RELIABILITY FOR ENGINEERING DESIGN

Quality Control refers to the actions taken throughout the engineering design and
manufacturing of a product to prevent and detect product deficiencies and product
safety hazards.

Conformance to design is achieved through quality control techniques. It is applied during

the manufacturing of the product.

How does the product perform during its intended lifetime?

Measure of product performance during the lifetime is reliability.

Quality Control is static, Reliability is time dependent.

Reliability may be defined as the probability that an item or piece of equipment will cary
out its specified function satisfactorily for the stated period when used under the
designed conditions.

If any machine satisfy design specifications at any time, it is said to be reliable. 100%
reliability is seldom possible.

Reliability of 0.999 implies that there is probability of failure of 1 part in every 1000
parts.

İlhan Konukseven ME 407 Mechanical Engineering Design 1

 You may design for 100% reliability, but production and operation requirements, and cost
factors make it impossible to implement, and never achieved in practice. Even at design
stage 100% reliability is rare, designers prefer to find an optimum.

Total Cost

Cost of design
And manufacture
Costs after delivery

The costs of design and manufacture increase with product reliability. The slope of the
curve increases, and each incremental increase in reliability becomes harder to achieve.
The cost of the product after delivery to the customer, (warranty and replacement
costs) decrease with increasing reliability. The summation of these two curves produces
the total cost curve, which has minimum at an optimum level of reliability.

İlhan Konukseven ME 407 Mechanical Engineering Design 2

 SELECTED TERMS, AND THE BATHTUB HAZARD RATE CURVE

Hazard rate : This is the rate of change of the number of failed parts divided by
the number of survived parts at time t

Reliability : This is the probability that an item will perform its specified function
stisfactorily for the stated period when used under the designed
conditions

Maintainability : This is the probability that a failed item will be repaired to its
satisfactory operational state.

Active redundancy : This term indicates that all redundant items are operating
simultaneously.

Failure : This is the inability of an item to operate within the initially stated
guidlines

Maintenance : This is all the actions required to keep an item in a defined condition,
or restoring it to that condition.

Downtime : This is the period of time during which an item is not in a condition to
perform its stated mission.

İlhan Konukseven ME 407 Mechanical Engineering Design 3

 The failure behavior of many items may be described by the bathtub hazard rate curve

(Burn-in period, infant morality Failures during this period The end of the useful life period.
period, debugging period) are assumed to occur randomly Causes
Reasons and remains constant over time - Poor maintenance
- Poor quality control Reasons - Incorrect overhaul practices
- Poor workmanship - Low safety factors - Aging
- Substandart parts - Undetectable factors - Friction
- Poor manufacturing methods - High unexpected random - Corrosion
- Inadequate debugging stress
- Human error - Natural failures
İlhan Konukseven ME 407 Mechanical Engineering Design 4
 RELEABILITY THEORY
The disipline of reliability engineering basically is a study of the cause, distribution, and
prediction of failure.

R(t) : Reliability with respect to time t

F(t) : Unreliability (probability of failure) in the same time t

R(t) + F(t) = 1

Math measure of Realibility

Time_Dependent Reliability
r : Rate of failure
P : Probability of survival P=1/r
m : Mean time between failures

No : Number of machines at the start of a time period t.

Nf : Number of failed machines during t.
Ns : Number of surviving machines after t N0  N s  N f

 N0  N f  N Ns
R     1 f
 or R
 N0  N0 N0
As time goes Nf increases, therefore R=R(t)

Rate of failure 1 dN f dN
r
dR
and r  
dR
 N 0   f
dt N 0 dt dt dt

İlhan Konukseven ME 407 Mechanical Engineering Design 5

 dN
Number of machines surviving: N s  N0  N f
dR
 N 0   f
dt dt
h(t) :The hazard rate is the number of failures per unit time per the number of
items exposed for the same time.
f(t) : is the failure density function
R(t) : Reliability with respect to time t
F(t) : Unreliability (probability of failure) in the same time t ( R(t) + F(t) = 1 )
dN f dF ( t ) d 1  R( t )  dR( t )
N dR( t ) 1 dR( t ) f (t )   
h( t )  dt   0  dt dt dt
Ns N s dt R( t ) dt f (t ) f (t )  dR( t ) 1
h( t )   
R( t ) 1  F ( t ) dt R( t )
dR( t ) Boundary conditions;
t
h( t )dt   at t=0 R=1 (new machines) R( t )  1   f ( t )dt
R( t ) 0

