s'i

 

2‘

..
4531

H3 1‘7:

21532;.

J

" ' “‘fi
D

i.

3‘ .- -
can.
u... 5“

.—
M

‘u

90'

.._~iy'31 . .

.-...—...o ‘—

king:

.

.. _-.‘ I.
'-« ow—‘u

-.-v
n...

«.4.

——'~“
V

-‘

7‘- .L

- "OM-‘v
P9313"

‘1' ..1_.4 I
u. \WYnA

, -

 

"'u IV

5.4—.--

~"» -.
an‘

‘ ~‘
vv

WW

VI—‘v—l

 

_-._r-

 

   
    
 
   
 
       
 

 

 
     
    
     

  
      
       

 
 

 

   

     
       
    
 
  
   
       

 

  

    

     

:3' :3 ' :; 1""13‘3733- "1717'“ 3": .. ‘3'77‘~7~"'-'.1".3~ 7'17" :1 1
1111:1'11»'17’:"1'-1137!113"13113133133111.3133? 1' 35131333133” .3333373"‘.3133...~w.t,.. 11’1

} < 31', ... o ‘ O l 3} ,5 I i

3331" 33 33- :‘3 1 " " 3313333 3.333313-33‘33'1? 3'33 . .,,31r '331 3 3 ' 3
}},1111'11111111111l"3~11}:11r V71 31111” 31"1}}‘..1}:"I1£~ 1:3: .I1.1.}'131v11,}: 17511.3!l11g‘11‘1’ 1}..111 : .31 .1 11133111,;5;1L315}}L?1 ‘31,}.
}, }_‘ ’ , 3. 1.""!' }3}.I3 ,.F-' 1111113311 .111.111-1"1 1};1:1;1-1113111:}11.".1- }_'} 5}}15111zh'111' I 3}“ 1111111111}. ti“

}.I .333:‘..1<!71."{“ “‘0 }11[1' 11‘f.'111!11|!1’}111111‘31t}ny’all“...‘31 41] 33,-1.1”, .} italifivé’lh‘fii1 '

3t ' -‘ 'fi'l ‘1': '11 1'1'111"1.1 1 1 I 1L} 31|:"13'.111.1}1111.3 1} 1 91

1 . . 3 33'11213333333133’}1.3333.1"" 1.1333371333231333333333333.73:3»..3. 53's E"? 33333“ 3331—1 } 3‘51
333.. ‘ 3. 3.9; 3 4.33.3.3 .3... 3. ”33:33 3:333 31 3313‘» 33:33:33? 1E1 3 3 } - ~41-
. WAN. 1. , "1}1 }I 11’31'1E1 1‘23 1111111111111111"11-}} . I, 1,1,} . .131, l11}111b".11'}m1‘°"1~1}1}!1f}i§’§ {as}; :.}:m“ .513... - F
191"} 11}‘1‘}'1"'i'1}}}},r1}1}1}}} 11}}1 I} ,2. '1'1‘11 \-}111‘1'1 1 fi;},.},1._1fi _ }a- . -l IQ is} :1- z ‘25
1111.1}}1}}13“111111h1“1“ :in lt}:lg. 1.'.,}11}}1§1}‘1--11,}311313332, 1}! }313H .}} .', ..}}}},:}1}}}.‘.I g};},}}}.g 3' 1:11:11; '11- '1 ' _ at».
3EHK§3M33333333 MHFE“3E¥31}333335 1} 1i 311 33333fi- ’1 'Miékffig
:3}, j 1111,1111”: 3 1.111}}I!“}1"131 "311311-1'311i11113 I111} E1113} J13} 111} 1i1.}1311-31| }11}}}'l1d}\}1}.1—1 , .--,_, LE.” {,4 "3'.-
'3 .1-3 3 }}..."...‘..H3~. v'|‘.'.1'3y. 111111111E1111111 I33 33.1.11? '3’“ 1} L511, 3 {1'11 . 3
, “ . ‘v 1 3 }}.}I}'I, Lit-4:3“ 11:}1"1111"‘}1‘\1':1.}111 }"2h’},}.«1~}3 11:1'41} mph} 1W%ti}t a}:

3. 33:33333333333333.1333 3’3‘*3 33333 3W 333.3.333. 33“ 333 3:3:=:1.-.i3333 3
} 313131.11.}33.}?31.}}-}1:;}1}}l1,1;}731,111, 3H,}; 13.3333'1‘331513} ”13131334313311”: 11"? .19.‘1c{‘}:1.1'1:1%‘}1€%‘¥t3 «ox-1 3}-

3’3. 1.333 "3.337333 3 3 33 39:3- 333 3.. . 3.3, 33333 '1": 33.3.- 33 33. 33.3}:- 3333 .33
E333 13313 111'11"}l}31.3 131" W'- 53 '31 '13113;‘:‘}3}13}'33. 3'13‘1‘ 5"}i11 $31.1 . 313133 HI 13111 1151" "
I}}}fl}!!! 51- [$1901. }}}t}3}1f “fig; 3.. i h; .1”: 1|}} 1111111111811F111}1}1}11111}11311111111 (I; séréa‘t 1&1g, '31:! ' r

m”1"fi}}1}}1)111}}}151>fi[}1 ., “1’1‘1‘ir1!!}'l' 12;, $.11{;a$§t'§'i‘i\iiihfi '21I3'L‘uJ‘

1131;}.13'11" 3. .3341 31, -3 3, ..: 13,,»3 117-11313 111331111“ 1’3}131'1 1,5. - 1““‘W3‘311'33‘1111' .333.

