chapter 7 overview of distributions and statistical processes
TRANSCRIPT
3/18/2013
1
Chapter 7
Overview of Distributions and
Statistical Processes
Introduction
• General knowledge of distributions can be helpful
when choosing a good test / analysis strategy to
answer specific question.
• Overview of some statistical distributions
• Hazard rate, homogeneous Poisson process (HPP),
and the nonhomogeneous Poisson process (NHPP)
with Weibull intensity
3/18/2013
2
7.1 An Overview of
the Application of Distributions
• The population of a continuous variable has an
underlying distribution (parent distribution), which
could be represented as a normal, Weibull, or
lognormal distribution.
• To characterize a population, samples can be
taken. A distribution that describes sampling
statistics is called the sampling distribution, or
child distribution.
7.1 An Overview of
the Application of Distributions
Continuous response:
• When making statement about the mean, the
shape of the parent distribution does not usually
need to be considered. (Central limit theorem)
• When estimating the percentiles of the population,
knowledge about the shape of the population
distribution is important. (normal, Weibull, or
lognormal distribution)
• Normal distribution is often encountered in
statistics. It is applicable to many sampling
statistical methodologies with continuous response.
3/18/2013
3
7.1 An Overview of
the Application of Distributions
Discrete response:
• Binomial distribution is another common distribution
for attribute pass/fail conditions.
• Hypergeometric distribution (similar to binomial)
addresses the situation when the sample size is
large relative to the population.
• When multiple defects or failures can occur, the
Poisson distribution is useful to design tests.
7.1 An Overview of
the Application of Distributions
Reliability tests: how long the item will perform before
failure?
• For not repairable items: the response of interest is
percentage failure as a function of usage. (Weibull,
lognormal distributions)
• For repairable items: the response of interest is a
failure rate model (intensity function).
• HPP for constant failure rate
• NHPP for increasing or decreasing instantaneous
failure rate
3/18/2013
4
7.2 Normal Distribution
The normal distribution (also called the Gaussian distribution) is by far the most commonly used distribution in statistics. This distribution provides a good model for many, although not all, continuous populations.
• A dimension on a part is critical. This critical dimension is measured daily on a random sample of parts from a large production process. The measurements on any given day are noted to follow a normal distribution.
• A customer orders a product. The time it takes to fill the order was noted to follow a normal distribution.
7
Probability Density Function, Mean, and
Variance of Normal Distributions
The probability density function of a normal
population with mean and variance 2 is
given by
𝑓 𝑥 =1
𝜎 2𝜋𝑒𝑥𝑝 −
(𝑥−𝜇)2
2𝜎2, −∞ < 𝑥 < ∞
If X ~ N(; 2), then the mean and variance
of X are given by 𝜇𝑋 = 𝜇, 𝜎𝑋2 = 𝜎2
8
3/18/2013
5
Probability Density Function, Mean,
and Variance of Normal Dist.
9 http://upload.wikimedia.org/wikipedia/commons/thumb/7/74/Normal_
Distribution_PDF.svg/720px-Normal_Distribution_PDF.svg.png
Cumulative Distribution Function,
Mean, and Variance of Normal Dist.
10 http://upload.wikimedia.org/wikipedia/commons/thumb/c/ca/Normal_
Distribution_CDF.svg/720px-Normal_Distribution_CDF.svg.png
3/18/2013
6
68-95-99.7% Rule
This figure represents a plot of the normal probability density
function with mean and standard deviation . Note that the
curve is symmetric about , so that is the median as well as
the mean. It is also the case for the normal population.
• About 68% of the population is in the interval .
• About 95% of the population is in the interval 2.
• About 99.7% of the population is in the interval 3.
11
Standard Units
• The proportion of a normal population that is
within a given number of standard deviations of
the mean is the same for any normal population.
• For this reason, when dealing with normal
populations, we often convert from the units in
which the population items were originally
measured to standard units.
• Standard units tell how many standard deviations
an observation is from the population mean.
12
3/18/2013
7
Standard Normal Distribution
In general, we convert to standard units by subtracting the mean and dividing by the standard deviation. Thus, if x is an item sampled from a normal population with mean and variance 2, the standard unit equivalent of x is the number z, where
z = (x - )/.