ln R( t )    h( t )dt
t

R( t )  exp   h( t )dt 
t

 0 
 0t h( t )dt
R( t )  e

a)Distribution of time to failure b)Cumulative distribution of time to failure

İlhan Konukseven ME 407 Mechanical Engineering Design 6
 Example:
Assume that the times-to-failure of an automobile engine are described by an exponential
distribution, as follows: (Known model-Exponentially distributed times to failure)
 ht
f ( t )  he
Where t is time, and h is the failure rate of the automobile engine. Obtain an expression
for the automobile engine’s time-dependent reliability, and the reliability value, for a 50-
hour mission, if the engine’s failure rate is 0.0005 failures per hour.
dF ( t )
f (t ) 
Solution: t h(t): hazard rate dt
R( t )  1   he  ht dt  e  ht f(t): failure density function
0 R(t): Reliability with respect to time t
F(t): Unreliability (probability of failure)
R(50)  e  ( 0.0005 )( 50 )  0.9753
The results means there is 97.53 percent chance that the automobile will not fail during
the 50-hour mission

Example:
Obtain an expression for the hazard rate of the automobile engine in previous example

h( t ) 
1
e  
 ht
he   h
 ht
dN f
N dR( t )
h( t )  dt   0 
1 dR( t )
Ns N s dt R( t ) dt
This result means that for exponentially distributed times-to-failure, the engine failure
rate is constant (Constant-failure-rate system)

İlhan Konukseven ME 407 Mechanical Engineering Design 7

 Mean Time to Failure (MTTF)

The mean time to failure of an item or system may be obtained using the following
relationship:


MTTF   R( t )dt
0

or

MTTF  lim R( s )
s 0 Where s is the Laplace transform variable.
R(s) is the Laplace transform of the reliability function

Example:
Consedering previous example calculate the automobile engine’s mean time to failure.
 
 ht 1
MTTF   e dt  MTTF   R( t )dt
0
h 0
1
MTTF   2,000hours
(0.0005)
This result means that for exponentially distributed times-to-failure, the engine failure
rate is constant

İlhan Konukseven ME 407 Mechanical Engineering Design 8

 BASIC NETWORK RELIABILITY

Most mechanical and electronic systems comprise a collection of components. The overall
reliability of the system depends on how the individual components with their individual
failure rates are arranged.

The components of a system may be interconnected in various forms called basic network.

SERIES NETWORK hi
1 2 3 n

Each block represents a single component of the system. If any one of the components
fail, the network fails. In other words, all the components must operate normally for the
series network to function successfully. Most consumer products exhibit series reliability.

For independent components, the network reliability, or the probability of success, is

Rsystem  R A  RB  . . . . . . . . . . .  Rn
Example: 20 components each with R=0.99, the system reliability is 0.9920=0.818
(Known model-
Exponentially distributed times to failure-
h(t): hazard rate Constant-failure-rate system)
t
Rsystem  e  h1t  e  h2t  . . . . . .  e  hnt Ri ( t )  1  Fi ( t )  1   f i ( t )dt
0
For a constant failure rate for each unit
n t
  hi t Ri ( t )  1   hi e  hi t dt  e  hi t
Rsystem  e i 1 0

İlhan Konukseven ME 407 Mechanical Engineering Design 9

 PARALLEL NETWORK

For the success of the system, at least one of the units must function normally. The
network is actually redundant system in that all of the units are active at one time; yet,
only one must function normally for system success. This is one of the approaches used to
improve system reliability.
1
Rsystem  1  (1  R A )(1  RB ) . . . . . . . . . . . (1  Rn )

hi 2
For a constant failure rate for each unit

Rsystem  1  (1  e  h1t )(1  e  h2t ) . . . . . . . . . . . (1  e  hnt )

n

k-OUT-OF-m NETWORK

A k-out-of-m network contains a total of m active units connected in parallel, and at least
k units must operate normally for the network (or system) to function successfully..

m
m j
Rk / m    j R 1  Rm  j
jk 

m m!
  
 j  j! ( m  j )!
Where Rk/m is the k-out-of-m network reliability

İlhan Konukseven ME 407 Mechanical Engineering Design 10

 STANDBY NETWORK

A standby network (or system) is an important type of redundancy network in which only
one unit is active or operating and m units are on standby (nonactive). If the operating unit
fails, it is immediately replaced by one of the standby’s. The system fails when all the
standby’s and the operating unit fail.

The reliability of a system of two components, one of which is on standby, is

R( t )  e  ht (1  ht )
Standby

Operating
(Known model- System
Exponentially distributed times to failure-
Constant-failure-rate system)
t
Ri ( t )  1  Fi ( t )  1   f i ( t )dt
0
t
Ri ( t )  1   hi e  hi t dt  e  hi t
0

İlhan Konukseven ME 407 Mechanical Engineering Design 11

 Example: A complex engineering design can be described by a reliability block diagram as
shown in the Figure. In subsystem A two components must operate for the subsystem to
function successfully. Subsystem C has true parallel reliability. Calculate the reliability of
each system and the overall system reliability.

