B Y R E N A T A C A... ˜ O,

advertisement
BY RENATA CARNIELETTO,
D AN I L O I G L E SI A S B R A ND ÃO ,
S I DD HA RTH SURYA NA RA YA NA N,
F E L I X A . F A R R E T,
& M AR CE LO G. S I M ÕE S
T
HE PRIMARY GOAL OF
this article is to discuss the
development of intelligent controls for a power electronic
inverter capable of interfacing a photovoltaic
(PV)-based unit to the utility grid. The inverter is designed as a single-phase, full-bridge
converter operating at 120 V, 60 Hz ac. The
control functionalities of the inverter are defined
under the perspective of the Smart Grid Initiative (SGI) of the U.S. Department of Energy and
ability to supply real and reactive power to local
loads, supply real and reactive power to other
utility loads up to the rated capacity of the
© ARTVILLE
inverter, provide voltage support at the point of
common coupling (PCC), store energy in a leadacid battery bank, and enable the provision of control options
Integration of distributed generation (DG) sources to
to the consumer based on near real-time electricity information
the electric distribution system has potential advantages
obtained from the utility through advanced metering devices.
including improved supply reliability, custom power
A multifunctional single-phase
voltage-source inverter
Digital Object Identifier 10.1109/MIAS.2010.939651
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
substantiated via case studies in this article as the
27
Date of publication: 28 June 2011
1077-2618/11/$26.00©2011 IEEE
1
Grid
Capacitor +
Banks
Protection
Breaker 2
2
C
C
1
2
1
Test
Breaker 1
Protection
Breaker 1
C
1
2
C
1
Brkr Pro
Brkr
2
Grid
Load
Grid
Test
Breaker 2 Measurements
Outin1
Secondary
VSI
Load
Block diagram of the voltage-source inverter with smart functionalities.
out2 in2
–
Breakers Control
Brkr
Brkr Pro
Brkr Zinv
Brkr Q
Islanding and Reclosure
Protection
Local Load Schedule
STATCOM Function
out1 in1
+
Battery
Bank
Bidirectional
Buck-Boost
VSI
IGBT Two-Bridge LCL
Filter Measurements
Inverter
Primary
VSI
Load
in2 out2
B
in1 out1
A
in2 out2
–
–
in1 out1
+
PV Model
Boost
+
g
Smart Inverter
STATCOM
Discrete
Controllers
Function
PWM Generator
Qref (kW)
Qref
Pulse Uref
pwm-out
Pref
Pref (kW)
Discrete,
Ts = 5e–007 s.
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
Outin1
28
quality, local use of thermal energy, the ability to off-load
electric energy from the transmission grid, and the provision
of an avenue to meet mandatory renewable portfolio
standards (RPSs) [1]. The U.S. federal government has ratified the SGI as its official policy to modernizing the electricity grid, which calls for increased levels of renewable energy
sources in the grid; provision of timely information and control options to consumers; deployment of smart technologies,
appliances, and advanced metering devices; and real-time
pricing of electricity [2]. The perspective of SGI is to develop
the functional controls for a power electronic inverter capable
of interfacing PV installations to the utility grid.
For the purpose of this article, smart controls of a voltagesource inverter are defined as the combined functional ability
to supply power to local loads, supply power to other utility
loads up to rated capacity of the inverter, provide voltage support at the PCC of the utility, store energy in a local lead-acid
battery bank, and provide control options to the consumer
based on near real-time electricity information obtained from
the utility through advanced metering devices. A general
modular design methodology for flexible inverters that may
cater to increasing demands in the smart grid is presented in
[3]; however, the combined smart functionalities described
here are deemed unique. The smart inverter functionalities
described in this study look beyond the recommendations of
the current national technical standard for interconnecting
DG sources to the grid—IEEE Standard 1547 [4]—in providing voltage support at the PCC—thus, offering an ancillary service in case of low-voltage scenarios. Traditionally,
voltage sags in distribution systems are corrected using
utility-owned (or) -operated capacitor banks; however,
with the advent of inverters with smart functionalities, the
ability to regulate voltage at the PCC is brought to the
customer. The authors have not probed the safety issues
stemming from performing voltage control on the grid side
using the proposed inverter setup. Based on real-time spot
pricing of electricity obtained from the utility using an
advanced metering device, the inverter control algorithm
determines the optimal operating mode. This algorithm
enables the inverter to: 1) schedule local loads and 2) determine either to locally store energy or sell energy to the grid.
Description of the Inverter Control
The voltage-source inverter is designed as a 5 kVA onephase, full-bridge converter operating at 120 V, 60 Hz ac
with current control and voltage control to function in two
modes: grid tied and islanded, respectively [5]–[7]. The
entire control is developed in the DQ frame with a virtual
Q axis (as the application is one phase). In the technical
literature, this second virtual quantity is obtained either by
the derivative of the fundamental signal [8] or by delaying
the real-axis quantity by one quarter of the line period [9].