The number z is sometimes called the “z-score” of x. The z-score is an item sampled from a normal population with mean 0 and standard deviation of 1. This normal distribution is called the standard normal distribution.
13
Finding Areas Under the Normal
Curve
• The proportion of a normal population that lies within a given interval is equal to the area under the normal probability density above that interval. This would suggest integrating the normal pdf, but this integral does not have a closed form solution.
• The areas under the curve are approximated numerically and are available in Table A, B, and C.
• Table A provides area in the right-hand tail of the curve from z=0 to any positive z.
• Table B provides some commonly used probability points of the normal distribution: single-sided (Variance known)
• Table C provides some commonly used probability points of the normal distribution: double-sided (Variance known)
14
3/18/2013
8
Normal Probabilities
Excel:
NORMDIST(x, mean, standard_dev, cumulative)
NORMINV(probability, mean, standard_dev)
NORMSDIST(z)
NORMSINV(probability)
Minitab:
Calc Probability Distributions Normal
15
Linear Functions of Normal
Random Variables
Let X ~ N(, 2) and let a ≠ 0 and b be constants.
Then aX + b ~ N(a + b, a22).
Let X1, X2, …, Xn be independent and normally distributed with means 1, 2,…, n and variances 1
2, 22,…, n
2. Let c1, c2,…, cn be constants, and c1 X1 + c2 X2 +…+ cnXn be a linear combination. Then
c1 X1 + c2 X2 +…+ cnXn
~ N(c11 + c2 2 +…+ cnn, c121
2 + c222
2 + … +cn2n
2)
16
3/18/2013
9
Distributions of Functions of Normal
Random Variables
Let X1, X2, …, Xn be independent and normally distributed
with mean and variance 2. Then
Let X and Y be independent, with X ~ N(X, X2) and
Y ~ N(Y; Y2). Then
𝑋 + 𝑌~𝑁 𝜇𝑋 + 𝜇𝑌 , 𝜎𝑋2 + 𝜎𝑌
2
𝑋 − 𝑌~𝑁 𝜇𝑋 − 𝜇𝑌 , 𝜎𝑋2 + 𝜎𝑌
2
17
𝑋 ~𝑁 𝜇;𝜎2
𝑛
7.3 Example 7.1:
Normal Distribution
• The diameter of bushings is =50 mm with a standard deviation of =10 mm.
• Estimate the proportion of the population of bushings that have a diameter equal to or greater than 57 mm.
18
3/18/2013
10
7.4 Binomial Distribution
• A binomial distribution is useful when there are only two results in a random experiment: pass/fail, compliance/noncompliance, present/absent, yes/no.
• A dimension on a part is critical. This critical dimension is measured daily on a random sample of parts from a large production process. To expedite the inspection process, a tool is designed either to pass or fail a part that is tested. The out put now is no longer continuous. The output now is binary; hence, the binomial distribution can be used to develop an attribute sampling plan.
19
Binomial Distribution
If a total of n Bernoulli trials are conducted, and
The trials are independent.
Each trial has the same success probability p.
X is the number of successes in the n trials.
then X has the binomial distribution with
parameters n and p, denoted X ~ Bin(n,p).
20
3/18/2013
11
Binomial Distribution
• The probability of exactly x defects in n binomial trails with
probability of defect equal to p is:
• For a random experiment of sample size n in which there
are two categories of events, the probability of success of
the condition x in one category, while there is n-x in the
other category is (Probability Mass Function):
𝑝 𝑥 = 𝑃 𝑋 = 𝑥 =
𝑛!
