Session13

Design for
Six Sigma – Contd..
 IIIrd Tool of Development -
Reliability Prediction
• Reliability
– Generally defined as the ability of a
product to perform as expected over
time
– Formally defined as the probability that a
product, piece of equipment, or system
performs its intended function for a
stated period of time under specified
operating conditions
 Reliability
Reliability: Probability of a failure occurring in operational
use.
Perceived reliability: Depends upon:
user behavior
set of inputs
pain of failure
 Reliability Metrics
Traditional Measures
• Mean time between failures
• Availability (up time)
• Mean time to repair
Market Measures
• Complaints
• Customer retention
User Perception is Influenced by
• Distribution of failures
Hypothetical example: Cars are less safe than airplanes in
accidents per hour, but safer in accidents per mile.
 Requirements Specification of
System Reliability
Example: ATM card reader
Failure class Example Metric
Permanent System fails to operate 1 per 1,000 days
non-corrupting with any card -- reboot
Transient System can not read 1 in 1,000 transactions
non-corrupting an undamaged card
Corrupting A pattern of Never
transactions corrupts
database
Cost of Improved Reliability

Rs.

Up time
99% 100%

Will you spend your money on new functionality or

improved reliability?
Example: Central Computing
System
A central computer serves the entire organization.
Any failure is serious.
Step 1: Gather data on every failure
• 10 years of data in a simple data base
• Every failure analyzed:
hardware
software (default)
environment (e.g., power, air conditioning)
human (e.g., operator error)
Example: Central Computing
System
Step 2: Analyze the data
• Weekly, monthly, and annual statistics
Number of failures and interruptions
Mean time to repair
• Graphs of trends by component, e.g.,
Failure rates of disk drives
Hardware failures after power failures
Crashes caused by software bugs in each module
Example: Central Computing
System

Step 3: Invest resources where benefit will be

maximum, e.g.,
• Orderly shut down after power failure
• Priority order for software improvements
• Changed procedures for operators
• Replacement hardware
 Reliability
Definition
– “The probability of a product performing without failure a
specified function under given conditions for a specified
period of time”
– In two formats
• Probability that system will perform on a given trial
– 99.999% for telephone system (Five-9 system)
• Frequency of successful performances of a system in a
given number of attempts.
– 10,000 operating hours
 Reliability – Contd..
Three dimensions
– Probability:
• frequency of successful uses out of a certain
number of attempts
• Likelihood of an item lasting a given amount of time
– Failure
• A situation in which an item does not perform as
intended
– Operating conditions
• How the product should be used?
• What is “normal operating conditions”
 Failures
• Definition
– Very clear in simple system, for example, light bulb
• No longer produces light
– Not very clear in other systems, for example,
automobile tire
• Disastrous failure? Blowout? Useful life?
• It is a good idea to use prescribed operating
conditions
– Some products have different reliability at different
condition.
– For example, some light bulb designed specially for
outdoor!!
 Why thing fails?
-- Design failure
• Because a characteristics of demand was overlook
and miscalculated
• A production line cannot cope with the demands
placed upon it
-- Facilities failure
• All the facilities of the operation are liable to break
down
• Such as machine, equipment, building and fitting
-- Staff failure
• Error - are mistakes in judgment
• Violation - contrary to defined procedures
 Why thing fails?
– Supplier failure
• Delivery, quality of incoming goods and service

– Customer failure
• Misuse of the products
• Misuse of service which the operation has created
for
• Inattention or incompetence
 Types of Failures
• Functional failure – failure that occurs at the
start of product life due to manufacturing or
material detects
• Reliability failure – failure after some period of
use

Types of Reliability
Inherent reliability – predicted by
product design.
Achieved reliability – observed during
use.
 Measuring failure
• There are 3 ways to measure failure:
– Failure rate - how often a failure occur (i.e. FR(%), FR(N))
– Reliability - the chance of a failure occur (MTBF)
– Availability - the amount of available useful operating time
• MTBF / (MTBF+MTTR)

