IEEE Transactions on Power Delivery, Vol. 12, No.

1, January 1997 273

Applying PWM to Control Overcurrents at Unbalanced Faults of
Forced-Commutated VSCs Used as Static Var Compensators
Y. Jiang, Member, IEEE A. Ekstrom, Fellow, IEZE
Department of Electrical Engineering
Royal Institute of Technology
S 100-44, Stockholm, Sweden

Abstrucf -This study is devoted to investigating the overcurrents to the converter. Tliis paper presents one way
the possibility of controlling the overcurrent of a to control the overcurrents to the Converter when the ac
forced-commutated voltage source converter (VSC) system is severely unbalanced. Moreover, it also shows that
by PWM when the ac system is undergoing large the second harmonic on the dc side of the converter caa be
unbalanced disturbance. The converter is supposed
to be used as a static var compensator at a high
reduced substantially when the overcurrent is controlled with
power level. A novel control strategy is proposed the proposed method.
for controlling the reactive current and the dc side
voltage independently. Digital simulation results 11. CONTROL STRATEGY
are presented and compared with the results by
using just the reactive current control with A. Cause of the overciirrent
fundamental switching frequency.
Fig. 1 shows the configuration of the studied forced-
I. INTRODUCTION commutated VSC. The operation principle has been well
described in [2,4,5,6], thus it will not be discussed here.
Gate turn-off-thyristors (GTO) have now reached power Normally, the converter generates balanced three phase
handling levels comparable to conventional thyristors. This voltages. It means that the output voltage of the converter
will make it more attractive to use voltage source converters contains only a positive sequence component and the negative
(VSC) for reactive power compensation, where the VSC is sequence component will be zero. Now, if the ac system
called Advanced Static Var Compensator (ASVC) or voltages are asymmetrical, with a high negative sequence
STATCOM. Compared with the conventional SVC, the component, a negative sequence current will be produced
ASVC or STATCOM offers several benefits [1,2,3,4]. according to eq. (1).
H o ~ e v e r when
, the ac system voltages are unbalanced due to
single phase faults in ac networks the 2nd harmonic
oscillations will appear on the dc side of the converter and the
currents to the converter will be very large in some phases.
This may lead to high voltage and current stresses, thus the
converter has to be blocked in order to protect from destroying where a bar on the top of symbols denotes a phasor, X is the
[4]. It should be noticed that in reality it is when the ac leakage impedance of the ccnverter transformer, and VL1+
system is undergoing large disturbances that the reactive represents the fundamental negative sequence component o f
power compensation is often most needed in order to support the ac line voltage.
the voltage and to assist quick recovery. Therefore, blocking If the converter is working at absorption or generation I pu
the converter at temporary ac faults is unfavourable. reactive power, the total phase currents to the converter might
In order that the ASVC can continue to operate at seriously easily exceed 1 pu in some phase as long as the ac system is
unbalanced ac system conditions, two problems need to be asymmetrical. Therefore, the negative current ought to be
solved: 1. Damping the 2nd harmonic oscillation at the dc controlled to zero in order to avoid overcurrents.
side, to which the solution can be using a properly designed 1 1 1
second harmonic filter connected to the dc side; 2. Limiting

This paper was presented at the 1995 Stockholm Power Tech

Conference held in Stockholm, Sweden, June 18-22, 1995. Fig. i . Coqfigurstim of the 1Zpiilse forced-corninutated VSC

0885-8977/97/$10.00 0 1995 IEEE

274
One way to control the negative sequence current to zero is of the converter gives eq. (7). Equation (8) describes the dc
to control the internal converter bridge voltages in such a way side of the converter.
that the same amount of the negative sequence voltage is:
generated at the converter bridges as the ac bus

