Lamb2 PDF

Reliability Engineering1
Leonard R. Lamberson, Ph.D., P.E.

College of Engineering and Applied Sciences
Western Michigan University
Kalamazoo, MI 49008
April 23, 2003
c
Copyright 1980,
1994, 1999, 2000 Leonard R. Lamberson
Contents
1 Reliability Engineering
1.1 Reliability Denition . . . . . . . . . . . . . . . . . . . .
1.1.1 Probabilistic Nature of Reliability . . . . . . . .
1.1.2 Perform Intended Function Satisfactorily . . . .
1.1.3 Specied Time Period . . . . . . . . . . . . . . .
1.1.4 Specied Conditions . . . . . . . . . . . . . . . .
1.2 Reliability Numbers . . . . . . . . . . . . . . . . . . . .
1.2.1 Indicators Used to Quantify Product Reliability .
1.3 Reliability and Quality . . . . . . . . . . . . . . . . . . .
1.3.1 Quality Characteristics . . . . . . . . . . . . . .
1.3.2 Product Defects . . . . . . . . . . . . . . . . . .
1.4 Customer Satisfaction . . . . . . . . . . . . . . . . . . .
1.5 Product Life and Failure Rate . . . . . . . . . . . . . . .
1.5.1 Infant Mortality Period . . . . . . . . . . . . . .
1.5.2 Useful Life Period . . . . . . . . . . . . . . . . .
1.5.3 Wear-out Period . . . . . . . . . . . . . . . . . .
1.6 Product Design and Development Cycle . . . . . . . . .
1.6.1 Product Development Cycle . . . . . . . . . . . .
1.7 Reliability in Design . . . . . . . . . . . . . . . . . . . .
2 Design Review
2.1 Policy . . . . . . . . . . . . . . . .
2.2 Appointment of Chairperson . . .
2.3 Timing of Reviews . . . . . . . . .
2.4 Scheduling of Reviews . . . . . . .
2.5 Advance Information . . . . . . . .
2.6 Agenda . . . . . . . . . . . . . . .
2.7 Checklists . . . . . . . . . . . . . .
2.8 Conduct of Design Review Meeting
i
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
1
1
2
3
4
4
6
7
7
7
8
9
11
11
14
15
16
16
18
.
.
.
.
.
.
.
.
20
23
23
23
25
25
25
25
26
CONTENTS
2.9
ii
Final Report Documentation . . . . . . . . . . . . . . . . . .
3 Reliability Measures
3.1 The Reliability Function . . . . .
3.2 The Expected Life . . . . . . . .
3.3 The Failure Rate . . . . . . . . .
3.3.1 Uses for the Failure Rate
27
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
28
29
34
35
38
4 Static Models
4.1 Static Reliability Modeling Process . . . . . . . .
4.1.1 Denition of Symbols . . . . . . . . . . .
4.2 Series System Analysis . . . . . . . . . . . . . . .
4.2.1 Basic Series Reliability Equations . . . . .
4.3 Parallel System Reliability Analysis . . . . . . .
4.4 Parallel and Series Combinations . . . . . . . . .
4.5 Design Considerations-Static Reliability Models .
4.5.1 Series System and Reliability . . . . . . .
4.5.2 Design Considerations - Series Systems . .
4.5.3 Parallel System and Reliability . . . . . .
4.5.4 Design Considerations - Parallel Systems
4.6 High level versus Low level Redundancy . . . . .
4.6.1 High Level Redundancy . . . . . . . . . .
4.6.2 Low Level Redundancy . . . . . . . . . .
4.6.3 Conclusion on Providing Redundancy . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
40
42
45
45
45
48
51
52
53
53
54
55
56
57
57
57
5 Life Distributions
5.1 The Exponential Distribution . . . . . . . . . . . .
5.1.1 Probability Density Function . . . . . . . .
5.1.2 Parameter . . . . . . . . . . . . . . . . . . .
5.1.3 Reliability Function . . . . . . . . . . . . .
5.1.4 Hazard Function . . . . . . . . . . . . . . .
5.1.5 Specied Time for a Reliability Level . . . .
5.2 The Normal Distribution . . . . . . . . . . . . . . .
5.2.1 Probability Density Function . . . . . . . .
5.2.2 Parameters . . . . . . . . . . . . . . . . . .
5.2.3 Hazard Function . . . . . . . . . . . . . . .
5.2.4 Standard Normal Density Function . . . . .
5.2.5 Standard Normal Geometric Relationships .
5.2.6 Inverse Use of Standard Normal Tables . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
59
59
59
60
60
60
61
62
63
63
63
63
65
66
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
CONTENTS
iii
5.2.7
Translation of any Normal Distribution to the Standard Normal . . . . . . . . . . . . . . . . . . . . . . .

Lognormal Distribution . . . . . . . . . . . . . . . . . . . . .
5.3.1 Lognormal Probability Density Function . . . . . . . .
5.3.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . .
5.3.3 Hazard Function . . . . . . . . . . . . . . . . . . . . .
5.3.4 Calculation of Probabilities for the Lognormal . . . .
Weibull Distribution . . . . . . . . . . . . . . . . . . . . . . .
5.4.1 Weibull Probability Density Function . . . . . . . . .
5.4.2 Parameters . . . . . . . . . . . . . . . . . . . . . . . .
5.4.3 Reliability Function . . . . . . . . . . . . . . . . . . .
5.4.4 Mean and Standard Deviation . . . . . . . . . . . . .
5.4.5 Hazard Function . . . . . . . . . . . . . . . . . . . . .
5.4.6 Specied Time for a Reliability Level . . . . . . . . . .
Distribution Selection . . . . . . . . . . . . . . . . . . . . . .
5.5.1 Bath-tub Hazard Function and Distribution Selection
67
70
70
70
71
71
72
74
74
74
74
74
75
76
76
6 Exponential Distribution
6.1 Mean Life Parameter . . . . . . . . . . . . . . . . . . . . . . .
6.2 Reliability Function . . . . . . . . . . . . . . . . . . . . . . .
6.3 Time and Mean Life Formulas . . . . . . . . . . . . . . . . . .
6.4 Failure Rate . . . . . . . . . . . . . . . . . . . . . . . . . . . .
6.5 Mean Life and Reliability . . . . . . . . . . . . . . . . . . . .
6.6 Applicability of the Exponential Model . . . . . . . . . . . . .
6.7 Warranty Expense Calculations . . . . . . . . . . . . . . . . .
6.8 Estimation of Mean Life . . . . . . . . . . . . . . . . . . . . .
6.9 Estimating the Reliability Function . . . . . . . . . . . . . . .
6.10 Condence Limits . . . . . . . . . . . . . . . . . . . . . . . .
6.10.1 Two-sided Failure Truncated Limits . . . . . . . . . .
6.10.2 Two-sided Time Truncated Limits . . . . . . . . . . .
6.10.3 One-sided Failure Truncated Limits . . . . . . . . . .
6.10.4 One-sided Time Truncated Limits . . . . . . . . . . .
6.11 Reliability Estimation With Condence . . . . . . . . . . . .
6.11.1 The No Failure Situation . . . . . . . . . . . . . . . .
6.12 Test Time to Demonstrate Reliability at a Condence Level .
6.13 Poisson Failure Calculations . . . . . . . . . . . . . . . . . . .
6.13.1 Normal Approximation to Poisson Failure Calculations
6.14 Combining Failure Rates . . . . . . . . . . . . . . . . . . . . .
77
77
78
78
79
79
80
80
81
82
83
83
83
84
84
84
86
87
87
88
89
5.3
5.4
5.5
CONTENTS
7 Success/Failure Testing
7.1 Binomial Probability Model for Reliability . . . . .
7.2 Success-Failure Testing . . . . . . . . . . . . . . . .
7.3 Approximate Condence Limits . . . . . . . . . . .
7.4 Success Testing . . . . . . . . . . . . . . . . . . . .
7.4.1 Conversion to the Exponential Distribution
iv
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
91
91
93
95
96
97
8 Weibull Distribution
100
8.1 Weibull Probability Density Function . . . . . . . . . . . . . . 100
8.1.1 Two-parameter Weibull . . . . . . . . . . . . . . . . . 100
8.1.2 Characteristic Life . . . . . . . . . . . . . . . . . . . . 104
8.2 Weibull Mean and Standard Deviation . . . . . . . . . . . . . 104
8.2.1 Weibull Mean . . . . . . . . . . . . . . . . . . . . . . . 104
8.2.2 Weibull Standard Deviation . . . . . . . . . . . . . . . 106
8.3 Three-parameter Weibull . . . . . . . . . . . . . . . . . . . . 107
8.3.1 Computerized Three-parameter Weibull Distributions 108
8.4 Graphical Estimation of Weibull Parameters . . . . . . . . . . 108
8.4.1 Plotting Positions . . . . . . . . . . . . . . . . . . . . 109
8.4.2 Summary of Weibull Plotting Procedure . . . . . . . . 111
8.4.3 Condence Limits for Graphical Analysis . . . . . . . 112
8.5 Non-zero Minimum Life Weibull Distributions . . . . . . . . . 113
8.6 Suspended Item Analysis . . . . . . . . . . . . . . . . . . . . 114
8.6.1 New Increment Method . . . . . . . . . . . . . . . . . 115
8.7 Sudden Death Testing . . . . . . . . . . . . . . . . . . . . . . 117
9 Maintainability Engineering
9.1 Maintainability Improvement Strategies . . . . . . . .
9.2 Determination of the MTTR . . . . . . . . . . . . . .
9.2.1 Mean Time to Repair from Operational Data .
9.2.2 Availability . . . . . . . . . . . . . . . . . . . .
9.2.3 Availability Calculation . . . . . . . . . . . . .
9.2.4 Basis for Availability Formula . . . . . . . . . .
9.3 Maintainability Specications . . . . . . . . . . . . . .
9.3.1 Parameters . . . . . . . . . . . . . . . . . . . .
9.3.2 Calculation of Probabilities for the Lognormal
9.4 Preventive Maintenance . . . . . . . . . . . . . . . . .
9.5 Overall Equipment Eectiveness . . . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
122
123
126
127
128
129
130
131
132
132
132
133
CONTENTS
10 Growth Management
134
10.1 Reliability Growth Relationships . . . . . . . . . . . . . . . . 135
10.2 Growth Curve Plotting . . . . . . . . . . . . . . . . . . . . . . 138
10.3 Risk Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
11 Life Cycle Costing
11.1 Application of the Life Cycle Concept . . . .
11.2 Some Engineering Economy Concepts . . . .
11.2.1 Interest and Annuity Relationships . .
11.2.2 A Series of Equal Payments . . . . . .
11.3 Life Cycle Cost Components . . . . . . . . . .
11.3.1 Acquisition Costs . . . . . . . . . . . .
11.3.2 Operating Costs . . . . . . . . . . . .
11.3.3 Maintenance Costs . . . . . . . . . . .
11.3.4 Conversion/Decommissioning Costs .
11.4 Acquisition Practices and Life Cycle Costing
A Mathematical Formulas
A.1 Derivatives . . . . . . . . . .
A.1.1 Denition . . . . . . .
A.1.2 Common Derivatives .
A.1.3 Partial Derivatives . .
A.2 Integrals . . . . . . . . . . . .
A.2.1 Gamma Function . . .
A.2.2 Factorial Numbers . .
A.3 Series . . . . . . . . . . . . .
A.3.1 Exponential Series . .
A.3.2 Geometric Progression
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
B Statistical Tables
B.1 Cumulative Normal Distribution . . . . . . .
B.2 Extended Normal Distribution . . . . . . . .
B.3 t-distribution . . . . . . . . . . . . . . . . . .
B.4 2 -distribution . . . . . . . . . . . . . . . . .
B.5 F -Distribution . . . . . . . . . . . . . . . . .
B.6 Weibull Tables . . . . . . . . . . . . . . . . .
B.6.1 Factor A1 for the Mean . . . . . . . .
B.6.2 Factor A2 for the Standard Deviation
B.7 Rank Tables . . . . . . . . . . . . . . . . . . .
B.7.1 Median Rank Table . . . . . . . . . .
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
146
. 146
. 148
. 148
. 149
. 151
. 151
. 152
. 153
. 154
. 155
.
.
.
.
.
.
.
.
.
.
156
. 156
. 156
. 156
. 157
. 157
. 157
. 158
. 158
. 158
. 158
.
.
.
.
.
.
.
.
.
.
160
. 161
. 162
. 163
. 164
. 168
. 172
. 172
. 173
. 174
. 174
CONTENTS
vi
B.7.2 5% Rank Table . . . . . . . . . . . . . . . . . . . . . . 175

B.7.3 95% Rank Table . . . . . . . . . . . . . . . . . . . . . 176
Chapter 1
Reliability Engineering
What do we mean when we say we have a reliable product?
Reliability n - may be relied on; trustworthy, authentic, consistent. Infallible, suggests the complete absence of error, breakdown or poor performance.
When we speak of a reliable product, we usually expect such adjectives
as dependable and trustworthy to apply. But to measure product reliability
we must have a more exact denition.
1.1
Reliability Denition
Reliability is the probability that a product will perform its intended function
in a satisfactory manner for a specied period of time when operating under
specied conditions.
Thus, the reliability of a system expresses the length of failure-free time that
can be expected from the equipment. Higher levels of reliability mean less
failures and consequently less downtime.
To measure reliability it is necessary to
Relate probability to a precise denition of success or satisfactory performance
Specify the time base or operating cycles over which such performance
is to be sustained
Specify the environmental or use conditions that will prevail.
1
50%
50%
25%
25%
0%
0%
200
75%
150
75%
100
100%
50
100%
R(t)
CHAPTER 1. RELIABILITY ENGINEERING
Time-hours (t)
Figure 1.1: Reliability Functions

Theoretically, every product has a designed-in reliability function. This
reliability function (or curve) expresses the system reliability at any point
in time. As time increases the curve must drop, eventually reaching zero
(see Figure 1.1).
1.1.1
Probabilistic Nature of Reliability
We can not say exactly when a particular product will fail, but we can say
what percentage of the products in use will fail by certain times. This is
analogous to the reasoning used by life insurance companies.
We can state reliability in various ways:
The probability that a product will be performing its intended function
at 5,000 hours of use is 0.95.
The reliability at 5,000 hours is 0.95 or 95%.
With a 5,000 hour reliability of 95%, if we place 1,000 units in use,

950 will still be operating with no failures at 5,000 hours.
For a longer period of time, the reliability decreases. We might nd
the reliability at 8,000 hours is 0.70.
The unreliability at 8,000 hours is 0.30.
From a service point of view, we may be interested in repair frequency
and then we say that 30% of the units will have to be serviced by 8,000
hours. Or the repairs per hundred units (R/100) is 30 at 8,000 hours.
The important point is that the reliability is a metric expressing the
probability of maintaining intended function over time and is measurable as
a percentage.
1.1.2
Perform Intended Function Satisfactorily
A product fails when it ceases to function in a way that is satisfactory to the

customer. Products rarely fail like a light bulb. Rather, they deteriorate
over time. This eventually leads to unsatisfactory performance from the
customers standpoint.
Understand product failure from the customers viewpoint.
satisfactory performance can result from:
Un-
Excess vibrations
Excess noise
Intermittent operation
Drift
Decreased accuracy
Torque loss
And, of course, catastrophic failure
and many other possibilities.
Product failure. There will be dierent kinds of product failure. Some
failures cause immediate halt to operations while others might provide a
mild inconvenience to the customer. The terms hard and soft are sometimes
used to categorize these failures.
Soft failure. A soft failure is where the product has degraded but is still
usable. For example, a light bulb may be ickering but is still giving o
light. An internal combustion engine may have a decrease in output and/or
an increase in oil consumption. Soft failures usually lead to catastrophic
failures if left unattended.
For a soft failure to be counted or categorized as a failure the degree
of degradation must be specied. For example, an automotive tire wears
with mileage so it is degrading. One might specify a thread depth at which
the tire must be taken out of service.
Hard failure. A hard failure immediately takes the product out of service.
For the light bulb example, if the bulb completely ceased to give light output
this would be a hard failure. For the tire example, a blow out would be a
hard failure.
Clearly dene unsatisfactory performance. Unsatisfactory performance must be clearly spelled out. The customers perspective must be
recognized in this process. There will usually be various levels of failure
based on the customers perceived level of severity. The levels of severity
are frequently grouped into categories such as
Major
Minor
The severity of the failure to the customer must be delineated in a Failure
Denition Document that precisely denes how each incident on a system
or equipment will be handled in regard to reliability and maintainability
calculations. Such a document should be developed early in a design and
development program so that all concerned are aware of the consequences
of incidents that occur during product testing and in eld use.
The design team must be able to use the failure denition document to
address product trade-os. If the severity of a failure to the customer can be
lowered by design changes, the failure denition document should promote
the analysis of this trade-o using cost and market impact.
1.1.3
Specied Time Period
Products deteriorate with use and even with age when dormant. Longer
lengths of usage imply lower reliability. For design purposes target usage
periods must be identied.
Typically identied usage periods are the

Warranty period(s): A contract supplied with the product providing
the user with a certain amount of protection against product failure.
Expected customer life: Customers have a reasonably consistent belief
as to how long a product should last.
Durability life: A measure of useful life, dening the number of operating hours (or cycles) until overhaul is required.
1.1.4
Specied Conditions
Dierent environments promote dierent failure modes and dierent failure

rates for a product. The environmental factors which the product will encounter must be clearly dened. The levels (and rate of change) at which
we want to address these environmental factors must also be dened.
Environmental Conditions Prole
The environmental prole must include the level and rate of change for each
environmental factor considered. There are many potential environmental
factors that might promote failure. A list of possibilities might include such
things as:
Temperature
Humidity
Vibration
Shock
Corrosive materials
Immersion
Pressures, vacuum
Salt spray
Dust
Cement oors/basements
Ice/snow
Lubricants
Perfumes
Magnetic elds
Nuclear radiation
Weather
Contamination
Antifreeze
Gasoline fumes
Rust inhibitors/under coatings
Rain
Soda pop/hot coee
Sunlight
Electrical discharges
Not all of these environmental conditions would be appropriate for a
particular product. Each product must be considered in its operating environment and scenario. The environmental prole must consider the environment induced from operating the product, the environment induced
from external factors, and the environment induced during delivery of the
product to the customer.
Example 1.1Dening product reliability.

Identify a part/product that you are familiar with and do the following:
A. Delineate the environment that should be considered when designing
this product.
B. Dene the term reliability as it applies to this product.
C. Decide on a reasonable value for the expected customer life.
1.2
Reliability Numbers
The reliability number attached to a product changes with

Usage and environmental conditions
Customers perception of satisfactory performance
At any product age (t) for a population of N products, the reliability at
time t denoted by R(t) is
R(t) =
Number of survivors
N
which is also equal to

Number of failures
= 1 Unreliability
N
This is the reliability of this population of products at time t. The real-world
estimation of reliability is usually much more dicult due to products being
sold over time with each having a dierent usage prole. Calendar time
is known but product life on each product is not while warranty systems
monitor and record only failure.
R(t) = 1
1.2.1
Indicators Used to Quantify Product Reliability
Several metrics are in common use to indicate product reliability. Some of

the metrics are actually quantifying unreliability. Some of the possibilities
follow:
MTBF - mean-time-between failures; also MTTF, MMBF, MCTF.
M T BF = 1, 500 hours; means that on the average a failure will occur
with every 1,500 hours of operation.
Failure Rate - The rate of failures per unit of operating time.
= 6.7104 /hour means that one failure will occur with every 1,500
hours of operation, on the average.
R/1,000 - The number of warranty claims per 1,000 products sold.
R/1,000 = 7 means that there are seven warranty claims for every
1,000 products sold.
Reliability number - the reliability of the product at some specic time.
R = 999% means that 999 out of 1,000 products work successfully for
the specied time.
Select a reliability metric for your company. Although all of the reliability metrics are mathematically related they do not all work equally well
to provide reliability information to all interested personnel involved in the
design and development of a product. A reliability metric must be selected
that works for your product and your industry. Then benchmark data,
warranty data, design documents and all useful information can be portrayed using this metric. The personnel within your company will become
acquainted with what is acceptable R&M and with how to use reliability
techniques employing this metric.
1.3
Reliability and Quality
The terms reliability and quality are frequently used interchangeable by

customers and product engineers. Ultimately the customer denes quality.
Customers want products that meet or exceed their needs and expectations,
at a cost that represents value. This expectation of performance must be
met throughout the customers expected life for the particular product.
Quality is usually recognized as a more encompassing term including
reliability.
1.3.1
Quality Characteristics
Psychological
Taste
Beauty, style
Status
Technological
Hardness
Vibration
Noise
Materials (bearings, belts, hoses, etc.)
Time Oriented
Reliability
Maintainability
Contractual
Warranty
Ethical
Honesty of repair shop
Experience and integrity of sales force
1.3.2
Product Defects
Quality defects are dened as those which can be located by conventional inspection techniques.
Reliability defects are dened as those which require some stress applied over time to develop into detectable defects.
What causes product failure over time? Some possibilities are:
Design
Manufacturing
Packaging
Shipping
Storage
Sales
Installation
Maintenance
Customer duty cycle
1.4
Customer Satisfaction
The ultimate goal of a product is to satisfy a customer from all aspects of

cost, performance, reliability and maintainability. The customer trades o
these parameters when making a decision to buy a product. Assuming that
we are designing a product to a certain market segment, cost is determined
within limits.
The trade-os are
10
Customer Satisfaction
Performance Parameters
Availability
Maintainability
Reliability
Serviceability
Spare Parts Availability
Administrative Policies
Figure 1.2: Customer Satisfaction Model
11
Performance parameters
Reliability
Maintainability
Availability
Performance parameters are the designed in system capabilities such as
acceleration, top speed, rate of metal removal, gain, ability to carry a 5-ton
payload up a 40 degree grade without overheating, etc.
The reliability of equipment expresses the length of failure-free time that
can be expected from the equipment. Higher levels of reliability mean less
failure of the equipment and consequently less downtime and loss of use.
Although we will attach reliability numbers to products, it should be recognized that the customers perspective interprets reliability as the ability of
a product to perform its intended function for a given period of time without
failure.
This concept of failure free operation is becoming more and more xed in
the mind of the customer. This is true whether the customer is purchasing
an automobile, a machine tool, a computer system, a refrigerator, or an
automatic coee maker.
Maintainability is dened as the probability that a failed system is restored to operable condition in a specied amount of down time.
Reliability is not concerned with repair and for repairable systems availability is a better measure of equipment performance. Availability is the
probability that at any time, the system is either operating satisfactorily or
is ready to be operated on demand, when used under stated conditions.
The availability might also be looked at as the ability of an equipment
under combined aspects of its reliability, maintainability and maintenance
supportto perform its required function at a stated instant of time. This
availability includes the built-in equipment features as well as the maintenance support function. Availability combines reliability and maintainability into one measure.
There are dierent kinds of availability that are calculated in dierent
ways1,2 . The most popular availabilities are achieved availability and inherent availability.
1
VonAlven, W.H. Editor; Reliability Engineering, Prentice-Hall, Inc., Englewood Clis,

NJ, 1964.
2
ANSI/IEEE Standard 100-1988, 4th Ed.; IEEE Standard Dictionary of Electrical and
Electronic Terms, The Institute of Electrical and Electronic Engineers, Inc., New York,
NY, 1988.
12
Achieved availability includes all diagnostic, repair, administrative and

logistic times. This availability is dependent on the maintenance support
system. Achieved availability can be calculated as
A=
Operating time
Operating time + Unscheduled downtime
Inherent availability only includes operating time and active repair time
addressing the built-in capabilities of the equipment. Inherent availability
is calculated as
M T BF
A=
M T BF + M T T R
where
MTTR = mean time-to-repair
and the M T T R is for the active repair time. Active repair time is that
portion of downtime when the technicians are working on the system to
repair the failure situation.
The dierent availabilities are dened for various time-states of the system. Figure 1.3 shows dierent phases of system operation.
Serviceability is the ease with which the product can be repaired. Here
repair includes diagnosis of the fault, replacement of the necessary parts,
tryout and bringing the equipment back on-line. Serviceability is somewhat
qualitative and addresses the ease by which the equipment, as designed, can
be diagnosed and repaired. Factors such as accessibility to test points, ease
of removal of the failed components and ease of bringing the system back
on-line inuence serviceability.
1.5
Product Life and Failure Rate
Let us assume that we have released a population of products into the marketplace. The failure rate is observed as the products age. The shape of the
failure rate is referred to as a bath-tub curve (see Figure 1.4). Here we have
over emphasized the dierent parts of the curve for illustration.
This bath-tub curve has three distinct regions
Infant mortality
Useful life
Wear out
Available
13
Unavailable
System Status
Free/idle time
Active repair time
Operating time
Adjustment/start up
Wait for maintenance
Diagnostic time
Out of stock/fuel
Administrative time/Planned shutdown
Wait for spare parts
Figure 1.3: System Status Over Time
14
3.000
Failure rate - h(t)
2.500
Infant mortality
2.000
Wearout
1.500
1.000
Useful life
0.500
2000.000
1500.000
1000.000
500.000
0.000
0.000
Life time (t)
Figure 1.4: Bath-tub Failure Rate Curve
1.5.1
Infant Mortality Period
During the infant mortality period the population exhibits a high failure
rate, decreasing rapidly as the weaker products fail. Some manufacturers
provide a burn-in period for their products to help eliminate infant mortality failures.
Early failures might be due to:
Poor welds
Cold solder joints
Nicks, voids, cracks
Incorrect part positioning
Product deterioration in shipment
15
Average
Strength
Stress/
Strength
Levels
Average
Stess
Stress
Profile
t = time
t1 = End of infant mortality
t2 = Start of wearout
Figure 1.5: Stress-Strength Prole Over Time

Metal chips from manufacturing
Contamination
Substandard materials
Improper installation
Porosity in castings
Inclusions
Fundamentally, these failures reect the manufacturability of the components and assembly of the product with the given manufacturing capabilities.
1.5.2
Useful Life Period
During this period the population of products exhibits a relatively low and
constant failure rate. This failure rate is due to the designed-in reliability.
The failure phenomena during this life period is illustrated in Figure 1.5.
Product failure is due to:
Low safety factors
16
Strength
Stress
y
Interference Area
Figure 1.6: Stress-strength Interference Model

