Page 1

Harmonic and Balance Compensation using

Instantaneous Active and Reactive Power Control

on Electric Railway Systems

A. Bueno, J. M. Aller and J. Restrepo

Grupo de Sistemas Industriales de Electrónica de Potencia

Universidad Simón Bolívar

Caracas 1080A, Venezuela

T. Habetler

School of Electrical and Computer Engineering

Georgia Institute of Tecnology

Atlanta, Georgia

Abstract—This work presents a general filtering and unbalance

compensation scheme for electric traction systems. The proposed

method uses an active filter controlled with the instantaneous

active and reactive power, to reduce the harmonic current

distortion and the negative sequence obtained by the system

under unbalanced operation in steady state. The proposed filter

is evaluated using open delta (V-V) and Scott transformers

in the power substation. The scheme has been simulated and

experimentally validated. Experimental and simulation results

show the controller advantages and the applicability of the

proposed method in railway systems .

Index Terms—Harmonics, Active filter, Transformer, Locomo-

tive, Traction application.

I. INTRODUCTION

Electric traction systems for passengers and goods use

different power transformer configurations, in order to feed

single phase systems from the three phase supply. In gen-

eral, three-phase to two single phase conversion schemes use

transformers connected in open delta (V − V ), Scott or Le

Blanc configurations [1]. In a practical application, the load

associated with each single-phase circuit does not compensate

each other, due to the variable demands in the transport

system and railroad line profile. Also, the use of uncontrolled

rectification to feed the traction load contribute to the total

unbalance seen from the three phase supply. This unbalance

is due mainly to the injection of current harmonics to the main

three-phase system depending on the transformer connection

and harmonic order [2].

It is then required the use of filters and unbalance com-

pensators to ensure proper system operation and to raise the

power quality [3].

These problems are usually addressed, in practice, with the

use of passive power quality compensators such as reactive

power compensation capacitors and passive filters, and they

are single-phase equipment installed in each feeder of the

traction substation. Usually, the coupling factor between two

feeders is negligible due to the independent operation of each

passive compensator. Moreover, passive equipment does not

have the dynamic capability to adjust to changes in load, where

over and under compensation happen frequently as a result of

continuous change in load conditions.

Different active power quality compensators have been

proposed in [4]–[6] to solve the unbalance problem. All of

them employ two single-phase converters that have a common

DC bus and the simultaneous compensation of harmonic

content and unbalance can not be achieved with these schemes.

Also, when the compensation is made from the single phase

side, the instantaneous active and reactive power definition is

difficult to use in the compensation of harmonics and power

unbalance [7] [8].

In this work a compensation scheme is proposed to pro-

vide simultaneous correction of harmonic content and load

unbalance for railroad systems using open delta or Scott

transformers in the power substation. This scheme is based

on the instantaneous active and reactive power description of

the system [9], using space vector representation of the state

variables, and the application of direct power control (DPC)

to attain the required correction by minimizing a cost function

obtained from the instantaneous active and reactive mismatch

[10]–[12].

The control strategies presented in this work are both,

simulated using a state variables model representation and

experimentally validated using a DSP based modular power

electronic system able to emulate the electric traction system

operating conditions, the open delta, the Scott transformer, the

filtering and the load balancing converters [13].

The generality of the proposed filtering technique using

instantaneous active and reactive power can be extended to

any other transformer configuration in the power substation.

Multilevel converter technology can facilitate the industrial

implementation because reduces the specifications of the

power electronics switches and the voltage stress (dv

the magnetic components like coupling transformers and/or

inductors [14].

dt) on

II. HARMONIC AND UNBALANCE COMPENSATION SYSTEM

Figure 1 shows the proposed control scheme. A shunt active

filter is used, directly connected to the power system using

a voltage rising transformer. The active filter uses a power

converter configured as an active three-phase PWM rectifier,

connected to the three-phase side.

978-1-4244-4783-1/10/$25.00 ©2010 IEEE 1139

Page 2

Figure 1: Proposed compensation scheme

(a) V-V Transformer(b) Scott Transformer

Figure 2: Proposed compensation scheme.

Figure 2 shows the open delta (V − V ) and Scott trans-

formers used frequently to connect a traction substation to the

electric grid. These connection schemes generate two single-

phase networks from the three-phase power system. Each

single-phase circuit is used to feed a 60 to 100 km rail track.

The simulation of the steady state and dynamic behavior

for the traction system under unbalance conditions and with

harmonic current injection uses a space vector model of the

open delta and Scott transformer, uncoupling the differential

equations in the transformer model [7]. Additionally, the filter

and its control have been modeled using a space vector

representation [15].

