Ajjarapu Reactive Power Pserc
Ajjarapu Reactive Power Pserc
Ajjarapu Reactive Power Pserc
Project Team
Venkataramana Ajjarapu, Project Leader, Iowa State University
A.P. Sakis Meliopoulos, Georgia Institute of Technology
October 2008
Acknowledgements
This is the final report for the Power Systems Engineering Research Center (PSERC)
research project S-28 titled Optimal Reactive Power Control against Voltage Collapse
Incorporating Protection System. We express our appreciation for the support provided
by PSERCs industrial members and by the National Science Foundations Industry /
University Cooperative Research Center program under the grants NSF EEC-0002917 at
Iowa State University and NSF EEC-0080012 at Georgia Institute of Technology.
The authors wish to recognize their postdoctoral researchers and graduate students that
contributed to the research and creation of the reports:
Georgia Institute of Technology
George Stefopoulos, GRA
Dr. George Cokkinides
The authors thank all PSERC members for their technical advice on the project,
especially the industry advisors for the project:
Executive Summary
In recent years, new attention has been given to system disturbances that have cascaded
due to voltage instabilities and to unwanted relay operations. Unwanted relay operations
due to voltage instabilities and transients have not been well understood. In this project,
voltage instability phenomena were studied to develop a comprehensive approach for
mitigating the effects of voltage transients and instabilities on designed operation of a
protection system. The comprehensive framework covers monitoring, predicting, and
assessing system performance for secure power system operation. The developed
framework and methodology provide advanced tools use optimal control strategies that
can be used to avoid voltage collapse with respect to system-wide voltage instability and
undesired protection system operations. The projects specific objectives were:
to develop realistic models that accurately model voltage dynamics and their effects
on protective schemes
to develop fast and flexible schemes for assessment of voltage stability and relay
status
and corrector based VSM estimation was very fast and accurate. Based on the relay
margin of each relay, critical relays were identified in the test system under several
different operating conditions. By performing the optimal reactive power control after
contingencies, the system voltage profile, the voltage stability margin at load buses, and
the relay margins were improved to insure that system operating criteria were met after
any of the contingencies.
Our research provides a proof of concept of the proposed framework. In future work we
can test the developed tools on a large utility system. With the above framework, we used
existing control sources to maintain relevant margins. In future work, we can identify
where and how much supply of additional control sources would be needed when the
existing control sources are not adequate to meet system operating criteria.
2. Incorporating Relays in a Power System Model with Dynamic Loads for Voltage
Transient and Stability Analyses (Volume II)
Relay operation due to transient voltage phenomena is unneeded if the voltage transient is
stable and the system will eventually return to normal voltages. If relay operation did
happen in these cases it will deteriorate system conditions and may result in a cascading
series of relay operations and eventual system collapse. Therefore relay operation must
be inhibited if the voltage transient represents a stable event.
In the second step of this project, we studied the impact of voltage transients and voltage
instabilities with an integrated power system simulation model that explicitly represents
the load dynamics (mainly motor loads) and the dynamics of reactive power sources.
Examples of reactive power sources include generators, over and under exciter limiters,
and static VAR sources. The power system model was augmented with relay models.
Two specific relays were modeled: (a) overcurrent relays and (b) distance relays.
Two test systems were developed to demonstrate use the augmented simulation model for
analysis of voltage instabilities and protective systems. The results illustrated that voltage
instabilities during recovery from disturbances can cause excessive current flows that
may affect the operation of overcurrent relays and of the impedance seen by distance
relays. Distance relays are especially vulnerable to mis-operation because these
phenomena exhibit simultaneously low voltages and high currents creating the possibility
of load encroachment. The level of these phenomena is dependent upon the specific
circuit parameters, and the type and amount of dynamic loads. As a result, it is difficult to
develop general guidelines for predicting the level and impact of these phenomena.
Based on these results, we conclude that the proper way to apply the proposed
methodology is to study specific systems that are heavily loaded with motor type loads.
The prototype power system simulation model has the capability to model a limited
number of relays since the objective of the project was to simply demonstrate the
feasibility of the approach. To achieve the capability of modeling and comprehensively
studying the response of a system, we recommend that the developed methodology be
augmented with a full set of relays and further developed into a commercial grade
computer program.
iii
Volume 1
Project Team
Venkataramana Ajjarapu, Project Leader, Iowa State University
Table of Contents
1. Introduction................................................................................................................... 1
1.1 Background .......................................................................................................... 1
1.2 Motivation ............................................................................................................ 2
1.3 Present Status of the Knowledge.......................................................................... 3
1.4 Power System Protection: A Review ................................................................... 4
1.5 Practical Criteria for Relay Margin and Voltage Stability ................................... 5
1.5.1 Dynamic Security Assessment Criteria......................................................... 5
1.5.2 Relay Margin Criteria for Post-Contingency Stability.................................. 5
1.5.3 Power Voltage (P-V) Curves...................................................................... 5
1.6 Power System Voltage Stability: A Review......................................................... 7
1.7 Overview of Voltage Stability with respect to Maximum Loading ..................... 8
1.8 Report Organization ............................................................................................. 9
2. Theoretical Methodology............................................................................................ 10
2.1 Relay Status Evaluation using Relay Margin..................................................... 10
2.1.1 Relay Margin and Relay Staying Time ....................................................... 10
2.1.2 Relay Margin Ratio and Relay Staying Time Ratio.................................... 13
2.1.3 Identification of Critical Relays .................................................................. 14
2.1.4 Critical Relay Identifications by the Power Flow and Time Domain
Simulation ................................................................................................... 15
2.2 Proposed Predictor-Corrector Framework for VSM Calculation....................... 16
2.2.1 Introduction ................................................................................................. 16
2.2.2 Overview of VSM Calculation.................................................................... 17
2.2.3 Problem Formulation................................................................................... 17
2.2.4 VSM Predictor............................................................................................. 18
2.2.5 VSM Corrector ............................................................................................ 20
2.2.6 Identification of VSM by Predictor and Corrector...................................... 22
3. Optimal Control Strategy against Voltage Instability................................................. 23
3.1 Introduction ........................................................................................................ 23
3.2 Overview ............................................................................................................ 23
3.3 Formulation of the Optimal Control Strategy .................................................... 23
3.3.1 Control Objective ........................................................................................ 23
3.3.2 System Constraints ...................................................................................... 23
3.3.2.1 Load Voltage Limit ..................................................................................... 23
3.3.2.2 Relay Margin Limit ..................................................................................... 24
3.3.2.3 Voltage Stability Margin Limit ................................................................... 24
3.3.2.4 On Load Tap Changer (OLTC) Limit ......................................................... 24
3.3.2.5 Shunt Capacitor Limit ................................................................................. 25
3.3.3 Mathematical Formulation of the Optimization.......................................... 25
3.4 Overall Framework of Optimal Control Strategy............................................... 27
4. Numerical Results....................................................................................................... 28
4.1 Numerical Results for Identification of Critical Relays ..................................... 28
4.1.1 23-bus Test System Description.................................................................. 28
4.1.2 Numerical Results ....................................................................................... 29
4.1.3 Conclusions ................................................................................................. 31
ii
iii
List of Figures
Figure 1 Voltage collapse sequence by additional tripping by protective relays ............... 3
Figure 2 Typical P-V curve for voltage stability limit........................................................ 6
Figure 3 Transmission line with mho relays..................................................................... 10
Figure 4 Illustration of relay margin of mho relay with = 0 .......................................... 12
Figure 5 Illustration of relay staying time of mho relay with = 0 .................................. 13
Figure 6 Flowchart of identifying critical relays by RMR and RSTR.............................. 15
Figure 7 Critical relay identification by power flow and time domain simulation.......... 16
Figure 8 Idea of predictor-corrector based VSM identification........................................ 18
Figure 9 Load bus and Thevenin equivalent system......................................................... 19
Figure 10 Flowchart of VSM correction by the corrector ............................................... 21
Figure 11 Identification of VSM by predictor and corrector............................................ 22
Figure 12 Framework of optimal control strategy against voltage collapse .................... 27
Figure 13 One-line diagram of the 23-bus test system ..................................................... 28
Figure 14 Impedance swing trajectory of relay 153-154_2 for contingency 1................. 30
Figure 15 Relay margin trajectory of relay 203-154 for contingency 1 ........................... 31
Figure 16 New England 39-bus system ........................................................................... 32
Figure 17 Predictor-corrector based VSM identification with contingency..................... 34
Figure 18 Predictor-corrector based VSM identification with load increase ................... 35
Figure 19 System voltage profile before and after optimal control (2-25 outage case) ... 36
Figure 20 Voltage stability margin before and after optimal control (2-25 outage case). 38
Figure 21 Relay margin before and after optimal control (2-25 outage case) .................. 38
Figure 22 System voltage profile before and after optimal control (26-29 outage case) . 39
Figure 23 Voltage stability margin before and after the optimal control (26-29 outage
case) .................................................................................................................................. 40
Figure 24 Relay margin before and after optimal control (26-29 outage case) ................ 40
iv
List of Tables
Table 1 Relay ranking for contingency 1 in 23-bus test system ....................................... 29
Table 2 Relay ranking for contingency 2 in 23-bus test system ....................................... 29
Table 3 Relay ranking for contingency 3 in 23-bus test system ....................................... 30
Table 4 Numerical results for the VSM calculation ......................................................... 33
Table 5 Optimal reactive power control variable settings ................................................ 37
1. Introduction
1.1 Background
The primary motivation for this project arises from the critical need to provide fast
calculation of voltage stability margin (VSM) and quick identification of critical relays
using real-time measurements. To achieve this objective one needs analytical tools to
enhance power system voltage stability incorporating protection systems.
Wide area blackouts in power systems are considered as system-wide events caused
by combination of diverse events such as severe system faults, excessive load demand,
and human/machine errors. In the 1996 WSCC blackout voltage collapse played a major
role. In the 2003 Northeast blackout even though it is not directly related to voltage
collapse, reactive power played an important role in the blackout. According to 2003 USCanada outage task force, reactive power supplies in the Northeast were exhausted but
the need for reactive power continued to rise as peak load increased. The conclusion is
that proper reactive power management would have helped to prevent the initial system
events and therefore would have delayed or possibly might even have prevented the
resulting blackout. In the experience of most system blackouts, the initiating event was a
protection device operation that played a very important role in triggering cascading
events and finally wide-area voltage collapse. This direct impact of protection system on
the phenomenon suggests that integration of protection systems into a unified framework
for control determination against voltage collapse.
This project proposes a comprehensive framework to monitor, predict, and assess
the system performance for secure power system operation, considering reactive power
aspects and protection schemes applied in power systems. The proposal provides
advanced tools that can be used to avoid voltage collapse with respect to system-wide
voltage instability as well as undesirable protective operation. In addition, the proposal
provides a framework for fast, flexible and reliable calculation of voltage stability
margin, which are composed of predictor, predicting the VSM directly by using Thevenin
Equivalent method and corrector, correcting the VSM to be close to the true maximum
loadability point. Particularly, the project first applied the concept of relay margin to
evaluate relay status and identify critical relays with different system contingencies. Then
the project develops a novel predictor and corrector based framework to calculate voltage
stability margin quickly, flexibly and accurately. Finally we propose an optimal reactive
power control scheme which incorporates the constraints of voltage stability margin and
relay margin to maintain voltage stability. Since the system protection scheme is
integrated in the system model, the method proposed here can also identify adequacy or
deficiency of control resources in the system.