‘ : :6, D‘I.1€.:.w’l

7}1i'}1$1:111}11,1. }1}1111}‘1'}‘}111:}11:11,1.:K}1§1}}111{Engéggg’k III‘EIi: "‘1! Ivy}::d§ ELK ‘I '1 I}? 3: .>%
:,= Ithn 1111!!! :, ' .111 . .f! "Kfi
1381111311}:3H}}}1111 11‘33113113“ 111-11131313 11 1111} (1111111 }11113‘ 1131313 ‘ i‘ i “1111}: 4.}1'1 £1111

£11311“! 11"}“1‘};:}2"11}}3}0}'L:'l:}I}Il}:'1;-‘-“13111}111:ki1.‘r1c’1}}111£' 'g})uun1£“ 1: 5-!}}}1‘1‘}}‘{ '2?9'}} iglt$315512111§

p}; .3: ‘ 33,.- ‘33 ,‘-‘ '31! 3 3. 31": .. .3. :7 ': ~11
1}}i’:;}3,};f}}1}11}1 ‘51.} 1}11}Hg111}11ufl 1111}}111'11.1_ 1}}.23'.323311-zr)} 01}? 1}}I. }}J}' 1}} “1”} I 1~11}}I(‘11§1§‘1}‘§11: 15I1§1;E'${1'11"111;1 1. I 1;?

I ' E 1 I 1' :{1'

3 v .
"V113. }I}3:,_,'_}‘:.‘3 1‘11“}
”‘}'*11} }33-3~3333}* 333? 333 Va! 333 ~33}1V11‘31333 33§3133333':} ‘““’1" 11
3 3'3 '3"

      

    
  
 
 
    
    
       
   
    

   

     

  

       

   

    

  
  

  

 

 
  
       

 
 

 

  
 
    
   

 
     
  

