STATISTICAL PROCESS CONTROL

Purpose of SPC

The objective is to find out what kind of variation is present in the process The purpose of statistical process control (SPC) is to indicate:
o When a process is working at its intended best (only common cause variation is present)
No corrective action is necessary Unnecessary actions may actually increase process variability

If special is considered as common o1. When a processcause is disturbed and needs corrective action of cause no investigation done. some time (Special cause variation is is present). 2. If common cause is considered as special cause tampering of the process takes place. Hence both the problems should be avoided.

Control Charts

A statistical tool used to distinguish between process variation resulting from common causes and variation resulting from special causes. Used to monitor inputs to process, parameters of a process, or process outputs Are used to recognize when a process has gone out of control Are used for identifying the presence of special cause variation within a process Do not tell us if we meet specification limits Neither identify nor remove special causes

Control Charts: Monitoring Process Parameter

Objectives: o Monitor a process parameter (e.g., the process mean, process variance, process proportion of defects) o Detect changes in process that are not due to natural variation Method: o Draw random samples from the process every time interval o A statistic (e.g., the sample mean) is computed from each sample, and plotted on a graph o If the point falls within 3 sigma from the target value, the process is considered in-control

Example: Control Chart (out of control)

Outlier

3 2 1

68% 95% 99.7%

-1 -2 -3

Outlier

Control Limits and Specification Limits

Control Limits

Specificatio n Limits

Control Limits and Specification Limits

In Specification

OCIS ICIS
Out of Control

In Control

OCOS

ICOS

Out of Specification

Selecting Control Charts

TYPE OF DATA

Count or Classification (Attribute Data)

Measurement (Variable Data)

Count

Classification

Defects

Defectives

Fixed Sample Size

Variable Sample Size

Fixed Sample Size

Variable Sample Size

Subgroup Size of 1

Subgroup Size < 8-10

Subgroup Size > 8-10

C Chart

U Chart

NP Chart

P Chart

I-MR

X-bar & R

X-bar & S

Basics of Statistical Process Control

Statistical Process Control (SPC)
monitoring production process to detect and prevent poor quality
UCL

Sample
subset of items produced to use for inspection
LCL

Control Charts
process is within statistical control limits
9

Statistical Process Control (SPC) Objectives

Quantify and reduce variation Improve understanding of products and processes Improve product and process design Monitor processes in real time Make statistically valid decisions Center the process Determine when and when not to take action on the process Determine the type of action to take (i.e., actions to eliminate special causes, actions to improve the overall process)
10

Statistical Process Control (SPC)

The statistical process control (SPC) tools achieve these objectives by collecting and analyzing data. The below figure depicts SPC as a machine that uses data as input and has information as its output. The SPC tools squeeze the information out of the raw data.

11

Variability
Random
common causes inherent in a

Non-Random
special causes due to identifiable

process can be eliminated only through improvements in the system

Variation in raw materials Variation in ambient

factors can be modified through operator or management action

Tool wear Large changes in raw materials Broken equipment
12

temperature and humidity

variation in electrical or pneumatic sources Variation within equipment, such

Special Cause & Common Cause

Common (natural) cause variation
is the result of complex interactions of variations in materials, tools, machines, information, workers, and the environment. usually is small and insignificant. accounts for 80-95% of the observed variation in a process. only management has the power to change systems and infrastructure that cause common cause variation.

Special (assignable) cause variation

arises from external sources that are not inherent in the process, usually is large, and can be traced to the source and corrected. accounts for 15-20% of the observed variation in a process. front-line employees and supervisors have the power to identify and solve special causes of variation. Examples poorly trained workers, machines out of adjustment, etc.
13

Common Causes

Special Causes

14

Special Cause & Common Cause

When special causes are present, the system said to be statistically out of control. If the variations are due to common causes alone, the process is said to be in statistical control.