𝑥! 𝑛 − 𝑥 !𝑝𝑥(1 − 𝑝)𝑛−𝑥, 𝑥 = 0,1,… , 𝑛
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
• When x=0, 𝑃 𝑋 = 0 = 𝑝0(1 − 𝑝)𝑛−0 = (1 − 𝑝)𝑛, is called
“first time yield” and equates to 𝑌𝐹𝑇 = 𝑞𝑛 = (1 − 𝑝)𝑛
21
Binomial Probability Histogram
(Probability Mass Function)
22 http://www.boost.org/doc/libs/1_51_0/libs/math/doc/
sf_and_dist/graphs/binomial_pdf_2.png
3/18/2013
12
Binomial Cumulative Distribution
Function
23 http://upload.wikimedia.org/wikipedia/commons/5/5
6/Binomial_distribution_cdf.png
Binomial Probabilities
Excel:
BINOM.DIST(number_s, trials, probability_s, cumulative)
Minitab:
Calc Probability Distributions Binomial
24
3/18/2013
13
Mean and Variance of
a Binomial Random Variable
Mean: 𝜇𝑋 = 𝑛𝑝
Variance: 𝜎𝑋2 = 𝑛𝑝(1 − 𝑝)
25
7.5 Example 7.2:
Binomial Distribution
• Calculate the probability of having the number 2 appear exactly three times in seven rolls of a six-sided die.
26
3/18/2013
14
7.6 Example 7.3:
Binomial Distribution
• A supplier claims a failure rate of 1 in 100.
• If this failure rate were true, the probability of observing exactly one defective part in a sample of 10 parts would be 0.091.
• The effectiveness of this test is questionable. The sample size is not large enough to do an effective job.
27
7.6 Example 7.3:
Binomial Distribution
• A supplier claims a failure rate of 1 in 100.
• If this failure rate were true, the probability of observing exactly one defective part in a sample of 10 parts would be 0.091.
• The effectiveness of this test is questionable. The sample size is not large enough to do an effective job.
28
3/18/2013
15
7.7 Hypergeometric Distribution
• Similar to Binomial distribution, except the sample size is
large (10%) relative to population size.
• A sample of size n is randomly selected without
replacement from a population of N items.
• In the population, r items can be classified as successes,
and N - r items can be classified as failures.
• A hypergeometric random variable, x, is the number of
successes that result from a hypergeometric experiment
Hypergeometric Probability Distribution
𝑝 𝑥 =
𝑟𝑥
𝑁−𝑟𝑛−𝑥𝑁𝑛
Where N = total number of elements in the population r = number of success in the population
N-r = number of failures in the population
n = number of trials (sample size)
x = number of successes in trial
n-x = number of failures in n trials
Let p=r/N, then 𝜇𝑥 = 𝑛𝑝, and 𝜎𝑥2 = 𝑛𝑝(1 − 𝑝)(
𝑁−𝑛
𝑁−1).
3/18/2013
16
Hypergeometric Probability Distribution
Example
Suppose we select 5 cards from an ordinary deck of playing cards. What is
the probability of obtaining 2 or fewer hearts?
Solution:
N = 52; since there are 52 cards in a deck.
r = 13; since there are 13 hearts in a deck.
n = 5; since we randomly select 5 cards from the deck.
x = 0 to 2; since our selection includes 0, 1, or 2 hearts.
We plug these values into the hypergeometric formula as follows:
𝑝 0 =
130
395
525
= .2215, 𝑝 1 =
131
394
525
= .4114, 𝑝 𝑥 =
132
393
525
= .2743
Hypergeometric Probability
in MINITAB
• Acceptance testing of ice cream cones Ice cream parlor
checks a batch of 400 waffle cones by checking 50 of
them. They will not buy them if more than 3 cones are
broken.
• What is the probability that the parlor will buy the cones if
35 of the 400 cones are broken.
– Define , n, r, N-r, x
– In MINITAB select: Calc-> Probability Distributions -
> Hypergeometric
3/18/2013
17
7.8 Poisson Distribution
One way to think of the Poisson distribution is
as an approximation to the binomial distribution
when n is large and p is small.
It is the case when n is large and p is small that
the mass function depends almost entirely on the
mean np, and very little on the specific values of n
and p.
We can therefore approximate the binomial mass
function with a quantity λ = np; this λ is the
parameter in the Poisson distribution.
33
Poisson Distribution Applications
There are a large number of critical dimensions on a part.
Dimensions are measured on a random sample of parts
from a large production process. The number of out-of-
specification conditions is noted on each sample. This
collective number-of-failures information from the samples
can often be modeled using a Poisson distribution.
