You are on page 1of 8

1

Enhanced Modeling and Control of VSIs in


Microgrids Operating in Grid-Connected Mode
Leonardo Rese, Student Member, IEEE, Antonio Sim oes Costa, Fellow, IEEE,
and Aguinaldo S. e Silva, Member, IEEE
AbstractThis paper focus on the inuence of lter dynamics
on the performance and stability of current control loops of three-
phase, pulse-width modulation (PWM) voltage source inverters
(VSIs) in grid-connected operating mode. The analysis considers
the application of those devices in microgrids. Based on a detailed
state-space model of VSIs and their companion lters, enhanced
controller design techniques are applied in order to prevent
the drawbacks of conventional microgrid control strategies. The
effectiveness of the proposed control strategy is veried via
computer simulations.
Index TermsVoltage Source Inverter (VSI), Inverter Control,
LCL Filter, Microgrids.
I. INTRODUCTION
T
HE evolution of regulatory and operational aspects re-
lated to the operation of power utilities and the emer-
gence of small and micro generation systems (PV systems,
microturbines, wind turbines, etc.) have brought about new
challenges and opportunities for small scale power producers.
In this scenario, the direct connection of diverse types of
distributed energy resources (DERs) to distribution networks
has evolved as an option to improve power quality and system
reliability [1]. Along with all positive aspects of distributed
generation (DG), several technical challenges for the operation
and control of distribution networks with DG resources have
arisen, such as: coupling between active power and voltage
in low voltage networks, changes in short circuit currents,
bidirectional power ows, mandatory disconnection of DG
units in case of a fault in the system, voltage sags and harmonic
distortion. [2].
In order to enhance the integration of DG in distribution
networks, changes in the paradigms of design and operation
of such systems have been proposed and implemented [3].
Among them is the widespread use of communication and
information technology to improve the controllability of the
devices connected to the network. As a result, the concept
of microgrids has been introduced in the late nineties [1]. A
microgrid can be seen as an application of the Smart Grid
concept to distribution networks [4].
A typical microgrid structure comprises a group of radial
feeders connected to the main grid through a single point of
connection, referred to as Point of Common Coupling (PCC).
A microgrid can operate in two modes: grid-connected or
islanded. In grid-connected mode, the microgrid imports or
exports power, depending on its internal load and generation
balance and according to proper market policies. In islanded
mode, the microgrid is disconnected from the main grid
and the load is supplied by local sources. If the available
generation is not enough to supply all loads, only selected
components of the load remain connected, while the others
must be disconnected, which points out to the need of a load-
shedding control [5].
In most cases, local microsources are connected to the grid
through power electronic-based interfaces in order to achieve
the required exibility, as well as to insure its operation as
a single aggregated system. Among such interface devices,
the most used is the three-phase (or single-phase) pulse-width
modulation (PWM) voltage source inverter (VSI), responsible
for converting energy available in DC form to AC, as well as
to enable the connection of diverse microsources to the main
grid. In grid-connected mode, the objective of the VSI is to
control the output power of the inverter in order to follow
an established reference (active and reactive power). This is
achieved by controlling the output current of the inverter, since
the output voltage is maintained by the bulk power system.
It is well known that the microsources have low or even null
inertia [2]. In this case, the dominant dynamics in microgrids
are in a range of frequencies higher than the range of study of
electromechanical stability problems in large power systems.
Considering this fact, it is clearly important include into such
studies the mathematical description of some fast dynamics of
microgrid components, such as lters and the network.
The design of microsource controllers able to respond in a
millisecond timescale using only local information [5], as well
as simultaneously taking into account the characteristics of
interface devices such as the VSI and companion lters, con-
stitutes a major challenge in the process of ensuring the proper
operation of microgrids. A number of authors have devoted
efforts to devise control strategies applicable to microgrids.
However, most of those studies are based on phasor models
for the VSI [6]. Accordingly, the lter is simply represented
by a constant impedance, and as a consequence its dynamic
behavior is neglected.
The objective of this paper is to study the inuence of lter
dynamics on the stability and performance of current control
loops in grid-connected operating mode. As a consequence,
this paper proposes the application of enhanced design meth-
ods capable of avoiding some drawbacks of classic microgrid
control strategies. Since the proposed methodology is based on
state-space system models, a detailed three-phase state-space
representation of the VSI and its companion lters is initially
presented, as well as its transformation to the dq0 reference
frame.
This paper is organized as follows. Section II presents the
detailed modeling of a three-phase VSI with an LCL output
978-1-4577-2159-5/12/$31.00 2011 IEEE
2
lter. Section III describes the current control problem in
grid-connected operating mode as well as a description of
a commonly used control strategy. Section IV describes the
control strategy proposed in this paper. Simulation results are
discussed in Section V. Section VI presents some concluding
remarks.
II. THREE-PHASE VOLTAGE SOURCE INVERTER
DESCRIPTION AND MODELING
Single-phase or three-phase PWM VSIs are commonly
used in microgrids as an interface for the connection of
microsources to the utility grid. In most studies related to
the application of VSIs in microgrids, the VSI is simply
represented by a controlled AC voltage source in series with
an impedance (representing the impedance of the lter). In
this paper, however, a more detailed representation of the VSI
that includes the lter dynamics is considered.
A general overview of a VSI is shown in Fig. 1. This
structure represents a three-phase VSI connected to the utility
grid through a third order LCL lter. In this case, the bulk
power system is represented by an ideal balanced three-phase
AC voltage source (representing an innite bus). Another
assumption of this model is that the DC side of the converter
can be represented by an ideal DC voltage source. In other
words, the dynamic behavior of the DC side will not be
considered in this paper.
Fig. 1. Topology of the three-phase VSI with LCL output lter
The output low-pass lter is employed to attenuate the
inverter high frequency switching harmonics [7]. An LCL
lter is more adequate than a simple L lter in microgrid
applications because, in addition to providing higher high-
frequency attenuation with the same inductance value, it is
also suitable to inverter operation in both grid-connected and
islanded modes [8].
Modeling of a PWM switching power converter can be basi-
cally accomplished through two distinct approaches: switched
(discrete-time) models or averaged (continuous-time) mod-
els [9]. In this paper, the VSI model is obtained via the State-
Space Averaging (continuous-time model) technique originally
proposed in [10] and detailed in [11].
A. State-Space Averaging Technique
Consider a PWM converter operating in the continuous
conduction mode. The converter circuit contains independent
states x(t) and is driven by independent sources u(t). During
the rst subinterval, when the upper-side switches (S
1
, S
3
and
S
5
in Fig. 1) are closed, the converter reduces to the linear
circuit shown in Fig. 2.
Fig. 2. Equivalent circuit for the rst subinterval (phase a)
The state equations that describe the circuit in the rst
subinterval are represented as given in (1) and detailed in (2):
x = A
1
x +B
1
u (1)
C
f
v
a
c
f
= i
a
1
i
a
2
L
f1

