Zero Defect Consultants Statistical Process Control

WELCOME TO ALL

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Training Contents
Day 1: Introduction to SPC Types of Process Controls

Introduction to Statistics
Understanding Mean, Mode, Median, Range, Standard Deviation Concept of Variation Special cause & common causes Stable & Unstable Process Approach towards identification of Special Causes Histogram An illustration Normal Distribution Process Capability
Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Training Contents
Day 2: Introduction to Control Charts Types of Control Charts Understanding application, methodology, interpretations of various types of charts (Variable and Attribute type covered) Exercises on Control Charts SPC implementation methodology Role of Operator in implementing SPC. Common mistakes done in implementing SPC Conclusion
Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Introduction to SPC
Process:Convert input to output using Man, Machine, Material, Method, Measurement. Input Output

Process
(Man, M/c, Material, Method)

Measure & give Feedback

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Process Control:Variation in the output of the process is natural. Hence the process needs to be controlled in order to ensure that output is meeting the customer requirements.

Tools for Process Control:Detection :A strategy that attempts to identify unacceptable output after it has been produced and then separate it from the good output.

Prevention :- A future oriented strategy by analysis and action toward

correcting the process itself so that unacceptable parts will not be produced.

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Need for Process Control
Detection :- Tolerates Waste Prevention :- Avoids Waste

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Techniques for Process Control
Mistake Proofing :- In this technique 100% process control is achieved by preventing all types of failures by using modern techniques to get defect free product. Here causes are prevented from making the effect.

100% Inspection : In this technique 100% checking of all the parameters of all products has been done to get defect free product. Here only defects are detected. Statistical Process Control : In this Statistical technique such as Control Chart, Histogram etc. are used to analyses the process to achieve and maintain state of statistical control to get defect free product. Causes are detected and prompting CA before defect occurs

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Statistics
Collection of Data, Analysis and Conclusion It is a value calculated from or based upon sample data (e.g. a subgroup average or range) used to make inferences about the process that produces the output.

E.g. Analysis of rejection data and initiating actions to reduce the rejection level.

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

What is Statistical Process Control
The use of Statistical techniques such as control charts to analyze a process or its outputs so as to take appropriate actions, to achieve and maintain a state of statistical control and to improve the process capability.

SPC is A tool to detect variation It identifies problems, it does not solve problems

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Introduction to Statistics
Data:
Any facts or numbers or observations made. Set of observations forms the data.

Types of Data:
Variable data Attribute data

Rev No:04, Date: 02.01.2010

10

Zero Defect Consultants Statistical Process Control

Variable Data
Data generated by Physically measuring the characteristic using an instrument Assigning an unique value to each item

Examples:
Hardness, Strength, Weight, Diameter, etc.

Rev No:04, Date: 02.01.2010

11

Zero Defect Consultants Statistical Process Control

Attribute Data
Data generated by Classifying the items into different groups based on some criteria No physical measurement is involved All the items classified into a group will have same value I.e OK or Not Ok. Examples: Sex, Shade Variation, Surface Defects, Go-No GO, etc.

Rev No:04, Date: 02.01.2010

12

Zero Defect Consultants Statistical Process Control

Statistical Properties of Data
Observations collected needs to be analyzed using various properties. Statistical properties of data helps in arriving at one value representing all observations. Two types of properties Measure of location Measure of Dispersion

Rev No:04, Date: 02.01.2010

13

Zero Defect Consultants Statistical Process Control

Measure of Location
Mean (Average) Median Mode

Measure of Dispersion
Range

Standard Deviation

Rev No:04, Date: 02.01.2010

14

Zero Defect Consultants Statistical Process Control

Mean:
Numerical value indicating the central value of data

Sum of all data / Total number of data

Suppose x1, x2, - - - xn be the data, then Mean = (x1+ x2 + - - -+ xn ) / n = xi /n Example Hardness Data Mean: = (55 + 65 + 59 + 59 + 57 + 61 + 53 + 63 + 59 + 57 + 63 + 55 + 61 + 61 + 57 + 59 + 61 + 57 + 59 + 63) / 20 = 1184 / 20 = 59.2
Rev No:04, Date: 02.01.2010

15

Zero Defect Consultants Statistical Process Control

Median:
Middle Value Value which divides data arranged in ascending or descending order into two equal halves Case 1: Total number of data is odd Median: Middle Value Case 2: Total number of data is even Median: Average of two middle values

Rev No:04, Date: 02.01.2010

16

Zero Defect Consultants Statistical Process Control

Median: Example Productivity Data 0.97 1.00 0.98 1.00 0.98 1.01 0.99 1.01 0.99 1.01 0.99 1.01 0.99 1.02 1.00 1.02 1.00 1.02 1.00 1.03

Total Number of data: 20 (even)

The middle Values : 1.00 & 1.00 (10th value and 11th value) Median: Average of 2 middle value (1.00 + 1.00) / 2 = 1.00

Rev No:04, Date: 02.01.2010

17

Zero Defect Consultants Statistical Process Control

Mode:
Highest no. of times an observation has occurred (Highest frequency) Mode: Example Productivity Data 0.97 1.00 0.98 1.00 0.98 1.01 0.99 1.01 0.99 1.01 0.99 1.01 0.99 1.02 1.00 1.02 1.00 1.02 1.00 1.03

0.97 - 1 0.98 - 2 0.99 - 4

1.00 - 5
1.01 - 4 1.02 - 3 1.03 - 1

- 1.00 is Mode as this occurred more no. of times

Rev No:04, Date: 02.01.2010

18

Zero Defect Consultants Statistical Process Control

Range: Definition
Range: Maximum value Minimum Value Range: Issues It depends only on extreme values Hence affected by outliers Example: 5 15 4 9 7 8 3 5 2 2

Maximum Value = 15 Minimum Value = 2 Range = 15 2 = 13

Rev No:04, Date: 02.01.2010

19

Zero Defect Consultants Statistical Process Control

Range: Issues

16 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10

Range

Rev No:04, Date: 02.01.2010

20

Zero Defect Consultants Statistical Process Control

Standard Deviation: Definition Square root of the average squared deviation from mean

Indicates On an average how much each value is away from the Mean
16 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10

Rev No:04, Date: 02.01.2010

21

Zero Defect Consultants Statistical Process Control

Standard Deviation: Example: Below data indicates the Money held by ten students in a class room.