A repairable system is known to have a constant failure
rate as a function of usage (i.e., follows an HPP). In a test
a number of systems are exercised and the number of
failures are noted for the systems. The Poisson
distribution can be used to design/analyze this test.
34
3/18/2013
18
Probability Mass Function, Mean,
and Variance of Poisson Dist.
If X ~ Poisson(λ), the probability mass function of X is
𝑝 𝑥 = 𝑃 𝑋 = 𝑥 = 𝑒−𝜆𝜆𝑥
𝑥! , 𝑥 = 0,1, 2…
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
𝑃 𝑋 = 0 =𝑒−𝜆𝜆0
0! = 𝑒−𝜆 = 𝑒−𝐷/𝑈 = 𝑒−𝐷𝑃𝑈
Mean and Variance: 𝜇𝑋 = 𝜆, 𝜎𝑋2 = 𝜆
Note: X must be a discrete random variable and 𝜆 must be a positive constant.
35
Poisson Probability Mass Function
36 http://brokensymmetry.typepad.com/photos/unca
tegorized/2008/07/31/800pxpoissoncdf.png
3/18/2013
19
Poisson Cumulative Probability
Function
37 http://www.boost.org/doc/libs/1_35_0/libs/
math/doc/sf_and_dist/graphs/poisson.png
Poisson Probabilities
Excel:
POISSON.DIST(x, mean, cumulative)
Minitab:
Calc Probability Distributions Poisson
38
3/18/2013
20
7.9 Example 7.4:
Poisson Distribution
A company observed that over several years they had a mean manufacturing line shutdown rate of 0.10 per day. Assume a Poisson distribution, determine the probability of two shutdowns occurring on the same day.
39
7.10 Exponential Distribution
• The exponential distribution is a continuous
distribution that is sometimes used to model the
time that elapses before an event occurs. Such a
time is often called a waiting time.
• A repairable system is known to have a constant failure
rate as a function of usage. The time between failures will
be distributed exponentially. The failures will have a rate of
occurrence that is described by an HPP. The Poisson
distribution can be used to design a test in which sampled
systems are tested for the purpose of determining a
confidence interval for the failure rate of the system.
40
3/18/2013
21
Exponential R.V.:
pdf, cdf, mean and variance
• The pdf of an exponential random variable X is
𝑓(𝑥) = (1
𝜃) 𝑒−𝑥/𝜃 , 𝑥 > 0
0, 𝑥 ≤ 0
• The cdf of an exponential random variable is
𝐹(𝑥) = 1 − 𝑒−𝑥/𝜃 , 𝑥 > 00, 𝑥 ≤ 0
• The exponential distribution is dependent on only one parameter (), which is the mean of the distribution (MTBF).
• The instantaneous failure rate (i.e., hazard rate) of an exponential distribution is constant and equals 1/.
41
Exponential Probability Function
42 http://upload.wikimedia.org/wikipedia/commons/thumb/e/ec/
Exponential_pdf.svg/325px-Exponential_pdf.svg.png
3/18/2013
22
Exponential Probabilities
Excel:
EXPONDIST(x, lambda, cumulative)
Minitab:
Calc Probability Distributions Exponential
43
7.11 Example 7.5:
Exponential Distribution
The reported time between failure rate of a system is 10,000 hours. If the failure rate follows an exponential distribution, the time when F(x) is 0.10 can be determined
𝐹 𝑥 = 1 − 𝑒−𝑥𝜃 = 0.10
Then x=1054 hours.
44
3/18/2013
23
Lack of Memory Property
The exponential distribution has a property known
as the lack of memory property:
If T ~ Exp(), and t and s are positive
numbers, then
P(T > t + s | T > s) = P(T > t).
45
Example
The lifetime of a transistor in a particular circuit has an exponential distribution with mean 1.25 years.
1. Find the probability that the circuit lasts longer than 2 years.
2. Assume the transistor is now three years old and is still functioning. Find the probability that it functions for more than two additional years.
3. Compare the probability computed in 1. and 2.
46
3/18/2013
24
7.12 Weibull Distribution
The Weibull distribution is a continuous random variable that is used in a variety of situations. A common application of the Weibull distribution is to model the lifetimes of components. The Weibull probability density function has two parameters, both positive constants, that determine the location and shape. We denote these parameters and .