15

Quality Measures
Attribute
a product characteristic that can be evaluated with a discrete response good bad; yes - no

Variable
a product characteristic that is continuous and can be measured weight - length

16

HISTOGRAMS: VARIATION FOR A PERIOD OF TIME

DEFINITION

A Histogram shows the shape, or distribution, of the data by displaying how often different values occur. Number of Days for Approval
40 30 20 10 0 1 2 3 4 5 6 7 8 Number of Days for Approval
17

EXAMPLE

Number of Occurrences

Histograms do not take into account changes over time.

Control charts can tell us when a process changes

18

Control Chart Applications

Establish state of statistical control Monitor a process and signal when it goes out of control Determine process capability

19

Capability Versus Control

Control Capability
Capable
In Control
IDEAL

Out of Control

Not Capable

20

Commonly Used Control Charts

Variables data
x-bar and R-charts x-bar and s-charts Charts for individuals (x-charts)

Attribute data
For defectives (p-chart, np-chart) For defects (c-chart, u-chart)
21

Charts for Attributes

Fraction nonconforming (p-chart)
Fixed sample size Variable sample size

np-chart for number nonconforming Charts for defects

c-chart u-chart
22

Developing Control Charts

1. Prepare
Choose measurement Determine how to collect data, sample size, and frequency of sampling Set up an initial control chart

2. Collect Data
Record data Calculate appropriate statistics Plot statistics on chart
23

Next Steps
3. Determine trial control limits
Center line (process average) Compute UCL, LCL

4. Analyze and interpret results

Determine if in control Eliminate out-of-control points Recompute control limits as necessary
24

Typical Out-of-Control Patterns

Point outside control limits Sudden shift in process average Cycles Trends Hugging the center line Hugging the control limits Instability
25

Shift in Process Average

26

Identifying Potential Shifts

27

Cycles

28

Trend

29

Final Steps
5. Use as a problem-solving tool
Continue to collect and plot data Take corrective action when necessary

6. Compute process capability

Pre Control Charts

LTL Red Zone UTL Red Zone

Green Zone

nominal value

Yellow Zones

Control Chart Design Issues

Basis for sampling Sample size Frequency of sampling Location of control limits

32

SPC Implementation Requirements

Top management commitment Project champion Initial workable project Employee education and training Accurate measurement system

33

P Chart
When to Use: Only when variable data cannot be obtained. When charting fraction rejected as nonconforming, from a varying sample size. When screening multiple characteristics for potential monitoring on variable control charts. When tracking the quality level of a process before any rework is performed. How: By counting the number of defective items from a sample and then plotting the percent that are defective.

P Chart
Conditions:

In order to be of help, there should be some rejects in each observed sample. The higher the quality level, the larger the sample size must be to contain rejects. For example, if 20% of a product is rejectable, a sample size of 5 will be needed. However, a sample of 1,000 will give an average of only one reject per sample if 0.1 % of the product is rejectable.
Calculation of Plot Points for the p Chart The p plot point is the fraction defective in a sample. The centerline is the average fraction defective in a series of samples.

P Chart: Assumptions

It is based on binomial distribution. Two attributes only. The expected proportion of item with the attribute is constant (the same for each sample). Occurrence of attribute is independent from item to item.

p-Chart (defectives)
This is also called Fraction Defective Chart. It is the ratio of number of defective articles found to the total number of

articles actually inspected. It is represented by p

The p chart is used for processes that generate attribute data. The sample size for the p-Chart can vary but usually consists

of 100 or more.
Formula (Constant Sample Size) Centre line = p = Average fraction defective = np / n (Where n is the number of items inspected in the sub group) Lower Control Limit (LCL) = p - 3 p (1-p )/n Upper Control Limit (UCL) = p + 3 p (1-p )/n

P Chart: Example 1
The customer service department of a company is monitoring the percentage of complaints resolved within five days. The data have been collected for several weeks and given in the file resolve.mtw Can we conclude that customer complaint resolving process is in control?

What is the interpretation of the control chart?

How can we use this information for future control

P chart example 2 and 3

2. A sales organsiation has 13 marketing analysts that provide the amounts for bidding on jobs. Data for the last nine months has been summarised. Do the analysts differ significantly from each other in their win-rates? Describe the actions you will take.