5 15
Step 1:

4 9

7 8

3 5

2 2

Calculate Mean Mean = 6

Rev No:04, Date: 02.01.2010

22

Zero Defect Consultants Statistical Process Control

Standard Deviation: Example: 5 15 Step 2: Take deviations from Mean -1 9 -2 3 1 2 -3 -1 -4 -4 4 9 7 8 3 5 2 2

Rev No:04, Date: 02.01.2010

23

Zero Defect Consultants Statistical Process Control

Standard Deviation: Example: Step 2: Take deviations from Mean
16 14 12 10 8 6 4 2 0 1 2 3 4 5 6 7 8 9 10

Rev No:04, Date: 02.01.2010

24

Zero Defect Consultants Statistical Process Control

Standard Deviation: Example: Step 3: Since some values are positive & rest are negative, while taking sum they will cancel out. So square the values & Sum 1 81 4 9 1 4 9 1 16 16

Sum = 142

Rev No:04, Date: 02.01.2010

25

Zero Defect Consultants Statistical Process Control

Standard Deviation: Example: Step 4:

Standard Deviation = (Sum of Squares / (n -1))

= (142 / (10 -1)) = 15.77 = 3.972

Rev No:04, Date: 02.01.2010

26

Zero Defect Consultants Statistical Process Control

Variation
No two things are exactly alike

It is impossible to produce or process two items exactly alike

Variation is natural.

Rev No:04, Date: 02.01.2010

27

Zero Defect Consultants Statistical Process Control

Example: Time to reach office : 9:30 am It is not possible to reach office exactly at 9:30 everyday Normally there will be a small variation around 9:30 as follows: 9:31 9:34 9:33 9:26 9:28 9:27 9:29 9:34 9:25 9:31

This small variation is difficult to explain.

Rev No:04, Date: 02.01.2010

28

Zero Defect Consultants Statistical Process Control

Generally Looking at the past values, it is possible to give some range around 9:30 am (e.g.: 9:30 5 Minutes) to reach office

Normally it is possible to reach office within this range

Suppose A particular day , there is vehicle break down that day it may not be possible to reach office at 9:30 5 Minutes Say, you reach office at 9:50 am In other words If you reach office too late (beyond normal range of 9:30 5 minutes), there will be some special reason for that or it is easy to find out the reason for such variation
Rev No:04, Date: 02.01.2010

29

Zero Defect Consultants Statistical Process Control

Assignable Cause of Variation (Special Causes)

Variations of large magnitude

Easy to identify the causes of variation Easy to eliminate the cause of variation

Common Cause of Variation

Variations of small magnitude Difficult to identify the causes of variation Difficult to eliminate the cause of variation

Rev No:04, Date: 02.01.2010

30

Zero Defect Consultants Statistical Process Control

Local Actions

Are usually required to eliminate special causes of Variation

Can usually taken by the people close to the process Can correct typically about 15% of process problem

Actions on the System

Are usually required to reduce the variation due to common causes Almost always require management actions for correction Are needed to correct typically about 85% of process problems

Rev No:04, Date: 02.01.2010

31

Zero Defect Consultants Statistical Process Control

Stable Process
No Assignable Causes are present

Process is operating under common causes only

Stable process will be predictable & Statistically under control. If a process is stable & the data follows normal distribution

Then the variation will be Mean 3 x Standard deviation

Unstable Process
Assignable Causes are present Process is operating under assignable & common causes.

Unstable process is unpredictable & not under Statistical control.

Rev No:04, Date: 02.01.2010

32

Zero Defect Consultants Statistical Process Control

Methodology to Identify Assignable Causes
The success of any SPC program is not in our ability to collect data, draw charts etc., but in effectively identifying and eliminating assignable causes. Assignable causes are those causes that do not allow one to predict the behavior of processes.

There is no meaning in calculating Process Capability without having a predictable process.

Many companies have initiated SPC charts. But the charts do not benefit them. One of the main reason for this is that they have not stopped the process when an assignable cause is indicated and eliminated the cause

Rev No:04, Date: 02.01.2010

33

Zero Defect Consultants Statistical Process Control

Methodology to Identify Assignable Causes
Before starting the SPC data collection, let us do the following steps: 1. Identify the characteristic for which SPC is to be done. 2. Have a brainstorming to list all the causes that may influence the variation in this characteristic 3. Prepare a Cause & Effect Diagram 4. Prepare a Master Cause Analysis Table (Annexure 1) 5. Prepare a Why-Why Analysis Table (Annexure 2) 6. Identify factors that may affect Average and those that may affect Range After completion of the above, plan for data collection & implementation of SPC.

