Beruflich Dokumente
Kultur Dokumente
Xia Jiang
Approved by the
Examining Committee:
March 2007
(For Graduation May 2007)
OPERATING MODES AND THEIR REGULATIONS
OF VOLTAGE-SOURCED CONVERTER BASED
FACTS CONTROLLERS
By
Xia Jiang
Examining Committee:
March 2007
(For Graduation May 2007)
c Copyright 2007
by
Xia Jiang
All Rights Reserved
ii
CONTENTS
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vi
ACKNOWLEDGMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
1. INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2. LITERATURE REVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1 Description of Shunt and Series Voltage-Sourced Converters . . . . . 5
2.2 Summary on Voltage-Sourced Converter Operating Modes . . . . . . 6
2.2.1 Shunt VSC Operating Modes . . . . . . . . . . . . . . . . . . 6
2.2.2 Series VSC Operating Modes . . . . . . . . . . . . . . . . . . 6
2.3 Overview of VSC-Based FACTS Controller Loadflow Models . . . . . 8
2.3.1 Decoupled FACTS Controller Loadflow Model . . . . . . . . . 8
2.3.2 Power Injection Model . . . . . . . . . . . . . . . . . . . . . . 10
2.3.3 Voltage Source Model . . . . . . . . . . . . . . . . . . . . . . . 13
2.3.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . 14
2.4 Overview of VSC-Based FACTS Controllers in Time-Domain Simu-
lation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.5 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 17
iii
3.4.2.1 Standalone Operation . . . . . . . . . . . . . . . . . 27
3.4.2.2 Coupled Operating Mode . . . . . . . . . . . . . . . 28
3.5 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.5.1 Voltage Stability Improvement by the SSSC . . . . . . . . . . 31
3.5.2 Operator Training Simulator (OTS) for NYPA’s CSC . . . . . 34
3.5.3 Maximum Dispatchbility for the UPFC and IPFC . . . . . . . 38
3.5.3.1 Maximum UPFC Dispatchability . . . . . . . . . . . 39
3.5.3.2 Maximum IPFC Dispatchability . . . . . . . . . . . . 40
3.6 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 43
4. FACTS CONTROLLER DYNAMIC MODELS AND SETPOINT CON-
TROL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
4.1 VSC Dynamic Modeling and Control . . . . . . . . . . . . . . . . . . 45
4.1.1 VSC Dynamic Model . . . . . . . . . . . . . . . . . . . . . . . 45
4.1.2 VSC Setpoint Controller Models . . . . . . . . . . . . . . . . . 48
4.1.2.1 Shunt VSC Model . . . . . . . . . . . . . . . . . . . 48
4.1.2.2 Standalone Series VSC Model . . . . . . . . . . . . . 49
4.1.2.3 Coupled Series VSC Model . . . . . . . . . . . . . . 52
4.1.2.4 The IPFC Model . . . . . . . . . . . . . . . . . . . . 53
4.1.3 DC Link Capacitor Dynamics . . . . . . . . . . . . . . . . . . 56
4.2 Numerical Computation . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.2.1 Nonlinear Dynamic Models . . . . . . . . . . . . . . . . . . . 58
4.2.1.1 Shunt Operating Modes . . . . . . . . . . . . . . . . 59
4.2.1.2 Standalone Series Dispatch Modes . . . . . . . . . . 60
4.2.1.3 Coupled Series Dispatch Modes . . . . . . . . . . . . 61
4.2.1.4 IPFC Operating Modes . . . . . . . . . . . . . . . . 62
4.2.2 Network Equations . . . . . . . . . . . . . . . . . . . . . . . . 65
4.2.3 Newton’s Method . . . . . . . . . . . . . . . . . . . . . . . . . 68
4.2.4 Integration Method . . . . . . . . . . . . . . . . . . . . . . . . 69
4.3 Simulation Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.3.1 FACTS Controller Dynamic Simulations . . . . . . . . . . . . 71
4.3.1.1 STATCOM Dynamics . . . . . . . . . . . . . . . . . 71
4.3.1.2 SSSC Dynamics . . . . . . . . . . . . . . . . . . . . . 72
4.3.1.3 UPFC Dynamics . . . . . . . . . . . . . . . . . . . . 72
4.3.1.4 IPFC Dynamics . . . . . . . . . . . . . . . . . . . . . 73
4.3.2 Transient Power Transfer Capability Analysis Example . . . . 73
4.4 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 75
iv
5. LINEARIZED MODELS AND MODAL DECOMPOSITION OF MULTI-
MACHINE SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.1 Small-Signal Linearization . . . . . . . . . . . . . . . . . . . . . . . . 84
5.2 System Modal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 85
5.3 Multi-Machine Modal Decomposition Approach . . . . . . . . . . . . 87
5.4 Application: A 20-Bus System Study . . . . . . . . . . . . . . . . . . 90
5.5 Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . 92
APPENDICES
v
LIST OF TABLES
6.1 MDI Indices for the UPFC Series Vd ,Vq Mode v.s. the STATCOM . . . 110
6.2 MDI Index Values for Measured Signals to IPFC Regulators . . . . . . 115
vi
LIST OF FIGURES
3.4 Injected Series Voltage Modification When the Slave VSC Cannot Sup-
port Enough Real Power . . . . . . . . . . . . . . . . . . . . . . . . . . 31
vii
4.3 Setpoint Control Schemes of a Shunt VSC . . . . . . . . . . . . . . . . 49
4.7 Setpoint Control Schemes of an IPFC in the Line Power Regulation Mode 54
4.8 Setpoint Control Schemes of an IPFC in the Fixed Injected Voltage Mode 55
4.17 Comparison of No FACTS and UPFC in V ,Vd ,Vq Mode When PL2 =3235
MW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
viii
6.7 STATCOM Damping Controller Signal . . . . . . . . . . . . . . . . . . 104
6.10 SSSC MDI Index Plots Varying Regulation Control Gains . . . . . . . . 107
6.21 IPFC Line Flows without and with Damping Controllers . . . . . . . . 119
ix
ACKNOWLEDGMENT
I would like to thank my advisor, Prof. Joe H. Chow, for the invaluable guidance
and inspiration that he has provided during the course of this work.
Special thanks are given to Dr. Abdel-Aty Edris (EPRI), Dr. Bruce Fardanesh
(NYPA), Dr. Sheppard J. Salon, Dr. Robert C. Degeneff, Dr. Murat Arcak, and
Dr. Jian Sun for their interacts of this work. I thank Ms. Edvina Uzunovic, Ms.
Jane (Jiyun) Sun, Ms. Liana Hopkins, Mr. Bruce Fardanesh, Mr. Mike Parisi, and
Mr. Mark Graham at NYPA for their great cooperation in the Operator Training
Simulator (OTS) project. I am also grateful to Dr. Graham Rogers of Cherry Tree
Scientific Software for providing the Power System Toolbox (PST).
I would like to thank my colleagues Dr. Xuan Wei, Mr. Xinghao Fang, and
Mr. Luigi Vanfretti for always being there to offer help.
I would also like to thank my husband Hui, parents Zhenghua and Juzhen,
sister Feng, and brother Liang, who have always been supportive in different stages
of my life.
This research is supported in part by Electric Power Research institute (EPRI)
and New York Power Authority (NYPA), and in part by National Science Founda-
tion (NSF).
x
ABSTRACT
xi
to their rated capacity.
Following the steady-state dispatch, dynamic regulator models of FACTS con-
trollers, which take into consideration the dynamics of DC Links, have been devel-
oped and implemented to evaluate their impact on transient stability during system
faults and lightly damped inter-area oscillations. Based on the dynamic models,
linearized models of FACTS controllers in multi-machine systems have been derived
using small-signal perturbations.
In addition to their capability of regulating power flow transfer, FACTS con-
trollers can be utilized to improve small-signal stability by providing damping con-
trol supplemental to their regulation controls. To study damping control effects of
FACTS controllers, a new modal decomposition approach, which fully decouples all
the modes in the system and considers the interaction of the other system modes
to the mode of interest, has been proposed to quantify levels of controllability, ob-
servability, and inner-loop gains of the linearized models. A comprehensive process
that examines the controller gain limitation, the selection of damping controller in-
put signals, and modal damping selectivity signal selection have been developed to
design damping controllers.
xii
CHAPTER 1
INTRODUCTION
1
2
the need for land acquisition and associated environmental concerns. Second, faster
control responses and lower output distortion can be achieved with suitable internal
controls. Third, they can improve dispatch flexibility by circulating active power
between their AC and DC terminals if there is a suitable power source or energy
storage connected to the DC terminals, or there are more than one VSCs coupled
together. Moreover, the combined VSC-based FACTS controllers, such as the UPFC
and CSC, provide complete controllability for controlling not only bus voltages but
also line flows. In this research, we focus our study on VSC-based FACTS controllers.
This research work focuses on steady-state and dynamic modeling, simula-
tion, and control of various operating modes and their regulations of VSC-based
FACTS controllers embedded in transmission networks. The steady-state modeling
and dispatch in part builds on Dr. Xuan Wei’s research work [19]-[20] of studying the
modeling, dispatch, and control of various VSC-based FACTS Controllers. In par-
ticular, she investigated dispatch strategies of these FACTS Controllers to optimize
the voltage profile and power transfer for both normal operation and rated capacity
operation conditions [21]-[23]. In her work, the FACTS controllers are classified by
type, which means for a new configuration or even for a new operating mode, a
complete set of codes for the model needs to be included.
In continuing the steady-state dispatch work in [19], three major improvements
are provided in this thesis work. First, an efficient control mode implementation has
been developed by advocating separate modeling for a shunt VSC and for a series
VSC. If the DC buses of two converters are coupled, then an appropriate active power
circulation constraint can be added to the VSC models. With this implementation,
we only need to select and combine the appropriate shunt VSC, series VSC, and
DC link coupling equations to form the specified FACTS controller and to operate
it in the desired operating mode. Second, in addition to the shunt voltage setpoint
control mode and the line power flow regulation mode, the reactive power setpoint
control mode and the reactive power reserve mode for the shunt VSC and the fixed
injected voltage control mode for the series VSC and their rated-capacity dispatch
have been incorporated into the control mode implementation. Third, line active
power priority rule is applied to re-calculate series injected voltage setpoints when a
3
coupled series VSC is operated at rated capacity, which does not require additional
optimization programs to specify the setpoints.
Besides the steady-state dispatch, this thesis work also extends the control
mode implementation to transient stability analysis for various FACTS controllers.
A comprehensive set of regulator models of FACTS controllers, which take into ac-
count the DC link capacitor dynamics, are proposed to evaluate their impact on
transient stability during system faults and lightly damped inter-area oscillations.
The shunt VSC controls and the series VSC controls are modeled as separate reg-
ulators. When a VSC changes its operating mode, only the input signals of the
corresponding regulator need to be adjusted. The comprehensive set of regulator
models are readily incorporated into a positive-sequence time-domain simulation
program.
Following the development of dynamic models in various control modes, lin-
earized models of FACTS controllers in multi-machine systems are derived based on
the dynamic models using small-signal perturbations. In this approach, dynamic
simulation and small-signal analysis are able to share a common set of codes for
FACTS controllers.
Furthermore, damping control design using VSC-based FACTS controllers for
damping inter-area modes are investigated in this thesis work. A new modal de-
composition approach, which fully decouples all the system modes and consider the
interaction of other state modes to the mode of interest, is proposed to to quantify
levels of controllability, observability, and inner-loop gain of the small-signal lin-
earized models of FACTS controllers in multi-machine systems. A comprehensive
process that will examine the controller gain limitation, the selection of damping
controller input signals, and modal damping selectivity signal selection have been
developed to design damping controllers.
This thesis is organized as follows. This chapter gives a short introduction
of VSC-based FACTS controllers and background of this research work. Chapter 2
introduced shunt VSCs and Series VSCs and their operating modes and reviewed
loadflow models and dynamic models for FACTS controllers. Steady-state dispatch
of FACTS controllers in both normal operation and rated capacity is described in
4
Chapter 3. Chapter 4 includes the details for dynamic simulation of FACTS con-
trollers. Chapter 5 presents linearized models and modal decomposition approach
for small-signal stability analysis of multi-machine systems. Damping control design
is discussed in Chapter 6. The main contributions of this thesis work and future
work recommendations are summarized in Chapter 7.
CHAPTER 2
LITERATURE REVIEW
~ ~ ~ ~ ~ ~ ~
V2 V1 V3 V3 V1 V2 V4
Z1 Z2 Z3 Z4
Ssh
From-bus From-bus Sse To-bus
+ _
Vdc
+ _
Vdc
5
6
The series VSC injects its output voltage into the transmission line via a series
coupling transformer. By regulating the amplitude and phase angle of its output
voltage, it exchanges both reactive and active power with the transmission system.
If the series VSC operates standalone as an SSSC and is not integrated with other
energy storage systems, the output voltage will be in quadrature with the line current
and thus no active power exchanges between the converter and the AC system.
If a VSC is integrated with an energy storage system or coupled with other
VSCs via DC link capacitors, active power will circulate between their AC and DC
terminals.
(Sh1) Control the shunt bus voltage with a reference value Vref and a droop α, that
is, V1 = Vref − αIshq , where Ishq is the reactive current injected by the shunt
VSC. The droop function can be turned off by setting α = 0.
(Sh2) Control the Var output of the shunt VSC to a desired value Ishqref .
(Sh3) Operate in the Var reserve mode which is the operating mode (Sh1) with the
Var output of the shunt VSC limited to [Ishqmin , Ishqmax ]. The operating V -I
characteristic of mode (Sh3) is shown in Figure 2.2(a).
~ Vq
V1 V1
ICshqres ILshqres
Vd ~
Vm
Vref
Iline
(Se1) Control the line active power flow to a desired value Pdes .
(Se2) Fix the injected voltage magnitude, in either a quadrature leading or lagging
direction with respect to the line current.
The standalone series mode also applies to the series VSC operating as the
“slave” in an Interline Power Flow Controller (IPFC) [26].
If the series VSC is coupled to another VSC, then the possible operating modes
include:
(SeC1) Control the line active and reactive power flow to the desired values Pdes and
Qdes .
(SeC2) Fix the magnitude of the d-axis and q-axis injected voltages at Vd and Vq ,
determined with respect to the from-bus voltage vector V1 (Figure 2.2(b)).
The (SeC1) mode is the most commonly cited mode of operation of a UPFC in
the published literature, as influenced by the UPFC operation in Inez [14]. Fixing the
line P and Q flow, however, may prevent the line from carrying a higher amount of
8
flow in case of a contingency. The operating mode (SeC2) would allow for additional
line power flow when appropriate.
The operating modes listed here must also respect the VSC operating limits.
When a shunt VSC is at its operating limit, it may not be able to control the bus
voltage to its desired value, and when a series VSC is at its operating limit, it may
not be able to control the line active power to its desired value. Instead, their
injected voltages will be fixed at their maximum values. For coupled VSCs such as
a UPFC, operation at capacity limits can be more involved.