Higher than anticipated stresses
Lower random strength (product variability)
Abuse
Operator error
Defects which escape even the best available detection methods
Misapplication
Maintenance practices - (will also promote wearout)
Skipping preventive maintenance actions
Misalignment
Improper assembly
Lack of lubrication
The useful life period is also explained using the stress-strength interference model for reliability as shown in Figure 1.6. Here the stress distribution
represents the combined stressors acting on a system at some point in time.
The strength distribution represents the piece-to-piece variability of components in the eld. The interference area is indicative of a potential failure
when stresses exceed the strength of a component.
1.5.3
17
Wearout Period
At the onset of wearout, the failure rate starts to increase rapidly. When the
failure rate becomes high, replacement or major repair must be performed
if the product is to be left in service.
Wearout is due to a number of forces such as:
Corrosion or oxidation
Fatigue
Shrinkage or cracking such as in plastics
Hardening of rubber components
Frictional wear
Chemical change
Maintenance practices - (See also, useful life period)
Contaminated lubricants
Misaligned gear teeth
1.6
Product Design and Development Cycle
Developing a product that can be manufactured economically and consistently to be delivered to the marketplace in quantity and that will work
satisfactorily for the customer takes a well established and precisely controlled design and development cycle. Events must be scheduled to occur at
precise times to phase the product into the marketplace. To develop a new
internal combustion engine for an automobile takes about a three-year design cycle (down recently from ve years), while a new minicomputer takes
about 18 months.
Although the timing may be dierent for dierent companies, there are
similarities in the activities comprising a design and development cycle. The
following is representative of these activities.
1.6.1
Product Development Cycle
Market Research
Forecast need
18
Forecast sales
Understand who the customer is and how the product will be used
Dene the customer usage prole and duty cycle
Dene the external environment
Set broad performance objectives
Establish program cost objectives
Establish technical feasibility
Establish manufacturing capacity
Understand Governmental regulations
Understand Corporate objectives
Establish reliability and maintainability (R&M) requirements
Benchmark the competition - identify the best-in-class
Identify and dene the customers perceptions concerning
R&M
Dene realistic R&M objectives considering total program
factors - (Management cost and prot objectives, competition, past performance of company concerning R&M)
Concept Phase
Formulate project team
Formulate design requirements
Update and rene real world customer usage prole
Develop and consider alternatives
Open a Design Failure Mode Analysis (DFMEA) activity for the
program
If done correctly, this is the most important tool that can be
used to prevent R&M eld problems
Must be done as a team using experienced technical personnel
Must be regarded as a high visibility activity throughout the
product design/development cycle
Rank alternatives considering R&M requirements
Review and update quality and reliability history on past products
Past performance is indicative of future performance
19
Past eld problems provide information useful as a basis for

product improvements
Past serviceability problems provide information for maintainability improvements
Assess feasibility of R&M requirements
Using all of the available information, update the R&M metrics
Identify and document R&M risk areas in the product design
Design Phase
Preliminary design
Design calculations
Rough drawings
Select nal alternatives to pursue
Manufacturing feasibility of design approach (design for manufacturability and assembly)
Detailed design
Complete detailed design package
Update DFMEA to reect current design and details
Develop design verication plan
Develop detailed R&M model for the product
Evaluate product R&M using current design approach
Update R&M metrics using current level of information
Identify R&M risk areas in design
Prototype Program
Build components and prototypes
Write test plans
Component/subsystem tests
System tests
Eliminate design weaknesses identied in tests
Update R&M metrics based on test data
20
Identify R&M risks

Estimate achievable R&M using growth techniques
Estimate warranty expenses based on R&M
Re-evaluate feasibility of achieving R&M goals
Manufacturing Engineering
Process planning
Assembly planning
Capability analyses
Perform a process Failure Mode Analysis (PFMEA)
Finalized Design
Consider test results
Consider manufacturing engineering inputs (design for manufacturability/assembly)
Finalize design changes
Freeze design
A necessity to stabilize the product design/development program
Only violate the design freeze for situations impacting safety,
violation of government regulations, or documented instances that
will lead to severe customer irritation
Release to Manufacturing
Validate that product as manufactured meets customer requirements
Engineering Changes issued based on hard evidence
Manufacturing experience - make the produce better or more economical
Field experience - correct undesirable eld situations
Maintain detailed warranty history
Needed for continuous improvement
Identify exactly what failure, how it failed and why it failed
21
Strategic Planning
Concept Design
Design\Development
Engineering
Prototype/Subsystem/System
Testing
Pilot Production
Production
Maximize reliability efforts
Maximize quality efforts
Figure 1.7: Simplied Overview of a Product Design and Development Cycle

A little good information is better than lots of imprecise information
Example 1.2Product design and development.

Briey outline the product design and development cycle as used by your
company.
1.7
Reliability in Design
The cost of unreliability is

High warranty costs
Field campaigns
Loss of future sales
Cost of added eld service support
It has been demonstrated in the marketplace that highly reliable products (failure free) gain market share. The evidence suggests that the mission
of a reliability program is to estimate, track and report the reliability of
hardware before it is produced. The reliability of the equipment must be
22
reported at every phase of design and development in a consistent and easy

to understand format.
Warranty cost is an expensive situation resulting from poor manufacturing quality and inadequate reliability. For example, the cost to the corporation for a Chrysler automobile may be as high as $850 per vehicle3 . From
this same article one could deduce that the cost per vehicle for General Motors is about $350 and for Ford $650. This would be to cover the 36,000
mile warranty currently in eect. For each car sold, the manufacturer must
collect and retain this expense in a warranty account.
Cost of Engineering Changes and Product Life Cycle. The stage of
product development/manufacturing and the cost of an engineering change
is typically illustrated as
Prototype <$20,000
After start of production >$100,000
Various articles in trade magazines estimate the magnitude of this cost
penalty to grow by a factor of ve to ten. Early detection of potential
R&M concerns is cost eective.
Warranty cost issue hurts Chrysler, USA Today, Tuesday, October 24, 1994, pg. 3B
Chapter 2
Design Review
The design process is an orderly, timed and controlled process that results
in the delivery of a product to the marketplace. The engineering steps in
creating the initial product and ensuring that it meets customer demand
and is ready for the marketplace include
Concept design
Detailed design/development
Prototype build/tryout
The production function is then
Build/production
and the customers function is
Operations/maintenance
In addition, another function is becoming increasingly important due to
environmental concerns. This function is
Retirement/reclamation
The responsibility for this function is still developing and may either be the
responsibility of the producer or the user. The design function must ensure
that all of these interests are addressed, and the design review procedure is
a means used for this purpose.
A design review is a formalized, documented and systematic study of the
technical design aspects of a product. The review is done by senior technical
23
CHAPTER 2. DESIGN REVIEW
24
managers to assess the design and development program from a technical

standpoint as opposed to cost and schedule, to identify any apparent design
weaknesses.
The review eort should be
formalized as part of the design procedure
scheduled as part of the design activity
multi-phased to address various aspects of product design in a timely
fashion
documented to provide a case history for all management and technical
personnel.
The design review procedures must be well developed, endorsed and participated in by top management.
In the development of design review procedures for a company, the following must be addressed:
Policy
Procedures
Chairperson
Participants
Timing of reviews
Objectives of each review
Documentation requirements
Development of checklists
Follow-up procedures
The procedure must be totally integrated into the product design and development cycle. This means that sign-o is necessary to proceed past gates
in the design and development program. The number and location of these
gates will depend on the particular product. For example, computer hardware must be developed in a very short time span of possibly a year to
18 months, while a new automotive transmission design might take three
to four years to develop and bring to market. Because the products are
dierent, they require a dierent structure to the review procedures.
Product Flow
25
Reliability Program Management
Reliability Considerations
Design Failure Mode

Analysis
Reliability/
Maintainability Estimation
Initial Product Concept
Preliminary
Design Review
Detailed Design/
Development
Intermediate
Design Review(s)
Reliability estimation
Maintainabiilty analysis
Shipping/transportability
Purchased parts reliability
Design for manufacturability/assembly
Prototype Build
System/Component Tests
Design Requirements
Parts Characteristics
Process Failure Mode

Analysis
Reliability/maintainability requirements
Performance requirements
Process Planning
Detailed Design
Review
Process Review
Process induced failure modes

Manufacturing capability (quality)
Assembly induced failures
Production
Figure 2.1: Product Planning Flow Chart
2.1
26
Policy
The design review procedure is a major tool for addressing and understanding the complex trade-os in a product design program and ensuring that
superior products are developed in a timely and cost-ecient fashion. Technical design reviews must be conducted on all new products and on major
revisions of existing products.
2.2
Appointment of Chairperson
The position of the chairperson requires a high level of technical management

visibility. This task requires tact, a broad understanding of design and
technical knowledge of the various disciplines involved.
Participants. The technical talent at each review meeting varies with the
product and objectives of the particular review. Generally the participants
will include senior personnel from engineering, manufacturing, tool design,
marketing, testing, purchasing, reliability and maintainability, quality and
value engineering. The review may also involve specialists from other activities in the company or from outside the company.
A review committee can easily get too large. No more than a dozen
participants should be involved in any one review.
An overview of a three phase review procedure and its participants is
given in Table 2.1.
2.3
Timing of Reviews
It is necessary to have design review meetings at critical points (milestones)

in the product design and development cycle. For example, a review may
be held when
marketing requirements are completed
design concepts have been identied and a feasibility study completed
specications and drawings are nearing completion
tooling concepts are completed
testing of prototypes has been completed, but before manufacturing
release
27
Table 2.1: Design Review Committee Members

Review member
Chairperson
Phase of design review

Concept Interm. Final
X
X
X
Design engineer
(of the product)
Design engineers
Reliability Maintainability
Manufacturing
engineer
Materials
neer
engi-
Marketing
Stress analysis
Quality control
Human factors
Safety engineering
Transportation
X
X
Field
nance
Corporate
sta
mainte-
legal
Responsibility
Conduct reviews, issue major reports, monitor follow-up
Prepare and present design approach, major trade-os, major concerns. May be called for input on
every review.
Review and verify adequacy of design, identify weaknesses, concerns.
Must evaluate existing design approach.
Provide estimates of the R&M levels
for the product.
Address manufacturability of the
product, check for tooling adequacy
and assembly problems.
Identify materials incompatibilities
which might promote corrosion,
stresses due to thermal expansion,
etc.
Review customer requirements, provide market demand analysis.
Provide a nal audit of all stress calculations, load levels, etc. Identify
unknowns in stress levels.
Identify inspection and testing problems, review tolerances and manufacturing capabilities.
Provide adequate considerations for
human operator, identify potential
human induced problems.
Audit safety analysis for product.
Identify potential packaging, shipping and handling problems.
Analyze maintainability of the product. Identify eld servicing problems.
Review governmental, safety and
commerce regulations.
28
assembly methods and hardware have been identied.
2.4
Scheduling of Reviews
The timing of the reviews must be incorporated into the product development cycle. Procedures for resolution of product problems in a timely
fashion must be established.
2.5
Advance Information
At least ten days before the actual date of the review, information will be
distributed to the participants. The information may include specications,
a reliability analysis or serviceability assessment, preliminary layouts, etc.
Distributing this information beforehand will help to ensure that participants are well prepared to contribute to the objectives of the review.
2.6
Agenda
The chairperson will have the agenda prepared for the meeting and will
distribute it with the advance information. At the meeting, the agenda will
be followed religiously.
2.7
Checklists
Checklists must be developed as an independent activity. These checklists

are then used by design engineers and the design review committee to ensure
that important points have been addressed. Checklists should be developed
for major technical areas of the product such as
hydraulics
electrical power distribution
transmission-engine matching
corrosion considerations
manufacturability
diagnostic assists
29
power supply sizing

Checklists are developed by bringing together technical talent to address
a specic area. For example, a machine tool company might bring talent
together from various divisions to address the area of hydraulics as it appears
on many dierent machines. The hydraulics design checklist would provide
a listing of cautions to be used in auditing the design.
2.8
Conduct of Design Review Meeting
Introductory comments by the chair should set a constructive tone and climate for the meeting. The specic objectives of the design review should
be stated and should be related to the overall objectives, namely, achieving
optimum product design from the standpoint of reliability, maintainability,
function, cost and appearance.
A design review is not a redesign of the product but is simply an audit
of the design and design approaches to uncover any identiable weaknesses
or potential problems. Inferior products will result in a competitive disadvantage. Everyone benets from a good design review.
A secretary will be part of each review to take notes, summarize meeting
outcomes, keep records and distribute items for the review process. The
minutes must be distributed within 24 hours.
The chairperson should make sure that the discussions follow a systematic plan so that no major subject areas are omitted. The discussion should
follow the prepared agenda. Checklists should be used to prevent omission
of important design considerations.
Follow-up of Findings. The design function is responsible for investigating and incorporating those ideas which will aid in achieving optimum
product performance in the eld. A communication link must be maintained
between design and the review committee. Each action item identied by
the committee must be reported on formally to close the loop. The reports
are part of the design review documentation.
A time schedule for closing the loop on outstanding items must be part
of the total review procedure. The time schedule should mandate that open
items are forwarded up the administrative ladder in a timely fashion until
resolved.
2.9
30
Final Report Documentation
A nal report is mandatory and must cover the total review process. The
nal design review report and documentation provides a historical record of
the product design and development process. This record serves to improve
the process in the future and provide a basis for better reviews on similar
products.
Chapter 3
Reliability Measures
Reliability is the probability that a product will perform its intended function, satisfactorily, for a specied interval of time when operating under
specied conditions.
Reliability is concerned with the life of a system from a success point
of view. It is assumed that the time of failure can be precisely determined.
This requires a clear denition for all system failure modes.
The loading conditions on the system and the operating environment
must be clearly understood and dened. Commercial products typically
experience a broad range of loading conditions and environments. The levels
to be used in design must be determined through management decisions.
Then use of the product with higher levels than anticipated in design will
promote higher failure rates.
Typically, manufacturers specify loading conditions for the 95 percentile
customer and similarly for environmental conditions. This means that the
product is over designed for the average customer. A second approach is
to put cut-of devices in the product that shut down the system if severe
usage or environments are encountered. Of course, this approach can irritate
the customer.
The random variable used to measure reliability is the time-to-failure
random variable (t). The random variable might represent time, cycles,
copies, revolutions or some other suitable indicator of life for the product.
31
CHAPTER 3. RELIABILITY MEASURES
32
100%
R(t)
75%
50%
25%
5000
4000
3000
2000
1000
0%
Cycles to failure (t)
Figure 3.1: Reliability Functions
3.1
The Reliability Function
The reliability function is a mathematical equation that provides the reliability at any given time, t. Let
t = the time-to-failure random variable
then the reliability function is dened as R(t), where
R(t) = Pr(Product life > t)
Typical reliability plots are shown in Figure 3.1. These reliability curves
will initially be at 100% and go down as the product ages.
For example, consider the data of Table 3.1 resulting from testing 50
light bulbs to failure. We can plot a histogram of the data as shown in
Figure 3.2.
33
Table 3.1: Time-to-failure Data for 50 Light Bulbs

Time interval (hr) Midpoint (hr) No. failures Cumulative
201-400
300
2
2
401-600
500
8
10
601-800
700
9
19
801-1,000
900
10
29
1,001-1,200
1,100
13
42
1,201-1,400
1,300
6
48
1,401-1,600
1,500
2
50
30%
25%
15%
10%
5%
Time (t)
Figure 3.2: Light Bulb Failure Data
1500
1300
1100
900
700
500
0%
300
Percent failing
20%
34
Table 3.2: Time-to-failure Analysis for 50 Light Bulbs

Time
Midpoint Percent Cumulative Reliability
Interval
(t)
failing
failure
curve
201-400
300
4%
4%
96%
401-600
500
16
20
80
601-800
700
18
38
62
801-1,000
900
20
58
42
1,001-1,200
1,100
26
84
16
1,201-1,400
1,300
12
96
4
1,401-1,600
1,500
4
100
0
In basic statistics courses we are introduced to the cumulative distribution, F (t), dened as
F (t) = Pr(Product life t)
The relationship between R(t) and F (t) is
R(t) = 1 F (t)
For the light bulb data, we show the plot of the cumulative distribution
in Figure 3.3. This plot is made from the data in column 3 of Table 3.2.
Example 3.1Plotting reliability curves.

The following data is on timing belt failures as observed in a controlled
vehicle test. Plot the data and estimate the 100,000 mile reliability. Recommend when the belt should be replaced as a statement included in the
vehicle service manual. A low value may not help your sales while a high
value may aect your reputation.
35
Cumulative Percent
0.75
0.5
0.25
2000
1500
1000
500
Time-to-failure (hours)
Figure 3.3: Cumulative Failure Plot for Light Bulbs

Mileage X1, 000
40-50
50-60
60-70
70-80
80-90
90-100
100-110
110-120
120-130
130-140
140-150
150-160
160-170
No. failures
1
1
2
2
3
6
14
22
31
8
6
2
2
Cume F
%F
%R
36
Example 3.2Using the reliability function.

Given the reliability function
R(t) = et/1,200
where t = time-to-failure in hours
A. Find the 100 hour reliability.
B. Find the 200 hour reliability.
C. Find the 1,200 hour reliability.
D. If 1,000 devices are placed in operation, how many will still be operating
without failure at 300 hours?

For the reliability function,
R(t) = e(t/800) , t 0
2


Given the reliability function,
R(t) = et/32,000 , t 0
B. Find the 3,000 hour reliability.
C. Find the B10 life (10% failure time).

Given the reliability function,
R(t) = e(t/1,200) , t 0
2.3
37
Table 3.3: Time-to-failure Data

Time-to-failure (hr) No. failures
300
2
500
8
700
9
900
10
1,100
13
1,300
6
1,500
2
C. Find the 1,200 hour reliability.
3.2
The Expected Life
Let us look at the light bulb data in Table 3.3, and calculate the average
life of a bulb. We would calculate
300hr 2 + 500 8 + 700 9 + 900 10 + 1, 100 13 + 1, 300 6 + 1, 500 2
t =
50
t = 900 hours
This could have been written as,
2
8
9
10
13
6
2
t = 300hr +500 +700 +900 +1, 100 +1, 300 +1, 500
50
50
50
50
50
50
50
This calculation is the time to failure multiplied by the relative frequency

of occurrence of that time.
We see that the average life is found from sample data by
t =
t Pr(t)
For any theoretical probability model there is an analogous concept called

expected life. We will now look at this concept of expected life.
The expected life is a theoretical quantity associated with a particular
probability density function used to model time-to-failure. The expected life
is dened as

t f (t)dt
E(t) = =
all t
38
This quantity E(t) is called the mean-time-between-failures (MTBF) for

systems that are repairable and the mean-time-to-failure (MTTF) for throwaway systems. For example, an airplane has a MTBF while a microprocessor
has an MTTF.
Example 3.6Estimating average life from data.

Four lift gate cylinders were cycled to failure. The failure times are 16,005
cycles, 20,326 cycles, 30,975 cycles, 61,788 cycles. Calculate the average life
of these lift gate cylinders.
3.3
The Failure Rate
The failure of a population of elded products is due to many factors (or

forcing functions) such as
The product design
The manufacturing variability
The variability in customer usage
The maintainability policies as practiced by the customer
The environments encountered.
The failure rate can change over the life of the product due to dierent dominating forcing functions. For example, if there is considerable variability in
manufacturing and bad product slips through the system, then a high early
failure rate can be expected. As the inferior products drop out, the failure
rate will stabilize at its lowest level until the onset of wear out. Later in the
product life, the failure rate will increase due to wear out.
Let us now consider how we calculate the failure rate. The failure rate is
the rate at which failures occur in a certain interval of life for those devices
that are surviving at the start of the interval.
The failure rate is calculated from life data as follows,
Failure rate =
M
N T
where
M
= number of failures in the life time interval, T
= number of survivors at start of the time interval
= length of the life time interval
Number Failing
39
M
T
Time
NumberSurviving
N-M
Figure 3.4: Failure Rate Calculation
Dt
time
N R(t )
Figure 3.5: Time and Number Failing

While theoretically,
h(t) =
f (t)
R(t)
This can be easily rationalized by considering that, for an original number of

devices, N , the number surviving at some arbitrarily selected time, t would
be
N R(t)
while the number of the original devices failing in an interval of time t,
would be
N f (t)t
This is illustrated in Figure 3.5. Therefore, the failure rate over the time t
is
N f (t)t
h(t) =
N R(t) t
40
So, we see that

h(t) =
f (t)
R(t)
which is called the hazard function (or theoretical failure rate). The theoretical failure rate can be found for any pdf through application of the above
formula.
Every probability density function (theoretical life distribution model)
has an implied reliability function and failure rate.
The failure rate can be used to nd out the mode of failure infant
mortality, wear out. Or, it can be used to select an appropriate pdf.
Example 3.7Calculating failure rate.

In the following, 320 city buses were observed for mileage to rst failure
incident. Calculate the failure rate for this population of busses.
Mileage Interval (x1,000)
00 -10
10 - 20
20 - 30
30 - 40
40 - 50
50 - 60
60 - 70
70 - 80
80 - 90
90 - 100
100 - 110
110 - 120
120 - 130
130 -
Number of Failures
1
4
8
16
13
21
21
30
30
28
43
76
25
4
Failure Rate
Example 3.8The case of the cheaper fan.

The failure characteristics of a particular brand of fan was of interest. The
fan was for a computer cooling application. In a large mainframe installation, all fans were replaced with this new brand. In order to keep accurate
records, as a new fan failed it was replaced with the old brand of fan.
Thus, all originally (new) installed fans were monitored. A total of 214 fans
were installed at the start of this project. The fans operated 24 hour/day.
Failure date on cooling fans follow:

Week
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
3.3.1
No. Failed
7
3
3
2
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
2
0
0
0
3
4
7
9
12
9
10
18
10
19
21
41
Failure Rate
Uses for the Failure Rate
The failure rate can be used to

Identify the mode of failure (infant mortality, useful life, wear out).
Select suitable reliability models.
42
Failure Rate
Infant mortality - failure rate is decreasing
Wear out - failure rate is increasing
Useful life period - failure rate is constant
Product Life
Figure 3.6: Failure Rate and Product Life

Figure 3.6 shows the relationships between product life phases and the failure
rate.
Chapter 4
Static Models
Static reliability modeling is a preliminary form of reliability block diagram
modeling used to analyze the necessary subsystem (component) requirements
to provide a given system reliability, or conversely, to determine what system reliability might be expected from known subsystem (or component)
reliabilities.
In order to model the system we must
dene system failure
identify the time period of interest.
Consider the simple physical system in Figure 4.1. Lets say we are interested
+24V.
S1
S2
SOL1
Figure 4.1: Simple Circuit Diagram

in the switch system reliability.
In static reliability modeling, we look at a particular switch (component,
subsystem) as having only two possible states: success-failure, as shown
43
CHAPTER 4. STATIC MODELS
44
in Figure 4.2. We cannot model the system until we dene system failSuccess mode
Switch
Failure mode
Figure 4.2: Component Failure Modes

ure/success.
Lets consider the following:
If our concern is de-energizing the solenoid coil, the reliability model
for the switches would be as shown in Figure 4.3.
Switch A
Switch B
Figure 4.3: Parallel Reliability Block Diagram

This is a parallel reliability block diagram. This block diagram means that
our denition of system success with be achieved if either switch A or switch
B performs its function satisfactorily.
If our concern is activating the solenoid coil, the reliability model would
be as shown in Figure 4.4.
Switch A
Switch B
Figure 4.4: Series Reliability Block Diagram
45
We now have a series reliability block diagram model. This block diagram
means that both switch A and switch B must perform satisfactorily for the
system denition of reliability to be achieved.
4.1
Static Reliability Modeling Process
Consider the following system:

System: Heavy duty o-road vehicle
Reliability denition: Durability failures
Environment: O-road conditions
Subsystem: Propulsion
Component: Transmission and clutch
Time period: 50,000 miles
The reliability block diagram for this system might be as shown in Figure
4.5. The reliability block diagram shows the necessary relationships of the
Intermediate prop
shaft & U-joints
Engine
Transmission
Front prop shaft

& U-joints
Differential
and carrier
Transfer Case
Rear prop shafts

& U-joints
Differential
and carrier
Figure 4.5: Truck Power Train

subsystems for system success.
We then consider each subsystem/component in the reliability block diagram and assign a success probability considering the environment, application and time period as illustrated in Figure 4.6. As a second example
consider the electro-mechanical system in Figure 4.7.
46
100%
100%
75%
75%
50%
50%
25%
25%
0%
0%
50000
40000
30000
20000
10000
Reliability
90%
Mileage
Transmission system
Environmental profile
Application
Information
Usage profile
Figure 4.6: Reliability Block Diagram Modeling Process for Truck Drive
System
Switch
47
Motor A
Mechanical
System A
Motor B
Mechanical
System B
Computer
and
Interface
Electronics
Number of switchings
Hours of operation for electronics
Duty cycle for motors
Mechanical systems wear out modes
M-A
Switch
Mech - A
Electronics
M-B
Mech - B
Figure 4.7: Reliability System Modeling for Electro-mechanical System
4.1.1
48
Denition of Symbols
We now dene the following symbols

Ei = Event that subsystem i operates successfully
Ri = Pr(Ei ) = Subsystem reliability
Rs = System reliability
Qs = 1 Rs = System unreliability
4.2
Series System Analysis
In a series system all subsystems must operate successfully for the system
to function. The reliability block diagram is given in Figure 4.8. System
1
Figure 4.8: Series Reliability Block Diagram

reliability is then due to the logical event
Rs = Pr(E1 E2 En )
and assuming independent events, we have
Rs = Pr(E1 ) Pr(E2 ) Pr(En )
Or,
Rs = R1 R 2 Rn
4.2.1
Basic Series Reliability Equations
The reliability equation for a series system is

Rs = R1 R2 Rn
which we write as
Rs =
n
Ri
i=1
We see that system reliability is limited by

Rs min Ri
i
49
That is, the system reliability is limited by the least reliable subsystem.
Under equal apportionment of reliability to each of the subsystems, the
component reliability necessary to achieve a system reliability is calculated
by

Rc = n R s
where n is the number of components and Rc is the necessary level of component reliability.
Example 4.1Series system analysis.

Calculate the reliability of the following series system:
0.99
0.90
0.85
Example 4.2Series system analysis.

Let us say we have a water control subsystem consisting of a timer, a
solenoid-operated water valve and a level sensor. The ten-year reliability
for each of these devices is given below:
Timer
0.92
Water Valve
0.87
Sensor
0.96
Figure 4.9: Water Level Control

A. Calculate the system reliability.
B. Using an equal apportionment of reliability to each component, nd the
minimum component reliability necessary to achieve a system reliability of 0.97.
50
Example 4.3Series production system.

Calculate the reliability of the pressworking operation in Figure 4.10.
Destacker
0.89
Draw Press
0.92
Unloader
0.94
Figure 4.10: Pressworking Operation
Example 4.4Series machining line.

