You are on page 1of 51

1

7QC Tools
2
Check Sheets
Pareto Diagram
Cause and Effect Diagram.
Stratification.
Scatter Diagram.
Histogram.
Graphs and Control Charts.
7QC Tools
3
Check Sheet
Purpose
How often are certain events happening
Translating Opinion into Fact
Aid to the limited memory
Making information collection easy from data
(Format)
Good point to start most problem solving tools
4
Mark defect by
SN Class Mean Value Tally Marks Total
1 950-955 952.5 IIII 4
2 955-960 957.5 IIII III 8
3 960-965 962.5 IIII IIII IIII 15
4 965-970 967.5 IIII IIII 10
5 970-975 972.5 IIII I 6
Diagram type check sheet
Frequency check sheet
Sn Parameter Spec Judgement Remark
1 2 3 4 5
Part No: Process: Machine:
Sample No
Inspection check sheet
Types of Check Sheet
5
Pareto Diagram
A bar chart that helps to prioritize actions by
arranging elements in descending order of
occurrence
Sorts out the vital few from the trivial many
Separates important from non-important causes
contributing to a problem
6
How to make Pareto Diagram?
Define a problem and collect data on the factors that
contribute to it.
Historical records generally provide sufficient
information.
Classify the data by type, cost , percent, number of
occurrences, or whatever is appropriate for the
situation.
7
How to make Pareto Diagram?
Arrange the data in descending order
Calculate % contribution of each item
Calculate cumulative % of each item
Defect Defect Qty % Cum
Code description Rej Contribution %age
A Drill Broken 70 46.60% 46.60%
B Serration Mismatch 45 30% 76.60%
C Runout more 25 16.60% 93.20%
D Dia U/S 6 4% 97.20%
E Dia O/S 4 2.80% 100%
Total 150
Rejection details of Pinion
8
How to make Pareto Diagram
Draw bar graph showing constituent ratio on vertical
axis.
Connect cumulative percentage of each bar graph to
obtain Pareto curve.
75%
50%
25%
100%
n=150
A B C D E
Nos.
70
25
45
6
4
46.6%
76.6%
93.2%
97.2%
75%
50%
25%
100%
n=150
A B C D E
Nos.
70
25
45
6
4
46.6%
76.6%
93.2%
97.2%
n=150
A B C D E
Nos.
70
25
45
6
4
46.6%
76.6%
93.2%
97.2%
9
Types of Pareto Diagrams
1. Pareto Diagrams by Phenomenon
Quality: defects, faults, failure, complaints, RW etc.
Cost: amount of loss, expenses
Delivery: delay in delivery, stock shortages etc
Safety: accidents, mistakes, breakdowns etc
2. Pareto Diagrams by Causes
Operator: shift, group, age, experience, skill etc
Machine: machines, equipments, tools, instruments etc.
Material: manufacturer, plant, lot, kind etc
Process: conditions, orders, arrangements,methods etc.
10
Hints on Making Pareto Diagrams
1. Check various classifications and construct many kinds
of Pareto diagrams
Essence of a problem can be grasped by observing it from
various angles
2. It is undesirable that others represents a higher
percentage
A different method of classification should be considered
3. It is best to draw Pareto diagrams by assigning
monetary value
Cost is an important scale of measurement in
management. If financial implications of a problem are not
properly appreciated, the research itself may end up as
ineffective.
11
Hints on Using Pareto Diagrams
1. If an item is expected to be amenable to a simple
solution, it should be tackled right away even if it is of
relatively small importance.
It will serve as an example of efficient problem solving, and
the experience, information and incentives to morale
obtained through this will be of great assets for future
problem solving.
2. Do not fail to make Pareto diagram by causes
After identifying the problem by making a Pareto diagram
by phenomenon, it is necessary to identify the causes in order
to solve the problem. It is therefore vital to make a Pareto
diagram by causes if any improvements are to be effected.
12
Cause and Effect Diagram
A graphic tool used to represent the relationship
between an effect and the cause that influence it
Identifies various causes affecting a process
Helps groups in reaching a common understanding of a
problem
Helps reduce incidence of subjective decision making
13
How to make Cause & Effect
Diagram?
Define the problem or effect clearly and explicitly.
Generate the potential causes of the problem
through brain storming.
Encourage wild ideas
Quantity rather than Quality of ideas
Suspend judgment on Good or Bad
Ride on anothers idea
Do not interrupt
Record all responses
Categorize all causes into major causes and give
suitable names
14
How to make Cause & Effect
Diagram?
Construct the cause and effect diagram by:
Place problem statement in a box on right hand side
Draw the major cause category boxes on the left
hand side. Commonly used categories are man,
machine, method, material and measurement
Put all brainstorming ideas at respective major cause
category in Cause and Effect Diagram
Identify likely root cause(s) and circle them
Collect data to verify the most likely root causes
Cause and Effect Diagram
Nurses
Many patients
Long waiting
time after
completion of
ultrasonic scan
Methods Environment
People
Reception
work
Medical charts
hard to tell apart
Many charts to input
Shortage of
numbers
Inexperience
Patients
Two or more booked
in at same time
Record
findings
Charts all
returned together
Handling
Data input
takes time
Calculation
of bill
Charts from all
departments
come together
Test
Rooms
Far away
Busy
Computer
terminal
slow
Reception
16
Hints on Making Cause and Effect
Diagrams
Identify all the relevant factors through
examination and discussion by many people
Express the characteristic as concretely as possible
Make the same number of cause and effect
diagrams as that of characteristics
Choose a measurable characteristic and factors
Discover factors amenable to action
17
Hints on Using Cause and Effect Diagrams
Assign an importance to each factor objectively on the
basis of data
Examination of factors on the basis of your own skill and
experience is important, but it is dangerous to give importance
to them through subjective perceptions or impressions alone
(would have been solved by now if so). Assigning importance to
factors objectively using data is both more scientific and more
logical.
Try to improve the cause and effect diagram
continuously while using it
Actually using a cause and effect diagram will help in
seeing those parts which need to be checked, deleted or
modified. This will be useful in solving problems, and at the
same time, will help improve your own skill and to increase your
technological knowledge
18
Stratification
Stratification is a statistical technique of breaking down
values and numbers into meaningful categories or
classification
To identify the cause of problem if they come from a
particular source
To analyze root cause in conjunction with other
techniques like Pareto diagram, histogram and graphs
19
How to use Stratification
Regroup original data as per the source of data
(e.g. Machine wise, shift-wise, model-wise, supplier-
wise)
If required collect data afresh after making the
source from which they come.
Recreate histogram, Pareto charts and graphs on
classified data

