Beruflich Dokumente
Kultur Dokumente
4, OCTOBER 2008
Abstract—This paper presents a novel philosophy to compen- Phase angle of shunt injected current w. r. t.
sate the load-reactive power demand through a three-phase unified source voltage with PAC approach.
power quality conditioner (UPQC). Most of the UPQC-based ap- Phase angle of shunt injected current w. r. t.
plications show the dependency on shunt inverter for load-reactive
resultant load voltage with PAC approach.
power compensation, whereas the series inverter is always looked
as controlled voltage source to handle all voltage-related problems.
This paper proposes a new functionality of UPQC in which both I. INTRODUCTION
the shunt and series APFs supply the load-reactive power demand. HE term “power quality” (PQ) has gained significant at-
This feature not only helps to share the load-reactive power de-
mand, but also helps to reduce the shunt APF rating, and hence,
the overall cost of UPQC. This results in better utilization of the ex-
T tention in the past few years. The advancement in the
semiconductor device technology has made it possible to realize
isting series inverter. The theory and complete mathematical anal- most of the power electronics based devices/prototypes at com-
ysis termed as “power angle control (PAC)” is presented. The sim- mercial platform. As a rule of thumb in all areas of engineering,
ulation results based on MATLAB/Simulink are discussed in detail the proper utilization of the resources that we have in the most
to support the concept developed in the paper. The proposed ap-
proach is also validated through experimental study. efficient way has lead to great development and is the major con-
cern for most engineers in their respective fields.
Index Terms—Active power filter (APF), power angle control,
Reactive power compensation is one of the common yet very
power quality, reactive power compensation, unified power quality
conditioner (UPQC). important issues for power system engineers at transmission as
well as at distribution level. It is a well-known fact that a typical
distribution network consist of distribution transformer, motor
NOMENCLATURE loads, etc., demands reactive power. This load-reactive power
, Source voltage and current. demand level is mainly affected by the type of loads present
, Load voltage and current. on the network. The capacitor banks have been used to com-
, Series-injected voltage and shunt compensating pensate the load-reactive power demand. It is the simplest and
current. under certain conditions, is a very effective way to compensate
Phase angle of load current with respective to (w. the load-reactive power demand. This traditional way has cer-
r. t.) load voltage. tain major disadvantages, such as fixed compensation, possible
Phase angle of series-injected voltage w. r. t. occurrence of resonance condition with nearby loads, switching
source voltage. transient, bulky size, aging effect, etc.
Phase angle of shunt injected current w. r. t. load The power processing at source, load, and for reactive and
voltage. harmonic compensation by means of power electronic devices
, Resultant load voltage and current with power is becoming more prevalent due to the vast advantages offered
angle control (PAC) approach. by them. The shunt active power filter (APF) is usually con-
Injected shunt current with PAC approach. nected across the loads to compensate for all current-related
problems such as the reactive power compensation, power factor
Power angle between resultant load voltage and
improvement, current harmonic compensation, and load unbal-
source voltage.
ance compensation [1], [2], whereas the series APF is connected
Phase angle of resultant load current w. r. t.
in a series with a line through series transformer. It acts as con-
source voltage with PAC approach.
trolled voltage source and can compensate all voltage-related
problems, such as voltage harmonics, voltage sag, voltage swell,
Manuscript received October 30, 2006; revised November 27, 2007. First
published April 3, 2008; current version published September 24, 2008. This
flicker, etc. [1], [2].
work was supported by the Natural Sciences and Engineering Research Council A unified power quality conditioner (UPQC)[1], [2], [4]–[9],
of Canada. Paper no. TPWRD-00679-2006. is another mitigating device that is similar in construction to a
The authors are with the Department of Electrical Engineering, École de unified power flow controller (UPFC) [3]. A UPFC is employed
Technologie Supérieure, Montréal H3C 1K3, QC, Canada (e-mail: v_khad-
kikar@yahoo.com; ambrish.chandra@etsmtl.ca). in power transmission system where as a UPQC is employed
Digital Object Identifier 10.1109/TPWRD.2008.921146 in a power distribution system, to perform the shunt and series
0885-8977/$25.00 © 2008 IEEE
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2523
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2524 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
power angle difference. This causes the load current phasor ad-
vancement from to maintaining the same load phase angle
Fig. 4. PAC: current phasor representation.
relationship with load voltage. Therefore, the effective load
phase angle with respective source voltage boosts from to ,
resulting in reduced reactive power handled by the shunt APF.
