1934 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 50, NO.

3, MAY/JUNE 2014

A Method of Seamless Transitions Between

Grid-Tied and Stand-Alone Modes of Operation for
Utility-Interactive Three-Phase Inverters
David S. Ochs, Student Member, IEEE, Behrooz Mirafzal, Senior Member, IEEE, and
Pedram Sotoodeh, Student Member, IEEE

Abstract—A method for the seamless transition of three-phase inverters. During this transition, the magnitude and frequency
inverters switched between grid-tied and stand-alone modes of of the voltage across the local loads may experience large
operation is presented in this paper. In this method, only the transients. Thus, the inverter control system needs to implement
inverter current and voltage sensors are utilized, and no con-
trol over the grid-side static transfer switch is needed. The pre- a mechanism to provide seamless transitions between the two
sented method contains two strategies for grid-tied-to-stand-alone modes of operation, namely, grid-tied and stand-alone (island-
and stand-alone-to-grid-tied transitions. In the stand-alone-to- ing) modes. In other words, grid-tied inverters should be able to
grid-tied transition strategy, a novel algorithm is presented for secure power delivery with minimal disturbance to their local
estimating the grid angle nearly instantaneously, which allows loads. Seamless transition is one of the technical challenges in
the three-phase inverter to respond very quickly if the grid and
point-of-common-coupling voltages are out of phase. This fast the renewable energy system technology, and therefore, many
response allows the inverter to effectively eliminate the transient industrial and academic investigations have been performed
overcurrent that would normally occur if it was connected to the in recent years. A majority of reported work on seamless
grid without first being synchronized. The fast response also allows transitions has made use of a controllable static transfer switch
the inverter to return to normal operation very quickly after such (STS) [2]–[8], while two sets of voltage and current sensors
an event. The strategy for the seamless transition from grid-tied
to stand-alone mode is also presented. These strategies have been are utilized in both sides of the STS [see Fig. 1(a)]. In [2],
verified through experiments, and the results are presented in this a virtually seamless transition to and from islanded operation
paper. is demonstrated using a controllable STS with sensors on both
Index Terms—Grid-tied and stand-alone modes, seamless tran- sides. Many other investigators have expanded on this concept
sition, utility-interactive inverters. in various ways, all using a controllable STS. In [3], the voltage
magnitudes and phase angles of both sides of the STS are mea-
I. I NTRODUCTION sured, and their differences are fed back into the inverter control
system to become resynchronized for the inverter reconnection

T HE role of power electronics in electric power generation

has increased greatly in recent years. Wind turbines, pho-
tovoltaic (PV) panels, fuel cells, and other renewable resources
to the grid. In [4], an islanding detection method for three-
phase grids is presented, which can provide the capability of
an intelligent load shedding or intentional-islanding operation.
produce energy in forms not suitable for direct connection to the In [5] and [6], similar techniques for single-phase systems
power grid. Such systems often use solid-state inverters to inter- are implemented through detection of voltage magnitudes and
face with the grid. In a highly adaptable grid, utility-interactive frequencies at the grid and inverter sides. Moreover, the inverter
inverters should be able to provide power to both local loads and reference voltage is readjusted at the inverter positive zero-
the power grid [1]. Under abnormal conditions, the grid can be crossing voltage waveform for a stand-alone-to-grid-tied mode
disconnected, and the local loads should be energized by the change. For a grid-tied-to-stand-alone mode change, though,
first, the reference current signal is set to zero at the zero
Manuscript received April 15, 2013; revised July 8, 2013; accepted crossing of the grid current, and then, the inverter reference
August 23, 2013. Date of publication September 20, 2013; date of current voltage is adjusted [6]. In [7], an active technique is presented
version May 15, 2014. Paper 2013-IPCC-191.R1, presented at the 2013 IEEE
Applied Power Electronics Conference and Exposition, Long Beach, CA, USA,
for single-phase systems to determine the state of the grid,
March 17–21, and approved for publication in the IEEE T RANSACTIONS which also uses an STS to achieve mode transitions.
ON I NDUSTRY A PPLICATIONS by the Industrial Power Converter Committee However, there are several other techniques in which only
of the IEEE Industry Applications Society. This work was supported by the
Electrical Power Affiliates Program at Kansas State University.
one set of sensors is needed and the STS may not be required,
D. S. Ochs is with the Advanced Technology Center, General Motors as shown in Fig. 1(b). For example, a method of transitions
Powertrain, General Motors Corporation, Torrance, CA 90505 USA (e-mail: between grid-tied to stand-alone modes of operation is reported
david.ochs@gm.com).
B. Mirafzal and P. Sotoodeh are with the Department of Electrical and in [9]. The technique is based on monitoring the currents that
Computer Engineering, Kansas State University, Manhattan, KS 66506 USA flow in the inverter and shutting the inverter off (coasting) until
(e-mail: mirafzal@ksu.edu). a phase-locked loop (PLL) could converge. Once the PLL has
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. converged, it can provide the grid angle to the inverter’s control
Digital Object Identifier 10.1109/TIA.2013.2282761 system, and normal operation can be resumed. In [10], a similar