If = 1, the Weibull distribution is the same as the exponential distribution with parameter = .
47
7.12 Weibull Distribution
Example for Two parameter Weibull distribution:
• A nonrepairable device experiences failures through either early-life, intrinsic, or wear-out phenomena. Failure data of this type often follow the Weibull distribution.
Example for Three parameter Weibull distribution:
• A dimension on a part is critical. This critical dimension is measured daily on a random sample of pans from a large production process. Information is desired about the "tails" of the distribution. A plot of the measurements indicates that they follow a three-parameter Weibull distribution better than they follow a normal distribution.
48
3/18/2013
25
Weibull R.V.
The pdf of the 3-parameter Weibull distribution is
𝑓(𝑥) =𝑏
𝑘 − 𝑥0
𝑥 − 𝑥0𝑘 − 𝑥0
𝑏−1
𝑒𝑥𝑝 −𝑥 − 𝑥0𝑘 − 𝑥0
𝑏
And the cdf is
𝐹 𝑥 = 1 − 𝑒𝑥𝑝 −𝑥 − 𝑥0𝑘 − 𝑥0
𝑏
49
Weibull R.V.
The pdf of the 2-parameter Weibull distribution is
𝑓(𝑥) =𝑏
𝑘
𝑥
𝑘
𝑏−1
𝑒𝑥𝑝 −𝑥
𝑘
𝑏
And the cdf is
𝐹 𝑥 = 1 − 𝑒𝑥𝑝 −𝑥
𝑘
𝑏
50
3/18/2013
26
Weibull Probability Function
51
Weibull Probabilities
Excel:
WEIBULL(x, alpha, beta, cumulative)
Minitab:
Calc Probability Distributions Weibull
52
3/18/2013
27
7.13 Example 7.6:
Weibull Distribution
A component has a characteristic life of 10,000 hours and a shape parameter of 1. 90% of the systems are expected to survive x hours determined by substitution
𝐹 𝑥 = 1 − 𝑒𝑥𝑝 −𝑥
𝑘
𝑏
= 0.10 = 1 − 𝑒𝑥𝑝 −𝑥
10000
1
which is 1054 hours.
53
7.14 Lognormal Distribution
• For data that contain outliers, the normal distribution is generally not appropriate. The lognormal distribution, which is related to the normal distribution, is often a good choice for these data sets.
• If X ~ N(,2), then the random variable Y = eX has the lognormal distribution with parameters and 2.
• If Y has the lognormal distribution with parameters and 2, then the random variable X = lnY has the N(,2) distribution.
54
3/18/2013
28
7.14 Lognormal Distribution
Application
• A nonrepairable device experiences failures through metal fatigue. Time of failure data from this source often follows the lognormal distribution.
55
Lognormal pdf, mean, and
variance
The pdf of a lognormal random variable Y with parameters
and 2 is
𝑓(𝑥) = 1
𝜎𝑥 2𝜋𝑒𝑥𝑝 −
[ln (𝑥)−𝜇]2
2𝜎2, 𝑥 > 0
0, 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
Mean: 𝐸 𝑌 = 𝑒𝑥𝑝 𝜇 +𝜎2
2
Variance: 𝑉 𝑌 = exp 2𝜇 + 2𝜎2 − exp (2𝜇 + 𝜎2)
56
3/18/2013
29
Lognormal Probability Density
Function
57
=0
=1
Lognormal Probability Function
58 http://upload.wikimedia.org/wikipedia/commons/thumb/4/46/Lognormal_
distribution_PDF.png/325px-Lognormal_distribution_PDF.png
3/18/2013
30
Lognormal Probabilities
Excel:
LOGNORMDIST(x, mean, standard_dev)
Minitab:
Calc Probability Distributions Lognormal
59
Example
When a pesticide comes into contact with the skin,
a certain percentage of it is absorbed. The
percentage that is absorbed during a given time
period is often modeled with a lognormal
distribution. Assume that for a given pesticide,
the amount that is absorbed (in percent) within
two hours is lognormally distributed with of 1.5
and σ of 0.5. Find the probability that more than
5% of the pesticide is absorbed within two hours.