20
Month Model A Mode B
Apr 10 50
May 8 32
Jun 15 65
Jul 10 50
Aug 8 42
Sep 7 28
0
10
20
30
40
50
60
70
80
90
Apr May Jun Jul Aug Sep
Month
R
e
j

N
o
s
Mode B
Model A
Stratification
21
Impurity Amount
V
i
s
c
o
s
i
t
y
After Stratification
Impurity Amount
V
i
s
c
o
s
i
t
y
Before Stratification
Stratification
22
Scatter Diagram
A tool used to study the possible relationship between
two variables (Cause and Effect)
Diagram makes it clear whether a relationship exists
and shows the strength of that relationship
If data points are scattered very close to a trend, it
shows strong relationship otherwise it is weak
relationship
23
How to make Scatter Diagram
Collect 50 to 100 paired samples of data
believed to be related.
Construct a data sheet.
Draw the horizontal and vertical axis of the
diagram.
Label the axes.
Causeis usually plotted on the horizontal axis
and the effect variable on the vertical axis.
Plot the data on the diagram. If values repeat,
circle that point.
24
A. Randomly scattered points - No correlation
B. Possible Positive correlation
C. Strong positive correlation
Interpretation
n=15 r=0.06
n=18 r=0.54 n=14 r=0.96
A B
C
25
Interpretation
D. Possible negative correlation

