Escolar Documentos
Profissional Documentos
Cultura Documentos
• Brainstorm on X’s
• Find change of which X’s affect Y and in what
manner
• Ultimately find which X’s are critical to move the
Y in the desired direction
UNDERSTANDING A PROCESS
•To better understand
your process, you will:
– Create a flowchart of
your process.
– Identify which of
your process steps
are value-added and
which are nonvalue-
added.
– Determine cycle time and identify bottlenecks.
– Look for errors or inefficiencies that contribute to
defects.
104
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FLOWCHARTS
• Flowcharts are tools that make a process visible.
Yes
Decision Step 6 End
No
Step 4 Step 5
105
SQC UNIT BANGALORE, GREEN BELT PROGRAM
106
SQC UNIT BANGALORE, GREEN BELT PROGRAM
ACTIVITY FLOWCHARTS
Hotel Check-out Process Process Name
• Activity 1 2 YES 3
Is there
what happens in 5
7 Consistent
decision points, Start/End
Action/Task
Give room number
level of
detail
YES
11
Clear starting
Pay bill and ending
Date of creation points
or update &
name of creator
107
SQC UNIT BANGALORE, GREEN BELT PROGRAM
DEPLOYMENT FLOWCHARTS
• Deployment People or groups
listed across the top Invoicing Process
flowcharts show the Sales Billing Shipping Customer Elapsed
Time
detailed steps in a Steps listed in 1 Time flows
Delivers goods
process and which column of person or
group doing step or 2 8
down the
page
people or groups are in charge Notifies sales of
completed delivery
Receives
delivery
5 days
of invoice
12
people or functions, as
6
Receives and
records payment
Horizontal lines
clearly identify
they help highlight 7
Reviews weekly
report of overdue
handoffs
handoff areas. accounts
108
SQC UNIT BANGALORE, GREEN BELT PROGRAM
109
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FOUR PERSPECTIVES
111
SQC UNIT BANGALORE, GREEN BELT PROGRAM
COPY PROCESS
Yes
No
Select
Size
Select
Orientation
Select
Number Paper? No Find
Paper
Yes
Yes
Box No Knife? No Find Open
Open? Knife Box
Yes
112
SQC UNIT BANGALORE, GREEN BELT PROGRAM
VALUE-ADDED AND NONVALUE-ADDED STEPS
•Value-Added Step:
– Customers are willing to pay for it.
– It physically changes the product.
– It’s done right the first time.
•Nonvalue-Added Step:
– Is not essential to produce output.
– Does not add value to the output.
– Includes:
• Defects, errors, omissions.
• Preparation/setup, control/inspection.
• Over-production, processing, inventory.
• Transporting, motion, waiting, delays.
113
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLES
Value-Added Activities Nonvalue-Added Activities
an activity 1.
Stop
9.
Adjust
settings
packing line
flowchart into No
2. 5. 6. 7. 8. Yes 10.
an opportunity Same
product?
Yes
Change
length
Adjust
speed
Run test
cartons
Speed
OK?
Adjust
stapler
flowchart by 3.
No
4.
4.
Yes
No 12. 11.
Pick-up Timing Staple
highlighting appropriate
tools
Clean
okay?
machine
Closed?
test carton
Yes
the steps that 18. 13.
Load carton
Stop
add waste and line
No
19.
Start
production
115
SQC UNIT BANGALORE, GREEN BELT PROGRAM
OPPORTUNITY FLOWCHART
Steps that are essential
even when everything
Value-Added
flowchart is Yes Nonvalue-Added
separate No
Yes
Place Glass
value-added Original
No
Dirty? Clean
nonvalue- Select
Orientation
116
SQC UNIT BANGALORE, GREEN BELT PROGRAM
only section No No
Value-Added steps No
with an arrow if
there are no
Nonvalue-Added
steps in between
117
SQC UNIT BANGALORE, GREEN BELT PROGRAM
118
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Value-Added 2 2%
Nonvalue-added
Fixing errors 10 10%
Prep/Set-up
Control/Inspection 6 6%
Delay 52 52%
Transporting/Motion 30 30%
119
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
There are two primary flavors of FMEA:
Design FMEAs are used during process or product design and
development.
