Progress: 3%

FACULTY OF ENGINEERING
CAIRO UNIVERSITY

Reliability and Maintenance Engineering

Week#2: Statistics review
Ahmed Sakr, PhD, eng.
This presentation is adopted from Prof.Soumaya Yacout’s Fault analysis
and Maintenance course
Department of Mathematics and Industrial Engineering
Polytechnique Montreal

2024-02-01 Reliability and Maintenance Engineering 1

Progress: 6%

Contents
1. Introduction
2. Probability density laws
3. Reliability and failure

2024-02-01 Reliability and Maintenance Engineering 2

Progress: 9%

1.Introduction
• Sampling is the basis of statistics and probability.
• τ is a sample of failure time, among all possible times
• ℑ (total population), and 𝑓(𝑡) is a function that assigns a real number (t) to
each result 𝑇 ∈ τ, so 𝑇 is a random variable.
• Example : a sample τ of n=5 engine failure times: 𝑡i= 1100, 1200, 1150, 5000,
4600, 3718 hours, i=1,…5.
• Using a statistical function 𝑓(t), we can Calculate P, the probability
that T is > t=1355 hours Sample
τ

Total population
ℑ
2024-02-01 Reliability and Maintenance Engineering 3
Progress: 12%

1.Introduction
• The goal is to find a model that generalizes the failure times observed in the sample τ and
that applies to the entire population ℑ.

• By convention, the small 𝑡 letter associated with a random variable 𝑇, will be used to denote a
particular value of this variable.
• In the context of maintenance, T represents the time of failure.

• To be able to predict the time of the next failure, one looks for a function 𝑓𝑇 (𝑡) which assigns
a real number to each time 𝑡 ∈ 𝜏.

2024-02-01 Reliability and Maintenance Engineering 4

Progress: 15%

1.Introduction

Random extraction

Sample τ Total Population ℑ

Generalization model 𝑓𝑇 (𝑡)

2024-02-01 Reliability and Maintenance Engineering 5

Progress: 18%

Question 1
• A satellite is considered to fail if its contact with Earth is lost. During three
years of operation, this event is observed 10 times, at the age of 18569, 20035,
21956, 22456, 22987, 23103, 23412, 23670, 24770 & 25150 hours.

1) The cardinality (size) of the sample space 𝜏 is …

2) 𝑇 is the random variable that represents ….
3) At some time t=34134, The two possible events of T are …

2024-02-01 Reliability and Maintenance Engineering 6

Progress: 21%

1.Introduction.Probability density function

Function obtained by observation of equipment failure in our sample.
𝑑
𝑓𝑇 𝑡 = 𝐹𝑇 (𝑡)
𝑑𝑡
To ease the writing we omit writing T in 𝑓𝑇 𝑡 .
As a generalization, the equipment is considered to follow a probability function
normal, exponential, Weibull …

∞
Properties to consider : 𝑓𝑇 (𝑡) ≥ 0 , ‫׬‬−∞ 𝑓𝑇 𝑡 = 1

2024-02-01 Reliability and Maintenance Engineering 7

Progress: 24%

1.Introduction.Cumulative distribution
function
• The cumulative distribution function (also called the distribution function) of the 𝑇 variable at a moment t,
is the probability that this variable will become a value of less than or equal to t.
• Therefore, the probability that the failure will occur at or before a time t can be represented by :
𝐹 𝑡 = 𝑃(𝑇 ≤ 𝑡)
• This probability corresponds to the area below the curve of the less infinite probability density function at a
certain time t. That's why it can also be represented as a :
𝑡
𝐹 𝑡 = න 𝑓 𝑡 𝑑𝑡
0

And from there, we get that for both points :

𝑡2
𝑃 𝑡1 ≤ 𝑇 ≤ 𝑡2 = න 𝑓 𝑡 𝑑𝑡
𝑡1

2024-02-01 Reliability and Maintenance Engineering 8

Progress: 26%

1.Introduction. Cumulative distribution

function
𝑎 = 𝑡1
𝑏 = 𝑡2

2024-02-01 Reliability and Maintenance Engineering 9

Progress: 29%

1.Introduction. Formulas
• Mathematical (or average) 𝑇 expectation:
∞
𝐸 𝑇 = න 𝑡𝑓 𝑡 𝑑𝑡
0

