.
doing so, the reactive power droop changes to a reactive current droop.
Moreover, the active power loops also resemble each other, so that Tm /kp is equivalent to the inertia
Moreover, the active power loops also resemble each other, so that Tm/kp is equivalent to the
time constant and 1/k is equal to the mechanical feedback damping parameter D. As an extension to
inertia time constant pand 1/kp is equal to the mechanical feedback damping parameter D. As an
the basic functionality, forward damping is achieved, due to the angle feedforward. This damping is
extension to the basic functionality, forward damping is achieved, due to the angle feedforward. This
much more effective against oscillation, as it bypasses the integrator. Consequently, a better stability
damping is much more effective against oscillation, as it bypasses the integrator. Consequently, a
and transient performance was observed when using such a technique. The damper windings have a
better stability and transient performance was observed when using such a technique. The damper
similar effect, since they inherently produce a counter torque according to the slip (deviation from the
windings have a similar effect, since they inherently produce a counter torque according to the slip
synchronous rotation) of the machine.
(deviation from the synchronous rotation) of the machine.
It is noteworthy, that with synchronous machines, system parameters such as time constants
It is noteworthy, that with synchronous machines, system parameters such as time constants
and reactance are subject to the mechanical construction. At best, the feedback damping parameter
and reactance are subject to the mechanical construction. At best, the feedback damping parameter
D of the synchronous machine can be increased when the power system stabilizer (PSS) is designed
D of the synchronous machine can be increased when the power system stabilizer (PSS) is designed
properly [8] (pp. 766–767). In contrast, in the case of grid-forming inverters, the system parameters
properly [8] (pp. 766–767). In contrast, in the case of grid-forming inverters, the system parameters
can be designed freely according to higher requirements.
can be designed freely according to higher requirements.
Values typically chosen for the Selfsync control algorithm can be found in Table 1. Except for the
Values typically chosen for the Selfsync control algorithm can be found in Table 1. Except for the
time constants, the values are given per unit with respect to inverter nominal power and voltage. Thus,
time constants, the values are given per unit with respect to inverter nominal power and voltage.
the control can easily be realized for different power ratings and voltage levels. The time constant Tm
Thus, the control can easily be realized for different power ratings and voltage levels. The time
is chosen according to the acceleration time of the grid (~10 s). Hence, it is a measure for the inertia
constant Tm is chosen according to the acceleration time of the grid (~10 s). Hence, it is a measure for
contribution in relation to the nominal power of the grid-forming unit. The time constant for the
the inertia contribution in relation to the nominal power of the grid-forming unit. The time constant
voltage amplitude is chosen lower. The droop parameters kp and kq mean that, at nominal power,
for the voltage amplitude is chosen lower. The droop parameters kp and kq mean that, at nominal
the frequency or voltage drop is 5% and 10%, respectively. Additionally, an adequate feedforward
power, the frequency or voltage drop is 5% and 10%, respectively. Additionally, an adequate
damping is achieved with a value around 5 pu.
feedforward damping is achieved with a value around 5 pu.
Energies 2020, 13, x FOR PEER REVIEW 7 of 22

Energies 2020, 13, 2589 7 of 21

Table 1. Exemplary parameterization of the Selfsync control algorithm.

Parameter Value
Table 1. Exemplary parameterization of the Selfsync control algorithm.
Tm kp·10 s
kp
Parameter 0.05
Valuepu
Tm Te kp1·10
s s
kkpq 0.1
0.05pupu
Tkep’ 5 1pu
s
kq 0.1 pu
kp ’ 5 pu
Although only one specific droop-based method is analyzed here, other known droop-based
approaches can be interpreted similarly, since the basic functional principle is the same. Nevertheless,
Althoughoronly
the damping one specific
superior droop-based
control features method
may differ, due is to
analyzed here, other
some additional known
control droop-based
loops.
approaches can be interpreted similarly, since the basic functional principle
In [22], an often-referenced approach concerning droop control is described. One basic element is the same. Nevertheless,
the damping
therein is the or superiormatrix
rotational control T.features
This wasmay differ, due
developed for to some additional
operation in resistivecontrol
grids loops.
in order to adapt
In [22], an often-referenced approach concerning droop control
the control action to the specific transfer function of the system. It means that not the measuredis described. One basic element
output
therein is the rotational
power values are applied matrix T.droops
to the This was butdeveloped
virtual valuesfor operation
resultingin resistive
from grids in in
the rotations order to adapt
the complex
the
planecontrol actionto
according to the
the R/X
specific
ratiotransfer function
of the power of the
line. As system.
a result,Itanmeans thatinnot
increase thethe measured output
effectiveness of the
power values are applied to the droops but virtual values resulting from
droops is achieved. However, this has a negative effect on the load sharing, since the droops control the rotations in the complex
plane
only theaccording
virtual to the R/X
values notratio of the
the real power
power line. As
values. a result,the
Moreover, an increase
consistency in the
witheffectiveness of the
the conventional
droops is achieved. However, this has a negative effect on the load sharing,
power system is lost, since the analogy to the synchronous machine is partially reversed. Therefore, since the droops control
only therecommendable
it is not virtual values not forthe real power
practical values. in
applications Moreover,
the mainthe grid.consistency with the conventional
power system
Other is lost,
authors since the analogy
supplement the droops to the
bysynchronous
differentiators machine
for theispurpose
partiallyofreversed.
improvedTherefore,
transient
itresponse
is not recommendable
[23,24] for practical applications in the main grid.
Other authors supplement the droops by differentiators for the purpose of improved transient
dP
response [23,24] = 0 − k p (P − P0 ) − k p,d , (6)
dPdt
f = f0 − kp (P − P0 ) − kp,d , (6)
dt dQ
U = U0 − k q (Q − Q0 ) − k q,d dQ ,
(7)
U = U0 − kq (Q − Q0 ) − kq,d dt, (7)
dt
When looking at Equation (6), it becomes clear that the effect of kp,d is similar to the angle
When looking
feedforward by kp’.atThe Equation
effect of(6),
kq,dit isbecomes
that theclear that the
reactance effect
of the of kp,dcurrent
reactive is similar
pathtoisthe angle
virtually
feedforward by k 0 . The effect of k is that the reactance of the reactive current path is virtually
further increased in p relation to the rate q,d of current change during a transient process. The consequence
further increased in relation to the
is that the time development of the reactance rate of current change
is not duringand
so smooth a transient process. The consequence
more dynamic.
is thatOther
the time development of the reactance is not so smooth and more
approaches combine the droop control with virtual impedance loops to improve load dynamic.
sharingOther approaches
or droop operationcombine theTherefore,
[25,26]. droop control with virtual
the underlying impedance
impedance loops tomay
parameters improve
differ,load
due
sharing or droop operation [25,26]. Therefore, the underlying
to the fast-acting additional loops, but the essential control structure is retained. impedance parameters may differ, due to
the fast-acting additional loops, but the essential control structure is retained.
4.2. Power Synchronization Loop
4.2. Power Synchronization Loop
A further technique to realize a grid-forming control is the so-called power synchronization
A further technique to realize a grid-forming control is the so-called power synchronization
control depicted in Figure 7 [27]. Additional insights can be found in [28]. Other authors improved
control depicted in Figure 7 [27]. Additional insights can be found in [28]. Other authors improved
this approach by orienting closer towards the synchronization mechanism of the swing equation and
this approach by orienting closer towards the synchronization mechanism of the swing equation and
include a further transfer function in order to match the dynamic order with an appropriate damping
include a further transfer function in order to match the dynamic order with an appropriate damping
capability [29].
capability [29].

ω0t U0

KI ϑ KE U
pref uref
s - 1 + TE s
-

p |upcc |

Figure 7. Power synchronization

Figure 7. synchronization control
control loop
loop according
according to
to [27].
[27].
 upcc

KE
TE s + 1
Li I PCC
jωLi Ia
Energies 2020, 13, 2589 jωLi Ir 8 of 21

U
With regard to the angle
U regulation, a strong similarity to the droop-based approaches becomes
U The power measurement filter and
obvious, because it also consists of aUdirect P/f-droop-characteristic.pcc
an angle feedforward are not mentioned, so that the appropriate parameters are 𝑎 assumed to be zero.
The voltage amplitude regulation consists of a cascade of a proportional voltage controller and
optionally of a superior reactive power controller. Due to the proportional voltage controller there
𝑗𝑟
is also a “droop” characteristic [27], but in the opposite direction. The drooping action stems from
the fact
. that the amplitude at the PCC is almost a linear function of the reactive current. For a better
understanding, refer to the phasor diagram in Figure 8.

uref
-
upcc

KE
TE s + 1
Li I PCC
jωLi Ia
jωLi Ir

U
U
U Upcc
𝑎

𝑗𝑟
Figure 8. Schematic voltage control in the power synchronization control loop.
Figure 8. Schematic voltage control in the power synchronization control loop.
It can be seen that the voltage component on the horizontal r-axis only has a negligible impact on
It can
the total be seen that
amplitude. Thethe voltage
main component
effect on the horizontal
on the amplitude is broughtr-axis
aboutonly hasa-axis
by the a negligible
voltageimpact
drop,
on the results
which total amplitude. The main
from the reactive effect on
current. the amplitude
Consequently, theisfeedback
brought loopabout the a-axisthe
byregulating
for voltage drop,
amplitude
which resultstofrom
corresponds the reactive
an impedance current. Consequently,
reduction, opposite to thatthe of feedback
a Q/U-droop. loopThe
for regulating
parameter the amplitude
KE defines the
corresponds
negative gaintofor anthe
impedance
impedance reduction,
Li, so thatopposite to that
in a steady of athe
state, Q/U-droop.
new valueThe i /(1+KE ))Kand
parameter
is (ωL E defines
TE is
Energies
the 2020, 13,
negative
responsible for x the
gain FOR
forPEER
the REVIEW
impedance
continuous Li, sowith
time delay that which
in a steady state, the new
the impedance value isThis
is adapted. (ωLcani/(1+K
beE)) 9 ofT22
and
deduced E

is responsible
from the transfer forfunction
the continuous
from thetime delaycurrent
resulting with which
flow tothe theimpedance is adapted.
provided voltage drop.This
The can
closedbe
closed voltage
deduced
voltage from the
amplitude amplitude
transfer
loop canloop can from
function
therefore therefore
be the be restated
resulting
restated (Figure
current
(Figure flow
9) under 9)to
under
the the the assumption
provided
assumption voltage that
drop.
that Ia has Ia The
onlyhas a
only small
very a veryinfluence
small influence on the voltage
on the voltage amplitude amplitude at theofpoint
at the point commonof common
coupling.coupling.