The power invariant space vector transformation is defined

as,

? x =

?

2

3

?xa(t) + αxb(t) + α2xc(t)?

α = ej2π

3

(1)

A. V-V Transformer space vector model

For the ideal V − V transformer configuration shown in

Fig. 2a, its model can be obtained considering the transformer

ratio and using Ampere and Faraday Laws [1]:

vab=N1

N2vT1; vbc=N1

N2vT2; ia=N2

N1iT1; ic=N2

N1iT2

(2)

The voltage and current space vectors calculated in the

transformer’s primary winding as function of the secondary

winding voltages and currents are:

?

?is=

3

N1

? vs=

?

2

3

N1

N2

?vT1− α2vT2

?

2

N2

?(1 − α)iT1+?α − α2?iT2

?

(3)

B. Scott Transformer space vector model

For the ideal Scott transformer shown in figure 2b, its model

can be obtained considering the transformer ratio and using

Ampere and Faraday Laws [1]:

vab=N1

√3

2

N2vT1; vco=

N2ic= iT2;1

√3

2

N1

N2vT2;

N1

2

N1

N2(ia− ib) = iT1

(4)

The voltage and current space vectors calculated in the

transformer’s primary winding as function of the secondary

winding voltages and currents are:

?

?is=

3

N1

? vs=

?

3

2

N1

N2

?(1 − α)iT1+√3α2iT2

1

1−α2(vT1− jvT2)

2

N2

?

(5)

C. Active and reactive power control

The DPC controller is based in the instantaneous apparent

power from the current and voltage space vectors definitions

[7]:

? s = ? vs·?i∗

From Fig. 1, the active filter can be modeled as,

s= (vsα+ j vsβ) · (isα+ j isβ)∗= p + jq

(6)

? vs= ? vr+ R?is+ Ld?is

dt

(7)

A discrete time version of this equation is obtained by

replacing the derivative with a first order Euler approximation,

and the estimated supply current for the next control cycle

becomes

??is(k + 1) =?is(k) + ∆??is(k)

∆??is(k) =Ts

From (6), the estimated active and reactive power for the

next sampling period can be written as,

(8)

where

ˆL

??? vs(k) −?? vr(k) −ˆR?is(k)

?

(9)

? s(k + 1) = ? s(k) + ∆? s(k) = ? s(k) + ∆p(k) + j∆q(k) =

= ? vs(k) ·?is(k)∗+ ∆?? vs(k) ·?is(k)∗+?? vs(k + 1) · ∆??is(k)∗

Replacing (9) in (10), the change in apparent power is:

(10)

∆? s(k)=∆?? vs(k) ·?is(k)∗+ ···

(11)

?∗

···+

?? vs(k + 1) ·Ts

ˆL

??? vs(k) −?? vr(k) −ˆR?is(k)

For a sinusoidal voltage source power supply, the estimated

?? vs(k + 1) is obtained by rotating in ∆θ = ωTsrads.

1140

Page 3

?? vs(k + 1) = ? vs(k) · ejωTs

∆?? vs(k) = ? vs(k)?ejωTs− 1?

(12)

(13)

The complex apparent power? s(k) is a function of the supply

voltage, and also changes with the rectifier voltage?? vr(k) that

j qref. Defining ∆? s0(k) as an independent voltage vector term

in?? vr(k), the following change in active and reactive power is

Ts

ˆL

can be manipulated to obtain the commanded value pref+

obtained

∆? s0(k)=

|? vs(k)|2· ejωTs+ ···

(14)

?

··· +? vs(k) ·?is(k)∗

??

1 −TsˆR

ˆL

?

ejωTs− 1

For a given reference in active and reactive power the

change in power for proper compensation becomes a function

of the converter voltage?? vr(k). The apparent power variation

the following sampling period, prefand qref, are given by the

following expressions

needed to change from the actual to the demanded value in

∆? s(k) = ∆? s0(k) −Ts

ˆL

??? vs(k + 1) ·?? vr(k)∗?

+j qref− ℑm{? s(k)}

?

(15)

? ǫs(k) = pref− ℜe{? s(k)}

?

In the OVSS algorithm, also known as predictive direct

power control, a cost function is evaluated for a set of the

converter voltages ? vr, and the value of this voltage providing

the minimum value for the cost function is employed in the

next control cycle [16]–[18]. In this case the cost function is

???

ǫp(k)

???

ǫq(k)

(16)

J(k) = ηp(ǫp(k) − ℜe|∆? s(k)|)2+ηq(ǫq(k) − ℑm|∆? s(k)|)2

(17)

where ηpand ηqcontrol the relative importance of the active

and reactive parts in the system.

