Documentos de Académico
Documentos de Profesional
Documentos de Cultura
H.A.Darwish, M.A.Rahman,
Fellow, IEEE
Memorial University of Newfoundland,
Faculty of Engineering and Applied Science,
St . John's, Newfoundland.
CANADA, AlB-3X5.
Menoufiya University,
Faculty of Engineering,
Shebin El-Kom,
EGYPT.
I. INTRODUCTION
1187
2
Q
t2-i
1 . 5 ................i...................................
i..................
I O
15
20
25
30
m u l t i p l e o f tap value current ( I )
35
40
(a)
2.
................. j ...................................
.................
(2)
0,5
~---~r
1 . 5 ..........................
a
!..........................................
.....................
............................
j .................
. . . . . . .
10
15
m u l t i p l e of
20
25
30
35
40
tap v a l u e c u r r e n t ( I )
(bl
(3) Calculate the twelve relations of Tp-I and then store them.
(4) Start with a high number of the linear portions m, say 20,
and identify their corresponding terminating current
values.
(5) Calculate the n+ 1 polynomial coefficients at each of the
m+ 1 current values using the curve fitting function.
(6) Calculate twelve values of T, by substituting in the precalculated polynomial equations with the corresponding
TDS taking into considerationthat the value of T, of the
base curve is equal to 1. This is repeated at each of the
m + 1 of the terminating current values.
(7) Calculate the tripping times (tr,) for all the twelve curves
by manipulating both tb values and the calculated T,. from
step 6 in (2) and using linear interpolation, if required.
(8) Calculate the mean value of the absolute error between qc
and 5.
(9) Reduce the number m and repeat from steps 5 to 8.
Fig. 2 shows the effect of varying number of linear
portions from 0 to 19 on the mean value of the absolute error
considering n is equal to 6 and the base curve is the curve of
1188
number
of
l i n e a r p o r t i o n s [m)
(a)
polynomial
degree [n)
(a)
polynomial
degree [n)
(bl
D. Table Length
The data length of the table corresponding to the base curve
is one of the important parameters during the implementation
stage of these standard curves in the microprocessors. Also,
the data table should be stored inside the processor RAM
equi-spaced and its length should be multiples of two. This
facilitates the job of the programmer and reduces the storage
memory size as the corresponding current values will be
implicitly defined. Moreover, it reduces the on-line calculation time for the processor. That is because, the fault current
value equivalent to a number stored in the processor, will
directly offset the jump in the look-up table. This requires
only shifting the number and initializing a stack pointer.
Furthermore, no division operation is required for the linear
interpolation process when the current value does not match
a specific value in the look-up table. That is because, the
table values can be decided on the current difference before
storing them and then division will be manipulated by
multiplication.
Fig. 4 shows the effect of reducing the data length of the
look-up table on the mean value of the absolute error. This
error is also computed over 5964 values. The linear interpolation is used in the simulation for the time values corresponding to the in-between values of the current. Fig. 4 shows the
relatively flat error value from 512 to 128 length of table
data. Therefore, the suitable data length is selected as 128
which corresponds to a fault current values started from zero
up to 40 times the tap value current. Consequently, the table
values corresponding to the current magnitudes lower than the
pick up value (
are filled with zeros or do not care. So,
the actual used ength of data in the look-up table for the
standard $-I curves of the overcurrent relay CO11 is 124
value. The model parameters for this extremely inverse
characteristics are:
Part 1 A table of 124 value length of the tripping times corresponding to 6-1 curve of TDS equals to 6.
Part 2 A 35 relating coefficients corresponding to 5 polynomial equations of the six degree are computed at
current values of 1.25, 2, 10, 20, and 40. Table I1
contains the values of the coefficients obtained from
the curve fitting.
..............
2 ......................
0
0
.......................
........................
.......................
.......................
.................
.......................
........................
.......................
.......................
......................
100
300
400
200
length o f Table data
50 0
600
Fig. 4. Effect of varying length of data in the table on the mean value of
error.
TABLE
I
Set (1)
112
1.9679%
15.55%
Set (2)
5 at I=1.25, 2, 3 , 16, 40
2.3361 %
13.3%
Set Number
Set (3)
1.2664%
15.62%
Set (4)
1.1175%
10.31%
set (5)
5 at I=1.25, 2, 10, 2 0 , 40
1.2153%
9.14%
Set (6)
1.3441%
11.82%
Set (7)
1
I
I
Set (8)
Set (9)
Set (10)
I
I
I
I
1.1736%
I
I
I
1.2197%
1.2905%
1.3391%
7.94%
I
I
9.59%
12.47%
Set (11)
10
1.2558%
14.62
Set (12)
11
1.3182%
14.80
1190
10.06%
1-1
111. EFFECTIVENESS
OF THE
PROPOSED
MODEL
bII
zn-1
b2
bl
t =b +-+-+...+O
where, bo
I-1
12-1
(5)
".
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
"I
log(t,) =bo+L
+b2 + ...+bn
(logo (logo2
(logo*
b,
(I-1)
tr=bo+-+-
b2 +...+(Z-1)2
1621
lo"
(4)
bn
(I-l)*
,:
:'
i
,
.:f
i i i i i
i.
lo'
m u l t i p l e a f tap value current ( I )
(3)
++++
---
standardvalues
using previous method in reference [8].
using the proposed method
Fig. 5. Standard and computed 4-1 curves.
TABLE11
TABLE
HI
COMPARISON BETWEEN
THE PROPOSED
Model type
Mean value of
the absolute error
Presented model
1.25%
E
I
M
o
d
.
e
o
181
f
r
e
f
l
I 1.095%
Number of
polynomial
coeff.'~
Number of stored
current values
35
868
Total stored
numbers
124
I
1191
164
868
Number of valid
curves
Max. absolute
error measured
for time values
Idkite
7.945%
10.21%
Infinite
VI.
++ + +
-
REFERENCES
standard values
simulated values
Fig. 6. Standard and fitted $-I base curve.
TABLE
N
VALUES OF THE COEFPICIENTS OF THE BEST FIT OF THE
BASECURVE
V. CONCLUSIONS