Ieee std 1366-2012 (Revision of ieee std 1366-2003) ieee guide for Electric Power Distribution Reliability Indices

B.8 Five years of data 
From a statistical point of view, the more data used to calculate a threshold, the better. However, the 
random process producing the data changes over time as the distribution system is expanded and operating 
procedures are varied. Using too much historical data would suppress the effects of these changes. 
The addition of another year of data should have a low probability of changing the MED classification of 
previous years. A result from order statistics gives the probability that the kth largest value in m samples 
will be exceeded f times in n future samples. It is given in Eq. (B.6): 
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
−
+
+
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
⎟⎟
⎠
⎞
⎜⎜
⎝
⎛
−
+
=
f
k
n
m
n
f
n
k
m
f
k
n
k
p
n
k
m
f
,
,
|
(B.6) 
For example, if M = 3 years of data, then m = 1 095 samples. If f = 3 MEDs/year, then the largest non-MED 
is the k = 1 095 - 9 = 1 086
th
ordered sample. The probability of f = 3 days in the next year of n = 365 
samples exceeding the size of the largest non-MED is found from the equation to be 0.194 (19.4%). In 
Figure B.5, p is plotted against M for several values of f. 
Authorized licensed use limited to: North China Electric Power University. Downloaded on February 16,2022 at 10:52:41 UTC from IEEE Xplore. Restrictions apply. 

---

IEEE Std 1366-2012 
IEEE Guide for Electric Power Distribution Reliability Indices 
Copyright © 2012 IEEE. All rights reserved. 
30
0.05
0.1
0.15
0.2
0
5
10
15
M , years
p
(f
MED
s)
f =3
f =5
f =10