0093-9994 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
Seeuse
Authorized licensed http://www.ieee.org/publications_standards/publications/rights/index.html for more
limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE information.
Xplore. Restrictions apply.
OCHS et al.: METHOD OF SEAMLESS TRANSITIONS BETWEEN GRID-TIED AND STAND-ALONE MODES 1935

the transitions between operating modes, it is assumed that the

dc-bus voltage at the inverter’s input remains constant. The in-
verter is controlled using a space-vector pulsewidth-modulated
(SVPWM) switching pattern. The only required sensors are the
two current transformers (CTs) and two potential transformers
(PTs), shown in Fig. 1(b). If the neutral point of the local load
is grounded, a third CT will be needed to account for possible
unbalanced conditions. In general, the local load can vary;
however, herein, it is assumed that the inverter can fully provide
the maximum load in stand-alone mode. A circuit breaker is
placed at the point of common coupling (PCC) in order to
connect the system to the grid. Moreover, the inverter controller
has no control over the circuit breaker nor any knowledge of its
status, i.e., open or closed.
The inverter has two different control modes: 1) voltage-
controlled mode for stand-alone operation and 2) power-
controlled mode for grid-tied operation. Each controller is
Fig. 1. Single-line schematics of possible configurations for three-phase grid- formulated in a dq rotating reference frame.√ If the inverter
tied inverters. (a) Grid-tied inverter with a controllable STS. (b) Grid-tied output voltages are sinusoidal [i.e., vab (t) = 2VLL cos(ωt)],
inverter without a controllable STS.
a PLL provides a reference angle θ that is zero at the peak
of vab whether the grid is present. Therefore, the synchronous
transition method is proposed without using a controllable STS dq transform can be performed on the line-to-line voltages
and with significantly distorted currents up to five cycles after measured at the PCC as follows:
transitions. In [11] and [12], a technique based on adding a ⎡ ⎤
  vab  
seventh harmonic component into the inverter reference signal vsLd 0
⎣
= Tdq vbc = ⎦ √ . (1)
is presented. In this technique, in grid-tied conditions, the local vsLq 2VLL
load voltage is dictated by the grid, and thus, its waveform is vca
not altered by the injected harmonic. However, in stand-alone Herein, the synchronous dq transform is given by
conditions, the voltage at the local loads follows the waveform    
2 sin(θ) sin θ − 2π sin θ + 2π
of the harmonic component injected by the inverter. In [13], a Tdq =  3 3 (2)
technique based on the grid impedance estimation is presented 3 cos(θ) cos θ − 2π 3 cos θ + 2π3
for mode transitions regardless of whether the grid is connected. where θ = ωt. Notice that, for line-to-line voltages, vab +
Typically, the major challenge in these techniques is reducing vbc + vca = 0, but the same cannot be applied for inverter phase
the calculation time. voltages due to a nonzero common-mode voltage magnitude at
The main contribution of this paper is to present a method- the inverter ac side.
ology that allows a three-phase grid-tied inverter to move from 1) Stand-Alone Control: In stand-alone mode, the inverter
grid-tied to stand-alone mode and vice versa, with smaller tran- acts as a voltage source to maintain the load voltage at its
sient times and overcurrent magnitudes than those of previously nominal value. From (1), it is clear that the line-to-line rms
reported techniques with only one set of sensors. The proposed value VLL can be controlled in the arbitrary dq reference
∗
√ frame
∗
technique can meritoriously be implemented for three-phase by holding vsLd to zero and controlling vsLq as vsLq = 2VLL ,
utility-interactive inverters. ∗
where VLL √ is the nominal line-to-line rms voltage of the system
In addition to the introduction, this paper contains five sec- and the 2 factor is included because (1) is written based on
tions. In Section II, a high-level description of the system, as peak values. An asterisk superscript denotes a reference value
well as voltage and power control schemes, is provided. In throughout this work. In practice, a closed-loop controller has
Section III, the strategies for transitioning between the different to be used to regulate the inverter terminal voltage magnitude,
operating modes are presented, along with a novel method with the block diagrams shown in Fig. 2. As shown in Fig. 2,
for quickly estimating the grid angle. Experimental results ∗
vsLq is fed forward to the output of the controller to achieve fast
confirming the validity of the proposed strategies are presented ∗ ∗
voltage control. The outputs of the controllers viLd and viLq
in Section IV. Lastly, a conclusion section completes this paper. are the set points for the inverter terminal voltages. These are
realized through the widely used SVPWM scheme, the details
of which may be found in several references [14], [15].
II. S YSTEM D ESCRIPTION AND C ONTROL S CHEMES
2) Grid-Tied Control: In grid-tied operation, the inverter
The case study system consists of a dc power source rep- acts as a voltage-controlled current source. As such, the active
resenting the inverter dc-bus voltage, a three-phase insulated- and reactive power outputs are controlled in the synchronous
gate bipolar transistor (IGBT)-based bridge inverter, an LCL reference frame by controlling the values of viLq and viLd . To
output filter, a local load, and the grid. The two possible circuit illustrate this implementation, consider Fig. 1 assuming that the
schematics for grid-tied inverters are shown in Fig. 1. Because filter capacitor has been neglected and the inverter has been
this work is concerned only with very small time spans around approximated as a three-phase voltage source. The inverter
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
1936 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 50, NO. 3, MAY/JUNE 2014

