Industrial Organisation Class Notes Uottawa MCG 5171B

EMP5103 Lecture 1
Thursday, January 10, 2013
EMP 5103 Reliability, Quality and Safety Engineering

Dhillon, B, Reliability, Quality and Safety for Engineers, CRC Press, Boca Raton, Florida, 2004
Marks
Report
Mid Term
Final Exam
30%
30%
40%
Mar 28
Feb 28
Apr 4
Report Format
1.
2.
3.
4.
Title
Summary, it is not an introduction to the report. What is the main highlight of the
document? What is contained in the report?
Introduction familiarize reader on topic sections.

Main Body 2.1, 2.2, etc
Conclusions - future research and studies.
References
Books
1. Gorman, D, Human Reliability, Program Press, New York, 1986.
Journal Articles
2. Williams, R, Engineering Design, Journal of Mechanical Engineering, Vol.10,
1982, pp 10-20.
Conference Proceedings
3. Reiche, H., Reliability Management, Proceedings of the American Society for
Quality Control Annual Conference, 1982, pp 10-30.
Report
4. Gorman, S., Technical Management, Report No.1801, 1987, Available from
the Department of Mechanical Engineering, University of Ottawa, Ontario,
Canada.
EMP5103 Lecture 1
Course Outline
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
Reliability background and introductory mathematics.

Failure data collection and Sources
Static Reliability Evaluation Models
Series system
Parallel system
K-out-n system,
series-parallel system,
parallel-series system
Dynamic Reliability Evaluation Models
Exponential distribution
Weibull distribution
Series, Parallel.
K-out-n,
series-parallel,
parallel-series,
Standby
Redundancy- Related Topics
Reliability Testing
Reliability and Maintainability Management
Safety Management
Robot Reliability and Safety
Quality Management
Quality Control Charts
1. Reliability Background and Introductory mathematics

Reliability Definition
Reliability is the probability that an item will perform its function adequately for the desired period
of time when operated according to specified conditions.
History of Reliability
Power system problems (1930s)

World War II (V1 and V2 rockets)
U.S. Department of Defense (1945-1950)
Advisory Group on the Reliability of Electronic Equipment (AGREE)1950
AGREE report (1957) ( it took AGREE 7 years to report on the reliability)
EMP5103 Lecture 1
History of Quality Control
C.N. Frazeo of Telephone Laboratories made use of statistical methods in inspection

problems1916
Walter A. Shewhart developed quality control charts1924
Etc
History of Safety
Pling, the Elder (AD 23-79) - Historia Naturalis wearing of protective masks by workers to
stop the inhalation of toxic substances
Miners Safety Lamp-Humphrey Davy-18th century
American Public Health Association1885 - Occupational Health and Safety
1st Text book-Industrial Safety by H.W. Heinrich1931
Publications
Reliability Publications
IEEE Transactions on Reliability (1953)

Microelectronics and Reliability (1962)
Reliability Engineering and System Safety
International Journal of Quality and Reliability Management
Reliability Reviews
Annual Reliability and Maintainability Symposium (1954)
Safety Publications
Journal of Safety Research (USA)

National Safety Needs (USA)
System Safety Conferences
Quality Control Publications

American Society for Quality Control (ASQC) Journals and Conferences
Design Reliability
Design specification (MTBF-mean time between failure ,MTTR-mean time to repair, Etc)
Reliability allocation
2000 hrs
1000 hrs
1000 hrs
Allocate top reliability of Arm to subsystems right down to components level.

3
EMP5103 Lecture 1
Data collection and Analysis writing information to databank and retrieving it

FMEA-Failure Mode and Effect Analysis and FTA- Fault Tree Analysis (methodologies)
Reliability growth build a prototype and test it
Reliability demonstration takes place at customer facility, performed at the system level
and is typically set up as a success test
Reliability warranty determine what percentage of the failure population can be covered
financially and estimating the time at which this portion of the population will fail
Failure data feedback citing the causes/reasons for failure occurrences
Reliability Areas and Applications:

1.
2.
3.
4.
5.
6.
7.
8.
Reliability general
Mechanical reliability
Software reliability
Human reliability
Reliability optimization
Reliability growth modeling (see graph above)
Power system reliability
Life cycle costing (LCC)
LCC = AC + OC
with AC = Acquisition Cost and OC = Ownership Cost
9. Failure data collection and analysis

10. Maintainability and maintenance
11. Structural reliability
12. Robot reliability and safety
13. Etc
Safety Areas and Applications:
EMP5103 Lecture 1
14. System safety

15. Safety management
16. Accidents
17. Etc
Quality Control Areas and Applications:
18. Statistical quality control
19. Food, textile, medical areas (health care), farm, etc
Mathematics
Probability Properties:
0 P( X ) 1
P( X 1 X 2 X 3 ... X n ) P ( X 1 ) P ( X 2 ) ... P ( X n ) (mutually exclusive events)
P( X 1 X 2 X 3 ... X n ) 1 (1 P( X i )) (independent events)
i 1
If n 2 : P ( X 1 X 2 ) P( X 1 ) P ( X 2 ) P( X 1 ) P( X 2 ) (independent events)
P(S ) 1
S = sample space
P( S ) 0
P( X 1 X 2 X 3 ... X n ) P ( X 1 ) P ( X 2 )...P ( X n ) (independent events)
Reliability Related Formulas:

Cumulative Distribution Function:
t
F (t )
f ( x ) dx
f(t):
F(t):
failure density function

cumulative distribution function
Failure density function:

f (t )
dF (t )
dt
Total Area:
F () 1
Reliability:
R F 1
R is reliability and F is Failure
EMP5103 Lecture 1

t
R (t ) 1 F (t ) 1 f ( x )dx f ( x) dx
t
f ( x)dx f ( x)dx 1
f ( x) dx 1 f ( x )dx
0
Hazard rate:
(t )
f (t )
f (t )
R (t ) 1 F (t )
Failure Time Distributions

Exponential Distribution:
f (t ) e t t: time
: Constant failure rate
F (t ) ( )e t dt 1 e t
(t)
R (t ) 1 F (t ) e t
(t )
f (t ) e t
t
R (t )
e
Example:
0.002 failures/year
If t 10
F (t ) 1 e t 1 e 0.002*10 0.0198
R (t ) e 0.002*10 0.9802
Weibull Distribution:
f (t )
(t )b1e
b
(t )
for t
and b,, > 0
b, , and are shape, scale, and location parameters, respectively
6
=0
=1
EMP5103 Lecture 1
F (t ) 1 e
(t )
b = 1: exponential distribution
b = 2: Reyleigh distribution
b
f (t )
(t )
R (t )
(t ) b 1 e
b 2; (t )
(t ) b 1
(t)
t b 1
b 1; (t )
b
(t )
For = 0:
b
(t )
(t )
b=1
1
t
t
(t)
See page 16a for function recapitulation.
EMP5103 Lecture 1
Bathtub Hazard Rate Curve:
(t)
Burn-in period
Useful life period
Wear out period
t
(t ) kct c1 (1 k )bt b1 e t
For b,c,, > 0

, and 0 k 1 t 0
And c = 0.5 and b = 1 to get the shape above for the bathtub
b= 0.5 : bathtub curve
b=1 : Extreme value distribution
b,c = shape parameters
, = scale parameters
t = time
Mean Time To Failure (MTTF):
E (t ) MTTF
tf (t )dt
0
MTTF
R(t )dt
0
MTTF lim R ( s )
s 0
Exponential Distribution:
EMP5103 Lecture 1
f (t ) e t
E (t ) MTTF te t dt
0
F (t ) 1 e
R (t ) e t
MTTF e t dt
0
1
R( s)
s
MTTF lim
s 0
1
1
2. Failure Data Collection and Sources

2.1 Purposes of Collecting Failure Data
1. To calculate hazard rate (failure rate) of an item.
2. To make decisions regarding the introduction of a redundant item.
3. To perform trade-off studies (cost-reliability studies).
4. To conduct the items replacement studies.
5. To perform preventive maintenance studies of the item.
6. To conduct effective design reviews.
7. To determine the maintenance needs of a new item.
8. To perform life cycle cost studies.
9. To predict the reliability and the availability of the item.
10. To recommend design changes for improving the items reliability.
11. Etc
2.2 Data Collection Sources
1.
2.
3.
4.
5.
6.
7.
8.
Past experience with similar or identical material.