P2 = −Pto (2.1a)
Q2 = −Qto (2.1b)
9
~ ~ ~ ~
V3 V1 V2 V4
Z3 Z4
PV bus PQ bus
P1, V1 P2, Q2
where Pto and Qto are the line flow of the series branch of the FACTS controller.
Assuming that the VSC operation is lossless, the active power injected from the
generator satisfies
P1 = P2 (2.2)
The reactive power Q1 injected from the generator is the amount required to keep
voltage V1 at its regulated value.
If the shunt VSC is operating in the reactive power setpoint control mode,
we can model the from bus as a P Q load bus with the equivalent load of P1 and
Q1 , instead of a P V bus. However, note that here P1 and Q1 do not equal to the
original shunt injections Psh and Qsh by the shunt VSC, so it is not able to obtain
appropriate setpoint for Q1 directly from the setpoint of Qsh .
A standard loadflow can be carried out with the equivalent P V bus and P Q
bus. After the loadflow has converged, the original control variables need to be
calculated from the set of FACTS controller steady-state nonlinear equations, which
requires an iteration process to solve in order to match the phases of V1 and V2 across
a reactance of the series transformer.
Although the decoupled model is capable of modeling the power flow regulation
modes for the series VSC, it is not applicable to the fixed voltage injection mode,
where the setpoints for P2 and Q2 of the equivalent load are not known or specified
in advance. Moreover, the decoupled model is not applicable to the rated-capacity
operating mode. In this model, the original FACTS control variables are calculated
only after the conventional loadflow converges. As a result, during the loadflow
10
iterations it is impossible to check or enforce the limits of the control variables and
other constraints such as MVA ratings and maximum current magnitudes.
~
Iu
P1 = Psh + rV12 sin(γ)/Xt2
(2.3)
Q1 = Qsh + rV12 cos(γ)/Xt2
P2 = −rV1 V2 sin(θ12 + γ)/Xt2
(2.4)
Q2 = −rV1 V2 cos(θ12 + γ)/Xt2
and the expressions of the active and reactive power supplied by the series VSC as
where r is the ratio of the series VSC inserted voltage magnitude to the from-bus
voltage magnitude, γ is the angle difference between the series VSC inserted voltage
angle and the from-bus voltage angle, θ12 = θ1 − θ2 is the difference between the
from-bus and to-bus voltage angles, and Psh and Qsh are the real and reactive power
injection by the shunt VSC, respectively.
11
Assuming that the DC link voltage is held constant and the VSC model is
lossless, the active power circulation is balanced between the shunt VSC and the
series VSC, that is
Psh + Pse = 0 (2.6)
Substituting (2.5) and (2.6) into (2.3), the power injections P1 and Q1 can be ex-
pressed as
P1 = rV1 V2 sin(θ12 + γ)/Xt2
(2.7)
Q1 = Qsh + rV12 cos(γ)/Xt2
Note that here we denote the power injections as P1 , Q1 , P2 , and Q2 because these
variables are different from the power injections in (2.1) and (2.2) in the decoupled
model.
The power flowing into the UPFC to-bus can be expressed as
Pto = Pu + P2
(2.8)
Qto = Qu2 + Q2
Note that in PIM, the UPFC line current Iu and flows Pu , Qu1 , and Qu2 solved from
the loadflow do not equal to those on the actual equipment.
In [29], the reactive power injection Q1 is set to be either zero or the maximum
value allowed by the shunt converter MVA rating, thus no longer regulating the from-
bus voltage regulation via the shunt VSC. The series VSC control variables Vm2 and
α2 are adjusted manually by trial and error in order to achieve a power flow solution
that matches the targeted requirements, which means searching through the feasible
solution space of the control variables to achieve the UPFC setpoints.
Another approach of implementing the voltage and power setpoint regulation
with the PIM is to use the following iterative algorithm:
1. Solve the loadflow with no power injections, and find the uncompensated power
Pu and Qu2 into the UPFC to-bus, as shown in Figure 2.4.
2. Set P2 = Pdes − Pu , Q2 = Qdes − Qu2 , P1 = −P2 , and Q1 = K(Vref − V1 ), where
K is a proportional gain to regulate V1 to its reference Vref .
12
3. Solve the loadflow using the power injections P1 , Q1 , P2 , and Q2 . Check
the setpoint mismatch criteria. If the criteria are not achieved, update the
uncompensated power Pu and Qu2 and go back to step 2.
The drawback of this method is that it introduces a Gauss update loop outside the
loadflow, which could lead to poor overall convergence of the solution.
~ ~ ~ ~
V3 V1
Z3 Pu, Qu1 jXt2 Pu, Qu2 V2 Z4
V4
~
Iu
The PIM is further developed by Xiao et al in [30, 31], where the decomposed
power injection model (DPIM) is proposed. As shown in Figure 2.5, the DPIM
separates the reactive power injection Q1 in the PIM (2.7) at the from-bus into two
components: Qsh and −∆Qsc , where Qsh is the reactive power injected by the shunt
VSC and −∆Qsc is the difference of Q1 and Qsh . From (2.7), we have
Another variation of the PIM is to transform the power injections at the UPFC
from-bus and to-bus into equivalent shunt admittances [39] (see Figure 2.6), resulting
in a π section representation of the UPFC. The admittance Ysh , Y1 , and Y2 are
functions of the corresponding power injections and bus voltages. A conventional
loadflow is then carried out with the pre-specified equivalent power injections.
In summary, to bypass the direct modeling of the voltage injected by a series
VSC, the PIM adds additional power injections at the FACTS controller from-bus
and to-bus in the conventional loadflow. An external iteration loop is necessary to
enforce the setpoint regulation, which could lead to poor convergence.
13
~ ~ ~ ~
V3 V1 V2 V4
Z3 Pu, Qu1 jXt2 Pu, Qu2 Z4
~
Pu,
Ysh Y1 Y2
For the same reasons of the decoupled model, the PIM is not applicable to
model rated-capacity mode for FACTS controllers.
The details of the VSM loadflow technique and the dispatch strategies for
FACTS controllers operating at various operating modes and rated capacity will be
described in Chapter 3.
time-domain studies.
The first one is the three-phase electromagnetic transient simulation approach,
in which electric power systems including FACTS devices have to be modeled in
detail, using some standard software, such as EMTP [41] and PSCAD/EMTDC
[42]. In these studies, the results should represent the time functions of physical
quantities with the fast transients. The system voltages and currents are represented
as sinusoidal functions, which requires a considerable amount of computational time
because of the comprehensive modeling and the short integration step size. As the
electro-mechanical response of a power system is relatively slower, such an approach
for studying transient stability is too computationally intensive.
The second one is the positive-sequence approach, in which an electric power
system is modeled as a balanced three-phase system. Since sinusoidal quantities are
not dealt with, the integration step size may be larger, and the modeling is simpler,
such that the simulation procedure is much faster than in the electromagnetic tran-
sient approach. With proper modeling, the results for transient stability should be
very close to those achieved in the electromagnetic transient approach. For these
reasons this approach is chosen as a basis for our investigation.
In this section, we will review and compare different models of VSC-based
FACTS controllers suitable for positive-sequence time-domain simulation.
A. The Instantaneous Control Model, without DC Link Capacitor Dynamics Involved
The instantaneous control model [29, 43, 44] determines the controllable vari-
ables of FACTS controllers instantaneously by solving algebraic equations in each
integration step. Those algebraic equations are usually solved by using optimization
techniques to the preliminary static models, which assume that the DC link voltage
maintains constant and the VSC models are lossless. Thus the DC link capacitor
dynamics is not involved in this type of model.
The application of the instantaneous control model is limited to those spec-
ified open-loop control strategies which drive the FACTS controllers to operate at
the rated capacity and thus the controllable variables of FACTS controllers are in-
stantaneously available. A new model, which is capable of modeling the closed-loop
setpoint regulation controls of the FACTS controllers in different operating modes,
16
is required.
B. The Regulator Model, without DC Link Capacitor Dynamics Involved
The regulator model [45]-[48] uses regulators to model the controls with feed-
back to determine the controllable variables of FACTS controllers. In this type
of model, the VSC controllers are usually modeled as voltage sources or current
sources. The independent variables of the voltage sources or current sources are
controlled by the regulators. These regulators can be represented as nonlinear dif-
ferential equations and then incorporated into the conventional positive-sequence
time-domain dynamic simulation program.
In [45]-[48], the DC link capacitor voltage is assumed to maintain a constant,
and thus the active power circulation equals zero. This active power balance equa-
tion is used to determine the dependent variables of the FACTS controllers. How-
ever, during transient stability studies, the DC link capacitor of FACTS controllers
will exchange energy with the system and consequently its voltage varies. Thus for
transient stability studies the active power balance equation would not apply.
C. The Regulator Model, with DC Link Capacitor Dynamics Involved
References [49, 50] include the DC link capacitor dynamics in the regulator
model. The DC link dynamics is expressed as a differential equation associated
with the DC link capacitor voltage. The constant DC link capacitor control for the
UPFC is regulated by controlling the firing angle of the shunt VSC.
[49] and [50] considered only the voltage and power flow regulation modes
for the UPFC. However, there are other control modes that are more appropriate,
such as the fixed reactive power setpoint control mode and the reactive power re-
serve mode for a shunt converter and the fixed injected voltage control for a series
converter.
D. A Comprehensive Set of Regulator Models
A comprehensive set of regulator models of FACTS controllers, which include
the DC link capacitor dynamics and take into account various operating modes, are
proposed in this thesis work. In the modeling, shunt VSC controllers and series VSC
controllers are modeled as controllable voltage sources with equivalent transformer
reactance.
17
18
19
bus power or current. It eliminates the potential hunting of the solution in case of
small VSC series transformer reactance and multiple UPFCs and IPFCs. Another
advantage is that the line current and power are readily computed, allowing direct
enforcement of equipment limits [51]. A further advantage is that if the loadflow
uses injected voltage source models, then all subsequent analysis such as sensitivity
computation, control design, and dynamic simulation, can also make use of the same
modeling framework and directly work with the injected voltage sources.
In this chapter, we will focus on the loadflow formulation for various regulation
modes of FACTS Controllers. The modeling details and setpoint regulation loadflow
equations for shunt and series VSCs are summarized in Sections 3.1 and 3.2. The
Newton-Raphson solution technique is described in Section 3.3. Rated-capacity
dispatch when a VSC reaches a limit is discussed in Section 3.4. Application results
are given in Section 3.5.
Vm1 − V1
Ish = (3.1)
jXt1
In the series configuration (Figure 3.2), Vm2 = Vm2 ejα2 is the complex injection
voltage, and Vi = Vi ejθi , i = 1, ..., 4, are the complex bus voltages. The reactance
Xt2 is the winding reactance on the high-side of the series transformer, which is
20
~ ~ ~
V2 V1 V3
Z1 Z2
~ ~
Ish , Ssh
jXt1
+~
_ Vm1
~ ~ ~ ~ ~
V3 V1 Vm2 _ ~ V2 V4
Z3 jXt2 S2 Z4
+
From-bus
~ ~ To-bus
Ise, Sse
typically very small. From Figure 3.2, the line current Ise is given by
Note that the model of the capacitor on the DC link of the VSC is not included
21
in (3.1) through (3.5). In coupled VSC operations, the DC bus would allow the
coupled VSCs to circulate active power. Additional constraints need to be placed
on (3.1) to (3.5) to represent isolated and coupled VSC operations.
(Sh1) Control the shunt bus voltage to a desired value with the droop α, that is,
where Ishq is the reactive current injected by the shunt VSC. The droop func-
tion can be turned off by setting α = 0.
(Sh2) Control the reactive current of the shunt VSC to a desired value Ishqref ,
(Sh3) Operate in the Var reserve mode which is the operating mode (Sh1) with
the reactive current of the shunt VSC limited to [ICshqres , ILshqres ], that is, if
ICshqres ≤ Ishq ≤ ILshqres , then (3.6) is applicable. Otherwise,
22
ILshqres if Ishq > ILshqres
Ishq = (3.8)
ICshqres if Ishq < ICshqres
Pcirc = 0 (3.9)
θ1 − α1 = 0 (3.10)
If the DC bus of the shunt VSC is integrated with an energy supply system, such
as a battery park, then the circulating active power Pcirc , instead of (3.9), is set to
the active power output PES of the energy supply device
If the shunt VSC is coupled to other VSCs, then the circulating power Pcirc is equal
to the active power collectively absorbed or generated by the other coupled VSCs.
When a shunt VSC is coupled with a series VSC, such as in a UPFC, the circulating
power is
Pcirc = −Psh = Pse (3.12)
(Se2) Fix the injected voltage magnitude, in either quadrature leading or lagging
23
For the standalone operation of a series VSC, we also need to set the series
VSC to operate with zero circulating power
Pcirc = 0 (3.15)
If the DC bus of the series VSC is integrated with an energy supply system, then
the active power injected into the line will become
In cases when the series VSC is operated as the “Slave” VSC in an IPFC,
(3.15) becomes
Pcirc = Pse1 = −Pse2 (3.18)
where Pse1 and Pse2 are the injected active power from VSC 1 and VSC 2, respec-
tively.
(SeC1) Control the line active and reactive power flow Pto and Qto to the desired
values Pdes and Qdes , respectively,
(SeC2) Fix the d-axis and q-axis of the injected voltage at Vd and Vq with respect to
the from-bus voltage vector V1
Vq
Vm2 = Vd2 + Vq2 , φ = tan−1 ( ) (3.21)
Vd
where φ is the angle between the injected voltage vector and the from-bus
voltage vector.
(SeC3) Fix the magnitude of the q-axis injected voltage at Vq , determined with respect
to the from-bus voltage vector V1 . Also, satisfy the real power circulation
balance between two VSCs, as shown in (3.18).
fP (v) = P
(3.22)
fQ (v) = Q
where
v = [V T θT ]T = [V1 V2 · · · VN θ1 θ2 · · · θN ]T (3.23)
f¯P (v̄) = P
f¯Q (v̄) = Q (3.25)
f¯VSC (v̄) = R
where
v̄ = [V T θT VmT αT ]T (3.26)
= [V1 · · · VN θ1 · · · θN Vm1 · · · VmM α1 · · · αM ]T
(3.27)
is a 2(N +M )−Ng −1 vector variable of bus voltage magnitudes and angles, and the
last (third) equation in (3.25) is determined by the VSC operating modes, where R is
a vector of the VSC-based controller setpoints or reference values. In our approach,
the P and Q (the first and second) equations in (3.25) will remain unchanged for
all operating modes. The VSC equations (3.6) to (3.21) constitute the formulation
of the third equation in (3.25).