Fig. 2. Stand-alone voltage control block diagram.

neutral point is not grounded due to the nonzero common-

mode voltage in these systems. Thus, KVL for the line-to-line
voltages yields
⎡ ⎤ ⎡ ⎤⎡ ⎤
viab R −R 0 ia
⎣ vibc ⎦ = ⎣ 0 R −R ⎦ ⎣ ib ⎦
Fig. 3. Simplified grid-tied power control block diagram.
vica −R 0 R ic
⎡ ⎤⎡ ⎤ ⎡ ⎤
L −L 0 pia vsab SVPWM similar to stand-alone mode. In order to find the
+⎣ 0 L −L ⎦ ⎣ pib ⎦ + ⎣ vsbc ⎦ (3) necessary set points, consider the virtual line-to-line currents
−L 0 L pic vsca in the dq synchronous reference frame
⎡ ⎤
  iab  
where p is the time derivative operator and R and L are the iLd √ sin(ϕ)
⎣ ⎦
= Tdq ibc = 6IL (7)
equivalent resistance and inductance values between the grid iLq cos(ϕ)
and inverter terminals, respectively. Herein, virtual line-to-line ica
currents are defined as follows:
where iab , ibc , and ica are the virtual line-to-line currents
⎡ ⎤ ⎡ ⎤⎡ ⎤
iab 1 −1 0 ia defined in (4) and Tdq is the same dq transform matrix used
⎣ ibc ⎦ = ⎣ 0
Δ
1 −1 ⎦ ⎣ ib ⎦ . (4) in (1). If the desired active and reactive powers, i.e., P ∗ and Q∗ ,
ica −1 0 1 ic are known, then the desired line current and its phase shift with
respect to the grid voltage are calculated from
It should be noted that the virtual line-to-line currents are √ ∗
herein defined to simplify the mathematics. Assuming that P ∗ + jQ∗ = 3VLL IL∗ ejϕ (8)
the
√ fundamental components of √ line-to-line voltages vab (t) =
2VLL cos(ωt) and vbc (t) = 2VLL cos(ωt − 2π/3), as well where VLL is √fixed by the grid. Substituting I ∗ =
√
as line currents i = 2I cos(ωt − π/6 − ϕ) and ib = ( P ∗2 + Q∗2 )/( 3VLL ) and ϕ∗ = arctan(Q∗ /P ∗ ) in
√ a L
2IL cos(ωt − 2π/3 − π/6 − ϕ), are measured, the virtual (7), i∗Ld and i∗Lq can be calculated. The inverter voltage
∗ ∗
line-to-line currents reference values viLd and viLq can be obtained using (6),
√ can be obtained as defined in (4), e.g.,
iac = ia − ic = 6IL cos(ωt − ϕ). √ It is clear that the virtual neglecting the derivative terms. These can then be used to set
line-to-line currents are scaled by 3 and shifted by +π/6 the SVPWM switching pattern. In order to improve the system
with respect to the line current, which is the same relationship robustness, i.e., reducing sensitivity of the controller to the
that exists between line-to-line and line-to-neutral voltages. circuit parameters, proportional–integral controllers must be
Applying (4) to (3) gives the following equation: implemented (see Fig. 3). It should be noted that the desired
⎡ ⎤ ⎡ ⎤ ⎡ ⎤ ⎡ ⎤ active power P ∗ is set by the maximum power of the input
vi,ab iab piab vs,ab dc source (e.g., PV systems and battery) and the maximum
⎣ vi,bc ⎦ = R · I ⎣ ibc ⎦ + L · I ⎣ pibc ⎦ + ⎣ vs,bc ⎦ (5) allowable power that can be injected to the grid, whichever
vi,ca ica pica vs,ca is lower. The desired reactive power Q∗ is typically set to
zero for small systems; however, for large systems, e.g., PV
where I = I3×3 is an identity matrix. Equation (5) can be farms, it is set by the grid demand for reactive power, which
referred to the dq synchronous reference frame as follows: indeed must be below Qmax = Srated 2 − P ∗2 , where Srated
       is determined by the inverter rated (nominal) values, i.e.,
viLd R + Lp −ωL iLd v
= + sLd (6) Srated = Vrated Irated .
viLq ωL R + Lp iLq isLq

where ω is the angular frequency of the grid. In order to control

III. P ROPOSED M ODE T RANSITION S TRATEGIES
the active and reactive powers injected into the grid, reference
values for i∗Ld and i∗Lq must be obtained. They can then be The inverter has three possible modes of operation: 1) stand-
∗ ∗
used in (6) to find viLd and viLq , which are realized through alone mode; 2) grid-tied mode; and 3) coast mode. In Fig. 4, one
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
OCHS et al.: METHOD OF SEAMLESS TRANSITIONS BETWEEN GRID-TIED AND STAND-ALONE MODES 1937