Inspection data produced by quality control and manufacturing groups.
Failure reporting system developed and used by customers.
Reports generated by the repair facility.
Data obtained during the breadboard or the development of the item.
Tests: field demonstration, environmental qualification, field installation.
Factory acceptance testing of equipment, modules, and assemblies.
Warranty claims:
Warranty Claims
Product failure
related data
General data
(product related)
9
EMP5103 Lecture 1
General Data:
o Serial number and identification of model
o Starting date of the warranty
o Date of production
o Name of the producing plant
Product Failure Related Data:
o Failed part
o Failure: date, type, and severity
o Cost of repair
o Warranty coverage type
o Age of the product at failure
o Etc
Government Industry Data Exchange Program (GIDEP)
Typical Deficiencies in Failure Data Sources:
1.
2.
3.
4.
5.
6.
7.
8.
Non-standardized data reporting.

Poor description of the item in question.
Time-consuming data feedback.
Lack of accuracy in repair times required for many parts.
Inconsistent and vague definitions and terms on reporting forms.
Difficulty in pinpointing the cause of failure.
Government security of company proprietary classification of the data.
Etc
Failure Data Forms:

1. Description of the item in question.
2. Location of the hardware.
3. Form number.
4. Serial number of the failed part.
5. Name of the manufacturer.
6. Failure description.
7. Date the failure occurred.
8. Date the failure was detected.
9. Date the failure was corrected.
10. Form completion date.
11. Serial number of the replaced part.
12. Manufacturer of the replaced part.
13. Time taken for repair.
14. Operating hours from previous failure.
15. Cost of repair.
16. Name of the repairperson.
17. Signatures.
18. Etc
See pp. 22 and 23.
10
EMP5103 Lecture 1
MIL-HDBK-217: very important document that includes mathematical formulas

(failure rates). Best source for reliability (pp 23).
Data Sources Information Documents:
1. Dhillon, B., Mechanical Reliability, American Institute of Aeronautics and Astronautics,
Washington, D.C., 1988. Lists over 55 data sources.
2. Dhillon, B., Human Reliability with Human Factors, Pergamon Press, New York, 1986.
Lists 24 human reliability related data sources.
3. Dhillon, B., Robot Reliability and Safety, Springer-Verlag, New York, 1991. List over 55
data sources.
4. Dhillon, B., Life Cycle Costing, Gordon and Breach Science Publisher. Lists over 20 data
sources.
See pp. 28.
MIL-HDBK-217:
Example:
Failure rate evaluation of an electronic component
ea
b C exp
kt
b: base failure rate
t: Absolute temperature
k: Boltzmanns constant
ea: activation energy
C: constant
b Fq Fe ...
Fq: the factor which takes into consideration the part quality level
Fe: the factor which takes into consideration the influence of environment
c Constant failure rate
Example:
Tunnel diode
c b Fe Fq failures/106 hours
Fq (quality level - Jan) = 5
note Jan is the unit
Fe (environment ground, Benign) = 1
note Benign is the unit
b (base failure rate) = 0.044 failures/106 hours
c 0.044 1 5 0.22 failures/106 hours
Note: failure rate is exponentially distributed.
11
EMP5103 Lecture 1
Failure Rate Evaluation of Equipment:

k
t q j ( g Fq ) j
j 1
k = the number of different generic part classifications

qj = the jth generic part quantity
Fq = the jth generic part quality factor
g = the constant failure rate of the jth component
3. Static Reliability Evaluation Models:

Series Configuration:
E1
E2
E3
Ek
Ei: Successful event i
P ( E1 E 2 E 3 ...E k ) P ( E1 ) P ( E 2 )...P ( E k )
Let : Ri P ( E i )
k
RS R1 R2 R3 ...Rk Ri
i 1
Example:
R1=0.8
R2=0.9
RS 0.8 0.9 0.72
IFF Ri 0.95
k
RS 1 Fi
(non-identical units)
i 1
Where Fi is the failure probability of unit i.

For identical units:
RS 1 kF
12
EMP5103 Lecture 1
Parallel Configuration:
E1
1
E1
2
E1
3
E1
Fp P ( E1 E 2 E3 ...E ) P ( E1 ) P ( E 2 ) P ( E3 )...P ( E )
Let : Fi P ( Ei )
Fp Fi
(for non-identical units)
i 1
R p 1 Fi
i 1
Rp 1 F
Example:
A parallel system is composed of four independent and active units. The unvreliability of units
1,2,3 and 4 is 0.4, 0.3, 0.2 and 0.1 respectively. Calculate the system reliability.
13
EMP5103 Lecture 1
0.4
0.3
0.2
0.1
R p 1 F1 F2 F3 F4 1 0.4 0.3 0.2 0.1 0.9976

Rp 1 F
As for identical units
1
0.95
Rp
=3
0.90
=2
0.85
=1
0.1
0.2
0.3
14
0.4
EMP5103 Lecture 1

not mentioned
1
0.90
Rp
=4
0.80
=2
0.70
=3
=1
0.5
0.6
0.7
0.8
0.9
F
K-out-of- -unit System:
15
EMP5103 Lecture 1

1 R i R i
i k i
Rk
where

!

i i!( i )!
n
(R
i 1
(identical units)
Fi ) 1
R F 3 R 3 3R 2 F 3RF 2 F 3 1
R2 R 3 3R 2 (1 R ).....F 1 R
3
=3
Parallel System
0.80
System
Reliability
0.60
2-ou-of-3 System
0.40
0.20
Series System
0.2
0.4
0.6
Unit Reliability
RS R 3 ( Series _ system)
R p 1 (1 R 3 )( Parallel _ system)
Rk R 2
( R F ) 3 R 3 3R 2 F 3RF 2 F 3
Stop here
16
0.8
EMP5103 Lecture 1
1
0.80
System
Reliability
0.60
R = 0.9
0.40
R = 0.7
0.20
R = 0.5
0
k (requirement for the system to work)

Example:
A parallel system is composed of three independent and active units. At least two units must be
functioning successfully for system success. The reliability of units 1, 2 and 3 is 0.8, 0.9 and 0.7
respectively. Calculate the systems reliability.
(R1 + F1)(R2 + F2)(R3 + F3) = 1
Ri: is the units reliability for i= 1, 2, 3
Fi: is the units unreliability for i= 1, 2, 3
R1 R2 R3 + R 1 R2 F 3 + R 1 R3 F 2 + R 2 R3 F 1 + R 1 F 2 F 3 + R 2 F 1 F 3 + R 3 F 1 F 2 + F 1 F 2 F 3 = 1
R2/3 = R1 R2 R3 + R1 R2 F3 + R1 R3 F2 + R2 R3 F1 = R1 R2 R3 + R1 R2 (1-R3) + R1 R3 (1-R2) + R2 R3 (1R1) = 0.8*0.9*0.7+0.8*0.7*(1-0.9)+..
Series-Parallel configuration:
17
EMP5103 Lecture 1

1
R pj 1 F ji
Fji: is the jth subsystem, ith units unreliability
i 1
j 1
j 1
i 1
Rsp R pj (1 F ji )
Rsp 1 F r
Example:
A system is composed of two independent, identical and active subsystems. Each subsystem consists
of two identical units in parallel. The units failure probability is 0.2. calculate the systems
reliability.
F = 0.2 and r = 2 and k = 2
R sp 1 F r
1 0. 2 2
0.9216
0.2
0.2
0.2
0.2
18
EMP5103 Lecture 1
r=3
F = 0.1
r=2
0.80
0.60
Rsp
r=3
0.40
0.20
r=2
F = 0.5
r=1
k
1
Rsj R ji
i 1
r
r
k
R ps 1 1 Rsj 1 1 R ji
j 1
j 1
i 1

r
R ps 1 1 R k where R is the units reliability
Example:
19
EMP5103 Lecture 1
k = r = 2, F = 0.2
0.2
0.2
0.2
0.2
R ps 1 1 R k 1 1 0.8 2 0.8704 R ps 1 1 R k
r
r=3
R = 0.9
r=2
0.80
Rps
r=1
0.60
r=3
0.40
0.20
r=2
R = 0.5
r=1
1. Dynamic Reliability Evaluation Models.