To apply the Newton-Raphson algorithm to the augmented system (3.25), the
Jacobian matrix J¯ becomes
∂ f¯P /∂V ∂ f¯P /∂θ ∂ f¯P /∂Vm ∂ f¯P /∂α
J¯ =
∂ f¯Q /∂V ∂ f¯Q /∂θ ∂ f¯Q /∂Vm ∂ f¯Q /∂α
(3.28)
∂ f¯VSC /∂V ∂ f¯VSC /∂θ ∂ f¯VSC /∂Vm ∂ f¯VSC /∂α
Note that the first 2 × 2 blocks of J¯ (namely, the (1,1), (1,2), (2,1), and (2,2)
entries) are identical to the Jacobian J in (3.24), except for the additional terms
due to the shunt and series VSC transformer reactance and injection terms. Thus
an attractive feature of this Newton-Raphson algorithm for solving loadflow is that
the formulation of the third equation of (3.25) can be readily built into an existing
conventional Newton-Raphson algorithm. For large data sets, sparse factorization
26
4. If the shunt VSC is coupled with an energy supply system or another VSC,
the active power transfer Psh is bound by |Psh | ≤ Pcirc max .
4. The voltage magnitude limits at adjacent buses: Vmin ≤ |V1 |, |V2 | ≤ Vmax
5. If the series VSC is coupled to an energy supply system or another VSC, the
active power transfer Pse is bound by |Pse | ≤ Pcirc max .
If any of these limits are violated, the voltage or flow setpoints of a VSC can no
longer be enforced. In such cases, the dispatch strategies to achieve a rated-capacity
loadflow solution have been derived.
i. Shunt VSC
• Reactive current control – Reset the reference current values if they are outside
the limits:
• Line active power control – If in controlling the desired line power flow ref-
erence results in Vm > Vmmax , switch to fixed injected voltage control and
enforce Vmref = Vmmax .
• Fixed injected voltage control: If Vmref > Vmmax , enforce Vmref = Vmmax .
• Line active and reactive power (P ,Q) control mode – If enforcing the desired
line P and Q values results in the series VSC insertion voltage magnitude
greater than its limit, adjustments need to be applied on the control strategy
– the line active power will be either enforced to its desired value or maximized
while the reactive power will deviate from the desired value. To accommodate
the modification, the control mode is switched to the Vd ,Vq setpoint control
mode. Because Vq affects strongly the line active power and the Vd affects
the line reactive power, in the real power priority strategy Vq should be either
kept constant or maximized and Vd is modified to satisfy the voltage magnitude
constraints, as described below:
– If series Vd2 + Vq2 > Vmmax and Vq ≥ Vmmax , enforce Vqref = Vmmax and
Vdref = 0 (Figure 3.3(a)).
– If series Vd2 + Vq2 > Vmmax and Vq < Vmmax , enforce Vqref = Vq and
Vdref = 2
Vmmax − Vq2 (Figure 3.3(b)).
• Vd ,Vq setpoint control – If 2
Vdref 2
+ Vqref > Vmmax , then scale them back pro-
29
D D
Vm Vd
Vm Vd
Vmmax Vdref
Q Q
Vq Vqref Vqref
(a) (b)
D
Vmref
Vdref
Vmmax Vdnew
ref
Q
Vqref Vqnew
ref
(c)
• The slave VSC can provide the required active power circulation – In this case,
we can modify the setpoint of the slave VSC to the limit and satisfy the active
power balance between the two VSCs at the same time.
Sshmax
Ishqref = Ishq , Psh = −Pcirc (3.31)
2
Psh + Q2sh
• The slave VSC cannot provide the required active power circulation – This
case often occurs when the master VSC is required to significantly impact the
line flow, which requires more active power circulation that can be provided by
the slave VSC. In such cases, the master VSC has to scale back its setpoints to
achieve active power balance in the coupled link, which means that it cannot
both regulate the line active power and reactive power [55]. Based on the real
power priority rule, we release the reactive power control for the master line,
so that
Vqref = Vq , Pse = Pcirc (3.32)
D D
Solution
without Limit
Line
Line
Current Q
Q Current
Solution Solution
with Limit without Limit Solution
with Limit
Figure 3.4: Injected Series Voltage Modification When the Slave VSC
Cannot Support Enough Real Power
3.5 Applications
The control mode implementation for the steady-state dispatch of FACTS
controllers is applied to a 4-bus test system and a 1673-bus test system used in the
Operator Training Simulator (OTS) [40]. Rated-capacity dispatch is enforced when
a VSC reaches its limit.
2 3
1
B 4
generator
L3
Sse
S1
Energy
Storage
S2
Table 3.1: Transmission Line Data of the 4-Bus Radial Test System
Note that by closing the Switch B the SSSC is bypassed, which is referred to
as the uncompensated system (base case). The SSSC is in service if Switch B is
open. By also closing the Switches S1 and S2, the SSSC is integrated with an energy
storage system.
Because the VSC is a reactive power source, the objective is to show the impact
of the VSC on the system voltage stability as the power transfer on the transmission
lines is increased. In a base case, Switch B is closed so that the VSC is not deployed.
Figure 3.6 shows the variation of the voltage V3 at Bus 3 when the load L3 on Bus
3 is increased from the base value of 400 MW. In particular, V3 drops to 0.95 pu
when L3 reaches about 565 MW. Next, Switch B is opened with Switches S1 and
S2 open and the SSSC is inserted and is set to carry 62% of the load L3. As shown
in Figure 3.6, V3 drops to 0.95 pu when L3 reaches about 615 MW, showing a 50
MW increase in the power transfer. Figure 3.7 shows the magnitude of the injected
33
1.15
No FACTS
SSSC
SSSC ES Pc = −10 MW
1.1 SSSC ES Pc = 0 MW
SSSC ES Pc = 10 MW
1.05
1
V3 (pu)
0.95
0.9
0.85
0.8
0.75
400 450 500 550 600 650 700 750
P (MW)
load
voltage as a function of the power dispatch. Note that the injected voltage Vm
reaches the maximum value of 0.1 pu (on the system base) at 510 MW and remains
at the maximum value for higher values of power transfer.
Then Swtiches S1 and S2 are closed to form a SSSC integrated with the energy
storage system. The load L3 on Bus 3 is again increased from the base load of 400
MW, with 62% carried on the series VSC. The dispatch results are shown in Figures
3.6. Without any power circulation, that is, the active power flowing out of the
energy storage system into the series VSC is zero, the dispatching is exactly the
same with the SSSC without the energy storage system. By circulating 10 MW
from the energy storage system to the series VSC, the power transfer is improved
by another 30 MW to 645 MW at V3 = 0.95 pu. On the other hand, if active
power circulates from the series VSC to recharge the energy storage system, the
power transfer is decreased. The magnitude of the injected series voltage source
is shown in Figure 3.7, depicting the saturation of the inserted voltage magnitude.
34
0.105
0.1
0.095
V (pu)
0.09
m
0.085
SSSC
0.08 SSSC ES Pc = −10 MW
SSSC ES Pc = 0 MW
SSSC ES Pc = 10 MW
0.075
400 450 500 550 600 650 700 750
Pload (MW)
There is one corner point at each curve. For example, the corner point of the curve
with squares (SSSC ES Pc = 10 MW) is at a load of 490 MW, which means that
the SSSC switches from line active power regulation mode to fixed series injected
voltage source magnitude mode at that point.
Marcy 345 kV
HSB
New
Scotland
TR-SE2 (UNS)
HSB 100 MVA
Coopers
Corners
TR-SE1 (UCC)
TR-SH 100 MVA LVB
200MVA
LVB
TBS1 TBS2
DC-SW
basis of a CSC Operator Training Simulator. The Training Simulator will allow
an operator to adjust the VSC-based controller using the manufacturer’s control
screens (see Figure 3.9 for a screen shot of Configuration 9) and see the impact of
the controller reflected on the station one-line diagrams. Such a training simulator,
which can be used to dispatch the CSC in different configurations and in different
modes, will provide NYPA system operators with an off-line tool to gain experience
in operating the CSC, which cannot be adjusted for training when it is in operation.
We now use the UPFC in Figure 3.9 to illustrate the versatility of this tool
to study power dispatch with the computation results shown in Table 3.3. The
simulator uses a 1673-bus power system, with VSC 1 in the shunt connection and
VSC 2 in the series connection on the UNS line. To start, the system operation is
solved without the UPFC. The Marcy bus voltage and the flows on the Marcy to
New Scotland (UNS) line are shown in the first row of Table 3.3. Then the UPFC
is inserted with the shunt voltage setpoint at the pre-insertion Marcy voltage and
with the series VSC voltage set to zero. The second row of Table 3.3 shows that the
UNS line power flow drops slightly because of the effect of the series transformer
37
reactance. Next the line flow on the UNS line is dispatched to P = 638 MW and
Q = −1 MVar, with the shunt bus voltage set to Vref = 1.0327 pu. The dispatch
result is shown in the third row of Table 3.3. From the P ,Q dispatch, we use the
series insertion quadrature voltages Vd and Vq as the setpoint to switch to the fixed
voltage insertion mode for the series VSC, as shown in the fourth row of Table 3.3.
In doing so, we expect the fixed inserted voltage mode dispatch to be identical to
the P ,Q mode dispatch. Then the ENS line, which is parallel to the UNS line and
carrying about 500 MW, is tripped. With the fixed inserted voltage dispatch, the
UNS line is able to carry an additional 180 MW. On the other hand, if the series
VSC is in the P ,Q mode, the rest of the network, in particular, the lower kV system,
needs to transport this 180 MW, which can cause low voltages in some portions of
the network.
38
D D Cases
1.0 Case 1.0 1,2,4
4 3 5 6 7
3 5 4
2 3 8
6 0.5
1.0 1.0 2 9
0.5
Q 1 7 Q 1 10
18 11
12 8 17 12
11 9 16 13
10 15 14
Qsh=0.5 Capacitive
150 Qsh=0.0
Qsh=0.5 Inductive
100 4
3
5
2
50
6
1
0
7
12
-50
8
11
9 10
-100
-150
-150 -100 -50 0 50 100 150
Line ∆P (MW)
The resulting dispatch of the master and slave VSCs is shown in Figure 3.12.
Both the master and slave VSC ∆P -∆Q curves are shaped like ellipses, with the
slave ∆P -∆Q ellipses being more narrow and the points corresponding to those in
Figure 3.10(b). Note also that Case 2, which has a higher Vqref , results in about 60
MW and 20 Mvar more power flow on the slave SVC line and about 10 MW less
on the master line than Case 1. The master SVC line reactive power flows for both
cases are very close.
5
50 50
18
9 1 18
0 0 1 9
18
1 9
-50 -50
14
5
5 Case 1
Case 2
-100 -100
-100 -50 0 50 100 -50 0 50 100 150
Master Line ∆P (MW) Slave Line ∆P (MW)
Figure 3.13 shows the d-axis and q-axis components of the injected voltage of
the slave VSC. When the slave VSC reference value is high as in Case 2, with the
master VSC simultaneously requiring large active power circulation (|Vd | > 0.6 pu),
the slave VSC voltage insertion will exceed its limit. Based on the power circulation
priority rule in Section 3.4.2.2.ii, the slave VSC Vq cannot keep its reference value
any more, and it will be reduced to ensure the slave VSC voltage satisfies its limit.
B. VSC 2 as the Master and VSC 1 as the Slave
In the second set of dispatch, VSC 2 is set as the master such that its magnitude
of the injected voltage reference is kept constant while its angle varies, as shown in
Figure 3.10(b). The dispatch is computed for two settings:
42
1 Case1
Case2
14 14
0.5
0 18 18
1 1
9 9
-0.5
5 5
-1
-1 -0.5 0 0.5 1
Slave Vq (pu)
100 100
Case 3 Case 3 A1
Case 4 Case 4
5
A2
50 50 5
11
9 1
0 9 1
0 1 10
18
18
2
-50 -50
14
14
-100 -100
-100 -50 0 50 -150 -100 -50 0 50 100 150
Slave Line ∆P (MW) Master Line ∆P (MW)
The resulting incremental P -Q curves of the master and slave VSCs are shown
in Figure 3.14. Case 4, which has higher injected voltage magnitude of the master
43
VSC, has a larger line flow dispatch region than Case 3. However, compared to Case
3, in Case 4 the top and bottom of the near-elliptical ∆P -∆Q curves are clipped
because of the limits of the slave VSC. Note that two additional reference points
A1 and A2 in Case 4 are added to the set of 18 values for a clearer illustration of
this limitation. Figure 3.15 shows the d-axis and q-axis components of the injected
voltages of the IPFC. When the master VSC reference Vdref is too high (|Vdref | > 0.6
pu), that is, the master VSC requires larger active power circulation, even though
the slave Vd is set to its limit, it is still unable to support the power circulation.
Based on the power circulation priority rule, the master VSC Vd will be reduced to
allow the slave VSC to provide enough active power circulation.
0 1(10) 0 1 1 10 10
1(10)
15 14
-0.5 -0.5
15 14
5(6)
65
-1 -1
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
Slave Vq (pu) Master Vq (pu)
45
46
Ssh
Sse
Idc1 Idc2
+ +
Vdc Vdc
C C
_ _
In the time scale of transient stability, in which the VSC switching dynamics
are neglected, the model of a VSC with modulation ratio γ and firing angle α can
be represented as a voltage source
where k is a factor which relates the inverter DC-side voltage to its AC-side terminal
voltage. Note that k is dependent on the modulation ratio γ.
The dynamic balanced positive-sequence model of a shunt VSC is shown in
Figure 4.2 (a). The shunt VSC is modeled as a controllable injected voltage source
Vm1 behind an equivalent transformer reactance Xt1 , where Vm1 can be expressed as
where k1 is the factor between the DC-side voltage Vdc and the AC-side voltage
magnitude Vm1 of the shunt VSC and θ1 is the angle of the from-bus voltage V1 .
The injected current Ish and the injected power Ssh from the shunt VSC into the
system are the same as given in (3.1) and (3.2) for the steady-state shunt VSC
model, respectively.
As shown in Figure 4.2 (b), the series VSC is modeled as a controllable injected
47
From-bus
From-bus To-bus
~ ~ ~ ~ ~ ~ ~
V2 V1 V3 V3 V1 ~ ~ V2 V4
Z1 Z2 Z3 jXt2 Vm2
+ _ S2 Z4
~ ~
Ish , Ssh ~ ~
Ise, Sse
jXt1
+ ~
_ Vm1
voltage source Vm2 behind an equivalent transformer reactance Xt2 , where Vm2 can
be expressed as
Vm2 = k2 Vdc j(αse +θ1 ) = Vm2 j(αse +θ1 ) (4.3)
where k2 is the factor between the DC-side voltage Vdc and the AC-side voltage
magnitude Vm2 of the series VSC and θ1 is the angle of the from-bus voltage V1 . The
line current Ise , the power injected by the series VSC Sse , and the power injected into
the to-bus (Bus 2) S2 are the same as given in (3.3), (3.4), and (3.5) for the steady-
state series VSC model, respectively. The series injected voltage Vm2 can be split
into two components: Vd is a component in phase with the from-bus voltage which
mainly affects the reactive power of the compensated line and Vq is a component in
quadrature with the from-bus voltage which mainly affect the active power of the
compensated line.