Fig. 4. Mode transition conceptual block diagram; mode = 0, −1, and +1 stand for coast (inverter turn-off), stand-alone (voltage control in Fig. 2), and grid-tied
(power control in Fig. 3) modes of operation, respectively.

can see how the proposed technique determines the inverter’s connected to the grid, ω0 and θ0 will be initialized to 120π and
mode of operation. In stand-alone mode, the inverter regulates zero, respectively.
the voltage at the PCC because the grid is disconnected (see As shown in Fig. 4, in stand-alone mode, the inverter intro-
Fig. 2). In grid-tied mode, the inverter operates as a power duces a dither signal, i.e., ωr = ω̃g + Δω, into the reference
source to inject certain amounts of active and reactive powers frequency of the voltage at the PCC. The magnitude of the
to the grid (see Fig. 3). In coast mode, all of the IGBTs are dither is ±0.08 Hz and the frequency of the dither signal is
turned off, which is necessary for proper transitions between 0.4 Hz; however, the magnitude of the dither must not adversely
stand-alone and grid-tied modes of operation. The direct path affect any load that is connected at the PCC. This dither signal
between coast and grid-tied modes is shown by a dashed line in is used in the transition from stand-alone mode to grid-tied
gray color, shown in Fig. 4. mode. As previously mentioned, the inverter captures the grid
The transition from grid-tied mode to stand-alone mode is the frequency and angle when a transition is made from grid-
simplest to execute. According to IEEE 1547-2003, the service tied mode to stand-alone mode. Therefore, in the event of a
voltage of a system should always remain between 0.918 and short disconnection, the grid and PCC voltages may still be
1.057 p.u., i.e., 191 < V < 220 for a 208-VLL system, and synchronized when the inverter is reconnected to the grid. For
the frequency should stay between 59.3 and 60.5 Hz [1]. If a longer disconnection time interval, the PCC and grid voltages
the inverter senses that either the voltage or the frequency has may or may not be synchronized. If the two are synchronized
surpassed the aforementioned ranges while it is in grid-tied prior to reconnection, there will not be an obvious transient
mode, the inverter controller switches to stand-alone mode, as when reconnection occurs. Rather, the frequency at the PCC
shown in Fig. 4. In stand-alone mode, the inverter still must will no longer follow the dither signal because it will be set by
have a reference angle for the control scheme. This angle can the grid. In other words, if the frequency error, defined as eω =
be interpolated from the grid conditions before the inverter is |ωr − ω̃g |, is greater than the threshold for mode transition
disconnected as follows: Δωth , the inverter shifts to grid-tied mode. Also, the PCC
voltage magnitude may change from its value in stand-alone
θr = ωr dt + θ0 (9) mode upon reconnection if the stand-alone voltage set point and
the grid voltage differ. Also, if the voltage strays too far from
its reference value while the inverter is operating in stand-alone
where ωr = ω̃g (tgrid−off ) is the grid angular frequency and mode, the inverter controller shifts to grid-tied mode, as shown
θ0 = θg (tgrid−off ) is the grid angle at the instant that the inverter in Fig. 4.
switched to stand-alone mode tgrid−off . Because the grid fre- If the PCC voltage and the grid voltage are not synchronized
quency may not be stable during a grid interruption, the inverter upon connection, this voltage difference will cause a large
stores the value of ω̃g in the microcontroller memory from transient current to flow unless an immediate action is taken to
several cycles. Thus, the inverter can use a reliable ω̃g value in prevent it. The basis of this action in the proposed technique
the event of a grid failure or disconnection. When the inverter is a voltage angle estimation technique in addition to the
senses a sudden change in voltage or frequency, it stores the
√ estimator. Assuming that the line–line voltage vab (t) =
PLL
present grid angle θ0 in memory and begins integrating ωr 2VLL cos(θg ) is the reference voltage, a first estimation for
according to (9). If the inverter controller switches to stand- the voltage angle θg can be obtained from
alone mode, θr is then considered to be the reference angle for √
use in the control system. If the inverter is started without being θ̃g1 = cos−1 (vab / 2VLL ) (10)
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
1938 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 50, NO. 3, MAY/JUNE 2014

where 0 < θ̃g1 < π using the cos−1 function. However, it is

possible that, for any given θ̃g1 , the actual voltage angle at the
PCC is −θ̃g1 . In order to find the actual reference angle, the
sign of sin(θg ) must be examined. If sin(θg ) is positive, then
the angle will be in quadrant I or II of a complex plane, meaning
that the actual angle is θ̃g1 . If sin(θg ) is negative, the angle
will be in quadrant III or IV, and the correct value is −θ̃g1 ,
accordingly. This can be mathematically described as

