1744 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO.

4, AUGUST 2018

Power System Compensation Using a

Power-Electronics Integrated Transformer
Mohamed Atef Elsaharty , Member, IEEE, Jose Ignacio Candela , Member, IEEE,
and Pedro Rodriguez, Fellow, IEEE

Abstract—This paper presents a new transformer, that is, the fast dynamic control provided through the installation of FACTS
custom power active transformer (CPAT)—which integrates shunt (Flexible AC Transmission Systems) and CPD (Custom Power
and series equivalent circuits within the transformer’s magnetic Devices) which can cope with the trending changes in the power
structure. Thus, it provides power system services using a single
transformer. The CPAT equipped with a power converter can be system [3]. With such devices playing an important role in the
utilized in distribution systems to control grid current and load- power system, it is essential to develop flexible integration ca-
voltage waveforms while operating as a step-up or step-down trans- pability into the distribution and transmission system.
former between the grid and load. Moreover, it can provide other Modern compensation devices utilize power converters con-
services that any typical shunt-series compensation arrangement nected in shunt and/or series to the electrical network to achieve
provides. The design and analysis of a single-phase CPAT are pre-
sented, showing the effect of coupling between windings and trans- the required compensation objective. Connection of compensa-
former parameters affecting CPAT operation. In this paper, control tion devices can be achieved through transformers [4] or with
of the CPAT in a unified power-quality controller application is in- a transformer-less approach [5]. Limitations in applicability of
vestigated to attenuate grid-current and load-voltage harmonics the transformer based approach is mainly due to the use of bulky
as well as compensate for reactive power requirements and mit- and complicated zigzag transformers to achieve the required VA
igate grid inrush current. Through simulation and experimental
implementation, the merits and performance of the CPAT were ratings and desired waveform. Meanwhile, non-isolated connec-
validated. tion of these systems on the high-voltage (HV) side of the power
system requires special complex converter structures which in-
Index Terms—Power transformers, magnetic circuits, power
conditioning, power distribution.
crease size and cost of installation. Converters installed on the
low-voltage (LV) side incorporate high compensation currents
increasing the size and cost of heat dissipation [5].
I. INTRODUCTION Considering the presence of coupling step-up and step-down
NE of the main challenges in the expansion of a power transformers in the power system, it is beneficial to utilize the
O system is the issue of eliminating transmission and distri-
bution system constraints and bottlenecks. With the increased
essential presence of these transformers while achieving the re-
quired coupling of FACTS devices and CPDs. Integrated mag-
peak customer demand, the distribution system suffers severe netics have been investigated in the literature in the form of
power quality and continuity of supply issues [1]. There is gen- coupled inductors, integrated inductor and transformer, and in-
eral consensus that the future power grid will need to be smart, tegrated transformers (several transformers sharing a common
fault tolerant, self-healing, dynamically controllable, and en- core) [6]. Integrated transformers would present a benefit in
ergy efficient [2]. Such requirements can be achieved through terms of reduced number of magnetic components which in
turn reduces the cost and volume of construction. Moreover, the
ability to combine several transformers into one, would facili-
Manuscript received May 5, 2017; revised July 31, 2017; accepted September tate sub-transmission and distribution transformers with special
16, 2017. Date of publication September 22, 2017; date of current version May
9, 2018. This work was supported in part by the Spanish Ministry of Economy functions of stability and quality improvements to the power
and Competitiveness under projects ENE2014-60228-R and ENE2013-48428- system.
C2-2-R. Paper no. TPWRD-00626-2017. (Corresponding author: Mohamed Integration of power electronics in transformers for such func-
Atef Elsaharty.)
M. A. Elsaharty is with the Electrical Engineering Department, Technical tions has been discussed in the literature and includes trans-
University of Catalonia, 08222 Barcelona, Spain, and also with the Electrical former core sharing [6], shunt integrated transformers [7], and
and Control Engineering Department, Arab Academy for Science, Technol- controllable reactor [8]. These configurations have been utilized
ogy and Maritime Transport, Alexandria 1029, Egypt (e-mail: mohamed.atef.
[email protected]). in FACTS and CPD applications in the form of series voltage
J. I. Candela is with the Electrical Engineering Department, Technical Uni- compensation [9], shunt reactive power compensation [10], se-
versity of Catalonia, 08222 Barcelona, Spain (e-mail: [email protected]). ries harmonic passive filtering [11], series power flow control
P. Rodriguez is with the Department of Engineering, Loyola University An-
dalusia, 41014 Seville, Spain, and also with the Electrical Engineering Depart- through controlled transformer saturation [12] and several other
ment, Technical University of Catalonia (UPC), 08222 Barcelona, Spain (e-mail: applications that involve series or shunt compensation. Another
[email protected]). approach is to avoid the use of transformers and utilizing high-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. power back to back converters to increase controllability [13].
Digital Object Identifier 10.1109/TPWRD.2017.2755725 These approaches present limitation in terms of achieving one