U0
Ir ωLi K E U
TE s + 1 + K E

Figure 9. Restated voltage–amplitude control loop.

Figure 9. Restated voltage–amplitude control loop.
4.3. Voltage Controlled Inverter (VCI)
4.3. Voltage ControlledtoInverter
This approach control(VCI)
grid-forming inverters was published in [30]. The concept will be
explained with the resulting dynamics
This approach to control grid-forming of the voltage
invertersangle
wasand amplitude.
published For the
in [30]. Thefrequency
concept control
will be
loop, the following
explained with the control structure
resulting shown
dynamics involtage
of the Figure 10, wasand
angle considered.
amplitude. TheForcontroller itself consists
the frequency control
of a PI-controller and can be optionally extended by a primary control loop as it is realized by frequency
loop, the following control structure shown in Figure 10, was considered. The controller itself consists
droops. A transformation
of a PI-controller and canofbetheoptionally
control algorithm
extended can
bybeafound in Figure
primary control11.loop as it is realized by
frequency droops. A transformation of the control algorithm can be found in Figure 11.

p f0

- KI ∆fi 2𝜋 ϑ
pref + KP
s 𝑠
-
 This approach to control grid-forming inverters was published in [30]. The concept will be
loop, the following control structure shown in Figure 10, was considered. The controller itself consists
explained with the resulting dynamics of the voltage angle and amplitude. For the frequency control
of a PI-controller and can be optionally extended by a primary control loop as it is realized by
loop, the following control structure shown in Figure 10, was considered. The controller itself consists
frequency droops. A transformation of the control algorithm can be found in Figure 11.
of a PI-controller and can be optionally extended by a primary control loop as it is realized by
frequency
Energies 2020, droops.
13, 2589 A transformation of the control algorithm can be found in Figure 11. 9 of 21
p f0

p- KI ∆fi f0 2𝜋 ϑ
pref + KP
s 𝑠
- ∆fi
- KI 2𝜋 ϑ
pref + KP
s 𝑠
-
KPrim

KPrim
Figure 10. Frequency controller of the voltage controlled inverter VCI according to [30].
Figure 10. Frequency controller of the voltage controlled inverter VCI according to [30].
Figure 10. Frequency controller of the voltage controlled inverter VCI according to [30].
p K P K prim p
f0 f0
∆fi ∆fi
pref p- - K P K prim KI
f0 2𝜋 ϑ pref p- KI f0 2𝜋 ϑ
s 𝑠 (1 + K P K prim )s 𝑠
- -
- - KI ∆fi 2𝜋 ϑ - KI ∆fi 2𝜋 ϑ
pref pref
s K𝑠p (1 + K P K prim )s K𝑠p
- -
KPrim Ki
KPrim Ki
Kp Kp
KPrim Prim K
Figure 11. Transformation
Ki of the VCI control structure. Ki
Figure 11. Transformation of the VCI control structure.

This transformation yields Figurethe11. same structure forthe

Transformation theVCI
frequency loop as in Figure 6. In addition to
This transformation yields the same structureoffor control structure.
the frequency loop as in Figure 6. In addition
the swing equation, it also includes the feedforward damping (realized by Kp /2πKi ) equally to that of
to the swing equation, it also includes the feedforward damping (realized by Kp/2πKi) equally to that
the Selfsync approach. Here,
This transformation K theequals
yields the feedback
same structure for thedamping
frequency factorloop D and
as in(1+K P KPrim
Figure 6. In)/K I is the
addition
of the Selfsync approach. Here,Prim KPrim equals the feedback damping factor D and (1+K PKPrim)/KI is the
inertia time constant.
to the swing equation, it also includes the feedforward damping (realized by Kp/2πKi) equally to that
inertia time constant.
of theThe voltage–amplitude
Selfsync approach. Here,regulation KPrim equals loopthe is shown damping
feedback below. With factorthe D andvoltage
(1+KPcontrol
KPrim)/KImode,
is the
The voltage–amplitude regulation loop is shown below. With the voltage control mode, the
the dynamical
inertia regulation of the voltage amplitude again is deduced by following the procedure
time constant.
dynamical regulation of the voltage amplitude again is deduced by following the procedure revealed
revealed
The in the power synchronization
voltage–amplitude regulationcontrol
loop is (see Figure
shown 8).
below. With the voltage control mode, the
in the power synchronization 0 control (see Figure 8).
Based regulation
dynamical on the assumption that only
of the voltage the reactive
amplitude again current
is deduced component
by following has athe considerable
procedure effect
revealedon
Based on the assumption that only the reactive current component has a considerable effect on
the voltage
in the power amplitude,
synchronization 0
a transfer
xd′ function
control (see from
Figure the reactive
8). current to the voltage amplitude is set up
the voltage amplitude, a transfer function from the reactive current to the voltage amplitude is set up
(Figure 12). on
Based Thethe effect of xthe
assumption PI controller
that only the results in ancurrent
reactive impedance reduction
component has according to the principle
a considerable effect on
(Figure 12). The effect of the d′ PI controller results in an impedance reduction according to the principle
in
theFigure
voltage
Energies 13.amplitude,
2020, 13,Due
x FORtoPEER
thea integrator,
transfer the effect
REVIEW function from ofthe
thereactive
internalcurrent
reactance in the
to the reactive
voltage currentispath
amplitude set is22
up
10 of
in Figure 13. Due to the integrator, the effect of the internal reactance in the reactive current path is
completely
(Figure 12). reduced
The
Energies 2020, 13,
to
effect
x FOR PEER
zero
of the during
REVIEWPI the
controllercontrol
results action.
in an In contrast
impedance to the synchronous
reduction according machine
to the (compare
principle
10 of 22
completely reduced to zero during the control action. In contrast to the synchronous machine
Figure
in Figure1), it
13.means
Due to that
thethe inner impedance
integrator, the effect is of
notthe
increased
internalduring
reactance the intransient time, but
the reactive eliminated.
current path is
(compare Figure 1), it means u xd that the inner impedance U0 is not increased during the transient U0 time, but
completely reduced to zero during the control action. In contrast to the synchronous machine
eliminated.u u - T ‘ t
(compare Figure 1), it means
ref x d that
TI
d U
the inner impedance
0 U I
is not increased
r ωLduring
U
the transient
0 U time, but
i (TI + sTP )
- + TP
Td ‘ t
eliminated. uqref s
U Ir (T P + 1)s + T P U
ref TI ωLi (TI + sTP )
- + TP
qref s (TP + 1)s + TP
q
-
q
Figure 12. Voltage control loop of the VCI.
Figure
Figure 12. Voltage control
12. Voltage control loop
loop of
of the
the VCI.
VCI.

0
0

xd′
xd′
t
Figure 13.Elimination
Figure13. Eliminationof
ofthe
theimpedance
impedanceby
bythe
theintegrator
integratoraction
actionwith
t appropriatetime
withthe
theappropriate timeconstant.
constant.
Figure 13. it
Basically, Elimination of thethat
can be stated impedance bycontroller
a voltage the integrator action
of the with the
terminal appropriate
voltage time constant.
eliminates the internal
Basically, it can be stated that a voltage controller of the terminal voltage eliminates the internal
impedance within the time constant of the PI-controller, whereas a reactive power controller corresponds
impedance within
Basically, it can the time that
be stated constant of the
a voltage PI-controller,
controller whereasvoltage
of the terminal a reactive powerthecontroller
eliminates internal
to an increase of the reactance value in the reactive current path.
corresponds to an increase of the reactance value in the reactive current path.
impedance within the time constant of the PI-controller, whereas a reactive power controller
corresponds to an increase of the reactance value in the reactive current path.
4.4. Virtual Synchronous Machine (VSM)
4.4. Virtual Synchronous
Directly emulatingMachine (VSM)
the synchronous machine is a straightforward approach to impart grid-
forming capabilities to inverters. Multiple control is
Directly emulating the synchronous machine algorithms have beenapproach
a straightforward proposedto[31–33],
impart which
grid-
carry this approach in their name. They differ in the depth of emulation and amount of additional
Energies 2020, 13, 2589 10 of 21