θ̃g = sgn (sin(θg )) · θ̃g1 . (11)

It should be noted that the value of θg is unknown at this

stage; however, sgn(sin(θg )) can be easily found from the
measured
√ line-to-line voltages. Again, assuming that vab (t) =
2VLL cos(θg ), one can write the following:
Fig. 5. Experimental test setup.
√
vab + 2vbc = 2VLL sin(θg ). (12)

Thus, using the measured voltages, the sign of sin(θg ) can be IV. E XPERIMENTAL S ETUP
identified as sgn(sin(θg )) = sgn(vab + 2vbc ), i.e.,
The proposed techniques were tested on a laboratory-scale
√ 1.2-kVA 208-V 60-Hz three-phase system. The experimental
θ̃g = sgn(vab + 2vbc ) · cos−1 (vab / 2VLL ). (13)
test setup is shown in Fig. 5. A dSPACE CLP1103 system
In both stand-alone and grid-tied modes, the inverter refer- hosted the controllers and feedback signals from the sensors.
ence angle θr (for the control system) is approximately equal The inverter controllers were run from a Simulink model linked
to θ̃g , i.e., θ̃g ∼
= θr , considering the inherent phase shift caused to the dSPACE system. The local load was purely resistive, fed
by the voltage sensors and feedback path delays. However, the through an LCL filter, and the inverter pulsewidth-modulation
only situation in which θr can differ from θg by a significant (PWM) frequency was 5 kHz. The inductance and capacitance
margin is the instant that the inverter is connected to the grid. values of the LCL filter in Fig. 1, i.e., Lf 1 , Lf 2 , and Cf , were
In this case, the error between θr and θg , defined as eθ , will be 10 mH, 5 mH, and 10 μF, respectively. It should be noted that
greater than the threshold set for mode transition Δθth . When the LCL filter resonance frequency, i.e., fr = 1/2π Lf Cf ,
the inverter senses this condition, it immediately enters coast where Lf = Lf 1 Lf 2 , should be much smaller than the PWM
mode in order to minimize transient overcurrent, as shown in frequency [19]. The capacitance value of Cf is limited by the
Fig. 4. This strategy would not be possible if a PLL scheme reactive power absorbed by the circuit inductances at rated
were used instead of θ̃g because even a well-tuned PLL scheme conditions. This means that, for a Y-connected capacitor bank,
2 2
needs time to converge to the correct grid angle [16]–[18]. approximately Cf Vph,rated < (Lf 1 + Lf 2 + Lg )Irated , where
In this technique, a time series of estimated angles, e.g., Lg is the grid inductance value. Meanwhile, the grid was
θ̃g (tk−1 ), θ̃g (tk ), and θ̃g (tk+1 ), can be scanned to minimize connected to the load through an isolated 1:1 transformer, as
the impact of measurement noises and voltage disturbances shown in Fig. 5. The inverter control system had no feedback
on the process of the grid angle estimation. Using the set of signals from the grid while it was in stand-alone mode. During
estimations, as θg is a linear function over each (0 2π)-radian transitions, it only had the information gleaned from the two
interval, any sudden change (except at zero and 2π) between PTs and two CTs at the PCC.
the estimated angles is considered outlier data. Furthermore, The experimental results were acquired with a LeCroy
in the proposed method, (13) is used only for several cycles Waverunner 64XI oscilloscope equipped with ADP305 differ-
until the PLL scheme is converged. ential high-voltage probes (100-MHz bandwidth) and CP031
Notice that a small mismatch between the frequencies of the high-current probes (50-MHz bandwidth), while the control
inverter reference signal and the grid during stand-alone mode low-voltage signals were collected with a LeCroy WaveAce 214
could make the inverter reference and the grid angles out of oscilloscope.
phase. The inverter also switches to coast mode if it senses
a certain level of current flowing at the PCC. Both possible
V. E XPERIMENTAL R ESULTS
transition paths to coast mode are shown in Fig. 4. Because θ̃g
provides a nearly instantaneous estimate of the grid angle, coast In these experiments, a signal called “Mode” indicates the
mode need only be used for a short duration. In experimental inverter operating mode. The values of the Mode signal are
results, the inverter was programmed to stay in coast mode for −1, 0, and +1 for stand-alone, coast, and grid-tied modes
half a cycle, i.e., 8.33 ms. This predetermined time margin can of operation, respectively. The results of five different mode
be identified based on the rated voltage and power values of the transition tests are shown in Figs. 6–10.
system. Upon leaving coast mode, θ̃g can be used in the grid- The results in Fig. 6 demonstrate the transition from grid-
tied mode controller for several cycles while the PLL scheme tied mode to stand-alone mode. As can be seen, the inverter
converges to the actual grid angle. is disconnected from the grid at approximately t = 0.025 s by
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
OCHS et al.: METHOD OF SEAMLESS TRANSITIONS BETWEEN GRID-TIED AND STAND-ALONE MODES 1939