During transient studies, the DC link capacitor of FACTS controllers will ex-
change energy with the system and consequently its voltage will vary. The variation
of the DC capacitor voltage is dependent on its current inflow, which can be modeled
as
dVdc
C = Idc (4.4)
dt
where Idc is the current flowing into the DC capacitor C from the VSC. In steady-
state operations when the power transfer is balanced, Idc = 0, and hence, dVdc /dt is
zero.
48
The FACTS dynamic models will be interfaced with the other dynamic com-
ponents in a power system, such as synchronous machines and excitation systems,
through the algebraic network equations. In using injected voltage sources for the
VSCs in the loadflow formulation, this transition to dynamic modes will be seamless,
because Vm1 and Vm2 are operational states in the loadflow models.
Vdc k1 V m1
V1 I shq
_ I*shq _ -α sh α1
Vref + Kv + Ki 1
Kp+ _
_ s s 1+Ts
+
θ1
α
Droop
(a) Voltage Regulation Mode
Vdc k1 V m1
I shq
_
I shqref Ki -α sh α1
+ Kp+ s
1
_
1+Ts
+
θ1
(b) Var Control Mode
filter 1/(1 + T s). The output of the inner current loop is the minus shunt inverter
voltage angle −αsh . The inverter voltage angle α1 can then be obtained with the
information of the from-bus voltage angle θ1 .
In the var control mode, the shunt reactive current setpoint Ishqref is directly
specified in the operator screens. The shunt reactive current Ishq is controlled to
∗
Ishq by an PI controller Kp + Ki /s and an LP filter 1/(1 + T s). This current loop
produces the angle information −αsh .
In steady state the angle αsh is zero, which means that the inverter output
voltage is kept essentially in phase with the from-bus voltage. Small transient posi-
tive or negative deviations in αsh cause nonzero active power to go through the DC
capacitor and thus result in an increase or decrease of the DC bus voltage Vdc .
modes are shown in Figures 4.4 and 4.5, respectively. The magnitude Vm2 and angle
α2 of the inverter voltage are generated by the control systems.
Vdc k2 V m2
θl
Pref ∆α
_ se
+ Ki 1 + α2
Kp+ s -1
_ 1+Ts + +
π
P 2
(a) Pref ≥ P0
Vdc k2 V m2
θl
Pref ∆α
_ se
+ Ki 1 + α2
Kp+ s 1
_ 1+Ts + +
P π
2
(b) Pref ≤ P0
The standalone series VSC is also operated with a constant k2 between its AC-
side voltage Vm2 and DC-side voltage Vdc , and hence the changes in the magnitude
of the inverter output voltage are achieved by charging or discharging the DC bus
capacitor to a different voltage.
An PI controller Kp + Ki /s and an LP filter 1/(1 + T s) are applied for the
angle control of the standalone series VSC. The input signal for the line active power
control mode is the difference of the line active power setpoint Pref and its measured
value P , while the input signal for the inverter voltage magnitude control mode is
the difference of the inverter voltage magnitude setpoint |Vm2ref | and k2 Vdc . In each
operating mode, the output signal from the LP filter is the angle deviation ∆αse .
In steady state, ∆αse is zero, which means that the inverter output voltage is kept
essentially in quadrature with the current of the compensated line. Small transient
positive or negative deviations in the phase of the inverter voltage cause nonzero
51
Vdc k2 V m2
θl
|Vm2ref| ∆α_ se
+ K 1 + α2
Kp+ i -1
_ s 1+Ts + +
π
k 2Vdc 2
(a) Vm2ref ≤ 0
Vdc k2 V m2
θl
|Vm2ref| ∆α
+ K 1 _ se + α2
Kp+ i 1
_ s 1+Ts + +
π
k 2Vdc 2
(b) Vm2ref ≥ 0
active power to go through the DC capacitor and thus result an increase or decrease
of the DC bus voltage.
In the line active power control mode, when the line active power setpoint Pref
is larger than the original line active power without compensation P0 , the inverter
voltage angle α2 is (Figure 4.4 (a))
π
α2 = θ − + ∆αse (4.5)
2
such that in steady state the inverter voltage is 90 degree lagging the line current
vector. When Pref ≤ P0 , the inverter voltage angle α2 is (Figure 4.4 (b))
π
α2 = θ + − ∆αse (4.6)
2
such that in steady state the inverter voltage is 90 degree leading the line current
vector.
In the inverter voltage magnitude control mode, a polarity is added to the
52
inverter voltage reference Vm2ref to indicate whether leading or lagging voltage in-
jection is required. When Vm2ref ≤ 0, the inverter voltage angle α2 is (Figure 4.5
(a))
π
α2 = θ − + ∆αse (4.7)
2
Thus in steady state the inverter voltage is 90 degree lagging the line current vector,
which means that it will increase the line active power. When Vm2ref ≥ 0, the inverter
voltage angle α2 is (Figure 4.5 (b))
π
α2 = θ + − ∆αse (4.8)
2
Thus in steady state the inverter voltage is 90 degree leading the line current vector,
which means that it will decrease the line active power.
P
_ Vq
Pref Ki V m2
+ Kp+
1
s 1+Ts Magnitude
and Angle α se α2
Qref Ki Calculator
+ Kp+
1
+ +
_ s 1+Ts Vd θ1
Q
(a) Line Power Regulation Mode
V m2
Vdref
Magnitude
and Angle α se α2
Calculator
Vqref + +
θ1
(b) Fixed Injected Voltage Mode
in Figure 4.7 and 4.8, respectively. One VSC of an IPFC is operated as the Master
VSC, and the other is operated as the Slave VSC. The magnitude Vm2 m and angle
α2 m of the Master inverter voltage and the magnitude Vm2 s and angle α2 s of the
Slave inverter voltage are generated by the control systems.
Q
_
Qref Ki V*d
+ Kp+
1
s 1+Ts
Vdc
_
Vdcref Ki ∆Vd + Vd
+ Kp+
1 + V m2_m
_ s 1+Ts Magnitude
and Angle α α2_m
Kα se
Calculator
P Droop + +
_ Vq
Pref + Ki θ1
1 V*q
Kp+
s 1+Ts
(a) The Master VSC
Vdc
Vdcref _ ∆Vd Vd
+ Ki 1 Vm2_ s
Kp+
s 1+Ts Magnitude
and Angle α
Calculator se α2_s
P + +
_ Vq
Pref + Ki 1 V*q
Kp+
s 1+Ts
(b) The Slave VSC
Figure 4.7: Setpoint Control Schemes of an IPFC in the Line Power Reg-
ulation Mode
In these two operating modes, a DC bus voltage regulation loop, which consists
of an PI controller Kp + Ki /s and an LP filter 1/(1 + T s), is applied for each VSC
of the IPFC. The control difference between the DC bus voltage regulation loops of
the Master and Slave VSCs is that there is a nonzero feedback droop for the Slave
VSC, while there is no such a feedback loop for the Master VSC. Thus the DC
bus voltage is more strictly controlled by the Slave VSC. The output signal of each
DC bus voltage regulator, denoted as ∆Vd , are the error signal to form the d-axis
inverter voltage Vd of the corresponding VSC.
55
Vdc Vdref
_
Vdcref
+ Ki
Kp+ s
1 ∆Vd
+ + Vd
V m2_m
_ 1+Ts Magnitude
Kα and Angle α α
Calculator se 2_m
Droop Vqref +
+
θ1
(a) The Master VSC
Vdref
Vdc
_
Ki ∆Vd + Vd
Kp+ s
1 + Vm2_s
Vdcref + 1+Ts Magnitude
and Angle α α
Calculator se 2_s
Vqref
Vq
+ +
θ1
(b) The Slave VSC
In the line power control mode, the Master line active and reactive power
P and Q regulations are implemented by independently controlling the q-axis and
d-axis voltages of the Master VSC Vq∗ and Vd∗ by using the PI controllers and LP
filters as shown in 4.7 (a). The Master inverter voltage can then be obtained as
Vd = Vd∗ + ∆Vd
Vq = Vq∗
(4.11)
Vm2 m = Vd2 + Vq2
αm2 m = θ1 + tan−1 ( VVdq )
For the Slave VSC, only its line active power P is controlled by the PI controller
and LP filter as shown in 4.7 (b). The regulator output is the q-axis voltage of
the Slave VSC vq∗ . The d-axis component vd∗ is not specified in order to meet the
power circulation constraint of the IPFC. Moreover, Vq of the Slave inverter voltage
is limited as a function of Vd to prioritize the transfer of power circulation. The
56
Vd = ∆Vd
V∗
q if Vm2 s ≤ Vm max
Vq =
sign(V ∗ ) · V 2
mmax − Vd ≥ Vm max
2 (4.12)
q if Vm2 s
Vm2 s = Vd2 + Vq2
αm2 s = θ1 + tan−1 ( VVdq )
Vd = Vdref + ∆Vd
Vq = Vqref
(4.13)
Vm2 m = Vd2 + Vq2
αm2 m = θ1 + tan−1 ( VVdq )
Vd = Vdref + ∆Vd
Vqref if Vm2 s ≤ Vm max
Vq =
sign(Vqref ) · V 2
mmax − Vd if Vm2 s ≥ Vm max
2 (4.14)
Vm2 s = Vd2 + Vq2
αm2 s = θ1 + tan−1 ( VVdq )
Thus the AC instantaneous active power flowing into a single shunt VSC, such as a
STATCOM, is
Pac = −Psh (4.17)
If the DC bus of a FACTS controller is coupled with M shunt VSCs and N series
VSCs, the AC instantaneous active powers flowing into the VSCs from the system
is given as
M
N
Pac = −( Pshi + Psei ) (4.19)
i=1 i=1
Assuming that the VSC model is ideal, the total AC instantaneous active
powers on the AC-side is equal to the DC-side active power, that is
dVdc 1
= Pac (4.21)
dt CVdc
1 1
Pac Vdc
CVdc s
ẋ = f (x, V ) (4.22)
I(x, V ) = Y V (4.23)
where I and V are complex injection currents and voltage vectors of dimension n,
respectively, and x is a state variable vector of dimension m. The number n is equal
to the number of nodes in the system and the number m depends on the number
and the type of the dynamic models used for the actual equipment. For example,
for a generator modeled with subtransient reactance, the state variables are its rotor
angle δ, speed ω, and direct- and quadrature-axis fluxes Eq , ψd , Ed , and ψq [56].
In the explicit integration approach, (4.22) is used to update the state variables
x and then the algebraic variables V in (4.23) can be solved iteratively by a Newton
method given by (4.62), at every integration step.
Kpj
.
zj zj
ej + + 1 1
. _ Tj s
xj
1
xj +
Kij s
x˙j = Kij ej
(4.24)
z˙j = (Kpj ej + xj − zj )/Tj
where xj and zj are the state variables for the jth regulators.
∗
We introduce three additional state variables Ishq , x1 , and z1 for a shunt VSC,
two state variable x2 and z2 for a standalone VSC, four state variables x3 , z3 , x4 ,
and z4 for a coupled series VSC, and ten state variables x5 , z5 , x6 , z6 , x7 , z7 , xM ,
zM , xS , and zS for an IPFC into the state variable vector x in (4.22). Also the DC
link dynamic state variable xdc = Vdc will be incorporated.
In this section we provide the equations for formulating the different shunt
and series VSC operating modes.
∗ ∗
I˙shq = Kv (Vref − V1 − αIshq )
∗
ẋ1 = Ki1 (Ishq − Ishq ) (4.25)
∗
ż1 = [Kp1 (Ishq − Ishq ) + x1 − z1 ]/T1
where Kv is the gain of the voltage regulator, Kp1 and Ki1 are the proportional
and integral gain coefficients of the PI controller, and T1 is the time constant
of the LP filter (Figure 4.3 (a)). The shunt injected voltage source can be
obtained as
Vm1 = k1 Vdc
(4.26)
α1 = θ1 − z1
where k1 is the constant ratio between Vm1 and Vdc and θ1 is the from-bus
voltage angle. Note that α1 is with regard to to the system swing bus angle.
(Sh2) Control the Var output of the shunt VSC to a desired value Ishqref : the dif-
∗
ferential equations of the state variables Ishq , x1 , and z1 can be expressed
60
as
∗
I˙shq =0
ẋ1 = Ki1 (Ishqref − Ishq ) (4.27)
ż1 = [Kp1 (Ishqref − Ishq ) + x1 − z1 ]/T1
where Kp1 and Ki1 are the proportional and integral gain coefficients of the PI
controllers and T1 is the time constant of the LP filter (Figure 4.3 (b)). The
shunt injected voltage source can be obtained as
Vm1 = k1 Vdc
(4.28)
α1 = θ1 − z1
where k1 is the constant ratio between Vm1 and Vdc and θ1 is the from-bus
voltage angle. Note that α1 is with regard to to the system swing bus angle.
where Kp2 and Ki2 are the proportional and integral gain coefficients of the
PI controller and T2 is the time constant of the LP filter (Figure 4.4). The
series injected voltage source can be obtained as
Vm2 = k2 Vdc
θl − π/2 + z2 if Pref ≥ P0 (4.30)
α2 =
θl + π/2 − z2 if Pref ≤ P0
where k2 is the constant ratio between Vm2 and Vdc and θ is the line current
angle.
(Se2) Fix the injected voltage magnitude, in either the quadrature leading or lag-
ging direction with respect to the transmission line current: the differential
61
where Kp2 and Ki2 are the proportional and integral gain coefficients of the
PI controller and T2 is the time constant of the LP filter (Figure 4.5). The
series injected voltage source can be obtained as
Vm2 = k2 Vdc
θl − π/2 + z2 if Vm2ref ≤ 0 (4.32)
α2 =
θl + π/2 − z2 if Vm2ref ≥ 0
where k2 is the constant ratio between Vm2 and Vdc and θ is the line current
angle.
where Kp3 , Ki3 , Kp4 , and Ki4 are the proportional and integral gain coefficients
of the PI controllers and T3 and T4 are the time constants of the LP filters
(Figure 4.6 (a)). The series injected voltage source can be obtained as
Vq = z3
Vd = z4
(4.34)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
62
(SeC2) Fix the d-axis and q-axis of the injected voltage at Vdref and Vqref with respect
to the from-bus voltage vector V1 : the differential equations of state variables
x3 , z3 , x4 , and z4 can be expressed as
ẋ3 = 0
ż3 = 0
(4.35)
ẋ4 = 0
ż4 = 0
Note that (4.35) is listed here only for completeness. The series injected voltage
source can be obtained as
Vq = Vqref
Vd = Vdref
(4.36)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
(SeM1) Control the Master line active and reactive power flow P and Q to their desired
values Pref and Qref , respectively: the differential equations of state variables
x5 , z5 , x6 , z6 , xM , and zM can be expressed as
where Kp5 , Ki5 , Kp6 , Ki6 , KpM , and KiM are the proportional and integral gain
coefficients of the PI controllers and T5 , T6 , and TM are the time constants
of the LP filters (Figure 4.7 (a)). The Master injected voltage source can be
63
obtained as
∆Vd = zM
Vq = z5
Vd = z6 + ∆Vd (4.38)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
(SeM2) Fix the d-axis and q-axis of the injected voltage at Vdref and Vqref with respect
to the from-bus voltage vector V1 : the differential equations of state variables
x5 , z5 , x6 , z6 , xM , and zM can be expressed as
ẋ5 = 0
ż5 = 0
ẋ6 = 0
(4.39)
ż6 = 0
ẋM = KiM (Vdcref − Vdc − Kα zM )
żM = [KpM (Vdcref − Vdc − Kα zM ) + xM − zM ]/TM
where KpM and KiM are the proportional and integral gain coefficients of the
PI controllers and TM is the time constant of the LP filter (Figure 4.8 (a)).