Rev No:04, Date: 02.01.2010

34

Zero Defect Consultants Statistical Process Control

Master Cause Analysis (Annexure 1)
Sl. N o. Cause Is there a spec? If so, what is the spec? Basis for the spec. Is it checke d and how? What is the actual? Diff. in Spec. Vs Actual Action plan

Rev No:04, Date: 02.01.2010

35

Zero Defect Consultants Statistical Process Control

Why Why Analysis (Annexure 2)
Sl. No. Cause WHY WHY WHY WHY WHY

Rev No:04, Date: 02.01.2010

36

Zero Defect Consultants Statistical Process Control

HISTOGRAM
Histogram is a graphical representation of data and shows the frequency of data. Histogram provides the easiest way to understand the distribution of data. It gives the Birds eye view of the variation in Data set. Portrays the information on location, spread and shape that enables the user to interpret the Process behavior. It indicates whether the process is operating under Normal / stable conditions.

Rev No:04, Date: 02.01.2010

37

Zero Defect Consultants Statistical Process Control

DEFINITIONS Class:- Is a category with lower and upper boundary value. Class Width:- Width of the class. Frequency:- No. of observations falling in the class.

Rev No:04, Date: 02.01.2010

38

Zero Defect Consultants Statistical Process Control

STEPS FOR CONSTRUCTING HISTOGRAM Collect Min 50 no. of readings (N). 50 readings should be continuous data. Determine Max value and Min value & Calculate Range. Range = Max Min.

Record the measurement unit (MU) used. This is usually controlled by

the measuring instrument least count.

Rev No:04, Date: 02.01.2010

39

Zero Defect Consultants Statistical Process Control

STEPS FOR CONSTRUCTING HISTOGRAM Contd.. Determine No. of classes (k), as below. No. of Class (k) = N Determine Class Width (CW), as below Class Width (CW) = Range / k

Construct the Frequency Distribution Table, as shown in the next

slide.

Rev No:04, Date: 02.01.2010

40

Zero Defect Consultants Statistical Process Control

Frequency Distribution Table
Class L1 U1 L2 U2 Upto Max value Total Where L1 = Minimum value (1/2*Measurement Unit) N Tally Frequency

U1 = L1 + Class Width
L2 = U1, U2 = L2 + Class Width & so on.

Rev No:04, Date: 02.01.2010

41

Zero Defect Consultants Statistical Process Control

STEPS FOR CONSTRUCTING HISTOGRAM Contd.. Determine the axis for the graph. Place Class on X axis and Frequency on Y axis. Mark off the classes, and draw rectangles with heights corresponding to the measurement frequencies in that class. Title the histogram. Give an overall title and identify each axis.

Rev No:04, Date: 02.01.2010

42

Zero Defect Consultants Statistical Process Control

HISTOGRAM GRAPHICAL REPRESENTATION.

Rev No:04, Date: 02.01.2010

43

Zero Defect Consultants Statistical Process Control

HISTOGRAM INTERPRETATIONS: NORMAL

Depicted by a bell-shaped curve. Most frequent measurement appears as center of distribution & less frequent measurements taper gradually at both ends of distribution. Indicates that a process is running normally (only common causes are present).

Rev No:04, Date: 02.01.2010

44

Zero Defect Consultants Statistical Process Control

HISTOGRAM INTERPRETATIONS: BIMODAL

Distribution appears to have two peaks. May indicate that data from more than one process are mixed together Materials may come from 2 separate vendors Samples may have come from 2 separate machines.

Rev No:04, Date: 02.01.2010

45

Zero Defect Consultants Statistical Process Control

HISTOGRAM INTERPRETATIONS: SKEWED

Appears as an uneven curve; values seem to taper to one side. Can be skewed left side or right side.

Rev No:04, Date: 02.01.2010

46

Zero Defect Consultants Statistical Process Control

Normal Distribution
Consider the following data on the case depth(mm) of 9 jobs: 2.3 2.7 2.4 2.6 2.5 2.5 2.4 2.5 2.6 Plot of the Data:
4 3 2 1 0 2.2 2.3 2.4 2.5 2.6 2.7 2.8

Rev No:04, Date: 02.01.2010

47

Zero Defect Consultants Statistical Process Control

Plot of the Data:
3

0 2.3 2.4 2.5 2.6 2.7

Bell Shaped

Symmetric
Total Area under the curve is 1 Then : Normal Curve & Data follows Normal Distribution
Rev No:04, Date: 02.01.2010

48

Zero Defect Consultants Statistical Process Control

Standard Normal Distribution: If

Data follows Normal Distribution then

(Data - Mean) / SD will follow Standard Normal Distribution

For Standard Normal Distribution:

Mean = 0 SD = 1

Rev No:04, Date: 02.01.2010

49

Zero Defect Consultants Statistical Process Control

Standard Normal Distribution: Properties
4 3 2 1 0 -3 -2 -1 0
68.26% 95.46% 99.73%
Rev No:04, Date: 02.01.2010

50

Zero Defect Consultants Statistical Process Control

Normal Distribution: Properties Between

Mean 1 SD : 68.26 % of Values will lie

Mean 2 SD : 95.46 % of Values will lie Mean 3 SD : 99.73 % of Values will lie

Rev No:04, Date: 02.01.2010

51

Zero Defect Consultants Statistical Process Control

Process Capability
A methodology to check whether a process is capable of meeting customer requirements

Determined by the variation that comes from common causes

Expressed as Process Capability Indices This needs to be demonstrated when process is being under statistical control Process Capability Indices 1. 2. 1. 2. Process Performance Index ( Pp & Ppk ) Process Capability Index ( Cp & Cpk ) Process performance Study (Pp & Ppk) Process Capability Study (Cp & Cpk)
Rev No:04, Date: 02.01.2010

Process Capability Study

52

Zero Defect Consultants Statistical Process Control

Process Capability Index (Cp) A methodology to check whether the process have the potential to meet the customer requirements Generally Customer requirements are given as specification on product characteristics Example Specification on Heat Treatment Process:

Hardness should be within 55 5 HRC

Rev No:04, Date: 02.01.2010

53

Zero Defect Consultants Statistical Process Control

Customer Requirement: Variation allowed by the customer Or Variation acceptable to customer Example: Specification: 55 5 HRC Meaning: As long as Hardness of the heat treated jobs are between 50 HRC to 60 HRC, Customer is satisfied Customer requirements are also expressed as Lower Specification Limit (LSL) = 50 HRC Upper Specification Limit (USL) = 60 HRC

Rev No:04, Date: 02.01.2010

54

Zero Defect Consultants Statistical Process Control

Process Capability Index (Cp): A process have the potential to meet customer requirement, if Total variation in process < Allowed variation Example: Specification: 55 5 HRC Allowed variation = 50 HRC to 60 HRC Total Variation = 52 HRC to 58 HRC Total Variation < Allowed variation Hence

Process have the potential to satisfy customer

Rev No:04, Date: 02.01.2010

55

Zero Defect Consultants Statistical Process Control

Example: Specification: 55 5 HRC Allowed variation = 50 HRC to 60 HRC Total Variation = 48 HRC to 62 HRC Total Variation > Allowed Variation

Then
Process doesnt have the potential to satisfy customer

Rev No:04, Date: 02.01.2010

56

Zero Defect Consultants Statistical Process Control

Process capability Index Cp: If the data is normally distributed, then

Total variation : Mean 3 SD

Example: Mean = 55 HRC & SD = 1HRC = 52 HRC to 58 HRC

Total Variation = 55 3 x 1 to 55 + 3 x 1

Rev No:04, Date: 02.01.2010

57

Zero Defect Consultants Statistical Process Control

Process Capability Index Cp: Definition Ratio of allowed variation to Total variation

Cp = Allowed variation / total variation

= (USL LSL) / ((Mean + 3 SD) (Mean - 3 SD)) = (USL LSL) / 6 SD

A Process has the potential to meet customer requirements if total variation < allowed variation 6 SD < (USL LSL) Cp > 1

Rev No:04, Date: 02.01.2010

58

Zero Defect Consultants Statistical Process Control

Process Capability Index Cp: Example 20 data on acid content (mm) is given in the table below. If the specification on acid content is 0.70 0.2 mm. Check whether the process has the potential to meet the customer requirement ? 0.85 0.80 0.75 0.75 0.80 0.70 0.65 0.70 0.75 0.75 0.60 0.75 0.80 0.85 0.70 0.60 0.75 0.50 0.60 0.65

Specification = 0.70 0.2 mm USL = 0.90 mm LSL = 0.50 mm Mean = 0.715 SD = 0.092

Rev No:04, Date: 02.01.2010

59

Zero Defect Consultants Statistical Process Control

Process Capability Index Cp: Example

Cp = (USL LSL) / 6 SD = (0.90 0.50) / (6 x 0.092) = 0.4 / 0.552 = 0.72

Rev No:04, Date: 02.01.2010

60

Zero Defect Consultants Statistical Process Control

Process Capability Index (Cp): Issues Cp checks only whether the process has the potential to meet the requirements Cp never checks whether the Process is actually meeting requirements

Rev No:04, Date: 02.01.2010

61

Zero Defect Consultants Statistical Process Control

Process Capability Index (Cpk): Definition Cpk = Min [Ppl, Ppu] Cpl = (Mean LSL) / 3 SD Cpu = (USL - Mean) / 3 SD Cpk checks whether the process is centered at the middle of the specification

Rev No:04, Date: 02.01.2010

62

Zero Defect Consultants Statistical Process Control

Process Capability Index Cpk: Graphical Representation
LSL USL

Cpl = a / c = (Mean LSL ) / 3 SD Cpu = b / d = (USL - Mean ) / 3 SD

a
1.2 1 0.8 0.6 0.4 0.2 0 1 2 3 4 5 6 7

3 SD
c

Mean
d

+ 3 SD

Rev No:04, Date: 02.01.2010

63

Zero Defect Consultants Statistical Process Control

6 12

Example: USL : 12
2
1.2 1 0.8 0.6 0.4 0.2 0 5 6 7 8 9 10 11

LSL: 6 SD : 1

Mean : 8 Cpu = 4 / 3 = 1.33 Cpl = 2 / 3 = 0.66

Cpk = Min [1.33,0.66] = 0.66

Cpk < 1, performance is not OK

- 3 SD 3

Mean 3

+ 3 SD
64

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

6 12

Example:

USL : 12
3
1.2 1 0.8 0.6 0.4 0.2 0 6 7 8 9 10 11 12

LSL: 6
SD : 1

Mean : 9 Cpu = 3 / 3 = 1 Cpl = 3 / 3 = 1

Cpk = Min [1 , 1] = 1 Cpk = 1

- 3 SD 3

Mean 3

+ 3 SD

Rev No:04, Date: 02.01.2010

65

Zero Defect Consultants Statistical Process Control

6 12

Conclusion: Cpu = 3 / 3 = 1 Cpl = 3 / 3 = 1

3
1.2 1 0.8 0.6 0.4 0.2 0 6 7 8 9 10 11 12

Cpk = Min [1 , 1] = 1 Cp = (USL LSL) / 6 SD = 6 /6 = 1

When Mean is at middle of Specification [(USL + LSL) / 2] then

Cpu = Cpl = Cpk = Cp Otherwise

Cpk < Cp
When Cpk < Cp Performance is not optimum
66

- 3 SD 3

Mean 3

+ 3 SD

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Process Performance Study Is a Short term Study performed during New Product Development. Indices are represented as Pp & Ppk. Standard deviation calculated using first principle formulae - Sigma(n-1). Acceptance value for Pp / Ppk is Min 1.66 Process Capability Study Is a Long term Study performed to monitor the ongoing Production. Indices are represented as Cp & Cpk. Standard deviation calculated using sigma = Rbar/d2 Acceptance value for Cp / Cpk is Min 1.33

Rev No:04, Date: 02.01.2010

67

Zero Defect Consultants Statistical Process Control

CONTROL CHARTS A tool for process control and improvement A statistical tool to know that process is stable or in control

A statistical tool to detect the presence of Assignable Causes in the process.

