Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Intelligent Underfrequency Load Shedding For 500kV Java-Bali Electrical Power System

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

1

Intelligent Underfrequency Load Shedding for 500kV Java-Bali Electrical


Power System
Dimas Fajar Uman P1) Ontoseno Penangsang2) Adi Soeprijanto3) Muhammad Abdillah4)
1) Department of Electrical Engineering, Faculty of Industrial Technology
ITS Surabaya Indonesia 60111, email: dimas10@mhs.ee.its.ac.id
2) Department of Electrical Engineering, Faculty of Industrial Technology
ITS Surabaya Indonesia 60111, email: zenno_379@yahoo.com
3) Department of Electrical Engineering, Faculty of Industrial Technology
ITS Surabaya Indonesia 60111, email: adisup@ee.its.ac.id
4) Department of Electrical Engineering, Faculty of Industrial Technology
ITS Surabaya Indonesia 60111, email: abdilah@elect-eng.its.ac.id

Abstract-- This paper presents a new method for solving underfrequency load shedding (UFLS) problem by using fuzzy logic
controller. Gradient frequency method is used for gaining the active power deficit, then fuzzy decision making will breakdown the
amount of the active power energy in every bus needed for load shedding. This method is implemented at 500 kV Java-Bali electrical
power system. The results showed during the disturbance for a trip of 1510 MW Paiton generation, the system voltage is decrease, the
load decreased 205 MW, df/dt is 0.5967 Hz/s and the total load to shed is 1003 MW. Comparing with load shedding from scheme of
PLN Java-Bali, the above disturbance give df/dt 0.6 Hz/s and total load to shed 1181 MW.

Index Terms-- underfrequency; electrical power system; load shedding;fuzzy decision making.

described. Meanwhile, Section IV applying the proposed


I. INTRODUCTION method to the system. The simulation results are discussed in
or emergency condition during disturbance, UFLS is the Section V. Finally, there are conclusions.

F only way to prevent an electrical power system (EPS)


from black out. Black out for underfrequency condition II. FUNDAMENTAL THEORY
due to deficit of active power generation. The fluctuation of The calculation for frequency respond of large power
frequency is represents the difference between the deficit systems can used low order respond model. In [3] a formula
active power generation and the active load. derived for searching the equivalent inertia constant for more
In Java-Bali EPS imbalances between active power than one generator become single equivalent generator as
generation and active load first taken over by the primary
control on generator, if imbalance still exist the second action n

is to active the governor and LFC control from generation. If  HiSi


i 1

the first and the second action cannot restore the balance of
Heq  n (1)
active power, then it needed a shedding.
 Si
i 1

UFLS schemes according to [1] can be categorized into three The quantities in (1) are:
groups:
 the traditional UFLS schemes; Heq : inertia constant of an equivalent generating unit
 the semi-adaptive UFLS schemes; H : inertia constant of the ith generating unit
 the adaptive UFLS schemes.
i

Si : apparent rated power of the ith generating unit


Traditional and adaptive schemes for UFLS are used in
Java-Bali EPS. But there is still no method to predict the drop n : number of all generating units
of frequency during a disturbance.
Many methods can be used to predict the drop of frequency A. Low-order Frequency Respond Model
during the disturbance conditions. The latest method uses
gradient frequency [2]. In paper [2] the gradient frequency is The low order respond can be derived from the first swing
used to calculate the active power deficit to estimate the total equation. From the equation the frequency gradients depend in
load to shed. The frequency-response model has been defined two parameters, there are frequency systems and the second is
and presented in [3]. active power deficit.
In many paper likes [2] and [4] give only the total loads to First, the frequency systems used is the mechanical rotating
be shed. It is better if there is a description for each bus about frequency of the equivalent generator. Second, considering [5]
the load must be shed. – [7] the active power deficit can be shown as
In this paper will be introduced the new intelligent UFLS
gradient frequency method using fuzzy decision making to 2 H eq df
Pdef  Hz S (2)
decide the amount for load shedding. eq
fN dt
In Section II, A brief discussion is presented on problem
formulation of the system. In Section III, proposed method is The quantities in (2) are:
2

Pdef : active power deficit in MW Pdef m


 Ui   100 i

Ptrip  100   PL 0 i    1 (5)


fN : nominal frequency in Hz PL 0 i 1  U 0 i   PL 0
df Hz
: gradient frequency
dt
 U  
i
df Hz fN m

