1670 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO.

3, JULY 2009

Stability Analysis Based on Bifurcation Theory of the

DSTATCOM Operating in Current Control Mode
Juan Segundo-Ramirez, Aurelio Medina, Senior Member, IEEE, Arindam Ghosh, Fellow, IEEE, and
Gerard Ledwich, Senior Member, IEEE

Abstract—This paper presents the stability analysis for a distri- collapse [3], subsynchronous resonance [4], chaotic oscillations
bution static compensator (DSTATCOM) that operates in current [5], ferroresonance oscillations [6], and design of nonlinear con-
control mode based on bifurcation theory. Bifurcations delimit the trollers [7]. Furthermore, this theory has been applied to assess
operating zones of nonlinear circuits and, hence, the capability to
compute these bifurcations is of important interest for practical the dynamical behavior of nonlinear components such as in-
design. A control design for the DSTATCOM is proposed. Along duction motors [8], load models [9], [10], tap changing trans-
with this control, a suitable mathematical representation of the formers [11], and flexible ac transmission systems (FACTS) de-
DSTATCOM is proposed to carry out the bifurcation analysis effi- vices [12], [13]. In addition, bifurcation theory has been used to
ciently. The stability regions in the Thevenin equivalent plane are analyze the stability of power converter and nonlinear switched
computed for different power factors at the point of common cou-
pling. In addition, the stability regions in the control gain space, circuits [14]–[22]. In [20] and [22] a wide collection of re-
as well as the contour lines for different Floquet multipliers are sults related to nonlinear phenomena in power electronics is pre-
computed. It is demonstrated through bifurcation analysis that the sented.
loss of stability in the DSTATCOM is due to the emergence of a Conventional stability analyses in power systems are basi-
Neimark bifurcation. The observations are verified through simu- cally based on brute force approach and eigenanalysis [23]. In
lation studies.
these analyses, the system is modeled using root mean square
Index Terms—Distribution static compensator (DSTATCOM), (rms) quantities and the network dynamics are neglected. In this
floquet multiplier, Neimark bifurcation, point of common coupling
contribution, the power system is represented through instanta-
(PCC), stability regions.
neous quantities, the network transients are taken into account
and the electric sources voltage are assumed to be sinusoidal.
I. INTRODUCTION As far as we know, there is no application of this theory to
analyze electric systems, including DSTATCOMs devices. The
DSTATCOM operating in current mode has been developed to

I N dynamic nonlinear circuits, it is frequently encountered

that a steady-state response, such as an equilibrium point or
a periodic response, abruptly changes its qualitative property by
mitigate some problems related to quality of power, such as to
cancel the effect of poor load power factor, to suppress the effect
of harmonic content in the load current so that currents drawn
continuous variation of the system parameters. Such a phenom-
from the source is nearly sinusoidal, to eliminate any dc offset
enon is known as the bifurcation of state [1] and is important in
in the load current so that the current drawn from the source
the analysis of nonlinear circuits. If the set of bifurcation values
has no dc component, and to balance an unbalanced load. Sev-
in the parameter space is known, it is possible design an elec-
eral papers [24]–[29] have proposed different control strategies
trical circuit with the optimal operating condition. An important
to enhance its performance. However, the nonlinear oscillation
problem in dynamic nonlinear circuit analysis is the investiga-
produced by the DSTATCOM has not been analyzed in detail.
tion of the set of bifurcation values in the parameter space. Upon
In this paper, the bifurcation theory [1] is used to assess the im-
obtaining the global feature of the bifurcation set, various non-
pact produced on the stability by the DSTATCOM operating in
linear phenomena, such as the coexistence of many stable states,
current control mode [24].
the jump behavior of periodic responses, the phenomenon of
This paper is organized as follows: Section II describes the
hysteresis and the appearance of chaotic states, etc., can be ob-
DSTATCOM operating in current control mode; in Section III, a
served.
simplified representation of the DSTATCOM is proposed, based
The bifurcation theory has been used in stability analysis [2];
on a continuous set of ordinary differential equations (ODE);
this facilitated the demonstration of chaotic motions in the two-
Section IV gives a concise description on the continuation tech-
degree freedom swing equations. Subsequent applications of
niques and the bifurcation theory; Section V presents the results
this theory have been directed to diverse studies such as voltage
of the bifurcation analysis; Section VI presents a discussion of
the results, and Section VII offers the conclusion drawn from
Manuscript received August 25, 2008; revised November 06, 2008. Current this investigation.
version published June 24, 2009. This work was supported by CONACYT.
Paper no. TPWRD-00618-2008. II. DSTATCOM OPERATING IN CURRENT CONTROL MODE
J. Segundo-Ramírez and A. Medina are with the Universidad Michoacana de
San Nicolás de Hidalgo (UMSNH), Facultad de Ingeniería Eléctrica, Morelia The DSTATCOM is a shunt-connected device similar to
58030, Mexico. the static compensator (STATCOM) [30]. However, there are
A. Ghosh and G. Ledwich are with the School of Engineering Systems,
Queensland University of Technology, Brisbane Qld 4001, Australia. important differences in the operating characteristic between
Digital Object Identifier 10.1109/TPWRD.2009.2016817 the DSTATCOM and the STATCOM. The STATCOM injects
0885-8977/$25.00 © 2009 IEEE

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
SEGUNDO-RAMIREZ et al.: STABILITY ANALYSIS BASED ON BIFURCATION THEORY 1671

Fig. 2. Compensation of the EAF when the source is nonstiff and the
DSTATCOM contains a passive filter.