B. Voltages represented in G@, p n and dq-jam

dipq
A set of three instantaneous phase variables can be fully v~~~- vvpq= Rips c L----
dt
+ Xipd
described by two independent quantities, if the zero sequence
or homopolar component is disregarded. Two such quantities
are the so-called a-p components. They can be obtained from
three phase variables (voltages or currents) by the following (7)
transformation (non-power invariantj:

where L and R are respectively the leakage inductance and

resistance of the converter transformer, zdc is the dc side
impedance, subscript "L" denotes variables on the line side,
It should be pointed out that the purpose for using the a-p subscript "v" denotes variables on the converter side,
conversion is that the positive and negative sequence subscript "p" denotes the positive sequence components, and
components can be separated faster in ap components than in subscript "dc" denotes variables on the dc side of the
other forms. The dq-components of the positive sequence can converter.
be obtained from the positive sequence ab components as eq. For small deviations around a chosen steady-state
(3),whereas the dq-components of the negative sequence can equilibrium point it is possible to linearize the above
be obtained by eq. (4). Fig. 2 illustrates the block diagram for equations. Assuming that the bus voltage is constant and the
calculating the positive and negative dq components. phase-shift angle 6 is equal to zero at steady-state and
neglecting the second order terms, the linearisation process
yields the following perturbation equations in Laplace form:
(3)

where 0 is equal to cot, and o is the fundamental angular

frequency.

C. Reactive current and dc side voltage control

The control system should, besides controlling of the

negative sequence current to zero, control the reactive current
in the positive sequence to the reference value and the dc side
voltage within its limitation. Following analysis will lead to where s is the Laplace operator, 0 in the subscript denotes the
the control laws. steady-state quantity. Solving equations (9-12) for Aipq,
In the dq coordinate system, the relationship between the Aipd a n d Avdc gives expressions for the following
voltages and currents of the converter can be formalized as eq. decaupled relationships:
(5) and eq. (6) [SI. Power balance between the ac and dc side

Fig. 2. Separation of the negR!ive arid positive sequence components

 According to eq. (13) the reactive current feedback control
loop can be formed as shown in Fig. 3.
The active current control loop can be designed in a similar
I I
way. It should be noted that the objective of the active current
control is to achieve a desired dc side voltage by delivering the
necessary active power. Thus, the active current reference is
derived from the controller of the dc side voltage. Therefore, Fig. 4. Control loops for active cdrrent and dc side volrage.
the whole control loop for the active current will be as
outlined in Fig. 4.
This kind of dc voltage control strategy has been adopted in
both [7,8] and some other PWM VSCs used as static var
2
compensators. The main disadvantage of the control scheme
is that not only two regulators have to be passed from the I
input to output, but also that the control action is delayed due _I
to the low-pass filter used for filtering harmonics in the active
current ipd. The current ipd may contain harmonics from the Fig. 5. dc voltage control via phase-shift angle.
lowest order, i.e., the second, to all higher orders during the
unbalanced operation. To obtain a low-noise input signal, at So far, the reactive current control and the dc side voltage
least a 3rd order low-pass filter with a cutoff frequency of 35 control has been discussed. The whole control circuit can now
Hz has to be used, which can introduce a delay of 6-8 ms. be outlined by Fig. 6, which includes the special control for
Moreover, due to the dq transformation the dynamics of the unbalanced operation. The control circuit includes two
phase-locked-loop is also introduced to the control loop and feedback control loops, which are the reactive current loop and
disturbs the control at transients. These disadvantages may the dc voltage loop, and a feedforward loop control of the
explain why the result was not so good even though the negative sequence component of the converter voltage.
optimized compensators were used in the control system [7].
In this study, it was also found to be very difficult to D.Calculation of the desired three phase 1 olrage vectors
coordinate two regulators and almost impossible to achieve as
good results as the other method that is the de voltage via In Fig. 6 the desired dq components in positive sequence
phase-shift angle 6 control [5,6]. are given by:
Fig. 5 illustrates the block diagram for the dc voltage via
phase-shift angle 6 control. The background for this control
scheme is that the transmitted active power which charges or
discharges the capacitor is determined by the phase-shift angle
6. This control scheme was proposed for the dc side voltage
control of a forced-commutated VSC with a fundamental
switching frequency, but it has not been used in a PWM VSC The desired negative sequence dq components are equal to
so far. that of the line voltages, i.e.,
The two alternative dc voltage control schemes shown in
Fig. 4 and 5 are basically functioning in a similar way.
However, the dc voltage controi shown in Fig. 5 has the
advantage of more direct control with less regulators compared
with the scheme shown in Fig. 4. Therefore, the control
Vvnq = VLnq (19)
scheme shown in Fig. 5 is introduced for controlling the dc
side voltage of the VSC with PWM during unbalanced faults. Phasors in the positive sequence car. be generated by using
of the following equations:

L---+G;sz+-J
Fig. 3. Reactive current control lcop
Phasors in the negative sequcnce ciin be obtained in the
same way.
276

I - I
Fig. 6 . Overcurrent control scheme with PWM.

The desired three phase voltages can be obtained by inverse

sequence decomposition:
r- 1

In order to keep the modulation indices (refer to ma, mb and

mc in Fig. 6) for all the bridge arms in the range of 0-1 and Fig. 7 Test system for simulation of a single-phasc fault
to reduce the switching losses as much as possible, the dc
In Fig. 8, trace (a) shows the fault phase boltage, wherecis
side voltage reference is chosen such that the bridge arm with
trace (b) shows one of the control variables that IS the reactive
the highest desired voltage has a modulation index equal to 1 ,
current. It can be seen that the reactive current responsc
which means that it will operate with fundamental switching
follows the reference (+1 pu) well except for transients See
frequency.
also trace (b) in Fig. 9, it can be found that w ~ t h o uthe ~
overcurrent control scheme the oscillation at the frequency of
111. SIMULATION RESULTS
the second harmonic is very large in the reactive current.
Trace (c) in Fig. 8 illustrates the measured dc side voliage.
In order to validate the proposed overcurrent control strategy,
the system illustrated in Fig. 7 has been tested with EMTP Compared with trace (c) in Fig. 9, it can also De observed that
the dc side voltage is less fluctuating with the overcurrent
simulation. It should be stressed that in high power
control.
application the commutation frequency of converters has to be
Traces (d)-(f) in Fig. 8 shaw the thrce phase curren!s to the
kept as low as possible in order to reduce the switching
converter. It can be seen that the overcurrents are i n good
losses, and it is often equal to the fundamental frequency. For
control except for one higher current peak at the beginning of
that reason, the PWM method presented in [9], which uses a
the fault, which is mainly due to the delay i n the
switchicg frequency of five times the fundamental frequency,
was adopted. Moreover, the PWM is only used for a short
measurement circuit. However, without the overcurrent
control the current in the fault phase is almost 3 times ar
period of time when the ac system is undergoing an
unbalanced fault. Considering the limited space of the paper, large as the rated current value which can be ceen on [lacs (d)
of Fig. 9. A close study may find that cven the Iiighcrt
only the results from one fault case will be presentcd. A close
current peak with the overcurrent control is half per-unit
single phase fault was made by closing the breaker BB during
the time interval 400-600 ms. The reactive current reference snialler than that without the overcurrent control Due to the
was set to be + I pu. Fig. 8 shows the results with t h e coupling of the ac network impedance the o-iercurrent also
proposed overcurrent control scheme during the fault, and Fig. appear on the non-faulted phase R without the oveiciirrent
control, as can he seen on trace (e) of Fig. 9
9 illustrates the results without the overcurrent control. All
the outputs arc scaled in per-unit.
 277