• Variance of 𝑇 (square deviation from 𝑇 value relative to its average):

∞
𝑉𝑎𝑟 𝑇 = 𝜎 2 = න 𝑡 2 𝑓 𝑡 𝑑𝑡 − [𝐸 𝑇 ]2 = 𝐸 [𝑇 − 𝐸 𝑇 ]2 = 𝐸 𝑇 2 − [𝐸 𝑇 ]2
0

(see Wikipedia for the demonstration)

𝑉𝑎𝑟 (𝑇) estimated by 𝑠 𝑇

2024-02-01 Reliability and Maintenance Engineering 10

Progress: 32%

2. Density laws. Exponential law

1 − 1 𝑡
𝑓 𝑡 = 𝑒 𝜃 ,𝑡 ≥0
𝜃

1
−(𝜃)𝑡
F 𝑡 =1−𝑒 ,𝑡 ≥0

Properties : 𝐸 𝑇 = 𝜃 , 𝑉𝑎𝑟 𝑇 = 𝜃 2
Scale Parameter: 𝜃 or 𝜆

2024-02-01 Reliability and Maintenance Engineering 11

Progress: 35%

2. Density laws. Exponential law

𝑓 𝑡 F 𝑡

X≡ t

2024-02-01 Reliability and Maintenance Engineering 12

Progress: 38%

2. Density laws. Weibull's Law (Two parameters)

𝑡 𝛽 𝑡 𝛽
𝛽 𝑡 − 𝛽𝑡 𝛽−1 −
• f 𝑡 = ( )𝛽−1 𝑒 𝜃 = 𝑒 𝜃 ,𝑡 ≥0
𝜃 𝜃 𝜃𝛽

𝑡
−(𝜃)𝛽
• F 𝑡 =1−𝑒 ,𝑡 ≥ 0

1 2 1
𝐸 𝑇 = 𝜃Γ(1 + ), variance = 𝑉𝑎𝑟(𝑇) = 𝜃2 Γ(1 + ) − (Γ(1 + )2 )
𝛽 𝛽 𝛽

Scale parameter: 𝜃 > 0(lambda) Shape parameter: 𝛽 > 0 (k)

∞
* 𝑔𝑎𝑚𝑚𝑎 function : Γ 𝑛 = ‫׬‬0 𝑒 −𝑥 𝑥 𝑛−1 𝑑𝑥 = . . . = 𝑛 − 1 !

2024-02-01 Reliability and Maintenance Engineering 13

Progress: 41%

2. Density laws. Weibull's Law (Two parameters)

𝑓 𝑡 𝐹 𝑡

Θ=0.5 β=2
Θ=1.0 β=2
Θ=1.5 β=3
Θ=3.0 β=4

Θ=0.5 β=2
Θ=1.0 β=2
Θ=1.5 β=3
Θ=3.0 β=4

If β=1 we get the special case of exponential function

2024-02-01 Reliability and Maintenance Engineering 14

Progress: 44%

2. Density laws. Weibull's Law (Three Parameters)

𝛽 𝛽
𝛽 𝑡−𝑡0 𝛽−1 − 𝑡−𝑡0 𝛽(𝑡−𝑡0 )𝛽−1 − 𝑡−𝑡0
•𝑓 𝑡 = ( ) 𝑒 𝜃 = 𝑒 𝜃 , 𝑡≥0
𝜃 𝜃 𝜃𝛽

𝑡−𝑡0 𝛽
−
• 𝐹 𝑡 =1−𝑒 𝜃 ,𝑡 ≥ 0
1 2 1
• 𝐸 𝑇 = 𝑡0 + 𝜃Γ(1 + ) , 𝑉𝑎𝑟𝑖𝑎𝑛𝑐𝑒 = 𝑉𝑎𝑟(𝑇) = 𝜃 2 2Γ(1 + ) − (Γ(1 + ))2
𝛽 𝛽 𝛽
Scale parameter: 𝜃 > 0 , Shape parameter: β > 0 , Position parameter: 𝑡0
∞
* 𝑔𝑎𝑚𝑚𝑎 function : Γ 𝑛 = ‫׬‬0 𝑒 −𝑥 𝑥 𝑛−1 𝑑𝑥 = . . . = 𝑛 − 1 !

