Process Capability Analysis

Process Capability

• Used to assess the degree to which the output of a

process conforms to specifications

• Natural spread of a process is defined as 6 sigma

• Determined w.r.t Specification limits (LSL, USL)

 Process Capability Index

A measure of the process capability which indicates

how the process is behaving relative to specifications.

However, it is a a valid measure only for in-control

processes.
 Process Capability

(a) (b)
specification specification

natural variation natural variation

(c) (d)
specification specification

natural variation natural variation

4
 Process Capability Index (PCI)

Allowable process variability

PCI =
Actual process variability

Many U.S. companies use PCI = 1.33 as a minimum acceptable

target and PCI = 1.66 as a minimum target for strength, safety, or
critical characteristics.

However, Some companies require that internal processes and

those at suppliers achieve a PCI = 2.0.

A process with PCI = 2.0 is referred to as a six-sigma process

because the distance from the process mean to the nearest
specification is six standard deviations.
 Specification Terminology

Allowable process variability

USL = the Upper Specification Limit,

and

LSL = the Lower Specification Limit.

Actual process variability

Process spread 6σ
where, σ is the process standard
deviation
 Process Capability Index, Cp

Cp= USL - LSL

6σ

– Suppose the process shown is

• in statistical control (i.e., stable and common cause
variations only)
• process mean is at 2000
• process standard deviation (std. dev.)= 500
• USL = 2750, LSL = 1250, Target = 2000

•Cp = (2750 - 1250)/(6x 500) = 0.5

– How about process std. dev. = 250, Cp =?
– How about process std. dev. = 125, Cp =?
7
Process Capability Index, Cp

Cp= USL - LSL

6σ

Cp < 1

8
Process Capability Index, Cp

Cp= USL - LSL

6σ

Cp = 1
Process Capability Index, Cp

Cp= USL - LSL

6σ

Cp > 2
Cp
 Limitations of the Cp Index
• Does not take into account the value (location) of the
process mean (where process is centered).

• If the process mean is not located at the mid-point

between USL and LSL, then Cp has little meaning.

Cp is same for both the cases

µ µ

LSL Target USL

Process Capability Index, Cpk

USL – µ
Cpu =
3σ

µ- LSL
Cpl =
3σ

Cpk = min{ C pl, C pu }

13
Comparison of Cp and Cpk In Off-Center Process
 Example

Target = 50, LSL = 30, USL = 70

Process 1 : Avg = 40, std dev = 1.66

Process 2 : Avg = 60, std dev = 1.66

Determine Cp and Cpk

 The Cpm Index
An improved capability ratio to measure process centering
is Cpm.

USL - LSL
Cpm =
6τ

where τ is the square root of expected squared deviation

from target:

τ 2 = E [ (x – T) 2 ] = σ 2 + ( µ - T) 2
T = ½ (USL+LSL)

USL − LSL
C pm =
6 σ 2 + (µ − T ) 2
 More About Process Centering

To simplify calculations, Cpm can be rewritten another way

USL − µ
C pmu =
3 σ 2 + (µ − T ) 2

µ − LSL
C pml =
3 σ 2 + (µ − T ) 2

Cpmk = min { C pml, C pmu }

 Estimating population standard deviation

Since population µ and σ are not known most of the time,

they must be estimated from data.

USL − LSL
cˆP =
6σˆ
Case 1 : Using sample standard deviation values
m
E(s) = s = ( ∑ s i ) / m σ = s / c4
i=1

Case 2 : Using sample range values

m
R=(∑
i=1
Ri)/m σ = R / d2
 Relation between PCI and DPMO

DPMO

500000

400000

300000
DPMO

200000

100000

0
0 0.5 1 1.5 2 2.5
PCI
 Relation between PCI and DPMO

DPMO DPMO

500000 3000

400000 2500

2000
300000
DPMO

DPMO
1500
200000
1000
100000 500
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
PCI PCI
 Relation between PCI and DPMO

DPMO DPMO

500000 3000

400000 2500

2000
300000
DPMO

DPMO
1500
200000
1000
100000 500
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
PCI PCI

DPMO

8
7
6
5
DPMO

4
3
2
1
0
0 0.5 1 1.5 2 2.5
PCI
 Relation between PCI and DPMO

DPMO DPMO

500000 3000

400000 2500

2000
300000
DPMO

DPMO
1500
200000
1000
100000 500
0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
PCI PCI

DPMO Chart Title

8 0.4
7 0.35
6 0.3
5 0.25
DPMO

DPMO

4 0.2
3 0.15
2 0.1
1 0.05
0 0
0 0.5 1 1.5 2 2.5 1.6 1.7 1.8 1.9 2 2.1
PCI PCI
1.5 sigma shift in the process average

2700 PPM 66800 PPM

1.5 σ shift

-3σ +3 σ

LSL USL
 A Six Sigma Process

Assuming a normally distributed process, the fallout of the

shifted process is 3.4 parts per million.

0.002 PPM 3.4 PPM

 Non-Normality and the PCI

• A capability Index based on the distribution of the

process is given by

USL - LSL
Cp (d)=
X 0.99865 – X 0.00135

m + 3s m - 3s

• In the case of the normal distribution Cp(d) reduces to

Cp