1.2 Motivation
The security of a bulk power system is threatened when it is loaded near to its
maximum capacity. Voltage instability and undesirable protective relay operation 1 are
two major interrelated phenomena that occur when the system is under stress. As reported
in many voltage collapse incidents [1-3], when the system experiences excessive voltage
drop after one or more severe faults, high loading conditions on system components tend
to operate protective relays and to further trip the corresponding components, leading to
spreading of cascading events of tripping. The lack of reactive power during heavy
loading conditions may trigger field limiters and overload protection to trip the
generators. Undoubtedly, this directly contributes to wide-area blackouts.
Even though it is not hard to find interrelation between voltage collapse and system
component protection, however, there are very few references to come up with
countermeasure both considering system wide voltage collapse as well as undesirable
protective operation. Generally, control determination for voltage stability enhancement
first detects the margin boundary with the given direction of load increase and then
calculates control strategies by applying the optimal reactive power control framework.
However, the control strategies may fail when another severe tripping due to protective
operation occurs. Thus, consideration of protection actions is required when deciding
control strategies against voltage collapse.
One of the main objectives in this project is to determine control strategies for
preventing further tripping events resulting from unwanted protective actions that make
systems more vulnerable in terms of voltage stability. Fig. 1 illustrates the sequence of
the events that are mainly concerned in this project. In the normal state, the system is
operated at the point a. After N-1 (or N-k) contingency , if the system is transiently
stable, short-term dynamics settles down to the point b, and because of load recovery
dynamics, the equilibrium of the short-term dynamics moves along the P-V curve of N-1
case. If the long-term load characteristic is constant power as shown in Fig. 1,
equilibrium point of total system dynamics will reach the point c and settles down.
However, during the transition from the point b to c, if one of the protective relays of
main transmission facilities violates its normal operational limits or seriously exceeds its
own rating, another event of tripping happens. If the trip of the transmission facility is in
a set of severe contingencies, the systems may be in great danger, and it may lead to
cascading events resulting in voltage collapse.
Undesired relay operation is here defined by action of protective relay that happens without any
faults in the systems and causes inappropriate trippings leading to aggravate situation during voltage
instability.
Inertial stability criteria. This concerns mainly the evolution of relative machine
angles and frequencies.
Voltage excursions (dip or rise) beyond specified threshold level and duration.
This includes separate voltage excursion threshold/duration pairs for voltage dip
and voltage rise, and maximum/minimum instantaneous excursion threshold.
Relay margin criteria. These are defined for pre-disturbance and post disturbance
conditions. If relay margin is violated for more than a maximum specified time
after the disturbance, it is identified as insecure.
programs can be used to generate a P-V curve. In certain situations it may be desirable to
manually generate a P-V curve to take into account specific remedies available.
A sample P-V curve is shown below. The critical point of the curve, or voltage
instability point, is the point where the slope of the P-V curve is vertical. As illustrated,
the maximum acceptable pre-contingency power transfer must be the lesser of:
a pre-contingency power transfer (point a) that is 10% lower than the voltage
instability point of the pre-contingency P-V curve, and
The P-V curve is dependent on the power factor. Care must be taken that the worst
case P-V curve is used to identify the stability limit.
The time frame of interest for voltage stability problems may vary from a few
seconds to tens of minutes. Therefore, voltage stability may be either a short-term or a
long-term phenomenon.
Long-term voltage stability involves slower acting equipment such as tapchanging transformers, thermostatically controlled loads, and generator current
limiters. The study period of interest may extend to several or many minutes, and
long-term simulations are required for analysis of system dynamic performance
Security of a power system refers to the degree of risk in its ability to survive
imminent disturbances (contingencies) without interruption of customer service. It relates
to robustness of the system to imminent disturbances and, hence, depends on the system
operating condition as well as the contingent probability of disturbances. System security
may be further distinguished from stability in terms of the resulting consequences. For
example, two systems may both be stable with equal stability margins, but one may be
relatively more secure because the consequences of instability are less severe.
The analysis of security relates to the determination of the robustness of the power
system relative to imminent disturbances. There are two important components of
security analysis. For a power system subjected to changes (small or large), it is
important that when the changes are completed, the system settles to new operating
conditions such that no physical constraints are violated. This implies that, in addition to
the next operating conditions being acceptable, the system must survive the transition to
these conditions. Hence, there are two types of analysis related to security:
7
2. Theoretical Methodology
2.1 Relay Status Evaluation using Relay Margin
Power system protection at the transmission system level is based on distance
relays. Distance relays are applied for both apparatus protection and system protection.
Significant power flow oscillations can occur on a transmission line or a network due to
major disturbances like faults and subsequent clearing. Relay margin related information
can be used to evaluate relay status and identify critical relays in the transmission system.
2.1.1 Relay Margin and Relay Staying Time
Traditionally, the impedance based relay margins are mainly used to quantify the
closeness of a system trajectory to a relay zone. However, relay margin can also be
formulated as a function of the bus voltage instead of the line impedance. This way we
can detect the effectiveness of controls applied to power systems for preventing possible
undesired relay tripping more efficiently. The relay status during system disturbances can
also be quantified based on the information of relay margin ratio and relay staying time
ratio.
Figure 3 is used to illustrate the voltage based relay margin. We consider that the
offset coefficient of mho relay is usually close to 0, so the 3-zone characteristic of mho
relay can be shown in Figures 4 and 5.
Vi
Zij
Sij
Bus i
Vj
Bus j
Relay
Relay
Z ij ( v i , v j , i , j ) =
vi i
vi i v j j
(1)
R ij + jX ij
where vi and i are the voltage magnitude and angle at bus i, and Rij and X ij are the
transmission line resistance and reactance.
From the design logic of distance relays, we know that the tripping of a distance
relay depends on: whether the system trajectory enters the relay tripping zone, which
10
zone of the three zones is entered and how long the trajectory stays in one particular
tripping zone. The relay at bus i along the line between buses i and j will operate when
Z ij
where
(2)
( Rij + jX ij )
(3)
is the center of the circle corresponding to the mho relay characteristic in Figure 4.
According to the design scheme of 3-zone distance relay, each protection zone is
designed to protect a different range of the transmission line which is referred as in
equation (3). From the relay setting, we know that usually =0.8 is set for zone 1 and
=1.2 for zone 2. Zone 3 setting depends on various factors and for zone 3 is usually
much higher than zone 2. When the zone 1 of a distance relay in the transmission system
is entered during a particular disturbance, the corresponding relay will trip the line
instantaneously. If the system trajectory enters the other two zones, it must stay in the
zones longer than the pre-set time before the relay tripping operation is activated.
However, the transmission system is considered as vulnerable any time when the three
zones are entered by system trajectory.
In order to evaluate the relay status during system disturbance, the concept of relay
margin is proposed in [8] and it can be applied to evaluate whether some relays in the
system are going to initiate the tripping operations. According to the original definition of
relay margin in [8], we make some adjustments about the variable. We define the relay
margin (RM) as the distance of system trajectory to the zone 3 of distance relay which
can be seen directly by the red line in Figure 4. The mathematical formulation of RM can
be expressed by
RM = Z ij
RM = Z ij
RM =
( Rij + jX ij )
(4)
( Rij + jX ij )
vii
11
(5)
(6)
RM (vi , v j , i , j , Rij , X ij , ) 0
(7)
When the system trajectory enters the three zones of distance relay during a system
disturbance, we also need to know whether staying time of the trajectory in one particular
zone exceeds the preset time for relay tripping. Therefore, we define the relay staying
time (RST) as the maximum staying time of a trajectory within one particular zone of
relay. Figure 3 shows the relay staying time directly.
The formulation of RST is defined as
(8)
12
RMR (c, r ) =
min RM (c, r , t )
t
RM 0
(9)
It will be referred to as RMR which samples the smallest relay margin over the time
span (t=0,1, nt ) divided by the relay margin of pre-contingency state ( RM 0 ) for each
relay. The proposed RMR provides a relative value of the current relay margin compared
with the pre-contingency relay margin. Actually, by applying the RMR we have given a
reference to the current relay margin compared with the pre-contingency state. The
threshold of RMR is less than or equal to 1 and greater than or equal to 0. If the RMR
value is less than 0, it means that the relay tripping zone is entered by system trajectory
and instead of using RMR, the following proposed RSTR will be applied to evaluate
relays.
Similarly to the definition of RMR, we also propose a scalar for evaluating the
system trajectory staying time within the relay tripping zones which is relay staying time
ratio (RSTR). The RSTR is defined as,
13
RSTR ( c , r , t ) =
RST ( c , r , t )
T set
(10)
where Tset is the relay pre-set time which is the maximum time for a system trajectory
staying within tripping zones before the relay operate. If the value of RSTR is greater
than 1, then the relay will initiate the tripping signal.
2.1.3 Identification of Critical Relays
After a contingency, if the tripping zones of some relays are entered, RSTR will be
calculated and applied to rank the corresponding relays in a descending order to obtain
the set of the most vulnerable relays (MVR). Otherwise, relays will be ranked by RMR in
an ascending order to obtain the set of the potentially most vulnerable relays (PMVR).
When the system operation condition is changed, some MVRs and PMVRs may change.
The above procedure will solve the problem of identification of critical relays. Therefore,
we only need to simulate the relay functionality of those critical relays for system
stability analysis. Figure 6 shows the procedure of identifying critical relays by RMR and
RSTR.
14
2.1.4 Critical Relay Identifications by the Power Flow and Time Domain
Simulation
The critical relay identification results by power flow and time domain simulation
are different. The details are published in our first of paper of our publication references.
As shown in the Figure 7, the relay margin calculation based on the power flow
basically calculates the relay margin at the two points (a and b) on the impedance
trajectory which is obtained by the time domain simulation. Point a refers to the precontingency state and point b refers to the post-contingency state. Therefore, the power
flow based relay margin calculation is just a snapshot of the time domain simulation at
one time instant.
The time domain simulation based relay margin calculation can capture all the relay
margin values at all the time instants in the time interval. For the identification of critical
15
relay based on time domain simulation, the minimum relay margin in the interested time
interval is picked up for each relay and used to rank them.
Therefore, the time domain simulation based critical relay identification is able to
consider the transient period which cannot be done by power flow based analysis. In the
optimal control part of this project, we apply the power flow based relay margin
calculation. In the next phase of work, we plan to formulate a dynamic optimization
model to incorporate the time domain simulation based relay margin as a constraint.
Figure 7 Critical relay identification by power flow and time domain simulation
16
f (X, S) = f (X) S =0
(11)
S = Pmax + jQmax
(12)
(13)
Qmax = pf Pmax
(14)
VSMP = Pmax PL
(15)
where S is the nodal power injection vector, Pmax is the nodal real power injection vector
, Qmax is the nodal reactive power injection vector after the prediction and correction. pf
is the load power factor vector, Ppredictor is the nodal real power injection vector after the
prediction, Pcorrector is the nodal real power injection vector after the correction, PL is the
present load vector and VSMP is the voltage stability margin vector in term of real power.
17
Initially, Ppredictor is derived by the VSM predictor to guess maximum power point.
Then, Pcorrector uses this initial guess to converge to the actual maximum point Pmax . Fig.
8 shows the basic idea about the predictor-corrector based VSM identification.