0885-8977 © 2017 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.
ELSAHARTY et al.: POWER SYSTEM COMPENSATION USING A POWER-ELECTRONICS INTEGRATED TRANSFORMER 1745

kind of compensation (series or shunt) and construction com-

plexity.
Modification to a transformer winding configuration in or-
der to incorporate series and shunt services has been discussed
in the development of the “Sen” transformer (ST) [14]. Using
tap changers to inject a compensation voltage, based on control
parameters the ST can operate as a UPFC with certain degrees
of freedom [15]. However, ST presents several limitations in-
cluding use of tap changers which introduces certain limitations
in the degrees of freedom, inability to provide extra services Fig. 1. Configuration of a single-phase CPAT.
to the grid such as inertia emulation and fast dynamic control
over power flow [16]. It is worth noting that no studies have dis-
cussed the integration of series and shunt compensation in sub-
transmission and distribution level transformers which could
provide a dynamically controllable transformer while consider-
ing minimization of size and cost of construction.
This paper presents a power electronics integrated trans-
former used in CPD applications. The CPAT can replace distri-
bution transformers to provide services that will help maintain
power system stability and reliability. The concept of CPAT re-
lies on combining shunt and series windings to the structure of
a transformer in order to incorporate the required services to the
grid. Such services are achieved by controlling the currents and
voltages in the auxiliary windings of a CPAT through passive
or active techniques. In this paper, the application of CPAT in
UPQC is investigated with a non-linear load and distorted grid
voltage. Using a power electronics converter, active control of
auxiliary windings current and voltage is achieved to attenuate
grid-current and load-voltage harmonics, load reactive power
requirements and grid inrush current to the transformer. Fig. 2. Equivalent model of a single-phase CPAT (a) equivalent magnetic
The paper is organized as follows: Section II presents the the- circuit and (b) winding and core-losses electric circuit.
ory of operation of CPAT for a single-phase system and its mag-
netic circuit model. Section III shows the application of CPAT as to be represented by reluctances. Such procedure has been ex-
a single phase UPQC with inrush current mitigation capability, plained in [17] and [18].
Section IV presents simulation evaluation of the single-phase The equivalent magnetic circuit in Fig. 2(a) represents the
CPAT and Section V shows its equivalent experimental evalua- transformer in Fig. 1 with F as the induced magneto-motive
tion. Conclusions are summarized in Section VI. force (mmf) from each winding, Φ as the flux in each limb and
 as the equivalent magnetic reluctance in the transformer core
II. CONCEPT OF CUSTOM POWER ACTIVE TRANSFORMER and air. Subscripts prim, sh, ser and sec represent parameters of
primary, shunt, series and secondary windings, respectively.
A. Configuration
Due to the existence of non-linear characteristics in the iron
Based on magnetic circuits theory, windings wound over a core, a non-linear reactance is used to represent each iron limb
common core are equivalent to shunt electrical circuits while and a linear reactance for leakage air path. Non-linear reluc-
windings wound over shunt cores are equivalent to series elec- tances Y and L represent yoke and limb iron core modelled as
trical circuits [6]. The construction of a single-phase transformer an opposing mmf source following the B-H characteristics of the
with auxiliary windings that are equivalent to shunt and series core material. Meanwhile, winding leakage reluctances l prim ,
circuits is presented in Fig. 1 which forms the basic concept l sec , l sh , l ser and core leakage reluctance 0 are linear
of a single-phase CPAT. The configuration in Fig. 1 shows the reluctances representing leakage flux through air. These reluc-
structure of a single-phase CPAT with primary voltage vprim , tances are calculated based on limb length and cross-sectional
secondary voltage vsec , shunt compensation current ish and se- area.
ries compensation voltage vser . In order to realize the theory The magnetic circuit is affected by the applied voltage and
of operation of the CPAT, an equivalent magnetic model of currents in the winding electric circuit shown in Fig. 2(b). Each
the structure is derived through the conventional modeling ap- mmf source in Fig. 2(a) is coupled to a winding electric circuit
proach of magnetic circuits using electric circuit analogy [17]. in Fig. 2(b) consisting of winding resistance (Rw ) and core-
However, the accuracy of the magnetic circuit is related to the losses equivalent resistance (Rc ). The effective voltage (ve ), is
assumptions made to transform the magnetic core into an equiv- based on the effective flux (Φe ) induced in the magnetic circuit
alent magnetic circuit. Transformation of this magnetic structure and number of turns of the winding (Nw ) as in (1). The mmf
to a magnetic circuit requires identification of magnetic paths generated in the magnetic circuit is dependent on the effective
1746 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO. 4, AUGUST 2018

current (ie ) in the electric circuit as in (2).

