INDUSTRIAL QUALITY

CONTROL
PROCESS CAPABILITY ANALYSIS
 LECTURE OBJECTIVES

 Know some of the commonly used process capability

measures.

 Know and understand the procedures used for their

computation and how to interpret them.

 Know and understand the different assumptions behind

their computation and use.
PROCESS CAPABILITY ANALYSIS
 INTRODUCTION

 The terms specification limits and tolerances are often used

interchangeably and are defined as the acceptable bounds on
quality characteristics.

 And they could be based on meeting customer needs or

functional requirements of the product/service.

 Tolerance limits are generally preferred in evaluating

manufacturing or service requirements, whereas specification
limits are more appropriate for categorizing materials,
products or services in terms of their stated requirements.
PROCESS CAPABILITY ANALYSIS
 Control limits identify the variation that exists between
samples, or subgroups, of measurements.

 Control limits reflect the variability of the process and have

no relationship to specification limits, which are chosen to
meet customer needs for the product or service.

 The figure below gives some indication and clarity on the

difference between control limits and specification limits.

 It shows a process could be in control, while its output not

meeting customer specification limits.
PROCESS CAPABILITY ANALYSIS
 Difference between control limits and specification
limits:
PROCESS CAPABILITY ANALYSIS
 PROCESS CAPABILITY ANALYSIS:

 The determination of process capability begins only after the

process has been brought to a state of statistical control.

 PROCESS CAPABILITY:

 Is determined by the total variability that exists because of all

common causes present in the system.

 Thus, it can also be viewed as the variation in the product

quality characteristic that remains after all special causes have
been removed.
PROCESS CAPABILITY ANALYSIS
 This allows us to determine the ability of the product to meet
customer expectations (specifications and tolerances).

 PROCESS CAPABILITY ANALYSIS:

 Involves estimating the process mean and standard deviation of

the quality characteristic.

 A true process capability study should involve collecting and

analyzing data that relate to process parameters e.g. depth of
cut, type of tool used, tool material, type of jigs or fixtures, or
rate of feed of the tool.

 Any problem with product characteristics can then be related to

process parameters, and remedial actions can be identified on a
timely basis.
PROCESS CAPABILITY ANALYSIS
 BENEFITS PROCESS CAPABILITY ANALYSIS:

 Uniformity of output.

 Maintained or improved quality.

 Product and process design facilitated.

 Assistance in vendor selection and control.

 Reduction in total costs.

PROCESS CAPABILITY ANALYSIS
 NATURAL TOLERANCE LIMITS:

 NTLs are established or influenced by the process itself also

called process capability limits.

 They represent the inherent variation in the quality

characteristic of the individual items produced by a process in
control.

 They are estimated based on the population of the values.

 For a normal distribution of the quality characteristic X, the

upper natural tolerance limit (UNTL) = µ + 3σ.

 And the lower natural tolerance limit (LNTL) = µ - 3σ.

PROCESS CAPABILITY ANALYSIS
 Where µ represents the process mean and σ represents the
process standard deviation, which is the standard deviation of
the individual items.

 Figure below shows the upper and lower natural tolerance

limits, or process capability limits.

 The assumption of a normal distribution of characteristic X

implies that nearly all (approx. 99.74%) of the items produced
will have a value within the bounds of the natural tolerance
limits.

 If the distribution is non-normal, first determine whether it

conforms to any known distribution.
PROCESS CAPABILITY ANALYSIS
 Natural tolerance limits and process spread:
PROCESS CAPABILITY ANALYSIS
 Then using the distribution that models the characteristic, the
NTLs can be found such that nearly the entire distribution is
contained within those limits.

 SPECIFICATIONS AND PROCESS CAPABILITY:

 Technically, there might not be any mathematical relationship

between process capability limits (or natural tolerance limits)
and specification limits.

 There is however a desirable relationship between these two

sets of limits.

 The specification limits should be preferably outside the

process capability (natural tolerance) limits, in which case
most of the units produced will be acceptable.
PROCESS CAPABILITY ANALYSIS
 There are three cases that arise regarding this relationship
and assuming a normal distribution of the quality
characteristic.

 Case 1: Process Spread Less than Specification Spread.

 In this scenario the process spread is less than the specification

limits, and the process is quiet capable (see the figure below).

 Case 2: Process Spread equal to Specification Spread.