5-3
‘311 3'. ”'1', ‘Eiu! l‘ix‘é‘g’} I“.
111311 311111131315‘5‘33'3 ,3 3.311%} $33.23... 313
1"‘1111 1 ““111. 1} 33W ~~.:,::?.'f3_.3,.,.,3,}§. 1.3331533: . ., “"3333“, ’EE 1'33"i,,..,;.,, ‘3
Nu} 111111}£}1‘1 {$111111}. ”Lin” . ‘1‘}'.1}.1'J(r.231}}3,}131,‘ 311:4}111l1u1i1géig {1131;1gr-1gfi'
.}°'11i}.}‘} ?} Y ,1 }} 11‘131 1’1ff'3}-1'71:1}111"' 1.1-}‘1'1$83
1: 3311111111111? }3'3..3 ~33} "£317211'31'33113'13.3-333'3".333"3311113311131” ‘ ~32}: 3333311133311
‘I"}1‘.} 1 11}}111111I1111 H Sidp‘ 1:1:131} 1111111‘111.}1}"111 11' ' 1ah11q“ '1 ‘3. x 333.12.1-'i},-.2:-,£! IS 111' ‘1’
3meE333H. 1E»133h “1~‘T‘ .3313EEH3L5333331-13’ 7.3. 33?33333
7}21.;3'5:}1§|1'I};}f,13f345..1:}I 1}.1}.3}‘1,‘17;3}1 ’l :1 } }3r33.:. 3.31151, 1'}... "‘33"! ’ '111'11'13131?}11E-1‘153'I3h‘1-5
i1}1}}11}1 33‘}: '; 1’1F1IL'111‘1I11I11311'1'1'1‘311111(1131'!11113 11.511‘}1}1.1111:1.:.}111'§; ' "1';'}:}.11»'.1: , I , E111}.:,1}11-1}E1,:}1111§1I1II1 ,
W111} }“ }Z'3~|},,‘}}3 11111l1 1311' 411113-11 ‘3’111'1 113'}lm'11 1111.1 El“ 1111:1‘1€}:‘.}I}; II‘Ix}'11:}L:11511-.
”.111 ”3,: 1.1,, , “.. . ‘1I3'-."3'*'111 ”‘33?! EIIH} '33‘133131’ .1. . 1.1;‘1:.-." '3': 1:1.
113131311113“: "1}“1111 }111 311 f11f1111":l1}11 3-33 3.31.1. :1 §'.' 333.3" 111‘ 1,1,- 33-...1 “.373? 3
'13- 33"'."'3~'.13113"3”.31 1 '3 1 ' "" .3’3:;'~31",3‘3'..333:
11:1}:}}111r1111}1-}',I1: :11}:}111I} :I1'}1.:}11 }:11111|111111J11}I~1}}1‘53}. }:.I11‘II} 1:11'1}'I1‘J17}I4. .' ‘1’}13111g}}::}1?}3:.
1333}“*133551'1“ "" " 1"”13'33 3.-~-313. “3333.33333.3 !
' " 111131" ' 33 ".' .-. 3» '33 3335‘ 1'11“ “33.3243 1.132 .,§. , E11.
1311 " 13,111 .3 3.1 '1 3 '1.'1 11'1" .71EL}.}1 133-1. 3; 3333.331 r}.v}331 1.41.111tx1‘1111. 1'. 3‘}
.1111” 11.1 111" "1 '11'31"111 113731"'.-.3:1.111'I'351111’13:31313133354311'13.“1.:1‘i1'3f11‘1'L1 1}1:‘11""~‘---...3 .
3 1“ 1. i 33,3}'.33333Ig,333:333I33xm}3}7“}333}§}333}33 1111": 33 "'1”3
E11111 1.11”} 13131111 1.,13, ' :1'1} -1 11 1'1-}':l.11 ' - '11’1'3 113:1“: 111531 }} 171.1.13131‘3 £1}, 1":1‘ 11 11 ‘:1,111:;E:11,1}1‘1'H:1.i:1112111(11:
3.131‘..33.~31333 3 333.'33333333}93333~.:3'.53}.}3.3}33~ 3 333£33....3.‘3333333333333
;}:}}-;3 1.111.}111‘11‘11111111 'gI‘}:3.,3 3}}3. ...'31' .1333}1:.1}11113.3 3111:1133":- '31j11 "3113}34-3.}1351‘:131353311311}}1331-
:1I}.1}1}I'}1111}1'111 111111111111EZ1}" 1}:1'3.1 ”‘3 11.3,}11‘ 1 1}:}:1}}1}!11:1'111111111111'I11 :41“;- 11:111': ’1‘: 1 :}.i}111;1;[x. ‘11-’r117‘9‘ZIFS1-J113 l
}1Q}1}'1-1}}11H111'1 },}i/ 11}I§1}1'L'3I131'1 1:11};1.111111113111111111111 '11V1'31511}.' {f}: I 1'1111”"111311'1191““high”;3 J; "1}
3 1111""1113} 3}}1}1,'1‘1’}|3‘ 1 11'1.}311};}1 11'1}}1:111111'11‘1 ..~ (33.5, .1331} ”1va .
“1112} 13,, 111.1} 1.11} 3. 114} l111111}11}1.1111111l111111311}15111111 '1’11.,1}}}11} 11}111}1}111 E1111}1} 5111} 11111I}I}11111'1 111111111.1} .( $1111? 11,11111}}1'13111i;33,1
3333.33“H1‘33333 “I: h “'3313}w 3. ‘“ . . . H1 1 .“‘3-333w3 3.33333
31%;”: 5}.1}}}I1:11111.311 I111 7'31'11"1i1}3} }}H1I}I111}1 11}11111"1‘}111}' H9 ) I}:}}!.u} }}}1}m}'}}} 111};""-}}}1}111311}1111}1111'11111111 '1‘}}.1111}11,1:}:}‘:1'1}L~},}}1"
.. 3" . z. '3' 3 3.:43}. . 3:!.,'1'.
11111111111111”1111111111111111 3:314} 3:3};3 .11} .1}? 31113111131111 1'} ....«3.3333. I}333}}333.§3,313§ 1 L },,}’ 1.1111111 “fi‘ditfiifii‘q
1:111. I}}"i{i1111.l11 ”1 I ‘01. o}:}u‘1'}',l1rn 1}. 3:13
33-33 3. "3’33 3333333333.3.3.3133» .33 3.33. 33 3331"} 333' .3 .3 .3333 33.. 3W3. 333333
}}1 1 1‘ '11111'11,11}1.‘}1'11:l-' “1111.1”? q}11111.i13}1"'a:u3.1‘1ll:11 .
1111111 :3 1}}II111}1111111'}111 1}” 11111 };11}1111}11’311}1}1}} 11M} WW} 3}'!1131}“1 }l1l}'l}11? 3,1111}11 111}1F1}3§}33}}}}.3 1}} 11111311,}? '11l111}}1 11,
I'Ul' 3%} ‘1} ya 1. v- “}:’
1}}}1111}1}}111}}13 '33‘111}‘}}}11]}}1}‘}}3}?13’}3}1,}13133""3'3‘33}33111}}}Ufi.3}131E}. }111}11}3’1}} L }},}1’1l!171'1}}}13} ,1} 1’ lI‘éi}}}I
11" }3 . 3'111‘1'1’113'133"‘}} }" } .3733.“ 3}}.1. «3‘. .3.3 3 _} }}}3:
1}}1111 1" 113331 3311133 33133 33"“ 33333 33333333. 33: 3.3- 333.333.3331 33
1W}. 31.33. "1}1135132} :1 3', " 3° 1'. 3'1.“ 11111111111
111}1}11111 111111 1' 1111111111111fl13111111111111 '11 .' } ‘3 3}} *‘}11‘13’§1.}113,3}‘:3‘.3.;}}
1111111 313 333 33" .3 333.. H! 333 3333'. 133313 333 1‘33 ~ 3 33 3‘ » 3 33333 3
33 nifllzi }} ...1.3}LR3EE}3.33 HLuiulil fiLL'341E' 3:3.' 1h3h

IHIilHll'lllllllllillll‘lllllll|lllll1|lllllllllllllllllllii

31293 00904 9390

This is to certify that the

thesis entitled

QUALITY IMPROVEMENT AND OPTIMIZATION OF A
RESINA CAPPER ON A PACKAGING LINE:
AN APPLICATION OF STATISTICS UTILIZING
THE TAGUCHI LOSS FUNCTION

presented by

CHRISTOPHER ANDREW IAN CLARKE

has been accepted towards fulfillment
of the requirements for

 

MM Mm

Major professor

 