Calculate the reliability of the machining system shown in Figure 4.11.
L oade r
0. 98
T ransfer No. 1
0. 99
Tr ansfe r No. 2
0. 98
Turn over
0. 92
Tr ansfer No. 3
0.96
Stacker
0.91
Figure 4.11: Machining Line
Example 4.5Welding assembly line.

A truck cab assembly line is under development. The line will use ve
welding robots.
A. If each robot has a 95% reliability, what is the total robot system reliability for the line?
B. To have a robot system reliability of 95%, what must each individual
robots reliability be?
51
Example 4.6Subsystem versus system reliability.

Consider the following system with 100 identical components each having
a reliability of 0.9999:
1
100
0.9999
0.9999
0.9999
0.9999
A. We would expect one subsystem out of ? to fail.

B. We would expect one system out of ? to fail.
Example 4.7Automotive engine transfer line.

An automotive engine transfer line has 23 stations. If each station has a
reliability of 0.90, what will be the transfer system reliability? What station
reliability is necessary to have a transfer system reliability of 0.90?
4.3
Parallel System Reliability Analysis
A parallel system is not considered to have failed unless all subsystems have
failed. The reliability block diagram for a parallel system is given in Figure
4.12.
1
Figure 4.12: Parallel or Redundant Reliability Block Diagram

To analyze the reliability for a parallel system, we consider the complementary event
Ei = Event ith subsystem fails
The probability of the event system failure would be
Qs = Pr(E1 E2 En )
52
and assuming independence, we have

Qs = Pr(E1 ) Pr(E2 ) Pr(En )
Or,
Qs = (1 R1 ) (1 R2 ) (1 Rn )
So, the basic equation for parallel (or redundant) system reliability analysis
is
Rs = 1 Qs = 1 (1 R1 ) (1 R2 ) (1 Rn )
which we write as,
Rs = 1
n
(1 Ri )
i=1
Example 4.8Parallel systems reliability analysis. The light source on an

optical check reader is a critical component. The machine stops immediately
if the light source fails. Heat and vibration are deleterious to the light source
and cause frequent failure. In order to improve the reliability, it is relatively
easy to add a redundant light source. The light source system reliability
block diagram is given below. Calculate the light system reliability.
Light Source
0.92
Light Source
0.92
Example 4.9Parallel system. Calculate the reliability of the following

parallel system.
53
PS - 1
0.92
PS - 2
0.92
PS - 3
0.92
Example 4.10System reliability gains through redundancy.

Calculate the reliability of each of the systems in Figure 4.13. Then calculate
the percent gain in reliability through addition of redundancy. Percent gain
would be calculated as
Percent Gain =
Rnew Rold
Rold
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
0.80
0.95
Figure 4.13: Redundancy and Gains in Reliability
4.4
54
Parallel and Series Combinations
Complex systems can be analyzed through application of the basic equations

for either series or parallel systems. Consider the following examples.
Example 4.11Complex system analysis.
0.90
0.70
0.80
0.60
0.80
0.70
0.90
0.60
0.60
0.90
0.95
0.85
0.92
55
Example 4.14Tailpipe forming line.

A tail pipe line is automated with an auto-loader, auto-crimp and cut-o.
However, the auto-crimp has experienced problems in the past so a backup
hand crimp is provided. The reliability block diagram follows:
Hand Crimp
0.90
Auto-Loader
0.99
Cut-off
0.98
Auto-Crimp
0.94
A. Calculate the system reliability assuming that the hand crimp operation
is an acceptable alternative.
B. Calculate the gain in reliability by adding the backup hand crimp station.

0.90
0.80
0.95
0.70
0.98
0..65
0.70
4.5
Design Considerations-Static Reliability Models
The product design and development function sets an upper limit on product
reliability. It is important to review our reliability concepts to see what they
tell us about product reliability in design.
Design determines:
A. Product reliability by
56
consideration of design alternatives

system conguration
selection of materials
use of safety factors
thoroughness of design specications.
B. Field maintainability and serviceability by
providing on board diagnostics (built-in-test)
designing for ease of replacement of failed components
modular replacement
module size considerations
quick disconnects.
4.5.1
Series System and Reliability
Single point failure. The failure of any series component/subsystem will

result in the failure of the system.
Reliability decreases rapidly as the number of components (or subsystems) increases (see Figure 4.14).
System reliability is always less than (or equal to) the least reliable
subsystem.
The customer frequently views a product as a series system, because
any failure is an irritant.
In a series system every subsystem is critical.
4.5.2
Design Considerations - Series Systems
Diligently attempt to minimize the number of components.

Combine components and subsystems into an improved subsystem
wherever possible.
Simplify design wherever possible.
Derate components to promote long life and high reliability.
57
100%
R=0.99
System Reliability
75%
R=0.98
50%
R=0.95
25%
40
30
20
10
0%
Number of Components
Figure 4.14: Series System Reliability Degradation

Make conservative choices for components and materials.
Use eld proven, o-the-shelf, components wherever possible.
...Thorne had selected the new lithium-ion batteries from Nissan, which were extremely ecient on a weight basis. But they
were still experimental, which to Eddie was just a polite word
for unreliable1 .
4.5.3
Parallel System and Reliability
Redundancy. The existence of more than one means for accomplishing a

given function/task. Each means of accomplishing the function/task need
not be identical.
Redundancy is a means of increasing reliability by providing alternative
ways to accomplish a given task.
1
Crichton, M.; The Lost World, Alfred A. Knopf, Inc., New York, 1995.
4.5.4
58
Design Considerations - Parallel Systems
The advantages in providing redundancy are:

Quick solution if time is a major concern
An easy and natural solution in some design situations
The only solution if the reliability requirement for the components and
materials are beyond the state-of-the art.
The disadvantages in providing redundancy are:
May not be feasible or possible from a design standpoint
Switching mechanism may be needed and may defeat reliability gains
In highly reliable systems the gains will be small (see Figure 4.15)
May exceed weight and size limitations
Cost may be prohibitive
Series failure rate will increase, resulting in increased maintenance
costs
Power consumption may increase.
100%
m=4
90%
Number of Components
m=3
80%
m=2
70%
m=1
60%
1.1
0.9
0.8
0.7
0.6
0.5
50%
System Reliability
Figure 4.15: Parallel System Reliability Improvement
4.6
59
High level versus Low level Redundancy
In order to illustrate a design strategy for redundancy, consider that we have

a completely uid design situation where we can either provide redundant
components or redundant systems. Figure 4.16 shows the concept of high
level versus low level redundancy. Lets analyze the two situations.
1
Figure 4.16: High Level versus Low Level Redundancy
4.6.1
60
High Level Redundancy
In the situation below, we provide backup systems rather than redundant

components. We term this high level redundancy.
4.6.2
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
Low Level Redundancy
In the situation below, we provide backup components and recalculate reliability. Less reliable components will give greater advantages in these calcu0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
0.95
lations, and the low level redundancy will always be better from a reliability
standpoint.
4.6.3
Conclusion on Providing Redundancy
In comparing the two systems, it is observed that low level redundancy will
always provide a higher reliability. Although designs are never completely
uid, the best approach is to provide redundancy at the lowest possible level
in the system.
Further Design Considerations
Series components/subsystems represent single point failures. That is, a
failure at this point, fails the system.
For very highly reliable systems (such as outer space systems), all single
point failure modes need to be identied and designed out or redundancy
needs to be incorporated.
61
A secondary approach is to take special care in controlling the manufacturing/assembly process through inspection and testing.
Single point failures would be identied using failure mode analysis.
Chapter 5
Life Distributions
The time-to-failure for a population of devices follows a distinctive pattern
unique to the particular device and failure phenomena. We model the timeto-failure with a probability density function, pdf.
Consider the histogram of Figure 5.1, representing a plot of 50 devices
tested to failure to illustrate this modeling process. The histogram is an
empirical probability density function obtained from test data. The best
t Weibull pdf to this data is shown in Figure 5.2. We could t a normal,
lognormal or exponential to this same data. Each pdf would attempt to t
the pattern of failure. The particular parameters of a pdf are what makes
the pdf t the data. In order to t the pdf to the data, the parameters must
be estimated from the data.
In this section we will cover four probability density functions used to
model time-to-failure. The probability density functions are the exponential,
normal, lognormal and Weibull.
5.1
The Exponential Distribution
The probability density function for the exponential distribution is pictured

in Figure 5.3.
5.1.1
Probability Density Function

1
f (t) = et/ , t 0
62
(5.1)
CHAPTER 5. LIFE DISTRIBUTIONS
63
10
Frequency
125
115
95
105
85
75
65
55
45
35
25
15
Time-to-failure
Figure 5.1: Histogram of Failure Data
5.1.2
Parameter
= Mean life
1
= Failure rate
=
5.1.3
Or
5.1.4
(5.2)
(5.3)
Reliability Function
R(t) = et/ , t 0
(5.4)
R(t) = et , t 0
(5.5)
Hazard Function
1
(5.6)
This is commonly called a failure rate and is constant for all values of time,
t.
h(t) = =
64
20
f(t)
15
10
100
75
50
25
Time-to-failure
Figure 5.2: Probability Density Function Fitted to Failure Data
Example 5.1.
A small monitoring and control unit consists of an MC6802 microprocessor,
an MCM2716 EPROM, an MC6821 PIA and support electronic components
such as decoders, resistors and capacitors. Considering the environment, the
failure rate for this system is estimated to be = 23.71 106 /hour.
A. What is the mean time-to-failure (M T T F ) for this module?
B. What are the chances that a system in continuous operation will last
for 30 days?
C. What are the chances that a system will last for 60 days?
5.1.5
Specied Time for a Reliability Level
The time t, at which a particular reliability level R, is achieved is given by

1
t = ln
R
(5.7)
65
0.006
0.005
f(t)
0.004
0.003
0.002
0.001
800
600
400
200
Time (t)
Figure 5.3: Exponential Probability Density Function
Example 5.2.
A targeting station with controls and display screen has a failure rate of
= 1.22 103 /hour in a specied usage environment.
A. What is the M T BF ?
B. If the station is on continuously, what is the targeting station reliability
for a 12 hour mission?
C. For a targeting station reliability of at least 95%, how long can the
mission last?
Example 5.3.
A hard disk card has an MTBF of 28,000 hours. The card is warranted
for one year. Assume an average annual usage of 640 hours, what is the
one-year reliability?
5.2
The Normal Distribution
The pdf for the normal distribution is given in Figure 5.4.
66
R(t)
t = time
Figure 5.4: Normal Distribution
5.2.1
Probability Density Function

1 t 2
1
f (t) = e 2 ( ) , < t <
2
5.2.2
Parameters
= Mean life
(5.9)
= Standard deviation
5.2.3
(5.8)
Hazard Function
h(t) =
(5.10)
R(t)
(5.11)
This is an increasing function of t.
5.2.4
Standard Normal Density Function

z2
1
z (z) = e 2 , < z <
2
(see Figure 5.5)

Use of the Standard Normal Distribution
Example 5.4.
Use of the standard normal tables.
(5.12)
67
s=1
z = standard normal deviate
Figure 5.5: Standard Normal Distribution
s=1
Tabled area
z
z=?
Figure 5.6: Standard Normal Tables
-1.76
68
+1.76
Figure 5.7: Symmetrical Relationship of Standard Normal

A. Pr(z 1.50) =
B. Pr(z > 1.50) =
C. Pr(z 1.92) =
D. Pr(z 2.12) =
E. Pr(z 2.92) =
5.2.5
Standard Normal Geometric Relationships
The standard normal distribution is centered at zero and symmetrical about

this value. This symmetry can be used to nd values in normal tables.
Example 5.5.
Use the standard normal tables to nd the following probabilities.
A. Pr(z 1.44) =
B. Pr(z > 1.44) =
C. Pr(z 0.65) =
D. Pr(z 2.17) =
69
0.8264
z
z=?
Figure 5.8: Inverse Use of Standard Normal
0.27
0
z
z=?
Figure 5.9: Finding the Z-value for a Given Area
5.2.6
Inverse Use of Standard Normal Tables
Here we have the area above a specied z-value.
Example 5.6.
Find the z-value for the following probabilities:
70
t = time-to-failure
z = standard normal
Figure 5.10: Use of the Standard Normal Tables

Probability (P )
0.10
0.40
0.75
0.80
0.90
0.95
5.2.7
Standard normal value (z)
Translation of any Normal Distribution to the Standard Normal
The equations to translate between the axes are

t
(5.13)
t = + z
(5.14)
z=
and
(see Figure 5.10).
Example 5.7.
We are concerned with a pressure sensor. We have determined that the
71
sensors turn-on when a pressure of 152 psi is reached (on the average).
We have found a standard deviation of 14 psi among sensors. Find the
following:
A. Pr(t 120 psi) =
B. Pr(t 160 psi) =
C. Pr(t 110 psi) =
D. The range in which 90% of the sensors will fall.
Example 5.8.
The average copies to failure for a plastic gear train is 400,000 with a standard deviation of 20,000 copies. This gear train is used on disposable cartridges for a particular brand of copier. Find the following:
A. The percentage of copier cartridges that last 350,000 copies without
requiring gear train maintenance.
B. The B10 life.
C. This gear train is attached to throw away cartridges that will be used
for no more than 12,000 copies. Will the reliability be adequate?
Example 5.9.
Carbon zinc batteries have a mean life of 28 hours and a standard deviation
of 3.5 hours when operating under a constant load current of 50 ma.
A. What are the chances that a battery will last longer than 24 hours?
B. For a 90% reliability, how many hours usage should be specied?
C. For the battery layout in Figure 5.11 (not a reliability block diagram)
what is the 24-hour reliability for the battery system?
Example 5.10.
A 2 12 -ton 4x4 truck has front leaf springs. The main leaf has exhibited a
life that can be reasonably approximated by the normal distribution. The
parameters are = 122, 250 miles and = 14, 000 miles.
A. What is the 100,000 mile reliability for the main leaf?
B. What is the 100,000 mile main leaf reliability for the vehicle?
72
V2
V1
150 ma
RL
Figure 5.11: Battery System
73
C. If we want the vehicle to experience no more than one failure per 100,000
vehicles over the 100,000 mile period, what mean life is required, assuming that the standard deviation does not change?
Example 5.11.
Fit a normal distribution to the following life data and estimate the chances
of surviving for 46,000 cycles. Also, specify the number of cycles for a 0.999
reliability.
Cycles to failure
46,079 45,985
45,907 45,931
46,096 45,929
46,021 45,846
45,961 45,958
5.3
Lognormal Distribution
The lognormal distribution is a nicely shaped model that would t many life
situations because it is almost as versatile as the Weibull; however, it is not
often used to model time-to-failure. The reason is probably partly historical
and partly by chance. Use of the Weibull among engineers was prompted
by an article in Applied Mechanics1 . In practice, the lognormal is used to
model time to repair for maintenance analysis.
5.3.1
Lognormal Probability Density Function
The log-normal probability density function is

1 ln t 2
1
f (t) = e 2 [ ] , t 0
t 2
5.3.2
(5.15)
Parameters
The parameters are and . These parameters are not the mean and
standard deviation but dene the mean and variance as follows:
Mean = e+
and
Variance = e2+
2
2
(5.16)

e 1
(5.17)
1
Weibull, W., A Statistical Distribution of Wide Application, J.Appl.Mech. V. 18,
1951.
74
f(t)
t = time to
failure
Figure 5.12: Lognormal Distribution
5.3.3
Hazard Function
The hazard function is given by

h(t) =
ln t
tR(t)
(5.18)
where z () is the standard normal pdf, and R(t) is the reliability at t. This
hazard function is not monotone increasing, but is used in situations where
an increasing hazard function would be appropriate.
5.3.4
Calculation of Probabilities for the Lognormal
The lognormal can be translated easily to the normal distribution allowing

standard normal tables to be used for nding probabilities. The translation
to the standard normal variate z is
ln t
Of course this formula can be rearranged to give

z=
t = e+z
(5.19)
(5.20)
75
Example 5.12.
Consider a lognormal distribution with = 2.0 and = 1.25, where t is
time-to-failure in kilometers. Find the following:
A. Pr(t 24 km) =
B. What is the mean life?
C. What is the B10 life.
Example 5.13.
The repair time for a system is represented by a lognormal distribution with
= 2.8 and = 1.01. Time is in minutes to diagnose and x a failure,
assuming that the spare parts are available.
A. What is the average time to repair?
B. What are the chances that a repair will take longer than one-half hour?
C. What are the chances that a repair will take less than the average time?
D. Find the time interval in which 90% of the repairs can be expected to
be accomplished.
Example 5.14.
The time to diagnose and repair the carriers on an electronic guided conveyor
system has been recorded as follows:
Time to repair in minutes
5.2
3.5
4.3
3.6
3.2
5.6
2.3
4.2
2.7
3.0
Find the parameters for the lognormal distribution and estimate the percentage of repairs taking longer than ve minutes. Also estimate a time in
which 90% of the repair actions will be accomplished.
5.4
Weibull Distribution
The Weibull distribution is a very popular model for time-to-failure of mechanical components. A representation of the probability density function
is shown in Figure 5.13.
76
Time
Figure 5.13: Weibull Probability Functions
200
150
100
50
0
0
pdf
5.4.1
77
Weibull Probability Density Function
The pdf for the two-parameter Weibull is

1
t
f (t) =

e(t/) , t 0
(5.21)
where t = time-to-failure.
5.4.2
Parameters
The parameters are

= Weibull slope
= Characteristic life
5.4.3
The reliability function is

R(t) = e(t/) , t 0
5.4.4
(5.22)
Mean and Standard Deviation
The mean and standard deviation involve Gamma functions. The formulas
for the mean and standard deviation are

1
= 1+
(5.23)
and
5.4.5
1+
2 1 +
(5.24)
Hazard Function
The Weibull hazard function is

1
t
, t0
(5.25)
For this function, we can see that for the dierent values of , the hazard
function will behave as follows:
h(t) =
> 1 Increases
= 1 Constant
< 1 Decreases
5.4.6
78
Specied Time for a Reliability Level
The reliability level will drop as time goes on. The relationship between
reliability R and time t is

1
t = ln
R

1/
(5.26)
Example 5.15.
A speed reducer used in a machine tool application has a Weibull time-tofailure distribution with = 2.71 and = 1, 320 hours.
A. What are the chances that the reducer will survive for 500 hours of
operation?
B. How often should the reducer be replaced so that there will be no more
than a 2% chance of failure?
Example 5.16.
The Weibull is used to predict gear life for distributor gears on a high-speed
wear test. For this test we have found = 3.12 and = 21.7 hours.
A. Find the B10 life.
B. Find the probability of failure prior to 10 hours.
C. Find the probability of surviving for 18 hours?
Example 5.17.
For each time-to-failure distribution dened below:
Exponential with = 5, 000 miles
Weibull with = 3 and = 5, 000 miles
Weibull with = 0.5 and = 5, 000 miles
Normal with = 5, 000 miles and = 500 miles
Find the following:
A. The probability that a vehicle will go 500 miles without failure.
B. The probability that a vehicle will survive for 500 more miles without
a failure, given that it has already survived for 5,000 miles with no
failures.
79
Failure Rate
Infant Mortality
Wear out
Machinery Life
Figure 5.14: Bath-tub Hazard (failure rate) Function
5.5
Distribution Selection
Distribution
Exponential
Failure rate
Constant
Normal
Lognormal
Increasing
Use where an increasing failure rate would
be appropriate
Either increasing, decreasing or constant
Weibull
5.5.1
Application
Use during useful life of product, not applicable in wear-out
phase, electrical components,
system level reliability.
Wear-out, fatigue.
Fatigue failures, maintenance
analysis.
Mechanical components, fatigue failure, components experiencing wearout.
Bath-tub Hazard Function and Distribution Selection
The bath-tub hazard function of Figure 5.14 shows the life regions in which
the pdf models apply.
Chapter 6
Exponential Distribution
The exponential distribution is a very popular model for reliability analysis.
It is important that this model be used correctly as it has denite limitations
due to its constant failure rate. The probability density function for the
exponential is shown in Figure 6.1.
The probability density function (pdf) models the time-to-failure phenomena. For the exponential distribution, the pdf is given by
1
f (t) = et/ , t 0
(6.1)
where
t = time to failure
The time variable (t) can be in hours, miles, revolutions, cycles, as appropriate for a particular application.
6.1
Mean Life Parameter
For products that are repairable,

= MTBF
the mean-time-between-failures. And for products that are not repairable,
= MTTF
the mean-time-to-failure. For example, we would say that an automobile
has an MTBF (or possibly MMBF, mean miles between failure) while an
MC6800 microprocessor has an MTTF.
80
CHAPTER 6. EXPONENTIAL DISTRIBUTION
81
12
10
f(t)
500
400
300
200
100
Time
Figure 6.1: Exponential Probability Density Function
6.2
The reliability function for the exponential distribution is given by
or
6.3
R(t) = et/ , t 0
(6.2)
R(t) = et , t 0
(6.3)
Time and Mean Life Formulas
The relationships between t, , and R are given by

t = ln
=
=
1
R
t
ln(1/R)
ln(1/R)
t
(6.4)
(6.5)
(6.6)

B10 = ln
6.4
1
0.90
82
(6.7)
Failure Rate
The failure rate is
1
=
So the failure rate and mean-time-to-failure are related by

h(t) =
(6.8)
(6.9)
and this is a constant for all t. This failure rate is calculated per mile, hour,
revolution, etc.
Example 6.1.
An automatic coee maker has a failure rate of = 2.7 105 /hour.

A. What is the MTTF?
B. What is the 5,000 hour reliability?
C. What is the one-year reliability for the customer?
D. What is the B10 life?
Example 6.2.
A universal I/O subsystem base module has a failure rate of 3.00104 /hour.
In a 7 24 operation, nd the six-week reliability.
Example 6.3.
A 5-ton 6 6 truck has an MMBF of 1,750 miles when used in cross country
terrain.
A. What is the reliability for a 75 mile mission?
B. What is the 1,750 mile reliability?
6.5
Mean Life and Reliability
The mean life and reliability are related in all exponential distributions. The
reliability function gives us
R(t) = Pr(Product life > t)
83
For the exponential, the reliability function is

R(t) = et/
and the mean life is
MTBF =
Then calculation of the reliability at the mean life gives
R() = e/ = e1 = 0.368
So the chance of surviving the mean life is 36.8% in the case of the exponential distribution.
Example 6.4.
An automotive transmission system should not experience more than one
failure per 1,000 vehicles in the eld, over the rst 50,000 miles of usage.
What is the implied MTBF?
6.6
Applicability of the Exponential Model
Used during the useful life of the product.

Wear-out must not be the dominant mode of failure.
The failure rate must be relatively constant.
Commonly used to model reliability of electronic systems/components.
Useful for system level prediction for large systems.
6.7
Warranty Expense Calculations
Let
= failure rate
tW
= warranty period
Cr = repair or replacement cost

The per-unit failure rate under warranty would be
tW
84
and the per-unit warranty cost would be

Cr tW
Note that and tW must have compatible units.
Example 6.5.
A smart terminal has an MTBA (mean-time-between-attentions) of 20,000
hours. These terminals are warranted for one year. A service call costs $130,
on the average. What should the manufacturer budget for warranty expense
for the one-year period?
Example 6.6.
A certain electronic module mounted on an automotive engine has an alleged
failure rate of = 4 106 /mile. If this module cost $85 to replace under
warranty, what should the manufacturer budget for warranty expense? The
warranty period is 50,000 miles.
6.8
Estimation of Mean Life
The mean life () is estimated from life data as follows:

T = total accumulated test and/or operating time
This value (T ) includes the time accumulated on both failed and unfailed
units.
r = total number of failures
Then the MTBF is estimated by
T
=
r
(6.10)
=1= r
(6.11)
The failure rate is estimated by
Example 6.7.
The following data indicate the occurrence of failures on seven trash compactors tested in a laboratory. All compactors were repaired when failure
occurred, and the test was continued. What is the MCBF (mean-cyclesbetween-failures) for this product?

Compactor No.
1
2
3
4
5
6
7
Cycles at failure
1,998; 4,308
No failures
3,223
1,933; 2,729
2,674; 5,328
No failures
297; 4,322; 5,047
85
Current test status (S/F)

5,084 cycles; S
5,482; F
4,771; S
4,360; S
5,536; S
4,121; S
5,089; S
Example 6.8.
Forty modules were placed on life test for 20 days (24 hours per day). Failed
modules were replaced on the test stands. The test produced two failures.
Estimate the MTBF.
Example 6.9.
Modules are burned-in for eight hours in production by running them on a
conveyor through a hot oven. In an eight-hour shift 3,200 modules were run
through the test with nine failures.
A. Estimate the failure rate relative to this burn-in test.
B. If the acceleration factor for this test is 2,000, estimate the eld failure
rate assuming that the failed units were not infant mortality failures.
C. If one-third of the failures were assumed to be due to infant mortality,
what would your answers be?
6.9
Estimating the Reliability Function
The reliability function is obtained by using the estimated value of . That

is, if
= estimate of the MTBF
then
= et/
R(t)
(6.12)
Example 6.10.
A local area network controller consists of a circuit card with processor,
interface electronics, connectors, etc. and a small amount of memory. Forty
of these modules were placed on life test for 60 days (operating 24 hours
per day). Failed modules were replaced on the test. Three failures were
observed.
86
A. Estimate the MTBF.