The primary objective is to uncover problems that will result
in potential failures within the new product or process.
Process FMEAs are used to uncover problems related to an
existing process.
These tend to concentrate on factors related to manpower,
systems, methods, measurements and the environment.
Although the objectives of design and process FMEAs may
appear different, both follow the same basic steps and the
approaches are often combined.
120
SQC UNIT BANGALORE, GREEN BELT PROGRAM
121
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
FMEA drives systematic thinking about a product or process by
asking and attempting to answer three basic questions:
What could go wrong (failure) with a process or system?
How bad can it get (risks), if something goes wrong
(fails)?
What can be done (corrective actions) to prevent things
from going wrong (failures)?
FMEA attempts to identify and prioritize potential process or
system failures. The failures are rated on three criteria:
The impact of a failure - severity of the effects.
The frequency of the causes of the failure - occurrence.
How easy is it to detect the causes of a failure -
detectability.
122
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Notice that only the causes and effects are rated - failure
modes themselves are not directly rated in the FMEA
analysis.
FMEA is cause-and-effect analysis by another name – avoid
being hung up on the failure mode.
The failure mode simply provides a convenient model, which
allows us to link together multiple causes with multiple
effects.
It is easy to confuse failures, causes, and effects, especially since
causes and effects at one level can be failures at a lower level.
Effects are generally observable, and are the result of some
cause. Effects can be thought of as outputs.
123
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Effects are usually events that occur downstream that affect
internal or external customers.
Root causes are the most basic causes within the process
owner’s control. Causes, and root causes, are in the
background; they are an input resulting in an effect.
Failures are what transform a cause to an effect; they are often
unobservable.
One can think of failures, effects, and causes in terms of the
following schematic:
124
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Note that a failure mode can have numerous distinct effects, and
that each effect has its own system of root causes.
125
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Keep in mind that a single failure mode can have several effects,
and that the cause-and-effect diagram on the previous slide
should be repeated for each effect of the failure!!!
127
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Once the RPN values are assigned to each of the cause-and-
effect pairs identified by the team, the pairings are
prioritized.
The higher the RPN value, the higher the priority to work on
that specific cause-and-effect pair.
The measurement scale for the SOD values is typically a 5 or
10 point Likert scale (an ordinal rating scale).
The exact criteria associated with each level of each rating scale
is dependent upon either a company designed rating criteria
or a specified rating criteria from an industry specific
guideline.
We recommend the use of a 10 point Likert scale for each of the
three rating criteria: Severity, Occurrence, and Detection.
128
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Note: The 10 point scale facilitates the ability of the team to
assign ratings in a timely fashion.
Too many levels can lead to a false sense of precision and a
lot of agonizing over the exact rating to be assigned for each
item.
The rating systems used should be developed to reflect the
specific situation of interest.
Recall the example of the completed FMEA presented earlier.
The FMEA was developed by a team studying OS/390 online
systems availability to end users.
On the following slides, we will see the ratings systems agreed
upon by the team.
129
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
The more critical the effect, the higher the severity rating.
130
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Severity ratings for the online systems availability FMEA.
131
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Occurrence, or frequency of occurrence, is a rating that
describes the chances that a given cause of failure will occur
over the life of product, design, or system.
Actual data from the process or design is the best method for
determining the rate of occurrence. If actual data is not
available, the team must estimate rates for the failure mode.
Examples:
The number of data entry errors per 1000 entries, or
The number of errors per 1000 calculations.
An occurrence value must be determined for every potential
cause of the failure listed in the FMEA form.
132
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
The higher the likelihood of occurrence, the higher the
occurrence value.
133
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Occurrence guidelines for the system availability FMEA:
4 1 in 2000
6 1 in 80
7 1 in 20
High High number of occurrences of cause
8 1 in 8
9 1 in 3
Very High Very high number of occurrences of cause
10 ³ 1 in 2
134
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
The detection rating describes the likelihood that we will detect
a cause for a specific failure mode.
