1. Statistical Process Control
Prepared By:
Malay Pandya
M.Pharm (QA)
Graduate School of Pharmacy
Gujarat Technological University
Enrollment no :192880824007
Guided by :
Dr Kashyap N Thummar
M.Pharm , PGDIM, PhD
Assistant Professor
Graduate school of Pharmacy
Gujarat Technological University
1
2. Content
• Introduction
• Statistical process control
• Importance of SPC
• Quality Measurement and Manufacturing
• The Shewhart control chart
• Advantages of statistical control
• Control Chart
2
3. Introduction[2]
• Statistical Quality Control (SQC) is the term used to
describe the set of statistical tools used by quality
professionals.
SQC Categories
1. Descriptive Statistics
2. Statistical Process Control
3. Acceptance Sampling
3
4. Descriptive statistics[3]: are brief descriptive coefficients that
summarize a given data set, which can be either a
representation of the entire or a sample of a
population. Descriptive statistics are broken down into
measures of central tendency and measures of variability
(spread).
Statistical process control (SPC)[4]: is a method of
quality control which employs statistical methods to monitor
and control a process. This helps to ensure that
the process operates efficiently, producing more specification-
conforming products with less waste (rework or scrap)
Acceptance sampling[5]: uses statistical sampling to determine
whether to accept or reject a production lot of material. It has
been a common quality control technique used in industry. It is
usually done as products leaves the factory, or in some cases
even within the factory.
4
5. Statistical process control[]
• Statistical process control is the application of
statistical method of to the measurement and analysis
of variation in a process. A process is a collection of
activities that converts inputs into outputs or results.
• Though use of control charts, statistical process
control assists in detecting special (or assignable)
causes of variation in both in process parameters and
end of the process (product) parameters.
5
6. Statistic Process Control Definition[2]
• Statistics: It is branch of mathematics concerned with
collecting, organizing, and interpreting data.
• Population: A process produce thousands product in one
shift – 8 hours - 480 minutes.
• Sample Selected: We want data for outer diameter of that
product. We decided to collect data of 10 parts every hour.
• Data Collector: OD of product is measured for 10 parts
every hour of that shift.
• Data organized: Make an excel sheet of data collected with
different ranges.
• Data interpreted: Organize data can be further interpreted
and use for process control.
6
7. • Process: Combination of people, materials, methods,
machines, environment and measurement to produce a
goods or services.
• Control: System/Procedure policy to achieve results
that confirms to requirements.
7
8. Importance of SPC[2]
Reduces waste.
Reduction in the time which is required to produce the
product.
Detecting errors at inspection.
Reduce inspection cost
Save cost of material by reducing number of rejects.
More uniform quality of production
Customer satisfaction
It provides direction for long term reduction in process
variability.
8
9. Quality Measurement and Manufacturing[6]
Quality measurement is central to the process of quality
control: “what gets measured gets done.”
Measurement is basic for all three operational quality
processes and for strategic management
1. Quality control measurement – provides feedback and
early warnings of problems.
2. Operational quality planning measurement quantifies
customer needs and product and process capabilities.
3. Quality improvement measurements can motivate
people, prioritize improvement opportunities, and help
in diagnosing causes.
9
10. Sr.
No.
Random (common) causes
Description
Assignable (special) causes
1 Consists of many individual causes Consist of one or just a few individual causes
2 Any one random cause results in minute
amount of variation (but many random
causes act together to yield a substantial
total).
Any one assignable cause can results in large
amount of variation.
3 Examples are human variation in setting
control dials, slight vibration in machines,
and NS slight variation in raw material.
Examples are operator, a faulty setup, or a batch of
defective raw materials.
Interpretation
4 Random variation cannot be eliminated from
a process economically.
Assignable variation can be detected; action to
eliminate the cause is usually economically
justified.
5 An observation within the control limits of
random variation means that the process
should not be adjusted.
An observation beyond control limits means that
the process should be investigated and corrected.
6 With only random variation, the process is
sufficiently stable to use sampling procedures
to predict the quality of total production or
do process optimization studies.
With assignable variation present, the process is
not sufficiently stable to use sampling procedures
for prediction.
10
11. The Shewhart control chart[1]
• It is most desirable to provide umpires with tools that can help
to distinguished between special causes and common causes.
• An elegant tool for this purpose is the Shewhart control chart
(or just control chart) shown in figure
• In figure the horizontal scale is time and the vertical scale is
quality performance.
• The plotted points show quality performance as time
progresses.
• The chart also exhibits three horizontal lines.