dΦe
v e = Nw (1)
dt
Fe = Nw ie (2)
Using Equation (1) and (2), the generated mmf by each wind-
ing (Fw ) and fluxes in each limb (Φk ) can be derived as shown
in (3) and (4) where w is the leakage reluctance per winding.
Using Ampere’s and Gauss’s laws, the coupling between wind-
ing voltage, current, fluxes and mmfs is derived in (5) and (6). Fig. 3. Equivalent electric circuit of a single-phase CPAT.
Since mmfs and fluxes are affected by both voltage and current
as shown in (3) and (4), to simplify the analysis, Rw is assumed Φser , the difference between Φprim and Φsec increases. Therefore,
minimal while Rc and w are assumed high. This assumption if the core reluctances are significant, the shunt winding would
attenuates the effect of voltage on mmf and current on flux. be required to compensate for harmonics present due to this
Equation (5) shows that the mmf on the secondary winding difference too.
(Fsec ) is a result of the sum of primary mmf (Fprim ) and shunt The presence of an equivalent shunt and series circuit from
mmf (Fsh ). In UPQC applications, with a harmonic demanding the configuration in Fig. 1 can be observed by transforming the
load current as in (7) with n harmonic orders, the effect on the magnetic circuit in Fig. 2 to a topologically correct equivalent
primary current can be attenuated by injecting harmonic currents electric circuit model through the principle of duality discussed
in the shunt winding. However, with a non-ideal transformer, the in [17]. The duality equivalent circuit shown in Fig. 3 is repre-
difference between primary and secondary voltage harmonics sented by the elements L as the equivalent inductance of each
combined with the non-linearity of core reluctances (L , Y ), reluctance. It can be observed that vsh is equivalent to a shunt
can introduce extra components in the current as shown in (5). circuit to the primary winding while vser is equivalent to a series
Nevertheless, this effect can be considered minimal based on circuit between the primary and secondary.
design consideration of the core. Moreover, these effects can be Extension of the single-phase CPAT to three-phase appli-
attenuated through the shunt winding as well when accompanied cations could be achieved by using three-single-phase CPATs
by an appropriate controller. where each CPAT represents a single-phase with series and
    shunt compensation. This is a suitable configuration when the
Rw vw
Fw = Nw iw 1 + − (3) three-phase CPAT is compared to a three-phase compensation
Rc Rc
 system consisting of multiple single-phase transformers. An-
1 Fw other approach is to stack all the three CPAT cores on top of
Φk = (vw − iw Rw ) dt − (4)
Nw w each other as in a typical three-phase shell type transformer. The
resulting structure would share common yokes between phases
Fprim + Fsh = Fsec + (Φprim − Φsec ) (L + 2Y ) (5)
as well as reduce required number of tanks, bushings, protec-
Fser − Φser L tion equipment and footprint. Since the three-phase application
Φprim + Φser + Φsec = Φ0 = (6)
0 is outside the scope of this paper, analysis and merits of such
configurations are left for future publications.

h=n
vw , iw = Ih , Vh sin (hωt + θh ) (7)
h=1 B. Characteristics
The relationship between limb fluxes in (6) shows that the The main characteristics of CPAT as compared to the con-
summation of all fluxes is equal to the core leakage flux (Φ0 ). ventional multi-transformer approach (series and shunt com-
Considering flux is mainly affected by the voltage, harmonic pensators connected to the grid using separate insolation
voltages from the primary can be attenuated at the secondary transformers with a distribution transformer in between) can
through the applied voltage on the series winding. This perfor- be summarized into size reduction and facilitating services to
mance is mainly affected by the core leakage reluctance (0 ), the grid.
a decrease in 0 would amplify the effect of L which is non- In a multi-transformer approach, two windings per phase and
linear and would add up harmonic components on the secondary per transformer are required. Therefore, the total number of
flux. However, 0 is typically low and with design considera- windings required in a single-phase system would be 6 p where
tions, core leakage can be minimized to achieve the required p is the number of phases. The CPAT requires 4 p windings
performance. to achieve the same objective. Assuming the number of turns
A decrease in w in (4), would require the shunt and series are identical, this yields a total number of windings reduction of
controllers to inject higher magnitude compensating current and 33%. Meanwhile, size of the resulting CPAT structure combines
voltage respectively due to the effect of the mmf on the fluxes the three single-phase core type transformers into one, replacing
in (5) and (6). two transformers with one central limb in the main transformer.
The coupling between the shunt and series winding can be The CPAT enables utilization of a single transformer with
realized in (5). As the series controller injects harmonic fluxes controlled flux to provide services to the load, transformer and
ELSAHARTY et al.: POWER SYSTEM COMPENSATION USING A POWER-ELECTRONICS INTEGRATED TRANSFORMER 1747

Fig. 4. Configuration of single-phase CPAT for UPQC application.

Fig. 5. Series compensation control loop of single-phase CPAT-UPQC.

grid such as reduced inrush currents, harmonics elimination,

voltage regulation, reactive power compensation and other grid
supportive services based on the control technique.
As compared to the multi-transformer approach, the CPAT
eliminates the requirement of high voltage/current handling
windings usually present in the grid side of series and shunt
isolation transformers that connects power electronic convert-
ers to the system.
The overall reduction in cost and the integration of power
converter with transformer as a single unit makes it an attrac-
tive approach to minimize the requirement of several cascaded
Fig. 6. Shunt compensation control loop for single-phase CPAT-UPQC.
compensation systems.