18
v
v
v
v
v
ETh VL v
VL (ETh VL )
P + jQ
v
v
= I L = ( v ) P + jQ =
ZTh
VL
ZTh
v
v
v
Maximal Power Transfer VL = ( ETh VL )
Ppredictor + jQ predictor
v
(V L ) 2
= v
Z Th
(16)
(17)
(18)
For the tracking of Thevenin equivalent, equations (19) and (20) are used,
v
v
v
v
E Th = V L + Z Th * I L
1
0
0
1
Er
d Ei
-c R Th
X Th
-c
-d
= a
b
(19)
(20)
where ETh = Er + jEi , V = a + jb, I = c + jdZTh = RTh + jXTh . Two or more measurements
taken
at
different
times
are
required
to
solve
for
unknown
parameters: E r , E i , R Th , X Th . For the online tracking of Thevenin equivalent
19
parameters, there are several parameters identification methods can be used, such as least
square method, Kalman Filter method etc.
Since there are several system measurement devices available, it opens up an
opportunity to make use of these measurement data to update the system status and
monitor the system stability on line. Given the measurement data from the system load
bus, the VSM predictor can quickly predict the voltage stability margin at load buses.
This ensures that when the new measurement data is arrived, the last calculation of VSM
has already been finished and the new measurements can be continuously used to update
the VSM value at local load buses. Being able to apply the real time measurements to
predict the VSM on line has several advantages compared with the traditional used
offline study method, such as continuation power flow method. One obvious advantage is
that, the assumption of load increase pattern is not needed as the load increase
information can be obtained and updated by the real time measurements. Therefore, this
VSM predictor makes the digital computer calculation based power system analysis
results closer to the real power system situations.
2.2.5 VSM Corrector
Since VSM Predictor using Thevenin equivalent which is only a linear
approximation (it can not consider the system constraints, such as, thermal limits of
transmission lines, requirements of maintaining bus voltages within certain ranges, and
limits of generator reactive power output limits) the VSM values at local load buses are
over estimated. The VSM corrector is proposed to correct the VSM by predictor and
provide a better estimate for VSM. By taking power flow convergence as an indicator,
the VSM corrector sets Ppredictor obtained by the VSM predictor and PL of the present
load as the upper and lower bounds for the search space of Pmax . Binary search algorithm,
which is fast and robust, is used to update Pcorrector and search for Pmax which draws the
system back onto the feasibility boundary. Fig. 10 shows the flowchart of VSM
correction by the corrector.
Since the corrector involves power flow convergence evaluation, it is able to
incorporate transmission line thermal limits, bus voltage limits, and generator reactive
power limits in the process of VSM correction. The power flow convergence evaluation
requires that the updated power flow data being available from the system state estimator.
During the process of VSM correction, the forecasted load for the next 24 hour or
possible contingencies can also be incorporated to calculate the VSM. Thus several
scenarios which the system may face in the next day can be considered to correct the
VSM obtained by the predictor. Therefore, the corrector makes the VSM calculation
being able to incorporate system constraints for online operation and consider possible
system scenarios for the next day planning.
Overall, the predictor-corrector based framework is able to provide fast, flexible
and reliable VSM calculation using the system real time measurements from local load
buses. The numerical section will provide the test results of the proposed method for
VSM calculation. It validates that the predictor- corrector based framework is suitable for
online calculation of VSM. The following section provides the procedure of using the
20
21
22
Vi
min
Vi Vi
max
(21)
where, Vi min and Vi max are the allowable minimum and maximum voltages, respectively.
23
c min c
(22)
where c represents the relay margin at a particular operating point and c min is the
required margin. The formulation of relay margin can be obtained from the previous
section and only voltage phasor measurements are needed to calculate the relay margin.
3.3.2.3 Voltage Stability Margin Limit
Here the voltage stability margin is given as:
c min c
(23)
where c represents the voltage stability margin for a given operating condition, and
c min is required margin. The formulation of voltage stability margin and the proposed
method to derive the margin can be obtained from the previous section and voltage and
current phasor measurements are needed to calculate the voltage stability margin.
3.3.2.4 On Load Tap Changer (OLTC) Limit
Most of the power system loads are voltage dependent and the system voltage drop
will cause a reduction in the load demand which may release the stress on the system.
However, OLTCs could act within tens of seconds after contingencies to bring the load
voltages back to their rated values, which consequently would cause further stress on the
system.
The tap ratio of OLTC can be used to control the reactive power and voltage profile
of the system,
Ti min Ti Ti max
(24)
where Ti is the tap ratio of the OLTC which is one of the three control variables in this
optimal reactive power control model.
24
(25)
where Qci is the reactive power output from the shunt capacitor installed in the system
which is also one of the three control variables in this optimal reactive power control
model.
3.3.3 Mathematical Formulation of the Optimization
The goal of optimal reactive power dispatch is achieved by proper adjustment of
reactive power control variables which includes generator terminal voltage magnitudes
(VGi), transformer tap settings (Ti), reactive power generation of the capacitor bank
(Qci).
The following optimization model is applied to obtain the optimal reactive power
dispatch in the power system. The voltage stability margin is incorporated as one of the
constraints. The relay margin is also included,
0 = g ( x, y, u )
Vi
Fj
min
Vi Vi
max
max
Fj Fj
max
Ti min Ti Ti max
c min c
c min c
where g represents power balance equations; x is the vector of state variables; y is the
vector of algebraic variables; u is the vector of control variables; PGi is the generator real
power output; QGi is the generator reactive power output; PLoss is the real power loss in the
network; F represents line flows; Vi is the load bus voltage; VG is generator terminal
voltage; Qci is reactive power output of shunt capacitors; T is OLTC tap ratio; is
voltage stability margin (in term of P); is relay margin. In this optimal reactive power
control scheme, the control variable vector u is made up by Qci , T, and VG .
26
27
4. Numerical Results
4.1 Numerical Results for Identification of Critical Relays
To verify the suitability of the proposed method for identification of critical relays,
a 23-bus test system is used.
4.1.1 23-bus Test System Description
28
RSTR
153-154_2
154-153_2
153-154_1
154-153_1
3008-154
154-3008
154-203
203-154
0.1079
0.1079
Relay
ranking
1
8
2
7
4
6
5
3
RSTR
153-154_2
154-153_2
153-154_1
154-153_1
3008-154
154-3008
205-154
154-205
154-203
203-154
0.0083
0.0083
0.0083
29
Relay
ranking
1
8
2
9
3
6
4
5
10
7
0.18
0.16
Reactance (p.u.)
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
-0.02
-0.1
-0.05
0.05
0.1
Resistance (p.u.)
0.15
0.2
0.25
30
0.35
0.3
0.25
0.2
0.15
0.1
0.05
3
Time (s)
Based on RMR values, we can identify the relays with smaller relay margin ratios
over the interested time span. Based on RSTR values, we can identify the relays which
have a longer staying time within the tripping zones. From Table 1 and Table 2, we can
see that the RMR based and relay margin based relay rankings are almost the same.
However, from Table 3, we can see that the two variables based relay rankings are
different. Therefore, we can have the conclusion that a smaller relay margin does not
necessarily mean that the corresponding relay have a smaller relay margin ratio. Actually,
from the single value of relay margin, we can only tell the distance of a trajectory to the
tripping zone at each time instant. However, we do not know the rate of relay margin
change. Figure 14 shows the impedance swing trajectory of the top 1 critical relay-153154_2 in the relay ranking list for contingency 1. We can see that after the tripping of line
154-205 caused by the three phase fault, the system trajectory stayed in the third tipping
zone of relay 153-154_2 for 10.79% of the preset time. Figure 15 shows the relay margin
trajectory of relay 203-154 for contingency 1. We can see that the post-contingency relay
margin is 0.0604 p.u. and its pre-contingency relay margin is 0.2853, thus the RMR is
0.2117. This satisfies the relay margin criteria for post-contingency [11].
4.1.3 Conclusions
Since power system protection at the transmission system level is based on distance
relays which are sensitive to the power flow variations on a transmission line, it is
important to evaluate the relay status during transmission system disturbances and
identify critical relays for system stability study.
Based on the practical dynamic system security criteria for relay margin, the
concept of voltage based relay margin is applied to redefine the two variables: relay
margin ratio and relay staying time ratio. These two variables are calculated and relays
are ranked accordingly. The critical relays are identified in the 23-bus test system with
three different contingencies.
31
iteration times and calculation time needed for VSM identifications are reduced in
general by about 75 percent using the predictor-corrector method.
Table 4 Numerical results for the VSM calculation
Base Case
Contingency
Line 2-25 outage
Contingency
Line 26-29
outage
Parameter
Pre-Cor method
CPF method
Pmax (p.u.)
21.6674
21.7956
No. of Iterations
19
0.4536
1.7338
Pmax (p.u.)
20.0379
21.0324
No. of Iterations
22
0.3438
2.4216
Pmax (p.u.)
20.3156
21.3069
No. of Iterations
20
0.3750
1.5902
Fig. 17 and Fig. 18 plot the predictor-corrector based VSM identification results.
The plots show the VSM of load bus 26 when the load is increased from initial value to
the voltage collapse point where power flow cannot converge.
Fig. 17 incorporates the outage of line 26-29 in the system. Fig. 18 considers the
scenario of all the other loads increased by 100 percent. Without considering any
contingencies or load increase, the maximum loading point for load bus 26 is at the
intersection point of predictor curve and load curve which indicates the predictor is
accurate at the collapse point. However, when the load is far away from the collapse
point, there is a relatively large error between the predictor based VSM and the true
VSM.
After applying the corrector to correct the VSM by incorporating system
constraints, the corrected VSM shows a better estimation of the true VSM value.
However, the corrected VSM still shows some errors compared with the true values.
Fig. 17 and Fig. 18 also indicate that different system operating scenarios may
affect the VSM as expected. Therefore, it is necessary to incorporate system constraints
and consider possible system scenarios for VSM identification. Through applying the
VSM corrector, system constraints can be incorporated and possible system scenarios can
be considered which make the VSM calculation more reliable and flexible.
33
45
P-Predictor
P-Corrector
P-Corrector-26-29 outage
PL
40
35
Per Unit
30
25
20
15
10
5
0
10
15
Incremental Load (Per Unit)
20
25
34
45
P-Predictor
P-Corrector
P-Corrector- +100% load
PL
40
35
Per Unit
30
25
20
15
10
5
0
10
15
Incremental Load (Per Unit)
20
25
4.2.3 Conclusions
This new VSM calculation scheme makes three contributions:
First, it formulates the VSM identification problem in a predictor-corrector
framework.
Second, it provides a method to incorporate system constraints and consider
possible system scenarios during VSM calculation which makes it more reliable and
flexible.
Third, it shows that the new method is able to obtain the VSM with a good accuracy
with reduced computational effort.
Generally, this scheme provides a new method for VSM identification which is fast,
reliable and flexible.