• The middle line is the average of past performance and is;
therefore, the excepted level of performance.
• The other two lines are statistical “limit lines” they are
indented to separate special causes from common causes from
causes, based on some chosen level of probability, such as 1
chance in 100.
11
13. Point within control limit
• Point A on the chart differs from the historical average.
However, since point A is within the limit lines, this
difference could be due to common causes (at a
probability of, more than 1 in 100).
• Hence we assume that there is no special cause. In the
absence of special causes, the prevailing assumption
include
• Only common causes are present.
• The process is in a state of “statistical control.”
• The process is doing the best it can
• The variations must be endured.
• No action need be taking action may make matters worse
(a phenomenon known as “hunting” or “tampering”).
13
14. Points outside of the control limits:
• Point B differs from the historical averages, but is
outside of the limit lines.
• Now the probability is against this being the result of
common causes, less than 1 chance in 100. Hence, we
assume that point B is the results of special causes.
Traditionally such “out-of-control” point becomes
nominations for corrective action.
14
15. Advantages of statistical control:[1]
• Provides means of detecting error at inspection.
• Leads to more uniform quality of production.
• Improves the relationship with the customer.
• It reduces cost.
• It reduces the number of rejects and saves the cost of material.
• It determines the capability of the manufacturing process
• It provides direction for long term reduction in process variability.
• It is stable process and operates with less variability.
• For some types of quality problems, the statistical tool are more than useful-
the problems cannot be solved at all without using the appropriate statistical
tools.
• The SPC movement has succeeded in training a great many supervisors and
workers in basic statistical tools.
• The resulting increases in statistical literacy have made it possible for them
to improve their grasp of the behavior of process and products.
• In addition, many have learned that decision based on data collection and
analysis yield superior results.
15
16. Control Chart[1]
• Traditional Shewhart control charts are divided into two
categories:
1. Variable charts (those using continuous, measurement
data),
2. Attribute charts (those using count data).
• Selecting the proper type of control chart is the different
types are described further below.
16
18. X̄ and R chart
• Also called the "Average and range" chart. X̄ refers to the average of a
rational subgroup and measures the central tendency of the response variable
over time. R is the range (Difference between highest and lowest values in
each subgroup), and the R chart measure the gain or loss in uniformity within
a subgroup which represents the variability in the response variable over time.
Note that, because specification limits apply to individual values rather than
average (averages inherently vary less than the component individual values),
control limits cannot be compared to specification limits which should not be
placed on control chart for averages
18
20. X̄ and s chart
• The average and standard deviation chart is similar to the
X̄ and R chart, but the standard deviation (instead of the
range) is used in the s chart. Although an s chart is
statistically more efficient than the range for subgroup
sizes greater than 2, a range chart is easier to compute
and understand and is traditionally used for subgroup
sizes smaller than about 10.
20
22. X-mR chart
Also known as I-mR chart, this charts individual measures
and a moving range. It is used when the rational subgroup size
= 1(such that there are no multiple measures from which to
obtain an average).
22
23. Z-mR Chart
This is similar to the X-mR chart, except that the
individual values are standardize through a Z transformation.
This is useful for short runs in which there are fewer than the
recommended 20 to 30 needed to establish one of the
preceding charts.
23
24. Example of Control Chart Attribute Data
• Whereas control charts for variable data require numerical
measurement (e.g., line width from a photo resist process),
control charts for attribute data require only a count of
observations of a characteristics (e.g., the number of
nonconforming items in a sample). There also are called
categorical data because units are classified into group such as
pass or fail.
24
25. p chart
• Also called a proportional chart, this tracks the proportion or
percentage of nonconforming units (percentages defective) in each
sample over time.
25
26. np chart
• This chart is used to track the number of nonconforming (defective)
units in each sample over time. An np chart should only be used
when the number of units sampled is constant (or nearly so).
26
27. c chart
• Used to track the number of nonconforming (i.e., defects, rather
than defective units as in the p chart).
27
28. u chart
• A variation of c chart, and analogous to the np chart, this chart
tracks the number of nonconformities (defects) per unit in a
sample of n units. As with the np chart, the number of units should
be approximately constant.
28
29. Conclusion
• From the above we concluded that SPC is a effective
tool of quality and process control in all type of
industry not only manufacturing industry where
quality and customer satisfaction is the major
concern. The power of SPC lies in the ability to
examine a process and the sources of variation in that
process, using tools that give weightage to objective
analysis over subjective opinions and that allow the
strength of each source to be determined numerically
29