The Master injected voltage source can be obtained as
∆Vd = zM
Vq = Vqref
Vd = Vdref + ∆Vd (4.40)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
(SeS1) Control the Slave line active power flow P to its desired value Pref : the differ-
64
where Kp7 , Ki7 , KpS , and KiS are the proportional and integral gain coefficients
of the PI controllers and T7 and TS are the time constants of the LP filters
(Figure 4.7 (b)). The Slave injected voltage source can be obtained as
∆Vd = zS
Vq = z7
Vd = ∆Vd (4.42)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
(SeS2) Fix the q-axis of the injected voltage at and Vqref with respect to the from-bus
voltage vector V1 : the differential equations of state variables x7 , z7 , xS , and
zS can be expressed as
ẋ7 = 0
ż7 = 0
(4.43)
ẋS = KiS (Vdcref − Vdc )
żS = [KpS (Vdcref − Vdc + xS − zS ]/TS
where KpS and KiS are the proportional and integral gain coefficients of the
PI controllers and TS is the time constant of the LP filter (Figure 4.8 (b)).
65
∆Vd = zS
Vq = Vqref
Vd = Vdref + ∆Vd (4.44)
Vm2 = Vd2 + Vq2
α2 = θ1 + tan−1 (Vq /Vd )
where Vg is the generator bus voltage vector and Vl is the bus voltage vector for all
the load buses.
If a shunt VSC is connected to Bus f of the power system, the bus admittance
equation is expanded to
Vg
Ygg Ygf Ygl 0 g I
Y Vf
fg Yff + 1
Yfl − jX1t1
=
0
(4.46)
jXt1
Vl
Ylg Ylf Yll 0 0
Vm1
If a series VSC is inserted into the line with from-bus f and to-bus t of the
66
If a VSC has the same from-bus or to-bus with some other shunt or series
VSCs, the effect of all these VSCs on the bus admittance matrix equation can be
added together.
Suppose the total number of distinct from-buses of FACTS controllers is L,
and the total number of distinct to-buses of FACTS Controllers is R in a specific
power system and let the from-bus fi have Nf i shunt VSCs and Mf i series VSCs
connected to it, for i = 1, . . . , L, and the to-bus tk have Mtk series VSCs connected
to it, for k = 1, . . . , R. The bus admittance matrix equation can be expressed as
Y
gg
YgF YgF Ygl V
g
I
g
YFg YFF YFT YFl VF IF
= (4.50)
YTg YTF YTT YTl
VT
IT
Ylg YlF YlT Yll Vl 0
67
where
Y f1 f1
...
0
YFF =
Y fi fi
(4.51)
...
0
Y fL fL
L×L
Yt1 t1
...
0
YTT =
Ytk tk
(4.52)
...
0
YtR tR
R×R
Yf1 t1 · · · Yf1 tk · · · Yf1 tR
.. ... .. ..
. . .
T
YFT = YFT =
Yfi t1 · · · Yfi tk · · · Yfi tR
(4.53)
.. .. ... ..
. . .
YfL t1 · · · YfL tk · · · YfL tR
L×R
Nf i M
1
fi
1
Yfi fi = Y fi fi + + , i = 1, . . . , L (4.54)
i=1 jXt1i i=1 jXt2i
M tk
1
Ytk tk = Ytk tk + , k = 1, . . . , R (4.55)
k=1 jXt2k
Yfi tk , no series VSCs in Line fi tk
Yfi tk =
M
1s (4.56)
Yfi tk −
, Ms series VSCs in Line fi tk
s=1 jXt2s
i = 1, . . . , L; k = 1, . . . , R
Nf i
Vm1i
Vm2i
Mf i
Ifi = +
i=1 jXt1i i=1 jXt2i
Nf i Mf i
Vm1i ej(α1i +θ1i )
Vm2i ej(α2i +θ1i )
= +
i=1 jXt1i i=1 jXt2i
68
i = 1, . . . , L (4.57)
Ntk
Ntk
Vm2i Vm2i ej(α2i +θ1i )
I
tk = − =−
jXt2i
i=1 i=1 jXt2i
k = 1, . . . , R (4.58)
where Yfi fi , Ytk tk , and Yfi tk are the nodal admittances and mutual admittances at the
bus fi and bus tk of the system without considering FACTS Controllers, respectively.
Next, we reduce the bus admittance matrix to the generator internal buses
and the FACTS controllers’ from-buses and to-buses. The corresponding reduced
bus admittance matrix equation takes the form
Ig
ȲGG ȲGF ȲGT E
Ȳ V = IF (4.59)
FG ȲFF ȲFT f
ȲTG ȲTF ȲTT Vt IT
where E is the generator internal voltage vector behind the transient or subtransient
reactance.
It is clear that in (4.57) and (4.58), Vm1 , α1 , Vm2 , and α2 are known from
the control outputs Vm1 , α1 , Vd , and Vq in Section 4-1. The only unknown is the
from-bus angle θ1 , which can be obtained from the network solution. An iterative
process can be applied to obtain the solutions of Vf , θ1 , and Vt .
Use the Newton’s method to solve for the variables VFre , VFim , VTre , and VTim itera-
tively as
VFre new = VFre old + ∆VFre
VFim new = VFim old + ∆VFim
(4.62)
VTre new = VTre old + ∆VTre
VTim new = VTim old + ∆VTim
−1
∂∆F1re ∂∆F1re ∂∆F1re ∂∆F1re
∆VFre ∂VFre ∂VFim ∂VTre ∂VTim
∆F1re
∆VF ∂∆F1im ∂∆F1im ∂∆F1im ∂∆F1im ∆F1
im
im
=
∂VFre ∂VFim ∂VTre ∂VTim
(4.63)
∂∆F2re ∂∆F2re ∂∆F2re ∂∆F2re
∆VTre ∆F2re
∂VFre ∂VFim ∂VTre ∂VTim
∂∆F2im ∂∆F2im ∂∆F2im ∂∆F2im
∆VTim ∂VFre ∂VFim ∂VTre ∂VTim
∆F2im
Then we get VF = VFre + jVFim and VT = VTre + jVTim . Substituting VF and
VT into the first equation of (4.59) gives the current injections Ig into the generator
internal buses
Ig = ȲGG E + ȲGF VF + ȲGT VT (4.64)
This multi-step scheme will result in a second-order accuracy of the solution, that
is, the local error of the method, which is the difference between the approximate
solution xk obtained by using this method and the exact solution x∗k of the differential
equation, is O((∆t)3 ) as ∆t → 0.
70
G3
G1 1 4 5
G2 2 3 11 6
VSC 2
12 VSC 3
13
VSC 1
10
22
14
8 7
G4
G5
15
9
19
G6 16
18
20
17
21
Load 1
Load 2
Figure 4.12 shows the simulation results of the STATCOM in var control mode
under the operating conditions in Table 4.1. A positive value of reactive power in-
jection reference Ishqset implies capacitive shunt reactive power compensation, while
a negative value implies inductive compensation.
When the reactive current reference changes from inductive to capacitive, the
shunt reactive power injected into the from bus by the VSC will respond to the
change, and thus will cause the from-bus voltage increase. Both the DC capacitor
voltage and the inverter voltage increase but the ratio between them is kept constant.
As shown in Figure 4.12, when the shunt reactive power compensation changes
from full inductive to full capacitive, the DC capacitor voltage increases from about
−20% below to 20% above nominal.
Note that the fast oscillations in the system voltage are due to the FACTS
controller and generator automatic voltage controllers, and the slower oscillations
72
are the effect of the superposition of the impact of all machine swing modes. This
also explains the voltage oscillations for all the following cases.
Figure 4.13 shows the simulation results of the SSSC in inverter voltage magni-
tude control mode under the operating conditions in Table 4.2. The polarity of Vmref
indicates that the insertion of the SSSC is either inductive when it is positive or is
capacitive when it is negative. We observe that the DC capacitor voltage decreases
from over 10 kV to zero and then goes back up, while the from-bus voltage keeps
decreasing from 1.0122 pu to 1.0058 pu. The line active power decreases about 160
MW with the SSSC control from 0.05 pu capacitive to 0.05 pu inductive.
Table 4.3: Operating Conditions of the UPFC in V ,Vd ,Vq Control Mode
Shunt Series Original Disturbances
Gains Gains Setpoint Event 1 Event 2 Event 3
Kv =500 Kp =0.01 Vset =1.025 pu At t=1 s, At t=7 s, At t=13 s,
Kp =0.01 Ki = 0.1 α=0.03 shunt Vset has series Vdref has series Vqref has
Ki = 0.1 T =0.02 Vdref =0.01 pu a step increase a step increase a step decrease
T =0.02 Vqref =−0.02 pu of 0.01 pu. of 0.02 pu. of 0.02 pu.
73
Figure 4.14 shows the simulation results of the UPFC in V ,Vd ,Vq control mode
under the operating conditions in Table 4.3. We observe that the from-bus voltage,
the series VSC voltage Vd , and the series VSC voltage Vq are independently con-
trolled to their reference values. The change of Vq mainly affects line active power,
while the change of Vd mainly affects line reactive power.
Figure 4.15 and 4.16 show the simulation results of the IPFC in inverter voltage
control mode under the operating conditions in Table 4.4. We observe that the
Master VSC Vd , the Master VSC Vq , and the Slave VSC Vq are independently
controlled to their reference values. Note that the DC capacitor voltage is also
controlled to its reference value. The step change (-0.01 pu) of the Master line Vdref
causes a 0.009 pu (3 kV in nominal) increase of the to-bus voltage of the Master
line and the same amount of decrease of the to-bus voltage of the Slave line at the
same time.
on the Bus 3 side of Line 3-13, which is a line paralleled with Line 4-11. Its near end
is cleared at t=0.15 s and remote end is cleared at t=0.17 s. The maximum load
on Bus 17 that the system can stand during the fault and the corresponding power
transfers on Line 4-11 and Line 4-12 are displayed in Table 4.5 for seven different
system configurations. The setpoints for these configurations are simply specified
on their rated capacity.
As shown in Table 4.5, a STATCOM with 200 MVA rating can support over
60 MW more load on Bus 17 than the configuration without a FACTS controller
(Config. 1). The corresponding power flows on Line 4-11 and Line 4-12 increase
about 30 MW and 20 MW, respectively. Compared with Config. 1, the system with
75
an 100 MVA SSSC in Line 4-11 (Config. 5) stands 40 MW more load on Bus 17.
The corresponding power flow on Line 4-11 increases 70 MW while that on Line
4-12 decreases 7 MW.
Considering the system with a UPFC, which consists a 100 MVA shunt VSC
on Bus 4 coupled with a 100 MVA series VSC in Line 4-11, if the series VSC of
the UPFC is operated in the line P ,Q control mode, the system will crash during
the fault because the system can not transmit enough power from the Northwest
generators to the Southeast loads with Line 3-13 tripped and the P setpoint of Line
4-11 fixed. So the series VSC of the UPFC should be operated in the inverter Vd ,Vq
control mode as in Config. 6 and Config. 7. Each of these two configurations can
support over 100 MW more load on Bus 17 than Config. 1 and transmit about 80
MW more power flow on Line 4-11 while keep that on Line 4-12 unaffected.
Config. 8 and Config. 9 are configurations with an IPFC, which consists of a
100 MVA series VSC in Line 4-11 coupled with a 100 MVA series VSC in Line 4-12.
Both configurations can stand over 60 MW more load on Bus 17 than Config. 1.
Each carries over 100 MW more the total power flows of Lines 4-11 and 4-12.
Figure 4.17 shows the dynamic simulations of Config. 1 and Config. 7 in the
same loading conditions where the active power of Load L2 on Bus 17 is 3235 MW.
The setpoints of Config. 7 is the same as shown in Table 4.5. We observe that the
UPFC reduces transient oscillations in voltages on Bus 3 and Bus 4 and power flows
on the critical paths Line 4-11 and Line 4-12 during the fault, in addition to its
capability on increasing post-fault bus voltages and power transfers.
To summarize, FACTS controllers can substantially improve the transient
power transfer capability of a transmission system during a fault.
the operators and equipment engineers, without performing the experiments on the
real hardware.