i
a
1
= v
dc
v
a
c
f
R
f1
i
a
1
L
f2

i
a
2
= v
a
c
f
v
a
R
f2
i
a
2
(2)
During the second subinterval, when the lower-side switches
(S
2
, S
4
and S
6
in Fig. 1) are closed, the converter reduces to
the linear circuit shown in Fig. 3 (notice the polarity of the
DC source).
Fig. 3. Equivalent circuit for the second subinterval (phase a)
The state equations that describe the circuit in the second
subinterval are represented as given in (3) and detailed in (4):
x = A
2
x +B
2
u (3)
C
f
v
a
c
f
= i
a
1
i
a
2
L
f1

i
a
1
= v
dc
v
a
c
f
R
f1
i
a
1
L
f2

i
a
2
= v
a
c
f
v
a
R
f2
i
a
2
(4)
Provided that the natural frequencies of the converter, as
well as the frequencies of the converter input variations, are
much lower than the switching frequency, the state-space
averaged model that describes the low frequency dynamic
behavior of the converter is given by (5) [11]:
x = [dA
1
+ (1 d) A
2
] x + [dB
1
+ (1 d) B
2
] u (5)
where d corresponds to the duty cycle of the PWM modulator.
By substituting (2) and (4) into (5), we have:
v
a
c
f
=
1
C
f
i
a
1

1
C
f
i
a
2

i
a
1
=
2d (t) 1
L
f1
v
dc

1
L
f1
v
a
c
f

R
f1
L
f1
i
a
1

i
a
2
=
1
L
f2
v
a
c
f

1
L
f2
v
a

R
f2
L
f2
i
a
2
(6)
where d (t) =
1
2
+
m
2
cos (t +), m stands for the modu-
lation index, is the angular frequency and is the phase
angle.
3
The equivalent circuit for the averaged model (phase a) is
shown in Fig. 4, whereas the complete averaged model for the
three-phase VSI with an LCL lter is given by Eq. (7), where
V
dc
is the DC source voltage and V
p
stands for the peak value
of the grid voltage.
Fig. 4. Equivalent circuit of the averaged model
The above model is not ideal for controller design purposes,
since in that case the objective of the control would be to
track a sinusoidal reference, making the control problem much
more complicated. In order to circumvent this problem, the
model is transformed to the synchronous reference frame (also
known as dq0 frame) through the use of a dq0 transformation.
The convention used in this paper for the dq0 transformation
considers the d-axis lagging the q-axis and also maintains
the one-to-one correspondence between the peak values of
voltages and currents in abc and dq0 frames. This convention
is the same used in [12].
The transformation matrix from abc reference to dq0 refer-
ence is dened as:
x
dq0
= Tx
abc
, (8)
where
T =
2
3
_
_
cos (t) cos
_
t
2
3
_
cos
_
t +
2
3
_
sin (t) sin
_
t
2
3
_
sin
_
t +
2
3
_
1
2
1
2
1
2
_
_
(9)
The inverse transformation matrix, that is, the one that
transforms the dq0 quantities to the abc reference frame is
dened by Eqs. (10) and (11):
x
abc
= T
1
x
dq0
(10)
T
1
=
_
_
cos (t) sin (t) 1
cos
_
t
2
3
_
sin
_
t
2
3
_
1
cos
_
t +
2
3
_
sin
_
t +
2
3
_
1
_
_
(11)
Dening the rotated state and input vectors as:
x
r
= Px (12)
u
r
= Pu (13)
where
P = diag {T, T, T} , (14)
we can obtain the state-space equations of the three-phase VSI
with the companion LCL output lter by substituting (12) and
(13) into (7). This transformation is summarized in (15):
x
r
=
_
PAP
1
P