4.4. Virtual Synchronous Machine (VSM)

Directly emulating the synchronous machine is a straightforward approach to impart grid-forming
capabilities to inverters. Multiple control algorithms have been proposed [31–33], which carry this
approach in their name. They differ in the depth of emulation and amount of additional current and
voltage control, but they all include some form of the swing Equation (2). Reference [34] gives an
overview of some of the published approaches and proves the equivalence of VSM and droop control
with regards to small signal stability [35].
Starting with [36], the concept of a virtual synchronous machine called VISMA was applied.
In this early implementation, the inverter control imposes an exact current response according to a
synchronous machine model of seventh order. The output voltage is used to feed the machine model
and calculate the current response in real time. By means of a hysteresis comparator, a fast adjustment
of the current is achieved. In [37] the machine model was simplified for the purpose of robustness and
stability and the suitability of island operation was shown.
Another well-known virtual synchronous machine concept is referenced as synchronverter [33].
The idea is again to implement the synchronous machine behavior in inverters. The control law
implemented on the inverter is composed of the dynamic swing equation and algebraic equations
for coupling the virtual rotor and stator side. In this, the damping of the virtual mass inertia is solely
performed by feedback of the difference between VSM and reference frequency. Thus, it incorporates a
frequency-power droop in terms of frequency containment reserve (previously referred to as “primary
control reserve”) [38] and cannot be designed freely. In a further development [39], an additional loop
is exploited from the virtual flux to the active power in order to improve the damping and a defined
dynamic response.
Another detailed realization of a virtual synchronous machine can be found in [40]. Here, the swing
equation is supplemented by a frequency droop. In addition, a more appropriate damping is performed
by a feedback of the difference between the grid and the VSM frequency. Since the grid frequency is
provided by a phase locked loop (PLL), it suffers from a minor delay. For interested readers, several
subsidiary controls were added in that paper, e.g., current control or virtual impedance. However,
they are more related to high-transient electromagnetic effects or internal impedance correction, both of
which are not the focus of this paper.
Since all VSM concepts utilize the swing equation, the damping of the virtual mass inertia is an
important criterion. A compilation of different damping strategies for virtual synchronous machines is
given in [38].
As an illustrative example for VSM control methods, the VSM structure applied in [32] is derived
in this paragraph. Firstly, the swing Equation (2) is expanded by damping power

dδm D
PD = DM = ω, (8)
dt 2π
which accounts for mechanical loss due to friction and windage (as in [9] (pp. 125–129)). This gives:

D dω
Pm − Pe − PD = Pm − Pe − ω = Jω0 , (9)
2π dt
which is also shown in Figure 2. The damping coefficient D is a property of the physical construction
of the synchronous machine and thus fixed. In VSM control, this parameter can be chosen freely. As it
directly correlates active power P and frequency f, it is used to define the slope of the P(f)-characteristic
of the VSM. It is renamed Kpf in order to reflect that change.
In the next step, J, ω0 and 2π are condensed into a single time constant Ta = 2H and a forward
damping constant Kdf is introduced, which can be dimensioned to effectively dampen oscillations
between generators.
 of the synchronous machine and thus fixed. In VSM control, this parameter can be chosen freely. As
it directly correlates active power P and frequency f, it is used to define the slope of the P(f)-
characteristic of the VSM. It is renamed Kpf in order to reflect that change.
In the next step, J, ω0 and 2π are condensed into a single time constant Ta = 2H and a forward
damping
Energies 2020,constant
13, 2589 Kdf is introduced, which can be dimensioned to effectively dampen oscillations
11 of 21
between generators.
In the Q-Branch, Ku is chosen as the slope of the Q(U)-characteristic and Tu as the excitation time
In the Q-Branch, K is chosen as the slope of the Q(U)-characteristic and T as the excitation time
constant. The resultingudiagram of the control scheme is shown in Figure 14. uIt is quite similar to the
constant. The resulting diagram of the control scheme is shown in Figure 14. It is quite similar to the
rearranged Selfsync control shown in Figure 6.
rearranged Selfsync control shown in Figure 6.

p q U0
- 1
f ϑ - 1 U
pref qref Ku
- Ta s s 1 + Tu s
f0 -

K pf K df

Figure 14. Virtual

Figure14. Virtualsynchronous
synchronousmachine
machinecontrol.
control.

4.5. Virtual Oscillator Control (VOC)

4.5. Virtual Oscillator Control (VOC)
The virtual oscillator control (VOC) differs from the previously mentioned approaches, as it is not
The virtual oscillator control (VOC) differs from the previously mentioned approaches, as it is
based on phasor
not based representation.
on phasor It is aIt sinusoidal
representation. timetime
is a sinusoidal domain
domainimplementation
implementation that that
is related to the
is related to
synchronization principle of coupled oscillators in complex networks [41]. The
the synchronization principle of coupled oscillators in complex networks [41]. The method uses method uses a Van dera
Pol
Vanoscillator
Energies 2020,
der with
x FORa PEER
nonlinear
Pol13,oscillator REVIEW
with differential
a nonlinear equation to
differential generatetothe
equation virtualthe
generate oscillator
virtual[42]. The principle
oscillator 12 of
[42]. 22
The
isprinciple
illustrated in Figure 15.
is illustrated in Figure 15.

+
αuc3 κi i
uc -1/σ L C uc

-
κi
κv κvε

U
[cos φ, -sin φ] PWM

Li11 Li2 I Ri

Ci

Figure 15. Illustration of the virtual oscillator circuit according to [43].

As a feedback parameter, only the output current is applied, which is used to feed the virtual
oscillator circuit. The parameters κii and κvv serve serve as
as scaling
scaling factors for the interface between the real
hardware and the virtual oscillator. The inductance and the capacitance of the oscillating circuit are
chosen in such a way that the resonant
resonant frequency
frequency matches matches the
the nominal
nominal frequency
frequency ω
ω00.
In [43] the dynamics of the VOC VOC is transferred into phasor representation, which makes it
is transferred into phasor representation,
comparable to phasor-based approaches. The The averaged
averaged dynamics
dynamics of of the
the VCO
VCO is captured by the
following equations:
dU . σ

β 3

κ κv
= U = σ U − βU 3 − κii κv Q,
dU (10)
dt =U= ̇ 2C 2
(U − U ) − 2CU Q, (10)
dt C CU
dϑ . κ κv
= ϑ = ω = ω0 − i 2 P, (11)
dtdϑ κi κv
2CU
= ϑ̇ = ω = ω0 − P, (11)
 −1 dt CU2
with β = 3α κ2v σ -1 and Φ = π/2. For this reason, a P/f- and Q/U-characteristic is achieved. Compared
2
with β = 3α(κv σ) and Φ = π/2. For this reason, a P/f- and Q/U-characteristic is achieved. Compared
to the 3rd-order model of the synchronous machine (or Selfsync for example), the state variable, which is
to the 3rd-order model of the synchronous machine (or Selfsync for example), the state variable, which
related to the mass inertia, does not exist.
is related to the mass inertia, does not exist.
For further analysis of the voltage amplitude loop, the model Formula (10) is linearized:
σ 3β 2 κi κv Q0 κi κv
∆U̇ = − U0 − 2 − 1 ∆U − CU ∆Q,
C CU0 0
(12)
K Z
Energies 2020, 13, 2589 12 of 21

For further analysis of the voltage amplitude loop, the model Formula (10) is linearized:
!
. σ 3β 2 κi κv Q0 κ κv
∆U = − U0 − − 1 ∆U − i ∆Q, (12)
2C 2 2CU0 2 2CU0
| {z } |{z}
K Z

with the nominal voltage U0 and the nominal reactive power Q0 . The block diagram of the VOC as the
averaged model is shown in Figure 16. The linearized Q/U-loop reduces to a PT1 element similar to
the electromagnetic behavior of the synchronous machine. In this way, 1/K corresponds to (xd − xd 0 )
and Z/K corresponds to the time constant Td0 0 .
As already mentioned above, the state variable that correlates with the frequency is not included
in this method. Thus, the time constant is set to zero. The resulting P/f-control loop achieves a similar
effect as it is done by the angle feedforward (feedforward damping), which is known from Selfsync or
some virtual synchronous machine and voltage-controlled approaches.
In steady state, the accordance of droops (compare Figure 4) can be easily shown, which is:
κi κv
kp = , (13)
2CU0 2

Energies 2020, 13, x FOR PEER REVIEW 2CU0 κi κv 13 of 22

kq = 4
, (14)
− 2CU0 2 )
σ(3βU0 C − κi κv Q0
However, it is important to note that the inertia time constant equals zero. As a result, an
However, it is important to note that the inertia time constant equals zero. As a result,
autonomous voltage phasor is formed, but it is not retarded, so it does not provide sufficient inertial
an autonomous voltage phasor is formed, but it is not retarded, so it does not provide sufficient
support.
inertial support.