1.1.General Reliability Function
t
( x ) dx
R t e 0
t: is time and is units hazard rate or the instantaneous failure rate

t
F (t )
f (t )
1 dR (t )
(t )
R (t )
R (t ) dt
t
f (t ) dt
dF (t )
f (t )
dt
R (t ) F (t ) 1
(t )dt R(t ) dR(t )
R (t ) 1 F (t )
At t=0; R(0) = 1
20
EMP5103 Lecture 1
R (t )
1
dR (t )
R (t )
(t )dt
0
(t ) ln R (t )
0
R (t ) e
( t ) dt
0
Example:
(t )
t
R(t ) e
dt
0
e t
1.2.Mean Time To Failure (MTTF)
MTTF R (t )dt
0
MTTF lim R ( s)
s 0
dont _ have _ to _ remember :

t
MTTF lim R (t ) dt
t
0
Where __ f (t ) R (t ) dt
0
Laplace _ transform :
lim f (t ) lim s f ( s )
t
s 0
R(s)
s
Then, _ MTTF lim R ( s )
but , _ f ( s )
s 0
Example:
21
EMP5103 Lecture 1
(t )
k (t k 1 )
where :
k : shape _ parameter
: scale _ parameter
t : time
t
R (t ) e
( t ) dt
0
MTTF
1
tk
kt
dt
1
tk
dt
k 1
1 t
e dt
Example:
k 1
where :
0.007 failure / hour

t 50hours
R(50) e (0.007 )(50) 0.7047
Series configuration:
1
1
For
1
2
1
3
1
k
,
k
Rs (t ) R (t ) e
i 1
1
t
i
as we have:
i 1
MTTFs Rs (t ) dt
e
i 1
1
t
i
R (t ) e
dt e
i 1
i t
( t ) dt
dt
22
1
1
i 1 i
k
1 dt
0
1
t
1
EMP5103 Lecture 1
1
1
t for ( t ) 0.05
i 1 i
i
Hazard rate:
f (t )
1 dRs (t )
s (t ) s
Rs (t )
Rs (t ) dt
1
let : i
i
k
Rs (t ) 1
s (t )
i
i 1
it
(e
it
i 1
i 1
s i
i 1
1
k
R (t ) e
1
t
i
i 1
1
1
t 0.05
t : true _ for :
i 1 i
i
k
Rs (t ) 1
Series System Non-Constant Failure Rate:
j (t ) j j t
23
EMP5103 Lecture 1
R (t ) e
( t ) dt
0
R j (t ) e
j t 1
jt
1
Rs (t ) e
j t 1
jt
1
i 1
Rs (t ) exp

k

1
t
j
k

j 1
jt
1
j 1

tot 1
Rs (t ) exp
1 tot tot

let :
tot i
j 1
k
tot i
j 1
ttot : gamma
Rs (t ) exp exp ttot e
it
j 1
Mean Time To Failure for Various Simple Unit and Series Systems:
System Structure
MTTF
1
Z(t)=
2k
Z(t)=kf
1
1
1
k m 1
(m 1)
m 1
m 1
Z(t)=kfm
1
n
Z1=1
Z2=2
Z3=3
11
24
EMP5103 Lecture 1
Z1 k1t
Z 2 k 2t
Z n knt
m
R p (t ) 1 Fi (t )
i 1
i 1
for : i t 0.05 :
R p (t ) 1 i t
i 1
For identical units
R p (t ) 1 1 e t
0.005 failures / hours
Eg: = 0.005 failures/hr
m 1
R p (t ) 1 1 e i t
ki
( m 1) 11
m
1
m 1
Parallel Network:
Ri (t ) e i t
t = 100 hours
R p (100) 1 1 e ( 0.005)(100) 0.845
2 redundancy can be the best increasing.

R p (t ) 1 (1 e t )
25
EMP5103 Lecture 1
1
0.80
Rp(t)
0.60
0.40
=4
=3
=2
0.20
=1
0.5
1.5
2.5
R p (t ) 2e t e 2t
0.80
R (t ) e t
0.60
Reliability
0.40
Rs (t ) 2e t
0.20
0.5
1.5
t
Mean Time To Failure (MTTF):

MTTF
R(t )dt
0
26
2.5
EMP5103 Lecture 1
2 t
R p (t ) 1 1 e 1t 1 e
R p (t ) e 1t e 2 t e (1 2 )t
MTTF p
1t
e 2 t e (1 2 )t dt
e 1t e 2 t e 1 2 t
1 2
2
1
1
1
1
1 2 1 2
k1tm
k2tm
m
1
1
1
1
m 1 m 1 1 1
1
m 1
k1 m 1 k 2 m 1 k1 k 2 m 1
MTTF
( ) t 1e t dt
0
Reliability and MTTF Functions for Various Active Configurations for Exponentially
Distributed Failure Times Identical Components:
Reliability Configuration
27
EMP5103 Lecture 1
MTTF
MTTF
1 m 1
i 1 i
1
n
28
MTTF
1 n
n
1
(1) j 1
j 1
j i 1 i
jm
EMP5103 Lecture 1
MTTF
11
12
MTTF
m
1
k _ out _ of _ m
MTTF
System Reliability Plots for Various Configurations:
29
1 m 1
1 k i
1 m 1
n i 1 i
EMP5103 Lecture 1
R(t ) e t
R p (t ) 1 1 e t
0.80
R 2 (t ) 3e 2t 2e 3t
3
0.60
Reliability
0.40
Rs (t ) e 3t
0.20
0.5
1.5
2.5
t
Standby System:
1
Assumptions:
19. switch perfect
20. standby units remain as good as new
21. unit failures are independent
22. identical units
Rs (t ) e t
MTTF
k 1
( t ) t
j!
j 0
Rs (t )dt
0
Approximate Reliability Equations for k-out-of-n type of Systems Identical Components:
30
EMP5103 Lecture 1
Situation
1 unit of 2 must
be working for
success
1 unit of 3 must
be working for
success
1 unit of 4 must
be working for
success
1 unit of n must
be working for
success
2 units of 3 must
be working for
success
3 units of 4 must
be working for
success
(n-1) units of n
must be working
for success
(n-2) units of n
must be working
for success
n
Formula R(t)
Approximation R(t)
2 e t e 2 t
1 t 2
3e t 3e 2t e 3t
1 t 3
4e t 6e 2t 4e 3t e 4t
1 t 4
n t
e

1 1
1 t n
3e 2t 2e 3t
1 3 t 2
4e 3t 3e 4t
1 6 t 2
n
t 2
2

ne ( n 1)t (n 1)e nt
n ( n 2 ) t
n 1 n t
e
e
(2n n 2 )e ( n 1)t
2
2
n
t 3
3
n!
Note:
( n i )!i!
i
Standby Systems:
Situation
1 unit of 2
must
be
working
for
success
1 unit of n
must
be
working
for
success
Formula R(t)
e t te t
e t te t
1
t 2 e t ... 1 t n 1 e t
2
(n 1)!
Fault Trees Analysis:

Developed in Bell labs (minuteman launch control system) in 1962.
31
Approximation
R(t)
1
t 2
2
t n
n!
EMP5103 Lecture 1
Fault Tree Symbols:

- AND Gate:
output
inputs
The AND gate denotes that an output event occurs if and only if all the input events occur.
- OR Gate:
output
inputs
The OR gate denotes that an output event occurs if any one or more of the input events occur.
- Resultant event:
A rectangle denotes an event which results from the combination of fault events through the input of
a logic gate.
- Basic fault event:
32
EMP5103 Lecture 1
A circle represents a basic fault event or the failure of an elementary component. The failure
parameters such as unavailability probability, failure and repair rates of a fault event are obtained
from the empirical data or other sources.
Example:
Build the constant a fault tree of a simple system concerning a room containing a switch and a light
bulb. Assume the switch only fails to close. In addition, the top event is dark room.
Switch fails
to close
Dark Room
Power off
Power
failure
Bulb burnt
out
Fuse failure
P ( A B ) P ( A) P ( B ) P ( A) P ( B )
n
P ( X 1 X 2 X 3 ... X n ) 1 1 P ( X i )
i 1
P ( X 1 X 2 X 3 ... X n )
P( X i )
i 1
output
inputs
P ( A.B ) P ( A) P ( B )
Example:
33
EMP5103 Lecture 1
A
1
6
D
7
4
B
8
C
E
F
0.1421
0.1296
0.36
0.0144
0.36
0.2
0.2
0.2
0.2
34
0.2
0.2
0.36
0.2
0.2
EMP5103 Lecture 1
R A 0.8 0.8 0.64

F A 1 R A 1 0.64 0.36
FB 1 R B 1 0.64 0.36
RC 1 F A FB 1 0.36 0.36 0.8704
FD 1 R D 1 0.8 0.8 0.36
R E 1 0.36 0.36 0.2 0.9856
R N RC R E 0.8578
FN 1 R N 0.1421
2. Miscellaneous Redundancy Related Topics.