35
Figure 19 System voltage profile before and after optimal control (2-25 outage case)
36
Optimal
settings
Optimal
settings
Control
Variables
Variable
Range
VG1
0.95 1.05
1.0139
1.0336
0.9737
1.0462
VG2
0.95 1.05
1.0369
1.0226
1.0357
1.0500
VG3
0.95 1.05
1.0331
1.0377
1.0459
1.0500
VG4
0.95 1.05
1.0316
1.0175
0.9980
1.0492
VG5
0.95 1.05
1.0247
0.9892
1.0438
1.0500
VG6
0.95 1.05
1.0368
1.0494
1.0335
1.0500
VG7
0.95 1.05
0.9923
1.0012
1.0431
1.0500
VG8
0.95 1.05
1.0341
1.0289
0.9852
1.0500
VG9
0.95 1.05
1.0131
1.0329
0.9944
1.0500
VG10
0.95 1.05
0.9958
1.0409
0.9796
1.0500
Qc4
0 1. 5
1.3000
1.5000
1.0000
1.5000
Qc8
0 1. 5
1.4000
0.8000
1.5000
1.5000
Qc16
0 1. 5
0.5000
1.1000
1.1000
1.5000
Qc20
0 1. 5
0.1000
0.6000
0.9000
1.4000
T7
0.9 1.1
1.0000
0.9600
0.9800
1.0200
T8
0.9 1.1
1.0600
0.9600
1.0600
1.0400
T9
0.9 1.1
1.0000
0.9600
0.9000
0.9800
T15
0.9 1.1
1.0000
1.0400
1.0200
0.9600
37
Figure 20 shows that after the optimal reactive power control, voltage stability
margin of every load bus is increased. By existing control sources, if the VSM
requirement still cannot be achieved, other control methods are needed, such as load
shedding.
Figure 20 Voltage stability margin before and after optimal control (2-25 outage case)
Figure 21 Relay margin before and after optimal control (2-25 outage case)
38
Figure 21 shows the relay margin before and after optimal control which indicates
relay margins have been increased by the optimal control after the contingency. By the
relay margin criterion, relay margin at bus 4 was improved by the optimal control to meet
the requirement.
The following three figures show the voltage profile, voltage stability margin and
relay margin before and after the optimal control which are used to further validate the
effects of optimal control against voltage collapse.
Figure 22 System voltage profile before and after optimal control (26-29 outage case)
39
Figure 23 Voltage stability margin before and after the optimal control (26-29 outage
case)
Figure 24 Relay margin before and after optimal control (26-29 outage case)
40
4.3.2 Conclusions
Voltage stability margin and relay margin are incorporated as constraints in the
optimization model. Optimal reactive power control is proposed to maintain voltage
stability and enhance relay margin. The optimal control strategy uses existing control
sources to improve the system voltage stability.
The numerical results from this section show that the optimal reactive power control
can effectively improve the system voltage profile after contingencies. Also, the voltage
stability margin and relay margin have been improved by adjusting the control variables
to the optimal settings. Two test cases with two contingencies are applied in the test
system which validate that the optimal reactive power control strategy is effective to
prevent voltage collapse.
4.3.3 Future Work
The power flow based relay margin calculation and critical relay identification have
been used in the process of solving the optimal reactive power control problem.
According to the relay margin criteria for post-contingency stability, the relay margin
which violates the criteria is incorporated in the optimization model as a constraint.
Through the optimal reactive power control, the relay margin has been increased after
contingencies to meet the requirement of the post-contingency stability criteria.
In the next phase of research work, we will formulate a dynamic optimization
model which incorporates the relay margin based on the time domain simulation as a
constraint. Through the dynamic optimization, the relay margin will be improved during
the interested time interval to prevent the unintended relay operation which may
aggravate the system operating condition under stressful situations.
41
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
N. B. Bhatt, August 14, 2003 U.S.-Canada blackout, presented at the IEEE PES
General Meeting, Denver, CO, 2004.
Western Systems Coordinating Council Disturbance Rep. For the Power System
Outages that Occurred on the Western Interconnection on July 2, 1996, 1424
MAST and July 3, 1996, 1403 MAST
X. Vieira et al., The March 11th 1999 blackout: short-term measures to improve
system security and overview of the reports prepared by the international
experts, in Proc. CIGR Session, SC 39Workshop on Large Disturbances, Paris,
France, Aug. 29, 2000.
M. Jonsson and J. E. Daalder, "An adaptive scheme to prevent undesirable
distance protection operation during voltage instability," Power Delivery, IEEE
Transactions on, vol. 18, pp. 1174-1180, 2003.
M. Larsson and D. Karlsson, "Coordinated system protection scheme against
voltage collapse using heuristic search and predictive control," Power Systems,
IEEE Transactions on, vol. 18, pp. 1001-1006, 2003.
A. G. Phadke and J. S. Thorp, Computer Relaying for Power Systems, 1st ed.
Baldock, Hertfordshire, U.K.: Research Studies, 1988, p. 31,263.
W. A. Elmore, System stability and out of step relaying, in Protective Relaying
Theory and Applications. New York: Marcel Dekker, 1994, pp. 319333.
M.A. Pai, P.W. Sauer, F. Dobraca, "A new approach to transient stability
evaluation in power systems", Proceedings of the 27th Conference on Decision
and Control, Austin, Texas, December 1988.
F. Dobraca, M.A. Pai, P.W. Sauer, Relay margin as a tool for dynamic security
analysis, Int. J. Electr. Power Energy Syst., vol. 12, no.4, pp. 226-234, Oct. 1990.
www.wecc.biz/documents/library/procedures/operating/WECC reliability criteria
MORC.pdf
www.ieso.ca/imoweb/pubs/marketAdmin/IMO req 0041 transmission assessment
criteria.pdf
P. Kundur, J. Paserba, V. Ajjarapu, G. Andersson, A. Bose, C. Canizares, N.
Hatziargyriou, D. Hill, A. Stankovic, C. Taylor, T. Van Cutsem, and V. Vittal,
"Definition and classification of power system stability IEEE/CIGRE joint task
force on stability terms and definitions," Power Systems, IEEE Transactions on,
vol. 19, pp. 1387-1401, 2004.
L.A.Ll. Zarate, C.A. Castro, J.L.M. Ramos and E.R. Ramos, Fast computation of
voltage stability security margins using nonlinear programming techniques, IEEE
Trans Power Syst, vol. 21, pp. 1927, 2006.
N. Flatab, R. Ognedal, and T. Carlsen, Voltage stability condition in a power
transmission system calculated by sensitivity methods, IEEE Trans. Power Syst.,
vol. 5, no. 4, pp. 12861293, Nov. 1990.
G. K. Morison, B. Gao, and P. Kundur, Voltage stability analysis using static and
dynamic approaches, IEEE Trans. Power Syst., vol. 8, no. 3, pp. 11591171,
Aug. 1993.
42
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
43
Project Publications
44
Volume 2
Project Team
A.P. Sakis Meliopoulos, Georgia Institute of Technology
Table of Contents
1. Introduction................................................................................................................... 1
1.1 Background........................................................................................................... 1
2. Voltage-Load Dynamics: System Modeling................................................................. 2
2.1 Introduction .......................................................................................................... 2
2.2 Quasi-Dynamic Quadratized Analysis ................................................................. 4
2.2.1 Overview of Quadratized Analysis ............................................................... 5
2.2.2 Quadratic Integration Method ....................................................................... 6
2.3 Three-Phase Induction Motor Model ................................................................... 7
2.4 Single-Phase Induction Motor Model ................................................................ 13
2.5 Synchronous Generating Unit Model ................................................................. 16
3. Voltage-Load Dynamics: Control............................................................................... 20
3.1 Introduction ........................................................................................................ 20
3.2 Problem Description ........................................................................................... 20
3.3 Mathematical Problem Formulation................................................................... 21
3.4 Solution Methodology ........................................................................................ 24
4. Numerical Examples ................................................................................................... 27
4.1 Test System Definition ....................................................................................... 27
4.1.1 Test System 1 Transmission System ........................................................ 27
4.1.2 Test System 2 Distribution System .......................................................... 36
4.2 Simulation of Voltage Recovery ........................................................................ 43
4.2.1 Test System 1 Transmission System ........................................................ 43
4.2.2 Test System 2 Distribution System .......................................................... 53
4.3 Summary............................................................................................................. 61
References......................................................................................................................... 62
Project Publications .......................................................................................................... 65
List of Figures
Figure 1. Possible Behavior of Voltage Recovery During and After a Disturbance .......... 4
Figure 2. Three-Phase Induction Motor Sequence Networks ............................................. 8
Figure 3. Induction Motor Input Data Form ....................................................................... 9
Figure 4. Single-Phase Induction Motor Input Data......................................................... 14
Figure 5. Single-Phase Induction Motor Physical Circuit ................................................ 15
Figure 6. Equivalent Circuit of Synchronous Generator Model ....................................... 18
Figure 7 Transmission Line Data...................................................................................... 29
Figure 8. Transmission Line Data..................................................................................... 30
Figure 9. Distribution Line Data....................................................................................... 30
Figure 10. Generating Unit Substation Configuration (Unit 1) ........................................ 30
Figure 11. Distribution Substation Configuration ............................................................ 31
Figure 12. Step-up, Three-Phase Transformer Data ......................................................... 32
Figure 13. Step-down, Three-Phase Transformer Data .................................................... 32
Figure 14. Three-Phase Distribution Transformer Data ................................................... 32
Figure 15. Load Data ........................................................................................................ 33
Figure 16. Three-Phase Induction Motor Data ................................................................. 34
Figure 17. Motor Protection Scheme ................................................................................ 35
Figure 18. Single-Line Diagram of Test System 2 ........................................................... 36
Figure 19. Underground Cable Model .............................................................................. 37
Figure 20. Mutually Coupled Distribution Line ............................................................... 38
Figure 21. Single-Phase Pole Transformer with Center-tapped Secondary ..................... 39
Figure 22. Secondary Bus Load........................................................................................ 40
Figure 23. Single-Phase Induction Motor......................................................................... 41
Figure 24. Additional Three-Phase Induction Motor Data ............................................... 42
Figure 25. Steady-State Analysis Results for Induction Motor 1 ..................................... 44
Figure 26. Steady-State Analysis Results for Induction Motor 2 ..................................... 44
Figure 27. Generating Unit Response after a Three-Phase Fault...................................... 45
Figure 28. Induction Motor Response after a Three-Phase Fault ..................................... 46
Figure 29. Transmission System Response after a Three-Phase Fault ............................. 47
Figure 30. Induction Motor Response after a Three-Phase Fault with Motor
Disconnection ................................................................................................................... 48
ii
iii
1. Introduction
1.1 Background
The primary motivation for this part of the project arises from the need to develop
analytical tools to study and enhance power system voltage stability incorporating system
dynamics (in particular load dynamics) and integrate the protection system for the
purpose of studying the response of relays during voltage instabilities. Emphasis is given
to detailed system modeling and in particular inclusion of distribution feeder models with
distributed dynamic loads to capture voltage phenomena that may have both local and
global impact on the system.
In most investigations after blackouts two observations emerged: (a) voltage
transients and/or lack of reactive power support was a contributing factor and (b) relay
unwanted operations have contributed to spreading the disturbance and impact on the
system. These observations are related. During voltage transients and/or voltage
instabilities flow of high currents are experienced in an un-faulted system together with
depressed voltages that may be recovering slowly. These phenomena can cause
operations of over-current relays or distance relays in the form of load encroachment. It is
important to fully understand these phenomena and to (a) determine the conditions that
make these phenomena manifest themselves and damage the system and (b) develop
mitigation methods to alleviate the impact on the system.
It is important to recognize that the best way to study these phenomena is by
developing integrated models of the power systems, the dynamics of the load and the
protective system. The work of this project focused on this approach. An integrated
model was developed and the phenomena were studied using this model as well as
techniques for mitigating the effects of these transients. The report presents the
methodology and preliminary results.
the system including the location of protective relays. Therefore the solution can be used
to determine the response of the relays during the recovery period, if any.
A new method has been developed for the solution of this problem. The method is
based on the quadrartized power flow and includes dynamic models of generating units
and electric loads. It provides a high fidelity response of the integrated system during the
disturbance and after the disturbance has been cleared as well as the response of the
relays if any. We have named the method Quasi-Dynamic Quadratized Analysis. This
method is described next.