B. Estimate the 1,200 hour reliability.
6.10
Condence Limits
In the case of the exponential, the condence limits use critical values from
the chi-square distribution. The values can be found in statistical tables
using the degrees of freedom and level of signicance (). The condence
level and level of signicance are related by
Condence level = C = 1
Or
=1C
For a 90% condence level, we would have = 0.10.
6.10.1
Two-sided Failure Truncated Limits

2T
2T
2
2/2,2r
1/2,2r
(6.13)
where
r = a preselected number of failures
6.10.2
Two-sided Time Truncated Limits

2T
2T
2
2/2,2r+2
1/2,2r
(6.14)
where
T = a preselected amount of testing time
Example 6.11.
Fifteen automotive A/C switches were cycled and observed for failure. The
test was suspended when the fth failure occurred. Failed switches were not
replaced. The failures occurred at the following cycles:
Cycles
1,410
1,872
3,138
4,218
6,971
87
When the fth switch failed, all of the remaining ten pieces on test were
suspended, each having accumulated the 6,971 cycles.
A. Estimate the mean-cycles-to-failure (MCTF).
B. Set a 90% condence interval about the MCTF.
C. Set a 50% condence interval about the MCTF.
6.10.3
One-sided Failure Truncated Limits

2T
2,2r
6.10.4
(6.15)
One-sided Time Truncated Limits

2T
2,2r+2
(6.16)
Example 6.12.
Forty modules were placed on life test for 20 days (24 hours per day). Failed
boards were replaced on the test stands. The test produced two failures.
A. Set an 80% lower condence limit on the MTBF.
B. Set a 50% lower condence limit on the MTBF.
Example 6.13.
interface chips, connectors, etc. and a small amount of memory. Forty of
these modules were placed on life test for 60 days (operating 24 hours a day).
Failed modules were replaced on the test. Three failures were observed.
A. Set an 80% lower condence limit on the MTBF.
B. Set a 60% lower condence limit on the MTBF.
6.11
Reliability Estimation With Condence
We know how to set condence limits on the MTBF. That is,

LCL U CL

Then
et/LCL R(t) et/U CL
88
(6.17)
We can also use one-sided limits. For

LCL
We have
et/LCL R(t)
Example 6.14.
interface electronics, connectors, etc. and a small amount of memory. Forty
of these modules were placed on life test for 60 days (operating 24 hours
per day). Failed modules were replaced on the test. Three failures were
observed.
A. Set a 90% lower condence limit on the MTBF.
B. Estimate the 1,200 hour reliability at 90% condence.
Example 6.15.
Forty microcomputer modules used to control an automotive fuel injection
system were placed on life test. The test was terminated at 2,160 hours.
The failure times are: 878.4 hour, 1,158.1 hour and 1,774.6 hour.
A. With 70% condence, what is the minimum MTBF?
B. If this microcomputer is to be used in an automotive application, what
is the 36,000 miles reliability, with 70% condence?
Example 6.16.
A new ignition module has been developed. In order to estimate the eld
reliability of this module, ve cars were driven around the clock starting
from the Detroit Metro area in the month of March. The cars were driven
cross country from Detroit to New England to California.
The failure history on the cars is:
Car No.
1
2
3
4
5
Odometer Miles at Failure

30,940
48,755
No failures
58,472
85,262; 86,214
Current Odometer Mileage

70,296
78,443
78,134
79,737
86,214
89
Using a suitable level of condence, answer the following questions.

A. Estimate the failure rate for this module.
B. Estimate the 50,000 mile reliability.
C. Estimate the warranty exposure cost. The replacement cost is $92.
Example 6.17.
Consider the data on a machining center where a shift is ten hours.
Shift no.
1
2
3
4
5
6
7
8
9
10
Fault
Hall eects sensor failure
Remove and clean sensor
Tool not seated in changer
No faults
Adjust positioning device
Spindle bearing swap out
Controller malfunction/change module
No faults
No faults
Controller malfunction/change module
Hydraulic cylinder alignment problem/shim cylinder
Production shut down
Malfunction on start up
Repair time
2 hr
4 hr
0.25 hr
0.75 hr
6 hr
2.5 hr
3 hr
4 hr
4 hr
0.75 hr
A. Calculate the MTBF.

B. Calculate the MTTR.
C. Calculate the availability.
E. Estimate the one-shift reliability at 80% condence.
6.11.1
The No Failure Situation
The time truncated condence limit can be used for the no failure situation.
The formula is
2T
(6.18)
2,2
Example 6.18.
A truck-tractor was rebuilt for testing a turbine engine. The truck was run
90
on a test schedule pulling a fully loaded trailer. The test was suspended at
27,600 miles because of a starter failure. No other failures were observed.
The starter failure was not credited as a failure of the core turbine engine.
Thus the turbine lasted for 27,600 miles with no failures attributed to it.
What can we say about the mean life for this engine?
6.12
Test Time to Demonstrate Reliability at a

Condence Level
We can design a qualication test to demonstrate that we have achieved a

certain MTBF or failure rate at a certain level of condence. To design a
no failure qualication test, we use the formula
2,2
T =
M T BFg
2
(6.19)
where M T BFg is the MTBF goal that must be demonstrated with a condence level of C = 1 . Here we must be operating in the constant failure
portion of a products life. The time (T ) must be achieved without failure.
If a failure does occur, the time clock is reset to zero.
Example 6.19.
To demonstrate an MTBF of 1,200 hours at 75% condence, how many
failure-free hours are needed? How many hours are needed at 50% condence?
Example 6.20.
To demonstrate a failure rate of no more than = 3.2 106 /mile, over the
rst 50,000 miles of vehicle usage, how many failure-free miles are needed
at 75% condence? How many vehicles are needed?
6.13
Poisson Failure Calculations
If we have a failure rate of , then in T accumulated hours the number of

failures observed (r) is probabilistically described by the Poisson distribution
(assuming that the failure rate is constant). We have
Pr(r) =
Example 6.21.
(T )r T
, for r = 0, 1, 2, . . .
e
r!
(6.20)
91
We have an electronic system in an automotive application that has a failure

rate of = 2.2 105 /mile. For the initial 50,000 mile warranty period:
A. What is the probability of having two failures of this electronic system
on a particular car?
B. What is the probability of having at least two failures on a particular
car?
C. What is the probability of having no failures?
D. What percentage of cars will have problems with this system in the rst
50,000 miles?
6.13.1
Normal Approximation to Poisson Failure Calculations
With a very large T (total accumulated hours) as would be obtained from

large warranty data bases, the use of the Poisson becomes dicult and it is
more convenient to use the Normal approximation to the Poisson. For the
Normal approximation, let
= T
and
=
The above denes the mean and standard deviation of the Normal distribution. We can let the standard Normal variate be
z=
r (T )
to calculate the probability of r or fewer failures. If we want to nd the

number of failures for a particular probability level, we can calculate
r = (T ) + z T
Example 6.22.
A command control computer has a failure rate of = 3.2 106 /mi. This
computer is an emission sensitive component and must be warranted for
50,000 miles. If 30,000 cars are sold during the rst month of production,
and each car is driven exactly 1,200 miles, with 80% probability, state the
maximum number of warranty claims that will be led.
6.14
92
Combining Failure Rates
If we have m subsystems with failure rates 1 , 2 , . . . , m , then the reliability

function for a subsystem would be given by
Ri (t) = eti
and if we assume a series system, the system reliability would be
Rs (t) = R1 (t) R2 (t) Rm (t)
Rs (t) = et1 et2 etm
Rs (t) = et(1 +2 ++m )
So, the system failure rate is simply calculated as
s = 1 + 2 + + m
(6.21)
and the reliability for the system is calculated from

Rs (t) = ets
If you are using MTBFs for each subsystem (i ), then the system MTBF
(s ) is found by
1
1
1
1
=
+
+ +
s
1 2
m
(6.22)
Example 6.23.
Let us say that we have memory chips that have a failure rate of 1.43
107 /hour. We will use nine of these chips in a control computer. What is
the memory system failure rate?
Example 6.24.
An ignition control system has three main modules. The MTBF for each
module follows:
Module No.
I
II
III
MMBF (miles)
79,000
74,000
82,000
93
The three modules are changed as a group when a failure is suspected.

Will there be a problem in providing a 50,000 mile warranty on this system?
To have no more than one failure per 1,000 vehicles in 50,000 miles of customer usage, what allowable module failure rate is needed? Use an equal
apportionment of the system failure rate.
Example 6.25.
A dryer has ve major subsystems. We want the dryer to have a 90%
reliability at 5,000 cycles of usage. What allowable failure rate is needed
for each of the ve subsystems? Use an equal apportionment of the system
failure rate.
Chapter 7
Success/Failure Testing
In success/failure testing the outcome is viewed as either success or failure.
Many testing schemes can be categorized this way. For example, we may
place electronic modules in a hot oven for 24 hours and then test to see if
any have failed. We might cycle 16 gas-lled cylinders in a xture for 10,000
cycles and then evaluate each cylinder at the end of the test to see if it is
still within specications. If any of the cyliners were out of specication,
this would constitute test failure.
Typically automotive engines are subjected to a dynamometer test of
xed length T in hours and this is equated to a useage mileage. The engines
are then tore down and inspected for any visible signs which might constitute
failure of the test.
7.1
Binomial Probability Model for Reliability
The binomial probability model applies to this situation where the outcome
is viewed as either success or failure. The binomial probability model as
applied to this reliability situation is

Pr(y) =
n
y
Ry Qny , y = 0, 1, 2, . . . , n
where
y = number of successes
and
R = reliability of the device
94
(7.1)
CHAPTER 7. SUCCESS/FAILURE TESTING
95
Succeed
Standardized
Test
Fail
Figure 7.1: Success-Failure Product Test

Q = unreliability of the device
n = number of devices
Of course,
R+Q=1
Also note that
n
y
n!
y! (n y)!
and
n! = n (n 1) (n 2) 1
Example 7.1.
The reliability of a four-way pneumatic solenoid operated valve in relation
to surviving a 10,000 cycle test in a hostile environment (heat, vibration,
moisture in air) is 0.92.
A. If ve valves are placed on test, what is the probability that exactly four
survive?
B. What is the probability that at least four survive?
C. What is the probability that none of the ve valves will fail?
96
Success
R
R = probability of a success
Device
Q = probability of a failure
Q
Failure
(n-y) = number of failures
Figure 7.2: Binomial Model for Success-Failure Testing
Example 7.2.
The probability of a Black Hawk helicopter surviving a mission is 0.87. If 8
helicopters are sent on a mission and 6 are needed for mission success, what
is the probability of mission success?
Example 7.3.
A light armored ghting vehicle has an MMBF of 800 miles. A mission is
100 miles in length and requires three vehicles to complete the mission for
mission success. How many vehicles should be sent out on the mission for a
mission reliability of at least 0.92?
7.2
Success-Failure Testing
Success/failure testing describes a situation where a product is subjected

to a test for a specied length of time (cycles, stress reversals, etc.). The
product either survives or fails the test.
Testing of this type can frequently be found in engineering laboratories
where a test bogy has been established and new designs are tested against
this bogy. The bogy will specify a number of cycles in a certain environment
at predetermined stress levels. The bogy might be established by attempting
to recreate a eld failure condition.
In success/failure testing, we want to determine the reliability demonstrated by the product. The point estimate of the reliability given y successes
N items
placed on
test
97
Standardized
Test
r = number of failures
Figure 7.3: Success-Failure Testing Scheme

out of n placed on test would simply be
= y
R
n
(7.2)
which is quoted as a single value with no concept of condence. We usually

want to quote this reliability at a certain condence level. The condence
limit is calculated from
RL =
y
y + (r + 1)F,2(r+1),2y
(7.3)
where
= level of signicance
This provides an exact one-sided lower condence limit on the reliability.
The F values are obtained from the appropriate tables. The appropriate
table is dened by
F,1 ,2
where
= determines the table to use
1 = determines the column nd this across the top of the table
2 = determines the row nd this down the left side margin
98
Example 7.4.
Twenty spot welded support brackets were subjected to a pull test of 100
pounds. Three brackets failed the test. Find the 90% lower condence limit
on the reliability of this bracket in relationship to this particular test.
Example 7.5.
Gas turbine engines are subjected to a ve hour burn in test after assembly.
Out of 20 engines produced in a week, one engine failed to pass the test. Set
a 90% lower condence limit on engine reliability relative to this test.
Example 7.6.
A range nder system, on command, is supposed to report a targets distance
to a digital message device (within a certain tolerance). Out of 30 sightings,
the range was inaccurately reported twice. Give a 90% lower condence
limit for range nder accuracy.
Example 7.7.
Cycle timers are subjected to testing in a hostile environment of heat, vibration, caustic soap and ammonia fumes. Each timer is continuously cycled for
5,000 cycles. Sixty timers were placed on test. Three timers malfunctioned
during testing. Set a 90% lower condence limit on cycle timer reliability.
7.3
Approximate Condence Limits
The F tables that are usually available are somewhat limited. Therefore,
it is convenient to have an approximation for the lower condence limit that
uses the standard normal tables. The lower condence limit on reliability
can be approximated by
RL =
y1

n + z
(7.4)
n(r+1)
(y2)
where
z = standard normal variate
n = sample size
The standardized normal (Z) values are reproduced here for easy usage.

Condence Coecient - C
95%
90%
80%
70%
60%
99
Z
1.645
1.281
0.841
0.525
0.250
The 50% condence limit can be quickly approximated as

RL =
y1
n
(7.5)
Example 7.8.
Eight hundred vehicles were monitored for 50,000 miles for emission system
failure to comply. The emission system on seven vehicles failed during the
50,000 mile test period.
A. With 90% condence, what is the lower limit on reliability for this
emissions system?
B. At 50% condence, what is the emissions system reliability?
Example 7.9.
Twenty laundromats were monitored for product failure problems for one
year. All failed parts were returned for inspection and analysis. In the
twenty laundromats there were 380 washing machines. Seven ow control
valves failed during this period of time. Set an 80% condence limit on
ow control valve reliability relative to this environment using the given
information.
7.4
Success Testing
A product qualication test is frequently specied as a no failure test. The

concern is usually on insuring that a minimum reliability has been demonstrated at a specied condence level. A no failure test is the quickest way
to do this.
For the special case where r = 0, (i.e. no failures), the lower 100(1 )%
condence limit on the reliability is
RL = 1/n
(7.6)
100
where
= level of signicance
and
Condence = C = 1
We can also specify a sample size necessary to demonstrate a reliability goal
n=
ln(1 C)
ln Rg
(7.7)
where
Rg = the reliability goal
Example 7.10.
We wish to demonstrate a minimum reliability of 0.95 with 90% condence.
Assuming no failures, how many do we need to place on test?
Example 7.11.
A cycle timer is supposed to have a 5,000 cycle reliability of 90%. We want
to design a qualication test for our suppliers.
A. How many should be placed on test?
B. How long should each be tested?
C. How can the sample size be reduced?
7.4.1
Conversion to the Exponential Distribution
In cases where it is reasonable to use an exponential time to failure distribution, it is possible to convert from success/failure testing data to the
exponential. Here we are assuming that the success/failure test is an accelerated test and that the equivalent length of usage under normal customer
conditions is designated as T . For example, one might specify 100,000 cycles
in a fatigue test for truck steer axles and this is judged to be equivalent to
T = 750, 000 miles of customer usage.
The point estimate of reliability is
= y
R
n
From the exponential, the reliability at time T is given by
R(T ) = eT /
101
So, as estimate of the MTBF can be obtained by equating

y
= eT /
n
Or,
T

(7.8)
n
y
ln
Condence limits can also be placed on the MTBF using the successfailure condence limit. Let
RL = the lower condence limit
Then
RL = eT /L
Or,
L =
ln
1
RL
(7.9)
will give the lower condence limit on the MTBF.

In the case of success testing, the lower condence limit on the MTBF
is
nT
(7.10)
ln(1/)
For any time t, the 100(1 )% condence limit on reliability for the no
failure situation is
t ln(1/)
(7.11)
e nT R(t)
Example 7.12.
Twenty armored personnel carriers were subjected to a cavalry mission prole. Two failed to complete the mission.
A. Find the mission reliability with 75% condence.
B. If the mission length is 50 miles, nd the 75% lower condence limit on
the MMBF.
Example 7.13.
Three engines were placed on an 800 hour durability test. All of the engines
survived the test. The test is said to be equivalent to 150,000 miles of
customer usage.
102
A. With 80% condence, what is the 150,000 mile reliability?

B. How many engines would we have to test with no failures to have an
80% reliability at 80% condence?
C. How many engines would be needed for 80% reliability at 50% condence?
Chapter 8
Weibull Distribution
The Weibull distribution is a very widely used model for time-to-failure
for mechanical components. This distribution was originally developed to
model fatigue failures.
8.1
Weibull Probability Density Function
The Weibull distribution is much more versatile than the exponential distribution and can model many dierent failure phenomena. Some general
shapes of the Weibull pdf are shown in the Figure 8.1.
8.1.1
Two-parameter Weibull
For the two-parameter Weibull, the pdf is given by

f (t) =
1 (t/)
t
e
, t0
(8.1)
where
t = time-to-failure
and
= Weibull slope, > 0
= characteristic life, > 0
R(t) = e(t/) , t 0
103
(8.2)
CHAPTER 8. WEIBULL DISTRIBUTION
104
60
50
f(t)
40
30
20
10
800
600
400
200
Time
Figure 8.1: Weibull Probability Density Functions

Some plots of this reliability function are shown in Figure 8.2.
The hazard function (theoretical failure rate) is
h(t) =
1
t
, t0
(8.3)
Various plots of this hazard function are shown in Figure 8.3.

For a given reliability R, the time is
1
t = [ln(1/R)]
and the B10 life is
(8.4)
1
B10 = [ln(1/0.90)]
(8.5)
This is the time at which 10% of the population will fail. The B10 life was
a terminology initially used to describe bearing reliability.
The value of (the Weibull slope) can give us an idea of the mechanism of
failure. The dierent values of the slope are related to the failure mechanism
in Table 8.1. For a of about 3.4-3.5, the Weibull distribution is reasonably
symmetrical and looks like the normal distribution.
105
50%
50%
25%
25%
0%
0%
200
75%
150
75%
100
100%
50
100%
R(t)
Time-hours (t)
Figure 8.2: Weibull Reliability Functions
Table 8.1:
Weibull Slope ()
<1
=1
>1
Weibull Slope and Failure Mechanism

Failure Mechanism
Decreasing failure rate; infant mortality
Constant failure rate; exponential distribution
Increasing failure rate; wear out, aging
106
0.25
Failure rate - h(t)
0.20
0.15
0.10
0.05
250
200
150
100
50
0.00
Time
Figure 8.3: Weibull Hazard Functions
Example 8.1.
A two-parameter Weibull is used to model time-to-failure for a module that
is believed to experience an increasing failure rate because of wear out caused
by thermal cycling. The parameters are = 4.22 and = 78, 000 miles.
A. Find the 12,000 mile reliability.
B. Find the 50,000 mile reliability.
C. Find the B10 life.
Example 8.2.
A two-speed synchronized transfer case used in a large industrial dump truck
experiences failures that seem to be well approximated by a two-parameter
Weibull distribution with = 98, 000 miles and = 1.7. For this situation,
answer the following questions.
A. What is the 20,000 mile reliability?
B. What is the 80,000 mile reliability?
107
C. If the Weibull slope can be changed to = 4.7 through tighter tolerancing, how will this aect the above answers?
D. For both slopes, nd the 98,000 mile reliability.
Example 8.3.
A gear box has a time-to-failure modeled by a Weibull distribution with
= 8, 000 cycles and = 3.6. This gear box experiences a usage of 500
cycles per year. What is the ten year reliability?
Example 8.4.
Given a module with a Weibull slope of 4.24 and a characteristic life of
89,000 cycles,
A. Find the B10 life.
B. Find the 12,000 cycle reliability.
8.1.2
Characteristic Life
Given that
R(t) = e(t/)
then for
t=
we have
R() = e(/) = e1 = 0.368
which provides some motivation for the term characteristic life. As can
be seen in Figure 8.4, the characteristic life always divides the population
into two areas of 0.632 before and 0.368 after.
8.2
8.2.1
Weibull Mean and Standard Deviation

Weibull Mean
The Weibull mean is given by

= (1 +
1
)
(8.6)
where (n) is a Gamma function with argument n. For integer n, we have

(n) = (n 1)!
108
pdf
63%
37%
200
150
100
50
Time
Figure 8.4: Weibull Characteristic Life and Population Split

Consider the situation where
= 0.50
and
= 1, 200 cycles
Then
=
In order to evaluate the mean for various values of , we have provided
a table of A1 factors where
A1 = (1 +
1
)
(8.7)
Using the tabled factors, the mean can be calculated by

= A1
Example 8.5.
For a bearing with a Weibull slope of = 3.12 and a characteristic life of
= 3 106 cycles, nd the mean cycles to failure.
8.2.2
109
Weibull Standard Deviation
The standard deviation for the Weibull is given by

(1 +
2
1
) 2 (1 + )
(8.8)
So, for
= 0.50
and
= 1, 200 cycles
Then
=
We can nd the standard deviation using the tabled A2 values where

A2 =
(1 +
2
1
) 2 (1 + )
(8.9)
So, the standard deviation is found by

= A2
Example 8.6.
For a bearing with a Weibull slope of = 3.12 and a characteristic life of
= 3 106 cycles, nd the standard deviation of the cycles to failure.
Example 8.7.
Given a Weibull distribution with a characteristic life of 127,000 hours and a
slope of 2.14, nd the mean and standard deviation. Also nd the probability
of surviving the mean life.
Example 8.8.
Given a module with a Weibull slope of 4.24 and a characteristic life of
89,000 miles,
A. Find the mean and standard deviation.
B. Find the B10 life.
C. Find the 12,000 mile reliability.
8.3
110
Three-parameter Weibull
The three-parameter Weibull has a non-zero minimum life. This means

that the population of products goes for an initial period of time without
failure. Many real-life products t into this category. The pdf for the threeparameter Weibull is given by
f (t) =
(t )1 e( ) , t
( )
(8.10)
where
t = time-to-failure (t )
= minimum life parameter ( 0)
= Weibull slope ( > 0)
= characteristic life ( )
t
R(t) = e( ) , t
(8.11)
For a given reliability

1
t = + ( ) [ln(1/R)]
(8.12)
and the B10 life is

1
B10 = + ( ) [ln(1/0.90)]
(8.13)
Example 8.9.
A timing gear belt in an automotive application has the following parameters:
= 52, 000 miles
= 98, 000 miles
= 3.72
where t represents miles-to-failure.
A. Find the 70,000 mile reliability.
B. If the Weibull slope is changed to 5.12, what will the 70,000 mile reliability be?
C. What is the 100,000 mile reliability using = 3.72?
D. What is the B10 life using = 3.72?
8.3.1
111
Computerized Three-parameter Weibull Distributions
The values for the parameters of this Weibull distribution are frequently
estimated using a computer program. In developing this program many
people use a modied version of our three-parameter Weibull. The modied
reliability function is
t
R(t) = e( ) , t
You can see that
=
and would not give 37% for the reliability function when substituted in for
t. However, programming the estimation process is slightly easier. Beware
that the documentation may refer to as the characteristic life.
8.4
Graphical Estimation of Weibull Parameters
With the Weibull distribution it is very easy to provide a means for graphical
analysis of failure data. Graphical analysis is useful for visualizing failure
phenomena, particularly with small samples.
Weibull probability paper is a special graph paper used to analyze failure data where the failure distribution is assumed to be the Weibull. The
procedure to construct this Weibull paper is outlined below.
The reliability function is given by
R(t) = e(t/)
Substituting the cumulative distribution function for the reliability function

gives
1 F (t) = e(t/)
reciprocating to remove the negative sign on the exponent
1
= e(t/)
1 F (t)
Then taking logarithms

1
ln
=
1 F (t)

t
and taking a second log

ln ln
1
= ln(t) ln()
1 F (t)
112
This is now of the linear form

Y = X A
This allows us to make Weibull paper by properly transforming the axes. A
typical piece of Weibull paper is shown in Figure 8.5.
Let us apply graphical analysis to the following data:
Order Number (j)
1
2
3
4
5
6
7
Time to failure (t) hours

91
122
195
220
261
315
397
Plotting Position (pj )
The order number is the ordering of a particular failure time with respect
to all of the failure times. The failure times are plotted on the abscissa
(X-axis), while the plotting position is the value on the ordinate (Y-axis).
We must select appropriate plotting positions.
8.4.1
Plotting Positions
There are two possibilities for assigning plotting positions to the failure
times. These are termed mean rank and median rank plotting positions.
The median rank plotting is most commonly used.
Median Rank
The median rank plotting position can be calculated by
pj =
j 0.3
n + 0.4
where
n = sample size
j = order number for the observation
Rank tables also appear in the appendix.
(8.14)
Figure 8.5: Weibull Paper for Plotting Failure Data
113
114
Mean Rank
The mean rank plotting position can be calculated by
pj =
j
n+1
where n, j are the sample size and order number, as used previously. The
mean rank plotting position is useful for a quick and dirty estimate of the
B10 life. If we place n = 9 devices on test and wait until the rst failure t1 ,
then
1
p1 =
= 0.10
9+1
so t1 is an estimate of the B10 life.
8.4.2
Summary of Weibull Plotting Procedure
1. Arrange the data points from the lowest to the highest value and number the data points sequentially. The sequential numbers are called
order numbers.
2. From the Table of Median Ranks, assign the appropriate median rank
value to each ordered data value. The median rank value is read under
the sample size.
3. On a sheet of Weibull Probability Paper establish a scale on the axis
that will include all the failure time values. The scale is logarithmic
and is laid out in major sections called decades. As the scale is read
from left to right the decimal point moves one place to the right at the
beginning of each new decade.
4. Plot each failure time using the recorded failure value and associated
median rank table as obtained from step 2.
5. By eye, draw the best tting straight line through the plotted points.
Place more emphasis on points near the center of the line than on
points towards the ends. This is the population line.
You can estimate the Weibull slope directly from the slope of the straight
line. The characteristic life can be estimated at the 63% point.
8.4.3
115
Condence Limits for Graphical Analysis
Let us imagine that we have two Weibull plots for the same test situation.
One Weibull plot is based on a sample size of ten, while the second plot is
based on a sample of 100. In which plot would we have more condence for
estimation purposes?
Obviously we have more condence in predictions made with larger samples. This leads us to the subject of condence limits. Condence can be
shown by placing condence limits about the Weibull plot.
The 5% and 95% rank values can be found in the tables.
Order No. (j)
1
2
3
4
5
6
7
5% Rank
95% Rank
Construction of Condence Limits

1. Copy the 5% and 95% rank values from the Rank Tables.
2. Horizontally intersect the Weibull population line (plotted line) at each
data point.
3. Using the 5% rank values, plot these vertically below each intersect on
the population line.
4. Connect these points with a smooth curve.
5. Using the 95% rank values, plot these vertically above each intersect
on the population line.
6. Connect these points with a smooth curve.
You now have constructed a 90% condence interval.
Example 8.10.
Consider the following data representing cycles to failure for turn signal
contacts.
116
Cycles to failure for 14 contacts

8,570
5,177
4,188
11,023
7,782
5,830
10,400
17,573
8,158
11,321
6,266
7,295
1,680
8,471
Analyze the above data on Weibull paper.
Example 8.11.
Four axles were fatigue tested to failure in a laboratory set-up. All readings
must be multiplied by 106 .
j
1
2
3
4
Cycles to failure
2.00
2.39
2.64
3.01
Plot on Weibull paper. Would you guarantee these axles to be failure free
for one million cycles?
8.5
Non-zero Minimum Life Weibull Distributions
So far, we have plotted data on Weibull paper that we assumed had a minimum life of zero. Let us now see what happens with the following data
when plotted on Weibull paper.

j
1
2
3
4
5
6
7
8
9
10
11
12
13
14
pj
4.8%
12
19
26
33
40
47
53
60
67
74
81
88
95
Cycles to failure (t)