P  PL 0     U 
i

There is four factors the influence the active power deficit  PL 0 i  1 (6)
 
tur
dt 2 H eq S eq i 1 0i
value [2]:
1. Initial mechanical turbine powers ( Ptur , 0 )
2. Initial active load power in the system ( PL 0 )
3. Spinning reserve ( Ptur ) III. METHODS
4. Active load power changing due to voltage variations
( PL )
Start

The active load power changing due to voltage variations


depend on the type of load [8].
Input data of
generation,
transmission line
& bus data
B. Gradient Frequency for Drawbacks the Frequency
Respond Model
The major drawbacks of using the frequency respond Perform load flow
calculation (Newton
model for UFLS purposes is load’s voltage dependence. On Rapshon Method)

the other hand, voltage drops rapidly when there is disturbance


in system, which is reflected in the load’s active and reactive
power.
Initializations Ui0,
The prompt reaction of voltage change to the active and Ptur0 and PL0

reactive load power cannot be neglected [2]. As the voltage


changing reflected in load changing and it reflects in the
amount of active power deficit and also in the calculating of
Perform disturbance load flow
frequency gradient. calculation after disturbance
(Newton Rapshon Method)
The mathematical model for the voltage dependent load
given in [8] as follows
Initializations Ui,
i i
m
 Ui  m
 Ui  Ptur, PL

PL  PL 0i   QL  QL 0i   (3)


i 1  U 0i  i 1  U 0i  Calculate:
Pdef
(using equations 4)
Pshed
The quantities in (3) are: (using equation 5)
df/dt
PL : active power value due to voltage variations (MW) (using equation 6)

QL : reactive power value due to voltage variations Perform Fuzzy


Decision Making
(MVAR) For gaining Pshed in
each bus
U 0i : voltage value before disturbance (p.u)
Ui : voltage value after disturbance (p.u)
Yes
i
Chage
: active power dependence factor Disturbance

i : reactive power dependence factor No

Stop
Knowing the load change after the disturbance, the value
of the deficit active power can be calculated
Figure 1. Flowchart for Intelligent Load Shedding
Pdef  Ptur  PL (4)
For intelligent UFLS method to the problem are shown
Inserting the load change due to voltage variations (3) to (4)
below:
and combine in (2), the value of the load to be shed Ptrip (in 1. Determining data on the 500 kV Java-Bali electrical power
percent) and the frequency gradient can be calculated system for the power flow calculation.
3

2. Performing power flow calculations to obtain the value of μc (Ui)


bus voltages, total generation and power load using Newton
Rapshon method.
3. Initializing the value of bus voltages, total generation and 1
power load for initial conditions.
4. Perform disturbance (loss generation) and performing power
flow calculations to obtain the value of bus voltages, total
generation and load power after disturbance using Newton
Rapshon method.
5. Initializing the value of bus voltages, total generation and
load power for conditions after disturbance. Vi (Ui)
Vmin Vmax
df Hz
6. Calculate Pdef , Ptrip and using equation 4, 5 and 6.
dt Figure 2. Membership function for FDM
7. Perform FDM to gaining Ptrip in each bus.
8. Updating the disturbance, if there is any updating go back to
step no 4. IV. EXPERIMENTAL RESULTS

A. Fuzzy Decision Making


A. 500 kV Java-Bali Electrical Power System Data
The fuzzy set theory has been implemented to derive
The 500 kV Java-Bali electrical power system consists of 8
efficiently a solution from the set of non-dominated solutions
generators, 30 lines, and 25 buses. The total numbers of
[9]. The solutions from fuzzy decision making function is
installed loads are 10,361 MW and 3,565 MVAR. Load and
given by the membership function μc.
one line diagram are shown in Table.1 and Fig. 3.
For the fuzzy variable, here we use the voltage difference as
MVA base of 1000 MVA and kV base of 500 kV are used in
a variable. If the difference is high then μc=1 and if there is no
this paper as the base value of the 500 kV Java-Bali electrical
difference then μc=0. This value indicates a bus near the
power system. Loads and generators are shown in table 1.
disturbance. The membership function of μc(Ui) defined:
TABLE I. LOADS DATA AND GENERATOR ON 500 KV JAVA-BALI
V max   c Ui  ELECTRICAL POWER SYSTEM [10]
μc(Ui) = (7) Loading Generator
V max  V min Bus
Bus Name
Bus
P Q P Q
No Types
(MW) (MVAr) (MW) (MVAR)
1  c Ui   0 1 Suralaya Swing 219 67 3.211,6 1074.1
2 Cilegon Load 333 179 0 0
FDM(Ui) =  c Ui  0   c Ui   1 (8) 3 Kembangan Load 202 39 0 0
4 Gandul Load 814 171 0 0
0  c Ui   0 5 Cibinong Load 638 336 0 0
6 Cawang Load 720 217 0 0
7 Bekasi Load 1126 331 0 0
Then the normalized membership function for FDM 8 Muara Tawar Generator 0 0 1.760,0 645.0
9 Cibatu Load 1152 345 0 0
FDM Ui 
10 Cirata Generator 597 201 948,0 200.0
FDM = FDM  m
(9) 11 Saguling Generator 0 0 698,4 150.0