In this implementation, the latter technique is employed.
Figure 1 depicts the block diagram of the inverter simulation on a popular modeling and simulation platform.
Phase-locked loops (PLLs) are used to supply the control
loops with phase angle information [10]–[12], as shown in
Figure 2. The current and voltage control shown use four proportional-integral (PI) controllers: two equal PIs for the
inverter current (id and iq ) and two equal PIs for the inverter
voltage (vd and vq ). The PI compensator is chosen for its
simplicity and ease in implementation, and the respective gains
p_ref
iVSI
(p.u.)
2
q_ref
Current Loop Control
Conversion
to DQ Frame
iVSI
i _dq
thetaVSI VSI
pq_ref
(p.u.)
1
Id_ref
Iq_ref
pq_ref/Idq_ref
vVSI
vVSI
thetaVSI
P1
1
pwm_out
Brkr
Conversion
to DQ Frame
PLL
m
φ
PWM Modulator
Current Control
Voltage Loop Control
vVSI thetaVSI
vVSI
i2dq
m
id_ref
Iq_ref φ (degree)
Switch
v2dq
vVSI_dq
[1 0]
m
vdvq_ref φ (degree)
vd_ref vq_ref
Voltage Control
m
φ
P1
PWM Modulator
2
Current control and voltage control loops for the inverter with smart functionalities.
The Smart Functionalities of the Inverter
The primary intent of the inverter development with smart
functionalities is to enable an efficient interconnection and
economical operation for dispersed PV-based DG installations to the utility grid. The motivation of this study comes
from a pilot program by a local utility in Colorado to implement some of the smart grid recommendations in a candidate
medium-sized metropolitan area [13]. Such an implementation is based on the ubiquitous deployment of PV installations at a residential level of the candidate city. Some
distinctive aspects of this pilot program are smart metering,
the incorporation of smart appliances, the provision of pricing information to consumers, the provision of some control
Voltage
Reference
350
Voltage Control Loop
+–
12,235×[1 2×683 683×683]
s 3 + 2×32,509s 2 + 32,509×32,509s
Out
Inverter
Control Algorithm
PWM
Generator 1
>
0
PWM
Generator 2
0
>
Current
Control Loop
2778×[1 1,700]
s 2 + 94,248s
Current
Reference
+
–
-C-
g
1
in1
1
S1
2
D2
v+
–
L = 430 µH
+ i
–
C2 = 1.3 mF
g
1
dc Link
Side (Vdc)
S2
2
in 2
3
D1
C1 = 560 µF
out1
2
Battery
Side (VB)
out2
4
3
Block diagram of bidirectional dc–dc buck-boost converter subsystem in the inverter setup.
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
options to consumers, and information exchange on a fully
networked system enabled by massively deployed sensors. It
is in this regard that the inverter with the aforementioned
smart functionalities is being proposed in this article.
The local load served by the inverter is modeled as two
components: primary and secondary VSI loads, which distinguishes the critical loads from others that can be scheduled at
the location. So if the inverter is operating at the islanded
mode and it does not have enough power to supply all local
loads, only the VSI primary load will be supplied. Another
convenience of this load set is the ability to operate in the economic mode. This will be explained in the following sections.
The input to the smart inverter is a steady-state voltage
of 350 Vdc , provided by PV panels [14], with a nominal
output voltage of 192 V. A dc–dc boost converter has been
used in the model to raise the PV voltage level to 350 Vdc.
The inverter setup also includes a lead-acid battery storage
bank with nominal voltage of 192 V and 24 Ah cells [15]
connected to the dc link through a bidirectional dc–dc
buck-boost converter, modeled as shown in Figure 3. A
can be tuned through extensive simulations. In the case of the
device shown in Figure 2, the PI compensators are used to
generate the references for a pulsewidth modulator (PWM).
According to Figure 2, the main control consists of two loops:
voltage loop, which is enabled when the inverter operates in
islanded mode, and current loop for a grid-connected condition.
29
the grid based on IEEE Standard 1547
[4]. The functionality is modeled as a
subsystem with the following input
[Qref]
parameters: frequency, phase, and DQ
+–
PI
1
frame voltage of the grid (denoted as
Brkr_Q
1
Discrete
Freq_Grid, Theta_Grid, and Vdq_Grid,
Discrete
Q
ref
PI Controller 0
PLL-Driven
respectively in Figure 5) and inverter
Switch
Fundamental Value
side (denoted as Freq_VSI, Theta_VSI,
freqg
Freq m
and Vdq_VSI, respectively in Figure 5).
