The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation...
Transcript of The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation...
![Page 1: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/1.jpg)
ESD.86The Weibull Distribution and
Parameter Estimation
Dan FreyAssociate Professor of Mechanical Engineering and Engineering Systems
![Page 2: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/2.jpg)
Weibull’s 1951 Paper
• “A Statistical Distribution Function of Wide Applicability”
• Journal of Applied Mechanics• Key elements
– A simple, but powerful mathematical idea– A method to reduce the idea to practice– A wide range of data
• Why study it in this course?Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
![Page 3: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/3.jpg)
Signs of a Struggle
"The objection has been stated that this distribution function has no theoretical basis. But ... there are – with very few exceptions –the same objections against all other df, at least in so far as the theoretical basis has anything to do with the population in question. Furthermore, it is utterly hopeless to expect a theoretical basis for distribution functions of random variables such as ..."
Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
![Page 4: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/4.jpg)
How was the Struggle Resolved?
The reaction to his paper in the 1950s was negative, varying from skepticism to outright rejection... Weibull's claim that the data could select the distribution and fit the parameters seemed too good to be true. However, pioneers in the field like Dorian Shainin and Leonard Johnson applied and improved the technique. ... Today, Weibull analysis is the leading method in the world for fitting life data.
Abernathy, Robert, 2002, The New Weibull Analysis Handbook
![Page 5: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/5.jpg)
Weibull’s Derivation
nn PP )1(1 −=−
xx
Let’s define a cdf for each link meaning the link will fail at aload X less than or equal to x as P(X≤x)=F(x)
Call Pn the probability that a chain will fail under a load of x
If the chain does not fail, it’s because all n links did not fail
If the n link strengths are probabilistically independent
Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
![Page 6: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/6.jpg)
Weibull’s Derivation
)()1())(1( xnnn ePxF ϕ−=−=−
)(1)( xexF ϕ−−=
xx
o
mu
xxx
exF)(
1)(−
−
−=
A cdf can be transformed into the form
This is convenient because
Among simplest functions satisfying the condition is
The function ϕ(x) must be positive, non-decreasing, and should vanish at some value xu which is often zero but not necessarily.
o
mu
xxx
x)(
)(−
=ϕ
So, a reasonable distribution to try is
![Page 7: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/7.jpg)
A Discussion Point
What is the probability of failure at a load of xu?
o
mu
xxx
exF)(
1)(−
−
−=Try to think of a situation where there is a physical or logical reason to suppose a non-zero value of xu exists.
![Page 8: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/8.jpg)
The Weibull Distribution
o
mu
xxx
exF)(
1)(−
−
−=
More common today to see
Weibull derived
kx
exF⎟⎠⎞
⎜⎝⎛ −
−−= λ
θ
1)(Shape
parameter
Location parameter
Scale parameter
If location parameter=0, we call it the “two parameter”Weibull distribution
Weibull reported for Bofors steel m=2.93. What is k or α?
α
β ⎟⎟⎠
⎞⎜⎜⎝
⎛−
−=x
exF 1)(OR
other notations also used, be
careful!
![Page 9: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/9.jpg)
Influence of the Shape Parameter
sott
etR⎟⎟⎠
⎞⎜⎜⎝
⎛ −−
= η)(
In reliability, the following"reliability function" is commonly defined (the complement of failure)
constant failure rate
![Page 10: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/10.jpg)
Useful Facts about the Two Parameter Weibull Distribution
kx
exF⎟⎠⎞
⎜⎝⎛−
−= λ1)(kxk
exkxf⎟⎠⎞
⎜⎝⎛−−
⎟⎠⎞
⎜⎝⎛= λ
λλ
1
)(
cdf
Mean Variance
SkewMedian
Support
);0[ +∞∈x
⎟⎠⎞
⎜⎝⎛ +Γ
k11λ
( ) k/12lnλ
22 21 μλ −⎟⎠⎞
⎜⎝⎛ +Γ
k
3
323 331
σ
μμσλ −−⎟⎠⎞
⎜⎝⎛ +Γ
k
How is this slide different for the three parameter distribution?
![Page 11: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/11.jpg)
The Procedure Weibull used for Parameter Estimation
If we assume
⎟⎟⎠
⎞⎜⎜⎝
⎛+−=⎥
⎦
⎤⎢⎣
⎡⎟⎟⎠
⎞⎜⎜⎝
⎛− 0
1log)log()(1
1loglogx
xxmxF u
o
mu
xxx
exF)(
1)(−
−
−=
It follows that
1. Starts with a list of values x for strength, size, life ...2. Assigns observed probabilities P≈F(x) to the values3. Transforms the P and x values as indicated above4. Fits a straight line to the data
Which is in the form of a linear realtionship, so Weibull:
![Page 12: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/12.jpg)
Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
x[1.275
kg/mm3]n P log(x-xu)
32 10
33 36
34 84
35 150
36 224
37 291
38 340
39 369
40 383
42 389
⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛− P11loglog
How do we estimate this probability? What are the potential bear-traps?