Fig. 8. Measured voltage and current waveforms in transition from stand-

Fig. 6. Measured voltage and current waveforms in transition from grid-tied alone to grid-tied mode, while the inverter and grid are 180◦ out of phase; coast
to stand-alone mode. mode is activated only by exceeding the current threshold.

Fig. 7. Measured voltage and current waveforms in transition from stand- Fig. 9. Measured voltage and current waveforms in transition from stand-
alone to grid-tied mode. alone to grid-tied mode, while the inverter and grid are 180◦ out of phase;
coast mode is activated by exceeding the current threshold, but θ̃g is used while
waiting on the PLL scheme to converge.
small distortions in the measured voltage and current wave-
forms. At that point, the voltage increases until it reaches
1.057 p.u., the upper threshold of acceptable voltage, at t = tied mode when the PCC and the grid are 180◦ out of phase
0.05 s. At t = 0.05, the inverter switches to stand-alone mode prior to the inverter’s connection to the grid (the worst case
and begins controlling the voltage at the PCC. scenario). In Fig. 8, the inverter is only allowed to switch to
In Fig. 7, the voltage and current waveforms are shown over coast mode when the PCC current exceeds its threshold value
the transition from stand-alone mode to grid-tied mode, when of 5.5 p.u. in this experimental setup. It should be emphasized
the inverter and the grid have been in phase accidentally prior to that the threshold value is practically set to a lower level, e.g.,
connection. In this test, at t = 0.05, the frequency error exceeds 2 p.u., to prevent any damage to the inverter. Regardless of the
the threshold as the inverter can no longer inject the dither threshold value, the peak current exceeds the acceptable level
signal because the PCC frequency is held steady by the grid. At during the transition. Then, only the PLL is used to provide
that point, the inverter switches to grid-tied mode and begins a reference angle for the inverter control system. This case is
injecting a set amount of real and reactive power, which, in this most comparable to the systems reported in the literature [9],
case, happens to be greater than that consumed power by the [10]. Because the controllers cannot function properly without
local load. an accurate estimate of the grid angle, the inverter is forced to
The next test, with results given in Fig. 8, is the first of stay in coast mode for two cycles while the PLL converges. This
three tests for the transition from stand-alone mode to grid- time may not still be enough as a significant transient (with a
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
1940 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 50, NO. 3, MAY/JUNE 2014

Fig. 11. Comparison between overcurrent magnitudes and transition times

for the three different scenarios shown in Figs. 8–10 that are enlarged in
this figure.

Fig. 10. Measured voltage and current waveforms in transition from stand-
alone to grid-tied mode, while the inverter and grid are 180◦ out of phase; coast
mode is activated by exceeding the |θ̃g − θr | threshold, as presented in Fig. 4.

