Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/336239537

Mitigation of SMIB Oscillations Using PSS Based on Particle Swarm


Optimization Algorithm

Article · October 2019

CITATIONS READS

0 75

2 authors, including:

Kamal Ramadan Doud


University of Khartoum
25 PUBLICATIONS   39 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Small Scale PV Grid connected System (case study: Sudan Distribution System) View project

Power systems analysis View project

All content following this page was uploaded by Kamal Ramadan Doud on 03 October 2019.

The user has requested enhancement of the downloaded file.


2019 International Conference on Computer, Control, Electrical, and Electronics Engineering (ICCCEEE)

Mitigation of SMIB Oscillations Using PSS Based on


Particle Swarm Optimization Algorithm
Abdel-Rahman Saifedin Alawad Ali Kamal Ramadan Daud
Power System Engineer, Sudanese Electricity Holding Associate Professor, Electrical and Electronics Department,
Company, Sudan University of Khartoum
arandas200@hotmail.com kamalramadan@uofk.edu

ABSTRACT— In this paper a technique dependent on Particle the poor damping of the electromechanical oscillation modes
Swarm Optimization (PSO) calculation is exhibited to find the of the power system. Poor damping could be brought out by:
ideal parameters of Power System Stabilizer (PSS), instead of (1) large amount of long-distance power transmission, (2)
classical control approaches. A fitness function based on
weak interconnection of large power sub-networks, and/or
Critical Damping Index (CDI) has been used as an optimization
(3) negative damping due to the fast-acting high-gain AVRs.
problem to get the PSS optimal parameters. PSS controller
combined with Single Machine Infinite Bus (SMIB) system is
Due to non-linearity and complexity of power systems and
used in order to mitigate any oscillations occurred in the they are regularly exposed to low frequency oscillations;
system. The eigenvalues and damping ratios for the consequently, the power system stabilizers(PSS) become a
electromechanical modes have been calculated and tabulated mandatory device than before. That is the motivation behind
for different cases of operation using MATLAB/SIMULINK. why designers invest significantly more energy and time in
All simulation outputs proved that the capability of the PSS designing power system stabilizers (PSS) [5-10]. In spite of
based on PSO in mitigation the system oscillations is better than their relative clearness, PSS might be one of the most
PSS based on conventional control methods. Also the results
misunderstood and misused pieces of generator control
show that the electromechanical damping ratios increased from
equipment. The PSS designed utilizing classical control
2.2113% without PSS control to 20.58% for system with PSS
based on conventional control method and to 65.94% for system
techniques performs well around nominal operating
with PSS based on PSO Algorithm. Also the settling time condition. however, the performance of such PSS is not
reduced to 4 seconds for system with PSS based on conventional guarantee if the system become more loaded [5]. lately,
control method and to 2 seconds for system with PSS based on Particle Swarm Optimizations (PSO) and Genetic Algorithm
PSO Algorithm. NEPLAN software has been used to prove the (GA) have been applied to the problem of PSS design [5].
effectiveness, robustness, and validity of the designed PSS
system based on PSO algorithm.
II. SYSTEM MODELLING AND ANALYSIS
Keywords – oscillation, PSS, SMIB, eigenvalues, particle swarm
A. SMIB:
optimization, The SMIB is an approximate representation of a kind of
real power systems, as represented in figure1 where, Vt∠ ,
I. INTRODUCTION the output voltage of the synchronous generator; ∠0, the
Due to complexity of power system oscillations they are not infinite bus voltage; Re, Xe, the transformer and
an easy to be analysis [1]. The analysis of electromechanical transmission equivalent resistance and reactance; and =
oscillations is one of the major parts of rotor angle stability − , the angle between Vt and . This system can be
issue [2]. A major issue here is the way wherein the power
modeled with different types of control by the following
output of synchronous machines changes as their rotors
blocks and equations [11]:
oscillate [3]. Study of power system oscillations is in order to
investigate the phenomena and to create avoidance measures
from the event of the issue. The oscillation can be happened
due to severe faults or by small perturbations under steady
state operations [4]. If the magnitude of power oscillation
increase with time, the power oscillation is said to have
Fig. 1. (SMIB system
negative damping and the power system is unstable. If the
1. SMIB System without Exciter
power oscillation shows constant magnitude, and then sets
i. Heffron–Philips Model
down, the power oscillation is said to have poor damping
figure 2 shown the block diagram of this model ,
though the power system is still stable. If the oscillation dies
where H is Inertia constant, Efd is field voltage, is
down rapidly inside a few seconds with a damping ratio
synchronous speed in radian, ∆w is the rotor deviation
greater than 0.1, the power oscillation is said to have well
speed, and K1-K4 are Heffron–Philips Model
damping [4]. The main cause of power system oscillations is
constants.