NP Chart
When to Use: Only when variable data cannot be obtained. When charting the number rejected as nonconforming from a constant sample size. When screening multiple characteristics for potential monitoring on variable control charts. How: By plotting the number of defectives from a series of equal size samples.

NP Chart
Conditions:

In order to be of help, there should be some rejects in each observed sample. The higher the quality level, the larger the sample size must be to contain rejects. For example, if 20% of a product is rejectable, a sample size of 5 will be needed. However, a sample of 1,000 will give an average of only one reject per sample if 0.1 % of the product is rejectable.
Calculation of Plot Points for the NP Chart The NP plot point is the number of defectives found in each sample. The centerline is the average number of defectives in a series of samples.

np-Chart (Defectives)
np-Chart is used for processes that generate attribute data. The np chart is used to graph the actual number of defectives in a sample. The sample size for the np-Chart is constant, with between 5 and 10 defectives per sample on the average.

Center line = np Upper Control Limit = np + 3np (1-p) Lower Control Limit = np - 3np (1-p)

NP Chart:Example 1

A software company in Bangalore involved in developing and marketing an accounting package has been monitoring customer satisfaction by surveying the views of a sample of 50 of their customers (who use their product) each month and asking them to rate their product on a scale of 1 to 4. 1 = Poor 2 = Fair 3 = Satisfied

NP Chart:Example 1

Sample 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23

Month Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov

Sample Size 3or4 Scores % 3or4 Scores 50 39 0.78 50 35 0.7 50 38 0.76 50 36 0.72 50 32 0.64 50 37 0.74 50 36 0.72 50 40 0.8 50 42 0.84 50 42 0.84 50 35 0.7 50 41 0.82 50 36 0.72 50 33 0.66 50 42 0.84 50 36 0.72 50 33 0.66 50 35 0.7 50 34 0.68 50 32 0.64 50 29 0.58 50 31 0.62 50 29 0.58

NP Chart:Example 1
NP Chart of 3or4 Scores
45 UCL=45.09

40
Sample Count

35

__ NP=35.46

30

25 1 3 5 7 9 11 13 Sample 15 17 19 21 23

LCL=25.82

Caution!
Not all percentage should be plotted on a P-chart, e.g. 1. % of time computer is down --- use individual chart.

2. % of product that is scrapped (where both product and scrap are weighed or measured, e.g tonne of steel, feet of paper, ) --- use

C and U Chart: Assumptions

c or u charts are based on assumption of Poisson distribution Can count occurrences but not non occurrences. Probability of an occurrence is relatively rare ( less than 10% of the time). Occurrences are independent .

U Chart
When to Use: Only when variable data cannot be obtained. When plotting the average number of defects found per unit. When screening multiple characteristics for potential monitoring on variable control charts. How: Each unit is examined and the average number of defects found are plotted.

U Chart
Conditions: Constant unit size, but any convenient number of units per plot point. Unit size is different from sample size. There are potentially several different types of defects per unit, but none of which would necessarily render the part a defective. In order for this type of analysis to be of help, there should be some defects in each observed unit. Calculation of Plot Points for the u Chart The u plot point is the average number of defects per unit in a sample of n units. The centerline is the average of all the plot points on the chart.

U Chart: Formulae

Chart
u

Control Limit
UCL = u-bar+ 3[u-bar/ni] LCL = u-bar - 3[u-bar/ni]

Center line
u-bar= Summation of plot points / Number of plot points

Plot Point
u = ci /ni = (number of defects)/ (number of equal units) = i th count(ci) / Area of opportunity ni

Sample size
Varying

U chart Example 1 and 2

1.A manufacturing company employs 22 operators in packaging. The number of errors made last month is totaled for each operator in file operator.mtw a)What is the appropriate control chart for this data? b)Produce the control chart in minitab. c) Interpret the chart d)What are some appropriate actions to take?