Date AUGUST 6, 1993

0-7639 MS U i: an Affirmative Action/Equal Opportunity Institution

 

 

 

LIBRARY
Michigan State
1 University

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

DATB'im DATE DUE DATE DUE

MAR 1 6 19fl9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Li

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

if??? i

MSU Is An Affirmative Action/Equal Opportunity Institution
cWWmS-pA

 

 

QUALITY IMPROVEMENT AND OPTIMIZATION OF A
RESINA CAPPER ON A PACKAGING LINE:
AN APPLICATION OF STATISTICS UTILIZING
THE TAGUCHI LOSS FUNCTION
By

Christopher Andrew Ian Clarke

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE

School of Packaging

1993

ABSTRACT

QUALITY MROVEMENT AND OPTIMIZATION OF A
RESINA CAPPER ON A PACKAGING LINE:
AN APPLICATION OF STATISTICS UTILIZING
THE TAGUCHI LOSS FUNCTION

By
Christopher Andrew Ian Clarke

The thesis reports an application of the Taguchi Loss Function to a Resina Capper
on a packaging line in a major United States pharmaceutical company. It calculates the
cost saving realized by reducing variability, improving the quality of removal torque, and
ultimately redrlcing process costs.

The design and structure of the experiments followed these principles:

1. Statement of the Needs - Written statement of intended knowledge to be
gained.

2. Conceptualizing the Design - Brainstorming, review of available data and
literature.

3. Evaluation of the Measurement Process - Gain an understanding of the process,
quality requirements, and operating characteristies.

4. Consider Potential Experimental Problems - Assess process inconsistencies.

5. Design the Experiment - Based on statistical and non-statistical methods.

6. Run the Experiment - Based on the design.

7. Analyze the Results - Utilizing statistical methods.

8. Taguchi Loss Function - Apply the Loss Function to the results to determine
possible cost savings, improved quality, and increased productivity.

Improved quality, increased output and Significant cost savings were realized as a
direct result of the experiments that were conducted on the packaging line. By
implementing the changes recommended, on all of the packaging lines utilizing Resina

Cappers, overall improvement of quality and productivity can be achieved, resulting in
‘ significant cost savings.

Copyright by
CHRISTOPHER ANDREW IAN CLARKE
1993

In memory of my father,
IAN EDWARD CLARKE,

who has provided me with continual motivation to succeed.

ACKNOWLEDGMENTS

I would like to express my eternal gratitude to Dr. Hugh Lockhart, my major
professor, for his keen interest, valuable advice, and guidance. I would also like to thank
my committee members Dr. Bruce Harte, Dr. Julian Lee, and especially Dr. Dennis
Gilliland for his invaluable statistical guidance. Without their help this paper could not

have been written.

To my mother, Shirley Clarke, I gratefully thank you for continually encouraging
me to reach for higher goals and providing me with the support, both financially and
spiritually, to be able to attain those goals.

To my wife, Jane, thank you for providing me with your undying support,
especially while being away from you during the research phase of this thesis.

To all those associated with this research project especially, Bill Gierke who made
the project possible, Jeff Fluharty and James Edelman for their help and enthusiasm. A
special thank you to all Operators who assisted in the running of the experiments. Without

your help the data could never have been collected.

TABLE OF CONTENTS

 

 

 

 

 

 

 

 

 

List of Tables -- - - - ...........................
List of Figures- - - - - -- -
List of Symbols -- - ...... - ...... - - - - x
1. Introduction _- -- - ..... 1
1.1 The American Pharmaceutical Industry ........................... 2
2. Literature Review - -- -- - - -- - - -- - ....... 3
2.1 Calculating the Taguchi Loss Function .......................................... 3
2.1.1 Mean Squared Deviation (MSD) .................................. 3
2.1.2 Loss Constant (K) -- - _ - ........ _ - 4
2.1.3 The Taguchi Loss Function - - - - 6
2.2 Calculating Savings Associated with Two Processes .................... 6
2.2.1 Signal-tO-Noise Ratio (SIN -Ratio) ............................... 6
2.2.2 Savings ........................................................................ 7
2.3 Calculating the Process Capability ............................................... 8
3. Company Background ............................................................................... 9
3.1 Company Introduction ................................................................. 9
3.2 Statement of the Needs ................................................................ 10
3.2.1 What are we trying to learn? .......................................... 10
3.2.2 What are the aims of the research? ................................ 10
4. Evaluation of the Measurement Process ...................................................... 12
4.1 Qualification of Measurement Device ............................................ 12
4.2 Results .................................... - ....... - 14
4.3 Conclusions ................................................................................. 18
5. Reducing Machine Set-up Variability .......................................................... 22
5.1 Designing a set-up gauge .............................................................. 24

Page

 

 

 

 

 

 

 