From Isosceles (Fig. 3)
In other words, with power angle lead between the source and
load voltages, the series APF now generates certain amount of (4)
reactive power. Thus, both the inverters of UPQC now take part
(5)
in load-reactive power compensation. The reactive power shared
by both the inverters depends on certain factors and are ana- (6)
lyzed in Sections III-A–C. The rating of series inverter is first From (Fig. 3)
defined based on the sensitive load requirements. Based on the
series APF rating, the maximum possible power angle lead (7)
that could be achieved, without affecting its rating, is computed. (8)
This gives the maximum amount of reactive power that
the series APF can share. Thus, this analysis promises NO addi- (9)
tional burden on series inverter rating. This results in a better uti-
(10)
lization of the existing series inverter at reduced shunt inverter
rating, i.e., a reduction in overall UPQC cost and better device (11)
utilization. (12)
Now
III. MATHEMATICAL COMPUTATION
(13)
In order to implement the Power Angle Control approach
in practical application it is necessary to estimate the power (14)
angle between source and load voltages, based on load-re-
active power demand, but in instantaneous manner. This paper
(15)
proposes an instantaneous angle determination based on in-
stantaneous load-active and -reactive power extraction. At first, (16)
it has been considered that PAC gives a power angle difference .
The instantaneous determination and the boundary condition Equations (12) and (16) give the required magnitude and
for maximum lead that can be achieved with PAC, without phase angle at which the series voltage should be injected in
affecting series APF rating, are presented later in the paper. order to achieve power angle lead, respectively; such that the
amplitude of will be at desired value. For a definite value
A. Series APF Parameter Estimation of , the series-injected voltage only depends on the power
Fig. 3 shows the detailed phasor representation for required angle .
series voltage injection estimation, where represents the rms
B. Shunt APF Parameter Estimation
value of the desired load voltage, which is a known quantity.
From (Fig. 3) Fig. 4 represents the phasor diagram for different currents due
to the power angle advancement. Without PAC, the load-reac-
(1) tive power demand is fulfilled by shunt APF alone by injecting
the shunt compensating current . With PAC, the load cur-
(2) rent phasor shifted to . The effective phase angle between
(3) source voltage and load current suggests that the reactive power
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2525
(33)
(34)
demand seen from the source side gets reduce. The shunt APF (35)
should now inject the compensating current in such a way
that NO extra active power should be taken from the source
Equation (35) gives the net reduction in kVA rating of shunt APF
during this operating period, i.e., the source current , should
as % of full load kVA rating with PAC scheme, without affecting
be same as that of without PAC. Fig. 5 represents the detailed
the existing series APF rating, discussed in Section III-C. Equa-
phasor representation to determine the shunt compensating cur-
tions (12), (16), (26), and (31) are performance equations of
rent and its phase angle with respect to resultant
PAC approach. The effectiveness of this approach is completely
load voltage
governed by proper reference signal generation based on these
quantities but in real time without delay in computation. For a
(17) particular load condition, the load parameters and are
constant; hence, these equations show that PAC approach is
From (Fig. 5) independent of other network parameters, resulting very robust
and parameter insensitive approach to compensate reactive
power demand of the load by utilizing both the shunt and series
(18) inverters simultaneously.
(19)
(20) C. Boundary Condition for Maximum
(21) The rating of series part of UPQC is mainly governed by
percentages of sag need to be compensated. If the maximum
(22)
voltage that can be injected through series APF is denoted by
(23) and the percentage of this maximum voltage limit in
(24) terms of desired load voltage by factor , then we can write
(36)
(25)
From (37)
(26)
(38)
(27) (39)
(28) (40)
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2526 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
(45) (59)
(46) (60)
Thus, with (40), (59), and (60), the maximum reactive power
B. With Power Angle Control burden on shunt APF can be reduced and fixed to
With PAC, both active as well as reactive power will flow based on series APF rating, without disturbing the other basic
through the series APF functions of both series and shunt APFs. If the load-reactive
power demand is less than , the shunt APF should com-
(47) pensate this reactive power demand alone. In such cases se-
ries APF should not take part in reactive power compensation.