The proposed control technique is based in the selection

of the voltage vector that minimized the cost function (17)

expressed by the active and reactive power errors. However,

since zero is the global minimum for the cost function, instead

of testing among several candidate vectors for the best choice,

the proposed techniques computes with a closed form formula,

the voltage vector for this minimum. Forcing to zero the cost

function (17), J(k) = 0,

? ǫs(k) − ∆? s(k) = 0

(18)

Replacing (15) and (16) in (18)

∆? s0(k) −Ts

ˆL

??? vs(k + 1) ·?? vr(k)∗?

= ǫp(k) + jǫq(k)

(19)

Finally, replacing (12) into (19), the absolute optimum

converter voltage required to attain the commanded active and

reactive power becomes

?? vr(k) = ? vrα(k) + j ? vrβ(k) =

This voltage is synthesized in the converter using standard

space vector modulation (SVM) [19]. As with other DPC

algorithms, the reactor parameters are required for computing

the estimated value of the power system voltages, the active

and reactive power and the update value for the converter

voltage indicated in (20).

The proposed algorithm has many advantages over existing

methods, among them it provides an instantaneous correction

of the active and reactive power flowing into the converter,

reduces the ripple in the instantaneous power and currents,

resulting in a low harmonic distortion and have low computa-

tional demands.

ˆL

Ts

?∆? s0(k) −? ǫs(k)

? vs(k) · ejωTs

?∗

(20)

III. SIMULATION RESULTS

The scheme shown in Fig. 1 has been modeled using the

space vector representation of the state variables [7]. Both,

the V − V and Scott transformers have been included in

these simulations. The rail road system is represented using

the measured harmonic currents distribution, injected to the

power system in the secondary side of each transformer [20].

The three phase power system is modeled using a space vector

Thevenin equivalent. Also, space vector representations of the

power transformer (V − V or Scott), IGBT converter and

the filter inductor are used in the simulation. The per unit

parameter used in simulations are shown in Table I.

Table I: Parameters of the filter scheme model

Lth

0.037

RTrx

0.01

LTrx

0.1

Rtrx

0.01

Ltrx

0.1

Rrec

0.005

Lrec

0.05

VDC

3

Table II shows the current Total Harmonic Distortion

(THD) and the unbalance relation between positive and

negative sequences (I2/I1) [21], for uncompensated and com-

pensated cases using V − V and Scott transformer. The

simulation uses maximum unbalance by operating on one

single phase circuit with the other under no load, which is the

most demanding operating condition. The active filter injects

the harmonic content used by the single phase rail road load.

The proposed control scheme reduces by more than 50% the

THD for both transformer connections.

Table II: Simulated total harmonic distortion and unbalance

Uncompesated

THD

.4065

.2054

.4015

.2264

Compensated

THD

.1804

.1110

.1228

.0747

Simulated Cases

V − V rectifier

V − V rail road

Scott rectifier

Scott rail road

I2/I1

.9238

.9216

.9235

.9384

I2/I1

.0835

.0732

.0838

.0813

1141

Page 4

Figure 3 shows the simulated instantaneous currents flowing

into the power system without compensation and with the

proposed active and reactive compensation. The simulations

show the balancing effect on the power system current as

well as the THD reduction obtained with the active filter

controlled by the instantaneous active and reactive power.

Both transformers (V − V and Scott) have a similar current

behavior when a single phase rectifier load is connected in one

secondary.

IV. EXPERIMENTAL RESULTS

For the experimental test, the proposed algorithm was

implemented on a custom build floating point DSP (ADSP-

21061-40 MHz) based test-rig. The power stage uses six 50A,

1200V, IGBTs with two 2200 µF 400 V series connected ca-

pacitors in the DC link. The input inductors have an 7mH; the

PWM signals are provided by a motion co-processor ADMC-

201 operating at 10 kHz. The rail road load was implemented

in only one single phase circuit using a single phase rectifier

bridge with an R-L (50-200 Ω, 40 mH) load in the DC side.

The sampling frequency is synchronized by the motion co-

processor at the beginning of each PWM cycle. Fig. 4 shows

the power module and the DSP based processing unit. The

electrical parameters for the power circuit in the experimental

tests are the same to those used in the simulations, and shown

in Table III. The V − V and Scott transformer connections

were built using two single-phase – 480 : 240 – 120 V, 1

kVA transformers. The Scott transformer was built with two

additional single phase variable transformers.