The algorithm compares the inputs with
sin cosg
Sin_Cos
IEEE Standard 1547 recommendations
φ
Fix Inverter Apparent
v3pu
In
and generates an output signal (Brkr) in
Power at 5 kVA
the required time frame according to
+–
2
25
sqrt
IEEE Standard 1547 to island or to
Pref
reclose the inverter to the grid. It is
[Qref]
Pref_kW
×
pertinent to note that: 1) the 1547
Saturation
recommendations are not reproduced
4
in this article and the attention of the
curious reader is pointed to [4] and 2)
Block diagram of the STATCOM function used in the inverter-based DG setup.
while the inverter looks beyond IEEE
battery model from a popular library simulation platform Standard 1547 in its ability to regulate the voltage at the
[16] was used for this purpose. The storage subsystem brings PCC, it conforms to IEEE Standard 1547 for grid connecflexibility to the system, e.g., the ability to supply local tion and disconnection.
loads when the inverter is islanded without enough power,
to store cheap energy, and to sell when the price is higher.
Bidirectional DC–DC Buck-Boost Converter
The bidirectional dc–dc buck-boost converter shown in
Figure 3 is responsible for controlling the charge and discharge
STATCOM Function
As affirmed by [17], if individual distributed energy (DE) sys- processes of the battery setup. This converter can behave either
tems are allowed to regulate reactive power, they can also be as buck or as boost converter depending on which switch (S1
used to provide voltage support at the low-voltage single-phase or S2 in Figure 3) is ON. When the battery is charging, switch
distribution level. Simulations with inverters for DG systems S1 is ON, and the converter runs in the buck mode. When
indicate that the inverter must be connected as a static var com- the battery is providing power to the inverter, the S2 is ON,
pensator (STATCOM), where the reactive power is injected and the converter works in the boost mode. The specifications
of the dc–dc converter are shown in Table 1.
into the ac grid to regulate voltage at the PCC.
The inverter control algorithm has been used to decide
The STATCOM function block used in this inverter setup
is shown in Figure 4; this block measures the voltage sag (grid the buck-boost converter action. For the buck functionality,
voltage magnitude minus nominal voltage reference), if any on a current control loop was designed based on the average
the grid side, to calculate the necessary reactive power for volt- current method [18]. For the boost functionality, a voltage
age compensation. The PI controller keeps the sag voltage control loop was designed based on the K factor [19].
The buck-boost inductor nominal current IL is calcunull. There may be times in the inverter operation when it
may not be profitable to perform a voltage compensation, such lated as shown in
as during times when the real power spot pricing is higher;
P
during such times, a controlled switch (see Figure 4) is used
:
(1)
IL ¼
VB
for bypassing the voltage compensation functionality of the
inverter, and only real power is supplied to the grid.
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
Voltage Sag
1 p.u. Measurement
30
Islanding and Reclosure Function
The islanding and reclosure function performs the task of connecting (and islanding) the inverter-based DG to (and from)
Freq_Grid
Freq_VSI
Theta_Grid
Theta_VSI
Vdq_Grid
Vdq_VSI
IEEE
Standard 1547
Inputs Output
Islanding and
Reclosure Conditions
Brkr
5
Islanding and reclosure subsystem of the inverter-based DG setup.
TABLE 1. SPECIFICATIONS OF THE BIDIRECTIONAL
DC–DC BUCK-BOOST CONVERTER USED IN THE
INVERTER-BASED DG SETUP.
Quantity (Symbol)
Value
Buck-boost nominal power (P)
5 kW
DC link nominal voltage (Vdc )
350 V
Battery nominal voltage (VB )
192 V
Switching frequency (fs )
15 kHz
Resistance of battery bank (rB )
0.2 X
Inductor resistance (rL )
0.02 X
Resistance of the C1 and C2
capacitors (rC )
0.05 X
pffiffiffiffi
where
the
pffiffiffi
ffi xz ¼ xGC K , the xp ¼
xGC K ; and K3 is type 3 dc gain,
THE PI
as shown in Figure 3. K is a constant
calculated using the boost-phase adCOMPENSATORS
vance [20].
VB ðVdc VB Þ
Figure 7 shows the Bode plot of the
ARE USED TO
:
(2)
L¼
transfer function for the open-loop comfs Vdc 0:5IL
pensated system, Tboost (s) as per unit
GENERATE THE
system for the type 3 control method.