B. DSTATCOM Control

The following nonlinear system periodically excited with

Fig. 1. Structure of the DSTATCOM.
a -periodic function, describes the dynamic behavior of the
equivalent circuit of the compensated system shown in Fig. 2

(1)
a set of three balanced quasi-sinusoidal voltages. On the other
hand, the DSTATCOM must be able to inject an unbalanced
where and are the -dimensional vectors, and is an -di-
and harmonically distorted current. Therefore, its control is
mensional parameter vector.
significantly different from that of a STATCOM.
In particular, for the electric system shown in Fig. 2, the func-
In the current control mode, the DSTATCOM compensates
tion is defined as
for any unbalance or distortion in the load, thus, the load draws a
balanced current from the system irrespective of any unbalance
or harmonic distortion in the load [24]. One of the most impor-
tant issues for the load compensation is the generation of the (2)
reference compensator currents. There are several techniques
proposed [24]–[29]. However, most of these methods assume
that the voltage at the PCC is stiff. Unfortunately, this is not a
valid assumption for most practical applications. and the state vector is defined as
For this particular analysis, the computation of the reference
currents will be done using instantaneous symmetrical compo- (3)
nents [25]. In addition, the source is not assumed to be stiff.
where is the transpose operator.
The nonlinear load is an electric arc furnace (EAF). The dy-
namic behavior of the – characteristic of the EAF is described
A. DSTATCOM Structure
by the differential equation introduced in [31]. This differential
equation is based on the principle of energy balance. Starting
In order to cancel out unbalance or harmonics in the line
from the power balance equation for the electric arc, the fol-
current the voltage-source converter (VSC) that constitutes the
lowing differential equation is derived in [31]:
DSTATCOM must be able to inject currents in one phase in-
dependent of the other two phases. From this point of view the
structure of a DSTATCOM is very important. (4)
The DSTATCOM structure adopted in our analysis is shown
in Fig. 1. It contains three H-bridge VSCs connected to a Here, the arc radius is chosen as a state variable. The arc
common dc storage capacitor. Each VSC is connected to the voltage is given by
network through a transformer. The purpose of including the
transformers is to provide isolation between the inverter legs. (5)
This prevents the dc capacitor from being shorted through
switches of the different inverters. The structure shown in Fig. 1 where is the arc conductance given by the following equation:
allows three independently current injections. It is to be noted
that due to the presence of transformers, this topology is not
suitable for canceling any dc component in the load current (6)
[24]. The inductance represents the leakage inductance of
each transformer and additional external inductance, if any. It is possible to represent the different stages of the arcing
The switching losses of an inverter and the copper loss of the process by simply modifying the parameters of and in (1).
connecting transformer are represented by a resistance . For The complete set of combinations of these parameters for dif-
more details about this structure, please see [24]. ferent stages of electric arc can be found in [31].

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
1672 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 3, JULY 2009

In (2), is the control signal constrained between and

. Once the reference currents are generated, they are tracked
in a hysteresis band current control scheme. The control signal
is computed through

(7)

where is the reference compensation current. These are given Fig. 3. Schematic representation for the simplified model.
by [25]

(8)

where is computed based on the load power factor and

Fig. 4. DC link model.

inconsistent initial condition can appear [32]. One way to miti-

In (8), is the average power drawn by the load, is the
gate the adverse effects related to the switching process is to use
power loss due to , and is the fundamental component
a small integration time step to carry out the simulation. How-
of , for .
ever, it takes a long simulation time.
The hysteresis function hys is defined by
The source of the numerical problems for the numerical in-
for tegration arises from the discontinuities and the nondifferentia-
(9)
for bility introduced by the ideal switch model [22].
For our analysis, the detailed model is not suitable because of
where is the hysteresis band. the difficulty to compute the limit cycle to high precision, and
The power loss is computed through the proportional for assessing its stability. For this reason, we propose a simpli-
controller [24], e.g., fied model, in which all of the aforementioned difficulties are
avoided. In the proposed representation, the three H-bridges are
(10) represented through controlled current sources, the link between
the dc side and the ac side is well represented using the energy
where is the reference dc voltage, and is the average preservation principle. In Fig. 3, the schematic representation
voltage across the dc capacitor. of the simplified DSTATCOM model in current control mode is
To compute , we introduce simple proportional-integral shown. Fig. 4 shows the schematic representation of the dc-link
control given by model. The power balance between the dc and ac side can be
given as
(11)

