Module 1 – Statistical

Process Control

Prepared by:
Ezrha C. Godilano
BSIE, CIE
[email protected]
 INTRODUCTION TO

PROCESS CONTROL
CONCEPTS
 PROCESS CONTROL
refers to any sequence of means to guide or influence
activities that aims at producing the elements of the process
results in order to obtain the desired result

STATISTICS
collection, analysis, and interpretation
of data
PROCESS CONTROL

The action to be done on the system

to manage a certain process in order
to obtain the desired results

The goal of Process Control System

is to make decision about actions
affecting the PROCESS
CONVENTIONAL CONTROL STRATEGIES

DETECTION PREVENTION
To determine the fact To act in advance
To check and recheck Avoids waste from happening
Expensive and time consuming Constructive preparation
Tolerates waste
Firefighting
 PROCESS CONTROL SYSTEM

ACTION ACTION
ON THE ON THE
PROCESS OUTPUT
DATA HANDLING

PROCESS
MAN METHOD MACHINE

OUTPUT FIELD
MANAGEMENT
PERFORMANCE

MATERIALS MEASUREMENT
ENVIRONMENT

PROCESS DATA PRODUCT DATA FIELD RESULT

DATA BASE
PROCESS CONTROL ACTIONS

ACTION ON THE OUTPUT

- Detection of non conformance
- Action on the result of non conformance
- Damage : already done
- Past oriented

ACTION ON THE PROCESS

- Prevention non conformance
- Performance improvement
- Damage : prevented
- Future oriented
PROCESS CONTROL ERRORS

Type I Error OVERCONTROL

taking action when not necessary

Type II Error UNDERCONTROL

failing to take action when the action is
necessary
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 IN ORDER TO ACHIEVE AND MAINTAIN
A STATE OF STATISTICAL CONTROL AND TO
IMPROVE PROCESS CAPABILITY
 PRINCIPLES
AND
EFFECTS
OF VARIATION
PRINCIPLES OF VARIATION

1. Things are never exactly alike.

2. Variation can be measured.
3. Variation forms a definite pattern.
4. Distribution can differ in location, spread
and shape.
5. In the absence of special causes of variation,
most measurable characteristics follow a
distribution called normal.
6. The output of a process with variation due to
special causes forms distributions that are
unstable and unpredictable.
7. The output of a process with variation due to
common causes forms distributions that are
stable overtime, and with predictable trend.
THINGS ARE NEVER EXACTLY ALIKE
VARIATION CAN BE MEASURED
DATA
facts and figures collected on some
characteristics of the population or sample

TYPES OF DATA
VARIABLES
measured in continuous
measurable
ATTRIBUTES
discrete measurements
countable
WAYS OF SUMMARIZING DATA

Stem and Leaf

Frequency Distribution
Histogram
Polygon
Box Plot
example :

TEST DATA
6.5 4.1
5.0 3.0
4.1 5.8
6.1 5.6
4.9 4.5
4.5 5.0
7.0 5.3
6.4 6.1
5.3 5.0
5.2 6.5

No. of data
Max.
Min.
example :
TEST DATA

7 0
6 5 1 4 1 5

5 0 3 2 8 6 0 0

4 1 9 5 1 5

3 0

No. of data 20
Max. 7.0
Min. 3.0
 DISTRIBUTION

VARIATION FORMS A DEFINITE PATTERN

 CHARACTERISTICS OF A DISTRIBUTION

CENTRAL TENDENCY

SHAPE

SPREAD

Notes:
 CHARACTERISTICS OF A DISTRIBUTION
most commonly used measures are :
CENTRAL TENDENCY
measures the tendency of data to cluster towards the middle
of frequency distribution.
mean ( X ) or average
median ( X0.5 )
mode ( X )
SPREAD
measures the degree by which the data vary from each other.
most commonly used measures are :
range ( R )
standard deviation ( s )
variance ( s2 )
SHAPE
the shape of a distribution can be characterized in terms of
skewness - bias and kurtosis - flatness or peakedness.
most commonly used measures are :
coefficient of skewness ( SK )
coefficient of kurtosis ( K )
 MEAN ( X ) or AVERAGE
sum of individual measurements divided
by the number of observations

x
 x i

x1  x2  x3  .....  xi
n n

where :
X = mean or average
Xi = value of ith observation
n = number of observation
 RANGE

the difference between the largest and the

smallest observations in a group of data

R = Xmax - Xmin

where :
R = range
Xmax = highest individual value
Xmax = lowest individual value
 STANDARD DEVIATION

 ( x  x)
i
2

s i 1
m 1
where :
X = overall mean or average
Xi = value of ith observation
m = total number of individual
measurements
= kxn
k = number of subgroups
n = subgroup sample size
 TARGET

LOCATION
TARGET

SPREAD
TARGET

SHAPE
DISTRIBUTIONS CAN DIFFER IN LOCATION,
SPREAD AND SHAPE
 CAUSES OF VARIATION

SPECIAL CAUSES
- assignable causes
- unusual to the process
- can be detected with simple statistical techniques
- unpredictable recurrence

COMMON CAUSES
- chance causes
- inherent to the process
- need detailed analysis
- stable distribution over time
 2.14% 34.13% 34.13% 2.14%

0.13% 13.6% 13.6% 0.13%

IN THE ABSENCE OF SPECIAL CAUSES

OF VARIATION, MOST MEASURABLE
CHARACTERISTICS FOLLOW A
DISTRIBUTION CALLED NORMAL
 PREDICTION

PRESENT

THE OUTPUT OF A PROCESS WITH

VARIATION DUE TO COMMON CAUSES
FORMS DISTRIBUTIONS THAT ARE
STABLE OVER TIME, AND WITH
PREDICTABLE TREND
 PREDICTION

PRESENT

THE OUTPUT OF A PROCESS WITH

VARIATION DUE TO SPECIAL CAUSES
FORMS DISTRIBUTIONS THAT ARE
UNSTABLE AND UNPREDICTABLE
MANAGEMENT
AND
CONTROL OF
VARIATION
 QUALITY

means

CONFORMANCE TO REQUIREMENTS

REJECT ACCEPT REJECT

LOWER TARGET UPPER

SPECIFICATION SPECIFICATION
LIMIT LIMIT
 BUT . . . . .
TO ACHIEVE
QUALITY IMPROVEMENT
WE HAVE TO REALIZE

MINIMUM VARIABILITY FROM THE

TARGET VALUE

REJECT ACCEPT REJECT

LOWER TARGET UPPER

SPECIFICATION SPECIFICATION
LIMIT LIMIT
 PRODUCT QUALITY COMPARISON

SONY COMPANY
UNITED STATES vs. JAPAN

JAPAN
US

Comparison of colour density between US and JAPAN SONY

 SONY COMPANY

UNITED STATES vs. JAPAN

JAPAN
US

Distribution of US SONY Company before reworking

 ACTIONS TO REDUCE VARIATION
CAUSES ACTIONS

SPECIAL CAUSE - LOCAL ACTIONS

- done by the people close to the process
- can correct about 15% of process
problems

COMMON CAUSES - ACTIONS ON THE SYSTEM

- usually require management action
- can correct about 85% of process
problems
CONTROL CHARTS
 CONTROL CHARTS

Allow us to distinguish between special

causes and common causes of variation
Indicate when to take action and when
not to, thus, preventing control errors.
Present trend which can be analyzed
to predict forthcoming results.
Powerful tool for prevention
TYPES OF CONTROL CHART

Variable Control Chart

Attribute Control Chart

Precontrol Chart