The buck capacitor, C1 , is designed
From Figure 6, it is observed that
to limit the ripple voltage at the battery
REFERENCES FOR
PM and the gain margin (GM) are infito 0.1% of the nominal battery voltage
A SINUSOIDAL
nite and 2.82 dB, respectively, which
(VB ), as shown in
indicates an unstable system. From
PULSEWIDTH
0:5IL
Figure 7, it is observed that PM is
C1 ¼
,
(3)
60.3° and GM is 9.72 dB, which is
8fs 0:001VB
MODULATOR.
satisfactory for a stable close system
because PM is almost in between 30°
and the boost capacitor, C2 , is intended
and 60° and GM is higher than 6 dB
to limit the ripple voltage at the dc link
to 0.1% of the nominal dc link voltage (Vdc ), as shown in [21]. This represents good robustness, which is important
for future implementation of the device in hardware protoPðVdc VB Þ
type. The prototyping of the device is beyond the scope of
(4)
C2 ¼
3 :
this article.
fs 0:001Vdc
The inductance (L) of the buckboost converter is designed to limit the
ripple current at the dc link to 50% of
the nominal current (IL ), as shown in
Current Control Loop
The type 2 controller G2 (s), shown in (5), has been used for this
method of current control of the buck-boost converter [20].
G2 (s) ¼
K2 (s þ xz )
:
s(s þ xp )
(5)
Voltage-Control Loop
The transfer function of the small signal dynamic boost
model, Tboost (s), can be expressed as [20]
1:3 3 104 s2 þ 0:22s þ 3:5 3 104
:
9:4 3 105 s2 þ 1:3 3 102 s þ 47:2
G3 (s) ¼
K3 (s þ xz )(s þ xz )
,
s(s þ xp )(s þ xp )
(6)
Open-Loop Bode
80
Magnitude (dB)
Figure 6 shows the Bode plot of the
open-loop transfer function for the uncompensated system, Tboost (s), as per
the unit proportional control method.
The zero-crossing gain frequency,
xGC , has been chosen as 1/20th of the
angular switching frequency (xs ¼
2pfs ). The phase margin (PM) has
been adopted at 60°. Thus, the
boost phase advance was calculated
to be 147° [20]. As the boost phase
advance is higher than 90°, a type 3
controller G3 (s) shown in (7) was
chosen [20] as
Phase (degree)
Tboost (s) ¼
60
40
20
0
360
315
270
225
180
102
103
104
Frequency (rad/s)
(7)
Open-loop Bode plot for the uncompensated system.
105
6
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
The high-frequency pole xp must be chosen close to the
switching frequency fs, the zero xz must lie in between one
half and one third of the resonant frequency, and the K2 is a
type 2 dc gain, as shown in Figure 3.
Premise of Operation of the Smart Inverter
The smart functionalities of the inverter are aimed at the
provision of real and reactive power support to local loads,
the provision of real and reactive power to grid loads up to
the rated capacity, the option to control voltage at the PCC
during voltage sags, and decision-making ability aided by
information of real-time pricing obtained through advanced
metering devices from the utility grid. Based on these functionalities, the inverter operation is governed by certain
rules, which determine the mode of operation, identified in
this article as supermodes and submodes. Depending on the
status connection to the grid as determined by compliance
with IEEE Standard 1547, there exist two supermodes:
stand-alone (S1) and grid-tied (S2) modes.
In supermode S1, i.e., the stand-alone mode, the
inverter is islanded (isolated) from the electric distribution
system, and it is subject to operation under one of the following three submodes, namely, s1, s2, and s3, depending
upon the available inverter active power (PINV ) and the
31
Magnitude (dB)
Phase (degree)
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
32
price to sell active power to grid ($PS ),
local active power demand (ZINV ), where
PINV represents the total power output
spot price to sell reactive power to
THE LOCAL LOAD
grid ($QS ), and a threshold value of
of the PV panels and the battery bank
output, and ZINV represents the sum of
the grid pricing of an electricity unit
SERVED BY THE
that will enable the consumer to decide
the primary VSI load plus secondary
which loads will be powered using the
VSI load.
INVERTER IS
inverter. In this case, it is assumed that
In submode 1 under supermode 1
the grid has infinite demand, i.e., the
(identified as S1s1), when PINV is lesser
MODELED AS
grid will purchase whatever the inverter
than ZINV , the power output of the PV
intends to sell.
panels and stored energy are not enough
TWO
In submode 1 under supermode 2
to supply the full demand of the local
COMPONENTS,
(identified as S2s1), when $QS is
load. In such circumstances, prioritization of local demand is effected, and
greater or equal than $PS , the inverter
PRIMARY AND
selected loads (primary VSI load) are
is controlled to provide voltage suppowered by the inverter. If, after the
port compensation to the grid. If there
SECONDARY VSI
selection of loads, there is any remainshould exist additional inverter capaing power from the PV panel, it will be
bility to provide real power, then the
LOADS.
directed to storage in the battery bank.
inverter is controlled so that ZINV is
Typically, such a redirection to the
supplied by the PINV , and any remainenergy storage depends on the level of
ing power is sold to the grid or stored
the state of charge (SOC). It is important to note that the in the battery bank [22].