An assessment of process controls gives an indication of the
likelihood of detection.
Process controls are methods for ensuring that potential
causes are detected before failures take place.
For example, process controls can include:
• Required fields or limited fields in electronic forms,
• Process and/or system audits, and
• “Are you sure” dialog boxes in computer programs.
If there are no current controls, the detection rating will be high.
If there are controls, the detection rating will be low.
135
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
The higher a detection rating, the lower the likelihood we
will detect a specific cause if it were to occur.
Detection guidelines developed for the system availability FMEA:
Detection Rank Criterion
Low 6
Low likelihood of detection
Very Low 7
Very low likelihood of detection
Remote 8 Cause is hard to identif y
136
SQC UNIT BANGALORE, GREEN BELT PROGRAM
137
SQC UNIT BANGALORE, GREEN BELT PROGRAM
FMEA
Conducting an FMEA: Basic Steps
1. Define the scope of the FMEA.
2. Develop a detailed understanding of the current process.
3. Brainstorm potential failure modes.
4. List potential effects of failures and causes of failures.
5. Assign severity, occurrence and detection ratings.
6. Calculate the risk priority number (RPN) for each cause.
7. Rank or prioritize causes.
8. Take action on high risk failure modes.
9. Recalculate RPN numbers.
138
SQC UNIT BANGALORE, GREEN BELT PROGRAM
139
SQC UNIT BANGALORE, GREEN BELT PROGRAM
140
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Man Machine
Problem
/ Effect
Material Method
Spark plugs
Impatience Heavy Contacts
Poor Bad Life
Craze
anticipation attitude
Body Technical
Wrong Poor details
skill Shape
Always gears Fuel mix
late Lack of Inexperience High H.P Carburetor
awareness Wrong
Riding on culture Engine
clutch
Cylinders High Petrol
Spurious Consumption
Crossings
Restrictions Spares Impurities
Traffic Incorrect
One way
No turn Tyres Inferior Octane no.
Frequent Petrol
Faulty
Circuitous stops Negligence
Speed Breakers pressure Additives
Road
Ignorance
Potholes Irregular Incorrect viscosity
Low pressure
Poor servicing
Clogged
condition Oil
False filters
Steep economy Not changed
Low level
Road Maintenance Materials
143
SQC UNIT BANGALORE, GREEN BELT PROGRAM
144
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Validation of Causes
146
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CORRELATION
If two variables X and Y, are related such that as Y
increases / decreases with another variable X, a
correlation is said to exist between them.
35
Mileage (km/Lit)
30
25
20
15
25 35 45 55 65 75
Speed (km/h)
148
SQC UNIT BANGALORE, GREEN BELT PROGRAM
SCATTER DIAGRAM
• A scatter diagram depicts the relationship as a
pattern that can be directly read.
• If Y increases with X, then X and Y are positively
correlated.
• If Y decreases as X increases, then the two types of
data are negatively correlated.
• If no significant relationship is apparent between X
and Y, then the two data types are not correlated.
149
SQC UNIT BANGALORE, GREEN BELT PROGRAM
150
SQC UNIT BANGALORE, GREEN BELT PROGRAM
DATA ON CONVEYOR SPEED AND SEVERED LENGTH
Sl. No. Conveyor Severed Sl. No. Conveyor Severed
Speed Length Speed Length
(cm/sec) (mm) (cm/sec) (mm)
1 8.1 1046 16 6.7 1024
2 7.7 1030 17 8.2 1034
3 7.4 1039 18 8.1 1036
4 5.8 1027 19 6.6 1023
5 7.6 1028 20 6.5 1011
6 6.8 1025 21 8.5 1030
7 7.9 1035 22 7.4 1014
8 6.3 1015 23 7.2 1030
9 7.0 1038 24 5.6 1016
10 8.0 1036 25 6.3 1020
11 8.0 1026 26 8.0 1040
12 8.0 1041 27 5.5 1013
13 7.2 1029 28 6.9 1025
14 6.0 1010 29 7.0 1020
15 6.3 1020 30 7.5 1022
151
SQC UNIT BANGALORE, GREEN BELT PROGRAM
1050
1045
1040
Severed Length (mm)
1035
1030
1025
1020
1015
1010
1005
1000
5 5.5 6 6.5 7 7.5 8 8.5 9
Conveyor Speed (cm/sec)
152
SQC UNIT BANGALORE, GREEN BELT PROGRAM
153
SQC UNIT BANGALORE, GREEN BELT PROGRAM
REGRESSION
Regression is the prediction of dependent variable
from knowledge of one or more other independent
variables.