f0 U0

𝜅𝑖 𝜅𝜐 - ∆ϑ 𝑍/𝐾 - U
2𝜋 ∆Q
∆P 1/𝐾 ∙ 𝑠 + 1
0 ∙ 𝑠 + 2𝐶𝑉0 𝑠 ∆U
∆f
Figure 16. Block diagram of the VOC in averaged representation.
Figure 16. Block diagram of the VOC in averaged representation.
4.6. Matching Control
4.6. Matching
The core Control
of matching control is the duality of the converter DC voltage and the generator angular
velocity
The[44,45]. A good explanation
core of matching of duality
control is the this concept
of theisconverter
given in [46], which is
DC voltage taken
and as the basis
the generator in the
angular
following.
velocity A further,
[44,45]. A goodmore analyticalof
explanation elaboration
this conceptcanisbe found
given in [47].
in [46], which is taken as the basis in the
The idea of this concept is to treat the physical dc-link capacitor
following. A further, more analytical elaboration can be found in [47]. as storage in the same way as the
moment
The of inertia
idea does.
of this For this
concept is topurpose,
treat thethe dc-capacitor
physical dc-linkvoltage
capacitoris used to adjust
as storage the same
in the frequency of
way as
the ac-converter bridge. The control law is given by [46]:
the moment of inertia does. For this purpose, the dc-capacitor voltage is used to adjust the frequency
of the ac-converter bridge. The control law is givenby [46]: 
s + KT 2
 
ref 2
ω = ωg + s + KT ( UDC ) − U DC 2 , (15)
ω = ωg +KJ s + KD [(UDC )2 − (UDC ref
) ], (15)
KJs + KD
where ω is the output frequency, ωg is the nominal or grid frequency and UDC is the dc-link voltage.
where ω is the
The control lawoutput frequency,
incorporates bothωthe
g is the nominal or grid frequency and UDC is the dc-link voltage.
dc-voltage regulation and the synchronization to the grid.
The control
The overall functional principle dc-voltage
law incorporates both the is depictedregulation
in Figure and the synchronization
17. The to the grid.
control law in Equation (15) is
The overall functional principle is depicted in Figure
supplemented by (U)Q-droop, in order to regulate the amplitude. 17. The control law in Equation (15) is
supplemented by (U)Q-droop, in order to regulate the amplitude.

Pin Pout
Li I
UDC 1 PCC
Eqn. (15) ϑ
s

DC-Link Uejϑ
QE U0
Q0 - U
KQ
 K s + KD
where ω is the output frequency, ωg is theJ nominal or grid frequency and UDC is the dc-link voltage.
The ω
where control
is the law incorporates
output frequency,both the
ωg is dc-voltage
the regulation
nominal or and theand
grid frequency synchronization to thevoltage.
UDC is the dc-link grid.
The law
The control overall functionalboth
incorporates principle is depicted
the dc-voltage in Figure
regulation and17. The
the control law intoEquation
synchronization the grid. (15) is
supplemented by (U)Q-droop,
The overall functional in order
principle to regulate
is depicted the amplitude.
in Figure 17. The control law in Equation (15) is
supplemented
Energies 2020, 13,by (U)Q-droop, in order to regulate the amplitude.
2589 13 of 21

Pin Pout
Pout Li I
Pin UDC 1 PCC
Eqn. (15) ϑ Li I
UDC 1 s PCC
Eqn. (15) ϑ
s
DC-Link Uejϑ
DC-Link QE U0 jϑ
Ue
Q0 QE - U0 U
KQ
Q0 - U
KQ

Figure 17. Functional principle of matching control according to [46].

Figure 17. Functional principle of matching control according to [46].
Figure 17. Functional principle of matching control according to [46].
Together with the power equation of the dc-capacitor, a relationship between power and voltage
Together with the power equation of the dc-capacitor, a relationship between power and voltage
phasor can with
Together be established
the power (Figure
equation18). It describes
of the the dynamics
dc-capacitor, withbetween
a relationship which the voltage
power phasor is
and voltage
phasor can be established (Figure 18). It describes the dynamics with which the voltage phasor is
adjusted.
phasor The
can be parameter(Figure
established KJ is the
18).inertia emulation
It describes coefficient,
the dynamics KT which
with is responsible for the
the voltage DC-link
phasor is
adjusted. The parameter KJ is the inertia emulation coefficient, KT is responsible for the DC-link
voltage The
adjusted. tracking and KDKisJ is
parameter thethe
damping
inertia coefficient.
emulation coefficient, KT is responsible for the DC-link
voltage tracking and KD is the damping coefficient.
voltage tracking and KD is the damping coefficient.
f0 U0
f0 U0
ω0 (s + K T ) - 2 ∆ϑ U
∆Pout ∆Q KQ
ω0C(s s(K
DC+ K Tj s) + K D ) - 2
s ∆ϑ
KQ ∆U U
∆Pout ∆Q
CDC s(K j s + K D ) s ∆U
Figure 18. Voltage phasor dynamics of matching control.
Figure 18. Voltage phasor dynamics of matching control.
4.7. PLL-Based ModifiedFigure 18. Voltage phasor
Current-Controlled dynamics of matching control.
Methods
4.7. PLL-Based Modified Current-Controlled Methods
There isModified
4.7. PLL-Based an ongoing discussion inMethods
Current-Controlled the scientific community regarding the need for a phase
locked loop (PLL) in grid-forming inverter types. Typically, grid-following inverters are associated
with a phase locked loop [48], whereas grid-forming control methods are assumed not to apply a
PLL [45], [49] or to apply it only for initial grid synchronization [27]. Another reference proposes
an improved grid-forming control approach by using frequency and angle measurements through a
PLL [50]. For parallel operation, all grid-forming units should incorporate some kind of synchronization
mechanism. The use of a PLL is therefore not necessarily a criterion against the grid-forming capability.
Even more, recent research showed that a power grid can be operated solely with an inverter using an
enhanced current-controlled control scheme [51]. Thus, we cannot say generally that grid-forming
units do not apply a PLL. This section will further analyze these aspects. It introduces the main concept
of using a PLL in order to provide an inherent inertial response with current-controlled inverters and
reports other approaches that make use of this concept.
The VSYNC project aimed to enable power system stability enhancement through inverters in
combination with short-term energy storage, providing an inertial-like active power response [3,52,53].
There, a grid-following control algorithm was equipped with a suitably parameterized PLL in order
to calculate an additional power set point for the inverter in analogy to a combination of the swing
Equation (2) in normalized notation and the transmission Equation (16). With Pe being the electrical
power between two voltage sources with amplitude U,E and angle ϑ1,2 over the inductive impedance Xi

U·E
Pe = sin(ϑ2 − ϑ1 ), (16)
Xi

the combination of both equations results in:

dω U·E
Jω0 = Pm − sin (ϑ2 − ϑ1 ). (17)
dt Xi | {z }
dϑ
 U∙E
Pe = sin(ϑ2 − ϑ1 ), (16)
Xi
the combination of both equations results in:
dω U∙E
Energies 2020, 13, 2589 Jω0 = Pm − sin (ϑ2 − ϑ1 ). 14 of 21
(17)
dt Xi
dϑ

With the PLL structure shown in Figure 19, while neglecting Kp and assuming the mechanical
With the PLL structure shown in Figure 19, while neglecting Kp and assuming the mechanical power
power Pm in Equation (17) to be zero, the similarity of the PLL structure and Equation (17) becomes
Pm in Equation (17) to be zero, the similarity of the PLL structure and Equation (17) becomes visible:
visible:
dωdω 11 U·E
U ∙ Esin(dϑ).
= −Ki KKd U1 sin(dϑ) = − (18)
dt dt = −K i d U1 sin(dϑ) = −Jω0 Xi
Jω0 Xi
sin(dϑ). (18)

By
By choosing
choosing the
the PLL
PLL parameters
parameters K
Kii, Kdd to

11 EE
Ki K=i = Jω; ; KdK=d = , , (19)
(19)
Jω0 0 XiXi
the PLL describes the same transient behavior as the mechanical part of a synchronous machine,
the PLL describes the same transient behavior as the mechanical part of a synchronous machine,
which is connected to a transmission line. The derivative of the frequency as inherent output of the
which is connected to a transmission line. The derivative of the frequency as inherent output of the
PLL without explicit differentiation is used as an additional active power setpoint in order to provide
PLL without explicit differentiation is used as an additional active power setpoint in order to provide
an inertial response [53].
an inertial response [53].

−Pref

dϑ2
ω2 =
Uejϑ1 Phase- Ki dt 1 ϑ2
Kd
detector s s

−U sin(ϑ2 − ϑ1 )
Kp

Figure
Figure 19. Phase locked
19. Phase locked loop
loop (PLL)
(PLL) structure
structure for
for calculation
calculation of
of an
an additional
additional active
active power
power setpoint
setpoint for
for
inertial response of inverters according to [53].
inertial response of inverters according to [53].

