See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286440848

Process Capability and Process Performance

Conference Paper · June 2006

CITATIONS READS

0 387

4 authors:

Kamran. Rezaie mm Rabbani

University of Tehran Semnan University
39 PUBLICATIONS   533 CITATIONS    112 PUBLICATIONS   1,272 CITATIONS   

SEE PROFILE SEE PROFILE

Mohammad Reza Taghizadeh Yazdi Ostadi Bakhtiar

University of Tehran Tarbiat Modares University
64 PUBLICATIONS   21 CITATIONS    23 PUBLICATIONS   199 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

‫ ﮔﺎﻧﻪ ﺷﻬﺮ ﺗﻬﺮان ﺑﻪ ﻣﻨﻈﻮر اﺳﺘﻘﺮار ﻣﺮاﮐﺰ ﻣﺎﻟﯽ و ﺗﺠﺎری‬22 ‫ ﺑﻪ ﮐﺎرﮔﯿﺮی روش ﻫﺎی ﺗﺼﻤﯿﻢ ﮔﯿﺮی ﭼﻨﺪ ﺷﺎﺧﺼﻪ ﺑﺮای رﺗﺒﻪ ﺑﻨﺪی اﻗﺘﺼﺎدی ﻣﻨﺎﻃﻖ‬View project

Operations Research- Simulation View project

All content following this page was uploaded by Mohammad Reza Taghizadeh Yazdi on 10 December 2015.

The user has requested enhancement of the downloaded file.

 The 36th CIE Conference on Computers & Industrial Engineering

Process Capability and Process Performance

K. Rezaie, M. Rabbani, M.R. Taghizadeh, B. Ostadi

Dep. of Industrial Engineering, Faculty of Engineering, Univ. of Tehran, Tehran, Iran.

P.O. Box: 11365/4563, Tehran, Iran, +98 21 88021067
krezaie@ut.ac.ir

Abstract

Once a process is in statistical control, that is producing consistently, you probably then want to
determine if it is capable, that is meeting specification limits and producing “good” parts. Process
capability indices are useful for assessing the capability of manufacturing processes. In this paper, the
process capability indices and their estimators are presented and related to process parameters. Also, the
capability indices have interpreted and steps in determining process capability have presented. Finally,
process performance indices are presented.

KEY WORDS : Process capability indices, process performance, estimators, steps of implementation

1 Introduction

Once a process is in statistical control, that is producing consistently, you probably then want to
determine if it is capable, that is meeting specification limits and producing “good” parts. You determine
capability by comparing the width of the process variation with the width of the specification limits. The
process needs to be in control before you assess its capability; if it is not, then you will get incorrect
estimates of process capability. You can assess process capability graphically by drawing capability
histograms and capability plots. These graphics help you assess the distribution of your data and verify
that the process is in control. You can also calculate capability indices, which are ratios of the
specification tolerance to the natural process variation. Capability indices, or statistics, are a simple way
of assessing process capability. Because they are unitless, you can use capability statistics to compare the
capability of one process to another.

Process capability compares the output of an in-control process to the specification limits by using
capability indices. The comparison is made by forming the ratio of the spread between the process
specifications (the specification "width") to the spread of the process values, as measured by 6 process
standard deviation units (the process "width"). Process capability attempts to answer the question: can we
consistently meet customer requirements? The number one limitation of process capability indices is that
they are meaningless if the data is not from a controlled process. The reason is simple: process capability
is a prediction, and you can only predict something that is stable.

Process capability indices, as a process performance measure, have become very popular in assessing the
capability of manufacturing processes in practice during the past decade. More and more efforts have
been devoted to studies and applications of process capability indices. For example, Rado (1989)
demonstrated how Imprimis Technology, Inc. used the process capability indices for program planning
and growth to enhance product development. The Cp and Cpk indices have been used in Japan and in the
US automotive industry such as Ford Motor Company (Kane, V.E., 1986a, 1986b). For more information
on Process capability indices, see Kotz and Johnson (1993), Kotz et al. (1993) and the references cited
therein.