Modulation (TL-UPQC SVM) is utilized due to its simplicity

III. APPLICATION OF CPAT and effectiveness. The resulting gate signals G a , G b and G c
A. Configuration are used to control the converter switches.
Assuming a non-linear load connected at the secondary wind-
In a distribution system, the single phase CPAT in Fig. 1 can
ing of the transformer, harmonic currents on the secondary wind-
be combined with a three-phase converter adopted from [20] for
ing arise. Since the secondary limb magnetic circuit is in parallel
UPQC application purposes as shown in Fig. 4. The objective of
to the primary limb, both primary winding and shunt winding
the shunt winding in the proposed UPQC is to achieve sinusoidal
would handle a similar mmf waveform.
grid current, damp grid current during transients, compensate
Injection of shunt compensation current adjusts the required
for reactive power requirements and maintain a constant DC
mmf from the primary and hence changing the required primary
bus voltage. The series winding would be required to achieve
current accordingly as in (5). With such an approach, harmonic
sinusoidal load voltage and regulate load voltage magnitude.
components present in isec can be compensated by ish such
In Fig. 4, leg ‘a’ of the converter is connected to the series
that the resulting iprim is harmonics free. Moreover, ish would
winding through an LC filter where the voltage across the capac-
be absorbing fundamental frequency current to charge DC bus
itor is controlled in order to achieve the required series compen-
vdc , compensating the reactive power required by the primary
sation voltage. Leg ‘b’ of the converter is connected to the shunt
winding and reducing the transient current required from the
winding through an LCL filter where the filter output current is
grid.
controlled according to the required shunt compensation. Design
Considering vprim consists of harmonic components, they can
of the series filter inductor (La ), shunt filter inductors (Lb1 and
be attenuated from arriving at the secondary winding by vary-
Lb2 ), series capacitor (Ca ) and shunt capacitor (Cb ) are based
ing vser such that the resulting Φsec is sinusoidal as in (6) and
on the required control bandwidth and the switching frequency
therefore the vsec would be sinusoidal too. Moreover, the mag-
of the converter. Passive damping of the LC and LCL resonance
nitude of the fundamental frequency of vsec can be controlled
is accomplished through damping resistance Ra and Rb .
by injecting the fundamental component in vser to compensate
Leg ‘c’ is the common leg between both windings and rep-
for voltage swells and sags from the primary.
resents the current return path of both phases. Having a com-
mon leg for both compensation presents a control challenge due
B. Control
to coupling between the converter legs. Among the strategies
proposed in the literature to decouple the control of each leg, The control architecture presented in Figs. 5 and 6 consists
the strategy in [21] based on Three-Leg UPQC Space Vector of two part controllers; shunt and series compensation. Series
1748 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO. 4, AUGUST 2018

compensation shown in Fig. 5 consists of two parts; magnitude TABLE I

SIMULATION PARAMETERS
of the secondary voltage Vsec is maintained constant at the refer-
∗
ence value Vsec through a PI controller. The resulting command
voltage is multiplied by the secondary voltage synchronizing Parameter Value
signal sin(ωt)sec in order to achieve an in-phase compensa- Simulation step time 1 μsec
tion of the secondary voltage magnitude. Meanwhile secondary Transformer rated power per winding 15 kW
Turn ratio 1:1
voltage harmonics are compensated through Resonant (R) con- System voltage 220 V
trollers tuned to required compensating frequencies nω where n Reference DC bus voltage 400 V
represents the harmonic order and a reference secondary voltage Switching frequency 10 kHz
∗ Controller sampling frequency 10 kHz
harmonics vsec h set to zero. Resonant controllers are utilized due Dead time 3 μsec
to their high gain at selected frequencies and no phase shift or Ra , La 15 mΩ, 1.5 mH
gain at other frequencies [22]. Therefore, the controller would Rc a , Ca 2 Ω, 40 μF
Rb 1 , Lb 1 60 mΩ, 6 mH
generate a command voltage at the selected frequency which Rb 2 , Lb 2 20 mΩ, 2 mH
would attenuate the error. The resulting sum of command volt- Rc b , Cb 4.7 Ω, 5 μF
ages from the PI and R controllers obtains the required injected C dc 5 mF
voltage vser , which is the output voltage from leg ‘a’ with respect
to leg ‘c’ in the single-phase setup in Fig. 4.
The shunt compensation controller in Fig. 6 consists of four
controllers; two controllers control the fundamental compo-
nent of primary current, one is used for compensation of har-
monic components and a shunt winding current controller which
achieves the required reference current from the previous con-
trollers. The DC bus voltage controller maintains a constant
∗
reference voltage vdc across the converter capacitor by absorb-
ing active power from the primary winding of the transformer.
The PI controller determines the required capacitor charging
current Iprim dc . Iprim dc is synchronized with the fundamental
primary voltage sin(ωt)prim to acquire the reference primary
current i∗prim . i∗prim and feedback primary current iprim are passed
on to a (Proportional Resonant) PR controller tuned at the fun-
damental frequency in order to obtain the required fundamental
shunt compensation current i∗sh f . The reference current would
maintain a constant DC bus voltage, zero-reactive power from Fig. 7. Voltage and current waveforms during 50% primary voltage sag.
the primary winding and would dampen the primary current
during transients. Primary harmonic currents are compensated
active and reactive current with the harmonic spectrum con-
through R controllers with a reference i∗prim h which is set to
sisting of 3rd, 5th, 7th and 9th with the proportions 23%,
zero. This obtains the required shunt compensation harmonic
6.34%, 1.56% and 0.85% of fundamental component respec-
currents i∗sh h . The combined compensation current i∗sh is con-
tively. Meanwhile, secondary voltage magnitude control was
trolled through a PR controller to obtain the required converter
examined by applying voltage disruption during operation. Pri-
shunt output voltage vsh .
mary current transient damping was investigated by suddenly
Primary frequency ω, synchronizing signal sin(ωt)prim , sec-
applying voltage to the primary winding to observe the inrush
ondary voltage magnitude Vsec and secondary voltage syn-
current effect with and without compensation.
chronizing signal sin(ωt)sec are obtained through a Double
Second-Order Generalized Integrator Frequency Locked Loop
(DSOGI-FLL). A. Harmonics Attenuation, Voltage Regulation and Reactive
Power Compensation
During a primary voltage sag of 50% illustrated in Fig. 7,
IV. SIMULATION RESULTS the secondary voltage magnitude controller increases the fun-
The configuration in Fig. 4 with parameters tabulated in Ta- damental frequency component of the injected series voltage
ble I was simulated to examine the operation of CPAT under vser to compensate for this drop. It should be noted that the
different scenarios. To investigate the effectiveness of series secondary voltage harmonic contents changes as the trans-
harmonic compensation, the primary voltage was generated former is loaded. This is typically due to the inductive action
through a combination of harmonic components 3rd, 5th, 7th of the transformer under load acting as a filter reducing the har-
and 9th with the proportions 25%,12.5%, 6.25% and 3.13% monic contents of the secondary voltage. However, this yields a
of fundamental component respectively. Shunt harmonics com- voltage drop to the load voltage and the presence of magni-
pensation was investigated by connecting a non-linear load at tude controller is essential. Based on the transformer design,
the terminals of the secondary winding to withdraw non-linear, such drop can be minimized by considering low core leakage
ELSAHARTY et al.: POWER SYSTEM COMPENSATION USING A POWER-ELECTRONICS INTEGRATED TRANSFORMER 1749