87,000
110,000
132,000
169,000
219,000
270,000
335,000
389,000
480,000
510,000
675,000
850,000
1,090,000
1,470,000
117
Adjusted cycles to failure (t )
For the above data, we would estimate the minimum life to be about
=
Let us plot the data and estimate the B10 life.
8.6
Suspended Item Analysis
Suspended failure times can result from

Test stand failure
Secondary failure
Accidents
Field data
A suspended item is an item that was taken o test for reasons other than
failure.
Consider the situation
No.
1
2
3
4
Time (hour)
84
91
122
274
Occurrence
Failure (F1)
Suspension (S1)
F2
F3
118
The possible sequences of S,F that could have occurred are

Possible Outcomes
I
II
III
F1
F1
F1
SF
F2
F2
F2
SF
F3
F3
F3
SF
The approach is to assign an average rank to each failure time. We
calculate the average ranks as
Average position (j)
Failure time (t)
Then the rank for each failure is calculated using the formula
pj =
j 0.3
n + 0.4
We could also interpolate in the median rank tables to obtain the same
values.
Then the failure data for plotting are summarized as
Failure time (t)
84 hour
122
274
8.6.1
Average rank (pj )
New Increment Method
The previous reasoning is easy with only one suspension but becomes unwieldy with several suspensions. The new increment method was developed
by Leonard Johnson as an easy procedure to nd the average order number
for each failure time. This method provides the average rank values in an
easy and systematic fashion. The new increment is
I=
(n + 1) Previous order number

1 + Number of items following suspended set
Let us apply this to the following set of data:
(8.15)

Hours to occurrence
544
663
802
827
897
914
939
1,084
1,099
1,202
Results
F
F
S
S
F
F
S
F
F
S
119
Calculation of average position
The median rank plotting position is found from the formula

pj =
j 0.3
n + 0.4
For the failure times and average positions, we calculate the average rank
Average position (j)
Average rank (pj )
Failure time (t)

544
663
897
914
1,084
1,099
Let us now try an example.
Example 8.12.
Seven experimental automotive transmissions were tested for mileage to rst
repair. Normal maintenance was done during testing as prescribed by the
service manual. During testing one test vehicle was in an accident. A second
failure was attributed to an assembly error that would not occur during
normal production. A third failure was attributed to a quality problem
due to a vendor supplying cracked valve seats. This problem would also be
eliminated during production.
The data follows:

Incident No.
1
2
3
4
5
6
7
Mileage (t)
48,721
79,291
99,430
112,982
114,271
116,782
198,218
120
Comment
Assembly error
Quality problem
Accident
Estimate the B10 life.
Example 8.13.
Ten prototype washing machine gear boxes are on life test. The current
status of the test is as follows:
No.
1
2
3
4
5
6
7
8
9
10
Cycles
1,114
2,280
3,347
3,819
4,271
5,487
6,244
6,934
7,234
7,891
Status
F
test stand broke, crushed housing
F
F
porous casting, taken o of test
F
still on test
still on test
still on test
still on test
Estimate the one year and ten year reliability.
8.7
Sudden Death Testing
The sudden death approach to life testing was proposed by Leonard Johnson
in the 1950s when he was working as a research scientist at the General
Motors Technical Center. This procedure helps to shorten the test duration.
Testing occurs in batches. The batch size might be determined by test
facility capacity. A batch is placed on test and the testing continues until
the rst failure occurs. When the rst failure occurs, the batch is suspended
and a second batch is placed on test. This procedure is followed successively.
Let us say you have 20 test pieces. Randomly divide the test pieces into
four batches of ve pieces each. Test each batch until the rst failure occurs
for the batch. Results are as follows:
121
Table 8.2: Data for First Failure in Five

t
j pj
1 16 17 hour
2 39
56
3 61
124
4 84
294
Table 8.3: Data for First Failure in a Batch of Ten
j
pj
t
1 13% 72 cycles
2
31
241
3
50
505
4
69
910
5
87
1,920
Batch no.
1
2
3
4
Time to failure
56 hour
124
294
17
Let us plot these four failure times on Weibull paper using the rank
values for n = 4. These rank values are found in the median rank tables
and are included in the following table along with the ordered failure times
for each batch. We plot these data to obtain a rst in ve plot. The plot is
shown in Figure 8.6.
The plot represents the median rank estimate of the rst in ve failure
times. The rank for the rst in ve failures is 12.9% as found from the
median rank values for n = 5 (the batch size). The rst in ve plot is
the median rank (50-tile) estimate of the distribution of the 12.9 percentile.
If we go across at 50% (the median) on the Weibull paper to the plotted
line and drop down to 12.9%, we have located the population line. This
population line is parallel to our plotted line. So, we can easily draw in the
population line.
Let us try this with a second situation. In this situation we tested 50
parts in groups of 10. Larger groups should produce the rst failure quicker.
The values in Table 10-3 represent the rst failure in a batch of 10 parts
placed on test. Here we had ve such batches. We looked up the median
rank values (pj ) in the table under n = 5.
Figure 8.6: Sudden Death Plot for First in Five
122
123
For a sample of size n = 10, which is the batch size, we nd that the
rst failure represents 6.7%. So, we go across at 50% (the median) and drop
down to the 6.7% value to locate the population line. The plot is given in
Figure 8.7.
Example 8.14.
Batches of 50 couplers are placed on life test in a severe application where a
drive motor is turned clockwise for 40 seconds and then reversed. The motor
is driving a ywheel to create inertia on the coupler. Only four failures are
available from four batches of couplers. Estimate the Weibull population
parameters and B10 life.
Order no.
1
2
3
4
Cycles to failure (106 )

4.597
5.184
6.463
7.615
pj
Figure 8.7: Sudden Death Plot for First in Ten
124
Chapter 9
Maintainability Engineering
High reliability ensures that the equipment will be failure free for a long
period of time, but if the equipment should fail, it must be designed such
that recovery from failure can be rapidly accomplished.
Products must be designed for quick diagnosis and fast repair. Designs
must address ease of maintenance in order to improve maintainability. Maintainability can only be achieved in design by anticipation of failure situations
and planning for the required maintenance actions for quick and easy recovery.
Maintainability is a characteristic of design and installation which imparts to a system an inherent ability to be maintained, so as to lower the
required maintenance man hours, skill levels, tools, facilities, and logistic
costs, to achieve higher system availability.
Various measures have been used to quantify maintainability such as
maintenance man hours per unit operating time (e.g., miles, rounds, ying
hours), maximum time to repair, minimum time between overhauls.
Maintainability is concerned with those actions taken by a design engineer, during development, to incorporate design features which will ensure
ease of maintenance. Maintenance is concerned with those actions taken by
a user to retain an existing system, or restore it to, an operable condition.
Maintainability is dened as the probability that a failed system is restored to operable condition in a specied down time.
Maintainability is rst of all a characteristic of design. Design determines
the location of components, the type of components used and the accessibility. In some cases, installation also plays an important part as in the case
of manufacturing machinery and equipment where much of the wiring and
piping runs are not called out in the drawings but are left to the skills of
125
CHAPTER 9. MAINTAINABILITY ENGINEERING
126
the installation personnel.
9.1
Maintainability Improvement Strategies
We will now list some of the strategies which might be useful during design
and development to improve maintainability.
Accessibility. Accessibility means having sucient working space around
a component to diagnose, troubleshoot, and complete maintenance activities
safely and eectively. Provision must be made for movement of necessary
tools and equipment with consideration for human ergonomic limitations.
Operators, maintenance and service personnel should be involved in evaluating the design for accessibility. These people have the best knowledge as
to how the repair job will be done and identify the accessibility problems.
Captive and Quick Attach/Detach Hardware. Captive and quick
attach/detach hardware provides for rapid and easy replacement of components, panels, brackets and chassis. The environment in which these devices
are used may restrict the type of device used. Spare parts and replaceable
subassemblies should also be congured with these devices pre-assembled.
Examples include
plate, anchor and caged nuts;
push and snap-in fasteners;
clinch and self-clinching nuts; and
quarter-turn fasteners.
Common Tools. Specialty tooling should be avoided. Standard tools
readily available to maintenance should be used. Standard tools are readily
available, eliminate further investment, and decrease the required inventory
of tools.
Color Coding. Color coding can help to speed up maintenance procedures. Applications include:
Orientation
Timing marks
127
Groupings
Processing stage
Source/destination
Size
Metric/standard English
Location
Date of inspection
Material type
Reference points
Torque requirements
Lubrication information
In preventive maintenance, color coding can make it easier for maintenance
personnel to nd servicing points.
Diagnostics. Diagnostic devices indicating the status of equipment should
be built in to aid maintainability support processes. The diagnostics can be
as simple as a visual display indicating the equipments status as a go/nogo condition, or as sophisticated as a knowledge-based expert system with
the capability of analyzing a problem and recommending the most likely
solution.
Diagnostic systems should indicate the specic component to replace or
repair.
Lubrication Points. All lubrication points should be brought out to a
readily accessible area and grease ttings should be clustered together. The
component being lubricated should be identied at the grease tting.
Modularity. Modularity requires that designs be divided into physically
and functionally distinct units to facilitate removal and replacement. It
allows design of components as removable and replaceable units for an enhanced design with minimum downtime. Modular design concepts typically
are thought of in terms of electrical black boxes, printed circuit boards and
128
other quick attach/detach electrical components. These concepts are also

applicable to the mechanical elements of equipment.
Modularity oers several advantages:
New designs can be simplied and design time can be shortened by
making use of standard, previously developed building blocks.
Specialized technical skill will be reduced.
Training of plant maintenance personnel is easier.
Engineering changes can be made quickly with fewer side eects.
Standardization. Designing equipment that incorporate component parts
that are commercial standard, readily available, and common from system
to system contributes to enhanced maintainability and to greatly reduced
investment and carrying costs associated with spare part inventories.
The use of exotic parts should be avoided. Here exotic parts are dened as those parts not readily available through the normal logistics channels.
All parts used should be traceable to the OEM, and the practice of some
system suppliers to place their unique numbers on purchased components
should be discouraged.
Example specication: The company must be able to obtain locally, o
the shelf, any component used in the system.
Training, Documentation and Job Aids. The most eective way to
reduce repair and diagnostic time is to ensure that appropriate personnel are
well trained and have clear and concise information about the operation of
the system. Prospective drawings and renderings created with the use of a
CAD system help to facilitate the documentation process. Product engineers
should be encouraged to use modern CAD facilities for development of their
documentation package.
The volume of maintenance manuals should be kept to a minimum, and
their quality should be high. Maintenance manuals are referred to during
a time of breakdown when quick answers are needed. Charts, pictures and
diagrams should be used for rapid assimilation of the maintenance information. With todays technology computerized maintenance manuals are
becoming standard.
9.2
129
Determination of the MTTR
The mean time-to-repair is used as an indication of the ease of maintainability for machinery and equipment.
Mean Time-To-Repair (MTTR). The average time to restore machinery or equipment to specied conditions.
The MTTR should be estimated during the design process. The approach for determining a MTTR might be as follows:
Based on your maintenance experience with similar equipment, identify how the equipment might fail (failure modes).
Estimate the frequency of occurrence for each failure mode.
Estimate the time to repair considering the way the equipment is designed.
Calculate the maintenance load for each mode of failure.
Initiate design action for failure modes that create a high load on the
maintenance function.
A maintenance analysis can be done during equipment design and development. The analysis is usually based on a Failure Mode Analysis. Once
the failure modes are identied, the analysis might proceed as follows:
Failure mode
Jammed slide
Bent ball screw
Pump seal failure
Failure rate (i )/hr

0.018
0.004
0.002
Time to repair (yi ) hr

0.50
6
24
The maintenance load is calculated as

L=
all
i yi
(9.1)
The mean time to repair calculation for the design is

MTTR =
i yi
all
i
all i i
In our previous example,

MTTR =
0.018 0.50 + 0.004 6 + 0.002 24

= 3.375 hr
0.018 + 0.004 + 0.002
Consider the following situation:
(9.2)
Subsystem
1
2
3
4
MTBF (hr)
3,000
2,000
1,500
2,500
Total
Failure rate
0.00033
0.00050
0.00067
0.00040
0.00190
Time to repair (hr)

16.0
10.0
2.5
4.0
130
Maint. load
0.0053
0.0050
0.0017
0.0016
0.0136
The MTTR would be

MTTR =
Example 9.1Calculation of MTTR.

Consider the following machining operation and calculate the MTTR for
this system.
Subsystem
Loader
Station 10
Transfer
Station 20
Palletizer
9.2.1
MTBF (hr)
7,000
6,000
4,000
3,400
8,200
Failure rate
Time to repair (hr)

0.25
3
1
0.75
4.25
Maint. load
Mean Time to Repair from Operational Data
The mean time to repair can be calculated from operational data and is the
average repair time. Repair time consists of ve basic elements:
Administrative time: The period between failure and onset of diagnosis. It includes time to become aware of the failure and preparation of
service, orders, assignment of people, etc.
Diagnostic time: The portion of repair time between administrative
and active repair time used to determine cause of failure and correction
method.
Logistic time: The time to acquire replacement parts.
Active repair time: Time required to physically perform a repair. Includes getting tools, gaining access, repairing. Assumes replacement
part availability.
131
Availability
Reliability
Maintainability
Figure 9.1: Reliability, Availability and Maintainability

Start-up time: The time required from completion of active repair
to begin producing satisfactory parts at design cycle time. Includes
warm-up and adjustments.
The MTTR is calculated by
MTTR =
Unscheduled downtime
r
(9.3)
where
r = Total number of equipment failures
and
MTBF =
9.2.2
Uptime
r
(9.4)
Availability
Availability (A) depends on R&M. It is also impacted by the planned maintenance schedule for the equipment. One of the goals of planned maintenance
is maximum availability of the equipment.
Availability is the probability that at any time, the system is either
operating satisfactorily or is ready to be operated on demand, when used
under stated conditions.
The availability might also be looked at as the ability of an equipment
under combined aspects of its reliability, maintainability and maintenance
supportto perform its required function at a stated instant of time. This
availability includes the built-in equipment features as well as the in-plant
maintenance support function. Availability combines reliability and maintainability into one measure.
9.2.3
132
Availability Calculation
During design/development, the availability is calculated from the design

data using the formula
A=
MTBF
MTBF + MTTR
(9.5)
For example, if
MTBF = 40 hours
MTTR = 2 hours
then
40
= 0.952
40 + 2
For installed equipment, availability is calculated using actual running
time and downtime as follows
A=
A=
Uptime
Uptime + Unscheduled downtime
(9.6)
It should be pointed out that the two formulas are equivalent.

Design Eorts to Improve Availability. The availability calculation
during design using equation 9.5 usually comes out high, but factors in the
eld tend to not produce a high number. Rather than concentrating on
availability, the design eort should be aimed at driving the MTBF up and
the MTTR number down.
Example 9.2Calculation of Availability.

Reconsider the previous machining system and calculate the availability.
Subsystem
Loader
Station 10
Transfer
Station 20
Palletizer
MTBF (hr)
7,000
6,000
4,000
3,400
8,200
Failure rate
Time to repair (hr)

0.25
3
1
0.75
4.25
Maint. load
133
lD t
Up
Down
mD t
Figure 9.2: Transitions in the System Status
9.2.4
Basis for Availability Formula
The availability formula is derived from the simplied situation shown in

Figure 9.2 where
= the failure rate
= the repair rate
If we dene
Pd (t) = the probability that the system is down at time t
Pu (t) = the probability that the system is up at time t
then
Pu (t + t) = Pu (t)[1 t] + Pd (t)t
Pd (t + t) = Pd (t)[1 t] + Pu t
Rearranging we have
Pu (t + t) Pu (t)
= Pu (t) + Pd (t)
t
Pd (t + t) Pd (t)
= Pd (t) + Pu (t)
t
Taking the limit as t 0 and at steady state, we have
0 = Pu + Pd
0 = Pd + Pu
Recognizing that
P u + Pd = 1
134
and that the availability is A = Pu gives

A=
which is
A=
9.3
MTBF
MTBF + MTTR
Maintainability Specications
The MTTR is not a good maintainability specication because the time-torepair distribution is skewed, and average values are not representative of the
true situation. Also, changing the MTTR in the availability formula appears
to have little eect, because the MTBF is much bigger than the MTTR. In
order to improve system reliability and maintainability performance one
must strive to drive the MTBF up and the MTTR down. One way to do
this is to specify limits on the time-to-repair for certain kinds of failures. For
example, the maximum allowable time to repair for common failure modes
and major overhauls might be specied.
Consider the following maintenance specication on a system:
Any failed component in the xyz system must be replaceable within
30 minutes of active repair time, by design, using the normal tools and
trades available in a ZZZ plant.
Any malfunction or adjustment must be accomplishable within 10 minutes of active repair time, by design, using the normal tools and trades
commonly available in a ZZZ plant.
The maintainability requirement must be addressed during concept/proposal
and shown through suitable analysis during design/development.
Another way is to specify the 90th percentile of the time-to-repair distribution. The lognormal distribution is used to model time to repair. The
lognormal probability density function is
f (y) =
where y = time to repair.
y 2
1 ln y 2
e 2 [
] , y0
(9.7)
9.3.1
135
Parameters
The parameters are and . These parameters are not the mean and
standard deviation but dene the mean and variance as follows:
Mean = e+
and
Variance = e2+
9.3.2
2
2
(9.8)

e 1
(9.9)
Calculation of Probabilities for the Lognormal
The lognormal can be translated easily to the normal distribution allowing

standard normal tables to be used for nding probabilities. The translation
to the standard normal variate z is
ln y
Of course this formula can be rearranged to give

z=
y = e+z
(9.10)
(9.11)
Example 9.3Use of Lognormal Distribution.

Consider a case where the parameters of the lognormal distribution are
given by = 2.5 and = 1.4. Find the 90th percentile of the time-to-repair
distribution. What is the average time-to-repair?
Example 9.4Calculation of Upper Percentile.

For the following parameters of the lognormal distribution, calculate the
MTTR and the 90th percentile:
1.0
2.0
9.4
2.2
1.7
Preventive Maintenance
So far we have been talking about unscheduled maintenance. Unscheduled maintenance results from equipment failure and produces unscheduled
downtime on the equipment.
The required scheduled maintenance of a machine must also be considered early in the design and development process. The following factors
should be addressed:
136
Frequency of required preventive maintenance (PM) actions (the less

required, the better)
Ease of performing PM (can personnel get to the required PM points
on the machine with ease)
Skill level required (reduce need for specialized skill)
Necessary training of personnel to perform PMs (the PMs should require a minimal of training, if any)
Any special tools or equipment required (avoid the need for special
tools or equipment)
9.5
Overall Equipment Eectiveness
The term Overall Equipment Eectiveness is being used to monitor production machinery and equipment and takes into consideration the availability
of the equipment. Overall Equipment Eectiveness (OEE) is a measure of
equipment performance relative to design performance for the net available
time the machine is scheduled to operate. The OEE is dened by
OEE = Availability Performance eciency Yield
(9.12)
where performance eciency is how fast the machine is running relative

to its design cycle and yield is percentage of the resulting product within
quality specications. The availability is simply a term in the calculation of
OEE.
Chapter 10
Growth Management
As a product goes through the various phases of design, development and
testing, its reliability should be estimated and projected to ensure that the
product will achieve an adequate reliability at an appropriate time in its
development cycle. The reliability values at the dierent phases in the life
cycle, when plotted on log-log paper, tend to fall in a straight line. The slope
of this line is termed the growth rate. The growth rate depends on the
product complexity and on the organization responsible for the development.
Desirable growth rates can be achieved through careful analysis, planning
and management of the development program.
Reliability growth is the result of an iterative design and development
process. The essential elements in achieving growth are:
Product testing/usage to surface defects/weaknesses.
Detection and analysis of failure modes.
Feedback of identied problems/weaknesses.
Redesign eort directed towards correction of deciencies.
Fabrication of new/rebuilt prototypes.
Continuation of the iterative process.
The growth in reliability is also evident from model year to model year on
many components, particularly those where an active and intensive improvement program is on-going. The use of surface-mounted electronic circuits
in automotive applications underwent a steep learning curve as engineers
became acquainted with tolerancing and mounting techniques, about good
137
CHAPTER 10. GROWTH MANAGEMENT
Reliability Level
Design &
Development
Production
138
Operation &
Maintenance
Inherent
Reliability
Operational
Reliability
Product Life Cycle
Figure 10.1: Reliability Growth Cycle During System Life

production practices, etc. All of this eort tends to be reected in the number of failures reported under a warranty system. If we take the failure rate
from year to year we can see a denite reliability improvement. This improvement rate can be projected into the future to calculate warranty cost
and to allocate eort to facilitate an accelerated improvement.
The benets of a reliability growth management program are:
It provides a procedure that takes advantage of the experience gained
in previous development programs.
It provides an evaluation and tracking program for reliability.
It provides a means for identifying the need for corrective action concerning the reliability level.
The reliability growth management process is a component of the overall
reliability program.
10.1
Reliability Growth Relationships
This section will serve to dene the reliability growth modeling relationships.
Since the model used here is the same as that used by Duane1 , we will term
1
Duane, J.T.; Learning Curve Approach to Reliability Monitoring, IEEE Transactions on Aerospace, V.2, No. 2, April 1964.
139
Prototype
and
Systems
Testing
Product Design
and
Development
Analysis of
Test Results
Identification of
Product Deficiencies
Root Cause
Analysis
Figure 10.2: Reliability Growth Feedback System
Phase 2
Phase 3
Product Reliability Level
Release to Production
Phase 1
Product Design & Development Phase

Actual Product Growth Performance
Acceptable Growth Bands
Figure 10.3: Reliability Growth Management
140
it the Duane Growth Model (DGM). J.T. Duane was a General Electric
engineer who recognized the reliability growth phenomena while working
on large scale electrical power generators. The original DGM was applied
to a cumulative failure rate which decreases as time and experience are
gained with the product. Most recent applications have used the mean time
between failure (MTBF), which increases. Obviously, both the MTBF and
failure are related. We dene
Mc (t) = cumulative MTBF at time t
N (t) = cumulative number of failures at time t
(10.1)
Then the cumulative MTBF is

t
N (t)
(10.2)
Mc (t) = Ktm
(10.3)
Mc (t) =
The basic DGM is
where
K = a constant determined by the growth curve

t = total operating time
m = the growth rate
Example 10.1.
For a growth rate of m = 0.2 and a constant, K = 4, 000, what would the
cumulative MTBF be at 10,000 hour of testing? If we doubled the growth
rate, what improvement would be realized in the cumulative MTBF?
Since
t
Mc (t) =
N (t)
then we have
N (t) =
t
Mc (t)
Or
t
1
= t1m
m
Kt
K
which is the expected number of failures up to time t.
N (t) =
(10.4)
141
Example 10.2.
For a growth rate of m = 0.3 and for a constant K = 200, ll out the
following table.
Cume t (hours)
Cume No. fail N(t)
5,000
10,000
20,000
50,000
The instantaneous rate of change of N (t) is the instantaneous (or inherent) failure rate of the system at time t and is

dN (t)
1
(1 m)
d 1 1m
= (1 m) tm =
=
t
dt
dt K
K
Ktm
but since
Mc (t) = Ktm
Then
Mi (t) =
Mc (t)
(1 m)
(10.5)
where Mi (t) is the inherent MTBF. If we suspended testing at time t and

xed all known faults/problems with the same diligence as in the past, then
Mi (t) would be the nal, realizable, MTBF of the system.
Example 10.3.
If we had a growth rate of m = 0.2 and a constant of K = 4, 000 as given in
Example 10.1, and we suspended design and development testing at 10,000
hour, what nal MTBF would we end up with? If we put on another 10,000
hours of development testing, what kind of an improvement would we see in
the MTBF?
10.2
Growth Curve Plotting
The growth phenomenon is easier to visualize when the curve is plotted on

log-log paper. Considering the cumulative MTBF equation, which is
Mc (t) = Ktm
and taking a logarithm we have
ln Mc (t) = ln K + m ln t
This equation will plot as a straight line on log-log paper with m being the
slope of the line measured on a linear scale. Eventually we will need to
estimate the parameters from a growth plot.
142
Example 10.4.
A high mobility 10-ton truck is under development for o-road terrain. Five
prototype vehicles are built for engineering development testing. Each vehicle is subjected to 60,000 mile of testing. The test produces 150 failures.
A redesign eort attempts to improve vehicle reliability and other shortcomings discovered during testing. Subsequently, ve more vehicles are built
and tested for 60,000 miles each, resulting in 42 failures.
Again the failures are studied and the design improved. To verify the
nal design, four more vehicles are tested for 75,000 miles each. The nal
test produces 32 failures.
A. Plot the reliability growth curve on log-log paper.
B. Estimate the growth rate.
C. Estimate the current MMBF for this vehicle assuming all outstanding
problems are corrected with the same eort as in the past.
D. If another ve vehicles are tested for 100,000 miles each, what nal
MMBF would result?
E. Is it possible to achieve a 20,000 mile MMBF?
Example 10.5.
A new vehicle development program is under way to produce a reliable
vehicle in a highly automated new factory. Vehicle system test results to
date are as follows:
Test
Eng. Feas.
DT I
DT II
Pre Prod.
Mi per vehicle
80,000
100,000
100,000
100,000
No. vehicles
3
5
10
10
No. failures
11
14
18
11
Consider the following questions:

A. What is the current MMBF of this vehicle?
B. Is a 100,000 mile MMBF a reasonable goal?
C. From another 17 test vehicles, each driven 100,000 miles, what would
the expected nal MMBF be?
143
When considering the reliability of a product, one does not usually think
in terms of a cumulative MTBF but rather an inherent MTBF. History
on mature products, use of failure rate data banks and experience are all
attuned to a mature MTBF, which is an inherent MTBF. The inherent
MTBF curve is always above the cumulative curve and has the same slope.
The relationship is
Ktm
(10.6)
Mc (t) =
(1 m)
taking logs