 FDM U 
12 Bandung Load 477 254 0 0
i Selatan
i 1 13 Mandiracan Load 293 65 0 0
14 Ungaran Load 193 118 0 0
The output from FDM is a value from load in each bus. 15 Tanjung Jati Generator 0 0 1321,6 90.0
16 Surabaya Barat Load 508 265 0 0
17 Gresik Generator 127 92 900,0 366.3
18 Depok Load 342 95 0 0
19 Tasikmalaya Load 133 33 0 0
20 Pedan Load 365 101 0 0
21 Kediri Load 498 124 0 0
22 Paiton Generator 448 55 3180,0 917.3
23 Grati Generator 180 132 398,6 100.0
24 Balaraja Load 732 287 0 0
25 Ngimbang Load 264 58 0 0

Transmission line parameters is measured in pu (per unit).


Data of transmission line 500 kV Java-Bali electrical power
system is shown in Table. 2.
4

TABLE II. TRANSMISSION LINE DATA OF 500 KV JAVA-BALI In this simulation the disturbance occur in Paiton power
ELECTRICAL POWER SYSTEM [10]
plant, the magnitude of the disturbance is 1510 MW, and the
No. Line R (p.u.) X (p.u.) ½ B (p.u.) spinning reserve is 300 MW. The result given:
1 1 2 0.000626496 0.007008768 0
2 1 24 0.003677677 0.035333317 0 B. UFLS scheme from PLN Java-Bali
3 2 5 0.013133324 0.146925792 0.003530571
4 3 4 0.001513179 0.016928308 0 Java-Bali EPS have a UFLS scheme listed below:
5 4 18 0.000694176 0.006669298 0
6 5 7 0.004441880 0.042675400 0 TABLE III. DATA UFLS OF 500 KV JAVA-BALI EPS [11]
7 5 8 0.006211600 0.059678000 0
8 5 11 0.004111380 0.045995040 0.004420973
9 6 7 0.001973648 0.018961840 0
10 6 8 0.005625600 0.054048000 0
11 8 9 0.002822059 0.027112954 0
12 9 10 0.002739960 0.026324191 0
13 10 11 0.001474728 0.014168458 0
14 11 12 0.001957800 0.021902400 0
15 12 13 0.006990980 0.067165900 0.006429135
16 13 14 0.013478000 0.129490000 0.012394812
17 14 15 0.013533920 0.151407360 0.003638261
18 14 16 0.015798560 0.151784800 0.003632219
19 14 20 0.009036120 0.086814600 0
20 16 17 0.001394680 0.013399400 0
21 16 23 0.003986382 0.044596656 0
22 18 5 0.000818994 0.007868488 0
23 18 19 0.014056000 0.157248000 0.015114437
24 19 20 0.015311000 0.171288000 0.016463941
25 20 21 0.010291000 0.115128000 0.011065927
𝑑𝑓
26 21 22 0.010291000 0.115128000 0.011065927 In the simulation the values of 𝐻𝑧 is 0.5967 Hz/s. If we use
𝑑𝑡
27 22 23 0.004435823 0.049624661 0.004769846
28 24 4 0.002979224 0.028622920 0
UFLS data from 500 kV Java-Bali EPS, the load must be
29 25 14 0.023479613 0.225580588 0.010097035 shedding is 1181 MW consist of:
30 25 16 0.005966652 0.057324466 0  From RJKB 463 MW
 From RJBR 194 MW
A one line diagram of 500 kV Java-Bali electrical power  From RJTD 156 MW
system is shown in Fig. 3.  From RJTB 299 MW
C. Intelligent UFLS
The loads to be shed are listed in table IV:
TABLE IV. DATA UFLS OF PROPOSED METHOD [11]
No bus Bus Name Load Shedding (MW)
1 Bus Generator Suralaya 0
2 Cilegon 0
3 Kembangan 48.828
4 Gandul 47.2159
5 Cibinong 45.4815
6 Cawang 51.9089
7 Bekasi 60.1495
8 Bus Generator Muara Tawar 0
9 Cibatu 26.6796
10 Bus Generator Cirata 32.17
11 Bus Generator Saguling 32.17
12 Bandung Selatan 51.5927
13 Mandiracan 74.7447
14 Ungaran 78.7922
15 Bus Generator Tanjung Jati 0
16 Surabaya Barat 13.48
17 Bus Generator Gresik 0
18 Depok 49.3349
19 Tasikmalaya 119.184
20 Pedan 118.2175
21 Kediri 97.6892
22 Bus Generator Paiton 0
Figure 3. One Line Diagram in 500kV Java-Bali electrical power system 23 Bus Generator Grati 0
[10] 24 Ngimbang 22.5327
25 Balaraja 32.9561
Total 1003.1274
5