P
1
_
x
r
+u
r
(15)
The complete model of the three-phase VSI in the rotating
reference frame is given by Eq. (16). Assuming balanced
operation, the 0 components can be omitted and the model
dimension is reduced to sixth order.
It is clear that this is a nonlinear model, since we have prod-
ucts of input and state variables as well as nonlinear functions
(sine and cosine) of the control variables. A simplication that
can be introduced into the model represented by Eq. (16) is to
redene the input vector as in (17), where v
d
i
= mV
dc
cos (),
v
q
i
= mV
dc
sin (), v
d
g
= V
p
cos () and v
q
g
= V
p
sin () and
is the grid voltage phase angle. In this case, the control
variables become v
d
i
and v
q
i
and the grid voltages become v
d
g
and v
q
g
.
u =
_
0 0
v
d
i
L
f
1
v
q
i
L
f
1

v
d
g
L
f
2

v
q
g
L
f
2
_
T
(17)
The linearized version of the three-phase VSI model with
an LCL output lter is given by Eq. (18). The inputs v
d
g
and
v
q
g
are not represented in the small signal model because,
as stated before, it is assumed that the inverter is connected
to a stiff grid, in this paper represented by an ideal balanced
three-phase AC voltage source. As a consequence, variations
v
d
g
and v
q
g
are null. Another important characteristic of this
model is the coupling between the d- and q-axis components,
which makes the control design more difcult. On the other
hand, however, the control problem in the dq-frame is simply
one of tracking constant references.
III. CURRENT CONTROL IN GRID-CONNECTED MODE
A. Controller Design in the dq0 Reference Frame
The objective of the inverter control in grid-connected
operating mode is to ensure active and reactive power injection
levels according to preset values. In this operating mode, the
voltage at the PCC is assumed to be constant and regulated by
the main grid [13]. Active and reactive power injection control
is achieved by regulating the output current of the inverter.
In the literature of microgrids, difculties associated to this
operating mode are often neglected, and most authors focus
attention only on the inverter control in islanded mode [14]
[19]. However, if the LCL lter dynamics is represented in
the VSI model, the control problem in grid-connected mode
becomes much more complicated due to limitations imposed
by the resonance frequency of the lter.
Usual VSI control strategies are based on proportional-
integral (PI) controllers. However, the design of such con-
trollers within the abc reference frame does not yield a null
steady-state tracking error for sinusoidal references. Although
such tracking error could theoretically be reduced by increas-
ing the controller gain and bandwidth, this probably would led
to an unstable control loop [8]. An alternative solution is to use
a proportional-resonant (PR) controller in the abc reference
frame [8]. Although this controller provides very low steady-
state tracking error, it works well only for grid frequencies
very close to the nominal frequency, a strong assumption in
real applications.
In this paper, the VSI current control is developed in the dq
reference frame. The reason behind this choice is that in this
case the control problem becomes a simple constant reference
tracking problem. On the other hand, one has to deal with the
coupling between d- and q-axis quantities introduced by the
dq0 transformation, as already mentioned in Section II. There
4
_

_
v
a
cf
v
b
cf
v
c
cf

i
a
1

i
b
1

i
c
1

i
a
2

i
b
2

i
c
2
_

_
=
_

_
0 0 0
1
Cf
0 0
1
Cf
0 0
0 0 0 0
1
Cf
0 0
1
Cf
0
0 0 0 0 0
1
Cf
0 0
1
Cf

1
Lf
1
0 0
Rf
1
Lf
1
0 0 0 0 0
0
1
Lf
1
0 0
Rf
1
Lf
1
0 0 0 0
0 0
1
Lf
1
0 0
Rf
1
Lf
1
0 0 0
1
Lf
2
0 0 0 0 0
Rf
2
Lf
2
0 0
0
1
Lf
2
0 0 0 0 0
Rf
2
Lf
2
0
0 0
1
Lf
2
0 0 0 0 0
Rf
2
Lf
2
_