978-1-7281-1006-6/19/$31.00 ©2019
eliminate it is steady state value as shown in figure 4 [1].
The structure of PSS is shown in figure 4, where KPSS is
PSS gain, Tw is wash out time constant, and T1 and T2are
lead-lag time constants.

Fig. 2. Heffron–Philips model of the synchronous machine without exciter Fig. 4. PSS block diagram

ii. State-space model 2. Combined SMIB model with PSS


Equation 1 shows the state space model, where: E is the SMIB system, including PSS, can be represented by a
transient voltage due to field flux linkage,δ is rotor single line diagram of figure 5, and modeled by
angle, ∆v is normalized speed deviation signal, T is Heffron–Philips Model of figure 6 excluding the
open circuit transient time constant, ω is synchronous washout block, and the system A-matrix (A_PSS) of
speed,and D is damping coefficient. this model is given by equation 3 [3].
1 K
− − 0 1
∆E K T T ∆E
T
∆δ = 0 0 ω ∆δ +
0
∆E
(1)
∆v K K Dω ∆v
− − − 0
2H 2H 2H
Fig. 5. SMIB system with PSS.
2. SMIB System with Exciter
Figure 3 shows the Heffron–Philips block diagram
model with exciter. Where, Vref is reference voltage, KA
is exciter gain, TA is exciter time constant, and K1-K6
are Heffron–Philips Model constants.

Fig. 6. Heffron–Philips model of the synchronous machine with exciter and


PSS

1 K 1
− − 4 0 − 0
'
K 3T do '
T do '
T do
0 0 ωs 0 0
Fig.3. Heffron–Philips model of the synchronous machine with exciter K K Dω s
− 2 − 1 − 0 0
A _ PSS = 2H 2H 2H (3)
K AK6 K A K5 1 KA
− − 0
Equation 2 represents the state space model for this system. TA TA TA TA
K T K KT K K K T Dω s 1
− 2 1 ( PSS ) − 1 1 ( PSS ) ( PSS − PSS 1 ) 0 −
T2 2H T2 2H 2H T2 2H T2
1 K 1
− − 0 −
K T T T 0 III. PARTICLE SWARM OPTIMIZATION
∆E ∆E
0 0 ω 0 0 The word particle denotes, for example, a bee in a colony or
∆δ = K K Dω ∆δ + 0 ∆V
∆v − − − 0 ∆v K (2) a bird in a flock. Each particle in a swarm can either use it is
2H 2H 2H ∆E
∆E K K K K 1 T
− − 0
own intelligence or the group intelligence of the swarm. As
T T T such, if one particle discovers a good path to food, the rest
of the swarm will also be able to follow the good path
B. Power System Stabilizer (PSS)
instantly even if their location is far away in the swarm. The
1. background
PSO algorithm originally proposed by Kennedy and
PSS signal is used with generator automatic voltage
Eberhart in1995 [12]. The velocity, ( ) and position, ( )
regulator reference input in order to produce an electrical
of particle j in the ith iteration can be calculated and updated
torque which proportional to the generator speed, Hence
respectively using equations 4 and 5. Figure 7 show the
the suitable inputs to the PSS are the generator speed, the
flowchart for the iteration procedure.
output power, or the generator bus frequency. The input
is first passed through the washout block in order to
V. SIMULATION AND RESULTS
V j ( i ) = V j ( i − 1 ) + c 1r1  Pbest , j − X j ( i − 1 )  +
All the simulations in this paper were carried out using
c 2 r2 G best − X j ( i − 1 )  (4) MATLAB m-file, SIMULINK, and NEPLAN softwares.
( ) = ( − 1) + ( ); = 1,2, ⋯ , (5) The scenario of simulation is as follows: First of all, the
where c1 and c2 are the cognitive and social learning rates, Heffron–Philips Model constants (K1-K6) are calculated
respectively, and r1 and r2 are uniformly distributed random and tabulated in table II, after that, the SMIB has been
numbers in the range 0 and 1, Pbest is the position studied in manual control mode (MVR case), and the study
corresponding to the best fitness, and Gbest is the overall best extended to include the effect of AVR in the stability of the
out of all the particles in the population [12]. system (AVR case), and the effect of installation PSS based
on classical method (Calculated case) with PSO
optimization method for parameters tuning (PSO case). In
each step the system eigenvalues and damping ratios were
observed and tabulated as in tables (III-VII) and the
response of electromechanical modes and other modes are
shown in figures (8-13). Also, different transient
disturbances such as three phase short circuit and loss of
excitation were carried out using NEPLAN software and the
results are shown in figures (15-24).
The simulations and results section has been divided into
the two following subsections
1. Simulation results using MATLAB
The results of the above simulation scenario are given in
Fig.7. Flowchart of PSO algorithm tables (III-VII) and graphs of figures (8-13).