where

(12)
(14)
(13)
where (15)–(17) are shown at the bottom of the next page.
From (14), the dc current is computed as the ac power
and is the desired reference power factor at the PCC bus
divided by the dc voltage capacitor , thus, this simplified
and PF is the load power factor at the PCC.
model based on the energy preservation principle is limited for
; in case of , the dc power becomes infinite,
III. SIMPLIFIED AND DETAILED DSTATCOM MODELS which is not physically feasible.
In the detailed model, the switching elements are explicitly The simplified model is advantageous in comparison with the
represented. The modeling of the switching devices can be per- detailed model, since the simplified models can be described
formed with different levels of detail. A very detailed model can only with an ODE set, while the detailed models require a hybrid
be justified and necessary when the phenomena associated with representation.
the switching process are to be analyzed, albeit at high compu- Conventionally, the stability analysis based on the bifurcation
tational cost. On the other hand, for power systems studies, the theory of power-electronics devices, including the switching
study of the switching phenomena is not necessary. Thus, it is process, is carried out with the discrete-time iterative mapping
advantageous to model the switches as open and short circuits. approach [15]–[19]. Under this approach, the switching con-
However, this model has some disadvantages; for example, an verter is essentially modeled as piecewise switched circuits. The

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
SEGUNDO-RAMIREZ et al.: STABILITY ANALYSIS BASED ON BIFURCATION THEORY 1673

number of possible circuit topologies is usually fixed. This re-

sults in a nonlinear time-varying operating mode, which natu-
rally demands the use of nonlinear methods for analysis [20].
In the discrete-time iterative mapping approach, it is assumed
that between switching instants, the network is linear and time
invariant; thus, the solution between switches can be found in
closed form. The nonlinearity comes from the switching control. Fig. 5. Comparison in the time domain between the detailed and the simplified
For the case of the DSTATCOM, it is not easy to represent model for P in steady state for different integration steps.
its mathematical model through a discrete-time iterative map-
ping approach, since the network between switching instants
is nonlinear because of the nonlinear load in the network. For
this reason, the continuous model based on the energy preser-
vation principle is used to carry out the stability analysis based
on the bifurcation theory, using a continuation scheme to trace
the stability boundaries. The bifurcation analysis based on con-
tinuation schemes allows us to identify, not only the stability
boundaries, but the type of dynamics in each region, as well as
the type of bifurcation that emerges in the system. Obviously,
there are some disadvantages with the simplified DSTATCOM
model. Basically, the high-frequency phenomena are neglected.
However, to avoid erroneous interpretations in our analysis, the Fig. 6. Comparison in the time domain between the detailed and the simplified
bifurcation diagrams are validated against time domain simu- model for (a) compensation current i and (b) dc capacitor voltage v .
lation, carried out with the detailed model. If the dynamics of
network become close to switching frequency, this simplifica-
tion is likely to be in error, but in many cases, the switching sw is closed and the DSTATCOM starts the compensation. Se-
frequency significantly exceeds network transient frequencies, lected waveforms are presented in Figs. 5 and 6. Fig. 5 shows
thus the simplification would be able to be reliable. the steady-state solution of the power loss , with different
In the simplified DSTATCOM model, only the switching integration steps in the detailed model and with an integration
process is neglected, but the nonlinearities are retained as step of 65 s for the simplified model. From this figure, it is
demonstrated in Section V. easy to notice that if the commutation process has to be taken
into account, a very small integration step size (0.52 s) should
A. Comparative Analysis of Models be chosen with the detailed model; otherwise, the solution will
In this section, the performance of the simplified model is contain a high numerical error due to the commutation process.
compared against the detailed model. The test system is shown Fig. 6(a) shows the compensation current , while Fig. 6(b)
in Fig. 2. The system parameters and the DSTATCOM param- shows the dc voltage with the simplified model and with the
eters are given in Table I. The hysteresis band for the detailed detailed model for an integration step size of 65 s and 1 s, re-
model is 1 A. spectively. Good agreement between the two models has been
Initially, for , the switch sw is open and the elec- achieved, even though the simplified model has a considerably
tric circuit is in periodic steady state. At 0 s, the switch larger integration step size.