E. Strong Negative correlation
n=22 r=- 0.5 n=18 r=-0.92
D
E
26
S(xy)
S(xx).S(yy)
r=
S(xx) = (x -x)
i
2
n
i=1
S(yy) = (y -y)
i
2
n
i=1
S(xy) = (x -x) (y y)
i
n
i=1
i
If r=0 No correlation
If r=1 Very strong correlation
If r=-1 Very strong negative
correlation
If 0<r>1 Possible correlation
Correlation Coefficient
27
Significance of r
If calculated value of r is more than the table
value of r at 1% or 5% significance level at (n-2)
degrees of freedom, then at 99% or 95%
confidence we can say that correlation exists
between two or vice-versa.
28
Notes on Correlation Analysis
1. Coordinate Axes:
Effect of choosing scale of axes
2. Stratification:
Stratify the data and then see the
correlation
3. Range of variables:
Select range of variables carefully as it
affects correlation
29
Regression Analysis
Estimating the exact relationship between
dependent and independent variables
Line of best fit joining data
points on a scatter diagram is a
regression line having equation
y=a+bx
where y is dependent
variable, x is independent
variable, a is a constant and b is
regression coefficient
y
x
30
Regression Analysis
Calculations:
1. Calculate x and y
2. Calculate S(xx) and S(xy)
3. Calculate b
b=S(xy)/S(xx)
4. Calculate a
a=y-b x
Then, equation of line is y=a+bx
Note: For both Correlation and Regression
Analysis, draw scatter diagram first
31
Histogram
A bar chart that displays the variation within the
process
Also called a frequency distribution because the
frequency of occurrence of any given value is
represented by the height of the bars
Allows one to quickly visualize whats going on within
a large amount of data
Provides clues to causes of problems
May be used to show the relationship between the
engineering tolerance and the capabilities of the
process
32
How to make Histogram?
Collect measurements(variable data)from a process or
key characteristic.
Thirty or more measurements are preferred.
Construct check sheet to record the data.
Find the range by subtracting the smallest
measurements from the largest.
Using this guide determine the proper number of class
intervals.

33


Observations No. of classes(K)
25 to 50 5 to 8
51 to 100 6 to 11
101 to 250 9 to 13
251 and over 11 to 15
K=R/h +1 (R = Range)
Select h such that K is between 5~8
How to make Histogram
34
Class Class limits Tally Number of observations
1 0.51 to 5.50
IIII IIII
10
2 5.51 to 10.50
IIII IIII IIII IIII
20
3 10.51 to 15.50
IIII IIII IIII IIII IIII
25
4 15.51 to 20.50
IIII IIII IIII IIII
20
5 20.51 25.50
IIII IIII IIII
15
6 25.51 to 30 .50
IIII IIII
10
How to make Histogram
Construct a frequency table by properly making class
boundaries. Tally the number of observations found in
each class.
35
0.5 5.5 10.5 15.5 20.5 25.5 30.5
5
10
15
20
5
25
How to make Histogram
Taking the class interval on horizontal axis, draw the
height of the bar corresponding to frequencies in
interval on the vertical axis.
36
General Type Comb Type Positively Skew Type
Left-hand
Precipice Type
Plateau Type Twin Peak Type
Isolated Peak
Type
Types of Histogram
READING HISTOGRAMS
A. General Type
Shape symmetrical (Bell shaped). If your vendor has less variability but
centre is shifting you can help him to do right setting.
B. Bimodal or Twin Peak Type
Two Distributions with widely different mean values mixed.
C. Comb Type Multimodel
Number of units of data included in class varies from class to class.
Rounding off. Incorrect least count of measurement system
D. Positive Skew
Occurs when lower limit is controlled either theoretically or by
specification value or when values lower than certain value do not occur
E. Left Hand Precipice Type
100% screening has been done because of low process capability or when
positive skewness becomes more extreme. (Check your vendor if in receipt
supply having such pattern)
F. Plateau Type
Mixture of several distributions having different mean values, or 100%
screening on both sides
38
Normal Distribution
x
f(x)
Frequency is the highest in the middle and becomes gradually lower towards the tail.
It is symmetrical
It is denoted by N (, o
2
)
: The centre of the distribution (the mean)
o : The spread of the distribution(the standard deviation)
To obtain a probability in a normal distribution, we standardize by transforming x to a
variable, u= x-
o , We then have a standard measure u, which is distributed as the
standard normal distribution N(0, 1
2
). The normal distribution table gives probabilities in the
standard normal distribution.
1o :- 68.3% of area
2o :- 95.4% of area
3o :- 99.7% of area

