Exercise 5

1. Derive the probability that the insulation structure will pass the 15/2 test (2 flashovers in 15
impulses). What is the probability with 125 kV test voltage when 50% breakdown voltage is
130 kV and standard deviation of breakdown is 3 %.

2. In order to determine an insulator’s 50% breakdown voltage, a series of tests were conducted
where voltage was increased steadily from 200 kV until breakdown occurred. The following
breakdown values were obtained (in kV): 478, 487, 503, 499, 481, 518, 530, 512, 495, 480,
471, 535, 505, 507, 491, 498, 506, 521, 482, 493. Determine the insulator’s 50% breakdown
voltage and its standard deviation using the probability sheet. Also, calculate the mean and
experimental standard deviation using the measured data.

3. For the purchase of a 123 kV air insulating device, the impulse test voltage was set according
to IEC as 450 kV. According to IEC, a 450 kV test voltage for an impulse voltage test
correlates to 10 % breakdown probability. Acceptance testing was conducted using up and
down method. The test produces the following document (x = breakdown, o = no breakdown):

494 kV x x
480 kV x o o x
466 kV x o o x x x
452 kV x o o o x o o
438 kV o o o
424 kV o
410 kV o

Withstand strength is assumed to follow normal distribution when standard deviation   3% .

Did the device pass the test?

Hint: Respective breakdown voltage Up for breakdown probability p can be estimated using
the mean and standard deviation according to the following table.

U p  U50  k

p/% 50 15,9 10 2,3 0,13

k 0 1 1,3 2,0 3,0
The table shows (0,1) normal distribution function:

( ) ˆ probability that (0,1) distributed random value falls within  ... .

(  ,  ) normal and (0,1) normal random value is correlated as:

x


where x ̂ (  ,  ) distribution random value.
 Answers
1.

Let p ̂ probability that the given voltage will produce breakdown.

Binomial distribution probability:

 n  n! 
   
 p   n  p  ! p ! 
15 
P0  P(0 breakdown)    p 0 (1  p )15  (1  p)15
0
15 
P1  P(1 breakdown)    p1 (1  p )14  15 p (1  p)14
1
15 
P2  P(2 breakdown)    p 2 (1  p )13  105 p 2 (1  p)13
2
P(at most 2 breakdown)  P0  P1  P2
2
15 
P     p i (1  p )15i  (1  p)13 (1  13 p  91 p 2 )
i 0  i 

x 125  130

   1, 28  P(1, 28)  (1  0.799) / 2  0.10  p
 0, 03 130  ( )

Substitute p into P:
P  0,913 (1  1,3  0,91)  0,8159  82%
2.
U2 / kV pc / % Arrange the measured values in order of magnitude. Calculate for each value
471 2,5 the respective cumulative probability pc.
478 7,5
480 12,5 Cumulative probability for each breakdown value can be attained by starting
481 17,5 from 0 and 5 % midpoint and increasing the value 5 % for the subsequent
482 22,5 breakdown voltage. (100% / n = 100% / 20 = 5%)
487 27,5
491 32,5 Plot the values on the probability sheet and draw the trend line.
493 37,5
495 42,5
498 47,5
499 52,5
503 57,5
505 62,5
506 67,5
507 72,5
512 77,5
518 82,5
521 87,5
530 92,5
535 97,5

The voltage correlating to the 50 % breakdown probability can be observed from the trend line:

Mean: U 50%  499kV  500kV

Standard deviation can also be observed from the probability sheet:

 STD is the difference between 50% and 16% values

Normal distribution:   1  ( )  0.6826  68%

100% - 68% = 32%

32% / 2 = 16% 482 kV
500 kV – 482 kV = 18 kV

STD:   18kV (ˆ 3,5%)

Estimates using the given data:

n = 20
1
Mean: U l 
n
 ul  499, 60kV  500kV
1
Experimental standard deviation: s 
n 1
 (U  U )2  17,60kV  18kV ( ˆ 3,5%)
3.

In order to pass, the 50 % breakdown voltage of the device has to be larger than calculated IEC 50%
test voltage.
U50%,device  U50%test
According to IEC, a 450 kV test voltage for an impulse voltage test correlates to 10 % breakdown
probability.

IEC test voltage U 50% :

U p  U50%  k
U10%  U50%  1,3

  3%  0.03U50%

U10%  U50%  1.30.03U50% 

U10%,test 450kV
 U 50%,test    468,3kV
1  1,3  0, 03 1  1,3  0, 03

”Up and down” test:

U 50 
n U i i

n i

where ni is the number of pulses at Ui , when at the beginning of the pulse series only ni  2 levels
are considered.
494 kV x x
480 kV x o o x
466 kV x o o x x x
452 kV x o o o x o o
438 kV o o o
424 kV o
410 kV o

3  438  7  452  6  466  4  480  2  494

U 50%,device  kV  462,8kV
3 7  6 4 2

Because U50%,device  U50%,test (462.8 kV < 468.3 kV), the device does not pass the test.