Note: How do we estimate this value?Recreating Weibull's Bofors Steel Plot
-2.5
-2
-1.5
-1
-0.5
0
0.5
1
0.3 0.5 0.7 0.9 1.1
Copyright © 1951 by ASME. Used with permission.
![Page 13: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/13.jpg)
Plot of Weibull's Dataempirical frequency
0
50
100
150
200
250
300
350
400
450
30 32 34 36 38 40 42 44
Strength
Num
ber o
f sam
ples
What P should we list for xvalue 42?
The convention for probabilty plotting is
(i-1/2)/n
![Page 14: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/14.jpg)
Test for "Goodness of Fit"as Conducted in Weibull's Paper
• Calculates the degrees of freedom10 (bins) -1 – 3 (parameters of the df) = 6
• Calculates the statistic• States the P-value • Comparison to alternative
∑ −=
estimatedestimatedobserved 2
2 )(χ
Note: Table is cumulative, χ2test requires frequency in bin
P=0.49 for Weibull
0 10 200
0.1
0.2
dchisq x 6,( )
0
0.2⎛⎜⎝
⎞⎟⎠
0
0.2⎛⎜⎝
⎞⎟⎠
x5.40
5.40⎛⎜⎝
⎞⎟⎠
,18.17
18.17⎛⎜⎝
⎞⎟⎠
,
P=0.008for Normal Copyright © 1951 by ASME. Used with permission.
![Page 15: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/15.jpg)
Another Good Check on Fit
• Plot the residuals– Check for patterns– Check for uniform variance
e=y-y_hat; plot(y_hat, e, 'or')
![Page 16: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/16.jpg)
Simple Distributions
Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
Copyright © 1951 by ASME. Used with permission.
Copyright © 1951 by ASME. Used with permission.Copyright © 1951 by ASME. Used with permission.
![Page 17: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/17.jpg)
Maximum Likelihood Estimates• Choose the estimate of the parameter θ
to maximize the likelihood function
• Or, if more convenient, maximize log(L(θ))
);,,,()( 21 θθ nXXXfL K=
n observed valuesparameter value(s)
to be estimated
Note: The estimated parameter value is not guaranteed to be the most likely one. It's the data that is made most likely by the parameter estimate.
θ̂
![Page 18: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/18.jpg)
Maximum Likelihood Estimateof the Weibull Distribution
• Write the likelihood function
leads to
Note: But there is no closed form solution in general and numerical methods must be used.
( ) 0),(ln =∂∂ λλ
kL
∏=
⎟⎠⎞
⎜⎝⎛−−
⎟⎠⎞
⎜⎝⎛==
n
i
Xki
n
ki
eXkkXXXfkL1
1
21 ),;,,,(),( λ
λλλλ K
kn
i
kiX
n
/1
1
1⎟⎠
⎞⎜⎝
⎛= ∑
=
λ
leads to( ) 0),(ln =∂∂ λkLk
1)ln()ln(
−
⎟⎟⎠
⎞⎜⎜⎝
⎛−= ∑
∑∑
nX
XXX
k ik
i
ik
i
![Page 19: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/19.jpg)
Some Terms Related to Estimation
• Consistent – for any c
• Unbiased –
• Minimum variance
( ) 0lim =≥−∞→
cPn
θθ)
θθ =)()
E
⎥⎥⎦
⎤
⎢⎢⎣
⎡⎟⎠⎞
⎜⎝⎛
∂∂
= 2)(ln1)var(
θ
θXfnE
)
MLEs are
MLEs are not always
MLEs are pretty close
![Page 20: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/20.jpg)
Concept Questionnumber_of_trials=10;number_of_samples=10;for t=1:number_of_trials
data = wblrnd(2.0,0.8,number_of_samples,1);[paramhat, paramci] = wblfit(data);shape(t)=paramhat(1); scale(t)=paramhat(2);
endmean(shape)mean(scale)
To make the mean(paramhat) equal to the parameters:1) raise number_of_trials2) raise number of samples3) both must be raised4) will not converge exactly in the limit even if both →∞
![Page 21: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/21.jpg)
Point and Interval Estimates
• Up to now, we have discussed point estimates only – a single real value for a parameter
• These are fine, but sometimes one would like to communicate information about degree of confidence
• For this, interval estimates are helpful• e.g., ±95% confidence intervals on paramters
![Page 22: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/22.jpg)
0 0.5 1 1.5 2 2.5
x 104
0
0.5
1
1.5
2
2.5
3
3.5
4Egypt Air Field Data
Hours of Operation
Non-FailureFailure
0 0.5 1 1.5 2 2.5
x 104
0
2
4
6
8
10
12UAL Field Data
Hours of Operation
Non-FailureFailure
Field Data on Engine Life
NOTE: What should we do about the non-failed items? What do we lose if we censor the data?
![Page 23: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/23.jpg)
104 105
0.001
0.003
0.01
0.02
0.05
0.10
0.25
0.50
0.75
0.90 0.96 0.99 0.999
Data
95% Confidence Bounds
UAL Field Failure Weibull Plot
![Page 24: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/24.jpg)
104
0.001
0.003
0.01
0.02
0.05
0.10
0.25
0.50
0.75
0.90 0.96 0.99 0.999
95% Confidence Bounds
Egypt Air Field Failure Weibull Plot
![Page 25: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/25.jpg)
Complex Distributions
Weibull, W., 1951,“A Statistical Distribution Function of Wide Applicability,” J. of Appl. Mech.