1.00
0.95
Voltage (pu)
0.90
0.85
0.80
0.75
0.70
Motors will trip
if voltage sags
for too long
0.65
0.60
Fault
-1.00
-0.50
Fault Cleared
0.00
0.50
1.00
1.50
2.00
Seconds
i k
k k
k
k
k
= Y x + x F2 x b
0
M
(2.1)
where i k is the current through the component, x k is the vector of the component states
and b k the driving vector for each component, which may contain past history terms, as
well, in the case of dynamical models. Matrix Y k models the linear part of the
component and matrices Fi k the nonlinear (quadratic) part. This model can refer to a
passive component of the system (no dynamical equations) or a dynamic component of
the system, i.e. a component that is described with algebraic and differential equations.
The form (2.1) results from (a) the quadratization of the equations and (b) the integration
of the differential equations. The integration of the differential equations is described in
the next section. Two examples of this modeling approach are given in subsequent
paragraphs (induction motor and synchronous generator).
X T F1 X
T
0
0 = Y X + X F2 X b = G ( X ) ,
M
(2.2)
where
Fi
: driving vector.
(2.3)
where
: iteration step,
The iterative procedure terminates when the norm of the equations is less than a defined
tolerance. This iterative procedure is repeated at each time step. If the system exhibits no
dynamical behavior (i.e. is completely static), the analysis is equivalent to the load flow
analysis and the solution of the above system of equations provides the steady state
solution of the system.
2.2.2 Quadratic Integration Method
A new numerical integration scheme is employed for the solution of the dynamical
equations. It relies on a collocation-based implicit Runge-Kutta method (Lobatto family)
and is A-stable and order 4 accurate. The method is based on the following two
innovations: (a) the nonlinear model equations (differential or differential-algebraic) are
reformulated to a fully equivalent system of linear differential and quadratic algebraic
equations, by introducing additional state variables, as described in the previous section,
and (b) the system model equations are integrated assuming that the system states vary
quadratically within a time step (quadratic integration).
Assuming the general nonlinear, non-autonomous dynamical system:
x& = f (t , x ) ,
(2.4)
the algebraic equations at each integration step of length h , resulting from the quadratic
integration method, are:
h
h
5h
f ( t m , xm ) +
f (t , x (t )) = x (t h ) +
f (t h, x (t h ))
3
24
24
h
h
2h
x (t )
f (tm , xm ) f (t , x(t )) = x (t h ) + f (t h, x(t h ))
3
6
6
xm
(2.5)
The above equations are put in the matrix form of equation (2.1).
2.3 Three-Phase Induction Motor Model
Typically induction motors are represented in power system studies as constant
power loads. Although this is a valid representation for steady-state operation under
certain conditions, induction motors do not always operate under constant power,
especially when large deviations of voltage occur. In reality induction motors in steadystate operate at a point where the electro-mechanical torque of the motor equals the
mechanical torque of the electric load. As the voltage at the terminals of the induction
motor changes, the operating point will change. Here, we present an induction motor
model that can more accurately describe the motor behavior. The model is in quadratic
form, that is, it consists of equations that are at most quadratic [19]-[22], and can be
readily integrated into the power flow model. In addition, the model can be used to
determine the operation of the system at a specific instant of time assuming that the speed
of the induction motor is fixed (for example, after a disturbance). The reactive power
absorption of the induction motors is different at different slip values and therefore they
affect the voltage profile of the system. This behavior cannot be captured by a simple,
static, constant power load model.
A quadratic, three-phase induction machine model has been developed [22], as an
extension of a similar single-phase equivalent model [19]-[21]. This was done as part of a
previous PSERC project, S-24 and is briefly presented in this section. The model is based
on the typical steady state sequence circuits of the induction motor, shown in Figure 2.
Note that induction motors have in general little or no asymmetry, so their representation
with sequence networks is valid and accurate. The model input data include typical motor
nominal (nameplate) data, plus electrical parameters, and mechanical load data. The user
interface of the model is presented in Figure 3 and shows the model implementation
details. The model supports two mechanical loading modes: (a) torque equilibrium, and
(b) constant slip. In the torque equilibrium mode, the mechanical torque can be either
constant, or depend linearly or quadratically on the mechanical speed.
(+)
(-)
(0)
Figure 2. Three-Phase Induction Motor Sequence Networks
~
~
0 = ( g m + jbm ) E1 + E1
s
~ ~
( g s + jbs )(V1 E1 )
rr + jx r s
~
~
0 = ( g m + jbm ) E 2 + E 2
2s
~ ~
( g s + jbs )(V2 E 2 )
rr + jx r (2 s )
9
(2.6)
where:
T
1
0
= e j 240
j120 0
e
1
e
j120 0
e j 240
1
1 ,
1
,
rs + jxs
g m + jbm =
1
,
rm + jx m
g 0 + jb0 =
1
for grounded Y, and 0 otherwise.
rs + rr + jx s + jx r
An additional equation links the electrical state variables to the mechanical torque.
This equation is derived by equating the mechanical power (torque times mechanical
frequency) to the power consumed by the variable resistors in the positive and negative
circuits of Figure 2.
2
2
~
~
E1
E2
0=
srr
( 2 s ) rr Tem s
rr + jx r s
rr + jx r ( 2 s )
(2.7)
where:
sn
Tem
Two steady state operating modes are defined from the above equations:
(a) Constant Slip Model (Linear):
~
~
I abc = T 1 I 120
~
~
0 = Vabc T 1V120
~
~ ~
0 = I 1 + ( g s + jbs )( E1 V1 )
~
~
~
0 = I 2 + ( g s + jbs )( E 2 V2 )
~
~
0 = I 0 ( g 0 + jb0 )V0
(2.8)
~ ~
0 = ( g m + jbm ) E1 + E1
s
~ ~
( g s + jbs )(V1 E1 )
rr + jx r s
~
~
0 = ( g m + jbm ) E 2 + E 2
2s
~ ~
( g s + jbs )(V2 E 2 )
rr + jx r ( 2 s )
10
In the constant slip mode the motor operates at constant speed. The value of the slip
is known from the operating speed and therefore the model is linear. If a neutral exists at
~
the stator side (wye connection) the neutral voltage, Vn , is added as state, along with the
~ ~ ~ ~
~
equation I n = I A + I B + I C = 3I 0 .
(b) Torque Equilibrium Model (Nonlinear-Quadratic):
~
~
I abc = T 1 I 120
~
~
0 = V abc T 1V120
~
~ ~
0 = I 1 + ( g s + jbs )( E1 V1 )
~
~
~
0 = I 2 + ( g s + jbs )( E 2 V2 )
~
~
0 = I 0 ( g 0 + jb0 )V0
0 = Tem s + U 1 srr U 2 ( 2 s ) rr
~
~
~
0 = ( g s + jbs )V1 + ( g s + g m + j (bs + bm )) E1 + W1 s
~
~
~
0 = ( g s + jbs )V2 + ( g s + g m + j (bs + bm )) E 2 + W2 ( 2 s )
~
~
0 = rr Y1 + jx r sY1 1
~
~
0 = rr Y 2 + jx r ( 2 s )Y 2 1
~ ~~
0 = W1 Y1 E1
~
~~
0 = W2 Y 2 E 2
~ ~*
0 = W1W1 U 1
~ ~*
0 = W2W2 U 2
(2.9)
If the mechanical torque is not constant, but depends on the speed (slip), equation (2.10)
is also added completing the general model.
0 = Tm a b s c s + (b s + 2 s ) s s s 2
2
(2.10)
In the torque equilibrium model the motor electromechanical torque Tem is equal to
the mechanical load torque, Tm . The slip is not a known constant and thus it becomes part
of the state vector. Note that this model is nonlinear. Note also that the state vector and
the equations are given in compact complex format. They are to be expanded in real and
imaginary parts to get the actual real form of the model. Note also that the last equation is
real. As in the previous mode, if a neutral exists at the stator side (wye-connection) the
~ ~ ~ ~
~
~
neutral voltage, Vn , is added as state, along with the equation I n = I A + I B + I C = 3I 0 .
In order to capture the essential dynamic behavior of induction motor loads the
model described in the previous section is augmented by the dynamical equation (2.11)
describing the rotor motion:
J
d m
= Tem Tm ,
dt
(2.11)
11
where
J
Tem
Tm
(2.12)
The above transient (hybrid) model can capture the effects of the motor in the
voltage profile of the power system. The electrical transients in the motor are neglected,
as they do not have significant effect in the network solution, especially for the time
scales of interest, which are very long compared to the time scales of the electrical
transients. Phasor representation is therefore used for the electrical quantities. The
elimination of stator electrical transients makes it possible to interface the dynamic motor
model with the static network model (quasi steady state network model) yielding an
integrated hybrid model.
The model described above is based on the standard equivalent circuit of an
induction machine. This model is in general capable of representing a wide variety of
motors; however, there are several motor types that cannot be adequately represented
with this model, for example motors with double cage or deep bar rotors. For proper
modeling of such motors a slightly modified equivalent circuit has been used [25][29].
Here, a generalized model is used that assumes that the rotor parameters are not constant,
but depend on the slip (speed) of the motor [23]. A quadratic dependence is assumed for
the rotor resistance and a linear dependence for the rotor reactance:
rr ( s) = + s + s 2
(2.13)
x r ( s) = + s
where s is the operating slip. These equations are included in the motor model. Note that
in the constant slip operating mode the model is not significantly affected, since the slip
is known and thus the rotor impedance is simply computed for this slip value. In the
constant torque mode, however, the rotor parameters become part of the state vector after
the inclusion of equations (2.13). The values of the resistance reduce as the speed
increases, while the reactance may have some very small variation with speed. In fact the
reactance value changes slightly and remains mainly constant, as it is also linearly related
to the stator reactance which we assume constant. A similar change could also be
assumed for the stator reactance, to make the model more precise.
A model with slip dependent rotor parameters can adequately represent, in a unified
way, motors of every type and every NEMA design (A, B, C or D), including motors
with double cage, or deep bar rotors. Designs A and D can be accurately represented
using constant parameter models; for designs B and C the slip-depended model is used
for more realistic representation.
12
Single-phase induction motors are very common in residential and commercial load
areas (representing a substantial part of the electric load) and have been identified to have
major contribution to many voltage related problems, especially when they operate as airconditioning compressors. A comprehensive single-phase induction motor model was
developed as part of this project to allow modeling of such conditions. The model input
data include typical motor nominal (nameplate) data, plus electrical parameters, and
mechanical load data. The model supports four mechanical loading modes, in steady
state: (a) Constant torque, (b) Constant power, (c) Constant slip, and (d) Speed-dependent
torque. For dynamic analysis the constant electric power and constant slip modes have no
meaning and therefore only the constant torque or slip dependent torque modes are used.
The model also supports the options of including a running and a starting capacitor. The
model incorporates four starting methods: (a) Split Phase, (b) Capacitor Start, (c)
Permanent Split Capacitor, and (d) Capacitor Start, Capacitor Run. The input data form is
illustrated in Figure 4.