(48) If the load-reactive power demand increases above
(49) limit, the series APF should take part in load-reactive power
compensation
(50)
(51) At any condition,
(61)
The negative sign in (51) implies that during PAC of UPQC,
the series APF consumes certain amount of active power. Due In the above equation, is a constant value as fixed by
to this active power the dc link voltage can be increased. To (60) and is computed instantaneously
maintain the dc link voltage at constant level the shunt APF
should now inject the compensating current in such a way that (62)
the extra active power should fed be back to supply side (i.e., it
should circulate between series and shunt active filters through (63)
the dc link). Thus, for ideal condition, without any losses, the
active power consumed by series APF should be equal to the (64)
active power fed back by the shunt APF and hence, the source
current would be at constant level. In actual practice there can
The source should supply only load-active power demand
be a slight increase in source current due to losses associated
as the load-reactive power demand will be shared by both the
with both the inverters, dc link, coupling inductances, and series
APFs. Thus, from (61)–(64), if we could extract the load-active
transformer
as well as -reactive power demands instantaneously, can be
estimated instantaneously.
(52)
(53) V. UPQC CONTROLLER DEVELOPMENT
This section describes how to generate the reference signals
(54)
for both series and shunt part of UPQC.
(55)
A. Instantaneous Determination
From (55), the reactive power generated by series APF mainly The load on a particular plant where UPQC is supposed to
depends on the power angle . The higher the value of , the be installed may not be constant all the time. The load-reactive
higher will be the reactive power compensation through series power demand, therefore, can vary based on the plant load vari-
APF, but at the same time higher will be the magnitude of series- ations. As discussed in the previous section, when the load-re-
injected voltage by (12). active power demand is well within the shunt APF compensa-
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2527
tion limits, the series APF should not act during such condi- where and represent the dc components that are respon-
tions (i.e., ). Now, if the plant load changes such that sible for fundamental load-active and -reactive power, whereas,
the load-reactive power demand exceeds the shunt APF max- and represent the AC components that are responsible for
imum limit, the series APF should act immediately to take over harmonic powers.
the additional load-reactive power demand. Therefore, the per- The per-phase fundamental instantaneous load-active and
formance of PAC approach solely depends on the extraction of fundamental instantaneous load-reactive power can be extracted
power angle in real time. In this paper the instantaneous de- from and by using a low-pass filter (LPF), respectively.
termination is proposed by extracting instantaneous load-active Once these quantities are known, the instantaneous power
as well as instantaneous load-reactive power demand utilizing angle can be determined very easily by using (64). The in-
single phase - theory [10], [11]. For a balanced three-phase stantaneous determination block diagram is shown in Fig. 6.
system, this theory results into less computation time, since only Here, the care is taken to sense power angle only when
one phase instantaneous load-active and instantaneous load-re- , not shown in the figure.
active power extraction is enough.
In single-phase - theory, each phase voltages and currents B. Reference Voltage Signal Generation for Series APF
of original system are defined as a pseudo two-phase system Fig. 7 shows the series-injected voltage reference signals
by giving lead. Thus, the resultant two-phase system can generation utilizing determined instantaneous . With standard
be represented in coordinates. The actual load voltage mathematical computation the required series-injected voltage
and load current are considered as —axis quantities, whereas magnitude and its phase angle are calculated as given by (12)
the lead load voltage and lead load current are con- and (16) respectively. These computations are based on RMS
sidered as —axis quantities. This paper assumes a balanced values. To generate time varying, 60-Hz sinusoidal signal with
three-phase system, hence, considering for only one phase, here estimated phase angle , the Matlab -function blocks are
phase- . used. The and signals, at unity magnitude, from the
Phase-a load voltage representation in coordinates by phase-locked loop (PLL) are used to maintain the synchro-
lead to nization between the generated reference signal and the supply
voltage. This signal multiplied with computed series voltage
(65) magnitude gives the required series-injected voltage signal
with desired phase angle shift. Similarly, with phase
angle difference the reference signals for other two phases are
Similarly, for Phase- load current
generated. These three reference series-injected voltages are
compared with sensed three-phase series-injected voltages and
the errors are then processed by hysteresis controller to generate
(66)
the required switching signals for series inverter switches.