6. Critical Parameter Design ........................................................... 27
6.1 Evaluating The Closure System ..................................................... 27
6.2 Determining Removal Torques: Target, Minimum, and Maximum 28
6.3 Predicting the Target Value (Application and Removal) ................. 30
6.4 Predicting the Minimum and Maximum Immediate Removal
Torque Values -- - -- . 36
7. Taguchi Process Evaluation ........................................................................ 38
7.1 Assumptions ................................................................................ 38
7.2 TheCurrentProcess -- - .......... -- 39
7. 3 Incorporating a Better Measurement Device ................................. 40
7. 4 Incorporating the Set-up Gauge into the Process .......................... 42
8. Estimating the Loss and Savings Generated by the Improved Processes ...... 46
8. 1 Calculating the Loss Constant 46
8. 2 Comparing the Current Process tO the Improved Process ............. 48
8.3 Comparing the Improved Process to the Optimum Process .......... 49
8.4 Comparing the Current Process to the Optimum Process ............. 50
9. Recommendations ..................................................................................... 52
9.1 Measurement Device ................................................................... 52
9.2 Resina Set-up Gauge ............ - - - - _ 52
9.3 Critical Controls .......................................................................... 52
9.4 Implement an Aggressive Statistical Process Control Program
(SPC) .......................................................................................... 53
9.5 Purchase a New Capper? ...... - ...... . 54
9.6 Future Research ..... -- - - -- -- -- -- - 54
10. Summary of Materials Used ..................................................................... 56
10. 1 Materials Tested ........................................................................ 56
10.2 Equipment Used ........................................................................ 56
10.3 Testing Conditions ..................................................................... 56
10.4 Analytical Tools ......................................................................... 56
1 1. Appendix .................................................................................................. 57
11.1 Appendix A - Raw Data Tables ................................................... 57
Bibliography ..................................................................................................... 74

LIST OF TABLES

 

 

 

 

 

 

 

1. Operator Variability Secure Pak (Hand Method) ......................................
2. Operator Variability Sure Torque -- ..........................
3. Torque Tester Variability: Secure Pak vs. Sure Torque .................

4. Removal Torque Study vs. Time Showing Removal Torque Decay ...........
5. Predicted Removal Torques

6. Summary of Recommended Process Parameters -- _- _ _- _ - -- _- -

7. Current Process Data Summary ........ - _- _ ......

8. Improved Process Data Summary _- _- -- -- .......

9. Optimum Process Data Summary .............................................................
10. Summary of Process Controls .............. - .............................

13
13
14

30
37
39
41
42
53

LISI‘ OF FIGURES

1. Loss Functions: Specification vs. Taguchi ...........................
2. Operator Variability Secure Pak ...............................................................
3. Operator Variability Sure Torque ..............................................................

 

4. Torque Tester Variability -

 

 

 

 

 

 

 

 

5. Paired Roller Set-up - - - - -
6. Resina Set-up Gauge ........... - ........................................
7. 12 TIP Application: Predicted Curve - -

8. 15 TIP Application: Predicted Curve _ .....
9. 18 TIP Application: Predicted Curve ....... - - -- - - ..................
10. Actual Process Density Functions ........ -- - ..........

11. Adjusted Process Density Functions -- - - _ ............................
12. The Taguchi Loss Function for Removal Torque .......................................

MSD
0(Classical)
Mast-chi)

USL

LIST OF SYMBOLS

= Mean Squared Deviation

= Classical Standard Deviation
.-. Taguchi Standard Deviation
= Deviation of Process Average from Target Value
= Taguchi Loss Constant

= Taguchi Loss Function

= Signal-tO-Noise Ratio

= Process Capability Index

= Process Average

= Target Value

= Lower Specification Limit

= Upper Specification Limit

1. INTRODUCTION

In today's market, organizations compete in three main areas: quality, delivery, and
price. An organization's reputation is directly tied to the quality of its products. Quality is
often used to signify "excellence" and in some companies quality indicates that a product
conforms to certain guidelines established by the company. These guidelines often
overlook the customer who ultimately defines the meaning of quality. A company's
definition of quality should be to meet the requirements of the customer. Traditionally
American companies have not given quality the attention it requires. The quality control
revolution in Japan has now forced American companies to adopt different manufacturing
methods to enable them to compete in the same markets. Statistical Process Control
(SPC) is one method that has been applied to American manufacturing processes. SPC
addresses the area of on-line quality control and refers to the monitoring of current
manufacturing processes to verify the quality levels produced. SPC is a method to help
control processes [1]. Off-line quality control refers to the improvement of quality in the
product and process development stages [10]. This step is not usually incorporated into
traditional SPC methods and often excludes the customer's requirements.

Genichi Taguchi, a Japanese statistician, has developed a methodology that
incorporates off-line and on-line quality control into the design and manufacturing
processes. Taguchi methods have a positive impact on cost that is obtained by improving
quality in the developmental phases of a product cycle. Taguchi emphasizes the
importance of moving quality improvements upstream from the manufacturing process.
The goal is to design quality into every product and the processes that build them. A key
component of Taguchi's philosophy is reduction in variability, this incorporates three

central ideas [13]:

1. Products and processes should be designed so that they are robust to external
sources of variability.

2. Experimental design methods are an engineering tool to help accomplish this
objective.

3. Operation on target is more important than conformance to specifications.

The ultimate goal of the Taguchi philosophy is to have a process that operates with
minimal variation around a desired target value [12]. The Taguchi loss function generates
minimal loss when a process has the least variation and is producing on target.