0.97
0.85
0.97
0.97 0.85

0.97
0.85
0.97
Network Reduction Method A B C

Solution: Subsystem A is an k-out-of-m model for which k=2 and m=4.

4
 4 j  4 2 4 2  4  3  4 4
R A    R 1  R     R 1  R    R (1  R )    R  0.998
  4 j
   
j  2 j   2  3  4
RC  1  (1  R1 )(1  R2 )(1  R3 )  1  (1  R1 ) 3  0.9966
Rsystem  R A  RB  RC  (0.998)(0.970)(0.9966)  0.965

İlhan Konukseven ME 407 Mechanical Engineering Design 12

 FAULT TREES

A fault tree is a powerful tool for evaluating the reliability of systems during their design
phase.

Basic symbols used to construct the fault tree:

Resultant fault error: A rectangle denotes a fault event that results from a combination of
failure events through the input of a logic gate.

Basic fault event: A circle denotes a basic fault event or the failure of an elementary
component.

AND Gate: This symbol denotes that an output fault event occurs if all the input fault events
accur.

OR Gate: This symbol denotes that an output fault event occurs if one or more of the input
fault events occur.

The four basic steps involved in developing a fault tree are as follows:

1. Define the top undesired event of the system to be studied.

2. Develop a through understanding of the system under consideration
3. Determine the logical interrelationships of higer-level and lower-level fault events.
4. Construct the fault tree using logic sysmbols.

İlhan Konukseven ME 407 Mechanical Engineering Design 13

 Example: Develope a fault tree for a system comprising a windowless room with one
switch and three light bulbs. The switch can only fail to close, and the top
undesirable event is the room without light.

Solution: Each individual capital letter in the figure denotes the corresponding fault
event.
Room without light
(top event)
The output fault occurrence probabilities
for OR and AND gates, respectively

Switch without All bulbs Switch

m power Burnt out Fails to
FAND  1   (1  Fi ) close
i 1

m
FOR   Fi
i 1
Fuse Bulb A Bulb B Bulb C
Fuse
Board w/o Burnt Burnt Burnt
failure
power out out out

The four basic steps involved in developing a fault tree are as follows:

1. Define the top undesired event of the system to be studied.

2. Develop a through understanding of the system under consideration
3. Determine the logical interrelationships of higer-level and lower-level fault events.
4. Construct the fault tree using logic sysmbols.
İlhan Konukseven ME 407 Mechanical Engineering Design 14
 MECHANICAL FAILURE MODES AND SAFETY FACTORS

A failure mode is a physical process or combination of processes whose effects induce

failure.

Categories I:
Hardware failure : failure of component to function as designed
Software failure : failure of the computer software to function as designed
Human failure : failure of human operators to follow instructions or respond
adequately to emergency situations
Organizational failure : failure of the organization to properly support the system.
(overlooking defective components, slowness to bring corrective
action, ignoring bad news)

Categories II:
Structural : structural failure modes are: fracture, and excessive deflection
Thermodynamic : Two assosiated failure modes are: overheating, and reduction in
efficiency
Fluid : The associated failure modes are: leakage, and distorted flow.
Hydrolic actuation : The failure modes include: fitting leakage, static seal leak, fluid
dirt contamination, and actuator cylinder rapture.

İlhan Konukseven ME 407 Mechanical Engineering Design 15

 SAFETY FACTORS

The parameter used to ascertain the safeness of a member or component is called the safety
factor. There are many ways to define a safety factor.

Definition I The safety factor Sf is defined as

Maximum safe load
Sf 
Normal service load
Definition II

Mean strength
Sf  1
Mean Load

İlhan Konukseven ME 407 Mechanical Engineering Design 16

 ENGINEERING DESIGN RELIABILITY GUIDELINES

1. Test new designs carefully; devise effective test procedures

2. Simplify a design as much as possible
3. Use standart parts as much as possible
4. Use well-tried parts as materials
5. Allow for human error
6. Do not use cost-saving approaches at the expense of reliability
7. Evaluate maintenance with respect to reliability
8. Use redundancy if required
9. Evaluate manufacturing with respect to reliability
10. Consider the effects of transportation, handling and storage
11. Carefully analyze all data received from the field
12. Carefully consider the design of critical parts
13. Incorporate inspection facilities
14. Carefully consider available field data when making modifications and improvements to a
design

İlhan Konukseven ME 407 Mechanical Engineering Design 17