(15)

(16)

(17)

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
1674 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 3, JULY 2009

IV. CONTINUATION TECHNIQUES AND BIFURCATION THEORY V. DSTATCOM STABILITY ANALYSIS

BASED ON BIFURCATION THEORY
The transient and steady-state response of a system repre- In this section the bifurcation theory will be applied to the
sented by an ODE set can be computed by conventional numer- electric system shown in Fig. 2 to compute the stability regions
ical integration methods; this method is known as a brute force of the electric system including the DSTATCOM operating in
approach [1]. Therefore, the stability of any system can be com- current control mode. The simplified DSTATCOM model will
puted through time-domain simulations. Conventionally, in the be used to construct the bifurcations diagrams. The stability re-
case of the DSTATCOM, the system (1) has to be integrated gions computed through bifurcation analysis are also compared
to know the performance of the electric systems including the against the time domain simulation using both the detailed and
DSTATCOM. However, it is possible with bifurcation theory to the simplified DSTATCOM models.
predict the behavior of trajectories and orbits around the oper-
ating points without resorting to the numerical integration solu-
A. Bifurcation Analysis for DSTATCOM in Current Control
tion of the system (1). The results obtained with this analysis can
Mode
be represented in a bifurcation diagram, which provides quali-
tative information about the behavior of the steady-state solu- 1) Stability Regions in the Plane: The network of
tions (limit cycles), as physical parameters are varied. At cer- Fig. 2 has been represented through its Thevenin equivalent.
tain points (bifurcation points), infinitesimal changes in system The network upstream from the PCC towards the source side
parameters can cause significant qualitative changes in periodic may contain different feeders and loads. Thus the radial line and
solutions. In general terms, the construction of a bifurcations the source shown in Fig. 2 is a Thevenin representation of the
diagram consists of the following steps [1], [33]: 1) finding a upstream network, where , and represent the Thevenin
first periodic steady-state solution of (1); 2) based on the first equivalent looking towards the left into the network.
solution, find other equilibrium solutions based on a continua- Since the Thevenin equivalent can change any time de-
tion scheme [1], [33]; and 3) determining the stability of each pending on the load at left side of PCC, it is desirable to assess
solution. a set of , and , for which the DTATCOM performance
Continuation schemes are used to determine how the solu- is stable.
tions of a system, such as (1), vary with a certain parameter. For the electric system shown in Fig. 2, only the Neimark
Implementing a predictor-corrector scheme, a continuation al- bifurcation was located in the parametrical space used in this
gorithm can trace the path of an already established solution as analysis. In analogy with the Hopf bifurcation, bifurcation is
the parameters are varied. In this paper, the sequential method expected at a critical value since the limit cycle loses its sta-
[33] is used as the predictor; in this method, the periodic solu- bility, so that an attracting torus is born; this is the secondary
tion determined in the previous step is used as an initial guess for Hopf bifurcation or a Neimark bifurcation. Besides, the bifur-
the periodic solution to be determined in the next step. After the cated solution can be either stable and supercritical or unstable
third point, an extrapolation method based on the cubic spline is and subcritical [33].
used as a predictor. The Newton method, based on the numer- Fig. 7 shows the bifurcation set on the plane. This
ical-differentiation (ND) process [34], is used as the corrector. figure shows the stability regions for different power factor cor-
The stability of a periodic solution is computed from its Flo- rections with Volts, where is the peak value. The
quet multipliers; they describe the stability near the limit cycle solid line represents the Neimark bifurcation set. Inside the con-
of interest. Floquet theory is based on the observation that a pe- tour line the solutions are -periodic and the gray zones repre-
riodic solution can be represented through a fixed point of an sent the unstable regions. The stable region in the plane
associated Poincaré map [1], [33]. Consequently, the stability changes according to the power factor at the PCC. For instance,
of a periodic solution can be determined by computing the sta- Fig. 7(d) shows that for a 0.822 lead power factor, an unstable
bility of the corresponding fixed point of the Poincaré map. The region within the stable region exists.
Floquet multipliers are the eigenvalues of the Jacobian of this To corroborate the bifurcations diagrams shown in Fig. 7,
Poincaré map. Stable periodic solutions correspond to Floquet time-domain simulations were carried out. Fig. 8(a) shows the
multipliers inside the unit circle; on the other hand, unstable pe- phase portrait in the plane with 87.77
riodic solutions have at least one characteristic multiplier out- mH, and and Fig. 8(b) shows the phase portrait
side the unit circle. Therefore, loss of stability is encountered in the plane with 30 mH, and
when a multiplier leaves the unit circle; this can occur in three . It is to be noted that only the parameters men-
different ways: A fold bifurcation is encountered when a single tioned before are changed, while the rest of the parameters are
real Floquet multiplier crosses the unit circle at . The flip bi- those given in Table I. These solutions agree with the bifurca-
furcation or period-doubling bifurcation is given when a single tion analyses which predict quasiperiodic solutions. A compar-
real Floquet multiplier crosses the unit circle at . At this ison between the detailed and the simplified models is shown;
bifurcation point, the prevailing solution branch becomes un- good agreement is achieved between both solutions.
stable and a new branch is born. Solutions on this new branch Fig. 9 shows the simulated waveform for the compensation
have twice the period of the previous limit cycle. The general- current , the dc capacitor voltage , and the terminal voltage
ized Hopf bifurcation or Neimark bifurcation [1], [33] is found , for 87.77 mH, and . For
when two complex conjugated Floquet multipliers leave the unit this operating point, the dc capacitor voltage suddenly collapses
cycle. This bifurcation corresponds to a quasiperiodic solution. around 0.65 s. This operating point is in the unstable region;

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
SEGUNDO-RAMIREZ et al.: STABILITY ANALYSIS BASED ON BIFURCATION THEORY 1675

Fig. 9. Simulated waveforms for PF = 0 =

; L 87.77 mH, and R = 25 .
(a) Compensator current i . (b) DC capacitor voltage v . (c) Terminal voltage
v .