SOC control of battery banks has not been developed by
In submode 2 under supermode 2 (identified as S2s2),
the authors.
when $QS is lesser than $PS , the inverter is controlled to fix
In submode 2 under supermode 1 (identified as S1s2), the reactive power reference to zero and to supply real power
when PINV is greater than ZINV , the available inverter to local loads ZINV , and any remaining power is sold to the
power is greater than the local demand. In this case, the grid or stored in battery banks [22].
excess power is routed to the battery banks for storage.
There is another submode under this supermode, i.e.,
In submode 3 under supermode 1 (identified as S1s3), S2s3. This is related to the option of powering local loads
PINV is equal to ZINV , wherein the available inverter power using the inverter versus the option of buying active
is equal to the local load demand. In this case, the inverter power from the grid when there is power available from
powers the local loads without storage. In this case, priori- DG. This can be chosen based on the comparison of the
tization of loads may be effected if there is a need to store real-time electricity pricing obtained from the grid ($PB )
some of the energy for later use.
with a threshold value, such as the marginal cost of
In supermode S2, i.e., the grid-tied mode, the inverter electricity production or a set customer preference, deis interconnected to the electric distribution system and noted as MCP. If $PB is lesser than MCP, then the inverter
is subject to operation under one of the following four load is supplied by the grid, and PINV is stored in a batsubmodes, s1, s2, s3, and s4, depending on the available tery storage for consumption or selling to the grid later,
inverter active power (PINV ), local active power demand possibly during isolation from the grid or when real-time
(ZINV ), and economic considerations for trading active pricing of electricity is conducive to profitability, respecand reactive power with the grid on a spot-pricing basis. tively. Or if there is no PINV available, then the electric
The variables for economic consideration include the spot energy can be purchased from the grid and stored in a battery for later use. If $PB is greater
than the MCP, then the inverter is so
controlled that ZINV is supplied by
PINV , and the stored energy in the
Open-Loop Bode
battery bank and any remaining
60
power is sold to the grid. The use of
40
the production marginal cost may
20
not be applicable in the case of PV
0
systems; however, if the installation
–20
considers customer preferences as
–40
input as in [22], then the above
rationale can be upheld as shown in
360
case study 1.
270
Submode 4 under supermode 2
(identified as S2s4) refers to the oper180
ation at an economic mode; this mode
90
is being proposed as an alternative
101
102
103
104
105
106
under the mode S2s3. This typically
Frequency (rad/s)
occurs when the cost of purchasing
7
electricity from the grid is relatively
Open-loop Bode plot for the compensated system.
expensive, i.e., such as described in
Voltage (p.u.),
Current (p.u.)
Inverter Side
Voltage (p.u.),
Current (p.u.)
Simulation Results
Based on the foregoing discussion,
three case studies describing the
smart functionalities of the inverter
are presented in this section. The
power and voltage bases used were
5 kWand 120 V, respectively. The gridside quantities are assumedly measured
at the PCC.
THE SMART
FUNCTIONALITIES
OF THE INVERTER
ARE AIMED AT THE
PROVISION OF
REAL AND
REACTIVE POWER
SUPPORT.
associated with this operation mode
are shown in Figure 8.
It is noticeable in Figure 8 that the
voltage and current waveforms on the
inverter side are shifted by 180°; this
implies that the inverter is buying electric
energy from the grid. As the voltage and
current are both at 1 per unit (p.u.), then
the inverter-active power is 5 kW, i.e., at
the inverter nominal power. At the grid
side, the voltage is 1 p.u. and the current
is 1.6 p.u. indicating that the grid-active
power is 8 kW. This corresponds to summation of the VSI loads (3 kW) and the
power supplied to the battery bank
(5 kW). The reactive power demands
at the inverter loads were set to zero.
1
Voltage (p.u.)
Current (p.u.)
0
–1
0.7
0.71 0.72 0.73 0.74 0.75 0.76 0.77 0.78 0.79
Time (s)
0.8
Grid Side
2
Voltage (p.u.)
Current (p.u.)
1
0
–1
–2
0.7
0.71 0.72 0.73 0.74 0.75 0.76 0.77 0.78 0.79
Time (s)
0.8
8
Current and voltage at inverter and grid side for Case 1.
1
0.5
0
–0.5
–1
0.58
Voltage (p.u.)
Current (p.u.)
0.6
0.62
0.64
0.66 0.68
Time (s)
0.70
0.72
0.74
0.76
Grid Side
Voltage (p.u.),
Current (p.u.)