Regression Analysis is a statistical technique for
estimating the parameters of an equation relating
a particular value of dependent variable to a set of
independent variables. The resulting equation is
called Regression Equation.
Linear regression is the regression in which the
relationship is linear.
Curvilinear regression is the regression in which
the best fitting line is a curve.
154
SQC UNIT BANGALORE, GREEN BELT PROGRAM
155
SQC UNIT BANGALORE, GREEN BELT PROGRAM
156
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONTROL CHARTS
FOR ANALYSIS
157
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONTROL CHARTS
Type of Data Common Type of Chart
Variable X-R, IX-MR
Attribute p, np, c, u
The control charts
• Isolate sources of unwanted product and process variation
• Aid in deciding on process capability
• Serve as basis for making judgement w.r.t. product and
process variation
159
SQC UNIT BANGALORE, GREEN BELT PROGRAM
160
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Target
161
SQC UNIT BANGALORE, GREEN BELT PROGRAM
162
SQC UNIT BANGALORE, GREEN BELT PROGRAM
164
SQC UNIT BANGALORE, GREEN BELT PROGRAM
W. Edwards Deming
165
SQC UNIT BANGALORE, GREEN BELT PROGRAM
How:
By plotting sample averages (X, pronounced X-bar) and
ranges (R) on separate charts. This allows for independent
monitoring of the process average and the variation about
that average.
Conditions:
• Constant sample size.
• One characteristic per chart.
• Should have no less than 20 samples before calculating
control limits.
166
SQC UNIT BANGALORE, GREEN BELT PROGRAM
167
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Conditions:
• All parts on the chart are the same units and have similar
average ranges ( R’s).
• Constant sample size.
• Should have no less than 20 samples before calculating
control limits.
170
SQC UNIT BANGALORE, GREEN BELT PROGRAM
UCL = Coded X-
dbar + A2.
Coded X-dbar Coded X= (X
X
= Coded /k target value)
LCL = Coded X-
dbar - A2. 2 to 9 but
3 to 5
preferred
UCL = D4.
R = R/k R
LCL = D3.
171
SQC UNIT BANGALORE, GREEN BELT PROGRAM
172
SQC UNIT BANGALORE, GREEN BELT PROGRAM
174
SQC UNIT BANGALORE, GREEN BELT PROGRAM
175
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Example:
If yield X goes down:
•Check for Pressure
•Check for Temperature
•Check for Material flow
176
SQC UNIT BANGALORE, GREEN BELT PROGRAM
TEST OF HYPOTHESIS
177
SQC UNIT BANGALORE, GREEN BELT PROGRAM
TEST OF HYPOTHESIS
A Hypothesis is a statement about the probability law of
a random variable, which can be sampled.
Test of Hypothesis is sampling the random variable
whose probability law is referred to and on the basis of
the sample, deciding to accept or reject the stated
hypothesis.
If sample results seem consistent with the hypothesis,
accept the stated hypothesis.
If sample results do not seem consistent with the stated
hypothesis, we would like to reject it.
178
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
• A development man, following up a bright idea, will often get
small sample evidence which favours his hunch. A sensible
and cautious man will at once try to put bias out of court by
considering the possibility of the apparent value of his hunch
being due to pure chance. Suppose he were asking whether a
new process he had thought of were better than the existing
process, then he might sensibly adopt what the statistician
calls a Null Hypothesis, i.e., he would assume that there was no
real significant difference between his pet process and the
standard. He would assume, provisionally, that the sample
results obtained by his new process might well have come from
the same population as results obtained by the standard
process. The position then would be that his pet process had
produced a sample of above average quality.