Fig. 7. Stability regions for the DSTATCOM operating in current control for
=
different power factors at the terminal bus with jV j 440 V.

Fig. 10. Stability regions for the DSTATCOM operating in current control for
jV j = = =
440 V, for jV j 400 V, and for jV j 300 V with PF = 1 .

In this section, we compute the stability region in the

space, and in the space, as well as the contour
lines for different Floquet multipliers.
Fig. 8. Phase portrait for different operating points. (a) Phase portrait in the
Fig. 11(a) shows the stability regions in the
v 0 i plane with PF = 1 ;L = 87.77 mH, and R = 12 1
: . (b) Phase space, and Fig. 11(b) shows the stability regions in the
portrait in the i 0 v plane with PF = 0 822 =
: ; L 30 mH, and R =1. space. Also, in these figures, contour lines are presented
for different Floquet multipliers to show the different speed of
response. For example, from Fig. 11(a), it is easy to notice that
however, since it is far from the Neimark bifurcation, the dc ca- the pair of gains 80 000 and 1040 give the
pacitor voltage collapses. In Fig. 9, only the solution with the fastest response.
detailed model is presented, since as we mentioned before, the Fig. 12(a) shows time-domain simulations of the convergence
simplified model does not give exact time-domain solutions for error for 1040 and different . It can be observed
. However, it has predicted the loss of stability cor- that this agrees with the bifurcation diagram of Fig. 11(a). From
rectly, as shown in Fig. 7(c). In addition, the bifurcation diagram Fig. 11(b), it is easy to notice that inside the stable region, there
[Fig. 7(c)] predicts the loss of stability due to the emergence of a is an important area for which the maximum Floquet multiplier
Neimark bifurcation. Please notice the presence of oscillations is constant. This means that for this area, the speed of response
in the time-domain solutions shown in Fig. 9. It indicates the should almost be the same. To corroborate this observation, the
existence of a Neimark bifurcation, which is in accordance with convergence error for and different are shown
the predicted behavior by the simplified model. in Fig. 12(b); as expected, the convergence error is almost the
To show the impact of the Thevenin equivalent voltage on same in this area.
the stability regions in the plane, Fig. 10 shows the Fig. 13 shows the torus solution for compensation current
bifurcation set for 440 V, for 400 V, and for for , and
300 V, with 1. Fig. 10 shows that the stable regions in the . This operating point corresponds to a quasiperiodic
plane decrease as the Thevenin voltage decreases. This solution. Note that the detailed model and the simplified model
is an important observation, since voltage sags can collapse the are in very good agreement even in the unstable regions.
system if this is operating close to the Neimark bifurcation. The correlation between the speed response of the
2) Stability Regions in the Gains Plane: There is a strong DSTATCOM and its stability is shown in Fig. 14. This
correlation between the speed response of the DSTATCOM and figure shows two bifurcation sets in the plane for two
its stability. In general, the set of initial conditions for which the different sets of gains. The first one corresponds to the nominal
solution is attracted to the stable limit cycle as well as the stable gains given in Table I, and the second one corresponds to the
regions decreases as the response becomes faster. fastest set of gains for the nominal electrical parameters given

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
1676 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 3, JULY 2009

Fig. 14. Comparison between the stability regions for two different sets of
gains.

Fig. 15. Comparison between the stability regions for different dc capacitors.

Fig. 11. Stability regions for the DSTATCOM operating in current control
mode in (a) the K 0K space and (b) K 0K
space.

Fig. 16. Comparison between the stability regions for different ac capacitors.