Table II: K1-K6 Calculated Values for SMIB


IV. PSS PARAMETERS OPTIMIZATION USING PSO
K1 K2 K3 K4 K5 K6
In order to get the PSS optimal parameters, the critical
0.93166 1.0966 0.339 1.650 -0.08391 0.457
damping index CDI must be minimized, which is defined by
equation 6.
CDI = (1 − ) (6) Table III: Eigenvalues and damping ratios result for MVR case
−σ
ξ = (7) Case Eigen Values Damping Mode
σ +ω Ratio
Where: ζ is damping ratio, σ is the real part of the eigen [Manual -0.055266 1.0 Flux Decay Model
Voltage -0.15661+7.0806i 0.022113 Electromechanical( )
value(λ), and ω is the imaginary part (λ = σ ± jω). Regulator -0.15661-7.0806i 0.022113 Electromechanical( )
There are three tuning parameters of the PSS namely, the (MVR)]
controller gain (KSTAB), lead and lag time constants T1 and Table IV: Eigenvalues and damping ratios result for AVR case
T2. The optimization problem can then be formulated as Case Eigen Values Damping Mode
follows: Ratio
Minimize CDI as in (6) using the parameters given in table I, [Automatic -51.7782 1.0 Flux Decay Model
Voltage -16.047 1.0 Electromechanical( )
Subject to
Regulator( 0.395+7.6735i -0.051408 Electromechanical( )
K ≤ K ≤ K AVR)] 0.395-7.6735i -0.051408 Exciter
T ≤ T ≤ T
T ≤ T ≤ T
Table I: PSO Algorithm Parameters
MVR AVR
PSO Parameter Value/Type 100 2000
Rotor Angle Deviation

Cost Function @(x)SMIBEIG(x) 50 1000


Number of Decision Variable 3
Lower Bound of Variables [0.2 0.02 0.1] 0 0
Upper Bound of Variables [1.5 0.15 50] 0 4 8
Maximum Number of Iteration 100 -50 -1000
Population Size 500 -100 -2000
Inertia Weight 1 Time
Inertia weight Damping Ratio 0.99
Personal Learning Coefficient 1.5 Fig. 8. Dynamic responses of the rotor angle deviation
with and without AVR
Global Learning Coefficient 2
MVR AVR PSS_Calc PSS_PSO
1000 15000 1
10000
Speed Deviation

500 0
5000

Rotor Angle
0 1 2 3 4 5
0 0 -1
0 4 8 -5000
-500 -2
-10000
-1000 -15000 -3
Time
-4 Time
Fig. 9. Dynamic responses of the rotor speed deviation
with and without AVR Fig. 12. Dynamic response of the rotor angle deviation
with AVR and PSS

3 MVR AVR 20000


Field Voltage

10000 PSS_Calc PSS_PSO


2 200
0
1 150

Field Voltage
-10000
100
0 -20000
50
0 4 8
Time 0
Fig. 10. Dynamic response of the field voltage -50
with MVR and AVR 0 1 2
Time(s) 3 4
Fig. 13. Dynamic response of the field voltage with AVR and PSS
Table V: Results of PSS parameters based on
calculations and PSO method
Parameter T1 T2 KSTAB 2. Testing of the PSS-Model using NEPLAN Software
Calculated Method 0.1829 0.0500 6.1620 In this subsection, NEPLAN Software has been used to study
PSO Method 0.2000 0.0200 14.0219 different types of disturbances with two scenarios as follow:
First Scenario: Loss of Excitation
Table VI: Eigenvalues and damping ratios results for AVR and SMIB with proposed PSS and AVR is implemented in
PSS_Calculated case NEPLAN as shown in figure 14. The disturbance assumed to
Case Eigen Values
Damping Mode occurs after one second, which is a loss of excitation
Ratio disturbance last for 70 milliseconds. The transient response
[Automatic -55.3366 1.0 Flux Decay Model
Voltage
results of generator rotor angle, rotor angular speed, field
-1.5828+7.5281i 0.2058 Electromechanical ( )
Regulator(A -1.5828-7.5281i 0.2058 Electromechanical ( ) voltage, and active and reactive power are presented in
VR) and
-14.2665+8.3617i 0.8627 Exciter figures 15-19.
PSS_Calc]
-14.2665-8.3617i 0.8627 PSS