Fig. 8. Active Power variation during a 50% primary voltage sag.

Fig. 11. Voltage and current waveforms with shunt harmonics compensation.

Fig. 9. Reactive Power variation during a 50% primary voltage sag.

Fig. 12. Voltage and current waveforms with series harmonic compensation.

harmonic voltages into the series winding as illustrated in Fig. 12

to attenuate the secondary voltage harmonics. The resulting har-
monic components present in the primary current were reduced
to 0.05% and secondary voltage to 0.5% which shows the effec-
tiveness of the CPAT in harmonics attenuation.
It should be noted that the series harmonics compensator not
only compensates for harmonic voltages present in the primary,
but also harmonic voltages induced due to the non-linear cur-
Fig. 10. Converter DC voltage and current during 50% primary voltage sag. rent absorbed by the load and passing through the transformer
equivalent reluctance as in (6). Controller parameters and fil-
ters affect the bandwidth over which compensation is possible.
impedance which yields a higher reluctance path of yokes and
Moreover, inverter output voltage and current harmonics will
end limbs as discussed in Section II.
affect the resulting compensation.
The effect of the sag can be observed on the power exchanged
Enabling the reactive power compensator, shunt reactive cur-
between the converter and the grid in Fig. 8 and Fig. 9. The
rent was injected into the transformer eliminating the require-
shunt leg of the converter absorbs more active power to supply
ment of reactive power from the primary side as shown in
the series leg to supply the difference in the voltage. This can be
Fig. 13. The required reactive power by the transformer is de-
also be observed in the DC bus voltage and current in Fig. 10.
pendent on the transformer design and equivalent impedance.
As the series leg demands more power, the DC bus voltage
decreases until the shunt controller provides sufficient power to
the DC bus. B. Inrush Current Mitigation
Enabling the shunt harmonics controller results in the injected The reactive power compensation controller presented in
current shown in Fig. 11 which includes harmonic components Fig. 6 can achieve damping of primary current during tran-
required by the load. The series harmonics controller injects sients. Therefore, during transformer excitation, the required
1750 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO. 4, AUGUST 2018

Fig. 13. Voltage and current waveforms with reactive power compensation. Fig. 15. Voltage and current waveforms during inrush transient with compen-
sation.

Fig. 14. Voltage and current waveforms during inrush transient without com-
pensation.