2.1.Determining Number of Parallel Units for Specified System Reliability
System
R S 1 1 R n
1 R n
(1 RS )
n ln(1 R ) ln(1 RS )
n
ln(1 R S )
ln(1 R )
Example:
RS=0.98
RS=0.85
ln(1 0.98)
2units
ln(0.85)
RS 1 (1 0.85) 2 0.9775
35
EMP5103 Lecture 1
2.2.Triple Modular Redundancy (TMR)vs. Simplex System

TMR:
R
R
R
Simplex System:
R
R 2 3R 2 2 R 3
3
R F 3
R 3 3R 2 F 3RF 2 F 3 R 3 3R 2 F R 3 3R 2 (1 R) ...
RSimplex R
R 2 RSimplex
3
3R 2 2 R 3 R
3R 2 R 2 1
2 R 2 3R 1 0
b b 2 4ac
2a
R1, 2 1or1 / 2
R1, 2
R (t ) e t
1 / 2 e t
t 0.693
36
EMP5103 Lecture 1
RSimplex (t ) e t
1
Reliability
0.80
R 2 (t )
3
0.60
0.50
0.40
0.20
t=0.693 (mission time)
0
0.5
1.5
2.5
2.3.Common-Cause Failures and Redundancy

Definition:
A common-cause failure is defined as any instance where multiple units or components fail due to a
single cause.
Some-causes:
1. Equipment design deficiency
2. Operations and maintenance errors
3. External normal environment (dust, temperature, humidity, vibrations, etc)
4. External catastrophe Fire, Flood, Earthquake, Tornado
5. Common Manufacturer
6. Common External Power Source
7. Functional Deficiency
2.4.Redundancy Equation Approximation (Effective Failure Rate or Average

Failure Rate)
With Repair:
k-out-of-n system:
23. All units are active with equal unit failure rate and (n-q)-out-of n required for success:
37
EMP5103 Lecture 1
n!q 1
( n q )
(n q 1)! q
1
MTTFk n
MTTFk
n
for : 0 :
( n q )
n
1
i
inq
24. Two active on-line units with different failure repair rates. One out of two is required for
success:
A
A,A
B,B
A B
12 A B A B
A B A B A B
38
EMP5103 Lecture 1
Unit A down
Unit B up
A
Both Up
Both down
B
B
Unit B down
Unit A up
A B 0
A2B AB
A 2 B 2 AB
12
Example:
A system has five active units, each with a failure rate of 220 failures/10 6 hours and only three are
required for successful operation. If one unit fails, it takes an average of three hours to repair it to an
active state. What is the effective failure rate of this configuration?
n5
q 53 2
35
5! 220 10 6
3 0.00575 failure / 10 6 hours
5 2 1! 13 2
Example:
A ground radar system has 2-level-weather-channel with failure rate of 50 failures/10 6 hours and a 6level-weather-channel with a failure rate of 180 failures/10 6 hours. Although the 6 level weather
channel provides more comprehensive coverage, the operation of either channel will result in
acceptable system operation. What is effective failure rate of the two-level-weather-channel if one of
two is required and mean time to repair is 1 hour?
A 50 10 6
B 180 10 6
A B 1
12
50 10 6 180 10 6 1 1 50 10 6 180 10 6
1 1 1 1 50 10 6 180 10 6
12 0.018 failures / 10 6 hours
39
EMP5103 Lecture 1
Example:
Determine the effective failure rate for 8 of 10 identical units required with no repair. The failure rate
of a single unit is 60 failures / 106 hours.
n 10
q2
60 10 6
(10 2 )
10
60 10 6
10
179 failures / 10 6 hours
1i
i 10 2
8units :
60 f / 10 6 hours
( failure _ rate 480 failures / 10 6 hours )
That means that by adding 2 (from 8 to 10) more units the failure rate will drop from 480 to 179
failures / 106 hours.
3. Reliability Testing.
Reliability tests can be divided into three categories:
1. Reliability Development and Demonstration Testing
2. Qualification and Acceptance Testing
3. Operational Testing
Standard used: MIL-STD-471
3.1.Reliability Development and Demonstration Testing

The objectives of reliability and demonstration tests are:
1. To determine if the design must be improved to meet the reliability requirement
2. To indicate any design changes needed
3. To verify improvements in design reliability
3.2.Qualification and Acceptance Testing

The objectives of qualification and acceptance tests are:
1. To determine if a part, assembly, or an end item should be accepted or rejected (either on an
individual or lot basis)
2. To determine if a particular design should be considered as qualified for its intended
application
3.3.Operational Testing
Operational Testing:
40
EMP5103 Lecture 1
The objectives of operational testing:
To verify the reliability analysis performed during the project.

To provide data indicating necessary modifications of operational procedures and
policies, as they affect reliability and maintainability.
To provide information to be used in later activities.
Tests for the Validity of Assumed Failure Time Distributions:
Bartlett test (exponential distribution only)

Kolmogorov-Smirnov test
Bartlett Test:
Bartlett statistic is defined as:
k
2
12k
7k 1
ln X
S bk
1 k
ti
k i 1
k
Y ln t i
i 1
where t i is the ith time to failure and k is the total number of failures in the sample.
A sample of at least 20 failures is necessary for the test to discriminate effectively. If the failure
times are exponentially distributed, then S bk is distributed as chi-square with k 1 degrees
of freedom. Thus, a two-tailed chi-square approach is utilized.
S bk
lower
limit
upper
limit
Example:
A sample of 20 failure times (in days) of an air traffic control system is given in Table 1.
Determine with the aid of the Bartlett test that the Table 1 data are representative of an
exponential distribution with 90% confidence.
7
8
35
46
85
86
41
142
186
EMP5103 Lecture 1
20
19
34
45
63
64
111
112
141
Table 1
185
266
267
1
7 8 20 19 ... 267 96.10
20
20
Y ln t k 82.8311
k 1
82.8311
20
14.43
(7)(20) 1
ln(96.10)
S bk (12)(20) 2
Upper Limit:
, ( k 1) where 1 confidence level 1 0.9 0.1

2
Lower Limit:
2 1 , ( k 1)
2
0.1
, (20 1) 30.14 (from Table 12.4 in handout #1)

2
2
Upper limit:
Lower limit: 1
2
0.1
, ( 20 1) 10.12 (from Table 12.4 in handout #1)
2
The above result exhibits that there is no contradiction to the assumption of exponential
distribution because S bk falls in between the upper limit and the lower limit.
Confidence Limits for Mean Time Between Failures:
2 p, fd
p : the quantity which is the function of the confidence coefficient
fd : degrees of freedom
f : the accumulated number of failures at time t* , where t* denotes the life test termination
f * : the number of failures which were preassigned
m : the quantity of components which were put on test at zero time (t 0)
: the mean time between failures or mean life

P : the acceptable error risk
1 p : the level of confidence
42
EMP5103 Lecture 1
To estimate confidence intervals, there are two cases to consider: (1) the test is terminated at a
specified time,
t * , and (2) the test is terminated at a specified number of failures.

t *:
Test Terminated at Time,
, 2 f 2

2
lower
limit
2
p
1 ,2 f
2
upper
limit
Test Terminated at Specified Failures,
2
2
,
p
2 p
2
,2 f 1 ,2 f