Table VII: Eigenvalues and damping ratios results


for AVR and PSS_PSO case
Damping
Case Eigen Values Mode
Ratio
[Automatic -74.5340 1.0 Flux Decay Model
Voltage -18.3381+20.9066i 0.6594 Electromechanical ( )
Fig. 14. SMIB with AVR and PSS in NEPLAN
Regulator(A -18.3381-20.9066i 0.6594 Electromechanical ( )
VR) and -2.9125+4.9826i 0.5046 Exciter
PSS_PSO] -2.9125-4.9826i 0.5046 PSS

MVR AVR+PSS_Calc AVR+PSS_PSO


3
PSS_Calc PSS_PSO
Rotor Angle
Rotor Speed
Deviation

20 2
0
1
-20
0 2 4
0
Time 0 2 4 6 8 10
Time
Fig. 11. Dynamic responses of the rotor speed deviation Fig. 15. Transient Response of the generator rotor angle
with AVR and PSS results after loss of excitation
1.01 PSS_Calc PSS_PSO
MVR AVR+PSS_Calc
1.009 1.005

Angular Speed
1
Angular Speed

0.999 0.995
0.99
0 2 4 Time(s) 6 8 10
0.989
0 2 4 Time 6 8 10 Fig. 21. Transient Response of the generator rotor angular speed result
after three phase short circuit followed by line removal
Fig. 16. Transient Response of the generator rotor angular
speed result after loss of excitation

PSS_Calc PSS_PSO
10 PSS_Calc PSS_PSO

Field Voltage
3
Field Voltage (pu)

0 -2

-7
-10 0 2 4 Time(s) 6 8 10
0 2 Time(s) 4 6
Fig. 22. Transient Response of the generator field voltage result after
Fig. 17. Transient Response of the generator field three phase short circuit followed by line removal
voltage result after loss of excitation

2 PSS_Calc PSS_PSO
Active Power(pu)

PSS_Calc PSS_PSO
1.3
1
Active Power

0.8
0
0 1 2 Time(s) 3 4 5 0.3
Fig. 18. Transient Response of the generator active power
result after loss of excitation -0.2 0 2 4 Time(s) 6 8 10
Fig. 23: Transient Response of the generator active power result
1 PSS_Calc PSS_PSO after three phase short circuit followed by line removal
Reactive Power(pu)

0.5 2 PSS_Calc PSS_PSO


Power(pu)
Reactive

-0.5
0 1 2 Time(s)
3 4 5 6 0
Fig.19. Transient Response of the generator reactive power 0 2 4 Time(s) 6 8 10
result after loss of excitation Fig. 24. Transient Response of the generator reactive power result after
Second Scenario: Three Phase Short Circuit Cleared by Line three phase short circuit followed by line removal

Tripped: VI. Discussion


In this scenario the SMIB system has been tested for a sever From the optimization and simulation results given in the
disturbance which is three phase short circuit in line 2 -after above tables and figures, the following observations can be
one second very close to the (Tx_Bus) as shown in figure 14, given:
after 70 milliseconds the fault was cleared by removal line 2. 1. When operating the system without the exciter (MVR
The results are shown in figures 20-24. mode) there is an inherent damping in the system and all
real parts of the eigenvalues lies in the left half of S-
plane (with negative sign), also as shown in figures (8)
PSS_Calc PSS_PSO and (9) respectively. But it is clear that the system has
1.8 slow time response (> 90 seconds) with decay
Rotor Angle