5431
 The 36th CIE Conference on Computers & Industrial Engineering

Because the sample mean and the sample variance S2 are not unitless, they can be cumbersome as
summary statistics for the process location and variance, respectively. This is particularly true in a
manufacturing process with many characteristics to be examined. Capability indices are unitless and
associate the process location and variance with one-sided or two-sided specifications, with or without a
target value for the process mean. These indices provide an effective means for communicating
assessments of the process capability. The capability indices relate the manufacturer’s specifications to
the natural tolerance of six standard deviations used in US quality control literature (Breyfogle, 1996;
Bissell, 1990).

In this paper, a number of process capability indices and their estimators will be presented, and
consideration will be made of some distributional properties of the estimators. Also, the capability indices
have interpreted and steps in determining process capability have presented. Finally, process performance
indices are presented.

2 Process Capability Indices

Process capability studies are used for monitoring the capability of a process. This implies that it has to be
based on some sort of collection of data from the process. In order to get a fair picture of the capability of
the process, it has to be stable when the data is collected. After the collection of data from a stable
process, the data may be assessed in several ways. One way to do the assessment is to use process
capability indices, which provide numerical measures of the capability.

2.1 Cp Index

The process capability index Cp is defined to be

USL - LSL
Cp =
6σ

where USL, LSL, and σ denote the upper specification limit, lower specification limit, and process
standard deviation associated with the measurements, respectively. A process is said to be capable if the
value of Cp associated with the process is at least 1.0 (Kane, 1986). Since the process standard deviation
is rarely known, it is estimated from a sample of n measurements X1… Xn and an estimate Ĉ P of the
process capability Cp is obtained by

USL − LSL
Cˆ P =
6σˆ

Typically, the sample standard deviation

12
⎡ 1 n
2⎤
S=⎢
⎢⎣ n − 1
∑ (X
i =1
i −X ) ⎥
⎥⎦
1
∑
n
is used to estimate σ (where X = X i ).
n i =1

2.2 C *p Index

5432
 The 36th CIE Conference on Computers & Industrial Engineering

The C *p index is the generalization of Cp to the case where the target value T for the process mean is not
necessarily equal to the midpoint m of the specification limits, and is defined by
min (T - LSL,USL - T)
C *p =
3σ
An estimate of can be obtained by

min (T - LSL,USL - T)
Ĉ *p =
3S
Note that when T =m, C *p = Cp, and Ĉ *p = Ĉ P .

2.3 CPL and CPU Indices

The lower process capability index (CPL) is defined to be

µ − LSL
CPL =
3σ
and the upper process capability index (CPU ) is defined to be
USL − µ
CPU =
3σ
Estimates of CPL and CPU, respectively, are obtained by

X − LSL
CP̂L =
3S
And
USL − X
CP̂U =
3S

Note that
Cp =
(CPL + CPU ) , Ĉ p =
(CP̂L + CP̂U )
2 2
The CPU index was developed in Japan and is utilized by a number of Japanese companies (Kane, 1986).

2.4 CPL* and CPU* Indices

The CPL* and CPU* indices are generalizations of CPL and CPU to the case where the target value T for
the process mean is not necessarily equal to the midpoint m of the specification limits. CPL* is defined
by
⎧ T − LSL ⎛ T −µ ⎞
⎪ ⋅ ⎜1 − ⎟, if T − µ > T − LSL
CPL* = ⎨ 3σ ⎜ T − LSL ⎟
⎝ ⎠
⎪
⎩0 , otherwise

and CPU* is defined by

⎧USL−T ⎛ T − µ ⎞
⎪ ⋅ ⎜1− , if T − µ > USL−T
CPU = ⎨ 3σ ⎜⎝ USL−T ⎠
*

⎪
⎩0, otherwise

Estimates of CPL and CPU, respectively, are obtained by

5433
 The 36th CIE Conference on Computers & Industrial Engineering

⎧ ⎛ T − X ⎞⎟
⎪⎪ T − LSL ⋅ ⎜1 − , if T − X > T − LSL
ˆ *
CPL = ⎨ 3S ⎜
⎜ T − LSL ⎟⎟
⎪ ⎝ ⎠
⎩⎪0, otherwise