IV. COMPARISON OF THE OSCILLATION

POWER BETWEEN AC AND DC SIDE OF THE
CONVERTER DURING UNBALANCED FALILT
WITH DIFFERENT CONTROL
-1 5
2.0 1 (b) For the dc voltage or reactive current control without
---.\rx-1
f
control of the negative sequence component [5,6],the intzrna!
converter voltage in fundamental frequency will only have a
0.0 ----.- --- ----C----+--_--t-----l- --J- positive sequence component. In this case the negative
2.0 sequen:e Component of the currents will be calculated
according to equation (I), and the oscillation power with the
V frequency of second harmonic will be:
00 L----+---..--_, -c-

: ~ dVJnej2Ut 3
------+------

p(rj =
(23)

On :he other hand, if the converter is controlled to produce

-3.0
both positive and the same amount of negative sequence

-3 0
/~~,~,,p,~fi~$~//+/$
r\,A A
wp"[
1 component as that in the ac bus voltage with the overcurrent
control, the negative sequence component of currents will be
zero. The second harmonic oscillation power will then be:

c__c__----c , : I -+
t It can be concluded that the amplitude of the oscillation
TIME. 350 - 700 ms. 35 ms/div. power with the frequency of second harmonic at rated current
Fig 8 Simulation results for a single-phase fault with overcurrent control. 'peration be "x (' in pu) times larger w i t h o u t
controlling the negative sequence current to the converter than
that with the presented overcurrent control. If X is equa: to
0.2 pu (usually the leakage inductance of the convertel
transformer is designed to have a value around 0.2 pu), the
amplitude of the oscillation power with the overcurrent
-1.5 1 c control can be reduced to one-fifth of that without controlling
2.0 1 'b) i the negative sequence current. When the converter works at no
reactive power generated or absorbed, the oscillation power
with the frequency of second harmonic will be zero if the
negative sequence current to the converter can bc fully
ellnunated.
Since the capacitor will be charged or discharged when the
power i s transferred to or transferred from the converter, the dc
side voltage oscillation will reflect this power oscillation
directly. Fig. 10 shows one example in which ac bus voltage
in phase A is reduced to 50% in the amplitude Fig 10a
shows the dc side voltage with the proposed overcuirenl
control and Fig. 10b shows the dc side voltage with just the
reactive current control [5]. In this simulation the reference
for the reactive current is set to zero, no second iiarmrtnic
filter is connected to the dc side and the dc side voltagc conml
loop is open. Those are the main difference between the
simulations given for Fig. 8, 9. Due to the diZcrcnt
skmulatIon conditions the dc side voltage shown i n 1.1g 10
are also different from that shown in trace (c) in Flg. 6 , 9
The important point is that Fig. 10b reveals the larg: S C C C ~
-3.0 t harmonic stress on the dc side filter in the case of F I ~9 .
~ ~p-e--+-i--(----c---*

TI!*IE: 350 - 700 ms.35 msldiv.