1
+ ln K + m ln t
ln Mc (t) = ln
1m
1
So the constant vertical oset on the log-log plot is ln (1m)
which is a
positive number. The equation for determining the slope would be
m=
ln Mc (t1 ) ln Mc (t2 )
ln t1 ln t2
Then K could be found by using m to change the inherent MTBFs to cumulative MTBFs. That is,
Mc (t) = (1 m)Mi (t)
(10.7)
Example 10.6.
Consider the following planned ight hours for a development program on
a ight actuator:
Test Phase
I
II
III
Flight Hours
100,000
200,000
400,000
For a growth rate of m = 0.32 and a 100 hour MTBF of 900 hours, what
would be the nal MTBF for this component?
An Application. Let us now reinforce these growth relationships by way
of a typical aircraft development situation. Suppose we believe that our
hardware system will ultimately achieve an MTBF of 4,000 hours by 100,000
hours of ight time. We believe that we will have a threshold MTBF of
800 hours; that is, we would expect our hardware to demonstrate initially
an MTBF of at least 800 hours. These MTBF values are interpreted as
instantaneous (or inherent) values. So we have
t2 = 100, 000 hours
144
Mc (t2 ) = 4, 000 hours

and
t1 = 1 hour
Mc (t1 ) = 800 hours
Then
m=
ln 4, 000 ln 800
= 0.14
ln 100, 000 ln 1
which is the slope of the growth curve (or the growth rate).
Now we change the instantaneous MTBFs to cumulative MTBFs by
Mc (t2 ) = (1 0.14) 4, 000 hours = 3, 440 hours
and
Mc (t1 ) = (1 0.114) 800 hours = 688 hours
Then we nd
ln K = ln 688 0.14 ln 1
or
K = 688
So the growth model for this situation is
Mc (t) = 688t0.14
The mean value function is
N (t) = 0.00145t0.86
At a value of t = 8, 000 hours, we nd
N (8, 000) = 3.30 failures
Or by 8,000 hours we will see 3.30 cumulative failures.
Let us say we are putting on 8,000 hours per month. The above value
of 3.3 failures would be the expected number of failures during the rst
calendar month of the program. By the end of the second calendar month
we would expect to have a total of
N (16, 000) = 5.98 failures
Or the expected number of failures during the second calendar month would
be
N (16, 000) N (8, 000) = 2.68 failures
145
In general we dene
ij = N (tj ) N (ti )
as the expected (or average) number of failures over the time interval [ti , tj ]
where (ti < tj ). The quantity ij will be used in the next section to develop
an approach to risk assessment.
Example 10.7.
In our aircraft situation of Example 10.6, what would be the expected number of failures in each phase of testing?
Test phase
I
II
III
10.3
Cume ight hours
Cume failures
Failures/phase
Risk Analysis
The number of failures in a particular period of time would be described

probabilistically by the Poisson distribution as
Pr(r) =
r e
, r = 0, 1, 2, . . .
r!
(10.8)
If r = 2.68 failures for a certain calendar month, then this situation is

described probabilistically as
Pr(r) =
2.68r e2.68
, r = 0, 1, 2, . . .
r!
and for various values of r, we can nd the probability of this number of

failures occurring as given in Table 10.1. Here we have dened risk as the
probability of exceeding a given level of failures, r.
An easier approach is to simply use the Normal approximation to the
Poisson. We can determine sparing levels by
S = + z
(10.9)
where z is the 100(1 ) percentile of the standard normal distribution
and can be easily found in Normal tables. Then the sparing level S would
have an associated risk level of
R = 100 %
146
Table 10.1: Poisson

Probability Values
r
r Pr(r)
Pr(i)
Risk level
i=0
0 0.069
0.069
0.931
1 0.184
0.253
0.747
2 0.246
0.499
0.501
3 0.220
0.719
0.281
4 0.147
0.866
0.134
5 0.079
0.945
0.055
Example 10.8.
In our aircraft situation of Example 10.6 and 7, quote the number of failures
expected at each phase with a 10% risk.
Test phase
I
II
III
Failures/phase
Failures at 10% risk
Example 10.9.
An ignition control module has been developed and placed on a limited
volume vehicle for the last three model years. Past production volumes and
number of eld failures under a 36,000 mile warranty are as follows:
No. cars
100,000
100,000
500,000
No. failures
41,857
28,214
83,317
Consider the following:

A. Plot this data on growth paper and determine the growth rate.
B. If this design is placed on one million cars in the next model year, how
many warranty claims can be expected for the 36,000 mile warranty?
Over a 50,000 mile warranty period? Quote these values as point
estimates and at 20% risk.
C. Assuming that the same improvement eort is applied, what will be the
nal failure rate after the one million cars?
147
D. Assuming that improvement eort is halted after the one million cars,
how much should be budgeted per car for warranty expense in subsequent years? It costs $87 to replace the module under warranty.
Example 10.10.
The following data was collected on a cylinder head machining line during
the second month of launch. A team including maintenance, production,
the equipment supplier and quality were responsible for a continuous improvement program.
Cumulative hours of uptime
16
40
80
Cumulative number of faults

2
4
6
A. Plot the data and determine the growth rate. Calculate the cumulative
MTBF and record in the following table:
t
16
40
80
Mc
Plot t on the x-axis and Mc on the y-axis on log-log paper. Measure

the growth rate as the slope of the plotted line.
m = growth rate =
B. Determine the inherent MTBF which has been achieved using the formula
Mc
Mi =
=
(1 m)
C. Calculate the one-shift reliability using the equation
R(t) = et/Mi =
D. The goal is to achieve a one-shift reliability of 85%. How long will it
take them at this present rate of growth? Use the formula
Mi =
t
=
ln(1/R)
148
Then calculate
Mc = (1 m) Mi
Locate this value on the y-axis and read out the cumulative uptime.
t=
Chapter 11
Life Cycle Costing

Life cycle cost (LCC) is the total cost of ownership of a system during its
operational life. A purchased system must be supported for its total life
cycle. The cost of support over the life cycle is usually much more than
the initial acquisition cost. Acquisition cost is primarily concerned with
the concept, design/development and build or production phases of product
development. This initial acquisition cost is a non-recurring cost factor while
support cost goes on until system conversion/decommission. It is said in the
literature that 80% to 95% of the total LCC is determined by decisions made
during the concept phase.
LCC analysis is a method of economic evaluation of alternatives which
considers all relevant costs (and benets) associated with each alternative
over its life. The purpose of a LCC analysis is to explore various alternatives
to identify the most cost-eective production machinery and equipment for
a specic application. Applying the LCC concept simply means identifying
and summing all costs associated with the equipments life cycle.
11.1
Application of the Life Cycle Concept
Application of the LCC concept requires the following steps:

1. Identify the alternatives.
2. Decide on a life period.
3. Determine the cost factors.
(a) One-time
(b) Recurring
149
CHAPTER 11. LIFE CYCLE COSTING
150
50%
Non-Recurring Costs
35%
Operation &
Support
Build &
Install
12%
3%
Design &
Development
Concept
Life Cycle Phases
Figure 11.1: Cost Expenditures over Equipment Life

1
151
n
Figure 11.2: Cash Flow Diagram

4. Convert costs and savings to a common time and common dollar measure.
5. Determine the most economical alternative.
6. Perform a sensitivity analysis.
The only way that a LCC can be performed is if the vendors are required
to provide an estimate for several necessary cost factors that are clearly
dened. These relevant factors must be decided on and a worksheet with
denitions developed. After the cost factors are obtained, the buyer can
perform the LCC study.
11.2
Some Engineering Economy Concepts
To perform an LCC study, it is necessary to understand some concepts from

engineering economy. Here we will provide the methods to convert costs
throughout the life cycle to present value dollars.
11.2.1
Interest and Annuity Relationships

i = rate of return per period
(11.1)
n = number of periods
(11.2)
Typically i is quotes as a per year return and n can be in either months or

i
.
years. If n is for months, the interest per month is taken as 12
Compound Interest
Compound interest is like interest on interest. Let us consider the situation
as given in Figure 11.2 and use compound interest.

1
152
Figure 11.3: Cash Flow Diagram for Periodic Expenses

Period Principal at beginning Interest
Amount at end
1
P
iP
P + iP = P (1 + i)
2
P (1 + i)
iP (1 + i)
P (1 + i)2
2
2
3
P (1 + i)
iP (1 + i)
P (1 + i)3
..
..
.
.
P (1 + i)n
n
We see that the relationship is
S = P (1 + i)n
and
P =
S
(1 + i)n
(11.3)
(11.4)
Example 11.1.
Let us say we put $1,000 in the bank at 3% compound interest. How much
would be there after 5 years?
Example 11.2.
Let us say we want $5,000 in the bank ten years in the future. How much do
we need to put in today to achieve this objective at 6% compound interest?
11.2.2
A Series of Equal Payments
Consider the cash ow in Figure 11.3 and let

A = cost at end of period
i = the per period interest rate

Then
153
A
A
A
A
+
+ +
+
2
3
(1 + i) (1 + i)
(1 + i)
(1 + i)n
P =
which can be written as

P
= (1 + i)n1 + (1 + i)n2 + + 1
A
(1 + i)n
(11.5)
Multiplying both sides by (1 + i) gives a new series

(1 + i)n+1
P
= (1 + i)n + (1 + i)n + + (1 + i)
A
(11.6)
and subtracting 11.5 from 11.6 gives

[(1 + i)n+1 (1 + i)n ]
P
= (1 + i) 1
A
which reduces to
i(1 + i)n P = A[(1 + i)n 1]
Or,
P =A
[(1 + i)n 1]
i(1 + i)n
(11.7)
A=P
i(1 + i)n
[(1 + i)n 1]
(11.8)
and, of course,
Example 11.3.
If you nance $10,000 at 8% interest for four years, what will be the monthly
payments?
Example 11.4.
The preventive maintenance cost for a machine tool is $150 per month. The
corporation desired a rate of return of 24%. The machine tool life is ve
years. What is the present value of this series of costs to the company?
Example 11.5.
The purchase price for a small broaching machine is $35,000. The recurring
cost of owning this machine is $75 per month. This machine will be scrapped
after ten years of use with a net salvage value of $3,000. What is the LCC
for this machine tool assuming these are the only costs? Use a 12% rate of
return.
154
Acquisition costs (A)

Operating costs (O)
LCC(L)
Maintenance costs (M)
Conversion/Decommission (C)
L=A+O+M++,-C
Figure 11.4: Life Cycle Cost Components
11.3
Life Cycle Cost Components
Life cycle cost is the sum of all cost factors over the expected life of the
equipment. Appropriate discount factors must be applied for present value
analysis.
Total life cycle cost is derived as shown in Figure 11.4. Break down of
the values A, O, M and C are discussed in the following.
11.3.1
Acquisition Costs
Acquisition costs are determined as follows:

Purchase price: The delivered price of the equipment excluding transportation costs. This purchase price should account for currency differences and cost of money associated with payment schedules.
Administration and engineering costs: Personnel, travel and runo
costs.
Installation costs: The costs uniquely associated with the installation
of the equipment.
Training costs: The costs associated with training of personnel to
operate or maintain the particular equipment.
155
Purchase price (P)

Administration/Engineering (AE )
Acquisition cost (A)
Installation (I)
Training (T)
Conversion (C)
Transportation (TR )
A=P+AE +I+T+C+T R
Figure 11.5: Acquisition Cost

Conversion costs: The costs associated with conversion of the equipment to a new use during its expected life.
Transportation costs: The costs associated with the movement of the
equipment from manufacturing location to user location.
11.3.2
Operating Costs
Operating costs are determined as follows:

Direct labor costs (operating): The total cost of direct labor to operate
the equipment over its expected life.
Utilities costs: The total utilities consumption cost over the expected
life of the production machinery and equipment. Includes air, steam,
electricity, gas and water.
Consumable costs: The costs of all consumable items used by the
equipment over the expected life. Cost items to consider are coolant,
lter media, expendable tooling, etc.
Waste-handling costs: The costs of collecting and disposing of waste
products associated with the equipment.
156
Direct labor (D)
Utilities (U)
Consumables (C C )
Operating cost (O)
Waste handling (W)
Lost production (LP )
Spare parts maintenance (S P )
O=D+U+CC +W+L P +SP
Figure 11.6: Operating Cost

Cost of downtime: Downtime is due to equipment failure and appropriate downtime costing must be used.
Cost of maintaining spare parts: The cost of carrying and maintaining
spare parts inventories to support the equipment.
11.3.3
Maintenance Costs
Maintenance costs are determined as follows:

Scheduled maintenance costs: Costs of material and labor associated
with the preventive maintenance schedule during the expected life.
Preventive Maintenance Schedule: The schedule of recurring maintenance actions to ensure long life of the equipment. The maintenance actions and frequency of occurrence are provided by the
equipment supplier.
Cost of Repair: Cost of labor and parts necessary to rectify the
equipment from a failed state.
Fixed Labor Cost: Cost associated with maintaining a pool of
skilled labor to service unscheduled breakdowns.
Unscheduled maintenance costs: Cost of material and labor associated
with unscheduled breakdowns during the expected life.
157
Life of equipment (L E )
Preventive maintenance schedule (P M)

Scheduled maintenance (M S )
Cost of repair (CR )
Fixed labor costs (FL )

Maintenance costs (M)
Life of equipment (LE )
Unscheduled breakdowns (UB )

Unscheduled maintenance (MU )
Average cost of repair (CRAV)
Parts per year (P Y )
Figure 11.7: Maintenance Cost

Unscheduled Breakdowns: Resulting from equipment failure.
Average Cost of Repair: The cost of repair actions caused by
unscheduled breakdowns where the cost is averaged over several
breakdowns.
Parts Per Year: The cost of all repair parts consumed in one
year.
11.3.4
Conversion/Decommissioning Costs
Conversion costs: Cost to convert the equipment to a new use.

Decommission costs: Cost to decommission the equipment. Might
include such items as scrap/salvage value, cleaning of site, and disposal
of waste by products.
11.4
ing
158
Acquisition Practices and Life Cycle Cost-
Application of the LCC concept requires that acquisition of equipment be

viewed from the total systems perspective. The common purchasing practice
of considering only initial purchase price must be expanded to encompass
the total life cycle cost elements for the system under consideration.
Appendix A
Mathematical Formulas
A.1
A.1.1
Derivatives
Denition
The derivative of y = f (x) with respect to x is dened as

f (x + x) f (x)
dy
= lim
x0
dx
x
A.1.2
(A.1)
Common Derivatives
d n
(u )
dx
d 1
dx u

d u
dx v
d
(uv)
dx
d u
(e )
dx
d
ln u
dx
d
G(y)
dx
d
dx
= nun1
1 du
u dx
du
dv
v dx u dx
v2
dv
du
u
+v
dx
dx
u du
e
dx
1 du
u dx
d
dy
G(y)
dy
dx
db
da
f (b(x))
f (a(x))
dx
dx
(A.2)
(A.3)
(A.4)
=
=
=
=
b(x)
f (t)dt =
a(x)
du
dx
159
(A.5)
(A.6)
(A.7)
(A.8)
(A.9)
APPENDIX A. MATHEMATICAL FORMULAS

d
du
1
sin1 u =
dx
1 u2 dx
d
du
1
cos1 u =
2
dx
1 u dx
A.1.3
160
(A.10)
(A.11)
Partial Derivatives
z = f (x, y, u, . . .)

2z
x2
2z
y 2
2z
x y
A.2
=
=
=
z
x x

z
y y

z
z
=
x y
y x
(A.12)
(A.13)
(A.14)
Integrals
un du =
udv = uv

un+1
+C
n + 1
du
u
vdu
= ln u + C
eu du = eu + C
(A.15)
(A.16)
(A.17)
(A.18)
(A.19)
A.2.1
Beta Function
1
0
A.2.2
xm1 (1 x)n1 dx =
(m) (n)
(m + n)
(A.20)
Gamma Function
Denition

(n) =
0
xn1 ex dx
(A.21)
161
Common Properties and Values

(z + 1)

1
2

3
2
(n)
A.2.3
A.3
= z(z)
(A.23)
2
= (n 1)!, for n = an integer
(A.25)
n! = n (n 1) (n 2) 1
(A.26)
0! = 1
(A.27)
Series
i =
i=1
n
i2 =
i=1
r
1
nj+1
j=1
n(n + 1)
2
(A.28)
1
n(n + 1)(2n + 1)
6
(A.29)
n + 12
ln
nr+
1
2
(A.30)
Exponential Series
x2 x3
+
+
2!
3!
x2 x3
= 1x+
+
2!
3!
ex = 1 + x +
ex
A.3.2
(A.24)
Factorial Numbers
n
A.3.1
(A.22)
(A.31)
(A.32)
Geometric Progression

i=1
i=0
i = 1 + + 2 + =
1
, where 2 < 1 (A.33)
1
ii1 = 1 + 2 + 32 + =
1
(1 )2
(A.34)

i=2
i (i 1)i2 = 1 2 + 2 3 + 3 42 + =
162
2
(A.35)
(1 )3
Appendix B
Statistical Tables
Index to Tables
Statistical table
Cumulative normal
Extended normal
t-distribution
2 -distribution
F -distribution
Weibull
Median rank
5% Rank
95% Rank
163
page
195
196
197
198
202
206
208
209
210
APPENDIX B. STATISTICAL TABLES
B.1
164
Cumulative Normal Distribution
.00
.01
.02
.03
.04
.05
.06
.07
.08
.09
.0
.1
.2
.3
.4
.5000
.5398
.5793
.6179
.6554
.5040
.5438
.5832
.6217
.6591
.5080
.5478
.5871
.6255
.6628
.5120
.5517
.5910
.6293
.6664
.5160
.5557
.5948
.6331
.6700
.5199
.5596
.5987
.6368
.6736
.5239
.5636
.6026
.6406
.6772
.5279
.5675
.6064
.6443
.6808
.5319
.5714
.6103
.6480
.6844
.5359
.5753
.6141
.6517
.6879
.5
.6
.7
.8
.9
.6915
.7257
.7580
.7881
.8159
.6950
.7291
.7611
.7910
.8186
.6985
.7324
.7642
.7939
.8212
.7019
.7357
.7673
.7967
.8238
.7054
.7389
.7704
.7995
.8264
.7088
.7422
.7734
.8023
.8289
.7123
.7454
.7764
.8051
.8315
.7157
.7486
.7794
.8078
.8340
.7190
.7517
.7823
.8106
.8365
.7224
.7549
.7852
.8133
.8389
1.0
1.1
1.2
1.3
1.4
.8413
.8643
.88493
.90320
.91924
.8438
.8665
.88686
.90490
.92073
.8461
.8686
.88877
.90658
.92220
.8485
.8708
.89065
.90824
.92364
.8508
.8729
.89251
.90988
.92507
.8531
.8749
.89435
.91149
.92647
.8554
.8770
.89617
.91309
.92785
.8577
.8790
.89796
.91466
.92922
.8599
.8810
.89973
.91621
.93056
.6621
.8830
.90147
.91774
.93189
1.5
1.6
1.7
1.8
1.9
.93319
.94520
.95543
.96407
.97128
.93448
.94630
.95637
.96485
.97193
.93574
.94738
.95728
.96562
.97257
.93699
.94845
.95818
.96638
.97320
.93822
.94950
.95907
.96712
.97381
.93943
.95053
.95994
.96784
.97441
.94062
.95154
.96080
.96856
.97500
.94179
.95254
.96164
.96926
.97558
.94295
.95352
.96246
.96995
.97615
.94408
.95449
.96327
.97062
.97670
2.0
2.1
2.2
2.3
2.4
.97725
.98214
.98610
.98928
.99180
.97778
.98257
.98645
.98956
.99202
.97831
.98300
.98679
.98983
.99224
.97882
.98341
.98713
.99010
.99245
.97932
.98382
.98745
.99036
.99266
.97982
.98422
.98778
.99061
.99286
.98030
.98461
.98809
.99086
.99305
.98077
.98500
.98840
.99111
.99324
.98124
.98537
.98870
.99134
.99343
.98169
.98574
.98899
.99158
.99361
2.5
2.6
2.7
2.8
2.9
.99379
.99534
.99653
.99744
.99813
.99396
.99547
.99664
.99752
.99819
.99413
.99560
.99674
.99760
.99825
.99430
.99573
.99683
.99767
.99831
.99446
.99585
.99693
.99774
.99836
.99461
.99598
.99702
.99781
.99841
.99477
.99609
.99711
.99788
.99846
.99492
.99621
.99720
.99795
.99851
.99506
.99632
.99728
.99801
.99856
.99520
.99643
.99736
.99807
.99861
3.0
3.1
3.2
3.3
3.4
.99865
.99903
.99931
.99952
.99966
.99869
.99906
.99934
.99953
.99968
.99874
.99910
.99936
.99955
.99969
.99878
.99913
.99938
.99957
.99970
.99882
.99916
.99940
.99958
.99971
.99886
.99918
.99942
.99960
.99972
.99889
.99921
.99944
.99961
.99973
.99893
.99924
.99946
.99962
.99974
.99896
.99926
.99948
.99964
.99975
.99900
.99929
.99950
.99965
.99976
B.2
165
Extended Normal Distribution
.93 = .999
Z
3.5
3.6
3.7
3.8
3.9
4.0
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
.00
.01
.02
.03
.04
.05
.06
.07
.08
.09
.93 7674
.93 8409
.93 8922
.94 2765
.94 5190
.94 6833
.94 7934
.94 8665
.95 1460
.95 4587
.95 6602
.95 7888
.95 8699
.96 2067
.96 5208
.937759
.938469
.938964
.943052
.945385
.946964
.948022
.948723
.951837
.954831
.956759
.957987
.958761
.962453
.965446
.937842
.938527
.940039
.943327
.945573
.947090
.948106
.948778
.952199
.955065
.956908
.958081
.958821
.962822
.965673
.937922
.938583
.940426
.943593
.945753
.947211
.948186
.948832
.952545
.955288
.957051
.958172
.958877
.963173
.965889
.93 7999
.93 8637
.94 0799
.94 3848
.94 5926
.94 7327
.94 8263
.94 8882
.95 2876
.95 5502
9 57187
9 58258
9 58931
.96 3508
.96 6094
.93 8074
.93 8689
.94 1158
.94 4094
.94 6092
.94 7439
.94 8338
.94 8931
.95 3193
.95 5706
.95 7318
.95 8340
.95 8983
.96 3827
.96 6289
.938146
.938739
.941504
.944331
.946253
.947546
.948409
.948978
.953497
.955902
.957442
.958419
.960320
.964131
.966475
.938215
.938787
.941838
.944558
.946406
.947649
.948477
.950226
.953788
.956089
.957561
.958494
.960789
.964420
.966652
.9 38282
.9 38834
.9 42159
.9 44777
.9 46554
.9 47748
.9 48542
.9 50655
.9 54066
.9 56268
.9 57675
.9 58566
.9 61235
.9 64696
.9 66821
.93 8347
.93 8879
.94 2568
.94 4988
.94 6696
.94 7843
.94 8605
.95 066
.95 4332
.95 6439
.95 7784
.95 8634
.96 1661
.96 4958
.96 6981
B.3
df
166
t-distribution
.40
.25
.10
.05
.025
.01
.005
.0025
.001
.0005
1
2
3
4
5
0.325
0.289
0.277
0.270
0.267
1.000
0.817
0.765
0.741
0.726
3.078
1.886
1.637
1.534
1.476
6.314
2.921
2.352
2.133
2.015
12.706
4.303
3.182
2.777
2.570
31.821
6.964
4.540
3.746
3.365
63.657
9.926
5.841
4.603
4.032
127.32
14.088
7.452
5.597
4.774
318.31
22.328
10.214
7.174
5.892
636.62
31.598
12.924
8.610
6.869
6
7
8
9
19
0.264
0.263
0.261
0.261
0.260
0.717
0.711
0.706
0.702
0.699
1.440
1.415
1.397
1.383
1.372
1.944
1.895
1.860
1.834
1.813
2.448
2.365
2.307
2.263
2.229
3.143
2.999
2.897
2.822
2.764
3.707
3.500
3.356
3.250
3.170
4.315
4.029
3.833
3.690
3.582
5.202
4.783
4.500
4.297
4.144
5.947
5.403
5.039
4.780
4.586
11
12
13
14
15
0.259
0.259
0.258
0.258
0.257
0.697
0.695
0.694
0.692
0.691
1.364
1.356
1.350
1.345
1.341
1.796
1.783
1.771
1.762
1.753
2.202
2.179
2.161
2.145
2.132
2.719
2.682
2.651
2.625
2.603
3.106
3.055
3.013
2.977
2.947
3.497
3.429
3.373
3.326
3.287
4.025
3.930
3.852
3.788
3.733
4.437
4.318
4.221
4.141
4.073
16
17
18
19
20
0.257
0.257
0.257
0.257
0.256
0.690
0.689
0.688
0.687
0.687
1.337
1.334
1.331
1.328
1.326
1.746
1.740
1.734
1.730
1.725
2.120
2.110
2.101
2.094
2.086
2.584
2.567
2.553
2.540
2.529
2.921
2.899
2.879
2.861
2.846
3.252
3.223
3.197
3.174
3.154
3.687
3.646
3.611
3.580
3.552
4.015
3.965
3.922
3.884
3.850
21
22
23
24
25
0.256
0.256
0.256
0.256
0.256
0.686
0.685
0.685
0.685
0.684
1.323
1.321
1.320
1.318
1.317
1.721
1.718
1.714
1.711
1.709
2.080
2.074
2.069
2.064
2.060
2.518
2.509
2.500
2.493
2.486
2.832
2.819
2.808
2.797
2.788
3.136
3.119
3.104
3.091
3.079
3.528
3.505
3.485
3.467
3.451
3.820
3.792
3.768
3.746
3.725
26
27
28
29
30
0.256
0.255
0.255
0.255
0.255
0.684
0.683
0.683
0.683
0.682
1.315
1.314
1.313
1.312
1.311
1.706
1.704
1.702
1.700
1.698
2.056
2.052
2.049
2.046
2.043
2.479
2.473
2.468
2.463
2.458
2.779
2.771
2.764
2.757
2.750
3.067
3.057
3.047
3.038
2.020
3.435
3.421
3.409
3.397
3.386
3.707
3.690
3.674
3.660
3.646
40
60
120
0.255
0.254
0.253
0.680
0.678
0.676
1.303
1.296
1.289
1.684
1.671
1.658
2.022
2.001
1.980
2.424
2.391
2.358
2.705
2.661
2.618
2.972
2.915
2.860
3.307
3.232
3.160
3.551
3.460
3.374
.50
.455
1.386
2.366
3.357
4.351
5.348
6.346
7.344
8.343
9.342
10.341
11.340
12.340
13.339
14.339
15.338
16.338
17.338
18.338
19.337
20.337
21.337
22.337
23.337
24.337
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
23.858
24.939
26.018
27.096
28.172
18.418
19.511
20.601
21.689
22.775
12.899
14.011
15.119
16.222
17.322
7.231
8.383
9.524
10.656
11.781
24.935
26.039
27.141
28.241
29.339
19.369
20.489
21.605
22.718
23.828
13.701
14.845
15.984
17.117
18.245
7.841
9.037
10.219
11.389
12.549
1.323
2.773
4.108
5.385
6.626
.25
26.171
27.301
28.429
29.553
30.675
20.465
21.615
22.760
23.900
25.038
14.631
15.812
16.985
18.151
19.311
8.558
9.803
11.030
12.242
13.442
1.642
3.219
4.642
5.989
7.289
.20
29.615
30.813
32.007
33.196
34.382
23.542
24.769
25.989
27.204
28.412
17.275
18.549
19.812
21.064
22.307
10.645
12.017
13.362
14.684
15.987
2.706
4.605
6.251
7.779
9.236
.10
5.024
7.378
9.348
11.143
12.832
14.449
16.013
17.535
19.023
20.483
21.920
23.337
24.736
26.119
27.488
28.845
30.191
31.526
32.852
34.170
35.479
36.781
38.076
39.364
40.646
12.592
14.067
15.507
16.919
18.307
19.675
21.920
22.362
23.685
24.996
26.296
27.587
28.869
30.144
31.410
32.671
33.924
35.172
36.415
37.652
.025
3.841
5.991
7.815
9.488
11.070
.05
38.932
40.289
41.638
42.980
44.314
32.000
33.409
34.805
36.191
37.566
24.725
26.217
27.688
29.141
30.578
16.812
18.475
20.090
21.666
23.209
6.635
9.210
11.345
13.277
15.086
.01
41.401
42.796
44.181
45.558
46.928
34.267
35.718
37.156
38.582
39.997
26.757
28.300
29.819
31.319
32.801
18.548
20.278
21.955
23.589
25.188
7.879
10.597
12.838
14.860
16.750
.005
B.4
1.074
2.408
3.665
4.878
6.064
.30