2024-02-01 Reliability and Maintenance Engineering 15

Progress: 47%

2. Density laws. Weibull's Law (Three Parameters)

Θ=2.0
β=5.0 Θ=2.0
t_0=5 β=5.0
t_0=10

2024-02-01 Reliability and Maintenance Engineering 16

Progress: 50%

2. Density laws. Weibull's Law (Three Parameters)

Discover Weibull distribution online!

2024-02-01 Reliability and Maintenance Engineering 17

Progress: 53%

2. Density laws. Normal law

1 (𝑡−𝜇)2
1 −( )
f 𝑡 = 𝑒 2 𝜎2 , −∞ ≤ 𝑡 ≤ ∞
𝜎 2𝜋

1 (𝑡−𝜇) 2
1 𝑡 −(2) 2
𝐹 𝑡 = ‫׬‬−∞
𝑒 𝜎 𝑑𝑡, −∞ ≤ 𝑡 ≤ ∞
𝜎 2𝜋

𝐸 𝑇 =𝜇, 𝑉𝑎𝑟 𝑇 = 𝜎 2 > 0

Note: The normal law represents a random variable that could take a value that varies from
− ∞ to ∞, But the time of failure is only positive

2024-02-01 18
Reliability and Maintenance Engineering
Progress: 56%

2. Density laws. Normal law

𝑓𝑋 𝑡 𝐹𝑋 𝑡

2024-02-01 Reliability and Maintenance Engineering 19

Progress: 59%

3. Reliability and failure. Reliability

Reliability is a characteristic of a system expressed by the likelihood that it will
perform the function for which it was designed, under given conditions and for a
given period of time.
In other words, it is the probability of survival of a system.

• Reliability = Probability (𝑇 > 𝑡𝑟 )

• 𝑇: Running time, random variable

• 𝑡𝑟 : a certain duration, for example 1085 hours

2024-02-01 Reliability and Maintenance Engineering 20

Progress: 62%

3.Reliability and failure.Three Milestone Failures

• Potential failure
• Start of a degradation that leads to failure.

• Functional failure
• Discontinuing the ability of equipment to perform a required function.

• Critical or catastrophic failure

• Failure that could cause personal injury, significant property damage or lead to other
consequences deemed unacceptable.

2024-02-01 Reliability and Maintenance Engineering 21

Progress: 65%

3.Reliability and failure.P-F Interval (conceptual)

2024-02-01 Reliability and Maintenance Engineering 22

Progress: 68%

3.Reliability and failure.P-F Interval

• Maintenance intervals must be smaller than the P-F interval, so that the repair
or maintenance action can be performed before the failure occurs.
• With better techniques, degradation can be detected in advance before failure.
• It doesn't work for all failure types!
• There are sudden failures, or the P-F interval is virtually non-existent.
• For sudden failures, see the actions described later in the course, in the RCM.

2024-02-01 Reliability and Maintenance Engineering 23

Progress: 71%

3.Reliability and failure.Mathematical definition

• 𝐹 𝑡 is the cumulative distribution function → F 𝑡 =𝑃 𝑇≤𝑡

• 𝑅 𝑡 is the survival function → 𝑅 𝑡 =𝑃 𝑇>𝑡

𝑭 𝒕 +𝑹 𝒕 =𝟏
𝑡
𝑅 𝑡 = 1 − 𝐹 𝑡 = 1 − න 𝑓 𝑡 𝑑𝑡
0

2024-02-01 Reliability and Maintenance Engineering 24

Progress: 74%

Question 2
1 −0.2𝑡
• For 𝑓 𝑡 = 𝑒
5

1) What is the probability of failure during the first 10 hours?

2) What is the probability of survival during the first 10 hours?
3) What is the probability of failure in the interval (10, 20)?
4) What is the probability of survival in the interval (10, 20)?