⎧ ⎛ T−X ⎞
⎪⎪USL − T ⋅ ⎜1 − ⎟ , if T − X > USL − T
CP̂U * = ⎨ 3S ⎜ USL − T ⎟
⎪ ⎝ ⎠
⎩⎪0 , otherwise

Note that when µ = T , CPL* = CPL and CPU * = CPU

2.5 Cpk Index

The process capability index Cpk is defined as

C pk = min (CPL , CPU )

Cpk describes a distance scaled by 3σ, between the process mean and the closest specification limit.
Assuming that µ is between the specification limits, let

k = 2 m − µ / (USL − LSL )
where m is the midpoint of the specification limits. The ratio k is used in Japan as an index describing the
amount that the process mean is off-center (Kane, 1986). It is easily seen that if 0 ≤ k ≤ 1 , then

C pk = (1 − k )C p
From this equation it is seen that Cpk is Cp reduced by the factor (1-k), where k is a scaled distance
between the process mean and the midpoint of the specification limits. The indices k and Cpk are
estimated by

( ) ( )
kˆ = 2 m − X / (USL − LSL ) , Ĉ pk = min CP̂L , CP̂U = 1 − k̂ Ĉ p
respectively. Note that when µ=m, Cpk= Cp.

2.6 C *pk Index

The C *pk index is the generalization of Cp to the case where the target value T for the process mean is not
necessarily equal to the midpoint m of the specification limits, and is defined to be

(
C *pk = min CPL* , CPU * )
Assuming that µ is between the specification limits, let

k * = T − µ / min(T − LSL ,USL − T )

The ratio k* is the generalization of k to the case where T is not necessarily equal to m. It is easily seen
that if 0 ≤ k ≤ 1 ,then

( )
C *pk = 1 − k * C *p

5434
 The 36th CIE Conference on Computers & Industrial Engineering

From this equation it is seen that C *pk is C *p reduced by the factor (1−k*), where k* is scaled distance that
the process mean is off-target. The indices k* and C *pk are estimated by

k̂ * = T − X / min(T − LSL ,USL − T )

and

( ) ( )
Ĉ *pk = min CP̂L* , CP̂U * = 1 − k̂ * Ĉ *p

Note that when µ = T , C *pk = C *p . Also, when T = m , C *pk = C pk .

2.7 Cpm Index

Chan et al. (1988) proposed Cpm as process capability index. The index Cpm is defined to be

USL − LSL
C pm =
6σ ′
(
where USL=upper specification limit, LSL=lower specification limit, σ ′ = E ( X − T )2 )1/ 2
, and T is the
target value for the process mean. Since the parameter is typically unknown, Chan et al. (1988) suggest
that a sample of n measurements be used to estimate by

1/ 2
⎛ n ( X i − T )2 ⎞
σˆ ′ = ⎜
⎜∑ n −1
⎟
⎟
⎝ i =1 ⎠
Consequently, an estimator for the Cpm index is
USL − LSL
Ĉ pm =
6σ̂ ′

2.8 C *pm index

The index C *pm is the generalization of Cpm to the case where T is not necessarily equal to m and is
defined to be
min(USL − T ,T − LSL )
C *pm =
3σ ′
An estimate of C *pm can be obtained by
min(USL − T ,T − LSL )
Ĉ *pm =
3σ̂ ′

3 Steps in Determining Process Capability

3.1 Determine if pecifications are currently being met

1. Collect at least 100 random samples from the process.

2. Calculate the sample mean and standard deviation to estimate the true mean and standard
deviation of the process.

5435
 The 36th CIE Conference on Computers & Industrial Engineering

3. Create a frequency distribution for the data and determine if it is close to being normally
distributed. If it is continue; if not, get the help of a statistician to transform the data or to find an
alternative model.

4. Plot the USL and LSL on the frequency distribution.

5. If part of the histogram is outside the specification limits, consider adjusting the mean to center
the process.

6. If the histogram indicates that the process spread is greater than the specification spread, the
process might not be capable.