2
2
lower
limit
f *:
upper
limit
The type of test determines the value of , for example, when a failure (failed units) is
replaced or repaired (replacement test) the value of is given by:
mt *
Similarly, for the non-replacement test, the value of gamma is given by:
f
m f t * T j where T j denotes the j th failure

j 1
Example:
Fifteen components were put on test at time t=0, and testing was terminated when the 8 th
failure occurred. That failure (8th) was observed at the 100th hour of testing. Calculate the
components mean time between failures and mean life upper and lower confidence limits with
90% confidence level.
Failed repaired back to its as good as new state:
(15)(100) 1500 hours

1500
187.5 hours
8
p 1 confidence level 1 0.9 0.1
2 f (2)(8) 16
43
EMP5103 Lecture 1
2
(2)(1500)
3000
2
376.9 hours
0.95,16 7.96
p
Upper limit:
(from handout #1)
2
1 ,2 f
2
2
(2)(1500)
3000
2
114 .07 hours

0.05,16 26.30
Lower limit:
(from handout #1)
2 p
,2 f
2
Example:
Twenty-five components were put on a test at time t=0. It was a non-replacement test and was
terminated after 90 hours. During the period, five components failed at 20, 40, 30, 70, and 80
hours of operation respectively. Calculate the components mean time between failures and
mean life upper and lower confidence units with 95% confidence level.
25 5 90 20 40 30 70 80 1800 240 2040 hours

2040

408 hours
5
2
Lower limit:
Upper limit:
, 2 f 2
2
(2)( 2040)
4080
174.81 hours
23.34
(handout 1)
2 0.05

,12
2
p
1 ,2 f
2
2
(2)(2040)
4080
1255.39 hours
3.25
0.05
(handout 1)
1
,10
2
Economics of Testing:
Quality Control Testing
Cost of Testing:
CT P
NL
n
Cost of Not-Testing:
C NT NR P I C R P W C R
44
EMP5103 Lecture 1
N : number of units under consideration

R : total fraction defective previously observed or estimated
P I : fraction defective which fails in plant
P N : fraction defective which fails during the warranty period
CR
: average cost of warranty repair
P : cost of test equipment

n : number of units tested per hour
L : labor and overhead rate per test hour
cost of testing cost of not - testing
C NT C T
N
R P I C R PW C R
Ln
Sources of Product Unreliability:

(excluding operation)
Distribution of Causes (%)
20-40
40-65
15-20
Source
Design and Development
Quality of Components
Quality of Workmanship
(t )
infant
mortality
failures
random
failures
wear-out
failures
t
Causes of Infant Mortality Failures:
Inadequate test specifications

Inadequate quality control
Inadequate manufacturing processes
Inadequate materials
Improper handling
Marginal components
45
EMP5103 Lecture 1
Overstressed components
Etc
Causes of Random Failures:
Insufficient design margin

Misapplication overstress
Wrong use environment
Cause unknown failures
Causes of Wear-Out Failures:
Material wear
Aging
Limited-life components
Inadequate or improper preventive maintenance
Etc
7. Reliability and Maintainability Management

To be continued next time
Handout #1: Table 12.4 Chi Square Distribution

Degrees of
Freedom
1
2
Probability
0.975
0.001
0.05
0.95
0.004
0.1
0.05
3.84
5.99
46
0.025
5.02
7.38
EMP5103 Lecture 1
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

0.22
0.48
0.83
1.24
1.69
2.18
2.7
3.25
3.82
4.4
5.01
5.63
6.26
6.91
7.56
8.23
8.91
9.59
12.40
0.35
0.71
1.15
1.64
2.17
2.73
3.33
3.94
4.58
5.23
5.89
6.57
7.26
7.96
8.67
9.39
10.12
10.85
13.85
7.82
9.49
11.07
12.59
14.07
15.51
16.92
18.31
19.68
21.92
22.36
23.69
25.00
26.30
27.59
28.87
30.14
31.41
36.42
9.35
11.14
12.83
14.45
16.01
17.54
19.02
20.48
21.92
23.34
24.74
26.12
27.49
28.85
30.19
31.53
32.85
34.17
39.36
Handout #2: Table 7.1 Chi-Square Distribution

Degrees
of
Freedom
Probability
0.975
0.05
0.48
1.24
2.18
3.25
4.40
5.63
6.91
8.23
9.59
2
4
6
8
10
12
14
16
18
20
0.95
0.10
0.71
1.64
2.73
3.94
5.23
6.57
7.96
9.39
10.85
0.90
0.21
1.06
2.20
3.49
4.87
6.30
7.79
9.31
10.87
12.44
0.05
5.99
9.45
12.59
15.51
18.31
21.03
23.69
26.30
28.87
31.41
0.025
7.38
11.14
14.45
17.53
20.48
23.34
26.12
28.85
31.53
34.17
0.01
9.21
13.28
16.81
20.09
23.21
26.22
29.14
32.00
34.81
37.57
7. Reliability and Maintainability Management

1. Dhillon, B., Engineering Reliability Management, IEEE Journal on Selected Areas in
Communications, Vol. 4, 1986, pp. 1015-1025.
2. Dhillon, B., Reiche, H., Reliability & Maintainability Management, Vari Nostrand, Reinhold,
New York, 1985.
History:
1958 USAF Reliability Program Management Document, Exhibit 58-10
Steps to Improve Reliability of Engineering Products:
47
EMP5103 Lecture 1
Westinghouse Electric Corporation.

Appoint a manager to look after reliability.
Direct the manager to review present effort.
Establish a high level reliability panel composed, of say, quality control, engineering,
purchasing, and manufacturing managers.
Define the panels responsibilities have it report to top management.
Establish an overall reliability and maintainability policy of the company.
Provide training to people in reliability.
Set reliability goals and refine them as necessary.
Design products that are reliable.
Conduct verification analysis of product reliability.
Produce the products that are reliable.
Audit reliability program on regular basis.
Make use of services provided by the central reliability staff.
Reliability Engineering Department Responsibilities:
Establishing reliability policy, plan, and procedures.

Reliability allocation.
Reliability prediction (MIL-HDBK-217).
Specification and design reviews with respect to reliability.
Reliability growth monitoring.
Providing reliability related inputs to design specification and proposals.
Reliability demonstration (MIL-STD-471).
Training reliability manpower and performing reliability-related research and
development work.
Monitoring subcontractors, if any, reliability activities.
Auditing the reliability activities.
Failure data collection and reporting.
Failure data analysis.
Consulting.
Etc
Tasks of a Reliability Engineer:

To be continued next class
Tasks of a Reliability Engineer:
25. Performing analysis of a proposed design.
26. Analyzing customer complaints with reliability.
27. Investigating field failures.
28. Running tests on the system, sub-system and parts.
29. Developing tests on the system, subsystem and components.
30. Budgeting the tolerable system failure down to the component level.
31. Developing a reliability program plan.
32. Determining reliability of alternative designs.
33. Providing information to designers or management concerning reliability.
34. Monitoring sub-contractors reliability performance.
35. Participating in evaluating requests for proposals.
36. Developing reliability models and techniques.
37. Participating in design reviews.
48
EMP5103 Lecture 1
38. Etc
Reliability Cost Categories:
Reliability cost = PC+ AC + IFC + EFC
PC: Prevention Cost
AC: Appraisal Cost
IFC: Internal Failure Cost
EFC: External Failure Cost
Prevention Cost:
39. Redundancy
40. Parts
41. Hourly cost and overhead rates for design engineers, reliability engineers, etc
Appraisal Cost:
42. Hourly cost and overhead rates for evaluation, reliability qualification, reliability
demonstration, life-testing, etc
43. Vendor assurance cost for new component qualification, inspection, etc
44. Etc
Internal Failure Cost:
45. Hourly cost and overhead rates for troubleshooting and repair, retesting, failure analysis,
etc
46. Replaced parts cost.
47. Spare parts inventory.
48. Etc
External Failure Cost:
49. Cost to failure or repair.
50. Replaced parts cost.
51. Cost of failure analysis.
52. Warranty administration and reporting cost.
53. Liability insurance.
54. Etc
Reliability Program Cost Estimation:
RAM (Reliability and Maintainability) Program Plan (in man hours):
= 2.73 (NOT)2
Where NOT = number of MIL-STD-785 tasks required.
Min 4
Max 22
Reliability Modeling and Allocation:
= 4.05 (MAC)2(NOV)
Where:
MAC: Modelling and Allocation Complexity
Series system = 1
Simple redundancy = 2
Very complex redundancy = 3
NOV: Number of items in allocation process
49
EMP5103 Lecture 1
Min 7
Max 445
Reliability Prediction:
= 4.54(LOD)2(RF)2(POC)
Where:
LOD: Level of detail
1 Prediction exists
2 Prediction made using similar system data
3 Full MIL- HDBK-217 stress prediction
RF: Report formality
1 Internal report
2 Formal report
POC: Percentage Commercial, Hardware used
4 025%
3 2650%
2 5175%
1 76100%
FMEA (Failure Modes and Effect Analysis):
= 17.79(NOI)
Where:
NOI: number of equipment for equipment level FMEA
NOI min 3 max 206
Reliability Testing:
= 182.07(HC)
Where:
HC: Hardware complexity
1 if < 15000 parts
2 if between 15000 and 25000 parts
3 if > 25000
8. Safety Management
In year 2000, in the USA:
55. 5,200 deaths
56. 2.9 million disabling injuries
57. It costs 131 billion dollars to the nation
History:
58. Pliny the Elder (23-79 AD): grinding wear masks
59. 1893 in the USA: Rail Safety Act
60. 1938 in the USA: Food, Drugs and Cosmetic Act
61. 1970: Occupational Safety and Health Act
62. Nuclear regulatory commission
63. Consumer product safety
64. Commission
65. National Transportation Safety Board
66. Federal Aviation Agency (FAA)
67. Etc
50
EMP5103 Lecture 1
Safety Periodicals:
68. Journal of Occupational Accident
69. Professional Safety
70. Concern
71. Journal of Safety Research
72. Protection
73. National Safety News
74. Nuclear Safety
75. Accident Prevention
76. Accident Facts
77. Safety Management Journal
78. Etc
Selective Texts on Safety:
1. Heinrich, H.W., Industrial Accident Prevention, McGraw Hill, New York, 1831
2. Handley, Industrial Safety Handbook, McGraw Hill, New York, 1969
3. Gloss, D.S, Introduction to Safety Engineering, Wiley, New York, 1994
4. Hammer, W., Product Safety and Engineering, Prentice Hall, New Jersey, 1980
5. Dhillon, B.S, Safety Assessment, A Quantitative Approach, Lewis Publish, New York.
1994
Safety-Related Data Sources:
79. Government Industry Data Exchange Program (GIDEP), Fleet Missile Systems Analysis,
and Evaluation Group, U.S. Navy, Corona, California.
80. International Occupational Safety and Health Information Center, Bureau International
du Travail, Geneva, Switzerland.
81. Loss Management Information System (LOMIS), Gulf Canada 800 Bay Street, Toronto.
Factors for Developing Safety Requirements:
82. Occupational Safety and Health Act regulations.
83. Environmental Protection Agency regulations.
84. Nuclear Regulatory Commission regulations.
85. Company safety policy with respect to plant and administrative procedures.
86. State and local government requirements.
87. Etc
Safety-Related Activities for the top management:
88. Safety training.
89. Safety Inspections.
90. Safety problem diagnosis ns solutions.
91. Accident investigations.
92. Employee participation in safety programs.
Safety Engineers Responsibilities:
93. Accident prevention and analysis.
94. Management of safety training.
95. Acting as a consultant to management on safety-related matters.
96. Design layout of equipment with respect to safety.
97. Study human factors (ergonomics) with respect to safety.
98. Focus on the interface between the workplace and environment.
99. Keep abreast of new literature on safety.
100.
Etc
51
EMP5103 Lecture 1
System Safety Analysis Methods:

101.
Fault tree analysis.
102.
Failure Mode and Effect Analysis (FMEA).
103.
Event tree analysis.
104.
Job safety analysis.
105.
Single point failure analysis.
106.
Etc
Fault tree analysis in Safety Studies:
?
Operator fails
to wear safety
glasses
Chip in eye (Grinding)
Person without safety glasses other than

operator s close to operation
Machine
operating
People in the
area (motive to
go into area by
non operator)
Safety glasses
not worn
52
Operator
fails to
stop
oeration
EMP5103 Lecture 1
Person enters
to bring item
to the area
Person enters
to carry away
items
Person enters area

for other reason
0.08012
0.00012
0.08
0.1
0.06
0.02
0.01
0.1
0.2
0.03
Formulas / Models Related to Safety:

107.
American National Standard Institute (ANSI), Z-1601, 1985. Title: Method of
Recording and Work Injury Experience.
Formula / Index:
Estimating the disabling-injury frequency rate (DIFR)
DIFR
NDI (100000)
EHE
EHE = the Employee-Hours of Exposure.

NDI = Number of Disabling Injuries.
Formula II:
DISR
D(100000)
EHE
53
EMP5103 Lecture 1
DISR = Disabling Injury Severity Rate

D = Total number of days charged
Shortcomings:
108.
Not sensitive enough to serve to serve as accurate indicator of safety
effectiveness.
109.
The smaller the work force, the less reliable the frequency rate, severity rate.
110.
Many accidents, particularly the less severe ones, are never reported.
111.
Etc
9. Robot Reliability and Safety

112.
5000 Egyptians built water-powered clocks. Chinese and Greeks built water and
steam powered toys.
113.
Aristotle: If every instrument could accomplish its own work, obeying or
anticipating the will of others
114.
1818 Many Shelley wrote the science fiction Frankenstein Machine Monster
115.
1920 Karl Capek wrote the science fiction Rossums Universal Robots.
116.
1942 Asimov 3 laws:
o A robot may not injure a person nor, through inaction allow a person to come to
harm.
o A robot must always obey orders from people except in circumstances in which
such orders conflict with the above (first) law.
o A robot must protect its own existence except in circumstances in which it is in
conflict with above two laws.
117.
1959 Planet Corporation commercialized the first robot in USA.
118.
1967 Japan got their fist robot.
119.
1970 Conference about robot.
120.
1971 Japanese Industrial Robot Association.
121.
1975 Robot Institution of America.
History of Robot Reliability and Safety:
122.
1985 Japanese Industrial Safety and Health Association.
123.
1986 American Nation Standard Industrial Robots and Systems Safety
Requirements.
124.
1986 Robot Safety, edited by Bonney and Yong.
Robot Population Worldwide:
125.
1981 US Automotive industry
o 30% of labor cost
126.
Japan (135 companies concerned with robots).
127.
US (100 companies concerned with robots).
128.
UK (30 companies concerned with robots).
129.
Today around 1 million
130.
2010 should be about 5 million robots.
Japan:
131.
132.
133.
134.
135.
Automobile (36%)
Electric machinery (30%)
Plastic molding products (10%)
Metal working
Steel making
54
EMP5103 Lecture 1
136.
137.
138.
139.
USA:
140.
141.
142.
143.
144.
145.
In 1990, the
Textile
Chemical
Ship building
Etc
Assembly (35 40%)
Arc welding (15 20 % )
Material handling (30 35%)
Paint spraying (5%)
Spot welding (3 5%)
Other areas (7 10%)
US had about 100,000 robots
Robot Accident Examples:

146.
A repair person climbed over a safety fence without shutting off the power to the
robot and worked in its area while it was temporarily stopped. When the robot
recommenced movement, it pushed the repair person into a grinding machine and,
consequently, the person died.
147.
A worker switched on a welding robot meanwhile another person was still in its
working area, consequently, that person was pushed into a positioning fixture by the
robot and died later.
Common Reason for Deaths:
148.
A human entered the danger zone to rectify fault.
149.
The human was either pushed into another machine or crushed against
something else by the robot. The robot itself did not kill the human.
150.
The human struck from behind by the robot in such a situation, the concerned
human was not aware that the robot was moving until it was too late.
151.
Even though the human was experienced, but through appropriate training
should have been alerted of existing dangers.
Robot Accident Around the World:
152.
Japan
153.
Sweden
154.
U.K.
155.
U.S.A.
Japan Accidents:
156.
4 persons died and several cases of injuries.
157.
1978 1982 (190 plants 4341 robots)
o 2 deaths
o 2 cases of lost time (3 days, 7 days)
Sweden Accidents:
Survey data (Jan 1976 June 1978) involving a total of 270 robots:
158.
7 accidents per year.
Survey data (1979 1983) involving a total of 36 robots:
159.
8 accidents per year.
Breakdown by industry:
160.
Foundries (4 accidents)
161.
Plastics (4 accidents)
162.
Automotive (9 accidents)
55
EMP5103 Lecture 1
163.
164.
Metal working (16 accidents)