1.1 The American Pharmaceutical Industry

The American pharmaceutical industry is a world leader in the manufacturing of
quality pharmaceutical products. Applicatiom of the Taguchi philosophy fit perfectly into
the industry because of the stringent manufacturing practices that are demanded by the
US. Food and Drug Administration (FDA). These practices are known as current Good
Manufacturing Practices (cGMP'S). By incorporating Taguchi's philosophy and Statistical
Process Control into the pharmaceutical industry, significant improvements in GMP'S can
be achieved. This not only satisfies the FDA'S GMP requirements but also reduces
variation in processes and lowers manufacturing costs. Currently the American
pharmaceutical industry leads in quality. To maintain this position, the industry must
adopt new manufacturing design and production methods to ensure that its quality is
continuously improved. This is the fundamental basis of the Taguchi philosophy: reducing
variation (improved quality) reduces the "loss to society ". The loss refers to the cost that
is incurred by society when the consumer uses a product whose characteristics differ from
the nominal [14]. Reducing this loss, in turn leads to the ultimate benefit of producing

quality products for less money.

2. LITERATURE REVIEW

The quality revolution in today's "Global Economy" has emphasized quality as the
major factor in deciding market share. Markets are no longer dominated by one particular
product. Companies around the world can compete in almost any market in terms of cost
and volumes of finished product. Supply often outweighs demand in most markets today.
As a result, product cost and cost differentiation between competing products are
minimized. One deciding factor that differentiates one product from another is that of
quality. Dr. Genichi Taguchi, defines quality as "the loss to society caused by a product
after it is shipped" [6,7]. A product with good quality will cause little loss to society and
one with poor quality will cause greater loss to society. The Taguchi loss function
recognizes the customer's desire to have products that are more consistent, and the
producer's desire to make a low-cost product. The loss to society is composed of the
costs incurred during the production process as well as the costs incurred. during the use
of a particular product by the customer i.e. repair, dissatisfaction, and lost business.
Taguchi claims that minimal loss is incurred when a process is producing close to or at the
set target. The further away a process is fi'om the target, the higher the resulting loss,
even when the process is within specification limits.

Traditional specification losses allow for a range of values that a process can
Operate within, generating zero loss within the specified range. In the traditional concept,
the only time a loss is incurred, is when the specification limits are exceeded. The loss
under the Taguchi philosophy is in part based on the Mean Squared Deviation (MSD) of a
given process from its target. Any deviation from the process target and/or increase in
variation within a process results in a proportional loss. This loss can be measured, even
for a process operating within specification limits. Any process can be improved by
, reducing the variation within the process, and saving generated by improving the process

are determined by comparing the Signal-tO-Noise ratios (SIN Ratio) of the process before

and after the change. This concept is presented in Figure l - Loss Functions: Specification
vs. Taguchi.

2.1 Calculating the Taguchi Loss Function

2.1.1 Mean Squared Deviation (MSD)

The Loss Function is based on the MSD of a given process. The MSD is a
measure of the amount of variation that a process contains and how far that process
average deviates from the process target. A larger amount of variation within a given
process, will generate a larger loss. The further a process average deviates from the target
value, the larger the loss incurred. The MSD is calculated as follows [6,7,10]:

MSD - (9('1"1181lchi))2 + (MY (1)
Where:

MTHgUChI) - «CINIcal/J—
1%: 1

M- --Target Average

2.1.2 Loss Constant (K)

The amount of loss determined by the Loss Function also depends on a constant
value that represents the cost per unit, during production, in relation to the specified
tolerance of a quality characteristic. To get an exact value for K, costs for the following
factors are needed: Materials costs, Product costs, Research and Development costs, and

Labor costs. The exact dollar value is difficult to obtain since most cost areas are usually

LLlSS LUNSLLLLNS: SPLCLLLCALLLLN VS. LAGUCHL
SPLCLFLCALIUN LDSS FUNCLLDN

$ - ------- L083 L088 -----

 

LSL TARGET USL
QUALITY CHARACTERISLIC

TAGUCHL LLlSS LUNCLLLLN

\ /
\ /

\ /

$ -------------- LDSS -------\ /m---~- LESS m-

\ /
\l/

LSL wits usi
QUALITY CHARACLLRLSLIC I

 

 

LLGURL L

spread out over the life of the product. This would require knowledge of previous and
future expected production volumes. The Loss Constant is calculated as follows
[6,7,10,11]:

K-_m (2)

(Tolerance)2
Where:
Tolerance = (USL - LSL)/2
2.1.3 The Taguchi Ines Function
The average loss per unit associated with a given distribution is calculated by

multiplying the Loss Constant (Equation 2) by the MSD (Equation 1) of a given process.
This Loss is calculated as follows [6,7,10]:

Loss = K(MSD) (3)
2.2 Calculating Saving Associated with Two Processes
2.2.1 Signal-to-Noise Ratio (SIN -Ratio)
The Taguchi method focuses on the reduction of noise factors that introduce

greater variation into the performance of a given product. There are three sources of

product performance variation: External factors, Internal factors, and Unit-to-Unit factors.

External factors encompass the environment in which the product is used or distributed.
Internal factors encompass product deterioration with age or use. Unit-tO-Unit factors
encompass variations in the manufacturing process [14]. To minimize the amount of
variation associated with a given product, all three of these noise factors must be
considered during the design phase Of a product. Product and production processes
should be designed so that they are less sensitive to noise factors. A useful calculation for
this is the signal to noise ratio (SIN).