179
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
• His next step would be to calculate the probability that the
standard process would give a sample as good as that obtained
by the new process. If it proved that the chance of the
standard process giving so good a sample were extremely low,
then, although his sample were small, he would be justified in
rejecting the Null Hypothesis, on the grounds that it seemed a
very unlikely explanation. It would then be fair - and
unbiased to conclude that his new process could be accepted as
having a real superiority to the standard process. On the
other hand, if it proved that such a sample might arise with
fair frequency from the standard process, it would be rash or
dishonest to claim the new process as superior to the standard
process.
180
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
When we get a result which is very unlikely to have arisen by
chance we say that the result is statistically significant. By this
we mean simply that it would be rather fantastic to ascribe it to
chance, that the difference must, in all common sense, be
accepted as a real difference. Since the judgment is based on
probability, falling short of absolute certainty, we indicate our
degree of confidence in the reality of the difference by describing
it as ‘significant’ or 'highly significant' depending on the
probability level associated with our judgment. Thus a result that
would only arise in one trial in twenty on the basis of pure chance
we should describe as ‘significant’. A result that would arise on
the basis of pure chance only once in a hundred trials we should
describe as 'highly significant'. The proper thing to do, of course
is not simple to use words of this kind but to quote the level of
probability p = 0.05, p = 0.01.
181
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
a. Formulation of Null Hypothesis (H0)
To find out whether a "real change" is indicated by the
experimental results, the hypothesis that there is "no real
chance" is formulated. This hypothesis of no real change is
called the null hypothesis. Whether the null hypothesis is
tenable in the light of experimental results is tested.
182
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
c. Decision on null hypothesis
183
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF SIGNIFICANCE
In practice,
i. the value corresponding to the probability level of 0.05 or
0.01 is obtained from the statistical tables relevant to the
null hypothesis.
ii. the statistical test criterion based on experimental data is
compared to the value obtained from the tables.
iii. the decision reject the null hypothesis (Ho) is taken when
the p value is less than 0.05, otherwise the null
hypothesis is accepted.
iv. When the null hypothesis is rejected, the result is said to
be ‘statistically significant’. The alternative Hypothesis
(Ha) is accepted.
184
SQC UNIT BANGALORE, GREEN BELT PROGRAM
HYPOTHESIS
If one is asked if one sample say (A) is similar to another
sample (B) the answer can be
Yes, both are similar (or)
No, both are not similar
185
SQC UNIT BANGALORE, GREEN BELT PROGRAM
186
SQC UNIT BANGALORE, GREEN BELT PROGRAM
187
SQC UNIT BANGALORE, GREEN BELT PROGRAM
188
SQC UNIT BANGALORE, GREEN BELT PROGRAM
189
SQC UNIT BANGALORE, GREEN BELT PROGRAM
H1 PICTORIALLY CALLED
2 tailed test
A B
190
SQC UNIT BANGALORE, GREEN BELT PROGRAM
DIFFERENCE BETWEEN
2 TAILED AND 1 TAILED TEST
• If = 5 % i.e. 0.05
2 TAILED 1 TAILED
191
SQC UNIT BANGALORE, GREEN BELT PROGRAM
PICTORIALLY
A A
B B B
Critical Critical Critical
Value Value Value
H0 : 1 2 t0
X 1 X 2
1 1
sp
n1 n2
H0 : 1 2
H 1 : 1 2
H0 : 1 2
H 1 : 1 2
194
SQC UNIT BANGALORE, GREEN BELT PROGRAM
195
SQC UNIT BANGALORE, GREEN BELT PROGRAM
196
SQC UNIT BANGALORE, GREEN BELT PROGRAM
197
SQC UNIT BANGALORE, GREEN BELT PROGRAM
199
SQC UNIT BANGALORE, GREEN BELT PROGRAM
200
SQC UNIT BANGALORE, GREEN BELT PROGRAM
d
t0
Sd n
1 n
where d dj
n j 1
2
n
d j
d j 1
n n
2
j d d 2
j
j 1 j 1 n
and S d2
n1 n1
203
SQC UNIT BANGALORE, GREEN BELT PROGRAM
H1 : 2 2
0
S 12
H 0 : 12 22 F0
S 22
H 1 : 12 22
204
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLE-TEST ON VARIANCES
• We will test the hypothesis that the variance of the road octane
numbers for the two gasoline formulations in the “example given
in Test on Means” are the same; that is,
H 0 : 12 22
H1 : 12 22
Since S12 1.34 and S 22 1.07, the test statistic is
S12 1.34
F0 2 1.25
S 2 1.07 P value = 0.3725
205
SQC UNIT BANGALORE, GREEN BELT PROGRAM
207
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Example
There was a change in the process and the process owner claimed that the average cycle time has
reduced.