Others (3 accidents)
U.K. Accidents:
165.
2623 robots
166.
Survey 37 robot systems different designs
167.
Approximately 22,000 robot production hours.
168.
73 accident occurrence
169.
1 led to human injury
170.
57 damage to machinery
171.
15 no damage accidents
172.
Approximately 25% of system production time lost due to accidents.
U.S.A. Accidents:
- 13,000 robots (1984).
- 1 death
- General Motors Corporation (used robots over 23 year)
- One serious injury.
- One minor injury.
U.S.A. Canada Accidents:
- 17 accidents:
o 1 death.
o 3 serious injuries.
o 5 minor injuries.
o 8 mal-functions.
Causes of Robot Accidents:
- Japanese study-causes of 18 near accidents.
- Incorrect action by the robot during manual operation.
- Incorrect movement of peripheral equipment during teaching or testing.
- Erroneous movement of the robot during teaching or testing.
- Sudden entry of the human to the robot area.
- Incorrect movement of peripheral equipment during normal operation.
- Etc
Possible Sources of Robot Accidents:
- Engineering Factors:
o Control panel failure
o Robot arms high speed (the speed is the factor)
o Poor software design
o Poor control panel design
o Etc
- Usability (user-friendliness) Engineering Factors.
- Organizational factors:
o Inadequate robot training programs-repairman, operators, programmers, etc
o Incorrect procedures for initial robots start-up.
o Operator carrying robot adjustments (they know how to operate but not how to
maintain).
o Etc
56
EMP5103 Lecture 1
Some
-
Solutions to Safety-Related Problems:

Improving robot reliability.
Improving design of mechanical hardware components.
Providing appropriate safety training to people concerned with maintenance, operation,
and testing.
Developing an effective sensory detection capability of the robots.
Paying proper attention to human factors during the design of human workstation
layout.
Paying more attention to human factor during the design of the robot systems.
Etc
Robot Reliability:
- Expected life: 40,000 hours.
- MTBF at least 400 hours.
- MTTR of 8 hours or less.
- Maximum MTBF: 2500 hours
- Cost of maintenance: approximately 11% of the procurement cost
- Availability: 0.98 (98%) (for power station 0.9999)
Publications:
1. Engelberger, J.F., Three Million Hours of Robot Field Experience, The Industrial Robot,
1974, pp. 164-168.
2. Pollard, B.W., RAM for Robots: Reliability, Availability, and Maintainability, Robotics Today,
1981, pp. 209-220.
3. Bonney, M.C., Yong, Y.F., Editors. Robot Safety, Springer, New York, 1985.
4. Dhillon, B., Survey On Robot Reliability and Safety, Microelectronics and Reliability, Vol.
27, 1987, pp.105-118.
5. Dhillon, 1991 (publications).
Robot Reliability Measures:

Mean Time to Robot Failure:
Rr ( s )
Rr (t )dt slim
0
MMTFR
0
t
Rr (t ) e
r (t ) dt
MTTFR
PHR DTDTRF
NRF
where :
PHR is the production hours of the robot.
DTDTRF is the downtime due to robot failure expressed in hours.
NRF is the number of robot failures.
Example:
PHR=15,500 hours, DTDRF=200 hours, and NRF=10
MTTFR = (15500-200)/10 = 1530 hours
Mean Time to Robot-Related Problems:
57
EMP5103 Lecture 1
MTRP
PHR DTDTRP
NRP
PHR is the production hours of the robot.

NRP is the number of robot-related problems.
DTDTRP is the downtime due to robot-related problems expressed in hours.
Example:
MTRP = (20000-280)/20 = 986 hours
Robot Reliability:
t
Rr (t ) e
r (t ) dt
0
r (t ) r
t
Rr (t ) e
r (t ) dt
0
e r t
Pareto Principle Robot Quality Assurance:

Alfredo Pareto (1848-1923)
His principle in relation to quality control work simply states that there are always a few kinds
of defects in the hardware manufacture which loom in occurrence frequency and severity.
Robot Model with Preventive Maintenance:
p
Robot down
for preventive
Robot
Operating
maintenance
Robot Failed
Andrei Markov (1856-1922):

- Assumptions:
1. The probability of more than one transition in time interval t form one state to the next
step is negligible.
2. The occurrences are independent.
3. The transitional probability from one state to the next state in the time interval t is
given by t, where is the constant failure rate associated with Markov states.
4. (t) (t) 0
i denotes the ith sate of the robot: i = O (operating)
i = p (preventive maintenance), i = f (failed)
Pi(t) is the probability that the robot is in state i at time t.
p is the robot preventive maintenance constant rate
f is the robot constant failure rate.
p is the robot constant repair rate due to preventive maintenance.
f is the robot constant repair rate.
Po (t t ) Po (t )(1 f t )(1 p t ) P f (t ) f t Pp (t ) p t
P f (t t ) P f (t )(1 f t ) Po (t ) f t
Pp (t t ) Pp (t )(1 p t ) Po (t ) p t
58
EMP5103 Lecture 1
Po (t t ) Po (t ) ( f p ) Po (t )t P f (t ) f t Pp (t ) p t
P (t t ) Po (t )
lim o
( f p ) Po (t ) P f (t ) f Pp (t ) p
t
t 0
dPo (t )
( f p ) Po (t ) P f (t ) f Pp (t ) p
dt
dPp (t )
dt
dP f (t )
p Pp (t ) Po (t ) p
f P f (t ) Po (t ) f
dt
At _ time _ t 0, Po (0) 1, Pp (0) P f (0) 0
Laplace _ transform :
p( s)
st
p (t ) dt
p (t )
e
at
dp (t )
dt
p (s )
1
sa
sp ( s ) p (0)
f (t ) lim sf ( s )
Final value Theorem: tlim
0
s 0
sPo ( s) Po ( s) ( f p ) Po ( s) Pp ( s ) p P f ( s ) f
sPo ( s) 1 ( f p ) Po ( s ) Pp ( s) p P f ( s ) f
Po ( s)
( s f )( s p )
s s 2 s( f p p f ) f p p f f p
A s 2 s( f p p f ) f p p f f p
f (s p )
Pp ( s )
sA
p (s f )
Pf (s)
sA
f p
Ass Po lim sPo ( s )
f p p f f p
s 0
B f p p f f p
Pf
Pf
f p
B
p f
B
59
EMP5103 Lecture 1
Po (t )
Pp (t )
P f (t )
f p
k1k 2
p f
k1k 2
f p
k1k 2
(k1 p )(k1 f ) k t (k 2 p )(k 2 f ) k t

e 1
e 2
k
(
k
k
)
k
(
k
k
)
1
1
2
2
1
2
p k1 p f k t f k 2 k t
e 1
e 2
k
(
k
k
)
k
(
k
k
)
1 1 2
2 1 2
f k1 f p k t f k 2 k t
e 1
e 2
k
(
k
k
)
k
(
k
k
)
1 1 2
2 1 2
( p f f p ) ( f p p f ) 2 4( f p p f f p )
k1 , k 2
2
Maximizing Income of a Robot System Subject to Failure and Repair:
Robot Up
0
Robot Down
1
dP0 (t )
P0 (t ) P1 (t )
dt
dP1 (t )
P1 (t ) P0 (t )
dt
At _ time _ t 0, P0 (0) 1, P1 (0) 0
e ( )t availability

U (t ) P1 (t )
e ( )t unavailability

A(t ) P0 (t )
60
EMP5103 Lecture 1
A(t)
0
t
Ass
1
MTTR
1
1
MTTF MTTR
MTTF
uptime
MTTR MTTF uptime downtime
Periodic Cost of the robot system maintenance crew is given by:
MC k
k
MTTR
where:
k is the robot system maintenance cost (constant) depending on the nature of the robot
system.
The expected periodic income from the robot system output:
EI I AV
where:
I is the periodic income from the robot system output, if the robot system worked full time.
MTTF
MTTF MTTR
EI I r AV I r
Thus the net income, NI of the robot system is:
61
EMP5103 Lecture 1
NI EI MC
I r MTTF
k
NI
MTTF MTTR MTTR

dNI
I MTTF
k
0
2
dMTTR
MTTF MTTR MTTR 2
MTTF
MTTR*
1/ 2
I MTTF
1
10. Quality Management.