The per-phase instantaneous load-active and instantaneous C. Reference Current Signal Generation for Shunt APF
load-reactive power can be represented by
The shunt APF is used to compensate the current harmonics,
reactive current and to maintain the dc link voltage at constant
level. Instead of calculating the shunt injected current magni-
(67)
tude and its phase angle, an alternative approach is used for
shunt part. In this approach reference source currents are gener-
where and can be expressed as ated directly. This indirect control of shunt compensating cur-
rents also helps to compensate the harmonics generated by the
(68) loads, if any. Therefore there is no need of load harmonic con-
(69) tent extraction. The other reason behind this consideration is to
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2528 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
Fig. 7. Reference voltage signal generation for series APF based on PAC approach.
(70)
(73) (74)
Equation (72) gives the reference fundamental load-active cur- The above equation gives the reference source current signals
rent signal that supplies the fundamental load-active power de- that would supply only fundamental load-active power demand
mand and the losses associated with UPQC. As we have dis- and losses associated with UPQC. The block diagram for refer-
cussed, the PAC approach gives a power angle boost between ence source current signals generation is shown in the Fig. 8. For
resultant load voltage and source voltage, maintaining the same an imbalanced system, each phase should be tackled separately;
voltage magnitudes. The load phase angle between and hence, the reference source current signal extraction would be
, and and is constant, but, the phase angle based on each phase fundamental active power demand sepa-
between the resultant load current with respect to source rately. These reference signals are then compared with sensed
voltage is now boosted to , as viewed from Fig. 4. Therefore, three-phase source currents and the errors are processed by hys-
in order to supply only required fundamental active power (load teresis controller to generate the required switching signals for
losses), the active power demand seen from the source side shunt inverter switches.
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2529
Fig. 9. Simulation Results: load reactive power compensation without PAC of UPQC. (a) Source voltage (v (t)). (b) Load voltage (v (t)). (c) Series injected
voltage (v (t)). (d) DC link voltage (v ). (e) Source current (i (t)). (f) Load current (i (t)). (g) Shunt compensating current (i (t)). (h) Scaled source and
load voltage relationship ( ).
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2530 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
Fig. 10. Simulation results: load reactive power compensation with PAC of UPQC. (a) Source voltage (v (t)). (b) Load voltage (v (t)). (c) Series Injected
Voltage (v (t)). (d) DC link voltage (v ). (e) Source current (i (t)). (f) Load current (i (t)). (g) Shunt compensating current (i (t)). (h) Scaled source and
load voltage relationship ( ).
Fig. 11. Simulation results: load reactive power compensation with PAC of UPQC. (a) Instantaneous power angle ( ). (b) Instantaneous series injected voltage
phase angle (' ). (c) Instantaneous Series injected voltage magnitude (jv j). (d) Instantaneous reference active power (p + p ). (e) Instantaneous source
active-reactive powers (p q ). (f) Instantaneous load-active-reactive powers (p q ). (g) Instantaneous series APF active-reactive powers (p q ). (h) Instan-
taneous shunt APF active-reactive powers (p ; q ).
Further, for better understanding, the load-active power demand At time , the load on the system is changed to 15 kW
is kept constant during all operating conditions at 15 kW and . Since the load-reactive power demand is well
only reactive power demand is changed. within the shunt APF VAR compensation limits (9.3 kVAR),
At time and both shunt and series the shunt APF should now acts immediately to compensate
APF are put ON, respectively, with a pure resistive load (15 kW) the load-reactive power alone. As viewed form Fig. 10(g),
on the system. During time t2 to t3, the UPQC is not doing any the shunt APF starts compensating the load-reactive power
compensation due to the presence of resistive load. This can be demand by injecting 90 leading current. The series APF does
noticed from Fig. 10(c) and (g), where both shunt and series APFs not take part in load-reactive power compensation as noticed
do not inject any compensating current and voltage, respectively. from Fig. 10(c).