peak current of ∼4 p.u.) is observed when the inverter enters

grid-tied mode at around t = 0.09.
The results of the second test are shown in Fig. 9. In this case,
the inverter is still only allowed to go to coast mode based on
PCC current but is allowed to use θ̃g for the grid angle instead
of waiting on the PLL scheme to converge. As can be seen in
Fig. 9, the inverter still experienced a large transient overcur-
rent, but it is able to return to normal operation after only half
a cycle (8.33 ms) because θr = θ̃g almost instantaneously. A
smaller transient is seen when grid-tied mode begins (at around
t = 0.06) than when the PLL was used to estimate the grid
angle. However, the large peak overcurrent, along with the fact Fig. 12. Grid voltage angle θg after entering coast mode; (a) when the PLL is
used and (b) when the proposed estimation method is applied.
that the transient current exceeded 2 p.u. for more than one-
third of a cycle, suggests that this method is also unsuitable for
In Fig. 12, a comparison of the estimated voltage angle θg
use in practical systems.
when (a) the PLL method and (b) the proposed approach in (13)
The final test, with results shown in Fig. 10, demonstrates the
are used is demonstrated. In Fig. 12(a), the PLL scheme takes
transition from stand-alone mode to grid-tied mode when the
approximately two cycles to converge to the grid angle, while in
PCC and the grid are 180◦ out of phase and the full capabilities
Fig. 12(b), θ̃g yields the grid angle almost instantaneously. As
of the proposed mode transition technique are used. At the
instant that the inverter is connected to the grid, θ̃g becomes can be seen in this figure, θ̃g contains some distortion around
equal to the actual grid voltage angle θg . It should be noted that π and 2π radians. However, it is inconsequential because the
estimated angle in (13) is only used for a very limited number
θ̃g can be different from the interpolated angle θr , as defined
of cycles until the PLL scheme can converge. Again, notice that
in (9). Thus, the threshold of |θ̃g − θr | is quickly reached, so
θ̃g is only used during transient time intervals for several cycles
the inverter changes to coast mode before a large transient
while the PLL scheme converges to the actual grid angle for
current can flow. Because the inverter does not have to wait
utility-interactive three-phase inverters.
on the PLL scheme to converge to the grid angle, it is able
to leave coast mode after only half a cycle, at which point
(t = 0.5) it begins to operate normally in grid-tied mode. The VI. C ONCLUSION
peak current in this transition is 3.667 p.u., and the current is The results presented in this paper serve as proof of concept
greater than 2 p.u. for only 0.428 ms or less than 1/38 of a cycle. for the possibility of achieving virtually seamless transitions
This suggests that the method presented in this paper may be between operating modes without a controllable STS and with
used in a practical system with only very minor oversizing of only one set of sensors. Furthermore, a novel technique has
components to account for current transients during transitions. been presented for making seamless transitions between grid-
The current waveforms in Figs. 8–10 over the time interval tied and stand-alone modes, which minimizes transient time
of [0.04 0.1] s are expanded in Fig. 11. As can be seen, the and overcurrent that normally flows when an inverter is con-
transient time and overshoot currents are significantly mitigated nected to the grid without first being synchronized. In the
using the proposed technique. experimental setup, using the presented technique, the transient
Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.
OCHS et al.: METHOD OF SEAMLESS TRANSITIONS BETWEEN GRID-TIED AND STAND-ALONE MODES 1941