load demand. Fig. 15 shows the stability regions in the

plane for different dc capacitor sizes.
4) AC Capacitor Impact on the Stability: The purpose of the
filter capacitor is to provide a path for the switching har-
monic current introduced by the DSTATCOM. However, it is
Fig. 12. Convergence error for different gains in the DSTATCOM controllers.
(a) For the dc capacitor voltage controller. (b) For the power factor controller. shown in [24] that this passive filter has an important impact on
the DSTATCOM performance and on its stability. High capac-
itances in the filter capacitor provide a low impedance path for
the harmonic currents. However, there are three problems re-
lated to high capacitances. The first one is the cost, the second
one is that the speed response becomes slower, and the third one
is that the stable region decreases as the capacitance becomes
larger. This is shown in Fig. 16, where the stability region in the
plane has been computed for three different capacitor
Fig. 13. Quasiperiodic solution for K = 0:5; K = 300;K = 1040, filters.
and K = 2:5 2 10 i vs i. .
VI. DISCUSSION OF THE RESULTS
With the tracing of bifurcation diagrams, an accurate portrait
in Table I. From Fig. 14, it is possible to observe that the stable is drawn for the dynamic system, and the operating zones have
region decreases as the speed response becomes faster. The set been delimited.
of gains can be selected through an assessment of bifurcation Variation in the Thevenin equivalent can cause the limit
diagrams, such as those shown in Fig. 14. cycle to lose stability. This is because a Neimark bifurcation
3) DC Capacitor Impact on the Stability: The impact of the appears. The periodic solutions in the unstable region close to
dc capacitor size in the stability regions in the plane is the Neimark bifurcation become quasiperiodic and an attracting
qualitatively shown through bifurcation analysis. This analysis torus is born. However, for the operating points in the unstable
shows that the stable region increases as the dc capacitor size region far from the Neimark bifurcation, the system collapses,
increases. However, as the capacitor size becomes larger than as shown in Fig. 9.
a certain value, the stable region remains constant. From this With the tracing of bifurcation diagrams in the gains plane,
analysis, the size of the dc capacitor can be chosen to suit the the operating zones are also delimited and the speed response

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
SEGUNDO-RAMIREZ et al.: STABILITY ANALYSIS BASED ON BIFURCATION THEORY 1677

of the DSTATCOM is known. It is also shown in Fig. 14 that for TABLE I

the slower speed response of the DSTATCOM, the stable region SYSTEM PARAMETERS
is increased.
The dc capacitor size increases asymptotically the stable re-
gion as this becomes larger. Beyond a certain value, the stable
region remains nearly constant.
An interesting result is shown in Fig. 16. The ac capacitor
filter connected to the PCC bus to eliminate the high-frequency
component of the current injected by the DSTATCOM has an
adverse effect on the stability. For a small ac capacitor filter, the
impedance for the harmonic current is high. Therefore, some
harmonic currents injected by the DSTATCOM remain in the
system. On the other hand, for a larger ac capacitor filter, the
impedance to the harmonic currents is low. In consequence, the however, it is not necessary to carry out time-domain simula-
harmonic current is efficiently drained by the passive filter. Un- tions for each case. From this comparison, we have shown that
fortunately, the stable region decreases as the ac capacitor filter the proposed simplified model of the DSTATCOM retains the
becomes larger. nonlinearities, even when the switching process is not taken into
The bifurcation diagrams have been successfully corrobo- account.
rated through time-domain simulations using the simplified The effect on the DSTATCOM performance and on its sta-
DSTATCOM model and the detailed DSTATCOM model, bility of the dc storage capacitor as well as the ac filter capacitor
demonstrating that the stability regions have been correctly has been shown. In the case of the dc storage capacitor, it has
computed by using the bifurcation theory. However, there been demonstrated that the capacitor size has a positive effect
are some aspects that must be emphasized to avoid misunder- on the DSTATCOM stability. On the other hand, the ac filter ca-
standing the results. For instance, for the stable region computed pacitor size has an adverse impact on the DSTATCOM stability.
through the bifurcation theory, there is a set of initial conditions It has been demonstrated that bifurcation theory can be suc-
for which the DSTATCOM properly compensates. This set of cessfully applied to assess nonlinear oscillations in distribu-
initial conditions is not known with the bifurcation analysis. tion systems containing DSTATCOM operating in current con-
However, for the unstable regions, there are not any initial trol mode. An assessment of qualitative effects of electrical pa-
conditions for which the DSTATCOM properly compensates. rameters on the stability and on the speed of response of the
For example, from Fig. 15, it can be observed that after cer- DSTATCOM has been achieved. This analysis allows an effec-
tain capacitance of , the stable region remains constant de- tive selection of the DSTATCOM parameter to ensure the rated
spite the size of . However, in this figure, it is not pos- operation condition of the DSTATCOM far away from a pos-
sible to notice that the energy in the storage capacitor is higher sible bifurcation.
as the capacitance of becomes higher. Consequently, the
DSTATCOM will be able to compensate larger and more serious ACKNOWLEDGMENT
disturbances in the network because the set of initial conditions The authors would like to thank the DEP-FIE of UMSNH
for which the DSTATCOM compensates properly increases. and the Faculty of Built Environment and Engineering of the
Queensland University of Technology for their facilities to carry
out this investigation.
VII. CONCLUSION
REFERENCES
A stability analysis of the DSTATCOM in current control [1] T. S. Parker and L. O. Chua, Practical Numerical Algorithms for
mode based on bifurcation theory using the proposed simplified Chaotic Systems. New York: Springer-Verlag, 1989.
DSTATCOM model has been presented. A state-space approach [2] Koppel and R. B. Washburn, “Chaotic motions in the two-degree
freedom swing equations,” IEEE Trans. Circuits Syst., vol. CAS-29,
has been used to represent in the time domain, the dynamics of no. 11, pp. 738–746, Nov. 1982.
the DSTATCOM connected to the system. A control has been [3] I. Dobson and H. D. Chiang, “Toward a theory of voltage collapse in
proposed. electric power systems,” Syst. Control Lett., vol. 13, pp. 253–262, 1989.
[4] M. Varghese, F. F. Wu, and P. Varaiya, “Bifurcations associated with
It has been shown that by adjusting system parameters, the subsynchronous resonance,” IEEE Trans. Power Syst., vol. 13, no. 1,
system exhibits loss of stability, in particular, a Neimark bifur- pp. 139–!44, Feb. 1998.
cation appears. [5] H. O. Wang, E. H. Abed, and A. M. A. Hamdan, “Bifurcation, chaos
and crisis in voltage collapse of a model power system,” IEEE Trans.
Bifurcation diagrams in the plane for different Circuits Syst., vol. 41, no. 3, pp. 294–302, Mar. 1994.
Thevenin voltages were presented. In addition, the bifurcation [6] C. Kieny, “Application of the bifurcation theory in studying and under-
diagrams in the space and in the space standing the global behavior of a ferroresonant electric power circuit,”
IEEE Trans. Power Del., vol. 6, no. 2, pp. 866–872, Apr. 1991.
have been presented. From these bifurcations diagrams in the [7] S. H. Lee, J. K. Park, and B. H. Lee, “A study on the nonlinear con-
gain space, the set of gains for the fastest response has been troller to prevent unstable Hopf bifurcation,” in Proc. IEEE Power Eng.
obtained. Soc. Summer Meeting, Jul. 2001, vol. 2, pp. 978–982.
[8] W. D. Rosehart and C. A. Cañizares, “Bifurcation analysis of various
Time-domain simulations have been only presented to cor- power system models,” Int. J. Elect. Power Energy Syst., vol. 21, no. 3,
roborate the solution obtained from the bifurcation diagrams; pp. 171–182, Mar. 1999.