PROPERLY USED CONTROL CHARTS CAN : Be used by operators for ongoing control of a process Help the process perform consistently and predictably Allow the process to achieve Higher Quality, Lower unit cost, Higher effective capability Provide common language for discussing the performance of the process

Distinguish special from common causes of variation, as a guide to local action or action on the system

Rev No:04, Date: 02.01.2010

68

Zero Defect Consultants Statistical Process Control

Theory of Control Charts For Normally distributed Data If the process is stable then Variation will be between Mean 3 x SD

Rev No:04, Date: 02.01.2010

69

Zero Defect Consultants Statistical Process Control

Control Charts A Graphical Tool with three horizontal lines 1. Lower Control Limit (LCL) 2. Center Line (CL) 3. Upper Control Limit (UCL) Control Chart UCL CL LCL

Rev No:04, Date: 02.01.2010

70

Zero Defect Consultants Statistical Process Control

Control Charts

UCL & LCL are such that if the value lies between UCL & LCL then the process is stable or in control

Control Chart UCL CL LCL

Rev No:04, Date: 02.01.2010

71

Zero Defect Consultants Statistical Process Control

Control Charts

For Normal Data UCL =Mean + 3 x SD CL = Mean LCL = Mean 3 x SD Control Chart UCL CL LCL

Rev No:04, Date: 02.01.2010

72

Zero Defect Consultants Statistical Process Control

Control Charts: Working 1. Calculate the Control Limits from past data 2. Plot the values in the chart 3. If the values are within the limits, the process is stable. Otherwise not.

Control Chart

6 4 2 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Rev No:04, Date: 02.01.2010

73

Zero Defect Consultants Statistical Process Control

Types of Control Charts Continuous Data Xbar & R Chart ( or Median R Chart) Individual X & Moving range Chart Xbar & S Chart ( or Median S Chart) Discrete Data

Control Chart for Defectives

p chart np chart Control Chart for Defects c chart u chart

Rev No:04, Date: 02.01.2010

74

Zero Defect Consultants Statistical Process Control

Selection Procedure for Control Charts
Determine Characteristics to be charted Is interest in monitoring % bad parts N Y Is the sample size constant Use np Chart Y N Y Is the sample size constant N Is it homogeneou s in nature Use p Chart Y Use C Chart N Can sub group avg. easily calculated Y Is s, calculated easily N Use Xbar-S Chart Use Xbar-R Chart Y Is the subgroup size 9 or more N Use Xbar-R Chart N Use Median Chart Use U Chart N Is interest in monitoring nonconformiti es

Are the data variable Y

Y Use X-MR Chart

Rev No:04, Date: 02.01.2010

75

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Methodology Conduct initial study : Decide on the Total Number of Samples N . (N >19) Decide on Sub Group Size n. (n > 3) Decide on Frequency of Sampling. ( eg: Once in a hour, Once in 2 hours, Once in every 50 items, etc.)

Collect Data & Calculate Control Limits

Plot Control Chart Calculate Process Capability Indices (Pp/Ppk). If Capable, Set Control Limits for Ongoing study. Monitor Process through plotting control chart.

Rev No:04, Date: 02.01.2010

76

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Process: Turning Sample Size N: 9 Step 1: Collect Data
Sample No. 1 2 3 4 5 6 7 8 9 Hour 8:00 9:00 10:00 11:00 12:00 13:00 14:00 15:00 16:00 x1 5.00 5.01 5.02 5.00 4.98 5.02 4.99 5.00 4.98 x2 5.01 4.98 5.01 5.00 4.98 4.99 4.99 5.01 5.00 x3 4.98 5.00 5.00 5.00 5.01 5.00 4.98 5.02 5.01 x4 5.00 5.00 5.00 5.00 4.99 4.98 4.98 5.00 4.98
77

Characteristic: Inner diameter (4.90 5.10) Sub Group Size n: 4

Frequency of Sampling: Once in a Hour

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 2: Calculate Sub Group Mean & Range

Sample No.
1 2 3

Hour
8:00 9:00 10:00

x1
5.00 5.01 5.02

x2
5.01 4.98 5.01

x3
4.98 5.00 5.00

x4
5.00 5.00 5.00

Mean
4.998 4.998 5.008

Range
0.03 0.03 0.02

4
5 6 7

11:00
12:00 13:00 14:00

5.00
4.98 5.02 4.99

5.00
4.98 4.99 4.99

5.00
5.01 5.00 4.98

5.00
4.99 4.98 4.98

5.00
4.990 4.998 4.985

0.00
0.03 0.04 0.01

8
9

15:00
16:00

5.00
4.98

5.01
5.00

5.02
5.01

5.00
4.98

5.008
4.993

0.02
0.03

Rev No:04, Date: 02.01.2010

78

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 3: Calculate Control Limits for R Chart