oscillatory mode and damping frequency of 1.127


1.3
rad/sec and damping ratio of 0.022113 which is less than
0.8 5% (lightly damped).
0 2 4 Time(s) 6 8 10 2. When the exciter was added (AVR mode), an additional
Fig. 20. Transient Response of the generator rotor angle results negative real eigenvalue introduced to the system, then a
after three phase short circuit followed by line removal negative feedback to the torque-angle loop also
introduced. As a result, the complex pair of eigenvalues VIII. APPENDIX: SMIB System Parameters
moves to the right half of s-plane (instability), with Generator:
negative damping ratios (-0.051408) as shown in table H=3.5s, D=0.00, Rs=0.003, =8s, = 377 rad, Xd=1.81,
IV and the real eigenvalues to the left, which pushes the =0.3,Xq=1.7
system toward instability as shown in the dynamic Transformer and Transmission lines:
response in figures (8) and (9), which shown Req=0.0, Xeq=0.4752, Vin = 0.9008V, Vt=1.0V, th=28.35
monotonically unstable situation due to insufficient Exciter:
damping. KA=200, TA=0.015s
3. Installation of PSS introduced an additional damping to
REFERENCES
the system and all oscillations mitigated, and the system
[1] Graham Rogers, Power System Oscillations, Kluwer
response become very fast as shown in figures (11, 12, Academic, 2000.
and 13). [2] P. Kundur, power system stability and control, McGraw Hill,
4. The damping ratios of electromechanical modes New York, 1994.
( ) were enhanced significantly after PSS [3] D. Mondal, Abhijit Chakrabarti, Aparjita Sengupta, Power
installed. And this is appearing clearly in their dynamic System Small Signal Stability and Control, Elsevier Inc.,
response. 2014.
5. In calculated (Conventional) approach the damping [4] Haifeng Wang, Wenjuan Du, Analysis and Damping Control
of Power System Low-frequency Oscillations, Springer
ratios of the critical modes (electromechanical mode)
science, 2016.
increased from 0.022113 (2.2%) without control to
[5] Lod Tapin, Dr. Ram Krishna Mehta, “Overview and
0.2058 (20.58%) with PSS based on calculated method. Literature Survey of Power System Stabilizer in Power
Also, the settling time reduced to about 4 seconds Systems”, International Journal of Engineering Research and
instead of (>90 seconds) without control and instability Development Vol. 10, Issue 6, PP-60-71, 2014.
case when AVR employed. [6] Sidhartha Panda, N.P. Padhy, “PSO-Based SSSC Controller
6. Furthermore, when PSO approach employed the for Improvement of Transient Stability Performance”,
damping ratios of electromechanical modes increased to International Journal of Electrical, Computer, Energetic.
0.6594 (65.94%) instead of 0.2058 (20.58%) in case of Electronic and Communication Engineering Vol: 1, No: 9,
200
PSO based on calculated method. Also, the settling time
[7] A. Jalilvand, M. daviran Keshavarzi, “PSO Algorithm-Based
reduce to about 2 seconds instead of 4 seconds in case of
Optimal Tuning of PSS for Damping Improvement of Power
calculated method. The electromechanical mode System”, IACSIT International Journal of Engineering and
damping ratios and settling time summarized in table Technology, Vol.2, No 6, December 2010.
VIII for different operational modes. [8] Nikhil Menaria, J. Kasera, “Design of PSS and SVC
Controller Using PSO Algorithm to Enhance Power System
Table VIII: Summary of electromechanical mode; Stability”, IOSR Journal OF Electrical and Electronic
damping ratio and settling time Engineering, Vol 10, Issue 2 Ver II, 2015.
Operation [9] Pothula Jagadeesh, Dr. M, Saiveenaju, “Particle Swarm
MVR AVR AVR+PSS_Calc AVR+PSS_PSO
Mode Optimization Based Power System Stabilizer for SMIB
Damping System”, IEEE, 2016.
2.2113 Negative 20.58 65.94
Ratio (%) [10] Serdar Ekinci, Aysen Demiroren, “PSO Based PSS Design
Settling Not
>90 4 2 for Transient Stability Enhancement”, IU-JEEE Vol.15 (1),
Time (s) available
2015.
[11] Saure, P.W., Pai, M.A. “Power System Dynamics and
Stability”, Prentice Hall, 1998.
VII. Conclusion
[12] Singiresu S. RAO, Engineering Optimization: Theory and
In this paper the parameters of PSS are tuned to mitigate the Practice, John Wiley & Sons, Inc., 2009.
SMIB oscillations employing two different approaches
which are the classical control approach (calculated
method) and modern optimization technique base on
Particle Swarm Optimization (PSO). The simulation results
show that, PSS based on PSO method is more effective in
damping out electromechanical oscillations rather than the
classical method (Calculated method). The
electromechanical damping ratios increased to 65.94% and
the settling time reduced to just 2 seconds based on PSO
method.

View publication stats

You might also like