inrush current can be supplied through the shunt winding to the

transformer and thus reducing its impact on the source grid. To
Fig. 16. Harmonic components of primary current during inrush transient
simulate inrush current, effect of saturation was taken into con- without compensation (solid line) and with compensation (dashed line).
sideration in the transformer model as in Section II with each
limb reluctance represented by a controlled mmf source oppos-
ing flux direction based on BH curve characteristics. The inrush
current through the transformer when rated primary voltage
was applied with no load is shown in Fig. 14. The magnitude,
harmonics and decay time of the primary current shows the
significant impact of the inrush current on the grid.
Enabling shunt compensation controllers would dampen
the primary current magnitude during such transient. This is
achieved through its reactive power compensation controller,
which acts directly on the primary current magnitude at funda-
mental frequency. The reference primary current at startup is
zero, since the DC bus is fully charged and thus the reactive
power compensation controller would maintain a zero primary
current at energization. Moreover, primary current harmonics
are mitigated through the primary current harmonics controller.
This can be observed in Fig. 15 where the primary current mag-
Fig. 17. Peak magnitude of primary current during inrush transient.
nitude and harmonics are significantly minimized. However,
the transformer still absorbs the same inrush current to energize
through the primary and shunt windings. beyond nominal rated values and below absolute maximum val-
Due to the high current passing through shunt winding in this ues. Moreover, the series winding may be shorted at start-up
compensation strategy, it should be noted that the simulation pre- to avoid connection of the filter circuit to the series winding
sented here is based on operating the converter at extreme condi- acting as a load. During this operation, DC bus control is dis-
tions where current and applied voltage to the filter are operating abled to avoid absorbing charging current at start-up through
ELSAHARTY et al.: POWER SYSTEM COMPENSATION USING A POWER-ELECTRONICS INTEGRATED TRANSFORMER 1751

Fig. 18. Single-phase CPAT experiment (a) diagram and (b) setup.

the shunt winding yielding a higher inrush current. Thus, this

operation requires a pre-charged DC bus. Another issue which
is not addressed in the presented controller is the decaying dc-
offset shown in Fig. 14 and Fig. 15, as the inrush current tends
to a steady-state.
A harmonic analysis of the inrush current in Fig. 16 illustrates
the capability of the shunt controller to mitigate fundamental
and harmonic components throughout the inrush current cycles;
this mitigates the peak current required from the grid during
each cycle as in Fig. 17. The performance of this mitigation is
reliant on the controller and converter bandwidth. An average
over 15 cycles shows a 51.5% reduction in the inrush current
peak magnitude with the utilized setup.

V. EXPERIMENTAL RESULTS
The hardware setup illustrated in Fig. 18(a) was implemented Fig. 19. Experimental waveforms during a 30% voltage sag.
as shown in Fig. 18(b) consisting of a 50 kW three-phase core-
type transformer of which four windings are used for primary,
secondary, shunt compensation and series compensation. Pri- primary winding as shown in Fig. 19 at the instant indicated by
mary voltage and compensating voltages are provided through the vertical line. With a reference of 195 V, it can be observed
a 21 kW (7 kW/phase) three-phase Power Amplifier to examine that the series controller injects an increased fundamental com-
the effect of injecting voltage into the series and shunt wind- ponent to compensate for the voltage sag. The resulting voltage
ing. The three phases of the amplifier U,V and W are used such recovery during the sag is shown to be similar to the typical
that phase U emulates primary voltage through the reference response of a dynamic voltage regulator.
∗ Harmonic compensation was performed with a non-linear
vprim , phase V injects a controlled shunt compensating current
∗ load connected at 200 V operating voltage. Series harmonic
through the reference vsh and phase W applies series compen-
∗ ∗ compensation controller performance is shown in Fig. 20 as it
sating voltage through the reference vser . vprim is generated by
OPAL-RT independent of the control algorithm with a defined was enabled at the instant indicated by the vertical line. It can
harmonic spectrum. Control of the amplifier is achieved through be observed in Fig. 20 that the transformation of the secondary
an OPAL-RT real-time simulator which includes all the control voltage from a harmonic-rich voltage to a sinusoidal voltage.
structure for shunt and series compensation presented earlier in As mentioned earlier, harmonic components present in the sec-
Fig. 5 and Fig. 6. The controller emulates the DC bus voltage ondary voltage arise from primary voltage harmonics and load
by measuring the output current from the amplifier phases iV current harmonics. This can be observed in Fig. 21, when a
and iW which represent shunt and series compensation currents sinusoidal voltage was applied to the primary while the har-
respectively. Data points of the results waveform have been monics controller of the series winding was disabled and shunt
captured for plotting and analysis. winding was enabled. Enabling the shunt harmonics controller
at the instant indicated by the vertical line in Fig. 21, shows
an attenuation in the primary current harmonics. However, sec-
A. Harmonics Attenuation, Voltage Regulation and Reactive
ondary voltage consisted of harmonic components induced by
Power Compensation
the secondary current as in (5).
To investigate the effectiveness of the series compensation Reactive power compensation through the shunt controller is
controller, a sustained 30% voltage sag was applied to the observed through the phase shift of the primary current with
1752 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO. 4, AUGUST 2018

Fig. 23. Experimental reactive power variation during reactive power con-
troller activation.

Fig. 20. Experimental waveforms with shunt-series harmonics compensation.

Fig. 24. Experimental active Power variation during reactive power controller
activation.

Fig. 21. Experimental voltage and current waveforms with shunt harmonic
compensation.

Fig. 25. Experimental primary inrush current without shunt compensation.