77
1.04 1.6
1.035
1.5
AC voltage per unit
DC voltage volts
1.03
1.4
1.025
1.3
1.02
1.2
1.015
1.1
1.01
1.005 1
1 0.9
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds time in seconds
STATCOM reactive current injection STATCOM ∆I = I* −I
shq shq shq
2 2
1
1
0.5
I*shq 0.5
0
I
shq
0
−0.5
−1 −0.5
−1.5 −1
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds time in seconds
1.4 2
0
1.3
AC voltage per unit
−2
angle rad
1.2
−4
1.1
−6
1
−8
0.9 −10
0.8 −12
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds time in seconds
STATCOM active power injection P STATCOM reactive power injection Q
sh sh
0.04 2
0.02 1.5
Reactive power per unit
1
Real power per unit
0
0.5
−0.02
0
−0.04
−0.5
−0.06
−1
−0.08 −1.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds time in seconds
1.012
AC voltage per unit
DC voltage volts
10000
1.01
1.008
5000
1.006
1.004 0
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10
time in seconds time in seconds
Series VSC voltage magnitude Series VSC voltage angle w.r.t. from−bus voltage angle
0.06 2
1.5
0.05
1
AC voltage per unit
0.04
0.5
angle rad
0.03 0
−0.5
0.02
−1
0.01
−1.5
0 −2
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10
time in seconds time in seconds
SSSC line active power into the to−bus SSSC line reactive power into the to−bus
8.5 −0.1
−0.2
8
power flow per unit
−0.3
7.5
−0.4
7
−0.5
6.5 −0.6
0 1 2 3 4 5 6 7 8 9 10 0 2 4 6 8 10
time in seconds time in seconds
1.04
DC voltage volts
1.035 1.6
1.03 1.5
1.025
1.4
1.02
1.3
1.015
1.01 1.2
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
UPFC line active power into the to−bus UPFC line reactive power into the to−bus
8.2 −0.25
−0.3
8.1
−0.35
8
−0.4
7.9 −0.45
−0.5
7.8
−0.55
7.7
−0.6
7.6 −0.65
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
Shunt VSC inserted voltage magnitude Shunt VSC injected voltage angle w.r.t. from−bus voltage angle
1.7 0.035
1.6
0.03
1.5
AC voltage per unit
0.025
angle rad
1.4
0.02
1.3
0.015
1.2
0.01
1.1
1 0.005
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
Series VSC inserted voltage Vd Series VSC inserted voltage Vq
0.035 −0.015
0.03 −0.02
AC voltage per unit
0.025 −0.025
angle rad
0.02 −0.03
0.015 −0.035
0.01 −0.04
0.005 −0.045
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
1.2
1.0245
AC voltage per unit
DC voltage volts
1.15
1.024
1.1
1.0235
1.05
1.023
1
1.0225 0.95
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
IPFC master to−bus voltage magnitude IPFC slave to−bus voltage magnitude
1.016 1.05
1.048
1.014
1.046
1.012
1.044
1.01 1.042
1.008 1.04
1.038
1.006
1.036
1.004
1.034
1.002 1.032
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
IPFC master line active power into the to−bus IPFC master line reactive power into the to−bus
7.65 −0.3
−0.35
7.6
−0.4
power flow per unit
7.55
−0.45
−0.5
7.5
−0.55
7.45
−0.6
7.4 −0.65
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
IPFC slave line reactive power into the to−bus
IPFC slave line active power into the to−bus 1.05
7.88
7.86
1
7.84
power flow per unit
power flow per unit
0.95
7.82
7.8 0.9
7.78
0.85
7.76
7.74 0.8
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
0.02 0
0.018 −2
AC voltage per unit
angle rad
0.016 −4
0.014 −6
0.012 −8
0.01 −10
0.008 −12
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
Slave VSC inserted voltage Vd Slave VSC inserted voltage Vq
−0.008 −0.028
−0.01
−0.03
−0.012
−0.032
AC voltage per unit
−0.014
angle rad
−0.034
−0.016
−0.036
−0.018
−0.038
−0.02
−0.022 −0.04
−0.024 −0.042
0 2 4 6 8 10 12 14 16 18 20 0 5 10 15 20
time in seconds time in seconds
0.9 0.9
voltage in pu
voltage in pu
0.8 0.8
0.7 0.7
14 9
12 8
10 7
8 6
0 1 2 3 4 5 0 1 2 3 4 5
time in seconds time in seconds
Figure 4.17: Comparison of No FACTS and UPFC in V ,Vd ,Vq Mode When
PL2 =3235 MW
CHAPTER 5
LINEARIZED MODELS AND MODAL
DECOMPOSITION OF MULTI-MACHINE SYSTEMS
83
84
all the swing modes in the sense that the inner loop does not include the interac-
tion of them and the dominant component is the network effect. In this thesis, a
modal decomposition approach is proposed based on the method in [59] by using
the canonical state transformation to fully decouple all the state modes. But in
the new approach, the interaction of other swing modes to the mode of interest, is
included in the inner-loop sensitivity KIL (s) to provide a better representation of
the linearized multi-machine systems.
In this chapter, small-signal linearization of a multi-machine system is dis-
cussed in Section 5.1. System modal analysis methods to study the linearized system
and to identify the inter-area mode are described in Section 5.2. The new multi-
machine modal decomposition approach is presented in Section 5.3. An application
example is given in Section 5.4.
this process is a single step in the dynamic simulation program. The input matrix
B, the output matrix C, and the feedforward matrix D can be determined in a
similar manner.
The linearized model of a multi-machine system can be expressed in the state
space form
ẋ = Ax + Bu, y = Cx + Du (5.1)
where u and y are the vectors of control and measurement variables, respectively,
and x is the vector of state variables, which can be arranged as
where the ∆δ’s and the ∆ωg ’s represent the perturbed generator angles and speeds,
respectively, and z is the vector of all the other state variables.
AUi = λi Ui (5.3)
The right eigenvector gives the mode shape, i.e., the relative activity of the state
variables when a particular mode is excited. For example, the degree of activity
of the state variable xj in the ith mode is given by the element uj,i of the right
eigenvector Ui . The magnitudes of the elements of Ui give the extents of the activities
of the n state variables in the ith mode, and the angles of the elements give phase
86
displacements of the state variables with regard to the mode. For a n-mode system,
the right eigenvectors for all the eigenvalues form a n×n matrix U = [U1 U2 . . . Un ].
A left eigenvector of the state matrix A is defined as a row vector Vi satisfying
Vi A = λi Vi (5.4)
The left eigenvector Vi identifies which combination of the original state variables
displays only the ith mode. Thus the jth element of the right eigenvector Ui mea-
sures the activity of the variable xj in the ith mode, and the jth element of the
left eigenvector Vi weighs the contribution of this activity to the ith mode. For
a n-mode system, the left eigenvalues for all the eigenvalues form a n × n matrix
V = [V1T V2T . . . VnT ]T .
B. Participation factor
Participation factors are nondimensional scalars that measure the interaction
between the modes and the state variables of a linear system. Participation factors
give the sensitivity of an eigenvalue to a change in the diagonal elements of the state
matrix.
Participation factors for an n-mode system are defined as [65] and can be
formed as
pk,i = uj,i vi,j (5.5)
of the eigenvector mean that the inter-area mode may be most readily observed
by monitoring those states. It does not necessarily mean that these states are
necessarily good for controlling the inter-are mode.
such that
T T
x = [∆δm1 ∆δm2 . . . ∆δmn ∆ωmg1 ∆ωmg2 . . . ∆ωmgn zm ] (5.7)
where the ∆δm ’s and the ∆ωm ’s represent perturbed modal generator angles and
speeds, respectively. This transformation produces a canonical state-space realiza-
tion of (5.1)
ẋc = Am xc + Bm u, y = Cm xc + Dm u (5.8)
where the real eigenvalues appear on the diagonal of the state matrix Am and the
88
and
Bm = T B, Cm = CT −1 , Dm = D (5.10)
where σi and ωi , i = 1, . . . , n, are the real and imaginary parts of system swing
modes λi , respectively, and Az is the state matrix where all the other modes appear
on its diagonal.
For a swing mode λi = σi + jωi corresponding to the inter-area mode of
interest, the state variables in xc can be rearranged such that the modal angle ∆δmi
and speed ∆ωmi correspond to become the first and second state variables, resulting
in the system representation
∆δ̇mi σi ωi 0 ∆δmi bmi1
∆ω̇ 0 + b
mi = −ωi σi ∆ωmi mi2 u (5.11)
żr 0 0 Ar zr Br
∆δmi
y= cm1 cm2 Cr
∆ωmi + Dm u (5.12)
zr
where zr is the vector of all the other state variables except the inter-area mode
state variables ∆δmi and ∆ωmi , and Ar , Br , and Cr are state matrices associated
with zr .
The main advantage of this system representation is that the state equations
for ∆δmi and ∆ωmi are decoupled from those of the other state variables. As a
89
result, we have
∆δ̇mi σi ωi ∆δmi bmi1
= + (5.13)
∆ω̇mi −ωi σi ∆ωmi bmi2
∆δmi
ymi = cmi1 cmi2 + Dmi u (5.14)
∆ωmi
where αi1 = −2σi , αi2 = σi2 + ωi2 , βi1 = bmi1 cmi1 + bmi2 cmi2 , and βi2 = −σi bmi1 cmi1 +
ωi bmi2 cmi1 − σi bmi2 cmi2 − ωi bmi1 cmi2 .
The numerator of Tmi (s) consists of the product of the controllability Kci (s)
and the observability Koi (s) of the inter-area mode of interest, that is
Kci(s) 1 1 Koi(s)
s s
_
αi1
_
αi2
Inter-Area Mode of Interest
+
+
KIL(s)
Inner Loop
Inner Loop
Feedback
u y
KPSDC(s)
522
G3
1433
G1 1 4
786 3
G2 2 11 6
VSC 2
12 VSC 3
13
VSC 1
7 10
14
628 8
G4
15
9
59 1177 19
G5 16
20
17
18
Load 1
1526 Load 2
2700
90 90
0.8 0.08
120 60 120 60
0.6 0.06
0.2 0.02
G4 G2
G3 G3
180 G2 0 180 G5 0
G1 G4
G5 G1
90 0.15 90
0.25
120 60 120 60
0.2
0.1
0.15
150 30 150 30
0.1
0.05 G2
G3 0.05
G5
180 G4 0 180 0
G1
G3
G2
G5
210 330 210 330
G4
G1
In [57] and [58] influence factors and control effectiveness were established to select
appropriate controller input signals for SVC and TCSC, respectively. Using these
techniques it has been demonstrated in several example power systems that for
effective damping control, shunt controllers (like SVC) should use flow (such as
line current magnitude) measurements and series controllers (like TCSC) should
use nodal (bus voltage) measurements [66], [67]. This duality in controllability
and observability is perhaps not surprising because intuitively a series controller
regulating line flows would need an orthogonal nodal signal to get the damping
information. The converse is true for a shunt controller.
Reference [68] presented the control strategies of the UPFC, controllable series
capacitor, and quadrature boosting transformer (QBT) for damping of electrome-
chanical power system oscillations based on the Control Lyapunov Functions (CLF).
The authors in [68] used Lyapunov function candidates in feedback design itself by
making the Lyapunove derivative negative when choosing the control. However,
considering the main functions of FACT controllers are to regulate system voltages
and power flows while damping power system oscillations is a supplemental function,
we aim to design the damping controllers based on their regulation controllers.
In this chapter, we use the new modal decomposition technique described in
Chapter 5 to analyze controllability, observability, and inner-loop gains for shunt,
series, and coupled VSCs in multi-machine systems to investigate the design of VSC-
based damping controllers supplemental to their regulation controllers for inter-area
modes. Although we will be interested in all FACTS Controllers, the design for
stand-alone shunt VSCs (STATCOMs) follows a similar line as in SVC [57]. The
focus here is on series VSCs, and series SVCs coupled to other VSCs (such as UPFCs
and IPFCs). Several papers [49], [70] on damping control design for UPFCs have
been published, but most of these papers only discussed the design mechanism.
Here we will pursue a comprehensive approach that will examine the selection of
95
96
damping controller input signals, the controller gain limitation, and modal damping
selectivity.
This chapter is organized as follows. Section 6.1 gives the block diagram
of a damping controller to be designed. For damping input signal selection, three
quantifiable indices based on the new multi-machine modal-decomposition approach
are described in Section 6.2. Design examples for the STATCOM, SSSC, UPFC,
and IPFC are discussed in Sections 6.3-6.5.
where
KPSDC = kGf (s)Gp (s)Gw (s) (6.2)
Error signal
+ FACTS Power
Regulator System
Damping +
signal u umax
1 1+Tns Tw s Measured
k
1+Tf s 1+Tds 1+Tw s signal y
umin
Gf (s) Gp (s) Gw (s)
KPSDC(s)
Tw s
Gw (s) = (6.3)
1 + Tw s
ẋw = (y − xw )/Tw
(6.4)
Fo1 = y − xw
where xw is the state variable. Its realization in block diagram is shown in Figure
6.2.
Measured
Fo1 + signal y
_
1 1
Tw .
x
s xw
w
1 + Tn s
Gp (s) = (6.5)
1 + Td s
where Tn and Td are constant coefficients. Its time-domain state equation is derived
as
ẋp = (Fo1 − Fo2 )/Td
(6.6)
Fo2 = xp + Fo1 Tn /Td
where xp is the state variable. Its realization in block diagram is shown in Figure
6.3.
Coefficients Tn and Td are designed based on phase compensation to generate
a proper damping signal. If phase-lead compensation is needed, we have
s + zld
Gp (s) = αld (6.7)
s + pld
98
Tn
Td
Fo2 +
1 1 + Fo1
+ xp s .
xp Td _
where ω is the frequency of the mode of interest to be compensated and αld deter-
mines the amount of the phase-lead compensation. The larger the αld , the higher
the phase-lead compensation is, although the relationship is not linear. In phase
compensation design, the values of αld are varied to find an optimal choice.
For a given try value of αld , the coefficients Tn and Td of a phase-lead com-
pensator can be calculated based on (6.7)-(6.9) as
√ 1
Tn = αld ω, Td = √ (6.10)
αld ω
1 s + zlg
Gp (s) = (6.11)
αlg s + plg
as follows
1 √
Tn = √ , Td = αlg ω (6.14)
αlg ω
The low pass filter has a transfer function
1
Gf (s) = (6.15)
1 + Tf s
where xf is the state variable. Its realization in block diagram is shown in Figure
6.4.
Fo3 .
xf 1
xf
1 + Fo2
s Tf _
State equations (6.4), (6.6), and (6.16) are incorporated into the dynamic
simulation program, together with other state equations for FACTS controllers in
Chapter 4. Then all the dynamic equations are linearized by performing small
perturbations.
Given a choice of shunt and series VSC configurations, two important issues
in damping control design are: which regulator configuration should be used and
which measured signal should be used as the damping input signal y. For a specified
regulator and measured signal, the damping controller as shown in Figure 6.1 can
be readily designed based on root-locus plot and bode plot techniques.
100
KPSDC (s)
Kei (s) = Kci (s) Koi (s) (6.17)
1 − KPSDC (s)KIL (s)
which describes the impact of a given damping controller KPSDC (s) on the ith swing
mode.
By analyzing the transfer functions and the effective control action at the
frequency of the inter-area mode, we create two useful indices which provide insights
to the performance of a damping controller with the given measurements.
A. Maximum Damping Influence (MDI) Index
By assuming that |KPSDC (s)KIL (s)| >> 1, the gain of the effective control
action in (6.17) can be simplified as
K (s)K (s)
ci oi
MDI = (6.18)
KIL (s)
which is denoted as the maximum damping influence (MDI) index [58]. The MDI
index is a measure of the eigenvalue shift per unit control gain achievable based on
the assumption. The value of the MDI index is that it indicates the effectiveness of
measurements having high observability gain and low inner-loop gain. The MDI is
a useful index to exclude those damping signal candidates who are not suitable as
input signals.
There is also the gain margin consideration that the control gain of KPSDC (s)
101
is limited, such that the assumption |KPSDC (s)KIL (s)| >> 1 is not applicable by
itself. Thus, it is necessary to create other indices to evaluate the candidates chosen
by the MDI index.
B. Controllability and Observability Gain Product Index
A KPSDC (s) with a smaller control gain is preferable because a smaller gain
would provide a higher gain margin. Considering that the candidates chosen by the
MDI index and inner-loop gain index usually have small inner-loop gain, we can
assume that |KPSDC (s)KIL (s)| << 1. The first derivative of the effective control
action gain to the gain k of KPSDC (s) = kGc (s) is expressed as
dK (s) K (s)G (s)K (s)
ei ci c oi
= (6.19)
dk (1 − kGc (s)KIL (s))2
Under the assumption of |KPSDC (s)KIL (s)| << 1, (6.19) can be simplified as
dK (s)
ei
= |Kci (s)Koi (s)| · |Gc (s)| (6.20)
dk
which indicates that as k increases from zero, the larger the product of controlla-
bility and observability gains is, the faster the control effect changes. This means
a relatively small k could achieve the desired damping improvement. Thus a large
value of the product of controllability and observability gains is preferred. Based
on this discussion, we select candidate signals with high values of the product of
controllability and observability gains from those ranked by the MDI index. They
become candidates for damping controller design for testing by dynamic simulations.