3.2 Determine the inherent variability using an r-chart

1. Get at least 40 rational subgroups of sample size, preferably at least 4 or 5.

2. Calculate the ranges for each subgroup, the average range, rbar, and the control limits for an r-
chart. Plot the data.

3. Discard any ranges outside the UCL ONLY if the undesirable special cause is identifiable and
can be removed from the process; otherwise include the offending range(s).

4. Recalculate the average range, rbar, and the control limits.

5. Repeat the last two steps until all ranges are in statistical control.

6. Estimate the process standard deviation, σ , with the formula rbar / d 2 .

7. Using the midpoint of the specifications as the process mean, assuming normality, draw a normal
curve, and estimate the percentage meeting specifications. This step assumes that the mean of the
process can be adjusted or recentered.

8. If the normal curve shows that the specifications can be met, determine what the specifications
are not being met. Create an xbar chart from the same subgroup data and hunt for clues. It may be
as simple as recentering the process. Perhaps special causes are present that can be removed.

9. If the specifications cannot be met, consider changing the process by improvement, living with it
and sorting 100% of the output, centering the mean, or dropping the product totally.

10. Set up a system of xbar-r charts to create future process improvements and process control.

4 Process Performance

Process Performance attempts to answer the question: does this sample from the process meet customer
requirements?

Pp = Process Performance. A simple and straightforward indicator of process performance.

Ppk = Process Performance Index. Adjustment of Pp for the effect of non-centered distribution.

5436
 The 36th CIE Conference on Computers & Industrial Engineering

Process Performance Index basically tries to verify if the sample that you have generated from the
process is capable to meet Customer CTQs (requirements). It differs from Process Capability in that
Process Performance only applies to a specific batch of material. Samples from the batch may need to be
quite large to be representative of the variation in the batch. Process Performance is only used when
process control cannot be evaluated. An example of this is for a short pre-production run. Process
Performance generally uses sample sigma in its calculation; Process capability uses the process sigma
value determined from either the Moving Range, Range, or Sigma control charts.

4.1 Pp Index

The ratio of the tolerance (specification, or permitted amount of variation) to the variation in a sample. A
value of 1 indicates the sample variation exactly equals the tolerance. Values of less than 1 indicate the
allowable variation (the tolerance) is less than the sample variation. Values of more than 1 indicate that
the sample variation is less than the tolerance. Use of Performance Indices is generally discouraged in
favor of Capability Indices wherever possible. your process (its natural limits) relative to its specification
limits (customer requirements). Following are the graphical details and equations quantifying process
capability

Normal distributions:
High Spec − Low Spec
Pp =
6σ X
Best Fit (Johnson) and True-Position (Rayleigh) distributions:

High Spec − Low Spec

Pp =
ordinate0.99865 − ordinate0.00135
Folded Normal distributions:

ordinate0.99865
Pp =
3
where: x is Sample Sigma. Ordinate0.99865 and ordinate0.00135 are the z-values of the non-normal
cumulative distribution curve at the 99.865 percentage point and the 0.135 percentage points,
respectively.

4.2 Ppk Index

A measure of both process dispersion and its centering about the average.

Ppk = Min [Ppl , Ppu ]

where
ZL Z
Ppl = − , Ppu = − U
3 3
Normal Distributions:
X − Low Spec High Spec − X
ZL = , ZU =
σX σX
Non-Normal Distributions:
Z L = Z normal , p , Z U = Z normal ,1− p

5437
 The 36th CIE Conference on Computers & Industrial Engineering

where x-double bar is the Grand Average and σx is sample sigma . Znormal,p and Znormal,1-p are the z-values
of the normal cumulative distribution curve at the p percentage point and the 1-p percentage points,
respectively.

4.3 Pr Index

A measure of the percentage of the tolerance actually used by the sample. Smaller numbers are best.

Pr = 100
Pp
where Pp is the process performance index.

4.4 Ppm Index

A measure similar to the Ppk index that also takes into account variation between the process average and
a target value. If the process average and the target are the same value, Ppm will be the same as Ppk. If the
average drifts from the target value, Ppm will be less than Ppk.