13
14
Circuit analysis yields the following equations, for the full case:
~ ~
~ ~
I 1 = I main + I C + I CS
~
~
~ ~
I 2 = I main I C I CS
~
~ ~
VS = V1 V2
~
~
Vmain = VS
~
~
VC = VS
~
~
VCS = VS
~
VC
~
IC =
j e C
~
VCS
~
I CS =
j e C S
~ ~
~
I C + I CS + I aux = 0
~
~
~
Vmain = I main Rmain + j e main
~
~
~
Vaux = I aux Raux + je aux
~
~
~
~
~
~ ~
main = Lmain jL2 main ,r K + + K I main + Lmain ,r Laux ,r K + K I aux
~
~
~ ~
~
~ ~
aux = Lmain,r Laux ,r K + K I main + Laux jL2 aux ,r K + + K I aux
~
~
~ + js e Lmain ,r I main + jLaux ,r I aux
I r1 =
2(Rr + js e Lr )
~
~
~ j (2 s) e Lmain ,r I main jLaux ,r I aux
I r1 =
2(Rr + j (2 s ) e Lr )
)]
[
]
15
)]
(2.14)
~
~
I r+2 = jI r+1
~
~
I r2 = jI r1
where
s e
,
(2.15)
2(Rr + js e Lr )
(2 s ) e
~
K =
,
(2.16)
2(Rr + j (2 s ) e Lr )
e is the synchronous electrical angular velocity, in electrical rad/sec,
and the slip, s , is related to the angular velocity and mechanical speed as:
m ns n
s = sm
=
,
(2.17)
sm
ns
where
sm : synchronous mechanical speed in rad/s,
m : motor mechanical speed in rad/s,
ns : synchronous mechanical speed in rpm,
n
: motor mechanical speed in rpm.
~
K+ =
An additional equation links the electrical state variables to the electrical torque produced
by the motor.
{(
)(
~ ~*
~ ~* ~ + ~
poles
2
2
Te =
Re L main, r I main I main + L aux , r I aux I aux K K
2
)(
~* ~
~ ~* ~ + ~
I aux I main I aux
K +K
+ jLmain, r Laux , r I main
(2.18)
2.5 Synchronous Generating Unit Model
A single axis generator model is used in this work. The unit is electrically described
as a source (controlled by its subsystems) behind an equivalent impedance, as illustrated
in the equivalent circuit of Figure 6. The rotor mechanical dynamics are the only
dynamics included in the model. The internal sources provide a set of balanced three
phase voltages, described with the state variables E , . The dynamic model is based on
a quasi-steady-state model that assumes that the generator is operating under sinusoidal
steady state conditions as far as the electrical system is concerned. Only the rotor
mechanical system dynamics are assumed, therefore the steady-state equations described
in the previous section also hold, with the augmentation of the system with the swing
equation of the rotor rotational movement. This equation defines the mechanical
rotational speed (t ) as well as the internal voltage angle (t ) which is now a time
varying quantity. The internal voltage magnitude E (t ) is specified by the excitation
16
)}
*
system, or may have constant value. Therefore the model compact equations are as
follows:
~
~ ~ ~
I a = ( g + jb )(Va Vn E a )
~
~ ~ ~
I b = ( g + jb )(Vb Vn Eb )
~
~ ~ ~
I c = ( g + jb )(Vc Vn E c )
~
~ ~ ~
~
I n = ( g + jb )( Va Vb Vc + 3Vn )
d (t )
= (t ) s
dt
J
d (t )
= Tm (t ) Te (t ) D( (t ) s )
dt
~ ~
~~
~~
0 = Pe (t ) + Re E a I a* + Eb I b* + Ec I c*
0 = Te (t ) (t ) Pe (t )
0 = E (t ) KE f (t )
where:
g=
R
R + 2 L2 ,
b=
L
R + 2 L2 ,
Furthermore,
~
E = E (t )e j (t ) = E (t ) cos (t ) + jE (t ) sin (t )
~
E a = Ee j
2
j ( )
~
3
Eb = Ee
j ( )
~
3
Ec = Ee
17
(2.19)
Vai
Vbr
Vbi
Vcr
Vci
Vnr
Vni
(t ) (t ) Pe (t ) Te (t ) E (t )]
~
Ia
Ls
~
Ea
Lm
~
Ec
~
Eb
Lm
Ls
Lm
Ls
~
Ib
~
Ic
b
c
n
19
Once the voltage recovery phenomena are modeled and studied the topic of
mitigation of such phenomena is addressed. Installation and operation of dynamic VAr
devices provides the means of locally controlling the voltage in real time, during
transients, and thus alleviating or even eliminating such problems. The focus of this part
of the report is the optimal operation of installed VAr resources for speeding up the
transient voltage recovery and minimizing the impact of these transients.
The overall problem is stated as follows: Assume a power system with static and
dynamic loads, generating units with specific VAr control capabilities and other VAr
control devices that is subject to a wide variety of exogenous disturbances. For a specific
disturbance the protective system will respond and will take the system through specific
switching operations. During this period the system will experience transients that may
include voltage recovery transients and possible voltage instability. For this sequence of
events, the needed control for the dynamic VAr sources is determined to ensure that the
system will not experience any voltage instabilities (operational problem).
This project briefly addresses this topic on a theoretical basis. The problem is
formulated as an optimal control problem that determines the minimum control effort that
provides the desired system behavior, based on specific operational criteria. Such criteria
are associated with the rate of recovery and the minimum time to recovery. Direct
transcription methods are used to create a discrete numerical approximation of the
continuous optimal control problem, using an implicit Runge-Kutta discretization
scheme. The mathematical problem formulation is presented considering a power system
quasi-steady state dynamic model, operational path constraints and specific objective
functions. The end result provides an insight of the optimal operation and control of
reactive support resources, under such transient phenomena.
3.2 Problem Description
20
2) During-fault phase: When a fault (or a disturbance in general) takes place the
system enters a transient operating condition. Typically the during-fault phase is
characterized by severely abnormal operating conditions, like e.g. very low voltage levels
or significant frequency excursions.
3) Post-fault phase: This is the most important and interesting phase of the analysis.
The fault is cleared by the protective system (or in general the disturbance is removed).
This is in general associated with a change in the system configuration, which may now
move to a new acceptable or unacceptable steady state or even become unstable, based on
the control action applied to the system. The final operating condition at the end of the
fault period is the initial conditions of the post fault system.
Voltage recovery following short circuits in electric power systems is one such
phenomenon. Short circuits of various types (three-phase, two-phase, two-phase to
ground, single phase to ground) are common events in electric transmission or
distribution systems. Such faults can cause significant voltage dips in their vicinity. They
are usually cleared by the operation of protective circuit breakers and possibly the
isolation of the fault location from the rest of the system (e.g. removal of faulted line).
However, voltage built-up after such a fault may be slow and even exhibit a strong
oscillatory behavior, depending on the dynamic characteristics of the system, and in
particular synchronous generating units and dynamic loads, like e.g. motor loads. Such
slow voltage recovery can result in system-wide problems, like voltage instability and
voltage collapse or local problems in particular weak system areas. That is, the system as
a whole may recover and appear to reach an acceptable new steady state, however,
specific load areas of the system may continue experiencing unacceptable operating
conditions. This might result from the fact that slow voltage recovery may have other
secondary effects and result in undesired protective relay operation like tripping of
sensitive loads.
Voltage behavior in a power system is mainly controlled via the synchronous
generating units, or via reactive support devices (VAr devices) like switched capacitors
deployed throughout the system. However, generators cannot provide local support, while
capacitors have to be switched on and off mechanically and thus cannot provide
continuous and fast, real time response. They are, thus, mainly used for controlling the
voltage based on steady state criteria, rather than during transients. They are also passive
elements, and their control ability depends on the system voltage at their locations.
Therefore, during faults that result in significant voltage dips their response might not be
considerable.
Therefore, dynamic VAr sources, in the form of FACTS devices, are the only
practical way of locally controlling the voltage. Such devices use power electronic
technology to control the reactive power they inject into the system. This allows fast
response times and thus practically real time control of the system, in the form of
continuous, rather than discrete control action.
3.3 Mathematical Problem Formulation
x& (t ) = f (x(t ), y (t ), u (t ), p, t )
0 = g (x (t ), y (t ), u (t ), p, t )
(3.1)
where
x (t ) are the dynamical system states;
y (t ) are the algebraic states;
u (t ) are the system control variables;
p
The number of dynamical states is nx, of algebraic states ny, and of control variables
nu. The number of differential equations is nf and of algebraic equations ng. For
implementation purposes it is convenient, and possible for most of the power system
applications, to convert the function f to a set of linear equations and move all the
nonlinearities to the algebraic equations, g. Furthermore, g is assumed to be a set of at
most quadratic equations. This can be, in general, achieved by introducing some
additional algebraic state variables to quadratize the model, without any approximations,
as also explained earlier in the report. Functions f and g, in general, also contain
switching functions representing elements that are switched on an off or change their
operating mode, or modifications in the system configuration during different stages of
the analysis and during the various scenarios under study. Changes in f and g define
different phases of the problem, so within a single phase the system dynamics are
considered unmodified.
The control vector can contain any available control of interest, like e.g. generator
controls, load controls (if available) both continuous and discrete. In our case, as
described in the previous section, the control variables are the desired reactive power
injections from existing dynamic VAr sources of interest. This is a continuous control, in
the form of a trajectory of reactive power injection during the during-fault and post-fault
phases of the analysis. During the pre-fault phase the control input is some specific
reference value that keeps the system operating under the desired steady state conditions.
The problem is to define the optimal (minimum effort) control trajectory for
dynamic VAr source injections that satisfies specific recovery criteria, associated with the
rate of recovery. The approach presented in [30]-[31] is followed. Specifically, the
control functions u (t ) are to be chosen to minimize the objective function
J = [x(t F ), y (t F ), t F ]
(3.2)
[x(t F ), y (t F ), p, t F ] = 0
(3.3)
22
where the initial conditions x(t I ) = x I and y (t I ) = y I are given at the fixed initial
time t I and the final time t F is free. The modified objective function of the problem is
defined as
J = + v T
] {
tF
tF
(t ) g + T (t )[x& (t ) f ] dt .
(3.4)
tI
Vector v is the vector of Lagrange multipliers for the discrete constraints, while
vectors (t ) and (t ) are the multipliers for the continuous constraints, referred to as
adjoint or costate variables. The necessary conditions for a constraint optimum are
obtained by setting the first variation J = 0 . The system Hamiltonian is defined as,
assuming that the differential-algebraic equation (DAE) system (3.1) is of index one.
H = T f + T g
(3.5)
(3.6)
[& & ] = [ H
T
x
0 = H uT ,
(t F ) = Tx
(t F ) = Ty
t =t F
t =t F
0 = ( y + H )
H Ty , (adjoint equations)
(3.7)
(control equations)
(3.8)
,
,
t =t F
(transversality conditions)
(3.9)
0 = (t I ),
0 = (t I ).
Equations (3.1), (3.5) and (3.8) comprise a DAE system with boundary conditions
at both t I and t F provided by (3.3) and (3.9). Altogether they form a two-point boundary
value problem.