In this simulation, the local demand
for an hour is 3 kW split into the
primary VSI load of 2 kW and the
secondary VSI load of 1 kW. As defined
in mode S2s3, the real-time electricity
pricing obtained from the grid ($PB ¼
US$1/kW) is lower than a set customer
preference (MCP ¼ US$2/kW). In this
case, the reference points of the
inverter are set such that the gridactive power is purchased to power
the local load and to charge the leadacid battery bank. Therefore, the
inverter is connected to the grid.
The bidirectional dc–dc buck-boost
converter is operated in the buck mode.
The voltage and current waveforms
Voltage (p.u.),
Current (p.u.)
Inverter Side
Case 1, Viz., S2s3
1
0.5
0
–0.5
–1
0.58
Voltage (p.u.)
Current (p.u.)
0.6
0.62
0.64
0.66 0.68
Time (s)
0.70
0.72
0.74
0.76
9
Current and voltage at the inverter and grid sides for Case 2.
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
S2s3; thus, the inverter is set up for
powering all its local loads while connected to the grid. However, in such a
case, if the inverter power is not
enough to supply its loads, the customer has an option to operate at an
economic mode, i.e., effecting prioritization of primary VSI loads and
shedding the secondary VSI loads
(as in S1s1). This submode is achieved
in the simulation by the use of a
flag variable; if the flag is set to zero,
it operates in the economic mode
using variable loads; and if the flag is
set to one, it operates in an always
supplying loads mode, as the inverter
power is not enough, power from the
grid is purchased to supply the
remaining loads.
It is pertinent to note that this
study does not consider the time line
for powering loads or charging storage, i.e., in terms of energy demanded
and supplied. Consideration of the
energy supplied and demanded is
inherently tied to hours of available
sunlight and to the PV installation
capacity and battery storage [22].
Since this article is aimed at describing the smart functionalities of an
inverter, the authors feel that the use
of power ratings is adequate for a
proof of concept.
33
as indicated by the grid voltage (1 p.u.) and grid current
(0.2 p.u.).
At 0.65 s, the inverter begins operating in the islanded
mode with the inverter voltage and current at 1 p.u. and
0.8 p.u., respectively; therefore, the inverter-active power
is 4 kW (3 kW from PV panels and 1 kW from batteries).
The grid-active power is 0.0 W (grid current is 0.0 p.u.).
As in Case 1, the reactive power demand is taken as zero
in this case as well. The transients seen around 0.65 s in
Figure 9 are caused by transition from the grid-tied mode
to the islanded mode.
Grid Voltage Magnitude
Voltage Magnitude (p.u.)
1.1
1
0.9
0.8
0.5
1
1.5
2
Time (s)
2.5
3
3.5
In this simulation, the primary and secondary VSI loads are
both set at 1 kW. A voltage sag was induced on the grid
side between 0.5 and 1.2 s. Operating in the S2s1 mode,
Voltage magnitude on the grid side for Case 3.
the inverter provides voltage support to the PCC. This phenomenon, including the transient performance, can be
observed in Figure 10. The grid-voltage magnitude (Vmag )
Case 2, Viz., S1s1
In Case 2 (S1s1), the primary VSI load is 4 kW, and the is kept at 1 p.u. by the action of the STATCOM function
secondary VSI load is 2 kW. The PV panels are set up to block of the inverter. The voltage and current waveforms assoprovide 3 kW. The inverter is initially in the grid-tied ciated with this mode of operation are shown in Figure 11.
Notice that the inverter is not islanded by the islanding
mode until 0.65 s when the islanding mode is reinforced,
possibly because of a far away fault in the grid. In the and reclosure subsystem because the magnitude and duraislanded mode, the PV panels are not able to supply tion of the voltage sag do not fall within the limits recomenough power to the loads, thus the bidirectional dc–dc mended by IEEE Standard 1547 [4].
From the simulation start-up to 0.5 s, the inverterbuck-boost converter will operate in the boost mode, and
the lead-acid battery bank will provide 2 kW. As the load active power is 5 kW, and the grid-active power is 3
values are higher than 5 kW, the inverter (or the kW since 2 kW is consumed by the VSI loads. The surcustomer) must buy 1 kW of active power. Initially, in the plus is injected to the grid side. When the grid has a voltgrid-tied mode, the inverter (or customer) purchases this age sag from 0.5 to 1.2 s, the inverter voltage and current
1 kW from the grid. However, when the inverter func- are both at 1 p.u. and 32.4° out of phase (see Figure 11);
tions in the islanded mode, the secondary VSI load drops this indicates that the inverter-active power is 4.2 kW
as described in mode S1s1. The voltage and current wave- and inverter-reactive power is 2.7 kvar. The grid-active
forms associated with this mode of operation are shown power is 2.2 kW, and grid-reactive power is 2.7 kvar.
So the inverter is selling active and reactive powers to
in Figure 9.