Fig. 9. Simulation results for a single-phase fault without overclirrent

control.
\
278
VII. REFERENCES

E Larsen, N Miller, S N h o n and S Lindgren, "Benefits of GTO-

Based Compensation Systems for EleLtnc Utility Applications", IEEE
Transactions on Power Delivery, Vol7, No 4, pp 2056-2064, October
1992
D Wuest, "An Improved PWM Optimization Method For a Reactive
Power Compensator with Self-commutated Inverter", 22nd Annlial
Power Electronics Specialists Conference, 91CH3008-O,MIT, MA,
24-27 June, 1991, pp 736-768
G D Galanos, C I Hatziadoniu, X-J Cheng, D Maratukuiam,
"Advance Static Compenmtor for Flexible AC Transmssion", IEEE
Transactions on Powzr Systems, Vol 8, No 1, pp 113-121, February
1993
Y Sumi, Y Harumoto, T Hasegawa, M Yano, "New Static Var
Control Using Forced-commutated Inverters", E E E Trans on Powei
Apparatus and Systems', Vol PAS-100,No 9 September 1981, pp
4216-4224
T I M E IN M I L L I P E C O N D B M M de Ohveira, Ake Ekstrom, " Transfer Fbnctions for Voltage-
(a) source Converter Operating as an SVC", 2nd Brazilian Power
Electronics Conference COBEP'93, Uberlandia, Minas Gerais, Brazil,
November 29 - December 2, 1993, pp 167 - 172
G Joos, L Moran, P Ziogas, A Solid Static Var Compensator with
Fast Dynamic Response", Proceedings of the 20th Anual IEEE Power
Electronics Specialists Conference (PESC), pp 400-409, Milwaukee-
WI, USA, June 1989
Rioual, P , Pouliquen, H., "Regulation of a PWM rectifier in the
unbalanced network state", Proceedings of the 24th Annudl IEEE
Power Electronics Specialists Conference (PESC). pp 641-647,
Seattle, Washington USA, June 1993
C Schauder, H Mehta, "Vector Analysis and Control of Advanced
Static Var Compensators", IEE Proceedings-C, Vol 140, No 4, July
1993, pp 299-306
Bouwknegt K , "Amplitude Control in Twelvepulse Voltage Source
Inverter", EPE Journal, Vol 2, No 1, March 1992, pp 34-39
, ! , , , I
--1,90

vm. BIOGRAPHIES
200.00 240.00 280.00 320.00 360.00 400.00

T I M E IN M I L L I S E C O N D S
Ying Jiang received her B Sc and M Sc degrees in electrical
engineenng from Huazhong University of Science and Technology, China,
(b) respectively in 1984 and 1987 She was with the Department of Electncal
Fig. 10. DC side voltages. a: with overcurrent control; b: without Engineering of Huaihai University from 1987 to 1991, as an assistant
overcurrent control. professor Since then she has been studying in the Department of Electncal
Engineering of Royal Institute of Technology in Sweden and has been one
V. CONCLUSIONS of the key participants in the EPRI project, "Evaluation of Performance of
Static Condensers" She has received the licentiate degree in 1994
Presently she is working toward her Ph D Her areas of interest are FACTS
This reveals that it is possible to control the overcurrent of devices with regarding time domain simulation and mathematical analysis,
and harmonic interactions in the power system with converters
the forced-commutated VSC by PWM when the ac system is
undergoing a large unbalanced disturbance The proposed Ake Ekstrbm received his M Sc degree in electrical engineering from
PWM control strategy for balancing of the negative sequence Royal Institute of Technology, Sweden, in 1957 He joined ASEA since
1956 He had been the chief engineer in the HVDC Design Department,
component performs well in controlling the overcurrent manager of ASEA'HVDC converter office, Presently, he is senior vice
during the asymmetrical faults. Another benefit of the this president in Technology at the Power Transmiwon and Distribution
Segment ABB He also worked in IBM Nordic Laboratory as a manager of
control strategy is that it can also reduce the second harmonic the Application Technology Department Mr Ekstrom is also the professor
oscillation power substantially during unbalanced faults. at the Royal Institute of Technology, head of the Department of High
Power Electronics since 1986 He is the manager of the EPRI project,
"Evaluation of Performance of Static Condensers" He is ClGRE Member
VI. ACKNOWLEDGEMENTS of Study Committee No 14, DC Links, he is a member of the Royal
Swedish Academy of Engineering Sciences, he had been charman of SC
The authors would like to appreciate for the financial 22F, IEC Mr Ekstrom has published and presented over 30 papers and
holds several patents in hvdc design
support provided by EPRI, USA and Vattenfall AB, Sweden.
The authors also wish to thank Prof. W. F. Long, University
OC Wisconsin-Madison. and Stig L. Nilsson, EPRI, USA, for
participating in the project with valuable discussions and
reviewing of the paper. The assistances from our colleagues:
L. Lindberg and M. M. de Oliveira at the Royal Institute of
Technology have been very helpful.