Chapter 10
Control Chart for Attributes
10.1 Introduction
Quality characteristics are monitored as a function of only two possible outputs.
• Yes-No
• Conforming-Non-conforming
• 1-0
• Pass-Fail
• Go-No Go
There are many types of attribute control charts. The most commonly used ones are-
1. P chart (proportion non-conforming)
2. np chart (number non-conforming)
3. c chart (count chart)
4. u chart
10.2 Basics of p chart(Fraction Non-conforming)
10.2 Basics of p chart(Fraction Non-conforming)
• Situaton 1: when p (population fraction non-conforming) is known
For a 3σ control limit:
• Situation 2: when p is unknown
UC
UC
Example 1: Tube Light Testing
San Marino Tube Lights Limited is a famous tube light manufacturing company in North Carolina, producing
around 5000 pieces of lights per day. The quality control expert planned to take samples of size 50 units each
every working day. The company worked 22 days in the month under consideration. To test for quality, the
expert also planned to use p-chart.
Table: Number and fraction defective (non-conforming)
Sample No. (i) No. of failures () Fraction non- Sample No. (i) No. of failures () Fraction non-
conforming conforming
1 3 0.06 12 1 0.02
2 2 0.04 13 3 0.06
3 3 0.06 14 2 0.04
4 2 0.04 15 5 0.10
5 3 0.06 16 4 0.08
6 2 0.04 17 3 0.06
7 5 0.10 18 9 0.18
8 3 0.06 19 4 0.08
9 7 0.14 20 2 0.04
10 2 0.04 21 1 0.02
11 1 0.02 22 0 0.0
UC 𝐿 𝑝 =0 .060909 +3
𝐶𝐿𝑝 =0 . 060909
√0 . 060909(1 −0 . 060909)
50
= 0 .1624
𝐿𝐶𝐿 𝑝 =0 .060909 − 3
√ 0 .060909 ( 1 −0 .060909 )
50
=−0 . 0405= 0
10.4 Control Charts for Non-conformities
A unit of product may have many nonconformities, still being
considered as acceptable, if the nonconformities are small and not
severe.
But presence of too many small nonconformities may definitely make it
unacceptable to a customer.
• Consider a case of a plastic door to be used in building.
• Similarly, a glass sheet to be used on the top of a table or mirror to be
used in a dressing table
10.4 Control Charts for Non-conformities
A glass sheet may have a number of nonconformities (say, scratches or broken side).
An inspector may record number of nonconformities, c, in one unit of glass sheet.
Probability of a nonconformity, p, at any particular location in the glass sheet is very small and constant.
Poisson
Whereas
Distribution
Number of opportunities for a nonconformity, n, in the full glass sheet is very large.
Mean, np=c
Control Charts under the condition may again be of two types-
1) c chart for number of nonconformities
2) u chart for number of nonconformities per unit
10.5 The c chart
As per standard Poisson distribution, probability of nonconformities x=i, is-
p(x=i)=……. Where
Mean=
Standard Deviation=
As known, the 3σcontrol limits can be:
UCL= c+3
CL=mean=c
LCL=c-3
Similar to p chart, if population c value is unknown, then it can be approximated by
UCL=c̅+3
CL=c̅
LCL=c̅-3
Example: c chart
A weight (to be used by grocery shops for measuring weights of materials to be sold to the customer)
manufacturing company produces iron weights, produced by melting and casting irons of different
grades. A weight may be considered nonconforming for many reasons- lower/higher weight than
designed, too irregular shape, visible fracture, bubbles, surface finish etc. may occur for improper
heating, air ventilation, mixing temperature, bad quality iron etc. The number of nonconformities on
22 samples, produced in January, 2007, yielded the following results. Thirty(30) pieces constitute
one inspection unit.
Table: Number of nonconformities on each inspection unit
Sample No. (i) No. of failures () Sample No. (i) No. of failures ()
1 12 12 10
2 9 13 12
3 25 14 9
4 16 15 8
5 10 16 11
6 13 17 24
7 11 18 13
8 18 19 10
9 12 20 11
10 8 21 11
Solve
UCL=c̅+3=12.36+3
CL=c̅=12.36
LCL=c̅-3=12.36-3=1.81
Solve
• For the revised 2nd trial, mean no. of nonconformities in each
inspection unit is-
c̅==11.15
The revised 2nd trial limits are given below, which are plotted in the
same figure.
Example: Type I and Type II Errors in c
chart (First Trial)
Mean=x̅=c̅=12.36; UCL=22.90; LCL=1.81
Standard Deviation, σ==3.52
It is to be noted that number of nonconformities is discrete in nature whereas the
following discussion is based on Normal (continuous distribution)
assumption/approximation.