ISYE6094 – Quality Engineering

Week 9

PROCESS CAPABILITY ANALYSIS

 PROCESS CAPABILITY
ANALYSIS

SUB TOPIC
1. SQC
2. STCEQ Concepts
3. STCEQ Requirement
4. STCEQ Structure
5. Specification Limits And Control Limits
6. Process Capability Analysis
7. Specifications And Process Capability
 1. Statistical Quality
Control (SQC)
• Statistical Quality Control (SQC) has been used for
more than 70 years around the world, and it has
greatly contributed to improving the quality of many
goods and services
• Statistical quality control – SQC (consisting of Statistical
Process Control, Process Capability Studies,
Acceptance Sampling, and Design of Experiments) is a
very important tool to obtain, maintain, and improve
the Quality level of goods and services produced by an
organization.
• Statistical Process Control includes Control Charts,
which monitor process performance, and some
‘‘opportunity tools,’’ like the Ishikawa diagram, and
statistical tools, like the Pareto diagram and the
histogram (those tools are used to find the causes of a
process misbehaving, perhaps an opportunity for
improving its quality).
• Process capability studies measure the process’s ability to produce
items according to specifications, which is crucial for quality
assurance.
• Acceptance Sampling can be defined as the group of statistical
techniques used to accept or reject lots of finished goods or raw
materials received from suppliers, based on random samples taken
from those lots.
• Acceptance Sampling does not estimate lot quality, it just
recommends a course of action: to accept or reject the lot.
• The design of Experiments is a broad statistical issue. Properly
designed statistical experiments can discover what variables are
causing a process to misbehave and also the magnitude of the
effect.
 STCEQ concepts
• an Artificial Intelligence (AI) approach could be a good
option to implement the computer environment for an
Intelligent Tutoring System (ITS). The reason for such a
decision was its ability to create a ‘‘learn-by-doing’’
environment, in which the student/trainee can develop
the needed SQC skills and long-lasting knowledge
• The ITS was called STCEQ. In its current version, STCEQ
includes only problem-solving activities for SPC and
Process Capability Studies. However, tutorials for
Acceptance Sampling and Design of Experiments were
added to provide a complete description of Statistical
Quality Control.
• STCEQ was developed using artificial intelligence to build
a learning environment for Statistical Quality Control, as
part of a teaching model.
• STCEQ is an Intelligent Tutoring System (ITS). Here an ITS
is understood as a computer program that uses artificial
intelligence techniques to simulate human thinking in a
certain domain, attempting to help an ‘‘apprentice’’ in
building strategies to solve problems or make decisions
• The idea was to create an environment in which a user
could freely search for SQC concepts and solve real SQC
problems under supervision (automated supervision).
 STCEQ requirements
• Friendly environment.
STCEQ provides a friendly learning environment, minimizing the time
needed to master the system interaction, thus allowing the user to
get down to business as soon as possible.
• Real SQC problems.
The problems embedded in STCEQ are ‘‘real’’ SQC problems, with
real production flows and real data.
• Supervision.
An SQC expert and a tutor are embedded in STCEQ. The former is
responsible for the ‘‘correct’’ answers to the questions, and the latter
compares the user’s and experts’ answers, calculates users’
performance, and suggests what he should do next
 STCEQ Structure
• STCEQ has been developed as
a stand-alone application
(shortly we intend to build a
web-based version) in a
Windows-based, IBM-PC-
compatible platform. An
expert-system shell,
• Intellicorp Kappa-PC 2.4, has
been used to develop the
prototype. Quality Control or
Statistical Quality Control
course.
• The tutorials are grouped into five main
areas: basic concepts about quality and
TQM, basic concepts about Statistical
Quality Control, Acceptance Sampling,
Statistical Process Control (with its
subdivisions), and Design of Experiments.
• SPC was divided into two major areas for
didactical reasons: Control Charts and
Capability Studies. Control Charts were
divided into Shewhart Variables Charts,
Other Variables Charts, Shewhart
Attributes Charts, and Other Attributes
Charts. The user can return to the main
menu (BACK), quit the system (QUIT), or
go to other screens, each one with all
tutorials from a specific area or
subdivision.
• The problems are grouped into three major types,
according to the techniques involved: Control Charts,
Capability Studies, and Control Charts and Capability
Studies.
• In the current version STCEQ has one difficulty level only:
the user answers questions about the results generated
by STCEQ. The problems that the user previously solved
are shown in a different color; they are highlighted to
prevent the user from solving them again. Moreover,
after knowing the previous performance the user can
make a more careful choice for the next step to take.
• Once a problem is chosen, STCEQ shows its description
to the user, and also a Microsoft PowerPoint
presentation about it. The user can decide to solve the
problem or to return to the previous screen. Once
deciding to solve it, the Problem module sends a
message to the Simulator module, which produces all
the results, like the X-bar and R control charts as shown
in the Figure attached
• The user will answer items based on
statistical results, for instance, if the
process from which the charts were
plotted is under statistical control and
why. The Expert module will also
answer the same questions, and the
Tutor module will compare both
answers (checking if there are
differences between them and how
different are they), evaluate the user’s
performance, and recommend the best
course of action. The user can anytime
quit STCEQ, or return to a previous step.
He/she can also ignore the tutor
STCEQs simulator module results – control charts for
recommendation, and choose not to
averages and ranges.
consult the listed tutorials.
2. SPECIFICATION LIMITS AND
CONTROL LIMITS
 2. SPECIFICATION LIMITS
AND CONTROL LIMITS