The data shows the follow
Old process: 15, 20, 12, 5, 30
New process: 8, 9, 12, 11
Is his claim tenable or true?
Ho(Null Hypothesis): There is no difference in the mean delivery time between old process & new
process:
Mu Old = Mu New
Steps for MINITAB
1. STAT>Basic Statistics>2 Variances>Click>enter variables>OK
2. Interpret P-Value in F-Test
3. If P-Value<0.05 reject Ho
4. Declare characteristics are different
5. Look at the Sigma Values in the Session
6. STAT>Basic Statistics>2 Sample t>Click>enter variables>Click/No click “Assume equal variances”
based on
Step 3 results> (Ensure Alternatives in Options is “ not Equal”)>Graphs>Boxplot> OK
7. If P-Value<0.05 reject Ho
8. Declare characteristics are different
9. Look at the Summaries in the Session
10. Use Box-plot graph as a presentation tool
208
SQC UNIT BANGALORE, GREEN BELT PROGRAM
10
209
SQC UNIT BANGALORE, GREEN BELT PROGRAM
210
SQC UNIT BANGALORE, GREEN BELT PROGRAM
THE CHI-SQUARE TEST
We talked of averages and standard deviations but we
may also have to deal with attributes.
For attribute data we use the Chi-Square Test ()
(Pronounced as Ki of Kite).
This Test, tests the frequency of the actual occurrence
vs. the frequency of expected occurrence to help decide
whether significant change has occurred.
E.g.: A coin is flipped 100 times. Heads occurs 40 times while
tail occurs 60 times.
Did it happen by chance? (or) Is the coin a trick coin ?
211
SQC UNIT BANGALORE, GREEN BELT PROGRAM
2 k (Oi - Ei )2
0
i1 E
i
It can be shown that χ 2 approximately follows chi square distributi on with k p 1 degress of
0
freedom, where p represents the number of parameters of the hypothesized distributi on estimated
by sample statistics . This approximation improves as n increases. We would reject the hypothesis
that X conform s to the hypothesized distributi on if 2 2 .
0 , k p1
212
SQC UNIT BANGALORE, GREEN BELT PROGRAM
0 1 2 3 4 5 6 7 8 9 Total n
Observed 94 93 112 101 104 95 100 99 108 94 1000
frequencies, Oi
Expected 100 100 100 100 100 100 100 100 100 100 1000
frequencies, Ei
213
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Since 2 16.92, we are unable to reject the hypothesis that the data come fom a discrete
0.05, 9
uniform distributi on. Therefore, the random number generator seems to be working satisfactorily .
214
SQC UNIT BANGALORE, GREEN BELT PROGRAM
215
SQC UNIT BANGALORE, GREEN BELT PROGRAM
216
SQC UNIT BANGALORE, GREEN BELT PROGRAM
OBS 15 21 45 13 94
BANG
EXP 22.51 20.99 38.94 11.5
OBS 26 31 34 5 96
HYD
EXP 22.9 21.44 39.77 11.81
OBS 33 17 49 20 119
MOHALI
EXP 28.5 26.57 49.29 74.63
= (4-1) x (3-1) = 6
The p value is 0.0039
Hence H0 is Rejected
220
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Example
Data were collected for revisions in the drawing Region wise
Eastern: 10 out of 25, Northern: 2 out of 10,
Is the efficiency of the Northern region in getting the drawing approved better?