 In this scenario the process spread is the same as the difference

between the specification limits, and we have an acceptable or
adequate situation in which there is no room for error (see the
figure below)
PROCESS CAPABILITY ANALYSIS
 Case 1: Process spread is less than specification spread:
PROCESS CAPABILITY ANALYSIS
 Case 2: Process spread is equal to specification spread:
PROCESS CAPABILITY ANALYSIS
 Case 3: Process Spread Greater than Specification Spread:

 In this scenario an undesirable situation exists when the

process spread is greater than the difference between
specification limits (see the figure.)

 The process could still be in control but there will be some

proportion of items not meeting specifications.
PROCESS CAPABILITY ANALYSIS
 Case 3: Process Spread Greater than Specification
Spread:
PROCESS CAPABILITY ANALYSIS
 There are several corrective approaches that could be taken
to remedy undesirable situation with some including the
following;

 1. The possibility of increasing the specification limits can be

explored (discussion with the customer).

 2. Measures to reduce the process spread can be investigated

(capital investment, better raw material, experienced
operators).

 3. If it is not economically feasible to reduce the process

variability through large capital investments, an alternative
may be to shift the process average to achieve a desirable
balance in the proportion of scrap and rework.
PROCESS CAPABILITY ANALYSIS
 PROCESS CAPABILITY INDICES:

 A process should first be analyzed to verify that it is in control

before its capability is estimated.

 This enables us to estimate the proportion of nonconforming

product using the right probability distribution function.

 There are several tests for pdf, for normality, normal

probability plots (pp plots can be used) as an example.

 The process capability index is an easily understood

aggregate measure of the goodness of the process
performance.
PROCESS CAPABILITY ANALYSIS
 The ability to meet specifications is the criterion used for
measuring the attractiveness of the process.

 Cp Index:

 It relates the process spread to specification spread, assuming

two-sided specification limits.

 It is given by:

𝑈𝑆𝐿 −𝐿𝑆𝐿
 Cp =
6𝜎

 Where USL and LSL represent the upper and lower specification
limits, respectively, and 𝜎 represents the process standard
deviation.
PROCESS CAPABILITY ANALYSIS
 When process 𝜎 is unknown it is replaced by its estimate, 𝜎,
which can be obtained by several means, some given below:

𝑅
 𝜎= or;
𝑑2

𝑠
 𝜎=𝑐
4

 Where c4 and d2 are factors used in the construction of standard

deviation and range charts respectively.

 A process that is centered between specification limits will

produce a minimum proportion of items that fall outside those
limits.
PROCESS CAPABILITY ANALYSIS
 It is desirable to have a Cp ≥ 1, indicating that the process is more
than capable.

 When Cp = 1, the process spread equals the specification spread,

and the process is said to be barely capable (case II).

 If the process is centered, only (1 – 99.74% = 0.26%) of the parts

will fall outside the specification limits.

 If the process is not centered, it is possible that even for a

process with Cp > 1, some proportion of the product will be
nonconforming.

 Though when Cp > 1, there is a great flexibility that the process

can go out of control, but still produce conforming items.
PROCESS CAPABILITY ANALYSIS
 If Cp < 1, it implies that the inherent variability in the process ,
as measured by the process spread 6σ, is greater than the
specification limit.

 For this situation, a process can be in control and still not

meet specifications (as evidenced by case III).

 From its formula it can be noticed that Cp is a measure of

process potential and not process performance.

 Therefore other capability indices are needed to measure

process performance such as CPU, CPL, Cpk, Cpm, and Cpmk.
PROCESS CAPABILITY ANALYSIS
 UPPER AND LOWER CAPABILITY INDICES:

 If only a single specification limit is given, indices can be

derived that measure shifts in the process mean 𝜇, relative to
the process spread.

 For a given upper specification limit, the upper capability

index (CPU, or Cp upper) is given by:

𝑈𝑆𝐿 − 𝜇
 CPU =
3𝜎

 It is desirable to have a CPU ≥ 1.

PROCESS CAPABILITY ANALYSIS
 Assuming normality, if CPU = 1, only 0.13% of the product will
be above the USL and thus nonconforming.

 Similarly, if only the lower specification limit is given, the

lower capability index (CPL, or Cp lower) is given by:

𝜇 −𝐿𝑆𝐿
 CPL =
3𝜎

 It is also desirable to have a CPL ≥ 1.

 In the above calculations, if the process parameters 𝜇 and 𝜎

are unknown, the sample mean 𝑋 and the sample standard
deviation s, respectively are used as estimates.
PROCESS CAPABILITY ANALYSIS
 The indices CPU and CPL are useful in evaluating the process
performance relative to the specification limits.

 Some recommended values for the process capability indices

are shown in table below.
PROCESS CAPABILITY ANALYSIS
 Many companies prefer a Cp ≥ 1.33.

 This ensures an extremely low (0.007%) nonconformance

rate.