60
3/18/2013
31
7.15 Tabulated Probability Distribution:
Chi-square Distribution
• Chi-square distribution is an important sampling distribution.
• Chi-square distribution is used to determine the confidence interval for the standard deviation of a population.
• Chi-square distribution is also used in determining the goodness of fit, independence, and homogeneity between data sets.
61
2 Distribution
• Let x1, x2, .. xn be a random sample from a normal distribution with and 2, and let s2 be the sample variance, then the random variable (n-1)s2/2 has 2 distribution with n-1 degrees of freedom.
• Probability Density Function, with degrees of freedom,
𝑓 𝑥 =1
2𝜈2Γ(
𝜈2)𝑥
𝜈2−1exp −
𝑥
2 𝑥 > 0
• = degrees of freedom
• = ; 2 = 2; Mode = -2 (when 3)
3/18/2013
32
7.15 Tabulated Probability Distribution:
Chi-square Distribution
63
7.15 Tabulated Probability Distribution:
Chi-square Distribution
• Table G gives the right-tailed percentage points of the chi-square distribution.
• Excel Functions: Chisq.dist left-tailed probability
Chisq.dist.rt right-tailed probability
Chisq.inv 2 from left-tailed probability
Chisq.inv.rt 2 from right-tailed probability
64
3/18/2013
33
7.16 Tabulated Probability Distribution:
Student’s t Distribution
• Discovered by W.S. Gosset (1908) and perfected by R.A. Fisher (1926)
• Student’s t distribution is used to determine the confidence interval for the population mean; and the difference in population means.
65
t Distribution
• Let x1, x2, .. xn be a random sample from a normal distribution with unknown and 2, the random
variable 𝑋 −𝜇
𝑠 𝑛 has a t- distribution with n-1 degrees of
freedom.
• Probability Density Function, with degrees of freedom,
𝑓 𝑥 =Γ (𝜈+1) 2
𝜋𝜈Γ(𝜈 2 )
1
(𝑥2 𝜈)+1 (𝜈+1) 2 − ∞ < 𝑥 < ∞
• = 0; 2 = 𝜈 (𝜈 − 2)
• Symmetrical; approaches normal as the degrees of freedom approaches ∞.
3/18/2013
34
t Distribution
www.boost.org/.../graphs/students_t_pdf.png
7.16 Tabulated Probability Distribution:
Student’s t Distribution
• Table D gives the right-tailed percentage points of the t distribution (single-sided).
• Table E gives the two-tailed percentage points of the t distribution (double-sided).
• Excel Functions: t.dist left-tailed probability
t.dist.2t two-tailed probability
t.dist.rt right-tailed probability
t.inv t from left-tailed probability
t.inv.2t t from two-tailed probability
68
3/18/2013
35
7.17 Tabulated Probability Distribution:
F Distribution
• F distribution is used to determine if two population variances are different in magnitude.
• Statistics that have an F distribution are ratios of quantities, such as the ratio of two variances.
• The F distribution has two values for the degrees of freedom: one associated with the numerator, and one associated with the denominator.
• The degrees of freedom are indicated with subscripts under the letter F.
• Note that the numerator degrees of freedom are always listed first.
69
F Distribution
70 http://upload.wikimedia.org/wikipedia/commons/thumb/f/
f7/F_distributionPDF.png/800px-F_distributionPDF.png
3/18/2013
36
7.17 Tabulated Probability Distribution:
F Distribution
• Table F gives the right-tailed percentage points of the F distribution.
• Excel Functions: F.dist left-tailed probability
F.dist.rt right-tailed probability
F.inv t from left-tailed probability
F.inv.rt t from two-tailed probability
71
7.18 Hazard Rate
• Hazard rate is the probability that a device on test will fail between (t) and (t+dt), if the device has already survived up to time (t).
• The general expression for the hazard rate () is
𝜆 =𝑓(𝑡)
1 − 𝐹(𝑡)
– Where f(t) is the PDF of failures, and F(t) is the CDF of failures at time (t); [1-F(t)] is the reliability of a device at time t (i.e., survival portion)
• The hazard or failure rate can be decribed by the classical reliability bathtub curve.