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.
1678 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 24, NO. 3, JULY 2009

[9] C. A. Cañizares, “On bifurcations, voltage collapse and load mod- [29] M. K. Mishra, A. Ghosh, and A. Joshi, “Load compensation for systems
eling,” IEEE Trans. Power Syst., vol. 10, no. 1, pp. 512–522, Feb. 1995. with non-stiff source using state feedback,” Elect. Power Syst. Res., vol.
[10] M. A. Pai, P. W. Sauer, and B. C. Lesieutre, “Structural stability in 67, pp. 35–44, 2003.
power systems-effect of load models,” IEEE Trans. Power Syst., vol. [30] N. G. Hingorani and L. Gyudyi, Understanding FACTS. Piscataway,
10, no. 1, pp. 609–615, Feb. 1995. NJ: IEEE, 2000.
[11] T. K. Vu and C. C. Liu, “Analysis of tap-changer dynamics and con- [31] E. Acha, A. Semlyen, and Y. N. Rajakovic, “A harmonic domain com-
struction of voltage stability regions,” IEEE Trans. Circuits Syst., vol. putation package for nonlinear problems and its application to electric
36, no. 4, pp. 575–590, Apr. 1989. arcs,” IEEE Trans. Power Del., vol. 5, no. 3, pp. 1390–1397, Jul. 1990.
[12] K. N. Srivastava and S. C. Srivastava, “Elimination of dynamic bifurca- [32] J. Vlach, J. M. Wojciechowski, and A. Opal, “Analysis of nonlinear net-
tion in power systems using FACTS,” IEEE Trans. Circuits Syst., vol. works with inconsistent initial conditions,” IEEE Trans. Circuits Syst.,
45, no. 1, pp. 72–78, Jan. 1998. vol. 42, no. 4, pp. 195–200, Apr. 1995.
[13] N. Mithulananthan, C. Cañizares, J. Reeve, and G. J. Rogers, “Compar- [33] A. H. Nayfeh and B. Balachandran, Applied Nonlinear Dynamics:
ison on PSSS, SVC and STATCOM controllers for damping power sys- Analytical, Computational, and Experimental Methods. New York:
tems oscillations,” IEEE Trans. Power Syst., vol. 18, no. 2, pp. 786–792, Wiley, 1995.
May 2003. [34] A. Semlyen and A. Medina, “Computation of periodic steady-state in
[14] A. Azzouz, R. Duhr, and M. Hasler, “Transition to chaos in a simple system with nonlinear components using a hybrid time and frequency
nonlinear circuit driven by sinusoidal voltage source,” IEEE Trans. domain methodology,” IEEE Trans. Power Syst., vol. 10, no. 3, pp.
Power Syst., vol. CAS-30, no. 12, pp. 913–914, Dec. 1983. 1498–1504, Aug. 1995.
[15] Z. T. Zhusubaliyev and E. Mosekilde, “Torus birth bifurcation in a
DC/DC converter,” IEEE Trans. Circuits Syst. I, vol. 53, no. 8, pp.
1839–1850, Aug. 2006.
[16] S. Jalali, I. Dobson, R. H. Lasseter, and G. Venkataramanan, Juan Segundo-Ramirez received the M.Sc. degree from the CINVESTAV
“Switching time bifurcation in a thyristor controlled reactor,” IEEE Guadalajara, Guadalajara, Mexico, in 2004 and is currently pursuing the Ph.D.
Trans. Circuits Syst. I, vol. 43, no. 3, pp. 209–218, Mar. 1996. degree at the División de Estudios de Posgrado, Facultad de Ingeniería Eléctrica
[17] M. di Bernardo and F. Vasca, “Discrete-time maps for the analysis of of the Universidad Michoacana de San Nicolás de Hidalgo.
bifurcations and chaos in DC/DC converters,” IEEE Trans. Circuits His area of research is the dynamic and steady-state analysis of power sys-
Syst. I, vol. 47, no. 2, pp. 130–143, Feb. 2000. tems.
[18] M. di Bernardo, F. Garofalo, L. Glielmo, and F. Vasca, “Switching, bi-
furcation, and chaos in DC/DC converters,” IEEE Trans. Circuits Syst.