39
Process Capability Index
After Histogram shows that it follows normal distribution, a study of
process capability is often undertaken. This is to find out whether the process
can meet specifications or not
Both-sided specifications (S
U
and S
L
)
C
P
= S
U
- S
L
/6s
One-sided specification (S
U
or S
L
)
C
P
= S
U
- x

/3s
Evaluation of process:
1) 1.33 <

C
P
Satisfiable enough
2) 1.00 < C
P
< 1.33 Adequate
3) C
P
< 1.00 Inadequate
DIFFERENCE BETWEEN Cp & Cpk

1. Cp is a ratio of tolerance and six sigma.
It does not talk about process setting. This can be
appreciated that even if variability is small & setting is
out we will have Cp very high. This is therefore not an
effective indicator.
Even though process setting may be totally out, Cp value
could be well above 1.33. Therefore, Cp can be best
described as process potential index.

2. Cpk takes care of setting as well as
variation. Therefore Cpk describes the actual condition

41
Graphs
Graphs are among the simplest and best techniques to
analyze and display data for easy communication

Y e a r R e j C o s t
1 9 9 6 5 0
1 9 9 7 4 5
1 9 9 8 7 0
0
2 0
4 0
6 0
8 0
1 9 9 6 1 9 9 7 1 9 9 8
Y e a r
R
e
j
.

c
o
s
t
Bar Graph
42
Graphs
Mont h Rej
A pr 0. 2
May 0. 6
Jun 0. 2
Jul 0. 2
A ug 0. 15
S ep 0. 2
0
0. 2
0. 4
0. 6
0. 8
Apr May Jun Jul Aug Sep
Month
R
e
j
e
c
t
i
o
n

%
a
g
e
Rej
Trend Chart
43
Graphs.
OTHERS
5.56%
RBS STRG.
ASSY.
10.83%
R&P ASSY.
17.62%
POWER
STRG.
4.92%
AXLE ASSY.
28.67%
PROP.SHAFT
ASSY.
6.15%
COLUMN
ASSY.
26.25
Pie Chart
44
Control Charts
A control chart is a line graph used to display variation
on time ordered fashion
A centerline and control limits are placed on the graph
to help analyze the pattern of the data
Purpose is to separate common causes from special
causes of variation and to help assign causes of
variation

45
How to make Control Charts
Define process parameter to be measured.
Define wherein the process the control characteristics
will be measured.
Select where control chart is to be used.
Determine sample size and frequency.
Take measurements.
Plot measurements on graph.
Connect dots.
After 20 plot points calculate center-line and control
limits.
Analyze pattern for special cause of variation
46
Chance Cause
Assignable Cause
Upper Control Limit
Lower Control Limit
Upper Specification Limit
Lower Specification Limit
Control Charts
47
LCL ( x - A
2
R )
UCL ( D
4
R )
LCL ( D
3
R )
x R Chart:
30
50
40
20
0
60
40
20
x
R
UCL ( x + A
2
R )
x
R
Control Charts
48
Coefficients for x-R Charts
Size of Sub-group X-Chart R Chart R Chart R Chart
n A
2
D
3
D
4
d
2
2 1.880 - 3.267 1.128
3 1.023 - 2.575 1.693
4 0.729 - 2.282 2.059
5 0.577 - 2.115 2.326
6 0.483 - 2.004 2.534
49

pn Chart Number of Defective

p Chart Fraction Defective

c Chart No. of Defects on a
fixed sized Product

u Chart No. of Defects on a
varying sized product

Other Charts
50
How to Read Control Charts
1. Out of Control Limits: Points outside the limits
2. Run: Continuously on one side of center line
Seven-Point length of run is abnormal
10 Out of 11 consecutive points on one side
12 Out of 14 consecutive points on one side
16 Out of 20 consecutive points on one side
3. Trend: Continuous upward or downward curve
51
How to Read Control Charts
4. Approach to the Control Limits
Two out of three points occur outside of 2-sigma
limits is abnormal
5. Approach to the Center Line
When most of the points are within central 1.5-sigma
lines, this is also abnormal. It indicates mixing odd
data in sub-groups
6. Periodicity
When the curve repeatedly shows an up and down
trend for almost same interval, this is also abnormal