72
3/18/2013
37
7.18 Hazard Rate
73 http://reliabilityweb.com/ee-assets/my-
uploads/art09/arc/SMRP_211.jpg
7.18 Hazard Rate
• For a non-repairable system, the Weibull distribution can be used to model portions of this curve.
• In the Weibull equation, a value of b<1 is characteristic of early-life manufacturing failures, a value of b>1 is characteristic of a wear-out mechanism, and a value of b=1 is characteristic of a constant failure mode (intrinsic failure period)
74
3/18/2013
38
7.19 Non-Homogeneous Poisson
Process (NHPP)
• For a repairable system, the failure rate (as a function of usage) is called the intensity function.
• The NHPP with Weibull intensity is a model that can consider system repairable failure rates that change with time.
• Application: A repairable system failure rate is not constant. The NHPP with Weibull intensity process often can be used to model this situation when considering the general possibilities of early-life, intrinsic, or wear-out characteristics.
• The NHPP with Weibull intensity can be expressed as
𝑟 𝑡 = 𝜆𝑏𝑡𝑏−1 – Where r(t) is instantaneous failure rate at time t, and is the intensity
of the Poisson process
75
7.20 Homogeneous Poisson Process
(HPP)
• HPP considers that the failure rate does not change with time (i.e., a constant intensity function)
• Application: A repairable system failure rate is constant with time. The failure rate is said to follow an HPP process. The Poisson distribution is often useful when designing a test of a criterion that has an HPP.
• The HPP is a specail case of the NHPP with Weibull intensity, where b = 1.
𝑟 𝑡 = 𝜆
• The intensity of HPP equates to the hazard rate of the exponential distribution.
76
3/18/2013
39
7.21 Applications for Various Types
of Distributions and Processes
77
Distribution
or Process Applications Examples
Normal
distribution
Can be used to describe various
physical, mechanical, electrical, and
chemical properties.
Part dimensions,
Voltage outputs,
Chemical composition
level
Lognormal
distribution
Shape flexibility of density function
yields an adequate fit to many types
of data. Normal distribution
equations can be used in the
analysis.
Life of mechanical
components that fail by
metal fatigue; Describes
repair times of
equipment
7.21 Applications for Various Types
of Distributions and Processes
78
Distribution or
Process Applications Examples
Binomial
distribution
Can be used to describe the
situation where an observation
can either pass or fail.
Part sampling plan where
the part meets or fails to
meet a specification criterion
Hypergeometric
distribution
For pass/fail observations
provides an exact solution for any
sample size from a population.
Pass/fail testing where a
sample of 50 is randomly
chosen from a population of
size 100
Poisson
distribution
Convenient distribution to use
when designing tests that
assumes that the underlying
distribution is exponential.
Test distribution to
determine whether a MTBF
failure criterion is met
3/18/2013
40
7.21 Applications for Various Types
of Distributions and Processes
79
Distribution or
Process Applications Examples
Weibull
distribution (2-
parameter)
Shape flexibility of density function
conveniently describes increasing,
constant, and decreasing failure rates as
a function of usage (age).
Life of mechanical
and electrical
components
Weibull
distribution (3-
parameter)
Shape flexibility of two-parameter
distribution with the added flexibility
that the zero probability point can take
on values that are greater than zero.
Mechanical part
tensile strength;
Electrical resistance
Exponential
distribution
Shape can be used to describe device
system failure rates that are constant as
a function of usage.
MTBF or constant
failure rate of a
system
7.21 Applications for Various Types
of Distributions and Processes
80
Distribution
or Process Applications Examples
HPP Model that describes
occurrences that happen
randomly in time.
Modeling of constant
system failure rate
NHPP Model that describes
occurrences that either decrease
or increase in frequency with
time.
System failure rate
modeling when the
rate increases or
decreases with time
3/18/2013
41
7.21 Distribution Approximations
81
Distribution Approximate
Distribution Situation
Hypergeometric Binomial 10n population size (Miller and Freund
1965)
Binomial Poisson n 20 and p 0.05. If n 100, the
approximation is excellent as long as np
10 (Miller and Freund 1965)
Binomial Normal np and n(1 - p) are at least 5 (Dixon and
Massey 1969)