The S/N ratio is the reciprocal of the variance of the measurement error. This
value is maximized when process parameters have a minimum error in their variance [6].
The S/N ratio consolidates several repetitions (at least two data points) into one value
which reflects the amount of variation present. The optimum process design will have a
S/N ratio that is smaller than the S/N ratio of a process that one is trying to improve. The
SIN-Ratio is calculated as follows [10,11]:

S/N Ratio = -1OIOg(MSD)* (4)
"' This equation is used in the calculation for S/N ratio based on the variance only.
2.2.2 Saving
The saving associated with an improved process can be calculated by comparing

the Signal-tO-Noise Ratios of each process. The saving are calculated as a percent cost

improvement of one process over another, according to the following equation [1 1]:

Saving = (1 —10HS’"='S’"0"°’)*100%
_ (MSD, - MSD
MSD,

(5)

 

2)‘100%

2.3 Calculating the Process Capability Index

The process capability index (Cpk) is based on the centering of a frequency
distribution and the ratio of the spread of the distribution within the width of the
Specification [12]. The frequency distribution and the Cpk will reflect control within a
process and the actual performance of a process after process modification has occurred.
The Cpk compares the centrality of the process mean within the tolerance spread, and how
capable the given process is of operating within the specified limits. A Cpk value of less
than one indicates that a process is not capable of producing within specified limits. A
Cpk value one indicates that a process is just capable of producing within the specified
limits. A Cpk value of greater than one indicates that a process is capable of producing
within specified limits. This latter case provides the most desirable situation. The Cpk is
calculated as follows [12]:

C k lNearer Specification Limit - X]
p ’ 30(Classical)

 

(6)

Where:

3. COMPANY BACKGROUND
3.1 Company Introduction

The research phase of the thesis was conducted at a major United States
pharmaceutical company located in the Midwest. The company produces prescription,
non-prescription, and animal healthcare products. The main emphasis is placed on the
prescription side of the pharmaceutical business. Products are packaged in a variety of
different packages: blisters, vials, glass bottles, and plastic bottles. The bottles are sealed
with a number of different closure styles made from metal and plastic components. The
Child Resistant Closure (CRC) is the most commonly used closure on both plastic and
glass bottles. The cap is made up of three components, an inner shell, an outer shell, and
the liner system. The inner component is made from a polypropylene resin and the outer
shell is made from a high density polyethylene resin The liner system is a laminate
composed of four separate layers: coating, facing, backing, and the glassine liner. The
caps are supplied to the company by an outside supplier.

The CRC is applied to the bottle utilizing a Resina Capper. The Resina is an in-
line capper which has consecutive pairs of rollers that spin the cap onto the bottle and
apply torque to the cap. The Resina Capper that this thesis. evaluates, has four pairs of
consecutive rollers that apply the torque successively. The first set of rollers Spins three
times as fast as thenext three pairs of rollers. The function of these rollers is to spin the
cap until the liner is seated on the land of the bottle finish. The first pair of rollers do not
actually apply any torque to the closure system. The next three pairs of rollers apply the
torque in succession, to the desired level. The separation between the two rollers in each
of the four pairs of rollers and the setting of the clutches on the individual spindles is
crucial. These two variables have the most pronounced effect on application torque.

There is a large number of variables associated with the torque application process

and each variable was studied to evaluate its effect on the removal torque of the CRC.

10

Areas of study include: the performance characteristics of the Resina capper, set-up of the
capper, and the variation caused by materials (caps and bottles).

The Food and Drug Administration (FDA) has instituted very strict regulations
governing closure removal torques. Effectively, if one CRC is found to be loose in the
field, an entire lot will have to be automatically recalled. A loose cap may lead to leaking
and deterioration of the contents of the package. A loose cap may also provide evidence
that a closure system has been tampered with. A recall can prove to be very costly and
careful consideration for the application process hasto be taken into account. The
variability of the process has to be controlled to reduce the probability that a closure
system might fail in the field.

3.2 Statement of the Needs

3.2.1 What are we trying to learn?

Historically the capping operation has not performed consistently to ensure that
there are no loose caps out in the field. This has resulted in a recall and increased process
costs. All the variables associated with the capper were studied to "learn" how each
variable affects the output (removal torque) of the process. Once the effects were
evaluated, those having the most significant effect were targeted to improve quality
(consistent removal torques) and reduce process costs.

The effects of supplier inconsistencies were also investigated to find out whether
the closure system contributes to machine down-time and decreased quality. Supplier

quality has been suspected as a probable cause for creating defects during production runs.

3.2.2 What are the aims ofthe research?
1. To improve the consistency of quality (removal torque) of the Resina Capping process.

2. To lower costs by reducing the probability of a recall occurring, by increasing the

11

output of the machine, and by reducing down-time associated with defective supplier
materials.
. To evaluate the performance characteristics of the Resina Capper:

a) Mechanical Performance - Evaluate the variables associated with the setup and
operation of the Resina Capper.

b) Operator Performance - Evaluate the effects of operator variability.
. To evaluate the supplier's materials for inconsistencies and defective components.
. To evaluate the variability of torque measuring methods and devices to ensure that
consistent and valid data are collected.
To predict possible cost saving by applying the Taguchi Loss function to the
collected data.

12

4. EVALUATION OF THE MEASUREIVIENT DEVICE

4.1 Qualification of the Measurement Device

The measurement device that was used to collect the data was a Sure Torque
Torque Tester Model number NEBT‘33 manufactured by the New England Machinery
company. The method of measuring removal torque is automated and requires little
human interaction while measuring removal torque. The Sure Torque has a constant
velocity rotating head. The torque (force) measuring system is calibrated against a known
mass and provides an accurate and repeatable method for measuring removal torque.
Operator variability is significantly reduced by the torque tester. The Sure Torque
measurement device was compared to the current removal torque tester being utilized.
This device is known as the SecurePak torque tester (Hand method) and when compared
to the Sure Torque it was found to be less precise.