overcurrent reduces to one-third of the overcurrent observed [18] P. Rodriguez, A. Luna, I. Candela, R. Mujal, R. Teodorescu, and
when previously published methods (without a controllable F. Blaabjerg, “Multiresonant frequency-locked loop for grid synchroniza-
tion of power converters under distorted grid conditions,” IEEE Trans.
STS) are used. Furthermore, the time interval for which the Ind. Electron., vol. 58, no. 1, pp. 127–138, Jan. 2011.
current is greater than 2 p.u. is reduced by more than a factor [19] M. Lesseri, F. Blaabjerg, and S. Hansen, “Design and control of an LCL-
of 12 compared to those of the same methods. These seamless filter-based three-phase active rectifier,” IEEE Trans. Ind. Appl., vol. 41,
no. 5, pp. 1281–1291, Sep./Oct. 2005.
transition results, along with the fact that only one set of sensors
is needed, make the grid-tied inverter with no controllable STS
an attractive configuration for inexpensive small systems.
David S. Ochs (S’09) received the B.Sc. and M.Sc.
R EFERENCES degrees in electrical engineering from Kansas State
University, Manhattan, KS, USA, in 2010 and 2012,
[1] IEEE Standard for Interconnecting Distributed Resources With Electric respectively.
Power Systems, IEEE Std. 1547, 2003. Since 2013, he has been an Electric Motor Con-
[2] R. Tirumala, N. Mohan, and C. Henze, “Seamless transfer of trol Design Engineer with the Advanced Technology
grid-connected PWM inverters between utility-interactive and stand- Center, General Motors Powertrain, General Motors
alone modes,” in Proc. IEEE Appl. Power Electron. Conf., Mar. 2002, Corporation, Torrance, CA, USA, where he works on
pp. 1081–1086. developing control algorithms for hybrid and electric
[3] Y. Li, D. Vilathgamuwa, and C. Poh, “Design, analysis, and real-time powertrains. His current research interests include
testing of a controller for multibus microgrid system,” IEEE Trans. Power motor control and thermal management in power
Electron., vol. 19, no. 5, pp. 1195–1204, Sep. 2004. electronic converters.
[4] C. Chen, Y. Wang, J. Lai, Y. Lee, and D. Martin, “Design of parallel
inverters for smooth mode transfer microgrid applications,” IEEE Trans.
Power Electron., vol. 25, no. 1, pp. 6–15, Jan. 2010.
[5] S. Jung, Y. Bae, S. Choi, and H. Kim, “A low cost utility interactive
inverter for residential fuel cell generation,” IEEE Trans. Power Electron., Behrooz Mirafzal (S’01–M’05–SM’07) received
vol. 22, no. 6, pp. 2293–2298, Nov. 2007. the B.Sc. degree in electrical engineering from
[6] Z. Yao, L. Xiao, and Y. Yan, “Seamless transfer of single-phase grid- Isfahan University of Technology, Isfahan, Iran, in
interactive inverters between grid-connected and stand-alone modes,” 1994, the M.Sc. degree (with first-class honors) in
IEEE Trans. Power Electron., vol. 25, no. 6, pp. 1597–1603, Jun. 2010. electrical engineering from the University of Mazan-
[7] D. Dong, T. Thacker, I. Cvetkovic, R. Burgos, F. D.Boroyevich, and daran, Babolsar, Iran, in 1997, and the Ph.D. degree
G. Wang, “Modes of operation and system-level control of single-phase in electrical engineering from Marquette University,
bidirectional PWM converter for microgrid systems,” IEEE Trans. Smart Milwaukee, WI, USA, in 2005.
Grid, vol. 3, no. 1, pp. 93–104, Mar. 2012. From 1997 to 2000, he was a Research Engineer,
[8] I. Balaguer, Q. Lei, S. Yang, U. Supatti, and F. Peng, “Control for as well as a Lecturer, with several academic insti-
grid-connected and intentional islanding operations of distributed power tutions in Isfahan. From 2005 to 2008, he was a
generation,” IEEE Trans. Ind. Electron., vol. 58, no. 1, pp. 147–157, Senior Development/Project Engineer with Allen-Bradley, Rockwell Automa-
Jan. 2011. tion, Mequon, WI, USA, where he was involved in research and development
[9] R. Teodorescu and F. Blaabjerg, “Flexible control of small wind turbines related to motor-drive systems. From 2008 to 2011, he was an Assistant
with grid failure detection operating in stand-alone and grid-connected Professor with Florida International University, Miami, FL, USA. Since 2011,
modes,” IEEE Trans. Power Electron., vol. 19, no. 5, pp. 1323–1332, he has been an Assistant Professor with Kansas State University, Manhattan,
Sep. 2004. KS, USA. He has published over 50 articles in professional journals and
[10] S. Hu, C. Kuo, T. Lee, and J. Guerrero, “Droop-controlled inverters with conference proceedings and is the holder of three U.S. patents. His current
seamless transition between islanding and grid-connected operations,” in research interests include applications of power electronics in modern energy
Proc. IEEE Energy Convers. Congr. Expo., 2011, pp. 2196–2201. conversion systems and motor drives.
[11] J. Vasquez, J. Guerrero, A. Luna, P. Rodriguez, and R. Teodorescu, Dr. Mirafzal was the recipient of the 2008 second best IEEE Industry
“Adaptive droop control applied to voltage-source inverters operating in Applications Society Transactions Prize Paper Award published in 2007, the
grid-connected and islanded modes,” IEEE Trans. Ind. Electron., vol. 56, best 2012 IEEE Power and Energy Society Transactions Prize Paper Award
no. 10, pp. 4088–4096, Oct. 2009. published in 2011, and a 2014 U.S. National Science Foundation (NSF)
[12] J. Kwon, S. Yoon, and S. Choi, “Indirect current control for seamless CAREER Award. He served as the Technical Co-Chair of the IEEE Interna-
transfer of three-phase utility interactive inverters,” IEEE Trans. Power tional Electric Machines and Drives Conference in 2009, and is currently an
Electron., vol. 27, no. 2, pp. 773–781, Feb. 2012. Associate Editor of the IEEE T RANSACTIONS ON I NDUSTRY A PPLICATIONS.
[13] S. Yoon, H. Oh, and S. Choi, “Controller design and implementation of
indirect current control based utility-interactive inverter system,” IEEE
Trans. Power Electron., vol. 28, no. 1, pp. 26–30, Jan. 2013.
[14] D. Holmes and T. Lipo, Pulse Width Modulation for Power
Converters: Principles and Practice. Hoboken, NJ, USA: Wiley, 2003, Pedram Sotoodeh (S’10) was born in Isfahan, Iran.
pp. 32, 259-333. He received the B.S. degree in electrical engineering
[15] P. Krause, O. Wasynczuk, and S. Sudhoff, Analysis of Electric Machinery from Shahrekord University, Shahrekord, Iran, in
and Drive Systems, 2nd ed. Piscataway, NJ, USA: IEEE Press, 2002, 2008, and the M.Sc. degree from Sharif University
pp. 506–510. of Technology, Tehran, Iran, in 2010. He is currently
[16] C-C. Hsieh and C. Hung, “Phase-locked loop techniques—A survey,” working toward the Ph.D. degree in the field of
IEEE Trans. Ind. Electron., vol. 43, no. 6, pp. 609–615, Dec. 1996. power electronics and renewable energy systems at
[17] F. Gonzalez-Espin, E. Figueres, and G. Garcera, “An adaptive Kansas State University, Manhattan, KS, USA.
synchronous-reference-frame phase-locked loop for power quality im- His areas of interest include utility application of
provement in a polluted utility grid,” IEEE Trans. Ind. Electron., vol. 59, power electronics, electrical drives, renewable en-
no. 6, pp. 2718–2731, Jun. 2012. ergy systems, and machine design.

Authorized licensed use limited to: KIIT University. Downloaded on May 17,2024 at 14:37:13 UTC from IEEE Xplore. Restrictions apply.