The difference to other df/dt-based inertial response controllers is the calculation of the frequency
derivative. Here, the frequency derivative is automatically determined within a PLL. Other approaches
on the basis of current-controlled control algorithms [54,55], measure the frequency with a PLL,
filter this measurement value with a low-pass, and formulate the derivative as, e.g., a multiplication
with the Laplace operator s. The latter approach is therefore delayed, compared to the direct PLL-based
method. Hence, the relationship between frequency derivative and power injection of df/dt-based
inertial response controllers does not initially follow the swing equation.
In [56] a similar PLL-based method has been recently introduced. In contrast to [3,52,53], the loop
filter has been equipped with a proportional gain Kp for additional damping and to improve the
frequency detection speed, see Figure 19. However, the design of the PLL gain factors follows the
same approach.
Therefore, assuming fast adjustment of the given active power setpoint, the inner angle of
both introduced direct PLL-based methods corresponds to the swing equation, and they behave in
the same way as synchronous machines in their frequency/active power relation. As a conclusion,
an active power/frequency relation can be deduced (see Figure 20), similar to synchronous machines or
grid-forming inverters.
However, similar to the synchronization torque in synchronous machines, the synchronization
principle of the PLL with its inherent relationship between power and angle can be used in a more
suitable sense. Assuming the provided voltage phasor is generated using the information (especially
the angle) from the PLL in the form of a voltage feedforward, by tuning the parameters of the PLL,
the dynamics of the provided voltage phasor can be chosen. As a result, a retarded following voltage
phasor can be created according to the slowness of the PLL. In inductive coupled grids, the way the
PLL-parameters work, can be derived again from Figure 20.
This relationship is implicitly exploited in modified current-controlled approaches, e.g., [57–60].
There, the inner PI current control loop has been additionally reduced to a proportional control.
Otherwise, the integrator will eliminate the retarded adjustment of the provided voltage phasor.
 Energies 2020, 13, x FOR PEER REVIEW 16 of 22
Energies 2020,13,
Energies2020, 13,2589
x FOR PEER REVIEW 15
16of
of21
22
According to the delayed (PT1) feedforward of the measured voltage, the provided voltage
According to the delayed
amplitude is a function of UPCC.(PT1)In turn, feedforward
the changesofinthe themeasured
amplitudevoltage,of UPCC the provided voltage
are determined by the
In general, the remaining proportional current controller acts as an additional resistive impedance
amplitude is a function of UPCC. In turn, the changes in the amplitude of UPCC are determined by the
in series and is comparable to the concept uref of a virtual impedance. It introduces a further damping
against electro-magnetic resonances. uref -
Furthermore, the amplitude of the- voltage feedforward isudelayed pcc in [60]. The quadrature-
component of terminal voltage is omitted, because the provided voltage
upcc phasor is completely projected
to the direct axis, which in turn is adjusted KE smoothly according to the PLL. Hence, a relationship
TEKsE+ 1
between reactive current and voltage amplitude can Li be Iderived, which describes jωLthe impedance
TE s + 1 PCC i Ia
evolution in time as for example seen in the synchronous machine.
I PCC The following considerations are in
Li jωLi Ir jωLi Ia
close correspondence to Figure 8. jωLi Ir
Under consideration of a small angle difference ϑ, the circuit equation yields for theU(reactive)
current through the impedance Xi : U U
U
UUPCC = U − Xi Ir , U pcc
U Upcc (20)
𝑎
assuming UPCC being the measured terminal voltage amplitude and U being the inverters 𝑎
voltage amplitude.
According to the delayed (PT1) feedforward of the measured voltage, the provided 𝑗𝑟 voltage
amplitude is a function
reactive current flow, as seen of U PCCin. In turn, the changes in the amplitude of U PCC are determined 𝑗𝑟 by the
reactiveAscurrent
reactive current
a result, flow,
the as
flow, asseen
seenininFigure
amplitude of the 8. internal provided voltage of the inverter can be formulated in
As
As aa result,
result, the
the amplitude
amplitude
relation to the reactive current flow: ofof the
the internal
internal provided
providedvoltage
voltageof ofthe
theinverter
invertercan canbe beformulated
formulatedin in
relation to the reactive current
relation to the reactive current flow: flow:
1 Xi
∆U = 1 ∆U = − X ∆Ir . (21)
1 + Ts PCC XT i∙ s
∆U∆U == ∆U ∆UPCCPCC= =−− i ∆I ∆I
r. r. (21)
(21)
1+1+ TsTs T∙s
In the Laplace domain, the initial state is not included,T·s so that only the reaction after an impulse
In the
is described. Laplace
In domain,
addition, thethe initial
initial state
voltage is
is not
taken
In the Laplace domain, the initial state is not included, so that included,
into so
account thathereonly
only the
bythe ,reaction
U0reaction
which equalsafteran
after an impulse
UPCC in idle
impulse
is described.
ismode or could
described. In addition,
be adjusted
In addition, the initial
in terms
the initial voltage is
of outer
voltage taken
is taken into
control account
into loops.
account here by U
Consequently, 0 , which
here by U0 , which equals
the provided U
equals UPCCPCC in idle
voltage
in
mode or
amplitude could
is be
given adjusted
as a functionin terms
of the of outer
reactive control
current loops.
deviation Consequently,
idle mode or could be adjusted in terms of outer control loops. Consequently, the provided voltage∆I r from the the
nominal provided
or voltage
initial value:
amplitudeisisgiven
amplitude givenas asaafunction
functionof ofthe
thereactive
reactivecurrent
currentdeviation
deviation∆I ∆Ir from
from thethe nominal
nominal or or initial
initial value:
value:
Xi r
U = − X ∆Ir + U0 . (22)
U=− T i∙ s∆I + U . (22)
Xi
U= ∙ sr +r U0 . 0
T∆I
The resulting block diagram (Figure 21)−canT·s be drawn from Equation (22). The interpretation is
(22)
that The resulting block
the impedance grows diagram
in time(Figure
after an21) can bereactive
evoked drawn current
from Equation
step. The (22). The interpretation
increase of the provided is
that The
the resulting
impedance block
grows diagram
in time (Figure
after an21) can
evoked be drawn
reactive from
current Equation
step.
voltage amplitude counteracts the cause for the current flow, so that in the closed loop, a negative The (22). The
increase interpretation
of the provided is
that the impedance
voltage
feedback amplitude
is achieved. grows
As ainresult,
counteracts time theafter
cause
the an for
evoked
inverter the reactive
current
is led current
backflow,
to theso step.
that in
operating Thethe increase
point closed
beforeof the
loop, theaprovided
negative
incident.
voltage
feedback is achieved. As a result, the inverter is led back to the operating point before theaincident.
amplitude counteracts the cause for the current flow, so that in the closed loop, negative
feedback is achieved. As a result, the inverter is led back to the operating point before the incident.
∆Pout 1 - ∆ω 1 ∆dϑ
∆Pout 10 s -
Jω ∆ω 1s ∆dϑ
-
Jω0 s s
-
Kp
Kp

Figure 20. Generalized transfer function between active power and voltage angle of PLL-based
Figure 20. Generalized transfer function between active power and voltage angle of PLL-based inertial
Figure
inertial20. Generalized
response transfer function between active power and voltage angle of PLL-based
methods.
response methods.
inertial response methods.
U0
U0
1 - U
∆Ir Xi U
T1∙ s -
∆Ir Xi
T∙s
Figure 21. Generalized transfer function between reactive power and voltage amplitude of inverters
with modified
Figure grid current
21. Generalized control.
transfer function between reactive power and voltage amplitude of inverters
Figure 21. Generalized
with modified transfer
grid current function between reactive power and voltage amplitude of inverters
control.
with modified grid current control.
Energies 2020, 13, 2589 16 of 21

In combination with a PLL, which might be used according to the aforementioned relationship
(Figure 20), the modified current-controlled approaches in [57–60] can provide inertia and could also
control a grid without the existence of other grid-forming units. For substantial inertia contribution,
the PLL must be slowed down in order to cause a retarded response of the provided voltage phasor.
The reduction of the classical PI-control structure to a proportional control is a plausible adjustment
in order to soften the effect of the current control in favor of a slowly-following voltage feedforward.
The remaining proportional controller can be regarded as an increase in the available resistive
coupling impedance.
For further frequency support in [57–59] outer control loops have been implemented in order to
adjust active power set-points dependent on frequency deviation and frequency derivative. In such
case [51] demonstrates that even operating in stand-alone mode could be achieved. Therefore, inverters
with enhanced current-controlled approaches with PLL for frequency and angle detection are capable
of acting as grid-forming inverters. Further research on the performance is necessary.

4.8. Direct Power Control (DPC)

Another approach to be discussed is the so-called direct power control (DPC). It has in common
with many other approaches the fact that the output power flow is measured to adjust the internal
power-source voltage phasor. This idea of controlling the inverter’s output power directly, without a
dedicated current controller and PLL reaches back to [61]. In this approach, the switching pattern of the
three-phase bridge, producing the internal voltage phasor, is gained from the hysteresis comparators
and the position of the estimated output voltage. The optimal switching table can be found in [61].
The hysteresis controller leads to a variable switching frequency, which may cause challenges in
designing the output filter.
Although, a dedicated PLL is not used, the estimation of the output voltage gives the reference for
the phase angle. As a result, the standard DPC can be categorized as a grid-following method [62].
Based on this concept, [63] uses a virtual flux similar to [18] by integrating the voltage in order
to obtain the phase angle information. Due to the low-pass behavior of the integrator, a smooth and
retarded reaction to voltage disturbances is achieved [64]. This effect is comparable to that of a slowly
following PLL as described in Section 4.7. Moreover, in [64] this concept is developed further to DPC
with a rotating-frame PI controller and space vector modulation. By doing this, the drawback caused
by variable switching frequency is eliminated.
Recently, a grid-voltage modulated direct power control (GVM-DPC) was suggested [65,66].
In contrast to the approach in [64], the output voltage is directly used to apply a kind of inverse park
transform and gain the vector position [67]. Consequently, the tracking performance is improved,
since the delay of a PLL no longer prevails. However, the downside thereof is losing the beneficial
behavior of a well-designed PLL, e.g., filtering of harmonic perturbation or inertia provision as
discussed in Section 4.7. For weak grids, it was proposed to use a band-pass voltage filter for
harmonic stability improvement, which in turn produces again a delay for the vector positioning [68].
Furthermore, reference [65] demonstrates the performance of the GVM-DPC after a sudden frequency
variation. The result is that even transiently no power deviation occurred. Hence, the inverter perfectly
follows the grid voltage angle to attain its operation point. In other words, without any retarded
voltage vector positioning, this technique acts as a grid-following inverter.
Another approach that is referenced as direct power control can be found in [69]. The control
scheme is given in Figure 22. There are two loops, i.e., the reactive power loop to control the voltage
amplitude and the active power loop to control the angle. The voltage amplitude does not follow fast
variations. After a sudden change on the grid-side, it retains its value. According to the time constant
T of the integrator, the previous operating point is slowly adjusted.
It should be noted that the angle ϑc is the phase angle shift related to the PLL angle θpll . Hence,
a retarded adjustment of the controlled angle ϑc is undermined, when the PLL follows the grid voltage
 Energies 2020, 13, 2589 17 of 21