_
_

_
v
a
cf
v
b
cf
v
c
cf
i
a
1
i
b
1
i
c
1
i
a
2
i
b
2
i
c
2
_

_
+
_

_
0
0
0
mVdc
Lf
1
cos (t +)
mVdc
Lf
1
cos
_
t +
2
3
_
mVdc
Lf
1
cos
_
t + +
2
3
_

Vp
Lf
2
cos (t +)

Vp
Lf
2
cos
_
t +
2
3
_

Vp
Lf
2
cos
_
t + +
2
3
_
_

_
(7)
_

_
v
d
cf
v
q
cf
v
0
cf

i
d
1

i
q
1

i
0
1

i
d
2

i
q
2

i
0
2
_

_
=
_

_
0 0
1
Cf
0 0
1
Cf
0 0
0 0 0
1
Cf
0 0
1
Cf
0
0 0 0 0 0
1
Cf
0 0
1
Cf

1
Lf
1
0 0
Rf
1
Lf
1
0 0 0 0
0
1
Lf
1
0
Rf
1
Lf
1
0 0 0 0
0 0
1
Lf
1
0 0
Rf
1
Lf
1
0 0 0
1
Lf
2
0 0 0 0 0
Rf
2
Lf
2
0
0
1
Lf
2
0 0 0 0
Rf
2
Lf
2
0
0 0
1
Lf
2
0 0 0 0 0
Rf
2
Lf
2
_

_
_

_
v
d
cf
v
q
cf
v
0
cf
i
d
1
i
q
1
i
0
1
i
d
2
i
q
2
i
0
2
_

_
+
_

_
0
0
0
mVdc
Lf
1
cos ()
mVdc
Lf
1
sin ()
0

Vp
Lf
2
cos ()

Vp
Lf
2
sin ()
0
_

_
(16)
_

_
v
d
cf
v
q
cf

i
d
1

i
q
1

i
d
2

i
q
2
_

_
=
_

_
0
1
Cf
0
1
Cf
0
0 0
1
Cf
0
1
Cf

1
Lf
1
0
Rf
1
Lf
1
0 0
0
1
Lf
1

Rf
1
Lf
1
0 0
1
Lf
2
0 0 0
Rf
2
Lf
2

0
1
Lf
2
0 0
Rf
2
Lf
2
_

_
_

_
v
d
cf
v
q
cf
i
d
1
i
q
1
i
d
2
i
q
2
_

_
+
_

_
0 0
0 0
1
Lf
1
0
0
1
Lf
1
0 0
0 0
_

_
_
v
d
i
v
q
i
_
(18)
are, however, means to get around that problem, as described
in the sequel.
The dq current control structure requires information about
the frequency and phase angle of the grid voltage in order to
perform the dq0 transformation and also synchronize the VSI
with the grid. This can be achieved by using a Phase-Locked
Loop (PLL). We assume that the phase angle and frequency
information used for the dq0 transformation are the same as the
phase angle and frequency of the grid voltage, so that v
q
g
= 0.
As a consequence, the active and reactive power expressions
in the dq frame become [20]:
P =
3
2
v
d
g
i
d
(19)
Q =
3
2
v
d
g
i
q
(20)
By analyzing (19) and (20) we conclude that the active and
reactive power injection should be regulated through the d-axis
current and the q-axis current, respectively.
B. PI Control Strategy Based on the Feedback of the Inverter-
side Current
A widely used strategy for current control is to feedback
the inverter-side current i
1
and use a feed-forward technique
to decouple d- and q-axis [18], [21], [22]. To better explain this
procedure, lets us analyze the state equations of the inverter-
side current given by:
L
f1

i
d
1
= R
f1
i
d
1
v
d
C
f
+L
f1
i
q
1
+v
d
i
L
f1

i
q
1
= R
f1
i
q
1
v
q
C
f
L
f1
i
d
1
+v
q
i
(21)
The decoupling of the above equations requires the feed-
forward of the following terms:
d-axis: v
d
C
f
L
f1
i
q
1
q-axis: v
q
C
f
+L
f1
i
d
1
By applying this strategy, the state equations (21) become:
L
f1