Quality Control:
This is a management function whereby control of the quality of manufactured item and raw
materials is expected to prevent the manufacture of defective items.
History:
- 1916: C.N. Frazee
- 1924: Quality control charts developed by Walter A. Shewhart.
- 1946: ASQC (American Society for Quality Control)
Journals (publications):
- Quality Progress
- Journal of Quality Technology
- IEEE Transactions on Reliability
- Annual Quality Congress transactions of the ASQC
Some
-
Elements of Quality Discipline:

Statistical Quality Control
Procurement Quality Control
Quality Costs
Applied Quality Control
Quality Circles
Etc
Present Quality Trends Facing Industry:

- 7 to 10% (of the total cost of production?)
- The cost of quality has risen to a very high level
- Customers quality requirements have been rising at a alarming rate
- Because of the above factors, present methods and practices associated with quality are
rapidly becoming outmoded.
Functions of the Quality Control Engineering:
- Process quality control (manufacturing aspect)
- Quality control in new design
- Quality control of incoming material
- Inventory evaluation quality control
- Special studies concerning quality control
Quality Control Manual:
- Benefits:
62
EMP5103 Lecture 1
It becomes useful when making quality-related decisions.

It serves as a reference document.
It can be used as a textbook when training quality personnel.
It helps in continuity of operations of the quality control organization despite the
personnel turnover.
Information:
o Responsibilities.
o Statistical methodology.
o Personnel.
o Organization charts.
o Quality policies and procedures.
o Vendor quality control procedure.
o Quality costs and inspection procedures.
o Measuring equipment.
o Defect prevention.
o Promote quality
o Etc
o
o
o
o
Quality Costs:
- Post delivery failure costs:
o Warranty charges
o Complain adjustment
o Returned material
o Etc
- Prevention costs:
o Quality planning
o Design review
o Supplier evaluation
o Process control
o Training
o Equipment calibration
o Etc
- Costs of internal failures:
o Repair and rework
o Scrap
o Re-inspection
o Downtime of facilities because of defects
o Etc
- Evaluation Costs:
o Incoming material inspection
o In process inspection
o Assembly inspection
o Review and recording data
o Auditing the quality system
o Etc
Indexes:
- Vendor Rating Program Index:
63
EMP5103 Lecture 1
QI
Cvq C p
Cp
QI : Value _ of _ the _ quality _ cos t _ performance _ index

Cvq : The _ vendor _ quality _ cos t
C p : The _ purchase _ cos t
QI = 1.1 or more indicates that there is an immediate need for corrective measures.
Example:
Vendor quality cost = $2,000
Purchase cost = $50,000
QI = (2000+50000)/50000 = 1.04
1.000 1.009: Excellent performance
1.01 1.03: Good performance
-
Evaluating Quality Costs:
a
100 100
b
: Quality _ Index
a : Quality _ cos t
b : The _ value _ of _ the _ output
In real life situation, a value of = 105 can readily be achieved.
= 110130: quality costs are ignored
- Determining Accuracy and waste of inspector:
Formula I:

100

where :

: the percent of defects correctly identified by the regular inspector.
: is the number of defects missed by the regular inspector as reveled by the check
inspector.
: is the number of defects discovered by the regular inspector.
: is the number of units without defects rejected by the regular inspector as revealed by
the check inspector.
Example:
A regular inspector inspected a number of units in a lot and found 60 defects. All units (i.e.:
good plus defective) of the lot were reexamined by the check inspector. Thus according to
the findings of the check inspector, the values of and were 10 and 15.
100 50100
76.92%
15 50 15
60 10 50
Formula II:
100
m
: the percent of good units rejected by the regular inspector.

m: the total number of units inspected.
64
EMP5103 Lecture 1
Quality Control Charts:

1. The p-chart.
2. The X -chart
3. The R-chart
4. The c-chart
The p-chart:
Components with defects versus components free of defects.
Upper control limit: UCLp = + 3
Lower control limit: LCLp = - 3
: the mean of the binomial distribution
: the standard deviation of the binomial distribution
99.7%
N
m
Where :
N: is the total number of defectives in classification
65
EMP5103 Lecture 1
m: is the size of the sample

: is the number of samples
1

m
Example: 8 samples were taken from a production line. Each sample consists of 40 mechanical
parts. After inspection, it was found that samples 1, 2, 3, 4, 5, 6, 7, and 8 contained 5, 6, 4, 2,
8, 10, 12, and 9 defectives respectively. Develop the p-chart.
The fraction of defectives in sample 1:
5
0.1250
40
Similarly, the fraction of defectives in samples 2, 3, 4, 5, 6, 7, and 8 are0.15, 0.10, 0.05, 0.20,
0.25, 0.30, and 0.225 respectively.
56
0.175
40 8
0.1751 0.175 2

0.06
40
UCL p 3 0.3552
1
UCL p 3 0 : because _ negative _ value
0.40
UCLp=0.352
0.35
0.30
0.25
0.20
=0.175
0.15
0.10
0.05
0
Sample number
66

Industrial Organisation Class Notes Uottawa MCG 5171B

Загружено:

Сведения о документе

Исходное описание:

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Industrial Organisation Class Notes Uottawa MCG 5171B

Загружено:

Авторское право:

Доступные форматы

EMP5103 Lecture 1

Thursday, January 10, 2013

EMP 5103 Reliability, Quality and Safety Engineering

Introduction familiarize reader on topic sections.

Thursday, January 10, 2013

Reliability background and introductory mathematics.

1. Reliability Background and Introductory mathematics

Power system problems (1930s)

Thursday, January 10, 2013

History of Quality Control

C.N. Frazeo of Telephone Laboratories made use of statistical methods in inspection

IEEE Transactions on Reliability (1953)

Journal of Safety Research (USA)

Quality Control Publications

Allocate top reliability of Arm to subsystems right down to components level.

Thursday, January 10, 2013

Data collection and Analysis writing information to databank and retrieving it

Reliability Areas and Applications:

9. Failure data collection and analysis

Thursday, January 10, 2013

14. System safety

P( X 1 X 2 X 3 ... X n ) 1 (1 P( X i )) (independent events)

P( X 1 X 2 X 3 ... X n ) P ( X 1 ) P ( X 2 )...P ( X n ) (independent events)

Reliability Related Formulas:

failure density function

Failure density function:

R is reliability and F is Failure

Thursday, January 10, 2013

Failure Time Distributions

and b,, > 0

b, , and are shape, scale, and location parameters, respectively

Thursday, January 10, 2013

See page 16a for function recapitulation.

Thursday, January 10, 2013

Bathtub Hazard Rate Curve:

Useful life period

Wear out period

For b,c,, > 0

Thursday, January 10, 2013

2. Failure Data Collection and Sources

Past experience with similar or identical material.

Thursday, January 10, 2013

Non-standardized data reporting.

Failure Data Forms:

Thursday, January 10, 2013

MIL-HDBK-217: very important document that includes mathematical formulas

Thursday, January 10, 2013

Failure Rate Evaluation of Equipment:

k = the number of different generic part classifications

3. Static Reliability Evaluation Models:

Ei: Successful event i

RS 0.8 0.9 0.72

Where Fi is the failure probability of unit i.

Thursday, January 10, 2013

(for non-identical units)

For identical units:

Thursday, January 10, 2013

R p 1 F1 F2 F3 F4 1 0.4 0.3 0.2 0.1 0.9976

Thursday, January 10, 2013

Thursday, January 10, 2013

Thursday, January 10, 2013

k (requirement for the system to work)