I, vol. 45, no. 2, pp. 133–141, Mar. 1998.
[19] H. H. C. Iu and B. Robert, “Control of chaos in a PWM current-mode Aurelio Medina (SM’02) received the Ph.D. degree from the University of Can-
H-bridge inverter using time-delayed feedback,” IEEE Trans. Circuits terbury, Christchurch, New Zealand, in 1992.
Syst. I, vol. 50, no. 8, pp. 1125–1129, Mar. 2003. He was a Postdoctoral Fellow at the University of Canterbury for one year and
[20] C. K. Tse, Complex Behavior of Switching Power Converters. Boca at the University of Toronto, Toronto, ON, Canada, for two years. Currently, he
Raton, FL: CRC, 2004. is a staff member of the Facultad de Ingeniería Eléctrica, UMSNH, Morelia,
[21] J. H. B. Deane and D. C. Hamill, “Instability, subarmonics, and chaos Mexico, where he is the Head of the Division for Posgraduate Studies. His re-
in power electronic systems,” IEEE Trans. Power Del., vol. 5, no. 1, search interests are the dynamic and steady-state analysis of power systems.
pp. 260–268, Jan. 1990.
[22] S. Banerjee and G. C. Verghese, Nonlinear Phenomena in Power Elec-
tronics, Attractors, Bifurcations, Chaos and Nonlinear Control. Pis-
cataway, NJ: IEEE, 2001.
[23] P. Kundur, Power System Stability and Control. New York: McGraw- Arindam Ghosh (S’80–M’83–SM’93–F’06) received the Ph.D. degree in elec-
Hill, 1994. trical engineering from the University of Calgary, Calgary, AB, Canada, in 1983.
[24] A. Ghosh and G. Ledwich, “Load compensating DSTATCOM in weak Currently, he is the Professor of Power Engineering at Queensland University
AC systems,” IEEE Trans. Power Del., vol. 18, no. 4, pp. 1302–1309, of Technology (QUT), Brisbane, Australia. Prior to joining the QUT in 2006,
Oct. 2003. he was with the Department of Electrical Engineering at the Indian Institute of
[25] A. Ghosh and A. Joshi, “A new method for load balancing and power Technology, Kanpur, India, for 21 years. His interests are the control of power
factor correction using instantaneous symmetrical components,” IEEE systems and power-electronic devices.
Power Eng. Rev., vol. 18, no. 9, pp. 60–62, Sep. 1998. Prof. Ghosh is a Fellow of the Indian National Academy of Engineering
[26] H. Akagi, Y. Kanazawa, and A. Nabae, “Instantaneous reactive power (INAE).
compensators comprising switching devices without energy storage
components,” IEEE Trans. Ind. Appl., vol. IA-20, no. 3, pp. 625–630,
May/Jun. 1984.
[27] H. Akagi, A. Nabae, and S. Atoh, “Control strategy of active power Gerard Ledwich (M’73–SM’92) received the Ph.D. degree in electrical engi-
filters using multiple voltage-source PWM converters,” IEEE Trans. neering from the University of Newcastle, Newcastle, Australia, in 1976.
Ind. Appl., vol. IA-22, no. 3, pp. 460–465, May/Jun. 1986. He has been Chair Professor in Power Engineering at Queensland University
[28] A. Ghosh and A. Joshi, “A new approach to load balancing and power of Technology, Brisbane, Australia, where he has been since 2000. His interests
factor correction in power distribution system,” IEEE Trans. Power are in the areas of power systems, power electronics, and controls.
Del., vol. 15, no. 1, pp. 417–422, Jan. 2000. Dr. Ledwich is a Fellow of I.E.Aust.

Authorized licensed use limited to: PONDICHERRY ENGG COLLEGE. Downloaded on August 9, 2009 at 01:19 from IEEE Xplore. Restrictions apply.