 A Cp value of at least 1.67 implies a process standard deviation

of no more than one-tenth of the specification spread.

 Companies designing a goal of a “six sigma process” are

aiming for a Cp value of 2.

 Cpk Index

 Although the Cp index does not incorporate the process

location, other indices do.
PROCESS CAPABILITY ANALYSIS
 One index that accounts for this location, the Cpk index, is
used when the process mean is not at the target value, which
is assumed to be half-way between the specification limits.

 The Cpk index, is given by:

𝑈𝑆𝐿 − 𝜇 𝜇 −𝐿𝑆𝐿
 Cpk = min ,
3𝜎 3𝜎

 = min 𝐶𝑃𝑈,𝐶𝑃𝐿

 It can be seen from the equation that Cpk represents the

scaled distance, relative to 3 standard deviations , between
the process mean and the closest specification limit.
PROCESS CAPABILITY ANALYSIS
 Desirable values are Cpk ≥1

 Whereas the Cp index represents the process potential, the Cpk

value represents the actual capability of the process with the
existing parameter values. It measures process performance.

 Figure below shows the distribution of a quality characteristic X

for a process that is not capable (Cpk < 1).

 As can be seen from this figure, some proportion of the product

will lie above the USL, even though it can be noted the process
spread 6𝜎 is less than the specification spread.

 With the Cp value greater than 1, indicating that the process can
potentially meet specifications.
PROCESS CAPABILITY ANALYSIS
 Process that is not capable (Cpk < 1):
PROCESS CAPABILITY ANALYSIS
 A corrective measure could be to adjust the process mean
downwards toward the midpoint m of the specification limits.

 Then virtually all the products will then be conforming.

 A measure of the deviation of the process mean from this

target value m is given by the scaled distance k;

𝑚−𝜇
 k=
(𝑈𝑆𝐿 − 𝐿𝑆𝐿)/2

 Where m = (USL + LSL)/2

PROCESS CAPABILITY ANALYSIS
 The relationship between Cp and Cpk is given by:

 Cpk = Cp(1 – k)

 If LSL ≤ 𝜇 ≤ USL, we observe that 0 ≤ k ≤ 1.

 If the process mean is at the target value, m, then k = 0 and

Cpk = Cp.

 If the process mean is at the USL or LSL, k = 1, and Cpk = 0.

 How do the Cp and Cpk indices compare?

 Cp ≤ Cpk - always
PROCESS CAPABILITY ANALYSIS
 The Cp value is a measure of the process potential; it does not
change as the process mean changes.

 A Cp ≥ 1 is desirable, and a value of Cp < 1 indicates that the

process is not capable.

 The Cpk value incorporates both the process mean and the
standard deviation to measure actual process performance.

 If the process is centered (i.e., the process mean is equal to

the target value), Cpk = Cp.

 A standard for benchmarking processes is a value of Cpk = 1, in

which case practically all the product is conforming.
PROCESS CAPABILITY ANALYSIS
 When the process mean is outside the specifications, Cpk is
negative.
PROCESS CAPABILITY ANALYSIS
 LECTURE TUTORIAL

 1. The diameter of a part has to fit an assembly. The

specifications for the diameter are 5 ± 0.015 cm. The samples
taken from the process in control yield a sample mean 𝑋 = 4.99
cm and a standard deviation s of 0.004 cm. Find the natural
tolerance limits of the process. Would you consider adjusting the
process center?

 1.a. Find the natural tolerance limits of the process.

 1. b. Would you consider adjusting the process center?

PROCESS CAPABILITY ANALYSIS
 LECTURE TUTORIAL CONT…..

 2. The relative humidity in a greenhouse is expected to be

between 65 and 85%. Random samples taken over a span of one
week yield the following values: 60, 78, 70, 84, 81, 80, 85, 60, 88, 75.

 2. a. Find and interpret the process capability index, assuming that

the process is in control.

 2. b. Supposing that the only vital specification limit to the

customer is the LSL; Find the lower capability index.

 2. c. If this is still not capable of meeting customer specifications,

what suggestions can you make to improve the situation, given that
the process standard deviation cannot be reduced?
PROCESS CAPABILITY ANALYSIS
 LECTURE TUTORIAL CONT……

 3. In an electrical circuit, the capacitance of a component

should be between 25 and 40 picofarads (pF). A sample of 25
components yields a mean of 30 pF and a standard deviation of
3 pF.

 3. a. Calculate the process capability index Cpk and comment on

the process performance.

 3. b. If the process is not capable, what proportion of the

product is nonconforming, assuming a normal distribution of the
characteristic?