Energies 2020, 13, Thus,

too quickly. x FOR PEER REVIEW
for analyzing the overall control performance, the dynamics of the PLL is of 18 of 22
great
importance. A closer look into the PLL and its parameterization is not given in the Reference.
However,
However,ififthe thePLL
PLLisisdesigned
designedininthetheaforementioned
aforementionedmanner manner(see
(seeSection
Section4.7),
4.7),ananinherent
inherent
reaction with active power injection is given, because of the retarded following
reaction with active power injection is given, because of the retarded following of the of the grid angle.
grid With
angle.
some
Withtime
somedelay, the PIthe
time delay, controller behind
PI controller adjusts
behind the former
adjusts the formeroperating point,
operating due
point, to to
due thethe
integral
integralpart
part
ofofthe controller. The proportional part serves as a type of feedforward control. With a
the controller. The proportional part serves as a type of feedforward control. With a well-designedwell-designed
PLL
PLLforforthe purpose
the purpose of aofdamped
a dampedinertia contribution,
inertia the proportional
contribution, part could
the proportional be neglected.
part could The
be neglected.
authors would like to emphasize that this method does not generally have a
The authors would like to emphasize that this method does not generally have a grid-forminggrid-forming capability.
However,
capability.with a well-designed
However, PLL, damped
with a well-designed inertia
PLL, dampedcontribution and even a stand-alone
inertia contribution operation
and even a stand-alone
could be realized.
operation could be realized.

fpll
fref -

1 Li I PCC
Rd

Pref G θpll
+G
- Js ϑc ud dq
P q
U
ϑ uq U
Qref 1
U d αβ

- Ts
Q

Figure
Figure22.
22.Direct
Directpower
powercontrol
controlaccording
accordingtoto[69].
[69].

5.5.Conclusions
Conclusionsand
andOutlook
Outlook
Both,synchronous
Both, synchronousmachinesmachinesand andsuitably
suitablycontrolled
controlledstaticstaticconverters
converterswith withsufficient
sufficientenergy
energy
reserves can autonomously form the grid voltage and provide electrical
reserves can autonomously form the grid voltage and provide electrical inertia. Grid-forming inertia. Grid-forming inverters
which incorporate
inverters electricalelectrical
which incorporate inertia can be realized
inertia can bewith different
realized withcontrol
different methods.
control Several
methods. approaches
Several
have been reviewed and the dynamic behavior of the provided
approaches have been reviewed and the dynamic behavior of the provided voltage phasor voltage phasor was analyzed forwas
both
voltage amplitude and voltage angle. In most cases, the dynamic behavior
analyzed for both voltage amplitude and voltage angle. In most cases, the dynamic behavior was was described in the form
of a transfer
described in theblock
form diagram from the
of a transfer blockactive and reactive
diagram from thepower active (or
andcurrent)
reactivetopower
the provided voltage
(or current) to
angle and amplitude, respectively. In that way, an illustrative representation
the provided voltage angle and amplitude, respectively. In that way, an illustrative representation in in terms of the dynamic
orderoforthe
terms damping
dynamic capability
order or was given.capability was given.
damping
Theretardedly
The retardedlyfollowing
followingvoltagevoltage angle
angle is ais keya key property
property of grid-forming
of grid-forming inverters.
inverters. In
In predominately inductive grids, such inverters inherently provide electrical
predominately inductive grids, such inverters inherently provide electrical inertia. The contribution inertia. The contribution
totomomentary
momentaryreserve reservecan canbebedefined
defineddirectly
directlythrough
throughthe theretarded
retardedreaction
reactionofofthe thevoltage
voltage phasor.
phasor.
Allmentioned
All mentioned methods
methods couldcouldformform a retardedly
a retardedly following
followingvoltage angle,angle,
voltage exceptexcept
for the for virtual
the oscillator
virtual
circuit and fast phase-locked-loops (PLL). Additionally, most of them
oscillator circuit and fast phase-locked-loops (PLL). Additionally, most of them leverage leverage a feedforward damping a
in any case, which
feedforward damping is equally
in anyascase,
effective as the
which is damper
equally windings
as effective in synchronous
as the damper machines, but with
windings in
the notable difference
synchronous machines,that butthese
withparameters
the notableare not restricted
difference that to physical
these construction
parameters are not limits. Free from
restricted to
physical construction limits. Free from these physical restrictions, electrical power systems couldtheir
these physical restrictions, electrical power systems could be more freely designed concerning be
timefreely
more constants and damping
designed concerning performance.
their time constants and damping performance.
The voltage amplitude branch isismore
The voltage amplitude branch moredifficult
difficulttotodeduce
deduceand andanalyze,
analyze,since sincetwo twoopposite
opposite
operational modes have to be considered: the reactive control and
operational modes have to be considered: the reactive control and the voltage amplitude control the voltage amplitude control
at
at the
the pointpoint of common
of common coupling.
coupling. In case
In case of the
of the droop
droop control,
control, a reactive
a reactive power
power valueis isused
value usedasas
feedback.Thus,
feedback. Thus, thethe derivation
derivationofofa dynamic
a dynamic transfer function
transfer from the
function from reactive power topower
the reactive the amplitude
to the
of the voltage phasor is straightforward. In contrast, in approaches
amplitude of the voltage phasor is straightforward. In contrast, in approaches with a feedback with a feedback of a voltage
of a
value, an analogy is not obvious. This was solved using the assumption
voltage value, an analogy is not obvious. This was solved using the assumption that in inductive that in inductive networks,
the voltage
networks, theamplitude deviation
voltage amplitude at the terminals
deviation is a function
at the terminals of the reactive
is a function currentcurrent
of the reactive flow, so that
flow,
so that again a relation between reactive current or power and the provided voltage amplitude is
established. Voltage droop control has the aim of defining the reactance for the reactive current
component in order to increase the reactive power sharing. In case of weak grids or in case of
electrically distant power sources, the control of the voltage amplitude at the point of common
Energies 2020, 13, 2589 18 of 21

again a relation between reactive current or power and the provided voltage amplitude is established.
Voltage droop control has the aim of defining the reactance for the reactive current component in order
to increase the reactive power sharing. In case of weak grids or in case of electrically distant power
sources, the control of the voltage amplitude at the point of common coupling is reasonable, so that
both modes have their application. Depending on the grid situation, either one or the other mode is
preferable. In all cases, the control action can be interpreted as an impedance evolution in time after a
reactive current step, either in a positive or in a negative direction. This was illustrated here in the
form of a dynamic transfer function from reactive current to the amplitude of the internally provided
voltage phasor.

Author Contributions: Conceptualization, P.U., M.N. and P.S.; methodology, P.U. and M.N.; formal analysis, P.U.,
M.N. and F.W.; writing—original draft preparation, P.U., M.N. and F.W.; writing—review and editing, P.S., M.N.
and P.U.; visualization, P.U., M.U. and F.W.; supervision, P.S.; project administration, P.S.; funding acquisition, P.S.
All authors agree to the published version of the manuscript.
Funding: We acknowledge the support of our work by the German Ministry for Economic Affairs and Energy
and the Projekträger Jülich within the project “Control and Stability in Inverter Dominated Interconnected
Systems-Grid Control 2.0” (FKZ 0350023A). Only the authors are responsible for the content of this publication.
This paper does not necessarily reflect the consolidated opinion of the project consortium “Netzregelung 2.0”.
The founding sponsor had no role in the design of the study; in the collection, analyses, or interpretation of data;
in the writing of the manuscript, and in the decision to publish the results.
Conflicts of Interest: The employer (non-profit research institute) of the authors is patent holder of one of the
listed grid-forming control approaches (Selfsync). The patent expires in August 2021.