B. Inrush Current Mitigation

Mitigation of inrush current effect on the grid through the
Fig. 22. Experimental voltage and current waveforms with reactive power shunt controller was investigated in the experimental setup by
compensation.
applying 60% of the rated transformer voltage at no-load to the
primary winding. Fig. 25 shows the peak current magnitude,
respect to the primary voltage as in Fig. 22. The shunt controller harmonic contents and transient decay, as in the simulation re-
injects reactive current to the shunt winding, eliminating the sults discussed previously. As the shunt controller was enabled,
required reactive power from the primary side for feeding the Fig. 26 shows the significant reduction in peak current magni-
load and transformer as shown in Fig. 23. Meanwhile, Fig. 24 tude and harmonic contents of the inrush current.
shows that active power was supplied through the primary and Further analysis of the resulting primary current waveform
absorbed through the shunt winding to maintain a constant DC before and after compensation is shown in Fig. 27. A significant
bus during this operation. reduction in the harmonic components of the primary current
ELSAHARTY et al.: POWER SYSTEM COMPENSATION USING A POWER-ELECTRONICS INTEGRATED TRANSFORMER 1753

shunt services in a single transformer. Simulation and experi-

mental results of a single-phase CPAT-UPQC show the ability
of the transformer to achieve compensation of primary current
harmonics, reactive power, secondary voltage harmonics and
secondary voltage magnitude. Moreover, shunt compensation
achieved inrush current mitigation during energization of the
transformer, reducing the effect of inrush current on the source
grid.

REFERENCES
[1] D. Volk, Electricity Networks: Infrastructure and Operations (Insights
Series), International Energy Agency, Paris, France, 2013.
Fig. 26. Experimental primary inrush current with shunt compensation.
[2] J. D. Van Wyk and F. C. Lee, “On a Future for Power Electronics,”
IEEE Trans. Emerg. Sel. Topics Power Electron., vol. 1, no. 2, pp. 59–72,
Jun. 2013.
[3] J. V. Milanovic and Y. Zhang, “Modeling of FACTS devices for voltage
sag mitigation studies in large power systems,” IEEE Trans. Power Del.,
vol. 25, no. 4, pp. 3044–3052, Oct. 2010.
[4] M. Morati, D. Girod, F. Terrien, V. Peron, P. Poure, and S. Saadate,
“Industrial 100-MVA EAF voltage flicker mitigation using VSC-Based
STATCOM with improved performance,” IEEE Trans. Power Del., vol. 31,
no. 6, pp. 2494–2501, Dec. 2016.
[5] F. Z. Peng, Y. Liu, S. Yang, S. Zhang, D. Gunasekaran, and U. Karki,
“Transformer-Less unified power-flow controller using the cascade multi-
level inverter,” IEEE Trans. Power Electron., vol. 31, no. 8, pp. 5461–5472,
Aug. 2016.
[6] Y.-S. Lee, L.-P. Wong, and D. K. W. Cheng, “Simulation and design of
integrated magnetics for power converters,” IEEE Trans. Magn., vol. 39,
no. 2, pp. 1008–1018, Mar. 2003.
[7] C. Liang et al., “Harmonic elimination using parallel delta-connected
filtering windings for converter transformers in HVDC systems,” IEEE
Trans. Power Del., vol. 32, no. 2, pp. 933–941, Apr. 2017.
Fig. 27. Experimental harmonic components of primary current during inrush [8] M. Tian, J. Yin, and Y. Liu, “Magnetic integration technology in control-
transient without compensation (solid line) and with compensation (dashed line). lable reactor of transformer type constituted by various magnetic materi-
als,” IEEE Trans. Appl. Supercond., vol. 24, no. 5, pp. 1–5, Oct. 2014.
[9] D. Li, Q. Chen, Z. Jia, and C. Zhang, “A high-power active filtering system
with fundamental magnetic flux compensation,” IEEE Trans. Power Del.,
vol. 21, no. 2, pp. 823–830, Apr. 2006.
[10] C. Wang, X. Yin, Z. Zhang, and M. Wen, “A novel compensation tech-
nology of static synchronous compensator integrated with distribution
transformer,” IEEE Trans. Power Del., vol. 28, no. 2, pp. 1032–1039,
Apr. 2013.
[11] J. Pleite, V. Valdivia, P. Zumel, and C. Gonzalez, “Transformer and series
inductance integration for harmonic filtering in PWM inverters based
in a simple design procedure,” in Proc. IEEE Int. Symp. Ind. Electron.,
Jun. 2007, pp. 1201–1206.
[12] A. Dimitrovski, Z. Li, and B. Ozpineci, “Magnetic amplifier-based power-
flow controller,” IEEE Trans. Power Del., vol. 30, no. 4, pp. 1708–1714,
Aug. 2015.
[13] M. Liserre, G. Buticchi, M. Andresen, G. De Carne, L. F. Costa, and Z. X.
Zou, “The smart transformer: Impact on the electric grid and technology
challenges,” IEEE Ind. Electron. Mag., vol. 10, no. 2, pp. 46–58, Jun. 2016.
[14] K. K. Sen and Mey Ling Sen, “Introducing the family of “Sen” transform-
ers: A set of power flow controlling transformers,” IEEE Trans. Power
Del., vol. 18, no. 1, pp. 149–157, Jan. 2003.
Fig. 28. Experimental peak magnitude of primary current during inrush tran- [15] K. K. Sen and M. Ling Sen, “Comparison of the “Sen" transformer with
sient. the unified power flow controller,” IEEE Trans. Power Del., vol. 18, no. 4,
pp. 1523–1533, Oct. 2003.
[16] J. Liu and V. Dinavahi, “Nonlinear magnetic equivalent circuit-based
can be observed which eventually resulted in a reduced peak grid real-time Sen transformer electromagnetic transient model on FPGA for
HIL emulation,” IEEE Trans. Power Del., vol. 31, no. 6, pp. 2483–2493,
current as shown in Fig. 28. An average over 15 cycles shows Dec. 2016.
a reduction of 22.3% in the inrush current from the primary [17] G. R. Slemon, “Equivalent circuits for transformers and machines includ-
winding. ing non-linear effects,” Proc. Inst. Elect. Eng., vol. 100, no. 77, pp. 577–
581, Oct. 1953.
[18] P. C. Sen, Principles of Electric Machines and Power Electronics. 2nd ed.
VI. CONCLUSION New York, NY, USA: Wiley, 1997.
[19] N. D. Hatziargyriou, J. M. Prousalidis, and B. C. Papadias, “Generalised
The Custom Power Active Transformer proposed in this pa- transformer model based on the analysis of its magnetic core circuit,”
per has shown the ability to combine the possibility of series and Proc. Inst. Elect. Eng., vol. 140, no. 4, pp. 269–278, Jul. 1993.
1754 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 33, NO. 4, AUGUST 2018