Furthermore, additional equality or inequality path constraints can be imposed,
within each phase, of the general form
hl h(x (t ), y (t ), u (t ), p, t ) hu ,
(3.10)
as well as simple bounds on the state and control variables, representing, e.g., upper and
lower limits of specific device outputs.
xl x(t ) xu
y l y (t ) y u
(3.11)
u l u (t ) u u
23
The number of grid points is N = M + 1 . The values of the state and control
variables are treated as a set of NLP variables. The differential equations are replaced by
a finite set of discretized equations, at each grid point, based on some numerical
integration scheme. For the trapezoidal scheme the NLP variables are
X T = x1T
y1T
u1T
... x MT
u MT ,
y MT
(3.12)
h
h
0 = x k k f k + x k 1 k f k 1
2
2
(3.13)
x k = x k 1 +
or
where the subscripts k and k-1 denote the values at the current time t k and the previous
time t k 1 = t k h respectively, and hk is the discretization step at grid point k. Usually
the step is constant, however, this notation also accommodates the case where the step is
variable. The stepsize can be defined as:
hk = k (t F t I ) = k t
(3.14)
x 2T
y 2T
h
h
5h
fm +
f k = x k 1 +
f k 1
3
24
24
h
h
2h
xk
f m + f k = x k 1 + f k 1
3
6
6
u 2T
T
... x Mn
T
y Mm
T
u Mm
x MT
xm
(3.15)
or
24
y MT
u MT
5h
h
h
f k + x m f m + x k 1
f k 1
24
3
24
h 2h
h
0 = xk + f k
f m + x k 1 f k 1
6 3
6
0=
(3.16)
The subscript m denotes the midpoint of the segment with endpoints k 1 and k . All the
algebraic equations are enforced at the grid points and appended to the algebraized
differential equations. This means that the following equations, (3.17) and (3.18), are
appended for the trapezoidal and quadratic discretization schemes respectively, as derived
from (3.1), (3.10) and (3.11).
0 = g ( x k , y k , u k , p, t k )
hl h( x k , y k , u k , p, t k ) hu
xl x k xu
(3.17)
yl y k yu
ul u k uu
0 = g ( x k , y k , u k , p, t k )
hl h( x k , y k , u k , p, t k ) hu
xl x k xu
yl y k yu
ul u k uu
(3.18)
0 = g ( x m , y m , u m , p, t m )
hl h( x m , y m , u m , p, t m ) hu
xl x m xu
yl y m yu
ul u m uu
Inequality constraints are treated by the introduction of slack variables that convert them
to equality constraints.
Therefore, the resulting equality constraint NLP problem is to chose the decision
variables X , as defined earlier, to minimize the objective function
F ( X ) = ( xM , y M ) ,
(3.19)
(3.20)
which in this case are defined by equations (3.13) and (3.17) or (3.16) and (3.18) for
k = 1 to M = N 1 , depending on the discretization scheme. It is assumed that the
provided initial condition, referring to the first grid point (at t 0 ) are consistent with all the
algebraic constraints.
25
The form of (3.19), though it appears simplistic and restrictive, is quite general and
can also accommodate objective functions or constraints defined via quadrature forms.
This can be done by introducing additional dynamical states and casting the optimization
problem in Mayer form [3.11]. So, for example, for an objective function defined as
tF
J = w(x(t ), y (t ), u (t ), p, t )dt
(3.21)
tI
an additional dynamic state variable, (t ) , can be introduced along with the differential
equation
& = w( x(t ), y (t ), u (t ), p, t )
(3.22)
(3.23)
L( X , ) = F ( X ) T c( X ) = ( x M , y M ) kT c k ( X ) .
(3.24)
k =1
(3.25)
The solution of the defined NLP problem is obtained via Newtons method.
The solution provides the control signals for the system to mitigate the voltage recovery
transients. The method will be demonstrated with examples in the next section.
26
4. Numerical Examples
4.1 Test System Definition
Two test systems have been utilized to demonstrate the methodology the dynamic
simulation of voltage recovery and mitigation of voltage transients. The first system is a
small transmission system with little representation of distribution parts that operates
under near-balanced conditions and presents very little asymmetry. The second system is
an extension of the first to include a more detailed modeling of distribution feeders
connected to the bulk power system. The distribution parts are asymmetric and operate
under unbalanced conditions. The two test systems are described next.
4.1.1 Test System 1 Transmission System
SUB03-S
SUB
SUB03-1 BUS03SUB03-2
SUB03-1-
SUB03-4
SUB01-3-
SUB03-4GEN2
SUB01-3
BUS01
GEN
GEN1
GEN
SUB02-3
BUS02
SUB01-4
SUB04-3-
SUB04-3
SUB
SUB04-1 BUS04
SUB04-5
SUB
BUS06SUB06-4
SUB
SUB05-4 BUS05
SUB05-7
SUB04-6
SUB06-7
BUS07
BUS08
BUS08-L
M
BUS08-M
IM
BUS07-L
M
BUS07-M
IM
3Ph
27
3Ph
The system consists of two generating units and a third generator (acting as slack unit in
steady state) representing the equivalent external network, where the system is connected
to, via a transmission line. There are two generating substations (buses 1 and 2), equipped
with step-up transformers, two transmission substations (buses 3 and 4), two distribution
substations (buses 5 and 6), with step down transformers, seven transmission lines and
two distribution lines in the system. Loads are connected to the system at the ends of the
distribution lines via transformers. Portion of the load is represented as constant
impedance load and another portion as induction motor loads. Motor 1 has a fan-type
mechanical loading, while motor 2 drives a constant torque mechanical load. The
generating units operate at 12 kV. The transmission system operates at 115 kV. The small
distribution feeders operate at 25 kV and 13.8 kV. The system data for all the system
components are presented in Figures 10 through 21.
28
29
SUB01-3
1
I
GEN1
SUB01L
SUB01H
SUB01
SUB01-4
SUB03-1
SUB03-S
SUB03-2
SUB03-4
(a)
SUB05-L1
SUB05-L2
SUB05-4
SUB05-7H
SUB05-7L
SUB05-7
(b)
Figure 11. Distribution Substation Configuration
(a) Bus 3, (b) Bus 5
31
33
34
35
The second test system is an extension of the test system 1 with more detailed
representation of the distribution system and additional dynamic loads three and single
phase. The test system 2 is illustrated in Figure 22.
SLACK
SUB03-S
BUS03
SUB03-1
SUB03-2
SUB03-1SUB
SUB03-4
SUB01-3-
SUB03-4GEN2
BUS02
SUB02-3
SUB01-3
GEN1 BUS01
SUB01-4
GEN
GEN
SUB04-3SUB04-3
BUS04
SUB04-1SUB04-5
SUB
SUB
BUS05 SUB05-9
SUB05-4
SUB
SUB06-9 BUS06
SUB06-4
SUB05-7
SUB04-6
SUB06-8
1
BUS07
BUS08
BUS08-L
BUS08-M
M
BUS07-L
BUS07-M
IM
PAD-01
2
HOUSE-P1
POLE3
Residential
IM
HOUSE-P2
M
PAD-02-L
1Ph
Residential
HOUSE-P3
M
PAD-03-L
1Ph
Residential
MCC-P2
PAD-4
HOUSE2
POLE4
B
PAD-03
POLE1
POLE2
PAD-02
SPOLE3
SPOLE2
SPOLE1
3 Ph
IM
3 Ph
IM
3 Ph
MCC-P3
IM
3 Ph
1Ph
PAD-3
Residential
Load
B
HOUSE-3
Residential
Load
36
37
38
39
40
41
42
Results with the proposed methodology applied to the two test systems are
presented in this section.
4.2.1 Test System 1 Transmission System
The steady-state operating conditions of the two induction motors are shown in
Figures 29 and 30. The system operates under steady state conditions when a three phase
fault occurs on the transmission line between buses 3 and 4, very close to bus 4. The fault
is cleared by the protection system after 200 ms (12 cycles) by opening the two circuit
breakers at the two ends of the line and removing the faulted line. The system response
during the pre-fault, fault and post-fault periods, for a total time of two seconds is shown
in Figures 31 through 34. Figure 31 shows the response of each generating unit in terms
of frequency, angle and power output. Figure 32 illustrates the behavior of the induction
motor loads in terms of terminal voltage, speed, current and absorbed power. The
important issue in the system behavior is the slow voltage recovery at the terminals of the
second induction motor. While the system, as a whole, remains stable and the voltage
recovers quickly at the terminals of the first motor, the voltage in fact never recovers to
an acceptable value at the terminals of the second motor and this leads to the motor
slowdown and eventually stalling. The transmission system response is illustrated in
Figure 33 in terms of bus voltages and line currents for all the transmission buses and
lines. Figure 33 illustrates the fact that the currents on the lines SLACK-BUS03, BUS01BUS03, and BUS02-BUS03 remain much higher than the steady-state loading values for
the whole transient period, even after the fault is cleared. In particular, the loading of all
the system lines at steady-state ranges from 20 up to 60 Amps. The currents of the three
lines mentioned above are 31, 14 and 22 Amps respectively. During the transient phase
these currents reach values up to 567, 507, and 230 Amps respectively, i.e. 10 to 35 times
their steady-state value. Such high values, above 300 Amps may last for 50 to 100 ms and
can possibly trigger overcurrent relay settings for line protection.
When a motor protection scheme is applied to the motors, that disconnects the
motor when the terminal voltage drops below 0.80 pu for more than 25 cycles of the
fundamental, the second motor is disconnected from the at about 0.2 seconds after the
fault is cleared and the rest of the system recovers again to an acceptable operating state,
as illustrated in Figure 34. Notice that in this case the voltage at the bus where motor 2 is
connected recovers to a value close to 0.9 p.u. This means that disconnection of the motor
has a beneficial effect of the rest of the loads connected to the same bus.
Finally, Figures 35 through Figures 36 show the impedance trajectory as seen by
one of the two impedance relays protecting each transmission line of the system. The
relay settings are also shown in the figures. Notice that except for the relay at the BUS03side of the faulted line, which trips, the relays of the lines connecting the equivalent rest
of the system (slack bus) and generating unit one also trigger the second protection zone,
but do not trip eventually. However, this depends significantly on the delay settings of
each relay and the situation could have been different if the settings where different.