i
d
1
= R
f1
i
d
1
+v
d
i
L
f1

i
q
1
= R
f1
i
q
1
+v
q
i
(22)
This procedure results in two completely decoupled state
equations. In most cases the measurement of the capacitor
5
voltage v
C
f
is impractical and the grid voltage measurement
is used instead (under the assumption that v
C
f
v
g
). As
shown by (22), this feed-forward strategy allows that the d-
axis current be controlled via d-axis inverter voltage v
d
i
, while
the q-axis current can be controlled via the q-axis inverter
voltage v
q
i
. In both cases, the resulting transfer function for
this system is given by:
G(s) =
I
d
1
(s)
V
d
i
(s)
=
I
q
1
(s)
V
q
i
(s)
=
1
L
f1
s +R
f1
(23)
Fig. 5 describes the overall control strategy, that comprises
both the dq-decoupling actions and a PI controller for each
axis. The PI controller is used due to its constant reference
tracking properties. In addition, its design can be easily
accomplished since the system is represented by a rst order
transfer function. A disadvantage of this control strategy is
that the regulated current is not the grid-side current. This
tends to cause the circulation of reactive currents between the
grid and the CL branch of the lter, something intolerable
in microgrid applications. Another disadvantage is that a
perfect dq-decoupling is seldom possible, due to the following
reasons: the exact value of L
f1
is usually unknown; the grid
frequency tends to uctuate around its nominal value, and the
same occurs with v
C
f
with respect to v
g
.
Fig. 5. Structure of the controller using i
1
as feedback variable
IV. PROPOSED STATE FEEDBACK CONTROLLER
In this paper, we take advantage of the detailed state-space
model developed in Section II to design a state feedback
strategy [23]. Since the state-space model takes into account
the interactions among all system variables without major
approximations, it is expected that the resulting control policy
outperforms conventional single variable-based controllers.
State feedback strategies require the measurement of all
state variables, which in our case are inverter- and grid-side
currents i
1
and i
2
and capacitor voltage v
C
f
. However, for
practical reasons, it is desirable to measure only the grid-
side current i
2
. This dilemma can be solved through a state
observer [23], which provides state estimates for the remaining
states by using the i
2
measurement and information about the
inputs. The use of the observer is made feasible by the fact
that the state-space model in Eq. 18 is fully observable, and
also by the low order of the plant [23].
Therefore, solving the controller design problem amounts
to determining appropriate gain matrices K and L for the
state feedback controller and the observer, respectively. A
pole placement procedure is employed to determine both
matrices [23].
A nal difculty to be circumvented is related to the fact that
the conventional state feedback controller design is a regulator
problem, whereas in this application we are interested in
accurately tracking a preset reference. In order to incorporate
the tracking requisite into the state feedback problem, we must
add integral control to the plant. This is achieved by placing
integrators in the forward path in series with the plant. Because
every added integrator increases the order of the plant by 1,
an augmented plant model is required. The state equations of
the augmented plant become:
_
x
x
I
_
=
_
A 0
C 0
_ _
x
x
I
_
+
_
B
0
_
u +
_
0
I
_
r (24)
where A and B are the state and input matrices in Eq. 18,
and C is given by:
C =
_
0 0 0 0 1 0
0 0 0 0 0 1
_
(25)
The controller design is then performed using the augmented
plant model. The result of the design procedure is the gain
matrix K of the state feedback controller along with a gain
matrix consisting of integrator gains (K
I
). The structure of
proposed controller is shown in Fig. 6.
Fig. 6. Proposed State Feedback Controller
Finally, it is important to emphasize that the controller
which results form the proposed application of the state
feedback procedure does not make use of any approximate
decoupling artice for the system state equations, which is an
advantage as compared to the conventional strategy described
in Subsection III-A.
V. SIMULATION RESULTS
The performances of the VSI controllers dealt with in
this paper have been assessed via computer simulation with
Simulink
c
in the Matlab
c
environment. The test system data
used in the case studies are presented in the Appendix. The
VSI nonlinear model developed in Section II and given by
Eq. (7) is used in all simulations. All simulations have been
carried out under continuous-time domain with an integration
6
step of 1 ms. The rst part of this section reports results
obtained for the PI control strategy using the inverter-side
current as feedback variable. Results related to the state space
feedback control plus state observer using the grid-side current
as feedback variable are presented in the second part of this
section. Performances of both control strategies are analyzed in
terms of their tracking capability, effects on system stability
and amount of reactive current circulating between the grid
and the lter.
A. PI Control Strategy Based on the Feedback of the Inverter-
side Current
Assuming dq-decoupling through the feed-forward tech-
nique described in Subsection III-B, the following rst-order
transfer function of the VSI model results:
G(s) =
1
L
f1
s +R
f1
=
1
0.00015s + 0.045
(26)
A PI controller can then be designed through the use of classic
control techniques, resulting in the following transfer function:
PI (s) = K
p
+
K
i
s
= 0.45 +
135.00
s
(27)
Simulations are then performed in the following conditions:
from 0 t < 0.05 seconds, the reference for the d- and q-
axis inverter-side currents are set to zero (I
d
ref
1
= I
q
ref
1
= 0).
At t = 0.05 seconds, steps of magnitudes I
d
ref
1
= 10A and
I
q
ref
1
= 5A are applied to the references of d- and q-axis
inverter-side currents, respectively. Results of the correspond-
ing simulations are presented in Figs. 7 to 10.
Figures 7 and 8 show the responses of the inverter-side
current in the dq and abc reference frames, respectively. The
plots clearly show that the controller is able to track the
references very well, since the transient responses die out
within one cycle of the current waveform and the steady-
state errors are zero. As for the grid-side currents shown in
Figs. 9 and 10, however, the controller falls short of the desired
performance. During the interval 0 t < 0.05 seconds, in
which the references for the inverter-side current are null, there
is a signicant circulating current between the grid and the
C
f
L
f2
branches of the lter. Such a large current cannot be
tolerated in grid-connected operating mode, as anticipated in
Subsection III-B.
0.045 0.05 0.055 0.06 0.065 0.07 0.075 0.08
2
0
2
4
6
8
10
12
i
1
(dq frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
1
d
i
1
q
I
d
ref
I
q
ref
Fig. 7. Inverter-side current in the dq reference frame (PI Controller)
Simulations have been also conducted considering the dq-
decoupling strategy applied to the grid-side currents. Although
detailed results are not reported in this paper due to space
0.045 0.05 0.055 0.06 0.065 0.07 0.075 0.08
15
10
5
0
5
10
15
i
1
(abc frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
1
a
i
1
b
i
1
c
Fig. 8. Inverter-side current in the abc reference frame (PI Controller)
0.045 0.05 0.055 0.06 0.065 0.07 0.075 0.08
4
2
0
2
4
6
8
10
12
i
2
(dq frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
2
d
i
2
q
Fig. 9. Grid-side current in the dq reference frame (PI Controller)
0.03 0.035 0.04 0.045 0.05 0.055 0.06 0.065 0.07 0.075 0.08
10
5
0
5
10
i
2
(abc frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
2
a
i
2
b
i
2
c
Fig. 10. Grid-side current in the abc reference frame (PI Controller)
limitations, it has been concluded that the PI controller yields
a poor transient response in this case, characterized by large
settling times. This is due to the fact that the controller
bandwidth is strongly limited by the resonance frequency of
the LCL lter.
B. State Feedback and State Observer to Control the Grid-side
Current
The objective of this control strategy is to ensure that the
eigenvalues of the closed-loop system satisfy specied damp-
ing ratios while keeping integral action tracking capability.
This is achieved by dening the desired closed-loop poles and
designing a state feedback controller that places the closed-
loop poles on the desired locations in the complex plane.
The rst step towards the application of the proposed control
strategy is to verify whether the system under study satises
the required controllability and observability properties [23].
This can be easily done by using the state-space model
developed in Section II, resulting in a fully controllable and
observable system. Next, the open-loop poles (or eigenvalues
of the state matrix) are determined and evaluated. They are
shown in Table I.
The next step for the controller design is the denition of
the desired location of the closed-loop eigenvalues. A mini-
mum damping ratio of 10% was specied for all closed-loop
7
TABLE I
OPEN-LOOP EIGENVALUES
Eigenvalue Damping Freq. (rad/s)
-3.00e+002 + j3.14e+002 6.91e-001 4.34e+002
-3.00e+002 - j3.14e+002 6.91e-001 4.34e+002
-1.50e+002 + j1.98e+004 7.58e-003 1.98e+004
-1.50e+002 - j1.98e+004 7.58e-003 1.98e+004
-1.50e+002 + j2.04e+004 7.35e-003 2.04e+004
-1.50e+002 - j2.04e+004 7.35e-003 2.04e+004
eigenvalues of the system. To meet that target, we make use of
state feedback control. The procedure employed in this paper
is to move the poor damped open-loop eigenvalues to the left
of the complex plane without changing their imaginary parts.
This amounts to calculating new values for the eigenvalues
real parts in order to meet the specied damping ratios, as
follows:

new
ij
=

min
{
ij
}
_
1 (
min
)
2

+j{
ij
} (28)
where
new
ij
corresponds to the desired closed-loop eigenval-
ues,
min
is the specied minimum damping ratio and {
ij
}
stands for the imaginary part of
ij
.
The desired closed-loop eigenvalues for the system under
study are given in Table II. It can be seen that all closed-loop
eigenvalues exhibit a minimum damping of 10%.
TABLE II
CLOSED-LOOP EIGENVALUES
Eigenvalue Damping Freq. (rad/s)
-3.00e+002 + j3.14e+002 6.91e-001 4.34e+002
-3.00e+002 - j3.14e+002 6.91e-001 4.34e+002
-1.99e+003 + j1.98e+004 1.00e-001 1.99e+004
-1.99e+003 - j1.98e+004 1.00e-001 1.99e+004
-2.05e+003 + j2.04e+004 1.00e-001 2.05e+004
-2.05e+003 - j2.04e+004 1.00e-001 2.05e+004
The design of the controller and the observer was carried out
through the use of a pole placement algorithm. The resulting
controller gain matrices are given as follows:
K =
_
0.0195 0.0298 0.8673 0.0535 0.8625 0.0126
0.0387 0.0503 0.0285 0.9324 0.1828 0.6948
_
(29)
K
I
=
_
12.0905 30.2718
69.5823 60.8464
_
(30)
In order to perform the observer design, the desired poles
of the state observer are dened as 2{} +j{} (real part
of the poles of the state observer is two times the real part of
the closed-loop poles of the system, while the imaginary part
of the poles of the state observer is equal to the imaginary part
of the closed-loop poles of the system). The resulting observer
gain matrix is given as follows:
L = 10
5