[20] V. Khadkikar, “Enhancing electric power quality using UPQC: A compre- Pedro Rodriguez (SM’10–F’13) received the M.Sc.
hensive overview,” IEEE Trans. Power Electron., vol. 27, no. 5, pp. 2284– and Ph.D. degrees in electrical engineering from the
2297, May 2012. Technical University of Catalonia (UPC), Barcelona,
[21] Y. Lu, G. Xiao, X. Wang, F. Blaabjerg, and D. Lu, “Control strategy for Spain. In 1990, he joined the UPC, where he became
single-phase transformerless three-leg unified power quality conditioner the Director of the Research Center on Renewable
based on space vector modulation,” IEEE Trans. Power Electron., vol. 31, Electrical Energy Systems and still collaborates with
no. 4, pp. 2840–2849, Apr. 2016. the UPC as a Visiting Professor. In 2005, he was a Vis-
[22] R. Teodorescu, M. Liserre, and P. Rodrı́guez Grid Converters for Photo- iting Researcher with the Center for Power Electron-
voltaic and Wind Power Systems. Hoboken, NJ, USA: Wiley, 2011. ics Systems, Virginia Polytechnic Institute and State
University. In 2006 and 2007, he was a Postdoctoral
Researcher in the Department of Energy Technology,
Mohamed Atef Elsaharty (S’12–M’13) received the Aalborg University (AAU). From 2007 to 2011, he was a cosupervisor of the
B.Sc. and M.Sc. degrees in electrical and control en- Vestas Power Program at the AAU. From 2011 to 2017, he was the Director of
gineering from Arab Academy for Science, Tech- Technology in the area of power systems in Abengoa Research, Spain. Since
nology, & Maritime Transport (AASTMT), Alexan- 2017, he has been with the Loyola University Andalucia, Seville, Spain, as a Full
dria, Egypt, in 2009 and 2012, respectively. Since Professor. He has coauthored one Wiley-IEEE book, more than 80 papers in ISI
2014, he has been working toward the Ph.D. degree in technical journals, and around 250 papers in conference proceedings. He is the
electrical engineering at Renewable Electrical Energy holder of 14 licensed patents. His research interests include distributed power
System Research Center, Technical University of Cat- systems, flexible transmission systems, and power conversion. Prof. Rodriguez
alonia (UPC), Barcelona, Spain. He is currently a is an IEEE Fellow for his contributions in the control of distributed generation,
Senior Teaching Assistant with the Department of an Associate Editor of the IEEE TRANSACTION ON POWER ELECTRONICS, and
Electrical and Control Engineering, AASTMT. His a member of the Sustainability and Renewable Energy Committee of the IEEE
research interests include renewable energy systems modeling, control, and Industry Application Society and the Renewable Energy Systems Technical
grid interface, linear and nonlinear control techniques, distributed control of Committee of the Industrial Electronics Society.
energy systems, active and passive power filters, power-electronics converters,
and robotics applications.

Jose Ignacio Candela (S’99–M’04) received the

B.S. and M.S. degrees in industrial engineering and
the Ph.D. degree in electrical engineering from the
Technical University of Catalunya (UPC), Barcelona,
Spain, in 1987, 2000, and 2009, respectively. In 1990,
he became an Assistant Professor at UPC, where he
later advanced to Associate Professor in 1993. Cur-
rently, he is part of the research group on Renewable
Electrical Energy Systems, Department of Electrical
Engineering. He has authored or coauthored more
than 30 published technical papers, and holds several
patents. His current research interests include power conditioning, integration of
distributed energy systems, and the control of grid-connected power converters.
Prof. Candela is a member of the IEEE Power Electronics Society, the IEEE
Industrial Electronics Society, and the IEEE Industry Application Society.