43
44
64.74
GEN1_-_FREQUENCY (Hz)
GEN2-_FREQUENCY (Hz)
SLACK_-_FREQUENCY (Hz)
61.40
58.05
54.70
31.29
GEN1_-_PHASE_ANGLE (Deg)
GEN2_-_PHASE_ANGLE (Deg)
SLACK_-_PHASE_ANGLE (Deg)
-36.04
-103.4
-170.7
103.3 M
GEN1_-_REAL_POWER (W)
GEN2_-_REAL_POWER (W)
SLACK_-_REAL_POWER (W)
43.50 M
-16.32 M
-76.14 M
553.9 M
GEN1_-_REACTIVE_POWER (VA)
GEN2_-_REACTIVE_POWER (VA)
SLACK_-_REACTIVE_POWER (VA)
357.9 M
162.0 M
-33.93 M
0.0
0.5
1.0
1.5
45
90.85
IM1_Voltage_A_pu (%)
IM2_Voltage_A_pu (%)
Threshold_0.8_pu ()
45.44
36.79 m
4.295 k
IM1:_CURRENT_A (A)
IM2:_CURRENT_A (A)
2.148 k
481.8 m
2.109 M
IM2:_REAL_POWER (W)
IM1:_REAL_POWER (W)
1.054 M
174.7 m
1.583 M
IM1:_REACTIVE_POWER (VA)
IM2:_REACTIVE_POWER (VA)
791.4 k
90.51 m
98.11
IM1:_SPEED (%)
IM2:_SPEED (%)
84.56
71.01
0.0
0.5
1.0
1.5
46
66.40 k
58.11 k
49.82 k
Voltage_A_SLACK (V)
Voltage_A_BUS01 (V)
Voltage_A_BUS02 (V)
Voltage_A_BUS03 (V)
Voltage_A_BUS04 (V)
Voltage_A_BUS05 (V)
Voltage_A_BUS06 (V)
41.53 k
33.24 k
24.95 k
16.66 k
8.366 k
76.60
3.035 k
2.656 k
2.276 k
Current_A_Line_SLACK_to_BUS03 (A)
Current_A_Line_BUS01_to_BUS03 (A)
Current_A_Line_BUS01_to_BUS04 (A)
Current_A_Line_BUS02_to_BUS03 (A)
Current_A_Line_BUS03_to_BUS04 (A)
Current_A_Line_BUS04_to_BUS05 (A)
Current_A_Line_BUS04_to_BUS06 (A)
Overcurrent_relay_setting (A)
1.897 k
1.518 k
1.138 k
758.8
379.4
-4.492 u
0.0
0.5
1.0
1.5
47
95.70
IM1_Voltage_A_pu (%)
IM2_Voltage_A_pu (%)
Threshold_0.8_pu ()
47.87
36.85 m
4.295 k
IM1:_CURRENT_A (A)
IM2:_CURRENT_A (A)
2.147 k
2.441 m
2.109 M
IM1:_REAL_POWER (W)
IM2:_REAL_POWER (W)
1.054 M
8.642 u
1.583 M
IM1:_REACTIVE_POWER (VA)
IM2:_REACTIVE_POWER (VA)
791.4 k
6.651 u
98.11
IM1:_SPEED (%)
IM2:_SPEED (%)
72.52
46.94
0.0
0.5
1.0
1.5
Figure 30. Induction Motor Response after a Three-Phase Fault with Motor
Disconnection
48
Figure 31. Impedance Trajectory and Relay Settings of BUS03-side Relay of Line
BUS03-BUS04 (faulted line)
Figure 32. Impedance Trajectory and Relay Settings of SLACK-side Relay of Line
SLACK-BUS04
49
Figure 33. Impedance Trajectory and Relay Settings of BUS01-side Relay of Line
BUS01-BUS04
50
Figure 34. Impedance Trajectory and Relay Settings of BUS02-side Relay of Line
BUS02-BUS03
Figure 35. Impedance Trajectory and Relay Settings of BUS04-side Relay of Line
BUS04-BUS05
51
Figure 36. Impedance Trajectory and Relay Settings of BUS04-side Relay of Line
BUS04-BUS06
52
The steady-state operating conditions of the five induction motors are shown in
Figures 41 through 50.
The system operates under steady state conditions when a three phase fault occurs
on the transmission line between buses 3 and 4, very close to bus 4. The fault is cleared
by the protection system after 200 ms (12 cycles) by opening the two circuit breakers at
the two ends of the line and removing the faulted line. The system response during the
pre-fault, fault and post-fault periods, for a total time of one second is shown in Figures
46 through 50.
Figure 46 shows the response of each generating unit in terms of frequency, angle
and power output. Figure 47 illustrates the behavior of the induction motor loads in terms
of terminal voltage, speed, current and absorbed power. The important issue in the system
behavior is the slow voltage recovery at the terminals of the second induction motor.
While the system, as a whole, remains stable and the voltage recovers quickly at the
terminals of motors MCC-P3, PAD-4, and BUS08-M, the voltage in fact never recovers
to an acceptable value at the terminals of the motors BUS07-M and MCC-P2 and this
leads to the motors slowdown and eventually stalling. When a motor protection scheme
is applied to the motors, that disconnects the motor when the terminal voltage drops
below 0.80 pu for more than 25 and 30 cycles of the fundamental for the two stalling
motors respectively, motors BUS07-M and MCC-P2 get disconnected from the network
at about 0.6 and 0.8 seconds after the fault is cleared and the rest of the system recovers
again to an acceptable operating state with voltages way above 0.8 p.u., as illustrated in
Figure 47. Finally, Figures 48 through 50 illustrate the imbalances that exist among the
three system phases, due to the system asymmetries and the existence of single-phase
loads. Notice there is a significant voltage imbalance on phase B, compared to the other
two phases, at the induction motor voltage, while there is significant imbalance of phase
A at the feeder currents.
The behavior of the single-phase induction motor is similar to the one of the threephase motors, however, since this motor is smaller and has smaller inertia constant is
decelerates faster, reaching a much slower speed, but it also accelerates much faster when
the fault is cleared and thus does not significantly affect the recovery process.
53
54
55
64.76
GEN1_-_FREQUENCY (Hz)
GEN2_-_FREQUENCY (Hz)
SLACK_-_FREQUENCY (Hz)
61.53
58.29
55.06
31.60
GEN1_-_PHASE_ANGLE (Deg)
GEN2_-_PHASE_ANGLE (Deg)
SLACK_-_PHASE_ANGLE (Deg)
-19.11
-69.82
-120.5
104.8 M
GEN1_-_REAL_POWER (W)
GEN2_-_REAL_POWER (W)
SLACK_-_REAL_POWER (W)
45.80 M
-13.25 M
-72.30 M
553.9 M
GEN1_-_REACTIVE_POWER (VA)
GEN2_-_REACTIVE_POWER (VA)
SLACK_-_REACTIVE_POWER (VA)
358.2 M
162.5 M
-33.16 M
0.00
0.20
0.40
0.60
0.80
56
102.2
51.13
IM_BUS08-M:_VOLTAGE_A__pu (%)
IM_BUS07-M:_VOLTAGE_A__pu (%)
IM_MCC-P2:_VOLTAGE_A__pu (%)
IM_MCC-P3:_VOLTAGE_A__pu (%)
IM_PAD-4:_VOLTAGE_A__pu (%)
Threshold_0.8_pu ()
20.48 m
4.269 k
2.135 k
IM_BUS08-M:_CURRENT_A (A)
IM_BUS07-M:_CURRENT_A (A)
IM_MCC-P3:_CURRENT_A (A)
IM_MCC-P2:_CURRENT_A (A)
IM_PAD-4:_CURRENT_A (A)
2.318 m
2.092 M
1.046 M
IM_BUS08-M:_REAL_POWER (W)
IM_BUS07-M:_REAL_POWER (W)
IM_MCC-P3:_REAL_POWER (W)
IM_MCC-P2:_REAL_POWER (W)
IM_PAD-4:_REAL_POWER (W)
13.00 u
1.565 M
782.7 k
IM_BUS08-M:_REACTIVE_POWER (VA)
IM_BUS07-M:_REACTIVE_POWER (VA)
IM_MCC-P3:_REACTIVE_POWER (VA)
IM_MCC-P2:_REACTIVE_POWER (VA)
IM_PAD-4:_REACTIVE_POWER (VA)
-10.99 u
99.22
66.35
IM_BUS08-M:_SPEED (%)
IM_BUS07-M:_SPEED (%)
IM_MCC-P3:_SPEED (%)
IM_MCC-P2:_SPEED (%)
IM_PAD-4:_SPEED (%)
33.48
0.00
0.20
0.40
0.60
0.80
57
252.4
IM_BUS08-M:_VOLTAGE_A (V)
IM_BUS08-M:_VOLTAGE_B (V)
IM_BUS08-M:_VOLTAGE_C (V)
126.2
93.31 m
281.6
IM_BUS07-M:_VOLTAGE_A (V)
IM_BUS07-M:_VOLTAGE_B (V)
IM_BUS07-M:_VOLTAGE_C (V)
140.8
100.7 m
284.8
IM_MCC-P2:_VOLTAGE_A (V)
IM_MCC-P2:_VOLTAGE_B (V)
IM_MCC-P2:_VOLTAGE_C (V)
142.4
56.76 m
277.8
IM_MCC-P3:_VOLTAGE_A (V)
IM_MCC-P3:_VOLTAGE_B (V)
IM_MCC-P3:_VOLTAGE_C (V)
139.0
111.8 m
277.8
IM_MCC-P3:_VOLTAGE_A (V)
IM_MCC-P3:_VOLTAGE_B (V)
IM_MCC-P3:_VOLTAGE_C (V)
139.0
111.8 m
251.3
IM_PAD-4:_VOLTAGE_A (V)
IM_PAD-4:_VOLTAGE_B (V)
IM_PAD-4:_VOLTAGE_C (V)
125.7
110.5 m
0.00
0.20
0.40
0.60
0.80
58
4.295 k
IM_BUS08-M:_CURRENT_A (A)
IM_BUS08-M:_CURRENT_B (A)
IM_BUS08-M:_CURRENT_C (A)
2.864 k
1.432 k
778.3 m
1.699 k
IM_BUS07-M:_CURRENT_A (A)
IM_BUS07-M:_CURRENT_B (A)
IM_BUS07-M:_CURRENT_C (A)
1.132 k
566.2
2.318 m
2.724 k
IM_MCC-P2:_CURRENT_A (A)
IM_MCC-P2:_CURRENT_B (A)
IM_MCC-P2:_CURRENT_C (A)
1.816 k
907.9
2.312 m
1.180 k
IM_MCC-P3:_CURRENT_A (A)
IM_MCC-P3:_CURRENT_B (A)
IM_MCC-P3:_CURRENT_C (A)
786.8
393.5
315.8 m
0.00
0.20
0.40
0.60
0.80
59
152.3
Current_A_-_Feeder_13.8_kV (A)
Current_B_-_Feeder_13.8_kV (A)
Current_C_-_Feeder_13.8_kV (A)
133.3
114.2
95.20
76.17
57.14
38.11
19.07
43.63 m
6.386
Current_A_-_Feeder_25_kV (A)
Current_B_-_Feeder_25_kV (A)
Current_C_-_Feeder_25_kV (A)
5.588
4.790
3.992
3.194
2.396
1.598
800.2 m
2.205 m
0.00
0.20
0.40
0.60
0.80
60
4.3 Summary
The simulation results clearly illustrate that voltage instabilities during recovery
from disturbances can cause excessive current flow that may affect the operation of
overcurrent relays as well as the impedance seen by distance relays. Distance relays are
especially vulnerable since these phenomena exhibit simultaneously low voltage and high
current creating the possibility of load encroachment.
The results presented in this section are preliminary. It is clear that they are
dependent upon the specific parameters of the circuits involved and the type and
magnitude of dynamic loads. Therefore it is difficult to develop general guidelines for
predicting the level and impact of these phenomena. For this reason it is suggested that
the proper way to apply the proposed methodology is to study specific systems that are
heavily loaded with motor type loads.
61
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
63
[28]
[29]
[30]
[31]
64
Project Publications
G. K. Stefopoulos, A. P. Meliopoulos, and G. J. Cokkinides, Voltage-Load Dynamics:
Modeling and Control, in Proceedings of the 2007 IREP Symposium on Bulk Power
System Dynamics and Control VII, (IREP 2007 Conference), Charleston, SC, USA,
August 19-24, 2007.
G. K. Stefopoulos, G. J. Cokkinides, and A. P. Meliopoulos, Voltage Recovery
Phenomena in Distribution Feeders, in Proceedings of the 2008 IEEE PES General
Meeting, Pittsburgh, PA, USA, July 20-24, 2008.
A. P. Meliopoulos, V. Farantatos, G.J. Cokkinides, S. Mohagheghi, and G. K.
Stefopoulos, A New Out-of-Step Protection Scheme via GPS-Synchronized Data, in
Proceedings of the 16th Power Systems Computation Conference, Glasgow, Scotland,
July 14-18, 2008.
G. K. Stefopoulos, G. J. Cokkinides, and A. P. Meliopoulos, Optimal Operation of
Dynamic VAr Sources for Mitigation of Delayed Voltage Recovery, submitted to the 6th
Mediterranean Conference and Exhibition on Power Generation, Transmission and
Distribution, Thessaloniki, Greece, Nov. 2-5, 2008.
65