References
1. Engler, A.; Hardt, C.; Strauss, P.; Vandenbergh, M. Parallel Operation of Generators for Stand-Alone
Single-Phase Hybrid Systems—First Implementation of a new control Technology. In Proceedings of the 17th
European Photovoltaic Solar Energy Conference, München, Germany, 22–26 October 2001; WIP-Renewable
Energies: Munich, Germany, 2001; pp. 2690–2693.
2. Visscher, K.; de Haan, S.W.H. Virtual synchronous machines (VSG’s) for frequency stabilisation in future
grids with a significant share of decentralized generation. In Proceedings of the CIRED Seminar 2008:
SmartGrids for Distribution, Frankfurt, Germany, 23–24 June 2008; pp. 334–337.
3. Engler, A.; Haas, O.; Kansteiner, B.; Raptis, F.; Sachau, J.; Zacharias, P. Control of Parallel Working Power
Units in Expandable Grids. In Proceedings of the 14th European Photovoltaic Solar Energy Conference,
Barcelona, Spain, 30 June–4 July 1997; Ossenbrink, H.A., Helm, P., Ehmann, H., Eds.; H.S. Stephens &
Associates: Bedfort, UK, 1997; pp. 2029–2032.
4. Anderson, P.M.; Fouad, A.-A.A. Power System Control and Stability, 2nd ed.; IEEE Press, Wiley-Interscience:
Piscataway, NJ, USA, 2003; ISBN 9780471238621.
5. Park, R.H. Two-reaction theory of synchronous machines generalized method of analysis-part I. Trans. Am.
Inst. Electr. Eng. 1929, 48, 716–727. [CrossRef]
6. Milano, F. Power System Modelling and Scripting; Springer: Heidelberg, Germany, 2010; ISBN 3642136680.
7. Weckesser, T.; Johannsson, H.; Ostergaard, J. Impact of model detail of synchronous machines on real-time
transient stability assessment. In Proceedings of the 2013 IREP Symposium Bulk Power System Dynamics
and Control—IX Optimization, Security and Control of Emerging Power Grid (IREP), Rethymnon, Crete,
Greece, 25–30 August 2013; Voumas, C., Ed.; IEEE: Piscataway, NJ, USA, 2013.
8. Prabha, K. Power System Stability and Control, 1st ed.; McGraw-Hill: Hightstown, NJ, USA, 1994.
9. Machowski, J.; Lubosny, Z.; Bialek, J.W.; Bumby, J.R. Power System Dynamics: Stability and Control, 3rd ed.;
John Wiley: Hoboken, NJ, USA, 2020; ISBN 9781119526346.
10. Van Cutsem, T.; Bo%νάς, K. Voltage Stability of Electric Power Systems, 1st ed.; Springer Science + Business
Media: Dordrecht, The Netherlands, 1998; ISBN 9780792381396.
11. Sauer, P.W.; Pai, M.A. Power System Dynamics and Stability, updated ed.; Stipes Publishing L.L.C: Champaign,
IL, USA, 2006; ISBN 9781588746733.
Energies 2020, 13, 2589 19 of 21

12. Crastan, V.; Westermann, D. Elektrische Energieversorgung 3. Dynamik, Regelung und Stabilität, Versorgungsqualität,
Netzplanung, Betriebsplanung und -Führung, Leit- und Informationstechnik, FACTS, HGÜ, 3rd ed.; Springer:
Berlin/Heidelberg, Germany, 2012; ISBN 3642201008.
13. Crastan, V. Elektrische Energieversorgung 1. Netzelemente, Modellierung, Stationäres Verhalten, Bemessung, Schalt-
und Schutztechnik, 3rd ed.; Springer: Berlin/Heidelberg, Germany, 2015; ISBN 366245985X.
14. Oeding, D.; Oswald, B.R. Elektrische Kraftwerke und Netze, 7th ed.; Springer: Heidelberg, Germany, 2011;
ISBN 3642192459.
15. Alberto, A. Arrangement of Parallel Static AC Power Sources Proportions. U.S. Patent 3864620A, 11
September 1973.
16. Torres, L.A.B.; Hespanha, J.P.; Moehlis, J. Power supply synchronization without communication.
In Proceedings of the 2012 IEEE Power & Energy Society General Meeting, San Diego, CA, USA, 22–26
July 2012.
17. Kawabata, T.; Higashino, S. Parallel operation of voltage source inverters. IEEE Trans. Ind. Appl. 1988, 24,
281–287. [CrossRef]
18. Chandorkar, M.C.; Divan, D.M.; Adapa, R. Control of parallel connected inverters in standalone AC supply
systems. IEEE Trans. Ind. Appl. 1993, 29, 136–143. [CrossRef]
19. Tayab, U.B.; Roslan, M.A.B.; Hwai, L.J.; Kashif, M. A review of droop control techniques for microgrid.
Renew. Sustain. Energy Rev. 2017, 76, 717–727. [CrossRef]
20. Engler, A. Device for Parallel Operation of Equal Range Single-Phase or Three-Phase Voltage Sources.
EP1286444B1, 21 August 2001.
21. Engler, A. Regelung von Batteriestromrichtern in Modularen und Erweiterbaren Inselnetzen. Ph.D. Thesis,
Universität Kassel, Kassel, Germany, 2002.
22. De Brabandere, K.; Bolsens, B.; van den Keybus, J.; Woyte, A.; Driesen, J.; Belmans, R. A voltage and frequency
droop control method for parallel inverters. IEEE Trans. Power Electron. 2007, 22, 1107–1115. [CrossRef]
23. Yajuan, G.; Weiyang, W.; Xiaoqiang, G.; Herong, G. An improved droop controller for grid-connected voltage
source inverter in microgrid. In Proceedings of the 2nd IEEE International Symposium on Power Electronics
for Distributed Generation Systems (PEDG), Hefei, China, 16–18 June 2010; IEEE: Piscataway, NJ, USA, 2010;
pp. 823–828.
24. Guerrero, J.M.; GarciadeVicuna, L.; Matas, J.; Castilla, M.; Miret, J. A wireless controller to enhance dynamic
performance of parallel inverters in distributed generation systems. IEEE Trans. Power Electron. 2004, 19,
1205–1213. [CrossRef]
25. Yao, W.; Chen, M.; Matas, J.; Guerrero, J.M.; Qian, Z.-M. Design and analysis of the droop control method
for parallel inverters considering the impact of the complex impedance on the power sharing. IEEE Trans.
Ind. Electron. 2011, 58, 576–588. [CrossRef]
26. Wang, X.; Blaabjerg, F.; Chen, Z. An improved design of virtual output impedance loop for droop-controlled
parallel three-phase voltage source inverters. In Proceedings of the IEEE Energy Conversion Congress and
Exposition (ECCE), Raleigh, NC, USA, 15–20 September 2012; IEEE: Piscataway, NJ, USA, 2012; pp. 2466–2473.
27. Zhang, L.; Harnefors, L.; Nee, H.-P. Power-synchronization control of grid-connected voltage-source
converters. IEEE Trans. Power Syst. 2010, 25, 809–820. [CrossRef]
28. Zhang, L. Modeling and Control of VSC-HVDC Links Connected to Weak AC Systems. Ph.D. Thesis, KTH
Royal Institute of Technology, Stockholm, Sweden, 2010.
29. Remon, D.; Cantarellas, A.M.; Rakhshani, E.; Candela, I.; Rodriguez, P. An active power synchronization
control loop for grid-connected converters. In Proceedings of the IEEE Power & Energy Society General
Meeting, National Harbor, MD, USA, 27–31 July 2014; IEEE: Piscataway, NJ, USA, 2014.
30. Laudahn, S.; Seidel, J.; Engel, B.; Bulo, T.; Premm, D. Substitution of synchronous generator based
instantaneous frequency control utilizing inverter-coupled DER. In Proceedings of the 7th International
Symposium on Power Electronics for Distributed Generation Systems (PEDG), Vancouver, BC, Canada,
27–30 June 2016; IEEE: Piscataway, NJ, USA, 2016.
31. Hesse, R.; Turschner, D.; Beck, H.-P. Micro grid stabilization using the Virtual Synchronous Machine (VISMA).
In Proceedings of the International Conference on Renewable Energies and Power, Valencia, Spain, 15–17
April 2009.
Energies 2020, 13, 2589 20 of 21