MINITAB Steps
Stat>Basic statistics>2 proportions > Summarise data>Enter Trials & successess(or failures whichever is
less)>OK
Interpret P-Value in the sessions as earlier
Test and CI for Two Proportions
Sample X N Sample p
1 10 25 0.400000
2 2 10 0.200000
221
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Example
There are three methods for welding a tube
Method 1 - Single-V Weld Prep – 80 joints inspected 5 were reported unacceptable
Method 2 – J-prep – 100 joints inspected 2 were reported unacceptable
Method 3 – Square ends – 50 joints inspected 10 were reported unacceptable
What do you conclude about the performance of the different methods?
222
SQC UNIT BANGALORE, GREEN BELT PROGRAM
ANALYSIS OF VARIANCE
(ANOVA)
223
SQC UNIT BANGALORE, GREEN BELT PROGRAM
225
SQC UNIT BANGALORE, GREEN BELT PROGRAM
226
SQC UNIT BANGALORE, GREEN BELT PROGRAM
227
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLE-ANOVA
• The cycle time in hours for closure of customer complaints from
the four regions were recorded as follows (Data are coded):
Manager feels that the regions affect the time for closure of a
customer complaint
REGION
E W N S
66 74 55 52
65 71 56 49
72 60 55 55
69 65 49 53
70 66 53 51
342 336 268 260
228
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLE-ANOVA
The method of computation for analysis of variance is as follows:
1. Sum of all observations: 66 + 65 + … + 51 = 1206.
2. Number of observations (N): 20
3. Square (1) and divide by (2): (1206)2/20 = 72721.8. This is
called correction factor (CF).
4. Square each individual observations and add:
662 + 652 +............ + 512 = 74060
5. Square each column total and divide their sum by the number of
observations in each column:
3422+3362+2682+2602 = 73856.8
5
229
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLE-ANOVA
6. Total sum of squares : (4) - (3) = 1338.2
degrees of freedom for this are N - 1 = 19.
7. Sum of squares between regions: (5) - (3) = 1135.0
degrees of freedom for this = No. of levels -1 = 4 - 1 = 3.
8. Residual sum of squares: (6) - (7) = 203.2
degrees of freedom for this = 16.
The analysis of variance table is now set up as follows:
Source of variation Sum of Degrees of Mean F
Squares freedom Squares
Between regions 1135.0 3 378.3 29.8**
Total 1338.2 19
230
SQC UNIT BANGALORE, GREEN BELT PROGRAM
EXAMPLE-ANOVA
231
SQC UNIT BANGALORE, GREEN BELT PROGRAM
Example
Case1 Ho: There is no difference among the methods w.r.t. variation in Hardness: 21= 22 = 23
Case 2 Ho : There is no difference among the methods w.r.t. average Hardness: 1= 2 = 3
233
SQC UNIT BANGALORE, GREEN BELT PROGRAM
234
SQC UNIT BANGALORE, GREEN BELT PROGRAM
210
200
response
190
180
methods
method1
method2
method3
235
SQC UNIT BANGALORE, GREEN BELT PROGRAM
CONCEPT OF
DESIGN OF
EXPERIMENTS (D.O.E.)
236
SQC UNIT BANGALORE, GREEN BELT PROGRAM
DESIGN OF EXPERIMENTS
Design of experiments (DOE) is a valuable tool to
optimize product and process designs, to accelerate the
development cycle, to reduce development costs, to
improve the transition of products from research and
development to manufacturing and to effectively
trouble shoot manufacturing problems. Today, Design
of Experiments is viewed as a quality technology to
achieve product excellence at lowest possible overall
cost.
237
SQC UNIT BANGALORE, GREEN BELT PROGRAM
IN DESIGN OF EXPERIMENT
239
SQC UNIT BANGALORE, GREEN BELT PROGRAM
OBJECTIVES OF EXPERIMENTATION
The following are some of the objectives of
experimentation in an industry :
241
SQC UNIT BANGALORE, GREEN BELT PROGRAM
242