• Failure rate (FR)

– calculated as the number of failures over a period time
• FR (N) = number of failures x 100% / operating time
• FR(%) = number of failures x 100% / total number of products
tested
 Measuring failure
• Example 1
To compute the value of FR, suppose that 10 units are
tested over 100 hours periods. Four units failed with
one unit each failing after 6, 35, 65 and 75 hours, the
remaining six units performed satisfactory until the
end of the test.
FR(N) = 4/ 781
= 0.0051 failure per hour

Total operation hours = 6x100+1x6+1x35+1x65+1x75 =

781
 MTBF
• An alternative measures of failure of a component
system is MTBF
• Defined as “the average length of time between failures
of a product or component”
– MTBF is the reciprocal of failure rate (FR(N))
– MTBF = operating hours / number of failure (unit hours)

• Example2
– In electronics components, the FR(N) was 0.000041. What is its
MTBF?

– MTBF = 1/F(N)) = 1/ 0.000041 = 24390 hours

– That is a, a failure can be expected once every 24390 hours on
average
 Availability
• Availability is the probability that a system is
operating satisfactorily at any point in them
when used under stated conditions.

• Availability = MTBF / (MTBF+MTTR)

• MTTR = the mean time to repair, which is the
average time taken to repair the operation, from
the time it fails the time it is operation again

• What is the causes of unavailability?

– Planned maintenance
– Change over
 Reliability and Availability

• A simple measure of reliability can be given as:

MTBF = MTTF + MTTR , where

MTBF is mean time between failures

MTTF is mean time to fail
MTTR is mean time to repair
 Example 3
• 20 air conditioning systems designed for use by a company in
NASA space shuttles were operated for 1000 hours. Two of
the system failed during the test – One after 200 hours and the
other 600 hours. Find the FR(%), FR(N) and MTBF
Solution
FR(%) = No of failure x 100%
No of units tested
= 2/20 = 10%
FR(N) = No of failure / total operation hours
= 2 / (18x1000 + 800+ 400) = 2/19200 = 0.00104 / per hour
MTBF = 9,600 hours
 Failure Rate Curve

“Infant
mortality
period”

Many electronic components commonly exhibit a high, but

decreasing, failure rate early in their lives
 Reliability Function
• The reliability function represents the
probability that an item will not fail within
a certain period of time, T. It is directly
related to the cumulative distribution
function: F(T) = 1 - e-T, that yields the
probability of failures. Since F(T) is the
probability of failure, the reliability
function, R(T) can be defined as the
complement, e.g. probability of not
failing:
• R(T) = 1 - (1 - e-T) = e-T
 Problem
• Assuming an exponential distribution, a
particular light bulb has a failure rate of
0.002 units per hour. What is the
probability of failure within 400 hours?
What is the reliability function?
• Data
• As shown in the problem statement:
 Problem
Analysis ---
The reliability function is R(T) = 1 - F(T) = e-T ;  =
0.002;
So: R(T) = 1 - e-0.002 T
R(T) = P(x < 400) = 1 - e-0.002 (400) = 1- 0.449 = 0.551

Conclusion ---
• This shows that there is a probability of 0.551 of the bulb
failing within 400 hours.

Recommendation ---
• Use these statistics for process analysis and adjustment
 Reliability Prediction
• System of components may be configured in
– Series
– Parallel
– Of mixed combination
• Series Systems
– All components must function or the system will fail
– The reliability of the system is the product of the individual
reliabilities, that is
– Rs = R1 x R2 x R 3 x … R n

Example
• A personal computer is compose of the CPU, modem,
and printer with reliability of 0.997, 0.98 and 0.95
respectively. The overall reliability of the system is given
by
• Rs= 0.997 x 0.98 x 0.975 = 0.953
 Example (service industry)
• The HSBC Bank processes Loan application
through 3 clerks set up in series:
R1 R2 R3
0.9 0.8 0.99 Rs
Rs = R1 x R2 x R3
= 0.8 x 0.9 x 0.94
= 0.713
= 71%
 Series Systems
1 2 n