_
0.1041 0.0127 0.2063 0.0047 0.1004 0.0066
2.4983 0.0297 0.7234 0.2526 0.7197 0.0612
_
T
(31)
The simulations are then performed under the following
conditions: from 0 t < 0.10 seconds, the reference for the d-
and q-axis grid-side currents are set to zero (I
d
ref
2
= I
q
ref
2
=
0). At t = 0.10 seconds, steps of magnitude I
d
ref
2
= 10A
and I
q
ref
2
= 5A are applied to the references of d- and q-axis
grid-side currents. The results of this simulation are presented
in Figs. 11 to 14.
Figs. 11 and 12 show the responses of the inverter-side
current in dq and abc reference frames, respectively. The plots
show that in this case the inverter-side current is not equal to
zero for a null reference for the grid-side current. This is due
to the reactive current circulating through the L
f1
C
f
branch
of the lter. This reactive current is needed to maintain the
capacitor voltage in such a value that the grid-side current is
able to track its predened references. The transient response
is very fast and well damped, dying out within one cycle of
the current waveform.
0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
0
2
4
6
8
10
12
i
1
(dq frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
1
d
i
1
q
Fig. 11. Inverter-side current in the dq reference frame (State Feedback
Controller)
0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
10
5
0
5
10
i
1
(abc frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
1
a
i
1
b
i
1
c
Fig. 12. Inverter-side current in the abc reference frame (State Feedback
Controller)
As for the grid-side currents shown in Figs. 13 and 14,
however, the reactive circulating current is null. The plots show
that the proposed controller is capable of tracking the current
references very well while keeping a minimum damping in
the transient response. As expected in the formulation of
this control strategy, this controller provides null reactive
circulating current between the grid and the lter, making this
controller suitable for microgrid applications. However, there
are still some practical issues that deserve further attention
concerning the use of the proposed methodology, such as the
level of robustness provided by the control structure. Such
issues are beyond the scope of this paper, and will be addressed
in future research.
VI. CONCLUSION
A detailed model of a three-phase PWM VSI including its
companion lter dynamics has been presented. It is shown
that this model is suitable for studies of more complex
8
0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
2
0
2
4
6
8
10
12
i
2
(dq frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
2
d
i
2
q
I
d
ref
I
q
ref
Fig. 13. Grid-side current in the dq reference frame (State Feedback
Controller)
0.08 0.09 0.1 0.11 0.12 0.13 0.14 0.15
10
5
0
5
10
i
2
(abc frame)
Time (s)
C
u
r
r
e
n
t

(
A
)