V. CONCLUSIONS
In this paper, the proposed method (Intelligent UFLS) is
used to determine the amount load must be shed in each bus.
Test results by using the 500 kV Java-Bali EPS show that the
Intelligent UFLS method gives better result to minimize the
total load to shed than the standard given by PLN Java-Bali.
The simulation show that for the disturbance of 1510 MW and
the spinning reserve is 300 MW the total load to be shed is
1003.1274 MW.

VI. REFERENCES
[1] B. Delfino, S. Massucco, A. Morini, P. Scalera, and F. Silvestro,
“Implementation and Comparison of Different Underfrequency Load-
Shedding Schemes,” in Proc. Power Engineering Society Summer
Meeting 2001, Jul. 15–19, 2001, vol. 1, pp. 307–312.
[2] Urban Rudez, Rafael Mihalic, “Analysis of Underfrequency Load
Shedding using a Frequency Gradient”, IEEE transactions on power
delivery, vol. 26, no. 2, april 2011.
[3] P. M. Anderson and M. Mirheydar, “A Low Order System Frequency
Response Model,” IEEE Trans. Power Syst., vol. 5, no. 3, pp. 720–729,
Aug. 1990.
[4] H. Seyedi, M. Sanaye-Pasand “Design of New Load Shedding Special
Protection Schemes for a Double Area Power System”, American
Journal of Applied Sciences 6 (2): 317-327, 2009.
[5] M. S. Pasand and H. Seyedi, “New Centralized Adaptive Under
Frequency Load Shedding Algorithms,” in Proc. Power Engineering-
Large Engineering Systems Conf., Oct. 10–12, 2007, pp. 44–48.
[6] M. Etezadi-Amoli, “On Underfrequency Load Shedding Schemes,” in
Proc. Power Symp.—22nd Annu. North American Conf., Oct. 15–16,
1990, pp. 172–180.
[7] A. Li and Z. Cai, “A Method For Frequency Dynamics Analysis And
Load Shedding Assessment Based On The Trajectory Of Power System
Simulation” in Proc. 3rd Int. Conf. Electric Utility Deregulation and
Restructuring and Power Technologies (DRPT), Apr. 6–9, 2008, pp.
[8] Chapter 14. Federico Milano, “Power System Modelling and Scripting,”
Electrical Power and Energy Systems, Springer-Verlag London 2010.
[9] P.K. Hota, A.K. Barisal, R. Chakrabarti, “Economic Emission Load
Dispatch Through Fuzzy Based Bacterial Foraging Algorithm”, Elsevier,
Electric Power and Energy Systems, 2010.
[10] Refi Aulia Krisida, Adi Soeprijanto, Heri Suryoadmojo, “Optimal
Reactive Power and Voltage Control in Java-Bali 500 kV
Interconnection System Using Quantum Behaved Particle Swarm
Optimization”, undergraduate research final report, Institut Teknologi
Sepuluh Nopember, 2011.
[11] PT PLN P3B Jawa-Bali, “Greed Code(aturan Jaringan)” PT PLN, 2007.

You might also like