Center Line
CL = Mean = Rbar = Sum of all Range Values / Total Number of Values = 0.21 / 9 = 0.0233 Upper Control Limit UCL = Mean + 3 SD = D4 Rbar, For n = 4, D4 = 2.282 = 2.282 x 0.0233 = 0.053 Lower Control Limit LCL = Mean - 3 SD = D3 Rbar, For n = 4, D3 = 0 = 0 x 0.0233 = 0.0
Rev No:04, Date: 02.01.2010

79

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 4: Plot R values in R chart as shown below:

R Chart

0.06 0.04 0.02 0 1 2 3 4 5 6 7 8 9

Step 5: If any value is beyond Control Limits, Do Homogenization Homogenization:

Remove the out of control value Recalculate the limits
Rev No:04, Date: 02.01.2010

80

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 6: Calculate Control Limits for Xbar Chart

Center Line
CL = Mean = Xdoublebar = Sum of all Means / Total Number of Values = 44.975 / 9 = 4.997 Upper Control Limit UCL = Mean + 3 SD = xdoublebar + A2Rbar, For n = 4, A2 = 0.729 = 4.997 + 0.729 x 0.0233 = 5.014

Lower Control Limit

LCL = Mean - 3 SD = xdoublebar - A2Rbar, For n = 4, A2 = 0.729

= 4.997 - 0.729 x 0.0233 = 4.98

Rev No:04, Date: 02.01.2010

81

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 7: Plot mean values in xbar chart as shown below:
xbar Chart

5.02 5 4.98 4.96 1 2 3 4 5 6 7 8 9

Step 8: If any value is beyond Control Limits, Do Homogenization Step 9: If all values are within limit, Calculate Standard deviation . = Rbar/d2 , Where d2 is constant = 0.023/2.059 = 0.0111
Rev No:04, Date: 02.01.2010

82

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Example Step 10: Calculate Process Capability Indices Cp = Tol / 6 = 0.2 / 6 * 0.0111 = 3.03 Cpk = Min { (Xbar LSL)/3 , (USL-Xbar)/3 } Cpk = Min { 2.91 , 3.09 } = 2.91. Process is Capable. Step 11: If Capable, Set the Control limits for ongoing monitoring.

Rev No:04, Date: 02.01.2010

83

Zero Defect Consultants Statistical Process Control

Control Chart Constants

n
2 3 4

d2
1.128 1.693 2.059

A2
1.880 1.023 0.729

D3
0 0 0

D4
3.268 2.574 2.282

E2
2.66 1.77 1.46

5
6

2.326
2.534

0.577
0.483

0
0

2.114
2.004

1.29
1.18

Rev No:04, Date: 02.01.2010

84

Zero Defect Consultants Statistical Process Control

Xbar R Chart: Exercise The following are the data on Time (in Minutes) to Process Transactions in a BPO company. Construct an Xbar R chart to monitor the process
Sample Number 1 2 3 4 5 6 7 8 9 10 x1 6.0 5.2 5.5 5.0 6.7 5.8 5.6 6.0 5.5 4.3 x2 5.8 6.4 5.8 5.7 6.5 5.2 5.1 5.8 4.9 6.4 x3 6.1 6.9 5.2 6.5 5.5 5.0 5.2 6.0 5.7 6.3 Sample Number 11 12 13 14 15 16 17 18 19 20 x1 6.2 6.7 6.1 6.2 4.9 7.0 5.4 6.6 4.7 6.7 x2 6.9 7.1 6.9 5.2 6.6 6.4 6.5 7.0 6.2 5.4 x3 5.0 6.2 7.4 6.8 6.6 6.1 6.7 6.8 7.1 6.7

Rev No:04, Date: 02.01.2010

85

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example For Short Run Process / Bulk Material Processing. Can be used when the testing method is a destructive type. Not possible to collect data in Sub Groups Process: Heat Treatment
Sample Number 1 2 3 4

Characteristic: Case Depth

Data 2.5 2.3 2.8 2.6

5
6 7 8
Rev No:04, Date: 02.01.2010

2.4
2.9 2.1 2.5
86

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example Step 2: Calculate Moving Range Sample Number Date MR 0.2 0.5 0.2 0.2 0.5 0.8 0.4

1
2 3 4 5 6 7 8

2.5
2.3 2.8 2.6 2.4 2.9 2.1 2.5

Rev No:04, Date: 02.01.2010

87

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example Step 3: Calculate Control Limits for M R Chart

Center Line
CL = Mean = MRbar = Sum of all MR Values / Total Number of Values = 2.8 / 7 = 0.4 Upper Control Limit UCL = Mean + 3 SD = D4 MRbar, For n = 2, D4 = 3.268 = 3.268 x 0.4 = 1.3072 Lower Control Limit LCL = Mean - 3 SD = D3 MRbar, For n = 2, D3 = 0 = 0 x 0.4 = 0.0
Rev No:04, Date: 02.01.2010

88

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example Step 4: Plot MR values in MR chart as shown below:

MR Chart

1.5 1 0.5 0 1 2 3 4 5 6 7

Step 5: If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

89

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example Step 6: Calculate Control Limits for Individual X Chart

Center Line
CL = Mean = Xbar = Sum of all Data / Total Number of Values = 20.1 / 8 = 2.512 Upper Control Limit UCL = Mean + 3 SD = xbar + E2MRbar, For n = 2, E2 = 2.66 = 2.512 + 2.66 x 0.4 = 3.58 Lower Control Limit LCL = Mean - 3 SD = xbar - E2MRbar, For n = 2, E2 = 2.66 = = 2.512 - 2.66 x 0.4 = 1.45
Rev No:04, Date: 02.01.2010

90

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Example Step 7: Plot individual values in Individual X chart as shown below:

Individual X Chart

4 3 2 1 0 1 2 3 4 5 6 7 8

Step 8: If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

91

Zero Defect Consultants Statistical Process Control

Individual X & Moving Range Chart: Exercise The data given below are surface Finish values of 30 jobs after chromium plating. Construct an Individual X & Moving Range chart to monitor the process?
0.078 0.075 0.072 0.079 0.078 0.075 0.077 0.08 0.068 0.076 0.081 0.073 0.074 0.08 0.074 0.072 0.079 0.081 0.069 0.082 0.076 0.075 0.073 0.08 0.078 0.078 0.074 0.077 0.074 0.07

Rev No:04, Date: 02.01.2010

92

Zero Defect Consultants Statistical Process Control

Control Charts for Defectives: np Chart
Used when sample size is constant Used to measure no of nonconforming items in an inspection. Example: Inspection results of video of the month shipment to customers for 10 consecutive days are given in table. The number of inspection each day is constant and is equal to 1000. Construct np chart to control the defectives?
Sample Number 1 2 3 4 5 6 7 8 9 10 Number of Defectives 47 42 48 58 32 38 53 68 45 37
Rev No:04, Date: 02.01.2010

93

Zero Defect Consultants Statistical Process Control

np Chart : Calculation of Control Limits CL = Mean UCL = Mean + 3 SD LCL = Mean 3 SD pbar = Sum of Defectives / total Number Inspected = 468 /(1000*10) = 0.0468

Mean = npbar = 1000 x 0.0468 = 46.8

SD = npbar(1-pbar) = (1000 x (0.0468 x (1-0.0468))) = 6.67 CL = npbar = 1000 x 0.0468 = 46.8 UCL = Mean + 3 SD = 46.8 + 3 x 6.68 = 66.84 LCL = Mean - 3 SD = 46.8 - 3 x 6.68 = 26.76
Rev No:04, Date: 02.01.2010

94

Zero Defect Consultants Statistical Process Control

np chart: Example Plot the number of defectives in np chart as shown below:

np Chart

80 60 40 20 0 1 2 3 4 5 6 7 8 9 10

If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

95

Zero Defect Consultants Statistical Process Control

np Chart: Exercise The following are the data on defectives in payment of dental insurance claims. Control the dental insurance payment process with np chart. Sample Size is 300
Sample Number 1 2 3 Number of Defectives 3 6 4 Sample Number 11 12 13 Number of Defectives

6 9 5

4
5 6 7 8 9

6
20 2 6 7 3

14
15 16 17 18 19

6
7 4 5 7 5

10

20
Rev No:04, Date: 02.01.2010

0
96

Zero Defect Consultants Statistical Process Control

Control Charts for Defectives: p Chart Used when sample size is not constant Example: The daily inspection results for electric carving knives are given below. Construct a control chart to monitor the process ?
Sample Number 1 2 3 4 Number Inspected 500 550 700 625 Number of Defectives 5 6 8 9

5
6 7 8 9 10

700
550 450 600 475 650
Rev No:04, Date: 02.01.2010

7
8 10 6 9 6
97

Zero Defect Consultants Statistical Process Control

p Chart : Calculation of Control Limits CL = Mean UCL = Mean + 3 SD LCL = Mean 3 SD pbar = Sum of Defectives / Total Number Inspected = 74 / 5800 = 0.0128 Mean = pbar = 0.0128 SD = pbar(1-pbar) / ni

Rev No:04, Date: 02.01.2010

98

Zero Defect Consultants Statistical Process Control

p Chart : Calculation of Control Limits
Sample Number 1 2 3 4 5 6 7 8 9 10 Number Inspected 500 550 700 625 700 550 450 600 475 650 Number of Defectives 5 6 8 9 7 8 10 6 9 6 p 0.010 0.011 0.011 0.014 0.010 0.015 0.022 0.010 0.019 0.009 LCL 0 0 0 0 0 0 0 0 0 0 UCL 0.028 0.027 0.026 0.026 0.026 0.027 0.029 0.027 0.028 0.026

Rev No:04, Date: 02.01.2010

99

Zero Defect Consultants Statistical Process Control

p chart: Example Plot the proportion of defectives in p chart as shown below:

p Chart

0.04 0.03 0.02 0.01 0 1 2 3 4 5 6 7 8 9 10

If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

100

Zero Defect Consultants Statistical Process Control

p Chart: Exercise Daily inspection results for the model 305 electric range assembly line are given in the table. Construct a control chart to monitor the process?
Sample Number Number Inspected Number of Defectives Sample Number Number Inspected Number of Defectives

1
2 3 4 5 6 7 8 9 10

171
167 170 135 137 170 45 155 195 180

31
6 8 13 26 30 3 11 30 36

11
12 13 14 15 16 17 18 19 20

181
115 165 189 165 170 175 167 141 159

38
33 26 15 16 35 12 6 50 26
101

Rev No:04, Date: 02.01.2010

Zero Defect Consultants Statistical Process Control

Control Charts for Defects: c Chart Used when sample size is constant Example: A leading bank has compiled the data in the table showing the count of nonconformities for 100 accounting transactions per day. Construct a control chart to monitor the process?
Day 1 2 3 4 5 6 7 8 9 10
Rev No:04, Date: 02.01.2010

Number of nonconformities 8 19 14 18 11 16 8 15 21 8
102

Zero Defect Consultants Statistical Process Control

c Chart : Calculation of Control Limits CL = Mean UCL = Mean + 3 SD LCL = Mean 3 SD cbar = Sum of nonconformities / Total Number Inspected = 138 /100 = 13.8 Mean = cbar = 13.8 SD = cbar = 13.8 = 3.71 CL = cbar = 13.8 UCL = Mean + 3 SD = 13.8 + 3 x 3.71 = 24.93 LCL = Mean - 3 SD = 13.8 - 3 x 3.71 = 2.67
Rev No:04, Date: 02.01.2010