The candidate that achieves the best dynamic performance will be finalized as the
selected signal for damping controller design.
Bus 4 of the same 20-bus test system as shown in Figure 5.2. The STATCOM is
operated in the Var control mode. The list of candidate signals contain three local
signals V4 , P4−11 , and Im4−12 .
64
Ki=0.005
Ki=0.1
Ki=1
Ki=5
60
56
52
26
25
24
Ki=0.005
Ki=0.1
Ki=1
23 Ki=5
5.85
Ki=0.005
Ki=0.1
Ki=1
5.8 Ki=5
5.75
5.7
5.65
0 0.05 0.1 0.15 0.2
Shunt Kp
In Figure 6.5, the four curves in each subplot are corresponding to the integral
gain Ki of the Var regulation loop set to 0.005, 0.1, 1, and 5, respectively. When
the proportional gain Kp of the Var regulation loop varies from 0.0 to 0.2, the MDI
103
68
Kp=0.005
Kp=0.01
64 Kp=0.1
Kp=0.5
60
56
52
26
25
Kp=0.005
24
Kp=0.01
Kp=0.1
Kp=0.5
23
6.0
Kp=0.005
Kp=0.01
5.9 Kp=0.1
Kp=0.5
5.8
5.7
5.6
0.001 0.5 1 1.5 2
Shunt Ki
indices for signals Im4−11 , P4−11 , and V4 converge to 53.0, 25.6, and 5.7, respectively.
For a higher Ki , the MDI indices of a signal are closer to the converged value. When
Ki is high enough, the MDI indices of a signal are not sensitive to the varying of
Kp .
The four curves in Figure 6.6 are corresponding to Kp =0.005, 0.01, 0.1, and
0.5, respectively. When Ki increases from 0.001 to 2.0, the MDI indices for signals
Im4−11 , P4−11 , and V4 converge to 53.0, 25.6, and 5.7, respectively. For a higher
Kp , the MDI indices of a signal are closer to its converged value. When Kp is high
enough, the MDI indices of a signal are not sensitive to the varying of Ki .
104
For any combination of Kp and Ki in Figures 6.5 and 6.6, we observe that
using Im4−11 as the damping input signal has the highest MDI indices. Thus, Im4−11
is selected as the damping input signal for the STATCOM damping controller to
design. It is consistent with the conclusions in [61], [66], and [67] that flow variables
are more suitable for shunt devices.
We show an example of designing the damping controller on the STATCOM
in the Var control mode with regulation gains Kp = 0.01 and Ki = 0.1. Based on
root-locus plots, the damping controller using input signal Im4−11 is designed as
0.1s 1
u = 20 · · (−0.1 ≤ u ≤ 0.1) (6.21)
1 + 0.1s 1 + 0.1s
damping signal
0.01
−0.01
per unit
−0.02
−0.03
−0.04
−0.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
Figures 6.8 and 6.9 show the dynamic simulation results without and with the
designed damping controller on the STATCOM. In the simulation, the STATCOM
Var reference Ishqref has a step increase from 0.0 pu to 0.1 pu at time t = 0.1
s. As shown in Figure 6.8, the damping controller results in substantial damping
improvement on the bus voltage V4 , line current Im4−11 , and line power flows P4−11
and Q4−11 .
Figure 6.9 shows the STATCOM variables including the shunt reactive power
injection Ishq , the DC bus voltage Vdc , and the inverter voltage magnitude Vm1
105
7.47
1.0325
AC voltage per unit
1.032
7.466
1.0315
7.464
no dmp
dmp input I
m4−11
1.031 7.462
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
line active power P line reactive power Q4−11
4−11
7.72
−0.014
7.715 −0.016
power flow per unit
7.71 −0.02
−0.022
7.705
−0.024
−0.026
7.7
−0.028
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
and angle αsh , all of which are affected to have larger oscillations due to injecting
the damping signal into the Var regulator. We also notice that with the damping
controller the voltage V4 and the line power flows P4−11 and Q4−11 have extra drops
between t=0.2 s and t=0.5 s (Figure 6.8). This is the side effect due to the high
injection of the damping signal at that time range. The high injection arises from the
quick drop of the damping input signal at the moment of the setpoint step change.
These are the costs needed to pay for improving system damping by building a
feedback damping controller on the STATCOM’s regulation control.
1.265
0.1
DC voltage volts
current per unit
1.26
1.255
0.05 1.25
1.245
1.24
0
1.235
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
invter voltage magnitude Vm1 −4 invter voltage angle αsh
x 10
2
1.06
1
1.055
0
AC voltage per unit
angle radians
1.05 −1
−2
1.045
−3
1.04
−4
1.035
−5
1.03 −6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
inverter Vm control mode. The MDI indices calculated for the SSSC using signals
V4 , P4−11 , and Im4−11 as the damping input signal are displayed by the 3 curves
in Figure 6.10, respectively. In this plot, the proportional gain Kp of the SSSC
regulator varies from 0.3 to 25 while the integral gain Ki is kept on Ki = 10Kp . As
Kp increases, the MDI indices for V4 , P4−11 , and Im4−11 converge to 56.5, 2.1, and
2.8, respectively. When Kp ≥ 1, signal V4 has much higher MDI index than the
other two signals. So V4 is selected to be the damping input signal for the SSSC.
It is consistent with the conclusions in [61], [66], and [67] that nodal variables are
more suitable for series devices.
We show an example of designing a damping controller on the SSSC in the
inverter Vm control mode with regulation gains Kp = 20 and Ki = 200. Based on
root-locus plots, the damping controller using input signal V4 is designed as
0.1s 1 + 0.381s 1
u = 10 · · · (−0.1 ≤ u ≤ 0.1) (6.22)
1 + 0.1s 1 + 0.0762s 1 + 0.1s
107
MDI Index
80
60
Signal V4
40
Signal P4-11
Signal Im4-11
20
0
1 2 5 10 15 20 25
Series Kp
Figure 6.10: SSSC MDI Index Plots Varying Regulation Control Gains
0
per unit
−2
−4
−6
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
Figures 6.12 and 6.13 show the dynamic simulation results without and with
the designed damping controller for the SSSC. In the simulation, the SSSC inverter
voltage reference Vmref has a step decrease from −0.01 pu to −0.02 pu at time t = 0.1
s. As shown in Figure 6.12, the damping controller results in substantial damping
improvement on the bus voltage V4 , line current Im4−11 , and line power flows P4−11
and Q4−11 .
Figure 6.13 shows the SSSC variables including the DC bus voltage Vdc and the
inverter voltage magnitude Vm2 and angle αse , all of which are affected to have larger
oscillations due to injecting the damping signal into the inverter Vm regulator. We
also notice that with the damping controller the current Im4−11 and the line power
108
7.3
1.03 7.15
7.1
1.0295 7.05
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
line active power P line reactive power Q
4−11 4−11
7.55 −0.05
7.5
power flow per unit
7.4
7.35 −0.15
7.3 no dmp
dmp input V
4
7.25 −0.2
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
flows P4−11 and Q4−11 have extra swells between t=0.27 s and t=0.7 s (Figure 6.12).
This is the side effect due to the high injection of the damping signal at that time
range. These are the costs needed to pay for improving system damping by building
a feedback damping controller on the SSSC’s regulation control.
5000
DC voltage volts
4000
3000
no dmp
dmp input I
m4−11
2000
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
series inverter voltage magnitude V
m2
invter voltage angle α
se
0.026
1.62
0.024
1.6
0.022
AC voltage per unit
1.58
0.02
angle radians
1.56
0.018
0.016 1.54
0.014 1.52
0.012 1.5
no dmp
0.01 dmp input V4 1.48
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
to Vref = 1.031 pu and the Var control setpoint is set to Ishqref = 0.0 pu. The
control gains in the shunt voltage control mode are Kv = 500, α = 0.03, Kp = 0.01,
and Ki = 0.1, and the control gains in the shunt Var control mode are Kp = 0.01
and Ki = 0.1. The setpoints of the series VSC are set to Vdref = 0.0199 pu and
Vqref = −0.0373.
Table 6.1 shows that the MDI index values of the UPFC are close to those
of the STATCOM. It implies that the coupled series VSC in Vd ,Vq mode has little
impact on the damping control of the shunt regulator. Using Im4−11 as the damping
input signal to the shunt regulator has the highest MDI indices for the UPFC in
Vd ,Vq mode and thus Im4−11 is selected as the damping input signal.
Table 6.1: MDI Indices for the UPFC Series Vd ,Vq Mode v.s. the STAT-
COM
Local MDI Values
Signals STATCOM UPFC Series in Vd ,Vq Control
Voltage Control Var Control Shunt Voltage Control Shunt Var Control
V4 4.92 5.70 4.89 5.65
P4−11 26.56 25.54 22.90 21.77
Im4−11 64.39 53.34 56.96 48.33
Kp = 0, the UPFC is actually degraded into Var,Vd ,Vq mode with the two series
regulators disabled. By increasing Kp , the impact of the series regulators increases.
The three plots in Figure 6.14 show the MDI indices for the shunt Var regula-
tor, series P regulator, and series Q regulator of the UPFC in Var,P ,Q control mode,
respectively. The three curves in each plot are corresponding to the MDI indices us-
ing V4 , P4−11 , and Im4−11 as the damping input signal, respectively, as Kp increases
from 0 to 0.2. In the simulation, the shunt Var setpoint is set to Ishqref = 0.0 pu,
and its control gains are Kp = 0.01 and Ki = 0.1. The series VSC setpoints are set
to Pref = 820 MW and Qref = −2 MVar.
For the shunt regulator, when Kp = 0, which is the case without the impact
of the two series regulators, the signal Im4−11 has the highest MDI indices. As Kp
increases from 0 to 0.2, which means the impact of the series regulators increases,
the MDI indices of Im4−11 , P4−11 , and V4 drop from 57.4 to 12.3, 20.9 to 14.8, and
5.6 to 4.4, respectively. Based on the above observation, we conclude that the signal
Im4−11 , which is a good damping input signal for the STATCOM and the UPFC in
Vd ,Vq mode, will be affected significantly by the coupled series P ,Q regulators.
In the second plot the MDI indices of V4 , Im4−11 , and P4−11 to the series P
regulator decrease from 23.8 to 6.3, 2.9 to 1.2, and 2.1 to 0.04, respectively, as Kp
varies from 0 to 0.2. Among these three signals, V4 has the highest MDI indices.
But the MDI indices of V4 is less than 10 when Kp ≥ 0.03, which is low for a good
damping input signal.
The third plot shows that the MDI indices of Im4−11 , P4−11 and V4 to the series
Q regulator decrease from 44.1 to 6.0, 13.6 to 5.8, and 6.6 to 5.0, respectively, as
Kp varies from 0 to 0.2. When Kp ≥ 0.5, all MDI indices of the three signals are
111
MDI Index
60
Signal V4
50 Signal P4-11
Signal Im4-11
40
30
20
10
0
25
Signal V4
20 Signal P4-11
Signal Im4-11
15
10
0
50
Signal V4
40 Signal P4-11
Signal Im4-11
30
20
10
0
0 0.2 0.4 0.6 0.8 1
Series Kp
less than 10. As a result, none of them is selected as the damping input signal when
Kp ≥ 0.5.
P ,Q regulators. Thus the UPFC with its series in the Vd ,Vq control mode is more
suitable to design a damping controller on its shunt VSC. We use the same 20-bus
system (Figure 5.2) to demonstrate this conclusion. Figures 6.15 (a) and (b) show
the root-locus plots of the four system swing modes in designing a damping controller
on the shunt VSC Var regulator for the UPFC series in the inverter Vd , Vq control
mode and the line P ,Q control mode, respectively. The shunt VSC is operated in
Var control mode with control gains Kp = 0.01 and Ki = 0.1. The series VSC
is inserted into Line 4-12. The control gains of the P ,Q regulators are Kp = 0.1
and Ki = 1. The damping input signal is Im4−11 and is the damping controller is
expressed as
0.1s 1
u=k· − 0.1 ≤ u ≤ 0.1 (6.23)
1 + 0.1s 1 + 0.1s
9 9
8 8
7 7
k=30
6 6 k=30
Imag
Imag
k=0
5 k=0
5
4 4
3 3
2 2
1 1
−2 −1.5 −1 −0.5 0 −2 −1.5 −1 −0.5 0
Real Real
(a) UPFC Series in Vd ,Vq Control (b) UPFC Series in P ,Q Control
As shown in Figures 6.15 (a) and (b), when the control gain k of the damping
controller increases from 0 to 30, the real part of the inter-area mode is moved from
−0.55 to −1.08 in Vd ,Vq control and from −0.27 to −1.59 in P ,Q control, respectively.
Thus, the former one has better damping effect on the inter-area mode.
113
Figures 6.16-6.18 show the dynamic simulation results without and with the
designed damping controller (k = 30) on the shunt Var regulator of the UPFC in
Vd ,Vq control. In the simulation, the shunt Var setpoint is set to Ishqref = 0.0 pu and
the series setpoints are set to Vdref = 0.0199 pu and Vqref = −0.0373 pu. A 0.001 pu
step increase of Vdref is applied at time t = 0.1 s.
damping signal
0.04
0.02
per unit
−0.02
−0.04
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
8.081
1.035 8.08
8.079
1.0345 8.078
8.077
1.034 8.076
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
line active power P4−11 line reactive power Q4−11
8.205
−0.2
8.2
−0.205
power flow per unit
8.195 −0.21
−0.215
8.19
−0.22
8.185
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds time in seconds
Figure 6.17 shows that the damping controller results in substantial damping
improvement on the bus voltage V4 , line current Im4−11 , and line power flows P4−11
114
and Q4−11 . Figure 6.18 shows the oscillations on inverter variables including the DC
bus voltage Vdc , the shunt inverter voltage magnitude Vm1 and angle αsh , and the
series d-axis and q-axis voltages Vd and Vq , all of which are affected to have larger
oscillations due to injecting the damping signal into the shunt Var regulator.
shunt reactive current I x 10
4
DC bus voltage V
shq dc
0.12 1.27
no dmp
0.1 dmp input Im4−11 1.265
0.08 1.26
DC voltage volts
current per unit
0.06 1.255
0.04
1.25
0.02
1.245
0
1.24
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds time in seconds
shunt inverter voltage magnitude Vm1 shunt inverter voltage angle αsh
1.06 0.0265
1.055
AC voltage per unit
1.05 0.026
angle radians
1.045
1.04 0.0255
1.035
1.03 0.025
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 0 1 2 3 4 5
time in seconds
time in seconds
0.0208
0.5
AC voltage per unit
AC voltage per unit
0.0206
0
0.0204
−0.5
0.0202
−1 no dmp
0.02
dmp input I
m4−11
−1.5
0.0198 0 1 2 3 4 5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
time in seconds
located in Line 4-11 and Line 4-12. The VSC in Line 4-11 is operated as the
Master and the VSC in Line 4-12 is operated as the Slave. The IPFC has 5 possible
regulators: Master capacitor Vdc regulator, Slave capacitor Vdc regulator, Master
line P regulator, Slave line Q regulator, and Slave line P regulator, denoted as R1 ,
R2 , R3 , R4 , and R5 , respectively. The control gains of these regulator are given in
Appendix B.