32. Glockler, C.; Duckwitz, D.; Welck, F. Virtual synchronous machine control with virtual resistor for enhanced
short circuit capability. In Proceedings of the 2017 IEEE PES Innovative Smart Grid Technologies Conference
Europe (ISGT-Europe), Torino, Italy, 26–29 September 2017; IEEE: Piscataway, NJ, USA, 2017.
33. Zhong, Q.-C.; Weiss, G. Synchronverters: Inverters That Mimic Synchronous Generators. IEEE Trans.
Ind. Electron. 2011, 58, 1259–1267. [CrossRef]
34. D’Arco, S.; Suul, J.A. Virtual synchronous machines—Classification of implementations and analysis of
equivalence to droop controllers for microgrids. In Proceedings of the IEEE Grenoble PowerTech, Grenoble,
France, 16–20 June 2013; IEEE: Piscataway, NJ, USA, 2013.
35. D’Arco, S.; Suul, J.A. Equivalence of Virtual Synchronous Machines and Frequency-Droops for
Converter-Based MicroGrids. IEEE Trans. Smart Grid 2014, 5, 394–395. [CrossRef]
36. Beck, H.-P.; Hesse, R. Virtual synchronous machine. In Proceedings of the 2007 9th International Conference
on Electrical Power Quality and Utilisation, Barcelona, Spain, 9–11 October 2007; IEEE: Piscataway, NJ, USA,
2007; pp. 1–6.
37. Chen, Y.; Hesse, R.; Turschner, D.; Beck, H.-P. Investigation of the Virtual Synchronous Machine in the island
mode. In Proceedings of the 2012 3rd IEEE PES Innovative Smart Grid Technologies Europe (ISGT Europe),
Berlin, Germany, 14–17 October 2012; IEEE, Ed.; IEEE: Piscataway, NJ, USA, 2012; pp. 1–6.
38. Ebrahimi, M.; Khajehoddin, S.A.; Karimi-Ghartemani, M. An Improved Damping Method for Virtual
Synchronous Machines. IEEE Trans. Sustain. Energy 2019, 10, 1491–1500. [CrossRef]
39. Dong, S.; Chen, Y.C. Adjusting Synchronverter Dynamic Response Speed via Damping Correction Loop.
IEEE Trans. Energy Convers. 2017, 32, 608–619. [CrossRef]
40. D’Arco, S.; Suul, J.A.; Fosso, O.B. A Virtual Synchronous Machine implementation for distributed control of
power converters in SmartGrids. Electr. Power Syst. Res. 2015, 122, 180–197. [CrossRef]
41. Sinha, M.; Dorfler, F.; Johnson, B.B.; Dhople, S.V. Uncovering droop control laws embedded within the
nonlinear dynamics of Van der Pol oscillators. IEEE Trans. Control Netw. Syst. 2017, 4, 347–358. [CrossRef]
42. Johnson, B.B.; Sinha, M.; Ainsworth, N.G.; Dorfler, F.; Dhople, S.V. Synthesizing virtual oscillators to control
islanded inverters. IEEE Trans. Power Electron. 2016, 31, 6002–6015. [CrossRef]
43. Johnson, B.; Rodriguez, M.; Sinha, M.; Dhople, S. Comparison of virtual oscillator and droop control.
In Proceedings of the IEEE 18th Workshop on Control and Modeling for Power Electronics (COMPEL),
Stanford, CA, USA, 9–14 July 2017; IEEE: Piscataway, NJ, USA, 2017.
44. Cvetkovic, I.; Boroyevich, D.; Burgos, R.; Li, C.; Mattavelli, P. Modeling and control of grid-connected
voltage-source converters emulating isotropic and anisotropic synchronous machines. In Proceedings of the
IEEE 16th Workshop on Control and Modeling for Power Electronics (COMPEL), Vancouver, BC, Canada,
12–15 July 2015; IEEE: Piscataway, NJ, USA, 2015.
45. Milano, F.; Dorfler, F.; Hug, G.; Hill, D.J.; Verbic, G. Foundations and challenges of low-inertia systems
(Invited Paper). In Proceedings of the 2018 Power Systems Computation Conference (PSCC), Dublin, Ireland,
11–15 June 2018; IEEE: Piscataway, NJ, USA, 2018.
46. Huang, L.; Xin, H.; Wang, Z.; Wu, K.; Wang, H.; Hu, J.; Lu, C. A virtual synchronous control for voltage-source
converters utilizing dynamics of DC-Link capacitor to realize self-synchronization. IEEE J. Emerg. Sel. Top.
Power Electron. 2017, 5, 1565–1577. [CrossRef]
47. Jouini, T.; Arghir, C.; Dörfler, F. Grid-friendly matching of synchronous machines by tapping into the DC
storage. IFAC-PapersOnLine 2016, 49, 192–197. [CrossRef]
48. Lasseter, R.; Chen, Z.; Pattabiraman, D. Grid-Forming Inverters: A Critical Asset for the Power Grid. IEEE J.
Emerg. Sel. Top. Power Electron. 2020, 8, 925–935. [CrossRef]
49. Mo, O.; D’Arco, S.; Suul, J.A. Evaluation of Virtual Synchronous Machines with Dynamic or Quasi-Stationary
Machine Models. IEEE Trans. Ind. Electron. 2017, 64, 5952–5962. [CrossRef]
50. Rokrok, E.; Qoria, T.; Bruyere, A.; Francois, B.; Guillaud, X. Effect of Using PLL-Based Grid-Forming Control
on Active Power Dynamics Under Various SCR. In Proceedings of the IECON 2019—45th Annual Conference
of the IEEE Industrial Electronics Society, Lisbon, Portugal, 14–17 October 2019; IEEE: Piscataway, NJ, USA,
2019; pp. 4799–4804.
51. Duckwitz, D. Power System Inertia. Ph.D. Thesis, Universität Kassel, Kassel, Germany, 2019.
52. Karapanos, V.; de Haan, S.; Zwetsloot, K. Real time simulation of a power system with VSG hardware in the
loop. In Proceedings of the IECON 2011—37th Annual Conference of IEEE Industrial Electronics Society,
Melbourne, Vic, Australia, 7–10 November 2011; IEEE: Piscataway, NJ, USA, 2011; pp. 3748–3754.
Energies 2020, 13, 2589 21 of 21

53. Van Wesenbeeck, M.P.N.; de Haan, S.W.H.; Varela, P.; Visscher, K. Grid tied converter with virtual kinetic
storage. In Proceedings of the IEEE Bucharest PowerTech (POWERTECH) 2009, Bucharest, Romania,
28 June–2 July 2009; IEEE: Piscataway, NJ, USA, 2009.
54. Duckwitz, D. Performance of df/dt-based inertia control during emergency islanding. In Proceedings of the
15th Wind Integration Workshop, Vienna, Austria, 15–17 November 2016; Betancourt, U., Ackermann, T., Eds.;
Energynautics GmbH: Darmstadt, Germany, 2016.
55. Duckwitz, D.; Fischer, B. Modeling and design of df/dt -based inertia control for power converters. IEEE J.
Emerg. Sel. Top. Power Electron. 2017, 5, 1553–1564. [CrossRef]
56. Laudahn, S. Dynamisches Verhalten von Wechselrichterbasierten Erzeugungsanlagen im Kontext Eines
Sicheren und Stabilen Netzbetriebs. Ph.D. Thesis, TU Braunschweig, Braunschweig, Germany, 2017.
57. Erlich, I.; Korai, A. Study on the minimum share of conventional generation units required for stable
operation of future converter-dominated grids. In Proceedings of the 2018 IEEE Power & Energy Society
General Meeting (PESGM), Portland, OR, USA, 5–10 August 2018; IEEE: Piscataway, NJ, USA, 2018.
58. Erlich, I.; Korai, A.; Neumann, T.; Koochack Zadeh, M.; Vogt, S.; Buchhagen, C.; Rauscher, C.; Menze, A.;
Jung, J. New control of wind turbines ensuring stable and secure operation following islanding of wind
farms. IEEE Trans. Energy Convers. 2017, 32, 1263–1271. [CrossRef]
59. Korai, A.W.; Erlich, I. Frequency dependent voltage control by DER units to improve power system frequency
stability. In Proceedings of the 2015 IEEE Eindhoven PowerTech, Eindhoven, The Netherlands, 29 June–2
July 2015; pp. 1–6.
60. Winter, P.; Wrede, H. Impact of power converter control on transient stability of power systems. In Proceedings
of the Conference on Sustainable Energy Supply and Energy Storage Systems, Hamburg, Germany, 20–21
September 2018; Schulz, D., Ed.; VDE Verlag GmbH: Berlin, Germany, 2018.
61. Noguchi, T.; Tomiki, H.; Kondo, S.; Takahashi, I. Direct power control of PWM converter without power-source
voltage sensors. IEEE Trans. Ind. Appl. 1998, 34, 473–479. [CrossRef]
62. Wang, X.; Guerrero, J.M.; Blaabjerg, F.; Chen, Z. A Review of Power Electronics Based Microgrids.
J. Power Electron. 2012, 12, 181–192. [CrossRef]
63. Malinowski, M.; Kazmierkowski, M.P.; Hansen, S.; Blaabjerg, F.; Marques, G.D. Virtual-flux-based direct
power control of three-phase PWM rectifiers. IEEE Trans. Ind. Appl. 2001, 37, 1019–1027. [CrossRef]
64. Malinowski, M.; Jasinski, M.; Kazmierkowski, M.P. Simple Direct Power Control of Three-Phase PWM
Rectifier Using Space-Vector Modulation (DPC-SVM). IEEE Trans. Ind. Electron. 2004, 51, 447–454. [CrossRef]
65. Gui, Y.; Kim, C.; Chung, C.C.; Guerrero, J.M.; Guan, Y.; Vasquez, J.C. Improved Direct Power Control for
Grid-Connected Voltage Source Converters. IEEE Trans. Ind. Electron. 2018, 65, 8041–8051. [CrossRef]
66. Gui, Y.; Wang, X.; Wu, H.; Blaabjerg, F. Voltage-Modulated Direct Power Control for a Weak Grid-Connected
Voltage Source Inverters. IEEE Trans. Power Electron. 2019, 34, 11383–11395. [CrossRef]
67. Gui, Y.; Wang, X.; Blaabjerg, F.; Pan, D. Control of Grid-Connected Voltage-Source Converters:
The Relationship Between Direct-Power Control and Vector-Current Control. IEEE Ind. Electron. Mag. 2019,
13, 31–40. [CrossRef]
68. Gui, Y.; Wang, X.; Blaabjerg, F. Vector Current Control Derived from Direct Power Control for Grid-Connected
Inverters. IEEE Trans. Power Electron. 2019, 34, 9224–9235. [CrossRef]
69. Ndreko, M.; Rüberg, S.; Winter, W. Grid forming control for stable power systems with up to 100% inverter
based generation: A paradigm scenario using the IEEE 118-Bus system. In Proceedings of the 17th Wind
Integration Workshop, Stockholm, Sweden, 17–19 October 2018; Betancourt, U., Ackermann, T., Eds.;
Energynautics GmbH: Darmstadt, Germany, 2018.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).