Component Search Method

(Simple Smooth Swapping)

1
Hierarchy of Defects

Time to Time

Unit to Unit

Within Unit
2
Principle of Component Search:
Story of train Travel through a tunnel in UK

3
Old Lady Young Woman

Soldier General

4
5
Long Train Tunnel

6
•British
•Formal people
•Will not talk to anyone
without proper introduction
•All was quiet in train
compartment
•Lights went out
•Soon thereafter, all four in
compartment heard sound of
kiss, followed by tight slap!
Now all four started to
speculate about who was
responsible

7
 1. Old lady’s theory:

“It has to be general, old goat that he is.

He kissed the young woman & young woman
slapped him back.”

8
 2. The young woman’s Theory:

“Damn, I'm pretty one. Handsome private tried

to kiss me, but missed me & got old bag
instead, who returned favor with slap”
9
 3.The general’s Theory:

He got slapped & He thought:

“It's got to be this young soldier, scum of army
He kissed young woman & I got brunt of attack
because I was sitting opposite her & she suspected
me”

10
 4. Soldier’s True Story:

As things were quiet & then became dark, he

seized his moment of opportunity
He kissed his own hand, slapped general &
got away with it!

11
•Principle of Component Search
•You actually swap parts & or
subassemblies from good unit to bad unit
& vice-versa
•See if problem follows swapped part or
remains in rest of unit

12
 QC?

*WOW ** ** **BOB
•Only extremes of distribution need be considered
•Capturing two extremes, captures full range of Green Y variation
•Causes can then be systematically distilled by swapping
components
 LSL USL
X X
WOW BOB
“WOW” & “BOB” units:
•May be “accept” & “reject” units

14
 LSL USL
X X
WOW BOB
“WOW” & “BOB” units:
•May be acceptable, but widely separated in terms
of Green Y measurement

15
Components Search in 12 Steps:
Step#1:
Select one BOB & one WOW assemblies
from day’s production

16
Step#2:
Disassemble and reassemble BOB & WOW twice &
measure their respective Green Ys twice more.
•In all how many Green Y Readings?

17
 Step#3: Test of Significance:

To determine if difference between BOB & WOW is statistically significant

3 Green Y's of BOB must all rank better than the Green Y's of WOW, with no
overlap

WOW BOB
Y
123 123 18
•D = Difference between medians of BOBs & WOWs
•Av = Common Delta = Average difference within the
ranges of 3 BOB & 3 WOW measurements = ( W +B) / 2
•D/Av to be > 1.25 (How did we get this number?)

DD

WOW BOB
123 123
W B 19
Step#4: If the D:Av < 1.25, Stage 1 of Components
Search has failed

Means that Green Y's do not remain constant

Means Red x - in Assembly process

Calls for progressive step-by-step disassembly &

reassembly to determine which assembly process
step is Red X (see Figure 10-2).

20
•If D:Av > 1.25, Red X is in one or more of
components

•Stage 1 is successful

•Swapping of components can begin

21
Step#5.
•Starting with most likely subassembly, rank
components in descending order of likelihood
•Your guess may or may not be accurate to
succeed

22
Step#6.
• Switch the top-ranked component (call it A) from the BOB to
the WOW & its counterpart from WOW to BOB.
• Measure and record the two new Green Y's

A B C D A B C D
WOW BOB

A B C D A B C D

23
Step#7.
There are three outcomes possible in Step 6:
(a)The BOB remains a BOB & WOW remains WOW

A B C D A B C D
WOW BOB
A B C D A B C D
WOW BOB
Interpret meaning of result
Your Conclusion? 24
Step#7.
Possibility #1:
BOB remains a BOB & WOW remains WOW

A B C D A B C D
WOW BOB
A B C D A B C D
WOW BOB
Component ‘A’ – unimportant
‘A’ is not the Cause 25
 Possibility #2:
WOW becomes BOB & BOB becomes WOW

A B C D A B C D
WOW BOB

A B C D A B C D
BOB WOW

Outcome means ?
26
 Possibility#2:
WOW becomes BOB & BOB becomes WOW

A B C D A B C D
WOW BOB

A B C D A B C D
BOB WOW

Outcome means that switched component ‘A’

- important & - solid Red X
Components Search - over 27
Possibility#3: WOW improves partially toward a BOB &
BOB deteriorates partially moving toward WOW

A B C D A B C D
WOW BOB

A B C D A B C D
Improvement Deterioration

Result means ?

28
Possibility#3: WOW improves partially toward a BOB &
BOB deteriorates partially moving toward WOW

A B C D A B C D
WOW BOB

A B C D A B C D
Improvement Deterioration

•Result means that ‘A’ may be important, but - not the only
problem component
•Some other component causing partial change needs to be
discovered in subsequent swaps
•Such partial change – also indication of interaction of ‘A’ with
another component 29
To determine extent of whether such partial
change - important or not, decision limits must
be calculated
Any swap that results in a point outside home
decision limit, for either BOB or WOW-means
that component ‘A’ is now definitely important,
along with one or more other components yet
to be discovered

30
Step# 8. In each of three possible outcomes of
Step 7, restore ‘A’ from original BOB back to
rest of BOB & ‘A’ from original WOW back to
rest of WOW to restore original Green Y's of
BOB & WOW

31
Step#9.
Repeat Steps 6, 7(a) or 7(c), and 8 with next
most likely subassembly, or component B, then
C, then D, etc.

32
Step#10.
Once two or more components are identified
as important in Step 7(c) & Step 9, switch
them simultaneously between BOB & WOW
until complete reversal occurs, i.e., BOB
becomes WOW & vice-versa
Red X - combination of these important
components & their interaction effects

33
Step#11.
Capping run - performed by testing all of
untested & unimportant subassemblies /
components as group as WOWs, against all of
important subassemblies / components as a
group as BOBs
Outcome should be close to Stage 1 BOBs.
Then flip-flop, with unimportant components as
BOBs & important components as WOWs. The
result should be close to Stage 1 WOWs.

34
Step#11.

Capping run - performed by testing all of untested & unimportant subassemblies

/ components as group as WOWs, against all of important subassemblies /
components as a group as BOBs
Outcome should be close to Stage 1 BOBs. Then flip-flop, with unimportant
components as BOBs & important components as WOWs. The result should be
close to Stage 1 WOWs.

A B C D R A B C D R
WOW BOB
A B C D R A B C 35D R
WOW BOB
Step#12.
Finally, full factorial analysis, using all data
generated in Stages 1 and 2, - used to
determine, quantitatively, magnitude &
direction of main effects & interaction effects
Only calculation, not new experiment

36
Thank You
pgschool@nttf.co.in
ROBUST DESIGN 37