• The terms specification limits and tolerance limits are often defined as
acceptable limits on quality characteristics
• Tolerance limits are generally for evaluating manufacturing or service
requirements, while specification limits are for categorizing materials,
products, or services in relation to specified requirements.
• Tolerance limits can be two-sided (with upper and lower limits) or one-
sided with upper or lower limits.
• Specification limits are determined by customer requirements. What
customers want in a product or service is analyzed through market
research and incorporated through product or service design
3. PROCESS CAPABILITY ANALYSIS
 3. PROCESS CAPABILITY
ANALYSIS

• Process capability represents the performance of a process under

statistical control, which is determined by the total variability that exists
due to all common causes present in the system.
• Common causes will always be attached to a system where there is a
special cause.
• Causes must be eliminated to determine the product's ability to meet
customer expectations.
• Process capability studies involve observing product quality
characteristics or are called product analysis studies
• Process capability studies are studies that involve the collection and
analysis of data related to process parameters
 Benefits of Process
Capability Analysis

1. Uniformity of output
• Variable can be tightly controlled.
• Any undesirable shapes in the distribution of quality characteristics are
evaluated possible changes to process parameters are made early.
2. Maintained or improved quality.
• consistent with quality improvement objectives
• Process capability analysis shows whether new equipment is needed
• When changes occur, new capabilities can be determined
3. Product and process design is facilitated.
• Process capability analysis provides critical feedback for design
• Product designers must be aware of the variations inherent in the process
• Inability to design product tolerances by process will result in longer
lead times in design
 Benefits of Process
Capability Analysis

4. Assistance with vendor selection and

control.
• information on the process capabilities
of the vendor is needed in selecting a
vendor.
• for selected vendors, regular reporting
of process capability information is an
effective way to control and improve
quality
5. Reduction of total costs.
• By continuously monitoring process
parameters, fewer nonconforming items
are produced, resulting in reduced
internal and external failure costs
4. PROCESS CAPABILITY INDEX
 4. PROCESS
CAPABILITY INDEX

• A process must first be analyzed to

verify that it is under control before its
capability is estimated.
• The process capability index is an easy
way to understand an aggregate
measure of process performance
• The index combines the location and/or
within-unit variation of a particular
process parameter. process.
 Index Cp

• A common measure to describe the potential of a process to meet

specifications relates the process spread (the difference between natural
tolerance limits) to the specification spread, assuming two-sided
specification limits.
• The formula:

• where USL and LSL represent the upper and lower specification limits,
respectively, and σ represents the process standard deviation
• If σ is unknown, it is replaced by an estimate of it which can be
obtained from the control chart information for the range chart when
the process is under control
 Upper and Lower
Ability Index

• If only a single specification limit is given, then the Index can be

derived from the shift of the average process size against the
process spread to determine the upper specification limit, the upper
capability index (CPU)
• The formula:

• Expected CPU > 1.

 Index CpL, CpU
• Similarly, if only the lower specification limit is given, the lower
capability index (CPL, or Cp is lower)
• The formula:

• Expected CPL > 1.

• The CPU and CPL indices are useful in evaluating process
performance relative to specification limits
• The CPU and CPL indices are useful in evaluating process
performance relative to specification limits, as well as assisting in
determining process parameter settings (such as process mean μ) or
process parameter requirements (such as process standard deviation
σ).
 Example 1

• The relative humidity in the greenhouse is estimated to be between 65%

and 85%. Random samples taken during one week yielded the following
values: 60,78,70,84,81,80,85,60,88,75.
Question :
• Define and interpret the process capability index.