When the inverter is connected to the grid, the inverter grid, and it is compensating voltage sag by supplying
voltage and current values are both 1 p.u., indicating that reactive power to PCC. After 1.2 s, the voltage sag event
the inverter-active power is 5 kW (3 kW from PV panels on the grid is assumed to be cleared; however, the inverter
and 2 kW from batteries). The grid-active power is 1 kW keeps providing reactive power; hence, the voltage
magnitude on the grid side increased
quickly. However, the STATCOM
action decreased the inverter-reactive
power to 0.0 var and consequently
Inverter Side
1
increased the inverter-active power to
Voltage (p.u.)
5 kW, helping the stabilization of
Current (p.u.)
Vmag at 1 p.u. Thus, the postevent
0
grid-active power is 3 kW and the
postevent grid-reactive power is 0 var.
–1
It is observed in Figure 11 that the
1.1 1.11 1.12 1.13 1.14 1.15 1.16 0.17 1.18 1.19 1.20
voltage waveform on the inverter side
Time (s)
leads the current waveform; this is
Grid Side
caused by the injection of reactive
1
Voltage (p.u.)
power from inverter to grid side. It is
Current (p.u.)
also noticeable that the voltage and
0
current waveforms on the grid side
are out of phase by approximately
–1
231°. This phase difference is not
1.1 1.11 1.12 1.13 1.14 1.15 1.16 0.17 1.18 1.19 1.20
exactly 180° as in Case 1 because in
Time (s)
this case the inverter injects (selling)
11 both active and reactive powers to
the grid.
Current and voltage at the inverter and grid sides for Case 3.
Voltage (p.u.),
Current (p.u.)
Voltage (p.u.),
Current (p.u.)
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
10
34
Case 3, Viz., S2s1
Conclusions
This article describes the topology, control philosophy,
operational algorithm, and simulation results of a voltage-source inverter, interfacing a PV-based DG system to
the grid, with certain smart functionalities such as the
ability to supply real and reactive power to local loads,
supply real and reactive power to other utility loads up to
rated capacity of the inverter, store energy in a battery
bank, provide voltage support at the PCC, schedule loads,
and provide control options to the consumer based on
near real-time electricity information obtained from the
utility through advanced metering devices. This VSI is
capable of automatically choosing the operation mode
based on a set of super and submodes corresponding to
system conditions and real-time pricing of electricity.
The smart functionalities are deemed to look beyond the
recommendations of the current national technical standard for interconnecting DG sources to grid, IEEE Standard 1547, i.e., providing voltage support to PCC, thus,
offering an ancillary service in case of low-voltage scenarios. This represents one of the tenets of the SGI, namely,
enabling active participation of consumers in the demand
response using timely information and control options.
Several case studies are presented to illustrate the functionalities of the proposed device.
References
[1] R. S. Thallam, S. Suryanarayanan, G. T. Heydt, and R. Ayyanar,
“Impact of interconnection of distributed generation of electric distribution systems—A dynamic simulation perspective,” in Proc. 2006
IEEE Power Engineering Society General Meeting, pp. 1–8.
[2] “Smart Grid, Title XIII,” in Proc. Energy Independence and Security Act of
2007 (EISA07), 110th Congr. of the United States, Dec. 2007.
[3] E. Ortjohann, M. Lingemann, A. Mohd, W. Sinsukthavorn, A.
Schmelter, N. Hamsic, and D. Morton, “A general architecture for
modular smart inverter,” in Proc. 2008 IEEE Int. Symp. Industrial
Electronics (ISIE), July 2008, pp. 1525–2530.
[4] IEEE Standard for Interconnecting Distributed Resources with Electric Power
Systems, Standard 1547-2003.
[5] S. Chakraborty, B. Kroposki, and W. Kramer, “Advanced power electronic interfaces for distributed energy systems,” Tech. Rep. NREL/
TP-55044313, Nov. 2008.
[6] R. Carnieletto, D. B. Ramos, M. G. Simões, and F. A. Farret,
“Simulation and analysis of DQ frame and P+Resonant controls
for voltage source inverter to distributed generation,” in Proc.
2009 Brazilian Power Electronics Conf., Sept. 2009, vol. 1,
pp. 104–109.
[7] A. Mohd, E. Ortjohann, W. Sinsukthavorn, M. Lingemann, N. Hamsic,
and D. Morton, “Supervisory control and energy management of an
Renata Carnieletto and Felix A. Farret are with the Universidade Federal de Santa Maria in Santa Maria, Brazil. Danilo
Iglesias Brandão is with the Universidade Estuadal Paulista
in São Paulo, Brazil. Siddharth Suryanarayanan is with Colorado State University, Colorado. Marcelo G. Simões (Msimoes@mines.edu) is with Colorado School of Mines in Golden,
Colorado. Suryanarayanan and Simões are Senior Members of
the IEEE. This article first appeared as “A Multifunctional
Single-Phase Voltage-Source Inverter in Perspective of the Smart
Grid Initiative” at the 2009 IEEE Industry Applications
Society Annual Meeting.