i
2
a
i
2
b
i
2
c
Fig. 14. Grid-side current in the abc reference frame (State Feedback
Controller)
dynamic phenomena that arise with the widespread use of
VSIs in microgrid applications. Through computer simulations
we show that the lter dynamics has a major inuence on the
performance of the current control loops in grid-connected
operating mode. As shown in Section V, a simple PI control
structure using the inverter-side current as feedback variable
presents a poor performance for microgrid applications due to
the resulting large circulating currents between the main grid
and the inverter output lter. In addition, there are also some
drawbacks related to underlying dq-decoupling artices.
The state feedback-based controller proposed in this paper
has proven to be a good alternative for the current control
problem in grid-connected mode, since that strategy allows
an effective control of the grid-side current. It also prevents
the difculties brought about by the need of dq-decoupling
techniques. The robustness properties of the proposed strategy
constitute a relevant issue that still requires further investiga-
tion.
APPENDIX
TABLE III
THREE-PHASE VSI PARAMETERS
Parameter Value
Grid voltage (RMS) 230 V
DC source voltage 450 V
Frequency 50 Hz
Inverter-side parameters R
f
1
= 0.045, L
f
1
= 150H
Grid-side parameters R
f
2
= 0.135, L
f
1
= 450H
Capacitor C
f
= 22F
REFERENCES
[1] R. H. Lasseter, MicroGrids, in Proceedings in IEEE Power Engineer-
ing Society Winter Meeting, vol. 1, 2002, pp. 305308.
[2] S.P.Chowdhury, P.Crossley, and S.Chowdhury, Microgrids and Active
Distribution Networks, ser. Renewable Energy Series 6. Institution of
Engineering and Technology, August 2009.
[3] R. H. Lasseter, A. Akhil, C. Marnay, J. Stephens, J. Dagle, R. Guttrom-
son, A. S. Meliopoulous, R. Yinger, and J. Eto, The CERTS MicroGrid
Concept, Lawrence Berkeley National Laboratory, Tech. Rep., April
2002.
[4] R. H. Lasseter, Smart Distribution: Coupled Microgrids, Proceedings
of the IEEE, vol. 99, no. 6, pp. 10741082, 2011.
[5] P. Basak, A. K. Saha, S. Chowdhury, and S. P. Chowdhury, Microgrid:
Control Techniques and Modeling, in Proceedings in 44th International
Universities Power Engineering Conference (UPEC), 2009, pp. 15.
[6] K. Karoui, V. Chuvychin, A. Sauhats, O. Samuelsson, M. H. J. Bollen,
N. Hatziargyriou, and N. Soultanis, Analysis and Modeling Challenges
Induced by a Growing Penetration of Inverter-Interfaced DER, in Proc.
IEEE Power Engineering Society General Meeting, 2006.
[7] A. Papavasiliou, S. A. Papathanassiou, S. N. Manias, and G. Demetri-
adis, Current Control of a Voltage Source Inverter Connected to the
Grid via LCL Filter, in Proc. IEEE Power Electronics Specialists Conf.
PESC 2007, 2007, pp. 23792384.
[8] S. Yang, Q. Lei, F. Z. Peng, and Z. Qian, A Robust Control Scheme
for Grid-Connected Voltage-Source Inverters, IEEE Transactions on
Industrial Electronics, vol. 58, no. 1, pp. 202212, 2011.
[9] C. K. Tse and M. Di Bernardo, Complex Behavior in Switching Power
Converters, Proceedings of the IEEE, vol. 90, no. 5, pp. 768781, 2002.
[10] R. D. Middlebrook and S. Cuk, A General Unied Approach to
Modelling Switching-Converter Power Stages, in Proceedings of the
IEEE Power Electronics Specialists Conference, June 1976, pp. 7386.
[11] R. W. Erickson and D. Maksimovi c, Fundamentals of Power Electronics,
2nd ed. Kluwer Academics, 2001.
[12] P. Kundur, Power System Stability and Control, ser. The EPRI Power
System Engineering Series. McGraw-Hill Education, 1994.
[13] C.-L. Chen, J.-S. Lai, Y.-B. Wang, S.-Y. Park, and H. Miwa, Design
and Control for LCL-Based Inverters with Both Grid-Tie and Standalone
Parallel Operations, in Proc. IEEE Industry Applications Society Annual
Meeting IAS 08, 2008, pp. 17.
[14] S. Barsali, M. Ceraolo, P. Pelacchi, and D. Poli, Control Techniques of
Dispersed Generators to Improve the Continuity of Electricity Supply,
in Proceedings in IEEE Power Engineering Society Winter Meeting,
vol. 2, 2002, pp. 789794.
[15] A. Engler and N. Soultanis, Droop Control in LV-Grids, in Proceedings
in International Future Power Systems Conference, 2005.
[16] J. A. P. Lopes, C. L. Moreira, and F. O. Resende, Microgrids Black Start
and Islanded Operation, in 15th Power System Computing Conference,
Li ege, August 2005, pp. 1 7.
[17] J. A. P. Lopes, C. L. Moreira, and A. G. Madureira, Dening Control
Strategies for MicroGrids Islanded Operation, IEEE Transactions on
Power Systems, vol. 21, no. 2, pp. 916924, 2006.
[18] N. Pogaku, M. Prodanovic, and T. C. Green, Modeling, Analysis and
Testing of Autonomous Operation of an Inverter-Based Microgrid,
IEEE Transactions on Power Electronics, vol. 22, no. 2, pp. 613625,
2007.
[19] I. A. Hiskens and E. M. Fleming, Control of Inverter-Connected
Sources in Autonomous Microgrids, in Proceedings in American Con-
trol Conference, 2008, pp. 586590.
[20] Y. K. H. Akagi and A. Nabae, Generalized Theory of the Instantaneous
Reactive Power in Three-Phase Circuits, in Proc. of International Power
Electronics Conference, Tokyo, Japan, 1983, pp. 13751386.
[21] F. Katiraei, M. R. Iravani, and P. W. Lehn, Small-Signal Dynamic
Model of a Micro-Grid Including Conventional and Electronically
Interfaced Distributed Resources, IET Generation, Transmission &
Distribution, vol. 1, no. 3, pp. 369378, 2007.
[22] T. Green and M. Prodanovic, Control of Inverter-Based Micro-Grids,
Electric Power Systems Research, vol. 77, no. 9, pp. 1204 1213, 2007.
[23] C.-T. Chen, Linear System Theory and Design, 2nd ed., ser. Electrical
and Computer Engineering. HRW, 1984.

You might also like