Copyright © 1951 by ASME. Used with permission. Copyright © 1951 by ASME. Used with permission.
![Page 26: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/26.jpg)
Looking for Further Evidence of Two Populations
No evidence of bi-modality in fatigue data
Clear evidence of bi-modality in strength data
Copyright © 1951 by ASME. Used with permission.
Copyright © 1951 by ASME. Used with permission.
![Page 27: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/27.jpg)
Reliability Terminology
• Reliability function R(t) -- The probability that a product will continue to meet its specifications over a time interval
• Mean Time to Failure MTTF -- The average time T before a unit fails
• Instantaneous failure rate λ(t)
∫∞
=0
)( dttRMTTF
) tosurvives System tosurvives SystemPr()( tdttt +=λ
∫=−
td
etR 0)(
)(ξξλ
![Page 28: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/28.jpg)
The “Bathtub” Curve
Time
Failu
re R
ate Constant failure rate
R(t)=e-λt
“Infant mortality” periodWear-out period
Failure rate rises with time
![Page 29: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/29.jpg)
Constant Failure Rates“When the system operating time is the MTBF, the reliability is 37%”
- Blanchard and FabryckyR(t)
1.0R(t)=e-λt
0.5
t
MTBF
![Page 30: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/30.jpg)
Series and Parallel Networks
CBA RRRR =
A B C
A
B
C
)1)(1)(1()1( CBA RRRR −−−=−
But ONLY if statistically independent!
![Page 31: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/31.jpg)
Reliability Growth (Duane Model)• As newly designed equipment is refined, it
becomes more reliable• J. T Duane [1964] published regressions for
aerospace items
See Ushakov,1994, Handbook of Reliability Engineering
30 Sep1975 31 Mar
1976
31 Mar1974
31 Mar1977
31 Mar1978
30 Sep1976
30 Sep1977y = 2.439 - .127x
3
2
1
6 7 8 9 10 11 12
x = Ln (Cumulative Flight Hours)
y =
Ln
Cum
ulat
ive
Failu
res
Cum
ulat
ive
Flig
ht H
ours
Reliability improvement of the F-15A, an example of Duane model growth.
Figure by MIT OCW.
![Page 32: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/32.jpg)
Problem Set #5
1. Parameter estimationMake a probability plotMake an estimate by regressionMake an MLE estimateEstimate yet another wayComment on "goodness of fit"
2. Hypothesis testingFind a journal paper uing the "null ritual"Suggest improvements (validity, insight, communication)
![Page 33: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/33.jpg)
Next Steps
• Between now and Weds– Read Gigerenzer "Mindless Satatistics"
• Wednesday 10:30-noon– Session on Hypothesis testing
• Friday– Recitation to support PS#5
• Monday – PS#5 due– Session on XXX
![Page 34: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/34.jpg)
The Wonderful One-Hoss Shayby Oliver Wendell Holmes
HAVE you heard of the wonderful one-hoss-shay, that was built in such a logical way it ran a hundred years to a day…
Now in building of chaises, I tell you what, there is always somewhere a weakest spot,-- in hub, tire, felloe, in spring or thill, in panel, or crossbar, or floor, or sill…
![Page 35: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/35.jpg)
But the Deacon swore .. he would build one shay to beat the taown 'n' the keounty 'n' all the kentry raoun'; It should be so built that it couldn' break daown! --"Fur," said the Deacon, "t 's mighty plain thut the weakes' place mus' stan' the strain; 'n' the way t' fix it, uz I maintain, is only jest t' make that place uz strong uz the rest"…
So the Deacon inquired of the village folk where he could find the strongest oak, That could n't be split nor bent nor broke,--
![Page 36: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/36.jpg)
Eighteen hundred and twenty came;-- Running as usual; much the same. Thirty and forty at last arrive, And then come fifty, and fifty-five…
There are traces of age in the one-hoss-shay–A general flavor of mild decay, But nothing local, as one may say. There couldn't be,--for the Deacon's art had made it so like in every part that there wasn't a chance for one to start…
And yet, as a whole, it is past a doubt in another hour it will be worn out!
![Page 37: The Weibull Distribution and Parameter Estimation The Weibull Distribution and Parameter Estimation Dan Frey Associate Professor of Mechanical Engineering and Engineering Systems](https://reader030.fdocuments.in/reader030/viewer/2022021509/5aa0f1767f8b9a8e178eb731/html5/thumbnails/37.jpg)
First a shiver, and then a thrill, Then something decidedly like a spill,--…What do you think the parson found, when he got up and stared around? The poor old chaise in a heap or mound, as if it had been to the mill and ground! You see, of course, if you're not a dunce, how it went to pieces all at once,-- all at once, and nothing first,--just as bubbles do when they burst.
End of the wonderful one-hoss-shay. Logic is logic. That's all I say.