IEEE INDUSTRY APPLICATIONS MAGAZINE SEPT j OCT 2011 WWW.IEEE.ORG/IAS
Acknowledgments
This work was supported in part by the Colorado Clean
Energy Project through the Colorado Renewable Energy
Collaboratory and the Xcel Energy Foundation and in part
by the U.S. National Science Foundation under grant
0757956. D. Brandão gratefully acknowledges F. Marafão
and D. Colón of the Universidade Estadual Paulista for
their insightful comments about this article and for their
help in designing the controller parameters described in
the “The Smart Functionalities of the Inverter: Bidirectional DC–DC Buck-Boost Converter” section.
inverter-based modular smart grid,” in Proc. 2009 IEEE Power and Energy
Society Power Systems Conf. and Exposition (PSCE), Mar. 2009, pp. 1–6.
[8] A. Roshan, R. Burgos, A. C. Baisden, F. Wang, and D. Boroyevich,
“A D-Q frame controller for a full-bridge single phase inverter used
in small distributed power generation systems,” in Proc. 2007 IEEE
Applied Power Electronics Conf. and Exposition (APEC), Feb. 2007,
pp. 641–647.
[9] U. A. Miranda, M. Aredes, and L. G. B. Rolim, “A DQ synchronous
reference frame current control for single-phase converters,” in Proc.
2005 IEEE Power Electronics Specialists Conf. (PESC), June 2005,
pp. 1377–1381.
[10] S. K. Chung, “Phase-locked loop for grid-connected three-phase
power conversion systems,” IEE Proc. Electr. Power Applicat., vol. 147,
no. 3, pp. 213–219, 2000.
[11] M. S. Pádua, S. M. Deckmann, and F. P. Marafão, “Frequency-adjustable positive sequence detector for power conditioning applications,”
in Proc. 2005 IEEE Power Electronics Specialists Conf. (PESC), June
2005, pp. 1928–1934.
[12] F. P. Marafão, S. M. Deckmann, J. A. Pomilio, and R. Q. Machado.
“A software-based PLL model: Analysis and applications,” in Proc.
XV Congresso Brasileiro de Automática (CBA), Gramado, 2004. v.
único.
[13] Xcel Energy Corporation. (2009, Jan.). SmartGridCity. [Online].
Available: http://smartgridcity.xcelenergy.com/
[14] M. Simões and F. Farret, Integration of Alternative Source of Energy, 1st
ed. Hoboken, NJ: Wiley, 2006, pp. 129–158.
[15] E. Lefter, L. M. Constantinescu, and C. Stoica. (2009, Nov.) The
experimental study of lead acid batteries for an electric hybrid drive
stand. [Online]. Available: http://electroinf.uoradea.ro/reviste%20CSCS/
documente/JCSCS_2008/JCSCS_2008_31_Lefter_1.Pdf
[16] Mathworks Inc. SimPowerSystems—Model and simulate electrical
power systems. [Online]. Available: http:www.mathworks.com/products/
simpower/
[17] B. Kroposki, C. Pink, R. DeBlasio, H. Thomas, M. Simões, and P. K.
Sen, “Benefits of power electronic interfaces for distributed energy
systems,” in Proc. 2006 IEEE Power Engineering Society General Meeting,
pp. 1–8.
[18] J. Sun and R. M. Bass, “Modeling and practical design issues for average current control,” in Proc. 1999 IEEE Applied Power Electronics Conf.
and Exposition (APEC), Mar. 1999, vol. 2, pp. 980–986.
[19] H. D. Venable, “The k-factor: A new mathematical tool for stability
analysis and synthesis,” in Proc. Powercon 10, San Diego, CA, Mar.
1983.
[20] R. H. Rosenback. (2009, Nov.). Conversor CC-CC bidirecional
buck-boost atuando como controlador de carga de bateria em sistema fotovoltaico. Master’s thesis Programa de Pós-graduação em
Engenharia Eletrica, Universidade Federal de Juiz de Fora MG
Brazil. [Online]. Available: http://www.ufjf.br/ppee/files/2008/12/
211038pdf
[21] K. Ogata, Engenharia de Controle Moderno, 4th ed. São Paulo, Brazil:
Prentice-Hall, 2003, pp. 444–448.
[22] J. Armas and S. Suryanarayanan, “A heuristic technique for scheduling a customer-driven residential distributed energy resource
installation,” in Proc. IEEE 15th Intelligent System Applications to Power
Systems (ISAP), Curitiba, Brazil, 2009, pp. 1–7.
35
Download