2024-02-01 Reliability and Maintenance Engineering 25

Progress: 76%

Question 3
• A printer's failure time T can be represented by a Weibull distribution with
parameters : 𝜃 = 40, 𝛽 = 2
1) What is the probability of failure in the interval (0, 40)?
2) What is the probability of survival in this interval?

2024-02-01 Reliability and Maintenance Engineering 26

Progress: 79%

3.Reliability and failure.Conditional probability

• Conditional reliability is useful in describing the reliability of a component following a burn-
in period or after a warranty period.
• A burn-in period refers to a practice of operating a system or component at elevated stress
levels initially to identify and eliminate potential failures. The goal is to improve the
reliability of the system by addressing potential issues early on.
𝑃[𝐴 ∩ 𝐵]
𝑃 𝐴|𝐵 = , PB >0
𝑃[𝐵]
• Events A and B are mutually exclusive (cannot happen at the same time) if :
0
𝑃 𝐴 ∩ 𝐵 = 0 , so: 𝑃 𝐴|𝐵 = =0
𝑃[𝐵]

• Events A and B are independent if :

𝑃 𝐴 𝑃[𝐵]
• 𝑃 𝐴 ∩ 𝐵 = 𝑃 𝐴 𝑃 𝐵 , so: 𝑃 𝐴|𝐵 = = 𝑃[𝐴]
𝑃[𝐵]

2024-02-01 Reliability and Maintenance Engineering 27

Progress: 85%

3.Reliability and failure. Conditional reliability

𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑡,𝑡+∆𝑡

• 𝑃
𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0,𝑡

→ P 𝐴|𝐵
𝑃[𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑡 + ∆𝑡 ∩ 𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0, 𝑡 ]
= =
𝑃[𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0, 𝑡 ]

𝑃[𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑡+∆𝑡 ] 𝑅 𝑡+∆𝑡

= =
𝑃[𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0,𝑡 ] 𝑅(𝑡)
2024-02-01 Reliability and Maintenance Engineering 28
Progress: 88%

3.Reliability and failure. Conditional failure

𝑓𝑎𝑖𝑙𝑢𝑟𝑒 𝑡,𝑡+∆𝑡
• 𝑃
𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0,𝑡

𝐴 𝑃[𝐹𝑎𝑖𝑙𝑢𝑟𝑒 𝑡+∆𝑡 ∩ 𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0,𝑡 ] 𝐹 𝑡+∆𝑡 −𝐹(𝑡)

• →P = =
𝐵 𝑃[𝑁𝑜𝑛 𝑓𝑎𝑖𝑙𝑢𝑟𝑒 0,𝑡 ] 𝑅(𝑡)

2024-02-01 Reliability and Maintenance Engineering 29

Progress: 91%

Question 4
• What is the probability of failure of a printer between the ages of 10 and 13? It
is assumed that it follows an exponential distribution with 𝜃 = 2 and:

1) We are in his first year of life (year 0)

2) We use it for 10 years (we are in year 10)

2024-02-01 Reliability and Maintenance Engineering 30

Progress: 94%

Question 5
• What is the probability of Survival of a printer between the ages of 10 and 13?
It is assumed that it follows a normal distribution with 𝜇 = 2 𝑎𝑛𝑑 𝜎 2 = 5:

1) We are in his first year of life (year 0)

2) We use it for 10 years (we are in year 10)

2024-02-01 Reliability and Maintenance Engineering 31

Progress: 97%

3.Reliability and failure. Bathtub curve

2024-02-01 Reliability and Maintenance Engineering 32

Progress: 100%

3.Reliability and failure. Bathtub curve

• Zone 1 : early life, which sees early breakdowns most often linked to an original
manufacturing defect, failures that are normally eliminated by a run-in before
going into production (wear-in).
• Zone 2 : the living environment, or useful life, which has few breakdowns,
constant rates, whose appearance remains unpredictable, and for which there
are therefore no possible preventive measures.
• Zone 3 : the end of life, which sees an increase in the number of (wear-out)
failures, and whose appearance can be avoided by the preventive replacement
of worn parts with preventive maintenance.

2024-02-01 Reliability and Maintenance Engineering 33