167
2 -distribution
60.304
66.792
73.190
79.512
85.769
57.359
63.706
69.971
76.167
82.305
88.391
94.433
100.436
106.403
112.338
118.244
124.125
129.980
135.814
141.627
147.421
153.197
158.956
53.207
59.345
65.414
71.424
77.384
83.301
89.181
95.027
100.843
106.632
112.397
118.139
123.861
129.565
135.250
140.920
146.574
152.215
49.798
55.755
61.653
67.502
73.309
79.080
84.819
90.530
96.216
101.879
107.521
113.145
118.751
124.342
129.918
135.480
141.030
146.568
46.034
51.780
57.480
63.141
68.770
74.370
79.946
85.500
91.034
96.550
102.050
107.536
113.008
118.468
123.917
129.355
134.782
140.201
41.802
47.295
52.757
58.194
63.610
69.006
74.387
79.752
85.105
90.446
95.777
101.097
106.409
111.713
117.009
112.299
127.581
132.858
40.221
45.615
50.984
56.333
61.665
66.982
72.286
77.578
82.860
88.132
93.396
98.653
103.902
109.145
114.381
119.612
124.838
130.059
38.860
44.166
49.453
54.725
59.983
65.229
70.466
75.693
80.912
86.124
91.329
96.529
101.723
106.911
112.095
117.275
122.451
127.623
34.338
39.337
44.337
49.336
54.336
59.336
64.336
69.335
74.335
79.335
84.335
89.335
94.335
99.335
104.335
109.335
114.335
119.335
35
40
45
50
55
60
65
70
75
80
85
90
95
100
105
110
115
120
151.956
157.814
163.654
122.337
128.310
134.257
140.179
146.078
91.970
93.122
104.230
110.300
116.334
48.290
49.645
50.993
52.336
53.672
45.642
46.963
48.278
49.588
50.892
41.923
43.194
44.461
45.722
46.979
38.885
40.113
41.337
42.557
43.773
35.563
36.741
37.916
39.087
40.256
31.795
32.912
34.027
35.139
36.250
30.434
31.528
32.620
33.711
34.800
29.246
30.319
31.391
32.461
33.530
.005
25.336
26.336
27.336
28.336
29.336
.01
26
27
28
29
30
.025
.05
.10
.20
.25
.30
.50

168
.995
.0000393
.0100
.0717
.207
.412
.676
.989
1.344
1.735
2.156
2.603
3.074
3.565
4.075
4.601
5.142
5.697
6.265
6.844
7.434
8.034
8.643
9.260
9.886
10.520
n
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
8.897
9.542
10.196
10.856
11.524
5.812
6.408
7.015
7.633
8.260
3.053
3.571
4.107
4.660
5.229
.872
1.239
1.646
2.088
2.558
.000157
.0201
.115
.297
.554
.99
10.283
10.982
11.688
12.401
13.120
6.908
7.564
8.231
8.907
9.591
3.816
4.404
5.009
5.629
6.262
1.237
1.690
2.180
2.700
3.247
.000982
.0506
.216
.484
.831
.975
11.591
12.338
13.091
13.848
14.611
7.962
8.672
9.390
10.117
10.851
4.575
5.226
5.892
6.571
7.261
1.635
2.167
2.733
3.325
3.940
.00393
.103
.352
.711
1.145
.95
13.240
14.041
14.848
15.659
16.473
9.312
10.085
10.865
11.651
12.443
5.578
6.304
7.042
7.790
8.574
2.204
2.833
3.490
4.168
4.865
.0158
.211
.584
1.064
1.610
.90
.75
.102
.575
1.213
1.923
2.675
3.455
4.255
5.071
5.899
6.737
7.584
8.438
9.299
10.165
11.306
11.192
12.792
13.675
14.562
15.452
16.344
17.240
18.137
19.037
19.939
.80
.0642
.446
1.005
1.649
2.343
3.070
3.822
4.594
5.380
6.179
6.989
7.807
8.634
9.467
10.307
11.152
12.002
12.857
13.716
14.578
15.445
16.314
17.187
18.062
18.940
17.182
18.101
19.021
19.943
20.867
12.624
13.531
14.440
15.352
16.266
8.148
9.034
9.926
10.821
11.721
3.828
4.671
5.527
6.393
7.267
.148
.713
1.424
2.195
3.000
.70
18.8
19.7
20.7
21.7
22.6
14.0
14.9
15.9
16.9
17.8
9.24
10.2
11.1
12.1
13.0
4.57
5.49
6.42
7.36
8.30
.27
1.02
1.87
2.75
3.66
.60

169
23.6
24.5
25.5
26.5
27.4
32.3
37.1
42.0
46.9
51.7
56.6
61.5
66.4
71.3
76.2
81.1
86.0
90.9
95.8
-
21.792
22.719
23.647
24.577
25.508
30.181
34.874
39.586
44.314
49.055
53.808
58.572
63.344
68.125
72.913
77.707
82.508
87.314
92.125
96.941
101.761
106.585
111.413
20.843
21.749
22.657
23.567
24.478
29.058
33.664
38.294
42.944
47.612
52.295
56.991
61.698
66.416
71.144
75.880
80.623
85.374
90.131
94.894
99.663
104.437
109.216
19.820
20.703
21.588
22.475
23.364
27.820
32.326
36.863
41.426
46.011
50.614
55.233
59.868
64.515
69.174
73.843
78.522
83.210
87.906
92.610
97.321
102.038
106.762
17.292
18.114
18.939
19.768
20.599
24.812
29.067
33.367
37.706
42.078
46.478
50.902
55.349
59.815
64.299
68.799
73.313
77.841
82.381
86.933
91.495
96.067
100.648
15.379
16.151
16.928
17.708
18.493
22.462
26.507
30.610
34.762
38.956
43.186
47.448
51.737
56.052
60.390
64.748
69.124
73.518
77.928
82.352
86.790
91.240
95.703
13.844
14.573
15.308
16.047
16.791
20.558
24.423
28.356
32.348
36.390
40.474
44.595
48.750
52.935
57.146
61.382
65.640
69.919
74.216
78.530
82.861
87.207
91.567
18.484
22.142
25.880
29.687
33.552
37.467
41.427
45.426
49.460
53.526
57.621
61.741
65.886
70.053
74.241
78.448
82.672
86.913
35.510
39.360
43.253
47.186
51.153
55.151
59.179
63.963
67.312
71.414
75.536
79.679
83.839
85
90
95
100
105
110
115
120
.60
.70
.75
.80
.90
12.198
12.879
13.565
14.256
14.953
11.160
11.808
12.461
13.121
13.787
17.156
20.674
24.281
27.962
31.708
26
27
28
29
30
35
40
45
50
55
60
65
70
75
80
.975
.95
.99
.995

170
10
12
F-distribution upper 1 % points

15
20
24
30
40
60
120
13.75
12.25
11.26
10.56
10.04
9.65
9.33
9.07
8.86
8.68
8.53
8.40
8.29
8.18
8.10
8.02
7.95
7.88
7.82
7.77
7.72
7.68
7.64
7.60
7.58
7.31
7.08
6.85
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
5.18
4.98
4.79
5.53
5.49
5.45
5.42
5.39
5.78
5.72
5.66
5.61
5.57
6.23
6.11
6.01
5.93
5.85
7.21
6.93
6.70
6.52
6.36
10.93
9.55
8.65
8.02
7.58
4.31
4.13
3.95
4.64
4.60
4.57
4.54
4.51
4.87
4.82
4.76
4.72
4.68
5.29
5.19
5.09
5.01
4.94
6.22
5.95
5.74
5.56
5.42
9.78
8.45
7.59
6.99
6.55
3.83
3.65
3.48
4.14
4.11
4.07
4.04
4.02
4.37
4.31
4.26
4.22
4.18
4.77
4.67
4.58
4.50
4.43
5.67
5.41
5.21
5.04
4.89
9.15
7.85
7.01
6.42
5.99
3.51
3.34
3.17
3.82
3.78
3.75
3.73
3.70
4.04
3.99
3.94
3.90
3.86
4.44
4.34
4.25
4.17
4.10
5.32
5.06
4.86
4.70
4.56
8.75
7.46
6.63
6.06
5.64
3.29
3.12
2.96
3.59
3.56
3.53
3.50
3.47
3.81
3.76
3.71
3.67
3.63
4.20
4.10
4.01
3.94
3.87
5.07
4.82
4.62
4.46
4.32
8.47
7.19
6.37
5.80
5.39
3.12
2.95
2.79
3.42
3.39
3.36
3.33
3.30
3.64
3.59
3.54
3.50
3.46
4.03
3.93
3.84
3.77
3.70
4.89
4.64
4.44
4.28
4.14
8.28
6.99
6.18
5.61
5.20
2.99
2.82
2.66
3.29
3.26
3.23
3.20
3.17
3.51
3.45
3.41
3.36
3.32
3.89
3.79
3.71
3.63
3.56
4.74
4.50
4.30
4.14
4.00
8.10
6.84
6.03
5.47
5.06
2.89
2.72
2.56
3.18
3.15
3.12
3.09
3.07
3.40
3.35
3.30
3.26
3.22
3.78
3.68
3.60
3.52
3.46
4.63
4.39
4.19
4.03
3.89
7.98
6.72
5.91
5.35
4.94
2.80
2.63
2.47
3.09
3.06
3.03
3.00
2.98
3.31
3.26
3.21
3.17
3.13
3.69
3.59
3.51
3.43
3.37
4.54
4.30
4.10
3.94
3.81
7.87
6.62
5.81
5.26
4.85
2.66
2.50
2.34
2.96
2.93
2.90
2.87
2.84
3.17
3.12
3.07
3.03
2.99
3.55
3.46
3.37
3.30
3.23
4.40
4.16
3.96
3.80
3.67
7.72
6.47
5.67
5.11
4.71
2.52
2.35
2.19
2.82
2.78
2.75
2.73
2.70
3.03
2.98
2.93
2.89
2.85
3.41
3.31
3.23
3.15
3.09
4.25
4.01
3.82
3.66
3.52
7.56
6.31
5.52
4.96
4.56
2.37
2.20
2.03
2.66
2.63
2.60
2.57
2.55
2.88
2.83
2.78
2.74
2.70
3.26
3.16
3.08
3.00
2.94
4.10
3.86
3.66
3.51
3.37
7.40
6.16
5.36
4.81
4.41
2.29
2.12
1.95
2.58
2.55
2.52
2.49
2.47
2.80
2.75
2.70
2.66
2.62
3.18
3.08
3.00
2.92
2.86
4.02
3.78
3.59
3.43
3.29
7.31
6.07
5.28
4.73
4.33
2.20
2.03
1.86
2.50
2.47
2.44
2.41
2.39
2.72
2.67
2.62
2.58
2.54
3.10
3.00
2.92
2.84
2.78
3.94
3.70
3.51
3.35
3.21
7.23
5.99
5.20
4.65
4.25
2.11
1.94
1.76
2.42
2.38
2.35
2.33
2.30
2.64
2.58
2.54
2.49
2.45
3.02
2.92
2.84
2.76
2.69
3.86
3.62
3.43
3.27
3.13
7.14
5.91
5.12
4.57
4.17
2.02
1.84
1.68
2.33
2.29
2.26
2.23
2.21
2.55
2.50
2.45
2.40
2.36
2.93
2.83
2.75
2.67
2.61
3.78
3.54
3.34
3.18
3.05
7.06
5.82
5.03
4.48
4.08
1.92
1.73
1.53
2.23
2.20
2.17
2.14
2.11
2.46
2.40
2.35
2.31
2.27
2.84
2.75
2.68
2.58
2.52
3.69
3.45
3.25
3.09
2.96
6.97
5.74
4.95
4.40
4.00
4052.
4999.
5403.
5764.
5859.
5625.
5928.
5981.
6022.
6056.
6106.
6157.
6209.
6235.
6261.
6287.
6313.
6339.
98.50
99.00
99.17
99.30
99.34
99.25
99.36
99.38
99.39
99.40
99.42
99.44
99.45
99.46
99.47
99.48
99.48
99.50
34.12
30.82
29.46
28.24
27.91
28.71
27.67
27.49
27.35
27.23
27.05
26.87
26.69
26.60
26.51
26.41
26.32
26.22
21.20
18.00
16.70
15.52
15.21
15.98
14.98
14.80
14.66
14.55
14.37
14.20
14.02
13.93
13.84
13.75
13.65
13.56
16.26
13.27
12.06
10.97
10.67
11.39
10.46
10.29
10.16
10.05
9.89
9.72
9.55
9.47
9.38
9.29
9.20
9.11
B.5
1
2
3
4
5
N2/N1 1

171
F -Distribution
10
12

15
20
24
30
40
60
120
5.99
5.59
5.32
5.12
4.96
4.84
4.75
4.67
4.60
4.54
4.49
4.45
4.41
4.38
4.35
4.32
4.30
4.28
4.26
4.24
4.23
4.21
4.20
4.18
4.17
4.08
4.00
3.92
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
3.23
3.15
3.07
3.37
3.35
3.34
3.33
3.32
3.47
3.44
3.42
3.40
3.39
3.64
3.59
3.56
3.52
3.50
3.98
3.89
3.81
3.74
3.68
5.15
4.74
4.46
4.26
4.10
2.84
2.76
2.68
2.97
2.96
2.95
2.93
2.92
3.07
3.05
3.03
3.01
2.99
3.24
3.20
3.16
3.13
3.10
3.59
3.49
3.41
3.35
3.29
4.76
4.35
4.07
3.86
3.71
2.61
2.53
2.45
2.74
2.73
2.71
2.70
2.69
2.84
2.82
2.80
2.78
2.76
3.01
2.96
2.93
2.90
2.87
3.36
3.26
3.18
3.11
3.06
4.54
4.12
3.84
3.64
3.48
2.45
2.37
2.29
2.59
2.57
2.56
2.55
2.53
2.68
2.66
2.64
2.62
2.60
2.85
2.81
2.77
2.74
2.71
3.21
3.11
3.03
2.96
2.90
4.39
3.97
3.69
3.48
3.33
2.34
2.25
2.18
2.48
2.46
2.45
2.43
2.42
2.57
2.55
2.53
2.51
2.49
2.74
2.70
2.66
2.63
2.60
3.10
3.00
2.92
2.85
2.79
4.29
3.87
3.58
3.37
3.22
2.25
2.17
2.09
2.39
2.37
2.36
2.35
2.34
2.49
2.46
2.44
2.42
2.41
2.66
2.61
2.58
2.54
2.51
3.01
2.91
2.83
2.76
2.71
4.21
3.79
3.50
3.29
3.14
2.18
2.10
2.02
2.32
2.31
2.29
2.28
2.27
2.42
2.40
2.38
2.36
2.34
2.59
2.55
2.51
2.48
2.45
2.95
2.85
2.77
2.70
2.64
4.15
3.73
3.44
3.23
3.07
2.12
2.04
1.96
2.27
2.25
2.24
2.22
2.21
2.37
2.34
2.32
2.30
2.28
2.54
2.49
2.46
2.42
2.39
2.90
2.80
2.71
2.65
2.59
4.10
3.68
3.39
3.18
3.02
2.08
1.99
1.91
2.22
2.20
2.19
2.18
2.16
2.32
2.30
2.27
2.25
2.24
2.49
2.45
2.41
2.38
2.35
2.85
2.75
2.67
2.60
2.54
4.06
3.64
3.35
3.14
2.98
2.00
1.92
1.83
2.15
2.13
2.12
2.10
2.09
2.25
2.23
2.20
2.18
2.17
2.42
2.38
2.34
2.31
2.28
2.79
2.69
2.60
2.53
2.48
4.00
3.58
3.29
3.07
2.91
1.92
1.84
1.75
2.07
2.06
2.04
2.03
2.01
2.18
2.15
2.13
2.11
2.09
2.35
2.31
2.27
2.23
2.20
2.72
2.62
2.53
2.46
2.40
3.94
3.51
3.22
3.01
2.85
1.84
1.75
1.66
1.99
1.97
1.96
1.94
1.93
2.10
2.07
2.05
2.03
2.01
2.28
2.23
2.19
2.16
2.12
2.65
2.54
2.46
2.39
2.33
3.87
3.45
3.15
2.94
2.78
1.79
1.70
1.61
1.95
1.93
1.91
1.90
1.89
2.05
2.03
2.01
1.98
1.96
2.24
2.19
2.15
2.11
2.08
2.61
2.51
2.42
2.35
2.29
3.84
3.41
3.12
2.90
2.74
1.74
1.65
1.55
1.90
1.88
1.87
1.85
1.84
2.01
1.98
1.96
1.94
1.92
2.19
2.15
2.11
2.07
2.04
2.57
2.47
2.38
2.31
2.25
3.81
3.38
3.08
2.87
2.70
1.69
1.59
1.50
1.85
1.84
1.82
1.81
1.79
1.96
1.94
1.91
1.89
1.87
2.15
2.10
2.06
2.03
1.99
2.53
2.43
2.34
2.27
2.20
3.77
3.34
3.04
2.83
2.66
1.64
1.53
1.43
1.80
1.79
1.77
1.75
1.74
1.92
1.89
1.87
1.84
1.82
2.11
2.06
2.02
1.98
1.95
2.49
2.39
2.30
2.22
2.16
3.74
3.30
3.01
2.79
2.62
1.58
1.47
1.35
1.75
1.73
1.71
1.70
1.68
1.87
1.84
1.81
1.79
1.77
2.06
2.01
1.97
1.93
1.90
2.45
2.34
2.25
2.18
2.11
3.70
3.27
2.97
2.75
2.58
161.45 199.50 215.71 224.58 230.16 233.99 236.77 238.88 240.54 241.88 243.91 245.95 248.01 249.05 250.09 251.14 252.20 253.25
19.02
19.19
19.30
19.33
19.28
19.37
19.43
19.46
19.50
18.51
19.40
19.42
19.47
19.36
19.39
19.45
19.48
19.49
9.55
9.28
9.00
8.94
9.11
8.85
8.71
8.64
8.56
10.13
8.79
8.75
8.62
8.89
8.81
8.67
8.60
8.58
6.95
6.59
6.26
6.17
6.39
6.05
5.86
5.77
5.66
7.71
5.97
5.91
5.75
6.10
6.00
5.80
5.72
5.69
5.79
5.41
5.05
4.95
5.20
4.82
4.62
4.53
4.40
6.61
4.74
4.68
4.50
4.88
4.78
4.56
4.47
4.44
6
7
8
9
10
1
2
3
4
5
N2/N1 1

172
4.06
3.78
3.59
3.46
3.36
3.29
3.23
3.18
3.14
3.10
3.07
3.05
3.03
3.01
2.99
2.97
2.96
2.95
2.94
2.93
2.92
2.91
2.90
2.89
2.89
2.88
2.84
2.79
2.75
2.71
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
39.86
8.53
5.54
4.54
N1
1
2
3
4
N2
53.59
9.16
5.39
4.19
3.62
3.29
3.07
2.92
2.81
2.73
2.66
2.61
2.56
2.52
2.49
2.46
2.44
2.42
2.40
2.38
2.36
2.35
2.34
2.33
2.32
2.31
2.30
2.29
2.28
2.28
2.23
2.18
2.13
2.08
3.78
3.46
3.26
3.11
3.01
2.92
2.86
2.81
2.76
2.73
2.70
2.67
2.64
2.62
2.61
2.59
2.57
2.56
2.55
2.54
2.53
2.52
2.51
2.50
2.50
2.49
2.44
2.39
2.35
2.30
49.50
9.00
5.46
4.32
2.14
2.09
2.04
1.99
1.94
2.18
2.17
2.17
2.16
2.15
2.25
2.23
2.22
2.21
2.19
2.36
2.33
2.31
2.29
2.27
2.61
2.54
2.48
2.43
2.39
3.52
3.18
2.96
2.81
2.69
55.83
9.24
5.34
4.11
2.05
2.00
1.95
1.90
1.85
2.09
2.08
2.07
2.06
2.06
2.16
2.14
2.13
2.11
2.10
2.27
2.24
2.22
2.20
2.18
2.52
2.45
2.39
2.35
2.31
3.45
3.11
2.88
2.73
2.61
57.24
9.29
5.31
4.05
1.98
1.93
1.87
1.82
1.77
2.02
2.01
2.00
2.00
1.99
2.09
2.08
2.06
2.05
2.04
2.21
2.18
2.15
2.13
2.11
2.46
2.39
2.33
2.28
2.24
3.40
3.05
2.83
2.67
2.55
58.20
9.33
5.28
4.01
1.93
1.87
1.82
1.77
1.72
1.97
1.96
1.95
1.94
1.93
2.04
2.02
2.01
1.99
1.98
2.16
2.13
2.10
2.08
2.06
2.41
2.34
2.28
2.23
2.19
3.37
3.01
2.78
2.62
2.51
58.91
9.35
5.27
3.98
1.88
1.83
1.77
1.72
1.67
1.93
1.92
1.91
1.90
1.89
2.00
1.98
1.97
1.95
1.94
2.12
2.09
2.06
2.04
2.02
2.38
2.30
2.24
2.20
2.15
3.34
2.98
2.75
2.59
2.47
59.44
9.37
5.25
3.95
1.85
1.79
1.74
1.68
1.63
1.89
1.88
1.87
1.87
1.86
1.96
1.95
1.93
1.92
1.91
2.09
2.06
2.03
2.00
1.98
2.35
2.27
2.21
2.16
2.12
3.32
2.96
2.72
2.56
2.44
59.86
9.38
5.24
3.94
1.82
1.76
1.71
1.65
1.60
1.87
1.86
1.85
1.84
1.83
1.94
1.92
1.90
1.89
1.88
2.06
2.03
2.00
1.98
1.96
2.32
2.25
2.19
2.14
2.10
3.30
2.94
2.70
2.54
2.42
60.19
9.39
5.23
3.92
10
1.77
1.71
1.66
1.60
1.55
1.82
1.81
1.80
1.79
1.78
1.89
1.87
1.86
1.84
1.83
2.02
1.99
1.96
1.93
1.91
2.28
2.21
2.15
2.10
2.05
3.27
2.90
2.67
2.50
2.38
60.71
9.41
5.22
3.90
12
F - distribution upper 10%
1.72
1.66
1.60
1.55
1.49
1.77
1.76
1.75
1.74
1.73
1.84
1.83
1.81
1.80
1.78
1.97
1.94
1.91
1.89
1.86
2.24
2.17
2.10
2.05
2.01
3.24
2.87
2.63
2.46
2.34
61.22
9.42
5.20
3.87
15
1.67
1.61
1.54
1.48
1.42
1.72
1.71
1.70
1.69
1.68
1.79
1.78
1.76
1.74
1.73
1.92
1.89
1.86
1.84
1.81
2.20
2.12
2.06
2.01
1.96
3.21
2.84
2.59
2.42
2.30
61.74
9.44
5.18
3.84
20
1.64
1.57
1.51
1.45
1.38
1.69
1.68
1.67
1.66
1.65
1.77
1.75
1.73
1.72
1.70
1.90
1.87
1.84
1.81
1.79
2.18
2.10
2.04
1.98
1.94
3.19
2.82
2.58
2.40
2.28
62.00
9.45
5.18
3.83
24
1.61
1.54
1.48
1.41
1.34
1.66
1.65
1.64
1.63
1.62
1.74
1.72
1.70
1.69
1.67
1.87
1.84
1.81
1.78
1.76
2.16
2.08
2.01
1.96
1.91
3.17
2.80
2.56
2.38
2.25
62.26
9.46
5.17
3.82
30
1.57
1.51
1.44
1.37
1.30
1.63
1.61
1.60
1.59
1.58
1.71
1.69
1.67
1.66
1.64
1.85
1.81
1.78
1.75
1.73
2.13
2.05
1.99
1.93
1.89
3.16
2.78
2.54
2.36
2.23
62.53
9.47
5.16
3.80
40
1.54
1.47
1.40
1.32
1.24
1.59
1.58
1.57
1.56
1.55
1.68
1.66
1.64
1.62
1.61
1.82
1.78
1.75
1.72
1.70
2.11
2.03
1.96
1.90
1.86
3.14
2.76
2.51
2.34
2.21
62.79
9.47
5.15
3.79
60
1.50
1.42
1.35
1.26
1.17
1.56
1.54
1.53
1.52
1.51
1.64
1.62
1.60
1.59
1.57
1.79
1.75
1.72
1.69
1.67
2.08
2.00
1.93
1.88
1.83
3.12
2.74
2.49
2.32
2.18
63.06
9.48
5.14
3.78
120
1.46
1.38
1.29
1.19
1.00
1.52
1.50
1.49
1.48
1.47
1.61
1.59
1.57
1.55
1.53
1.76
1.72
1.69
1.66
1.63
2.06
1.97
1.90
1.85
1.80
3.10
2.72
2.47
2.29
2.16
63.33
9.49
5.13
3.76