RS = R1 R2 ... Rn

Parallel Systems

n
RS = 1 - (1 - R1) (1 - R2)... (1 - Rn)
 Reliability Prediction
• Parallel system:
– The system will operate as long as one component functions
– The additional components are “redundant”
– In this kind of system, failure of an individual component is less critical
than in series system;
– Redundant is often build into system to improve the reliability
– Rs= Rm + Rb (1- Rm ) =R(main) + R(backup)xFail(main)
• Where Rs = reliability of the system
• Rm = reliability of the main sub-system
• Rb = reliability of the stand-by sub-system
• (1- Rm ) = probability that the main sub-system will fail
– For example:
– 0.8 + 0.8 (1-0.8) =0.8 + 0.16=0.96=96%
– Note the reliability has been increased from 80%
to 96%
 Reliability Prediction
• For example
– The computers on the space shuttle were
designed with built-in redundancy in case of
failure. Five computers were designed in
parallel. Thus, if the reliability of each
computer is 0.99. What is the combined
reliability Rs.
– Rs = 1- (1-0.99)5
– Rs = 0.9999999999
 Optimizing Reliability
• Standardization
• Redundancy
• Physics of failure

Tools for Design Verification

• Reliability testing
• Measurement systems evaluation
• Process capability evaluation
 Metrology - Science of
Measurement
• Accuracy - closeness of agreement
between an observed value and a
standard

• Precision - closeness of agreement

between randomly selected individual
measurements
 Accuracy vs. Precision
Accuracy Precision
How well a series of
How well a measurements
measurement agree with each
agrees with an other
accepted value
 Reliability testing
• Life testing
• Accelerated life testing
• Environmental testing
• Vibration and shock testing
• Burn-in (component stress testing)
 Measurement System
Evaluation
• Whenever variation is observed in
measurements, some portion is due to
measurement system error. Some errors are
systematic (called bias); others are random.
The size of the errors relative to the
measurement value can significantly affect
the quality of the data and resulting
decisions.
 Repeatability and
Reproducibility
• Repeatability (equipment variation) –
variation in multiple measurements by an
individual using the same instrument.

• Reproducibility (operator variation) -

variation in the same measuring
instrument used by different individuals
 Repeatability & Reproducibility
Studies
• Quantify and evaluate the capability of a
measurement system
– Select m operators and n parts
– Calibrate the measuring instrument
– Randomly measure each part by each
operator for r trials
– Compute key statistics to quantify
repeatability and reproducibility
 R&R Evaluation
• Under 10% error - OK
• 10-30% error - may be OK
• over 30% error - unacceptable
 Process Capability
• The range over which the natural variation
of a process occurs as determined by the
system of common causes
• Measured by the proportion of output that
can be produced within design
specifications

39
 Types of Capability Studies
• Peak performance study - how a process
performs under ideal conditions
• Process characterization study - how a
process performs under actual operating
conditions
• Component variability study - relative
contribution of different sources of variation
(e.g., process factors, measurement
system)
 Process Capability Study

1. Choose a representative machine or process

2. Define the process conditions
3. Select a representative operator
4. Provide the right materials
5. Specify the gauging or measurement method
6. Record the measurements
7. Construct a histogram and compute
descriptive statistics: mean and standard
deviation
8. Compare results with specified tolerances
 Key Idea

The process capability index, Cp

(sometimes called the process potential
index), is defined as the ratio of the
specification width to the natural
tolerance of the process. Cp relates the
natural variation of the process with the
design specifications in a single,
quantitative measure.
Process Capability Index

Cp = UTL - LTL
6
Cpu = UTL - 
3
Cpl =  - LTL
3
Cpk = min{ Cpl, Cpu }