Process Capability
Process Capability
Process Capability
In 6-sigma
Prof. S. K. Neogy
Indian Statistical Institute
Delhi Centre
Process Capability
Design
Specifications
Process
Process Capability
Design
Specifications
Process Capability
• Terminology
– Control limits
• represent the natural variation in a quality
characteristic
– Tolerance limits
• represent variation acceptable to the customer
– Capable Process
• In a capable process, specification limits are well
outside control limits
• The goal is to choose a capable process such that
the natural control limits of the process are within
the specifications of the customer
Process Capability
Index
• There are several indices that may be
used to measure process capability. We
will examine one.
X − LSL USL - X
Cpk = min or
3σ 3σ
Process Capability
Index
X − LSL USL - X
Cpk = min or
3σ 3σ
As a production process
produces items small shifts Shifts in Process Mean
in equipment or systems
can cause differences in
production performance
from differing samples.
Interpreting Process
Capability
• Interpretation of index values
– If Cpk = 1, then the natural control limits
and the customer specification are
exactly equal. The process is just
capable
– If Cpk < 1, the process is not capable
– If Cpk > 1, the process is highly capable
of meeting customer specifications
• Note: If the process is not under
control, then Cpk has no meaning
Process Capability
Measures
Process Capability Ratio
tolerance range
Cp =
process range
9.5 - 8.5
= = 1.39
6(0.12)
Process Capability
Measures
Process Capability Index
=
x - lower specification limit
,
Cpk = minimum 3σ
=
upper specification limit - x
3σ
Computing Cpk
Net weight specification = 9.0 oz ± 0.5 oz
Process mean = 8.80 oz
Process standard deviation = 0.12 oz
=
x - lower specification limit
,
Cpk = minimum 3σ
=
upper specification limit - x
3σ
Sigma Cpk
1.5 0.50
3.00 1.00
3.50 1.17
4.00 1.33
4.50 1.50
5.00 1.67
6.00 2.00
Assumptions
In the above table, we have assumed that the standard sigma shift of 1.5 is appropriate (the
calculator allows you to specify another value), the data is normally distributed, and the
process is stable.