103

Zero Defect Consultants Statistical Process Control

c chart: Example Plot the number of nonconformities in c chart as shown below:

c Chart

30 20 10 0 1 2 3 4 5 6 7 8 9 10

If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

104

Zero Defect Consultants Statistical Process Control

c Chart: Exercise 100 product labels are inspected every day for surface nonconformities. The data for the past 20 days is given below. Construct a suitable control chart to monitor the nonconformities
Day 1 2 3 4 5 6 Number of Nonconformities 22 29 25 17 20 16 Day 11 12 13 14 15 16 Number of Nonconformities 15 10 33 23 27 15

7
8 9 10

34
11 31 29

17
18 19 20

17
17 19 22

Rev No:04, Date: 02.01.2010

105

Zero Defect Consultants Statistical Process Control

Control Charts for Defects: u Chart Used when sample size is not constant Example: The inspection results for the surface finish of rolls of white paper for 10 lots is given below. Construct a control chart to monitor the process ?
Lot Number 1 2 3 4 5 6 7 8 9 10 Number Inspected 10 10 10 9 10 10 10 8 8 8
Rev No:04, Date: 02.01.2010

Number of Defects 45 51 36 48 42 5 33 27 31 22 106

Zero Defect Consultants Statistical Process Control

u Chart : Calculation of Control Limits CL = Mean UCL = Mean + 3 SD LCL = Mean 3 SD ubar = Sum of Defects / Total Number Inspected = 340 / 93 = 3.66 Mean =ubar = 3.66 SD = (ubar / ni)

Rev No:04, Date: 02.01.2010

107

Zero Defect Consultants Statistical Process Control

u Chart : Calculation of Control Limits
Lot Number 1 2 3 4 5 6 7 8 9 10 Number Inspected 10 10 10 9 10 10 10 8 8 8 Number of Defects 45 51 36 48 42 5 33 27 31 22 u 4.5 5.1 3.6 5.3 4.2 0.5 3.3 3.4 3.9 2.8 LCL 2.80 2.86 2.70 2.83 2.77 1.09 2.66 2.56 2.63 2.44 UCL 5.47 5.47 5.47 5.57 5.47 5.47 5.47 5.69 5.69 5.69

Rev No:04, Date: 02.01.2010

108

Zero Defect Consultants Statistical Process Control

u chart: Example Plot the defects / unit (u) of in u chart as shown below:

u Chart

6 4 2 0 1 2 3 4 5 6 7 8 9 10

If any value is beyond Control Limits, Do Homogenization

Rev No:04, Date: 02.01.2010

109

Zero Defect Consultants Statistical Process Control

u Chart: Exercise Construct a suitable control chart for the data in the table for empty bottle inspections of a soft drink manufacturer?
Sample Number 1 Number of bottles inspected 40 Number of Defects 45 Sample Number 11 Number of bottles Inspected 52 Number of Defects

55

2
3 4 5 6 7 8 9

40
40 40 40 52 52 52 52

40
33 43 62 79 60 50 73

12
13 14 15 16 17 18 19

52
52 52 40 40 40 40 40

74
43 61 43 32 45 33 50

10

52

54

20

52

28

Rev No:04, Date: 02.01.2010

110

Zero Defect Consultants Statistical Process Control

SPC Implementation:Select the characteristics & Process for SPC

Perform MSA Study Plan for Data Collection Collect Data & Plot Control Chart

Find Assignable Cause & Fix it N

Is Process Stable Y Is Process Capable Y Establish Control Limits

Improve the Process

Prepare Reaction Plan

Ongoing Process Control

Rev No:04, Date: 02.01.2010

111

Zero Defect Consultants Statistical Process Control

Operator Role in SPC:Collect Data

Plot on the Control Chart

Is Process in Control (I.e no special cause) N Refer Reaction Plan

Take Corrective Action

Take Disposition action, if required

Rev No:04, Date: 02.01.2010

112

Zero Defect Consultants Statistical Process Control

REACTION PLAN
Process:Parameter:Chart Condition Possible Causes Doc No.:Rev. No./Date:Corrective Action Disposition Action

Rev No:04, Date: 02.01.2010

113

Zero Defect Consultants Statistical Process Control

Common Mistakes in SPC Implementation SPC Chart plotting is just a show put up to customers and outsiders. No real analysis is done. Companies do Fill SPC Charts, even though 100% inspection is being done. Hi we are implementing SPC charts Hence we are going to reduce Rejection. Just by plotting charts, rejection level doesnt reduce. Charts are plotted at the end of the shift & analyzed. Inspection frequency & the SPC chart plotting frequency is different.

Management involvement is very less. It is just to meet the customer / certification requirement.
Charts & Capability indices show positive sign. But rejections/reworks are increasing????
Rev No:04, Date: 02.01.2010

114

Zero Defect Consultants Statistical Process Control

Common Mistakes in SPC Implementation Reaction plans are not prepared. Even if available, not referred / used. Corrective actions are not initiated even if the process is not stable / not capable. Process Capability results are not considered as feedback for new product development. Poor awareness at Operator level on usage of SPC charts & Its interpretation So would you like to avoid above mistakes????

&
Get Maximum benefit of SPC???

Rev No:04, Date: 02.01.2010

115

Zero Defect Consultants Statistical Process Control

Any Questions

Rev No:04, Date: 02.01.2010

116

Zero Defect Consultants Statistical Process Control

Thank You

Rev No:04, Date: 02.01.2010

117