Solution:
• The specification limits are LSL = 65% and USL = 85%. Assume that the
process is under control. The sample mean and standard deviation were
found to be 76.1 and 9.905, respectively.
• Then the Cp index is:

• A Cp value of less than 1 indicates that the process is unable to meet

specifications. Corrective actions that will reduce process variability
must be identified.
• Suppose the only specification is a lower limit of 65%. So it is necessary
to look for a lower ability index. Using the equation and replacing μ and
σ with estimates of X and s, respectively, we get:
• A calculated CPL value, which is less than 1, is undesirable. However, if
process variability cannot be reduced, another option is to increase the
process average such that it is sufficiently above the LSL; that is,
average humidity levels will be well above 65%.
• To increase the CPL to 1, assuming that the process standard deviation
cannot be reduced, the process average target value should be:
 Index Cpk

• One index that takes location into account is the Cp index, used when
the process average does not reach the target value, which is assumed to
be halfway between the specification limits. Cp or Cpk index:

• Cpk represents the scaled distance, relative to three standard deviations

(i.e., half the process spread), between the process average and the
nearest specification limit.
• The desired value is Cpk > 1. While the Cpk index represents the
potential of the process, the Cpk value represents the actual capability of
the process with the existing parameter values.
• If management assigns equal significance to values that are above the
USL or below the LSL, the optimal setting for the process mean is
midway between the specification limits. This measurement of the
deviation of the process mean from the target value is incorrectly given
by the scaled distance:

• Where, m=(USL+LSL)/2
• The estimate of k is obtained by using the sample mean X as an estimate
of the process mean μ
 Relationship between
Cp dan Cpk

• If LSL < μ < USL, we observe that 0 < k < 1. If the process average is at
the target value m, then k = 0 and Cpk = Cp. If the process mean is at
USL or LSL, k = 1 and Cpk = 0
 Example 2

• In electrical circuits, the capacitance of a component must be between

25 and 40 picofarads (pF). A sample consisting of 25 components
yields a mean of 30 pF and a standard deviation of 3pF. Calculate the
process capability index Cpk and comment on the process
performance.
Question :
• If the process is nonconforming, what proportion of the product is
nonconforming, assuming a normal distribution of characteristics?
 Answer

• Solution:
• The Cpk index is estimated:
•

• Since Cpk<1, the process cannot produce only products that comply
with the current settings. Corrective actions will bring the process
towards the target value of 32.5pF or, if possible, reduce process
variability.
 Answer

• The standardized normal value (Z value) on LLS is:

•

• The standard values in USL are:

 Answer

• After checking the standardized normal distribution in Appendix A-3,

it is found that the area below the LSL is 0.0475, and above the USL
is 1-0.9996=0.0004 (which is negligible).
• Thus, 4.75% of products are below LSL. Process performance can be
improved if the process average is shifted to the target value of
32.5pF. In this case, the Z value in LSL will be ZLSL=(25–32.5)/3=-
2.50, and in USL it will be ZUSL=2.50.
• As per Appendix A-3, the proportion of products below the LSL is
0.0062, with the same proportion above the USL. The total
proportion that did not comply was 0.0124, or 1.24 %.
 Conclusion

1. Statistical Quality Control (SQC) has been used for more than 70 years around the
world, and it has greatly contributed to improving the quality of many goods and
services
2. Statistical quality control – SQC (consisting of Statistical Process Control,
Process Capability Studies, Acceptance Sampling, and Design of Experiments)
3. An artificial Intelligence (AI) approach could be a good option to implement the
computer environment for an Intelligent Tutoring System (ITS) or STCEQ. In its
current version, STCEQ includes only problem-solving activities for SPC and
Process Capability Studies.
4. STCEQ requirements : Friendly environment, Real SQC problems, Supervision.
5. STCEQ has been developed as a stand-alone application (shortly we intend to
build a web-based version) in a Windows-based, IBM-PC-compatible platform.
 Conclusion

1. This topic has discussed important aspects relating to all processes of

determining the capabilities inherent in the process.
2. Capability analysis must be carried out on controlled processes, making it
possible to conclude whether the process will be able to produce items
that meet the desired specifications.
3. For Limitations, there are two sides, the process spread must be smaller
than the specification spread for the process to be considered capable,
where the process must be focused on the most desired locations to
minimize the production of non-conforming items.
4. For one-sided specification limits, process averages and spreads must
change the process average, which requires adjusting certain process
parameters, usually easier than reducing process variability.
5. Process sprawl can be reduced through fundamental changes in processes
initiated by management.
 REFERENCES

• Fundamentals of Quality Control and Improvement, Fourth

Edition, AMITAVAMITRA, Auburn University College of Business
Auburn, Alabama, Copyright 2016 by John Wiley & Sons, Inc. All
rights reserved. ISBN978-1-118-70514-8(cloth) Published by John
Wiley & Sons, Inc., Hoboken, New Jersey. Published
simultaneously in Canada
• K. S. Krishnamoorthi. (2019). A first course in quality engineering:
integrating statistical and management methods of quality. 03.cRc
press. Boca Raton. ISBN -13: 978-1-4987-6420-9
• Douglas C. Montgomery. (2013). Introduction to Statistical quality
control . 7th Edition John Wiley & Sons, Inc.. ISBN:
9781118146811.
• David L Goetsch (2013). Quality management for organizational
excellence : introduction to total quality. 07. pearson Education.
Boston. ISBN: 9780132870979
Thank You