Pp
Ppm =
1+
(x − T )2
σ X2
where T is the process target, x-doublebar is the Grand Average and σx is sample sigma .

5 Differences between Cpk and Ppk

It differs from process capability in that process performance only applies to a specific batch of material.
Samples from the batch may need to be quite large to be representative of the variation in the batch. Cpk is
for short term, Ppk is for long term. Ppk produces an index number (like 1.33) for the process variation. Cpk
references the variation to your specification limits. If you just want to know how much variation the
process exhibits, a Ppk measurement is fine. If you want to know how that variation will affect the ability
of your process to meet customer requirements (CTQ's), you should use Cpk. It could be argued that the
use of Ppk and Cpk (with sufficient sample size) are far more valid estimates of long and short term
capability of processes since the 1.5 sigma shift has a shaky statistical foundation. Cpk tells you what the
process is CAPABLE of doing in future, assuming it remains in a state of statistical control. Ppk tells you
how the process has performed in the past. You cannot use it predict the future, like with Cpk, because the
process is not in a state of control. The values for Cpk and Ppk will converge to almost the same value
when the process is in statistical control. That is because Sigma and the sample standard deviation will be
identical (at least as can be distinguished by an F-test). When out of control, the values will be distinctly
different, perhaps by a very wide margin.

Cp and Cpk are for computing the index with respect to the subgrouping of your data (different shifts,
machines, operators, etc.), while Pp and Ppk are for the whole process (no subgrouping). For both Ppk and
Cpk the 'k' stands for 'centralizing facteur'- it assumes the index takes into consideration the fact that your
data is maybe not centered (and hence, your index shall be smaller). It is more realistic to use Pp & Ppk
than Cp or Cpk as the process variation cannot be tempered with by inappropriate subgrouping. However,
Cp and Cpk can be very useful in order to know if, under the best conditions, the process is capable of
fitting into the specs or not. It basically gives you the best case scenario for the existing process. Cp
should always be greater than 2.0 for a good process which is under statistical control. For a good process
under statistical control, Cpk should be greater than 1.5. As for Ppk/Cpk, they mean one or the other and you
will find people confusing the definitions and you WILL find books defining them versa and vice versa.

5438
 The 36th CIE Conference on Computers & Industrial Engineering

You will have to ask the definition the person is using that you are talking to. I just finished up a meeting
with a vendor and we had a nice discussion of Cpk vs. PPk. We had the definitions exactly reversed
between us. The outcome was to standardize on definitions and move forward from there. My suggestion
to others is that each company has a procedure or document (we do not) which has the definitions of Cpk
and Ppk in it. This provides everyone a standard to refer to for WHEN we forgot or get confused.

6. Conclusions

Process capability indices are getting more and more popular in the sorts of quality and productivity
improvement. Capability indices help to prevent production of items which do not meet the specification
requirements and help to continuously monitor product improvement. Finally, indices provide a common
language for successful communication of information about the process capability and location. In order
to estimate process capability, you must know the location, spread, and shape of the process distribution.
These parameters are, by definition, changing in an out of control process. Therefore, only use Process
Capability indices if the process is in control for an extended period.

References

Rado, L.G., 1989. Enhance product development by using capability indices. Quality Progress., April, pp. 38-41.
Kane, V.E., 1986a. Process capability indices. J. Quality Technol., 18 (1), 41-52.
Kane, V.E., 1986b. Corrigenda. J. Quality Technol., 18 (4), 265.
Kotz, S., Johnson, N.L., 1993. Process capability indices. Chapman & Hall, London.
Kotz, S., Pearn, W.L., Johnson, N.L., 1993. Some process capability indices are more reliable than one might think.
J. Roy. Statist. Soc. 42 (1), 55-62.
Kane, V.E., 1986. Process capability indices, J. Qual. Technol., 18, 41–52.
Chan, L.K. S.W. Cheng, F.A. Spiring, 1988. A new measure of process capability: Cpm, J. Qual. Technol., 20,
162–175.
Bissell, 1990. How Reliable is Your Capability Index. Royal Statistical Society.
Breyfogle F., 1996. Measurement of Process Capability. Smarter Solutions.

5439

View publication stats