Table III: Parameter test-rig

Rrec

20mΩ

Lrec

7.0mH

CVDC

Ts

V

1100µF

200 ∼ 600V

100µs

LLOAD

17mH

208V

ffs

RLOAD

50Ω

CLOAD

2200µF

60Hz

100kHz

A. V − V Transformer

Figures 5a to 5d show the current waveforms and spectrums

measured on a three phase V − V transformer test bench

feeding a non-linear load, with and without the proposed

compensation scheme. The measurements were obtained us-

ing a power quality analyzer type “B” [22]. Comparing the

compensated and the uncompensated results, it can be ob-

served that the compensator reduces unbalance and harmonic

distortion in the system. The unbalance is reduced from 94.7%

(uncompensated value) to 16.8% (compensated value), and the

system’s current THD to values below 18.4% in all phases,

with a significant reduction in the third and seventh harmonics

which are the more significant components present in the

harmonic spectra generated by vector controlled converters

used in locomotives.

(a) Uncompensated (V − V transformer)

(b) Compensated (V − V transformer)

(c) Uncompensated (Scott transformer)

(d) Compensated (Scott transformer)

Figure 3: Simulated active filter effect on power system

currents feeding a single phase rectifier load

1142

Page 5

Figure 4: Experimental rig

B. Scott Transformer

Figures 5e to 5h show the current waveforms and the

harmonic spectrum, measured on a three phase Scott trans-

former test bench feeding a non-linear load, with and without

the proposed compensation scheme. The non-linear load was

the same used in the V − V transformer case. Comparing

the compensated and the uncompensated results, it can be

observed that the compensator reduces harmonic content and

balance the three phase load. The compensator reduces the cur-

rent unbalance from 94.7% (uncompensated value) to 16.8%

(compensated value), and the system’s current THD to values

below 18.4% in all phases, with a significant reduction in the

third and seventh harmonics which are the more significant

components present in the harmonic spectra generated by

vector controlled converters used in locomotives.

V. CONCLUSIONS

The proposed compensation scheme reduces negative se-

quence currents that circulate in the uncompensated system

feeding an electric traction system using a power system trans-

former connected in V −V or Scott configuration. The scheme

reduces the current THD to values allowed by international

regulations, and regulates the power factor observed in the

common coupling point between the traction substation and

the power system. The proposed compensation scheme im-

plementation using an instantaneous power control algorithm

with direct space vector representation, reduces the system’s

current THD to allowable ranges (< 20%) and reduces the

overall unbalance from 97% to 18% for worse case operation.

The compensation algorithm is able to control the power factor

measured at the coupling point under all considered conditions.

From the simulation and experimental results it is found that

there is a compromise between the amount of unbalance

correction and harmonic reduction that can be achieved. This

is due to the finite amount of energy stored by the active filter

in its input inductance and dc-link capacitor.

(a) Uncompensated (V − V )(b) Harmonics uncompensated (V −

V )

(c) Compensated (V − V )(d) Harmonics compensated (V − V )

(e) Uncompensated (Scott) (f) Harmonics uncompensated (Scott)

(g) Compensated (Scott)(h) Harmonics compensated (Scott)

Figure 5: Experimental active filter effects on power system

currents feeding a single phase rectifier load

ACKNOWLEDGMENT

The authors want to express their gratitude to the Dean

of Research and Development Bureau (DID) of the Simón

Bolívar University, for the annual financial support provided

to the GSIEP (registered as GID-04 in the DID) to perform

this work.

REFERENCES

[1] B.-K. Chen and B.-S. Guo, “Three phase models of specially connected

transformers,” Power Delivery, IEEE Transactions on, vol. 11, no. 1,

pp. 323–330, 1996.

[2] G. W. Chang, H.-W. Lin, and S.-K. Chen, “Modeling characteristics

of harmonic currents generated by high-speed railway traction drive

converters,” IEEE Transactions on Power Delivery, vol. 19, pp. 766–

733, apr 2004.

1143

Page 6

[3] R. E. Morrison, “Power quality issues on ac traction systems,” in

Proceedings Ninth International Conference on Harmonics and Quality

of Power, 2000, vol. 2, pp. 709–714, 2000.

[4] S. T. Senini and P. J. Wolfs, “Novel topology for correction of un-

balanced load in single phase electric traction systems,” in IEEE 33rd

Annual Power Electronics Specialists Conference, 2002. pesc 02. 2002,

vol. 3, pp. 1208–1212, 2002.

[5] M. Goto, T. Nakamura, Y. Mochinaga, and Y. Ishii, “Static negative-

phase-sequence current compensator for railwaypower supply system,”

in International Conference on Electric Railways in a United Europe,

1995., pp. 78–82, mar 1995.

