Data Reduction
9
from Table E.4
N=2
0.571a 2 = 0.886 w
Range w = 45,000 - 40,000 = 5,000 psi
(1.34)
J0.886(5,000)
a = 88.081 psi
The calculation Table E.5
,/0.8862(5,000)
~r-- v
a = 88.091 psi
If A and B values for two samples are calculated with KA = 37.094, from
Table E.1 and K~=20.581.
~A =/~- K~ a
aA = 42,500 -- 37.094 (88.09 psi)
(1.35)
aA = 39,232 psi
a~ = 42,500 -- 20.581 (88.09 psi)
(1.36)
a~ = 40,687 psi
These are compared to actual values of 42,000 and 43,000 for A and B basis.
These are now closer and could be used for designing.
5. k~rger Data Samples N>30
When more data is available, say >20-30, the estimates in Example 1.1 get
better for the t and x calculations to get a mean and standard deviation.
When ~ and s are derived from a log normal plot the highs and lows
for ~ and a or ~ can be placed on the log normal plot with the original data
and limits on the expectations can be made.
B. Weibull Distribution
The confidence limits for the infinite sample size Weibull curves Eqs (1.2),
(1.14) and (1.15) from the test samples [1.1,1.21] are shown in Table
The test sample B and 0 are estimated from a plot or a computer