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2531
TABLE I
DIFFERENT PARAMETER VALUES
At time , the load on the system is increased to 15 kW and its magnitude are shown in Fig. 11(b) and (c),
. The series APF should now share the part of the respectively. The extracted total instantaneous active power as
load-reactive power demand, since, it is more than the shunt needed in (73) is shown in Fig. 11(d). The instantaneous active
APF compensation limit. The series APF starts compensating and reactive powers supplied by the source and demanded by
the load-reactive power (2.7 kVAR) by injecting appropriate the load are shown in Fig. 11(e) and (f), respectively, whereas
voltage (61V) through series transformers at proper phase angle the instantaneous active and reactive power flow through series
as calculated by controller, as shown in Fig. 10(c). This causes and shunt APFs are shown in Fig. 11(g) and (h), respectively.
9.7 power angle advancement in load voltage with respect to The different parameters during all mentioned conditions are
the source voltage without changing the resultant load voltage shown in Table I. All the current and the voltage values are men-
magnitude as seen from Fig. 10(b). The 9.7 power angle dif- tioned on the per-phase basis and are represented as peak am-
ference between these two voltages can be noticed from scaled plitude. As noticed from Table I(e) and Fig. 9(g), with the full
Fig. 10(h). During this condition the shunt APF compensate load on the network, without considering the PAC control ap-
the load-reactive power up to its maximum limit. Thus, both proach, the required shunt APF compensating current in order
the shunt (9.3 kVAR) and series (2.7 kVAR) APFs shares the to compensate the load-reactive power demand is 29 A. From
load-reactive power demand effectively. Table I(d) and Fig. 10(g), the proposed PAC approach results
At time , the system is loaded up to its maximum in considerable reduction in shunt APF current magnitude from
rated condition (i.e., 15 kW ). During this condi- 29 A to 18.5 A. Thus, with proposed approach the shunt APF
tion, the series APF injects required voltage (140V) to compen- current can be reduced by 10.5 A. In other words, with PAC the
sate the increased load-reactive power demand. This causes fur- shunt APF kVA rating get reduced to 25.6% of total load kVA
ther increase in power angle from 9.7 to 21.5 between both rating by (35). The source current magnitude is almost constant
the source and load voltages, as discussed in the theory and can during all the operating conditions at unity power factor, since,
be noticed from the Fig. 10(h). Here, the shunt APF compen- the active power demand on the load is kept constant during all
sates 9.3 kVAR, while series APF compensates 5.7 kVAR of the conditions.
the total load-reactive power demand.
The source current, as shown in Fig. 10(e), is strictly at con- C. UPQC Performance Under Different Operating Conditions
stant value during all of the above conditions. The load cur- The performance of UPQC under different operating condi-
rent profile for all conditions is shown in Fig. 10(f). Thus, from tions is tested in this section. At time , the active
Fig. 10(e) and (f), the source delivers only the load-active power power demanded by the load is increased from 15 to 20kW. As
demand all the time (i.e., in other words, PAC approach does not the load-active power demand increases, with constant reactive
put any extra active power burden on the source). The dc link power demand, the load power factor angle reduces. Therefore,
voltage profile is shown in Fig. 10(d). The slight variation in dc the power angle also reduces, as given by (64). This interesting
link voltage is acceptable since the dc link feedback control loop observation can be noticed from the Fig. 12(c), where, the se-
is kept slower as the reference current signal extraction is based ries-injected voltage magnitude reduces accordingly, by (12).
on load-active power demand. Fig. 11(a)–(h) shows the instan- At time , distortion in supply voltage is intro-
taneous phase angle and different active and reactive powers. duced. The supply voltage has THD of 12.5% with dominant 5th
All these values are based on per-phase calculation. and 7th harmonics of 10%, and 7.5%, respectively. For better
Fig. 11(a) shows the instantaneous phase angle extraction visualization of transition of UPQC from PAC to voltage con-
using proposed determination approach. The power angle is trol mode the load on the system changed to 15 kW ,
almost zero till time t4, since, the load-reactive power demand is such that the series APF acts as voltage harmonics compensator.