[6] R. Cutri and L. M. Jr, “Reference currents determination techniques for

load unbalance compensation,” Tech. Rep. 11, 2005.

[7] J. M. Aller, A. Bueno, and T. Pagá, “Power system analysis using space

vector transformation,” IEEE Transaction on Power System, vol. 17,

no. 4, pp. 957–965, 2002.

[8] H. Li, F. Zhuo, L. Wu, W. Lei, J. Liu, and Z. Wang, “A novel

current detection algorithm for shunt active power filters in harmonic

elimination, reactive power compensation and three-phase balancing,”

in IEEE 35th Annual Power Electronics Specialists Conference, 2004.

PESC 04. 2004, vol. 2, pp. 1017–1023, jun 2004.

[9] J. M. Aller, A. Bueno, J. Restrepo, M. I. Giménez, and V. Guzmán,

“Advantages of instantaneous reactive power definition in three phase

system meausrements,” IEEE Power Engineering Review, vol. 19, no. 6,

pp. 49–50, 1999.

[10] S. A. Larrinaga, M. A. R. Vidal, E. Oyarbide, and J. R. T. Apraiz,

“Predictive control strategy for DC/AC converters based on direct power

control,” IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS,

vol. 54, pp. 1261–1271, June 2007.

[11] P. Antoniewicz, M. P. Kazmierkowski, S. Aurtenechea, and M. A.

Rodriguez, “Comparative study of two predictive direct power control

algorithms for three-phase AC/DC converters,” in Power Electronics and

Applications, 2007 European Conference on, (Aalborg), pp. 1–10, Sept.

2007.

[12] P. Cortés, J. Rodríguez, P. Antoniewicz, and M. Kazmierkowski, “Direct

power control of an AFE using predictive control,” IEEE Transactions

on Power Electronics, vol. 23, pp. 2516–2523, Sept. 2008.

[13] M. Giménez, V. Guzmán, J. Restrepo, J. Aller, A. Bueno, J. Viola,

A. Millán, and A. Cabello, “Plataforma: Development of an integrated

dynamic test system for power electronics systems performance analy-

sis,” Revista de la Facultad de Ingeniería – UCV, vol. 23, no. 3, pp. 91–

92, 2008.

[14] Y. Cheng, C. Qian, M. L. Crow, S. Pekarek, and S. Atcitty, “A

comparison of diode-clamped and cascaded multilevel converters for a

statcom with energy storage,” IEEE TRANSACTIONS ON INDUSTRIAL

ELECTRONICS, vol. 53, pp. 1512–1521, Oct. 2006.

[15] J. A. Restrepo, J. M. Aller, J. C. Viola, A. Bueno, and T. G. Habetler,

“Optimum space vector computation technique for direct power control,”

IEEE Transactions on Power Electronics, vol. 24, pp. 1637–1645, June

2009.

[16] J. Restrepo, J. Viola, J. M. Aller, and A. Bueno, “A simple switch

selection state for svm direct power control,” in Proceedings of the 2006

IEEE International Symposium on Industrial Electronics, ISIE 2006,

pp. 1112–1116, July 2006.

[17] S. Larrinaga, M. Rodríguez, E. Oyarbide, and J. Torrealday, “Predictive

control strategy for DC/AC converters based on direct power control,”

IEEE Trans. Ind. Electron., vol. 54, pp. 1261–1271, June 2007.

[18] P. Antoniewicz, M. P. Kazmierkowski, S. Aurtenechea, and M. A.

Rodriguez, “Comparative study of two predictive direct power control

algorithms for three-phase AC/DC converters,” in Proceedings of the

2007 European Conference on Power Electronics and Applications, EPE

2007, pp. P1–P10, Sept. 2007.

[19] M. Malinowski, “Adaptive modulator for three-phase pwm recti-

fier/inverter,” in Proceedings of the 9th International conference on

power electronics and motion control EPE-PEMC 2000, vol. 1, (Kosice,

Slovak Republic), pp. 35–41, Sept. 2000.

[20] S. R. Huang and B. N. Chen, “Harmonic study of the le blanc

transformer for taiwan railway’s electrif ication system - power delivery,

ieee transactions on,” IEEE TRANSACTIONS ON POWER DELIVERY,

vol. 17, no. 2, pp. 495–499, 2002.

[21] IEEE, IEEE 519-1992 Recommended practices and requirements for

harmonic control in electrical power systems. pub-IEEE-STD:adr: IEEE

Standards Coordinating Committee, 1992.

[22] IEC, IEC 61000-4-15 Testing and measurement techniques – Flickerme-

ter Functional and Desing Specifications. International Electrotechnical

Commission Standard, 1997.

1144