2. A paper mill uses a control chart to monitor the imperfection in finished rolls of paper. Production output is inspected for 20 days and the resulting data are shown in paper.mtw. Use these data to setup a control chart for nonconformities per roll of paper. Does the process appear to be in statistical control? What central line and control limits would you recommend for controlling current production? Set up a control chart based on average sample size to control this process.

C Chart
When to Use:

Only when variable data cannot be obtained. When plotting the number of defects found per unit. When screening multiple characteristics for potential monitoring on variable control charts.
How: Each unit is examined and the number of defects found are plotted. As a rule, the unit size should be chosen so that at least two defects per unit are found.

C Chart
Conditions: Constant unit size and constant sample size. A unit is different from a sample. There are potentially several different types of defects per unit, but none of which would necessarily render the part a defective. In order for this type of analysis to be of help, there should be some defects in each observed unit. Calculation of Plot Points for the c Chart The c plot point is the number of defects found in/on a unit. The centerline is the average of all the plot points on the chart.

c-Chart (Defects)
c-Chart is used for processes that generate attribute data. The cChart monitors the number of defects per sample taken from a process. 5 and 10 readings are required to construct c-Chart, and the sample size must be constant.

The c-Chart can be used in both low- and high- volume

environments. It is based on Poission Distribution
Formula Centre line = c = Total number of defects / Number of Items Inspected Lower Control Limit (LCL) = c - 3 c Upper Control Limit (UCL) = c + 3 c

Variable Control Chart

Variable Control Chart

In most cases, you will be two charts for

each set of continuous data.

The first chart will show the actual data

points or averages, the second chart will

shows the ranges or standard deviations.

For analysis and monitoring, we use both

begin with range chart.

Variable Chart - Advantages

Allows to control particular quality

characteristic.
No needs for 100% inspection. Use

sampling inspection.
Provides a very good measures of the

nature of the underlying universe.

Formulas for Variable Charts

Chart Type Centerline Upper Control Limit Lower Control Limit Range Chart

D4 R

D3 R

X Chart
X Chart Moving Range Chart

X
X

X A2 R
X 2.66 MR

X A2 R
X 2.66 MR

MR

*D R 4

D3 R

D3, D4 and A2 are called statistical constants for a particular sample size, refer control chart table
R = R / Number of samples

X = X / Number of samples

Constant Table
Control Charts Constant Values
n 2 3 4 5 6 7 8 9 10 A2 1.13 1.69 2.06 2.33 2.53 2.70 2.85 2.97 3.08 D3 0 0 0 0 0 0.08 0.14 0.18 0.22 D4 3.27 2.58 2.28 2.11 2.00 1.92 1.86 1.82 1.78

Exercise X bar R
In a steel rod cutting industry, four samples were collected every month for each of the four vendors and measured for its length. Check if the data is in control.
Month 1 2 3 4 5 6 7 Sample_1 93.14 63.74 69.62 40.22 40.22 28.46 54.92 49.04 7.88 51.98 Sample_2 43.16 69.62 81.38 99.02 22.58 81.38 60.8 78.44 19.64 37.28 Sample_3 66.68 66.68 75.5 40.22 75.5 40.22 66.68 34.34 81.38 49.04 Sample_4 72.56 22.58 96.08 78.44 57.86 49.04 66.68 28.46 22.58 57.86

Enter the data in MINITAB worksheet in Column C1, C2, C3, C4 and C5

8 9 10

After entering the data Go To Stat > Control Chart > X Bar R Chart
Sample Month 1 2 3 4 5 6 7 8 9 10 1 93.14 63.74 69.62 40.22 40.22 28.46 54.92 49.04 7.88 51.98 2 43.16 69.62 81.38 99.02 22.58 81.38 60.8 78.44 19.64 37.28 3 66.68 66.68 75.5 40.22 75.5 40.22 66.68 34.34 81.38 49.04 4 72.56 22.58 96.08 78.44 57.86 49.04 66.68 28.46 22.58 57.86

X Bar Chart

Range Chart

In both the charts, all the values are well within the control limit