First, the MDI index is used to exclude inappropriate signals. Table 6.2 shows
the MDI values at the inter-area mode frequency ω = 5.87 rad/s for a list of mea-
sured local signals to each regulator when both VSCs of the IPFC are operated in
power flow control mode. In this example only local measured signals, including bus
voltages, line currents, and line flows, are considered.
Table 6.2: MDI Index Values for Measured Signals to IPFC Regulators
MDI Index
y R1 R2 R3 R4 R5
V4 80.685 44.239 89.687 71.262 91.106
P4−11 1.136 1.238 0.343 1.188 14.628
Im4−11 0.785 0.834 1.492 0.748 48.15
P4−12 1.169 1.046 32.455 1.201 0.890
Im4−12 0.597 0.543 10.371 0.623 1.327
magnitude response to a 0.1 pu step change of Master line active power reference at
time t = 0.1 seconds. The five curves represent the cases without damping control
and with damping controllers designed by using the four candidate signals. The
details of the four designed damping controllers are given in Table 6.4. We observe
that the case using measured signal Im4−11 reduced the damping effect in the first 3
circles of oscillation. So this signal is not recommended. Compared the other three
cases with damping controllers, the cases using V4 achieve better damping effect on
the bus voltage. As a result, the two candidates V4 , R3 and V4 , R5 are selected for
final consideration.
Figures 6.20 and 6.21 show the line flows and DC capacitor voltage of the
two cases using V4 , compared with the case without damping control. The case
V4 ,R3 damps the inter-area mode by injecting a damping signal to the Master line
P regulator, which results in a 6.1 MW oscillation of the Master line P in the first 2
seconds. The case V4 ,R5 damps the inter-area mode by injecting a damping signal to
the Slave line P regulator, which results in a 4.4 MW oscillation of the Slave line P in
the first 2 seconds. Considering the operation of an IPFC, the main function of the
Master VSC is to regulate the Master line flow, while the more important function
of the Slave VSC is not to regulate the Slave line flow but to provide real power
circulation to the Master VSC. Thus a smooth Master line active power response is
117
more preferable than a smooth Slave one. As a result, V4 ,R5 is recommended and
considered as the best damping signal in this example.
4
x 10 IPFC capacitor voltage Vdc
1.2025
1.202 no dmp
dmp:V4,R3
DC voltage volts
1.2015
dmp:V4,R5
1.201
1.2005
1.2
1.1995
0 1 2 3 4 5
time in seconds
signal selection. The damping controller design processes for the the STATCOM,
SSSC, UPFC, and IPFC are discussed. The designed damping controllers show
substantial improvement of the damping effect on the system inter-area mode.
However, we notice that the damping controllers built on the regulators of
FACTS controllers will cause larger oscillations in the inverter variables. Due to
high injection of the damping signal at the moment of faults or oscillations, a large
disturbance may cause the damping controllers ineffective or even cause severe extra
system oscillations.
The damping controllers built on the basis of the regulator control of FACTS
controllers are a trade-off to the system. And the damping controllers need to be
designed carefully to not cause unexpected problems.
119
7.6
7.56
7.54 no dmp
dmp:V4,R3
7.52 dmp:V ,R
4 5
7.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
slave line active power P4−12
6.13
power flow per unit
6.12
6.11
no dmp
6.1 dmp:V4,R3
dmp:V4,R5
6.09
no dmp
dmp:V4,R3
−0.55 dmp:V ,R
4 5
−0.555
−0.56
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
slave line reactive power Q4−12
0.52
0.515
power flow per unit
0.51
0.505
no dmp
dmp:V4,R3
0.5
dmp:V4,R5
0.495
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
time in seconds
Figure 6.21: IPFC Line Flows without and with Damping Controllers
CHAPTER 7
MAIN CONTRIBUTIONS AND FUTURE WORK
RECOMMENDATIONS
In this research work, we focus on the loadflow and dispatch strategies, linearized
models, dynamic simulations, and damping control design for VSC-based FACTS
Controllers in various operating modes.
2. Except for shunt voltage setpoint control mode and line power flow regulation
mode, additional reactive power setpoint control mode and reactive power
reserve mode for the shunt VSC and fixed injected voltage control mode for
the series VSC have been incorporated into the control mode implementation
3. Efficient dispatch strategies are developed to optimize power transfer when one
or both VSCs are loaded to their rated capacity, which allows one to study
maximum dispatchability of FACTS controllers in large power systems under
all the operating constraints considered.
120
121
controls and the series VSC controls are modeled as separate regulators. When
a VSC changes its operating mode, only the input signals of the corresponding
regulator need to be adjusted. The VSC operating constraints due to various
ratings and operating limits are imposed in the VSC controls. The versatile
regulator model has been incorporated into the positive-sequence transient
stability simulation program to evaluate their impact on transient stability
under oscillations and faults.
6. A new modal decomposition approach, fully decouples all state modes and
considers the interaction of other state modes to the inter-area mode of in-
terest, is proposed to to quantify levels of controllability, observability, and
inner-loop gains of the linearized models.
4. Real-time simulation
To implement the strategies in real-time, more studies on the automatic control
system are needed.
LITERATURE CITED
123
124
[37] X-P. Zhang, “Modelling of the Interline Power Flow Controller and the
Generalized Unified Power Flow Controller in Newton Power Flow,” IEE
Proceedings-Generation, Transmission and Distribution, Vol. 150, No. 3, May
2003.
[41] K. K. Sen and E. J. Stacey, “UPFC - Unified Power Flow Controller: Theory,
Modeling, and Applications,” IEEE Transactions on Power Delivery, vol. 13,
pp. 1454-1460, 1998.
[45] S. Arabi and P. Kundur, “A Versatile FACTS Device Model for Powerflow
and Stability Simulations,” IEEE Transactions on Power Systems, vol. 11,
pp. 1944-1950, 1996.
[47] K. R. Padiyar and K. U. Rao, “Modeling and Control of United Power Flow
Controller for Transient Stability,” Int. J. Electr. Power Energy Stst., vol. 21,
no. 1, pp. 1-11, Feb. 1999.
[53] H. Cai, Z. Qu, and D. Gan, “Determination of the Power Transfer Capability
of a UPFC with Consideration of the System and Equipment Constraints and
of Installation Locations,” IEE Proceedings – Generation, Transmission and
Distribution, vol.149, pp. 114-120, January 2002.
128
[71] C. Schauder and H. Mehta, “Vector Analysis and Control of Advanced Static
VAr Compensators,” IEE Proceedings on Generation, Transmission and
Distribution, vol. 140, no. 4, pp. 299-306, July 1993.
[73] J. H. Chow, X. Wei, B. Fardanesh, and A.-A. Edris, “Power Flow Controller
Responsive to Power Circulation Demand for Optimal Power Transfer,” US.
Provisional Patent application filed by Rensselaer Polytechnic Institute,
Application No.60/483694, June 2003.
[77] J. H. Chow and K. W. Cheung, “A Toolbox for Power System Dynamics and
Control Engineering Education,” IEEE Transactions on Power Systems, vol.
7, pp. 1559-1564, 1992.
[79] J.-Y. Liu, Y.-H. Song, and P. A. Mehta, “Strategies for Handling UPFC
Constraints in Steady-State Power Flow and Voltage Control,” IEEE
Transactions on Power Systems, vol. 15, pp. 566-571, 2000.
[82] K. T. Wong, “Simple Rules for the Application of Series Voltage Sources in
Power Transmission Systems,” IEEE Transactions on Power Delivery, vol. 13,
pp. 1488-1493, 1998.
131
132
17 1.0428 -59.660 0.000 0.000 26.0 1.60 0 2.7 3 0.0 0.0 345 1.5 0.5;
18 1.0429 -53.770 0.000 0.000 0.00 0.00 0 0.0 3 0.0 0.0 345 1.5 0.5;
19 1.0448 -58.000 5.000 1.000 0.00 0.00 0 0.0 2 0.0 0.0 345 1.5 0.5;
20 1.0367 -58.000 0.000 0.000 18.26 1.00 0 2.7 3 0.0 0.0 345 1.5 0.5;
21 1.0345 -54.710 0.000 0.000 0.00 0.00 0 0.0 3 0.0 0.0 345 1.5 0.5;
22 1.02953 -33.0626 0.000 0.000 0.00 0.00 0 0.0 3 0.0 0.0 345 1.5 0.5;
];
FACTS number, from bus, to bus, FACTS mode, line active power setpoint
(MW), line reactive power setpoint (MVar), bus voltage setpoint (pu), maximum
shunt current (pu), maximum active power transfer (MW), minimum bus voltage
(pu), maximum bus voltage (pu), maximum series current (pu), series reactance
135
(pu), shunt reactance (pu), owner, maximum series inverter voltage (pu), maximum
shunt inverter voltage (pu), series MVA rating, shunt MVA rating, setpoint 1, set-
point 2, series reference code: 1 for bus voltage reference 2 for line current reference,
shunt Var setpoint (pu), voltage droop, shunt mode, series mode
IPFC number, Master line from bus, Master to bus, Slave line from bus, Slave
line to bus, IPFC mode, Master line active power setpoint (MW), Master line reac-
tive power setpoint (MVar), Slave line active power setpoint (MW), Slave reactive
power setpoint (MVar), maximum Master line current (pu), maximum Slave line
current (pu), Master reactance (pu), Slave reactance (pu), maximum Master in-
verter voltage (pu), maximum Slave inverter voltage (pu), maximum active power
transfer (MW), Master MVA rating, Slave MVA rating, Master d-axis inverter volt-
age setpoint (pu), Master q-axis inverter voltage setpoint (pu), Slave d-axis inverter
voltage setpoint (pu), Slave q-axis inverter voltage setpoint (pu), Master operating
mode, Slave operating mode
Master line active power regulator Kp , Ki , and T , Master line reactive power
regulator Kp , Ki , and T , Master DC bus voltage regulator Kp , Ki , T , and Kα ,
Slave DC bus voltage regulator Kp , Ki , and T , DC capacitor voltage (Volts), DC
capacitance (µF), maximum DC capacitor voltage (Volts), minimum DC capacitor
voltage (Volts)
ipfc con = [
1 4 11 4 12 1 7.9 -0.3 7.8 0.0 18.108 18.108 0.00034 0.00034
0.056 0.056 0.5 100 100 0.02 0.0 0.0 -0.03 1 1
.1 2 .02 .1 2 .02 .1 1 .02 100 .1 2 .02 .1 1 .02
12000 2820 14400 9600;
];
• (row 1): col 1 simulation start time (s), cols 2-6 zeros, col 7 initial time step
(s)
• (row 2): col 1 fault application time (s), col 2 bus number at which fault is
applied, col 3 bus number defining far end of faulted line, col 4 zero sequence
impedance in pu on system base, col 5 negative sequence impedance in pu on
system base, col 6 type of fault, col 7 time step for fault period (s)
sw con = [
0 0 0 0 0 0 0.002; % sets initial time step
0.1 3 13 0 0 6 0.002; % no fault fault at bus 3
20 0 0 0 0 0 0; % end simulation
];
APPENDIX B
DATA FILE OF A 20-BUS POWER SYSTEM
137
138
17 1.0428 -59.66 0.000 0.000 27.0 1.60 0 2.7 3 0.0 0.0 345 1.5 0.5;
18 1.0345 -54.71 0.000 0.000 0.00 0.00 0 0.0 3 0.0 0.0 345 1.5 0.5;
19 1.0448 -58.00 12.00 1.000 0.00 0.00 0 0.0 1 0.0 0.0 345 1.5 0.5;
20 1.0367 -58.00 0.000 0.000 15.26 1.00 0 2.7 3 0.0 0.0 345 1.5 0.5;
];
FACTS number, from bus, to bus, FACTS mode, line active power setpoint
(MW), line reactive power setpoint (MVar), bus voltage setpoint (pu), maximum
shunt current (pu), maximum active power transfer (MW), minimum bus voltage
(pu), maximum bus voltage (pu), maximum series current (pu), series reactance
(pu), shunt reactance (pu), owner, maximum series inverter voltage (pu), maximum
shunt inverter voltage (pu), series MVA rating, shunt MVA rating, setpoint 1, set-
point 2, series reference code: 1 for bus voltage reference 2 for line current reference,
shunt Var setpoint (pu), voltage droop, shunt mode, series mode
IPFC number, Master line from bus, Master to bus, Slave line from bus, Slave
line to bus, IPFC mode, Master line active power setpoint (MW), Master line reac-
tive power setpoint (MVar), Slave line active power setpoint (MW), Slave reactive
power setpoint (MVar), maximum Master line current (pu), maximum Slave line
current (pu), Master reactance (pu), Slave reactance (pu), maximum Master in-
verter voltage (pu), maximum Slave inverter voltage (pu), maximum active power
transfer (MW), Master MVA rating, Slave MVA rating, Master d-axis inverter volt-
age setpoint (pu), Master q-axis inverter voltage setpoint (pu), Slave d-axis inverter
voltage setpoint (pu), Slave q-axis inverter voltage setpoint (pu), Master operating
mode, Slave operating mode
Master line active power regulator Kp , Ki , and T , Master line reactive power
regulator Kp , Ki , and T , Master DC bus voltage regulator Kp , Ki , T , and Kα ,
Slave DC bus voltage regulator Kp , Ki , and T , DC capacitor voltage (Volts), DC
capacitance (µF), maximum DC capacitor voltage (Volts), minimum DC capacitor
voltage (Volts)
ipfc con = [
1 4 11 4 12 1 7.5 -0.558 6.123 0.0 18.108 18.108 0.00034 0.00034
142
• (row 1): col 1 simulation start time (s), cols 2-6 zeros, col 7 initial time step
(s)
• (row 2): col 1 fault application time (s), col 2 bus number at which fault is
applied, col 3 bus number defining far end of faulted line, col 4 zero sequence
impedance in pu on system base, col 5 negative sequence impedance in pu on
system base, col 6 type of fault, col 7 time step for fault period (s)
sw con = [
0 0 0 0 0 0 0.002; % sets initial time step
0.1 14 16 0 0 6 0.002; % no fault fault at bus 14
10 0 0 0 0 0 0; % end simulation
];