handled by shunt APF alone. At time t4 and t5, the load-reactive As noticed from the Fig. 12(b), after time t7, the load voltage
power demands are changed such that the series APF also take (2% THD) is maintained at desired value by injecting appro-
part in load-reactive power compensation. The change in power priate voltage through series APF as shown in Fig. 12(c). The
angle is instantaneous without much delay in the operation. voltage harmonics present in the source are determined by using
The power angle is solely determined by the instantaneous the PLL-based approach, as given in [7], and then added to ref-
load-active and instantaneous load-reactive power demand. erence signal generated for PAC approach. At time ,
The extracted instantaneous series-injected voltage phase angle the load is changed from linear RL to a nonlinear load. The load
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2532 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
Fig. 12. Simulation Results: UPQC performance under different operating conditions—(a) Source Voltage (v (t)). (b) Load voltage (v (t)). (c) Series-injected
voltage (v (t)). (d) Source current (i (t)). (e) Load current (i (t)). (f) Shunt compensating current (i (t)).
current has a THD of 26.3% with dominant 5th, 7th, and 11th
harmonics of 22%, 9.96%, and 7.4%, respectively. As viewed
from Fig. 12(d), after time t8, the shunt APF compensates the
harmonic generated by nonlinear load by injecting appropriate
compensating current.
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
KHADKIKAR AND CHANDRA: NEW CONTROL PHILOSOPHY FOR UPQC 2533
Fig. 14. Experimental results: load-reactive power compensation with PAC of UPQC. (a) Source and load voltage-current (UPQC OFF). (b) Source voltage-current
(only series APF ON). (c) Source voltage-current (UPQC ON). (d) Source and load voltage profile.
Fig. 15. Experimental results: shunt APF performance comparison with and without PAC of UPQC. (a) Shunt APF performance with PAC. (b) Shunt APF per-
formance without PAC of UPQC.
0.846 lagging. Fig. 14(c) shows the experimental results when VIII. CONCLUSION
both the inverters are in operation i.e., UPQC ON. As seen
from Fig. 14(c), the shunt APF now compensate the remaining This paper proposes a new functionality of UPQC in which
load-reactive power, thus making the unity power factor source both the shunt and series APFs share the load-reactive power
current. The power angle between load and source voltages can demand, termed as PAC of UPQC. The complete mathematical
be viewed from Fig. 14(d). Thus, this experimental investiga- analysis is presented in this paper. The proposed approach is val-
tion shows that the reactive power sharing feature of UPQC, idated through extensive simulation and experimental studies.
by appropriate control of both series and shunt inverters, can The major key points of PAC approach, for load-reactive power
be done effectively. compensation, are as follows.
The performance of shunt APF with, and without 1) Based on the given system, load and UPQC constraints,
PAC of UPQC is compared in Fig. 15. As noticed from the maximum series APF VAR compensation can be esti-
Fig. 15(a) and (b), the shunt current profiles suggest the mated. This helps to fix the maximum shunt APF compen-
reduction in compensating current magnitude. In other words, sation limit. Thus, results in the reduced shunt APF rating,
the PAC of UPQC thus helps to reduce the reactive power without affecting the existing series APF rating and, hence,
burden on shunt inverter and eventually its rating. the overall cost of UPQC. The simulation results for full
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.
2534 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 23, NO. 4, OCTOBER 2008
load of 15 kW show that the shunt APF kVA [7] V. Khadkikar, A. Chandra, A. O. Barry, and T. D. Nguyen, “Applica-
rating can be reduced up to 25.6%, by utilizing proposed tion of UPQC to protect a sensitive load on a polluted distribution net-
work,” in Proc. Annu. Conf. IEEE Power Eng. Soc. General Meeting,
PAC approach. 2006, pp. 867–872.
2) The shunt APF has always the priority to compensate the [8] M. Basu, S. P. Das, and G. K. Dubey, “Performance study of UPQC-Q
for load compensation and voltage sag mitigation,” in Proc. Annu.
load-reactive power demand up to its maximum limit. This Conf. IEEE Ind. Electron. Soc., 2002, vol. 1, pp. 698–703.
assures the full utilization of shunt APF for load-reactive [9] V. Khadkikar, A. Chandra, A. O. Barry, and T. D. Nguyen, “Concep-
power compensation. tual analysis of unified power quality conditioner (UPQC),” in Proc.