173
3.85
1.92
1.56
1.42
1.34
1.29
1.25
1.23
1.21
1.19
1.18
1.17
1.17
1.16
1.15
1.15
1.14
1.14
1.14
1.13
1.13
1.13
1.12
1.12
1.12
1.12
1.12
1.12
1.11
1.11
1.10
1.09
1.08
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
40
60
120
N2/N1 1
1.24
1.23
1.22
1.26
1.26
1.26
1.26
1.25
1.28
1.27
1.27
1.27
1.26
1.30
1.29
1.29
1.28
1.28
1.35
1.33
1.32
1.31
1.31
1.48
1.44
1.40
1.38
1.36
5.06
2.33
1.85
1.65
1.55
1.26
1.25
1.24
1.29
1.28
1.28
1.28
1.28
1.30
1.30
1.29
1.29
1.29
1.33
1.32
1.32
1.31
1.31
1.38
1.37
1.35
1.34
1.34
1.53
1.48
1.45
1.42
1.40
5.56
2.48
1.94
1.72
1.60
1.26
1.25
1.23
1.29
1.29
1.28
1.28
1.28
1.31
1.30
1.30
1.29
1.29
1.33
1.33
1.32
1.32
1.31
1.39
1.37
1.36
1.35
1.34
1.55
1.50
1.46
1.43
1.41
5.83
2.56
1.98
1.75
1.63
1.26
1.24
1.23
1.29
1.28
1.28
1.28
1.28
1.30
1.30
1.30
1.29
1.29
1.33
1.33
1.32
1.31
1.31
1.39
1.38
1.36
1.35
1.34
1.56
1.51
1.47
1.44
1.41
6.00
2.61
2.01
1.77
1.64
1.25
1.24
1.22
1.28
1.28
1.28
1.27
1.27
1.30
1.30
1.29
1.29
1.28
1.33
1.32
1.32
1.31
1.31
1.39
1.38
1.36
1.35
1.34
1.57
1.51
1.47
1.44
1.41
6.12
2.64
2.03
1.78
1.65
1.25
1.23
1.21
1.28
1.27
1.27
1.27
1.27
1.30
1.29
1.29
1.28
1.28
1.33
1.32
1.31
1.31
1.30
1.39
1.37
1.36
1.35
1.34
1.57
1.51
1.47
1.44
1.41
6.20
2.68
2.04
1.79
1.65
1.24
1.23
1.21
1.27
1.27
1.27
1.26
1.26
1.29
1.29
1.28
1.28
1.28
1.32
1.32
1.31
1.30
1.30
1.39
1.37
1.36
1.34
1.33
1.57
1.51
1.47
1.43
1.41
6.27
2.68
2.05
1.79
1.68
1.24
1.22
1.20
1.27
1.26
1.26
1.26
1.26
1.29
1.28
1.28
1.28
1.27
1.32
1.31
1.31
1.30
1.29
1.39
1.37
1.35
1.34
1.33
1.57
1.51
1.47
1.43
1.41
6.32
2.69
2.06
1.80
1.68
1.23
1.22
1.20
1.26
1.26
1.26
1.26
1.25
1.28
1.28
1.28
1.27
1.27
1.32
1.31
1.30
1.30
1.29
1.38
1.37
1.35
1.34
1.33
1.57
1.51
1.47
1.43
1.41
6.36
2.70
2.06
1.80
1.68
10
1.23
1.21
1.19
1.26
1.25
1.25
1.25
1.25
1.28
1.27
1.27
1.26
1.26
1.31
1.30
1.30
1.29
1.28
1.38
1.36
1.35
1.33
1.32
1.57
1.51
1.47
1.43
1.40
6.42
2.72
2.07
1.80
1.68
12

7
15
1.22
1.20
1.18
1.25
1.25
1.24
1.24
1.24
1.27
1.27
1.26
1.26
1.25
1.31
1.30
1.29
1.28
1.28
1.38
1.36
1.34
1.33
1.32
1.57
1.51
1.46
1.43
1.40
6.48
2.74
2.08
1.81
1.66
20
1.20
1.18
1.16
1.24
1.24
1.23
1.23
1.23
1.26
1.26
1.25
1.25
1.24
1.30
1.29
1.28
1.28
1.27
1.37
1.35
1.34
1.32
1.31
1.57
1.51
1.46
1.42
1.40
6.54
2.75
2.08
1.81
1.68
24
1.20
1.18
1.15
1.23
1.23
1.23
1.22
1.22
1.26
1.25
1.25
1.24
1.24
1.30
1.29
1.28
1.27
1.26
1.37
1.35
1.33
1.32
1.31
1.57
1.51
1.46
1.42
1.39
6.58
2.76
2.09
1.81
1.67
30
1.19
1.17
1.14
1.23
1.22
1.22
1.22
1.21
1.25
1.25
1.24
1.24
1.23
1.29
1.28
1.27
1.27
1.26
1.37
1.35
1.33
1.31
1.30
1.57
1.51
1.46
1.42
1.39
6.61
2.77
2.09
1.82
1.67
40
1.18
1.16
1.13
1.22
1.22
1.21
1.21
1.21
1.25
1.24
1.23
1.23
1.22
1.29
1.28
1.27
1.26
1.25
1.36
1.34
1.33
1.31
1.30
1.57
1.50
1.45
1.42
1.39
6.64
2.78
2.10
1.82
1.67
60
1.17
1.15
1.12
1.21
1.21
1.20
1.20
1.20
1.24
1.23
1.23
1.22
1.22
1.28
1.27
1.26
1.25
1.25
1.36
1.34
1.32
1.31
1.29
1.57
1.50
1.45
1.41
1.38
6.67
2.79
2.10
1.82
1.67
1.16
1.13
1.10
1.20
1.20
1.19
1.19
1.19
1.23
1.22
1.22
1.21
1.21
1.27
1.26
1.25
1.25
1.24
1.35
1.33
1.32
1.30
1.29
1.57
1.50
1.45
1.41
1.38
6.70
2.79
2.10
1.82
1.67
120

174
B.6
B.6.1
b
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
4
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
5
175
Weibull Tables
Factor A1 for the Mean
0
3.3234
2
1.5046
1.2658
1.133
1.0522
1
0.9649
0.9407
0.9236
0.9114
0.9027
0.8966
0.8922
0.8893
0.8874
0.8862
0.8857
0.8856
0.8859
0.8865
0.8873
0.8882
0.8893
0.8905
0.8917
0.893
0.8943
0.8957
0.897
0.8984
0.8997
0.9011
0.9025
0.9038
0.9051
0.9064
0.9077
0.9089
0.9102
0.9114
0.9126
0.9137
0.9149
0.916
0.9171
0.9182
0.01
3.1091
1.9295
1.473
1.2491
1.1231
1.0459
0.9959
0.9621
0.9387
0.9222
0.9104
0.902
0.8961
0.8919
0.8891
0.8872
0.8861
0.8857
0.8856
0.886
0.8866
0.8874
0.8883
0.8894
0.8906
0.8918
0.8931
0.8944
0.8958
0.8972
0.8985
0.8999
0.9012
0.9026
0.9039
0.9052
0.9065
0.9078
0.9091
0.9103
0.9115
0.9127
0.9139
0.915
0.9161
0.9172
0.9183
0.02
2.9213
1.8652
1.4436
1.2332
1.1137
1.0399
0.9919
0.9593
0.9368
0.9208
0.9094
0.9013
0.8956
0.8916
0.8888
0.8871
0.8861
0.8856
0.8857
0.886
0.8866
0.8874
0.8884
0.8895
0.8907
0.8919
0.8932
0.8946
0.8959
0.8973
0.8987
0.9000
0.9014
0.9027
0.9041
0.9054
0.9067
0.9079
0.9092
0.9104
0.9116
0.9128
0.914
0.9151
0.9162
0.9173
0.9184
0.03
2.7557
1.8062
1.4161
1.2183
1.1047
1.0342
0.988
0.9567
0.9349
0.9195
0.9085
0.9007
0.8951
0.8912
0.8886
0.8869
0.886
0.8856
0.8857
0.8861
0.8867
0.8875
0.8885
0.8896
0.8908
0.8921
0.8934
0.8947
0.8961
0.8974
0.8988
0.9002
0.9015
0.9029
0.9042
0.9055
0.9068
0.9081
0.9093
0.9105
0.9117
0.9129
0.9141
0.9152
0.9163
0.9174
0.9185
0.04
2.6091
1.7522
1.3904
1.2041
1.0961
1.0287
0.9843
0.9542
0.9331
0.9182
0.9076
0.9
0.8947
0.8909
0.8884
0.8868
0.8859
0.8856
0.8857
0.8861
0.8868
0.8876
0.8886
0.8897
0.8909
0.8922
0.8935
0.8948
0.8962
0.8976
0.8989
0.9003
0.9016
0.903
0.9043
0.9056
0.9069
0.9082
0.9094
0.9107
0.9119
0.913
0.9142
0.9153
0.9164
0.9175
0.9186
0.05
2.4786
1.7024
1.3663
1.1906
1.088
1.0234
0.9808
0.9517
0.9314
0.917
0.0067
0.8994
0.8942
0.8906
0.8882
0.8867
0.8859
0.8856
0.8857
0.8862
0.8868
0.8877
0.8887
0.8899
0.8911
0.8923
0.8936
0.895
0.8963
0.8977
0.8991
0.9004
0.9018
0.9031
0.9044
0.9058
0.907
0.9083
0.9096
0.9108
0.912
0.9132
0.9143
0.9154
0.9166
0.9176
0.9187
0.06
2.3619
1.6566
1.3437
1.1779
1.0801
1.0183
0.9774
0.9493
0.9297
0.9158
0.9059
0.8988
0.8938
0.8903
0.888
0.8866
0.8858
0.8856
0.8858
0.8862
0.8869
0.8878
0.8888
0.89
0.8912
0.8925
0.8938
0.8951
0.8965
0.8978
0.8992
0.9006
0.9019
0.9033
0.9046
0.9059
0.9072
0.9084
0.9097
0.9109
0.9121
0.9133
0.9144
0.9156
0.9167
0.9177
0.9188
0.07
2.2572
1.6142
1.3224
1.1658
1.0727
1.0135
0.9741
0.947
0.9281
0.9146
0.905
0.8982
0.8934
0.8901
0.8878
0.8865
0.8858
0.8856
0.8858
0.8863
0.887
0.8879
0.8889
0.8901
0.8913
0.8926
0.8939
0.8952
0.8966
0.898
0.8993
0.9007
0.9021
0.9034
0.9047
0.906
0.9073
0.9086
0.9098
0.911
0.9122
0.9134
0.9145
0.9157
0.9168
0.9178
0.9189
0.08
2.1628
1.5749
1.3024
1.1543
1.0655
1.0088
0.9709
0.9448
0.9265
0.9135
9.9043
0.8976
0.893
0.8898
0.8877
0.8864
0.8858
0.8856
0.8858
0.8863
0.8871
0.888
0.8891
0.8902
0.8914
0.8927
0.894
0.8954
0.8967
0.8981
0.8995
0.9008
0.9022
0.9035
0.9048
0.9061
0.9074
0.9087
0.9099
0.9111
0.9123
0.9135
0.9147
0.9158
0.9169
0.918
0.919
0.09
2.0774
1.5384
1.2836
1.1434
1.0587
1.0043
0.9679
0.9427
0.925
0.9125
0.9035
0.8971
0.8926
0.8895
0.8875
0.8863
0.8857
0.8856
0.8859
0.8864
0.8872
0.8881
0.8892
0.8903
0.8916
0.8928
0.8942
0.8955
0.8969
0.8982
0.8996
0.901
0.9023
0.9037
0.905
0.9063
0.9076
0.9088
0.9100
0.9113
0.9125
0.9136
0.9148
0.9159
0.917
0.9181
0.9191
B.6.2
b
0.4
0.5
0.6
0.7
0.8
0.9
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2
2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9
3
3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
4
4.1
4.2
4.3
4.4
4.5
4.6
4.7
4.8
4.9
5
176
Factor A2 for the Standard Deviation

0
10.4382
4.4721
2.6451
1.8512
1.4282
1.1711
1
0.8783
0.7872
0.7164
0.6596
0.6129
0.5737
0.5402
0.5112
0.4858
0.4633
0.4431
0.4249
0.4085
0.3933
0.3797
0.367
0.3552
0.3343
0.3341
0.3246
0.3156
0.3072
0.2993
0.2918
0.2847
0.278
0.2716
0.2656
0.2598
0.2543
0.249
0.244
0.2392
0.2345
0.2301
0.2258
0.2217
0.2178
0.214
0.2103
0.01
9.3768
4.1973
2.5371
1.7973
1.3969
1.151
0.986
0.868
0.7794
0.7102
0.6546
0.6087
0.5701
0.5372
0.5085
0.4834
0.4611
0.4412
0.4232
0.4069
0.392
0.3783
0.3657
0.3541
0.3432
0.3331
0.3236
0.3148
0.3064
0.2985
0.2911
0.284
0.2774
0.271
0.265
0.2592
0.2537
0.2485
0.2435
0.2387
0.2341
0.2297
0.2254
0.2213
0.2174
0.2136
0.2099
0.02
8.4749
3.9507
2.4373
1.7465
1.367
1.1316
0.9725
0.858
0.7717
0.7041
0.6496
0.6046
0.5666
0.5341
0.5059
0.4811
0.459
0.4393
0.4215
0.4054
0.3906
0.377
0.3645
0.3529
0.3422
0.3321
0.3227
0.3139
0.3056
0.2978
0.2904
0.2834
0.2767
0.2704
0.2644
0.2587
0.2532
0.248
0.243
0.2382
0.2336
0.2292
0.225
0.2209
0.217
0.2132
0.2096
0.03
7.7028
3.7285
2.3448
1.6987
1.3386
1.1129
0.9594
0.8483
0.7642
0.6981
0.6447
0.6005
0.5631
0.5311
0.5033
0.4787
0.457
0.4375
0.4198
0.4038
0.3892
0.3757
0.3633
0.3518
0.3411
0.3312
0.3218
0.313
0.3048
0.297
0.2896
0.2827
0.2761
0.2698
0.2638
0.2581
0.2527
0.2475
0.2425
0.2377
0.2332
0.2288
0.2246
0.2205
0.2166
0.2129
0.2092
0.04
7.0373
3.5276
2.2588
1.6535
1.3114
1.095
0.9467
0.8389
0.7569
0.6923
0.6399
0.5965
0.5597
0.5282
0.5007
0.4765
0.4549
0.4356
0.4182
0.4023
0.3878
0.3745
0.3621
0.3507
0.3401
0.3302
0.3209
0.3122
0.304
0.2962
0.2889
0.282
0.2754
0.2692
0.2632
0.2576
0.2522
0.247
0.242
0.2373
0.2327
0.2284
0.2242
0.2201
0.2162
0.2125
0.2089
0.05
6.4601
3.3453
2.1789
1.6108
1.2854
1.0777
0.9344
0.8297
0.7498
0.6866
0.6352
0.5925
0.5564
0.5252
0.4981
0.4742
0.4529
0.4338
0.4165
0.4008
0.3864
0.3732
0.361
0.3496
0.3391
0.3292
0.32
0.3113
0.3032
0.2955
0.2882
0.2813
0.2748
0.2686
0.2627
0.257
0.2516
0.2465
0.2415
0.2368
0.2323
0.2279
0.2238
0.2197
0.2159
0.2121
0.2085
0.06
5.9564
3.1794
2.1043
1.5703
1.2606
1.061
0.9225
0.8208
0.7428
0.681
0.6306
0.5886
0.553
0.5224
0.4956
0.472
0.4509
0.432
0.4149
0.3993
0.385
0.3719
0.3598
0.3485
0.3381
0.3283
0.3191
0.3105
0.3024
0.2948
0.2875
0.2807
0.2742
0.268
0.2621
0.2565
0.2511
0.246
0.2411
0.2364
0.2318
0.2275
0.2233
0.2193
0.2155
0.2118
0.2082
0.07
5.5145
3.028
2.0347
1.532
1.2368
1.045
0.9109
0.8121
0.736
0.6755
0.6261
0.5848
0.5498
0.5195
0.4931
0.4697
0.4489
0.4302
0.4133
0.3978
0.3837
0.3707
0.3586
0.3475
0.3371
0.3273
0.3182
0.3097
0.3016
0.294
0.2868
0.28
0.2735
0.2674
0.2615
0.2559
0.2506
0.2455
0.2406
0.2359
0.2314
0.2271
0.2229
0.2189
0.2151
0.2114
0.2078
0.08
5.1249
2.8894
1.9696
1.4956
1.214
1.0295
0.8997
0.8036
0.7293
0.6701
0.6216
0.5811
0.5465
0.5167
0.4906
0.4676
0.447
0.4284
0.4117
0.3964
0.3823
0.3694
0.3575
0.3464
0.3361
0.3264
0.3174
0.3089
0.3008
9.2933
0.2861
0.2793
0.2729
0.2668
0.2609
0.2554
0.2501
0.245
0.2401
0.2354
0.231
0.2267
0.2225
0.2186
0.2147
0.211
0.2075
0.09
4.7795
2.7622
1.9085
1.461
1.1921
1.0145
0.8888
0.7953
0.7228
0.6648
0.6172
0.5774
0.5434
0.514
0.4882
0.4654
0.445
0.4267
0.4101
0.3949
0.381
0.3682
0.3563
0.3453
0.3351
0.3255
0.3165
0.308
0.3001
0.2925
0.2854
0.2787
0.2723
0.2662
0.2604
0.2548
0.2495
0.2445
0.2396
0.235
0.2305
0.2262
0.2221
0.2182
0.2143
0.2107
0.2071
B.7
B.7.1
177
Rank Tables
Median Rank Table
SAMPLE SIZE (n)

j
1
2
3
4
5
6
7
8
9
10
1
50.000
2
29.289
70.711
3
20.630
50.000
79.370
4
15.910
38.573
61.427
84.090
5
12.945
31.381
50.000
68.619
87.055
6
10.910
26.445
42.141
57.859
73.555
89.090
7
9.428
22.849
36.412
50.000
63.588
77.151
90.572
8
8.300
20.113
32.052
44.015
55.984
67.948
79.887
91.700
9
7.412
17.962
28.624
39.308
50.000
60.691
71.376
82.038
92.587
10
6.697
16.226
25.857
35.510
45.169
54.831
64.490
74.142
83.774
93.303
17
3.995
9.678
15.422
21.178
26.940
32.704
38.469
44.234
50.000
55.766
61.531
67.296
73.060
78.821
84.578
90.322
96.005
18
3.778
9.151
14.581
20.024
25.471
30.921
36.371
41.823
47.274
52.726
58.177
63.629
69.079
74.529
79.976
85.419
90.849
96.222
19
3.582
8.677
13.827
18.988
24.154
29.322
34.491
39.660
44.830
50.000
55.170
60.340
65.509
70.678
75.846
81.011
86.173
91.322
96.418
20
3.406
8.251
13.147
18.055
22.967
27.880
32.795
37.710
42.626
47.542
52.458
57.374
62.289
67.205
72.119
77.033
81.945
86.853
91.749
96.594
SAMPLE SIZE (n)

j
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
11
6.107
14.796
23.578
32.380
41.189
50.000
58.811
67.620
76.421
85.204
93.893
12
5.613
13.598
21.669
29.758
37.853
45.951
54.049
62.147
70.242
78.331
86.402
94.387
13
5.192
12.579
20.045
27.528
35.016
42.508
50.000
57.492
64.984
72.472
79.955
87.421
94.808
14
4.830
11.702
18.647
25.608
32.575
39.544
46.515
53.485
60.456
67.425
74.392
81.353
88.298
95.169
15
4.516
10.940
17.432
23.939
30.452
36.967
43.483
50.000
56.517
63.033
69.548
76.061
82.568
89.060
95.484
16
4.240
10.270
16.365
22.474
28.589
34.705
40.823
46.941
53.059
59.177
65.295
71.411
77.525
83.635
89.730
95.760
B.7.2
178
5% Rank Table
SAMPLE SIZE (n)

j
10
1
2
3
4
5
6
7
8
9
10
5.000
2.532
22.361
1.695
13.535
36.840
1.274
9.761
24.860
47.237
1.021
7.644
18.925
34.259
54.928
0.851
6.285
15.316
27.134
41.820
60.696
0.730
5.337
12.876
22.532
34.126
47.820
65.184
0.639
4.639
11.111
19.290
28.924
40.031
52.932
68.766
0.568
4.102
9.775
16.875
25.137
34.494
45.036
57.086
71.687
0.512
3.677
8.726
15.003
22.244
30.354
39.338
49.310
60.584
74.113
11
12
13
14
15
16
17
18
19
20
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
0.465
3.332
7.882
13.507
19.958
27.125
34.981
43.563
52.991
63.564
76.160
0.426
3.046
7.187
12.285
18.102
24.530
31.524
39.086
47.267
56.189
66.132
77.908
0.394
2.805
6.605
11.267
16.566
22.395
28.705
35.480
42.738
50.535
58.990
68.366
79.418
0.366
2.600
6.110
10.405
15.272
20.607
26.358
32.503
39.041
45.999
53.434
61.461
70.327
80.736
0.341
2.423
5.685
9.666
14.166
19.086
24.373
29.999
35.956
42.256
48.925
56.022
63.656
72.060
81.896
0.320
2.268
5.315
9.025
13.211
17.777
22.669
27.860
33.337
39.101
45.165
51.560
58.343
65.617
73.604
82.925
0.301
2.132
4.990
8.464
12.377
16.636
21.191
26.011
31.083
36.401
41.970
47.808
53.945
60.436
67.381
74.988
83.843
0.285
2.011
4.702
7.969
11.643
15.634
19.895
24.396
29.120
34.060
39.215
44.595
50.217
56.112
62.332
68.974
76.234
84.668
0.270
1.903
4.446
7.529
10.991
14.747
18.750
22.972
27.395
32.009
36.811
41.806
47.003
52.420
58.088
64.057
70.420
77.363
85.413
0.256
1.806
4.217
7.135
10.408
13.955
17.731
21.707
25.865
30.195
34.693
39.358
44.197
49.218
54.442
59.897
65.634
71.738
78.389
86.089
SAMPLE SIZE (n)
B.7.3
179
95% Rank Table
SAMPLE SIZE (n)

j
10
1
2
3
4
5
6
7
8
9
10
95.000
77.639
97.468
63.160
86.465
98.305
52.713
75.139
90.239
98.726
45.072
65.741
81.075
92.356
98.979
39.304
58.180
72.866
84.684
93.715
99.149
34.816
52.070
65.874
77.468
87.124
94.662
99.270
31.234
47.068
59.969
71.076
80.710
88.889
95.361
99.361
28.313
42.914
54.964
65.506
74.863
83.125
90.225
95.898
99.432
25.887
39.416
50.690
60.662
69.646
77.756
84.997
91.274
96.323
99.488
11
12
13
14
15
16
17
18
19
20
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
23.840
36.436
47.009
56.437
65.019
72.875
80.042
86.492
92.118
96.668
99.535
22.092
33.868
43.811
52.733
60.914
68.476
75.470
81.898
87.715
92.813
96.954
99.573
20.582
31.634
41.010
49.465
57.262
64.520
71.295
77.604
83.434
88.733
93.395
97.195
99.606
19.264
29.673
38.539
46.566
54.000
60.928
67.497
73.641
79.393
84.728
89.595
93.890
97.400
99.634
18.104
27.940
36.344
43.978
51.075
57.744
64.043
70.001
75.627
80.913
85.834
90.334
94.315
97.577
99.659
17.075
26.396
34.383
41.657
48.440
54.835
60.899
66.663
72.140
77.331
82.223
86.789
90.975
94.685
97.732
99.680
16.157
25.012
32.619
39.564
46.055
52.192
58.029
63.599
68.917
73.989
78.809
83.364
87.623
91.535
95.010
97.868
99.699
15.332
23.766
31.026
37.668
43.888
49.783
55.404
60.784
65.940
70.880
75.604
80.105
84.366
88.357
92.030
95.297
97.989
99.715
14.587
22.637
29.580
35.943
41.912
47.580
52.997
58.194
63.188
67.991
72.605
77.028
81.250
85.253
89.009
92.471
95.553
98.097
99.730
13.911
21.611
28.262
34.366
40.103
45.558
50.782
55.803
60.641
65.307
69.805
74.135
78.293
82.269
86.045
89.592
92.865
95.783
98.193
99.744
SAMPLE SIZE (n)

Lamb2 PDF

Uploaded by

Copyright:

Available Formats

Lamb2 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lamb2 PDF

Uploaded by

Copyright:

Available Formats

Reliability Engineering1

Leonard R. Lamberson, Ph.D., P.E.

Final Report Documentation . . . . . . . . . . . . . . . . . .

Translation of any Normal Distribution to the Standard Normal . . . . . . . . . . . . . . . . . . . . . . .

B.7.2 5% Rank Table . . . . . . . . . . . . . . . . . . . . . . 175

CHAPTER 1. RELIABILITY ENGINEERING

Figure 1.1: Reliability Functions

Probabilistic Nature of Reliability

CHAPTER 1. RELIABILITY ENGINEERING

With a 5,000 hour reliability of 95%, if we place 1,000 units in use,

Perform Intended Function Satisfactorily

A product fails when it ceases to function in a way that is satisfactory to the

CHAPTER 1. RELIABILITY ENGINEERING

Specied Time Period

CHAPTER 1. RELIABILITY ENGINEERING

Typically identied usage periods are the

Dierent environments promote dierent failure modes and dierent failure

CHAPTER 1. RELIABILITY ENGINEERING

Example 1.1Dening product reliability.

CHAPTER 1. RELIABILITY ENGINEERING

The reliability number attached to a product changes with

which is also equal to

Indicators Used to Quantify Product Reliability

Several metrics are in common use to indicate product reliability. Some of

CHAPTER 1. RELIABILITY ENGINEERING

Reliability and Quality

The terms reliability and quality are frequently used interchangeable by

CHAPTER 1. RELIABILITY ENGINEERING

The ultimate goal of a product is to satisfy a customer from all aspects of

CHAPTER 1. RELIABILITY ENGINEERING

Spare Parts Availability

Figure 1.2: Customer Satisfaction Model

CHAPTER 1. RELIABILITY ENGINEERING

VonAlven, W.H. Editor; Reliability Engineering, Prentice-Hall, Inc., Englewood Clis,

CHAPTER 1. RELIABILITY ENGINEERING

Achieved availability includes all diagnostic, repair, administrative and

Product Life and Failure Rate

CHAPTER 1. RELIABILITY ENGINEERING

Active repair time

Wait for maintenance

Wait for spare parts

Figure 1.3: System Status Over Time

CHAPTER 1. RELIABILITY ENGINEERING

Failure rate - h(t)

Life time (t)

Figure 1.4: Bath-tub Failure Rate Curve

Infant Mortality Period

CHAPTER 1. RELIABILITY ENGINEERING

t2 = Start of wear- out

Figure 1.5: Stress-Strength Prole Over Time

Useful Life Period

CHAPTER 1. RELIABILITY ENGINEERING

Figure 1.6: Stress-strength Interference Model

CHAPTER 1. RELIABILITY ENGINEERING

Product Design and Development Cycle

Product Development Cycle

CHAPTER 1. RELIABILITY ENGINEERING

CHAPTER 1. RELIABILITY ENGINEERING

Past eld problems provide information useful as a basis for