Annu. Conf. IEEE Ind. Electron. Symp., 2006, vol. 2, pp. 1088–1093.
3) When the load-reactive power increases above the shunt [10] J. Liu, J. Yang, and Z. Wang, “A new approach for single-phase har-
APF limits, the series APF comes into the role to compen- monic current detecting and its application in a hybrid active power
sate the remaining load-reactive power demand with in a filter,” in Proc. Annu. Conf. IEEE Ind. Electron. Soc., 1999, vol. 2, pp.
849–854.
short period of time. The half-cycle delay in the series APF [11] M. T. Haque, “Single-phase PQ theory,” in Proc. Annu. Conf. IEEE
compensation is due to the delay in computation time to ex- Power Electronics Spec., 2002, vol. 4, pp. 1815–1820.
tract the change in load-reactive power demand.
4) The shunt and series APF operates in coordination to
achieve the common task—the load-reactive power com- Vinod Khadkikar (S’06) was born in Aurangabad,
pensation. Thus, better utilization of existing series APF India, in 1978. He received the B.E. degree in elec-
trical engineering from the Government College of
can be done during normal operating condition. Engineering, Dr. Babasaheb Ambedkar Marathwada
5) A power angle difference occurs between source and load University, Aurangabad, in 2000 and the Master of
Technology degree in power electronics from the
voltage, but the resultant load voltage magnitude is main- Indian Institute of Technology, New Delhi, India, in
tained constant at desired value. 2002. He is currently pursuing the Ph.D. degree in
6) The proposed approach does not put any extra active power the Department of Electrical Engineering, École de
Technologie Supérieure, Montréal, QC, Canada.
burden on the supply system for its operation. His research includes the application of power
electronics in distribution systems, power quality, and active power filters.
REFERENCES
[1] H. Akagi, “New trends in active filters for power conditioning,” IEEE
Trans. Ind. Appl., vol. 32, no. 6, pp. 1312–1322, Nov.-Dec. 1996.
[2] B. Singh, K. Al-Haddad, and A. Chandra, “A review of active power Ambrish Chandra (SM’99) was born in India
filters for power quality improvement,” IEEE Trans Ind. Electron., vol. in 1955. He received the B.E. degree from the
45, no. 5, pp. 960–971, Oct. 1999. University of Roorkee [presently the Indian Institute
[3] L. Gyugyi, “Unified power-flow control concept for flexible AC trans- of Technology (IIT)], Roorkee, India, the M.Tech.
mission systems,” Proc. Inst. Elect. Eng. C, Gen., Transm., Distrib., degree form IIT, New Delhi, India, and the Ph.D.
vol. 139, no. 4, pp. 323–331, Jul. 1992. degree from the University of Calgary, Calgary, AB,
[4] H. Fujita and H. Akagi, “The unified power quality conditioner: The Canada, in 1977, 1980, and 1987, respectively.
integration of series and shunt-active filters,” IEEE Trans. Power Elec- He worked as a Lecturer and later as a Reader at
tron., vol. 13, no. 1, pp. 315–322, Mar. 1998. the University of Roorkee. Since 1994, he has been a
[5] A. Ghosh, A. K. Jindal, and A. Joshi, “A unified power quality condi- Professor in the Electrical Engineering Department,
tioner for voltage regulation of critical load bus,” in Proc. Annu. Conf. École de Technologie Supérieure, Universié du
IEEE Power Eng. Soc. General Meeting, 2004, vol. 1, pp. 471–476. Québec, Montreal, QC, Canada. His main research interests are power quality,
[6] L. Monterio, M. Aredes, and J. Neto, “A control strategy for unified active filters, static reactive power compensation, flexible ac transmission
power quality conditioner,” in Proc. Annu. Conf. IEEE Industrial Elec- systems (FACTS), and renewable energy resources.
tronics Symp., 2003, vol. 1, pp. 391–396. Dr. Chandra is a member of the Ordre des Ingénieurs du Québec, Canada.
Authorized licensed use limited to: ECOLE DE TECHNOLOGIE SUPERIEURE. Downloaded on November 10, 2009 at 15:54 from IEEE Xplore. Restrictions apply.