An experiment was designed to evaluate the two torque measuring devices against
each other. A comparison of Operator variability was conducted for each device and a
comparison of torque tester variability between the two devices was also conducted. The
Operator variability test _on each device was conducted by two operators on each of the
torque testers. Two hundred caps were applied to two hundred glass bottles at 22 Torque
Inch Pounds (TIP) by the Sure Torque tester. Each Operator then measured the removal
torques of fifty bottles on each of the measuring devices.

The evaluation of the two measuring devices was made up of the combined sample
of the two operators on the same measuring device compared to the combined sample of
the two operators on the other measuring device. The raw data is listed in Table 1, Table

2, and Table 3 in Appendix A. A summary of the data is listed below.

Table 1 - Operator Variability Secure Pak (Hand Method)

13

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VARIABLE OPERATOR 1 OPERATOR 2
SAMPLE SIZE 50 50
AVERAGE 16.2 13.3
VARIANCE 2.44 1.32

STD. DEV. 1.56 1.15
MINIMUM 12.1 10.3
MAXIMUM 19.7 15.5

RANGE 7.6 5.2

Table 2 - Operator Variability Sure Torque

VARIABLE OPERATOR 1 OPERATOR 2
SAMPLE SIZE 48 50

AVERAGE 12.3 11.4
VARIANCE 2.50 1.18

STD. DEV. 1.58 1.09
MINIMUM 8.2 9.5
MAXIMUM 15.6 14.5

RANGE 7.4 5.0

 

 

 

 

 

14

Table 3 - Torque Teswr Variability: Secure Pak vs. Sure Torque

 

 

 

 

 

 

 

 

 

 

 

 

 

 

VARIABLE SECURE PAK SURE TORQUE
SAMPLE SIZE 100 98
AVERAGE 14.8 11.9
VARIANCE 3.97 1.98
STD. DEV. 1.99 1.41
MINIMUM 10.3 8.2
MAXIMUM 19.7 15.6
RANGE 9.4 7.4
KURTOSIS -0.460 -0.096
4.2 Results

The results of the operator variability test on the SecurePak torque tester show
that the average for Operator 1 is 16.2 TIP and the average for Operator 2 is 13.3 TIP. A
t-test was conducted to see whether there is a significant statistical difference in means.
The t-test showed that the difference between means is highly significant at an Alpha level
of 0.001. A further test was conducted to see whether the ratio of the variances of the
two operators was equal to one. A ninety five percent confidence interval for the ratio of
the variances does not include one, therefore the variances of the two operators are not
equal. The multiple Box-and Whisker Plot in Figure 2 shows graphically the difference in
medians represented by a visible separation Of the center lines of each box. The presence
of no overlap between each of the boxes reveals that there is a difference in variances

between each of the operators.

REM TOROIE

18

18

14

12

10

OPERATE]? 9mm SECIRE Pfl<

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FIGIREZ

own-no.0": m.m

(PERRTIRNU'BER

 

 

16

The results of the operator variability test on the Sure Torque torque tester Show
that the average for Operator 1 is 12.3 TIP and the average for Operator 2 is 11.4 TIP. A
t-test was conducted to see whether there is a statistical difference in means. The Meat
showed that the difference between means was not significant at an Alpha level of 0.001.
A further test was conducted to see whether the ratio of the variances of the two
operators was equal to one. A ninety five percent confidence interval for the ratio of the
variances does not include one, therefore the variances of the two operators are not equal.
The difference in variances can be attributed to the fact that the CRC cap must be
engaged, or in the locked position before it is placed into the Sure Torque tester.
Individual Operators may accomplish this by utilizing different methods of engagement and
this may result in the differences in variances. The multiple Box-and Whisker Plot in
Figure 3 shows graphically that the difference in medians represented by the separation of
the center lines of each box is not pronounced and not statistically significant The
presence of overlap between each of the boxes reveals a closer association between the
two operator distributions. This is not reflected in the standard deviation for each
operator on each measurement device. The standard deviation for each operator on the
SecurePak are highly conservative estimates Of a true production setting. The tests were
conducted under controlled conditions and the results present a "best case“ scenario. The
Sure Torque test data reflect actual results that can be expected in a production setting.
The latter part of the thesis provides further evidence by evaluating actual production
processes utilizing the SecurePak and Sure Torque torque testers, and show the above to
be true.

The results of the torque tester variability test show that the average of one
hundred samples for the SecurePak torque tester was 14.8 TIP and the average for the
ninety eight samples on the Sure Torque was 11.9 TIP. A t-test was conducted to see
whether there is a statistical difference in means. The t-test showed that the difference

between means was highly significant at an Alpha level of 0.001. A further test was

REMOVAL TOROUE

OPERHTOR VRRIRBILITY SURE TOROUE

 

 

16.2

 

 

 

14. a ...........

 

 

 

12.2

 

 

 

 

 

 

 

 

 

10.2

 

 

8.2

 

 

 

 

 

 

OPVRSURE.ONE OPVRSURE.THO

OPERRTOR NUHBER

FIGURE 3

18

conducted to see whether the ratio of the variances of the two torque testers was equal to
one. A ninety five percent confidence interval for the ratio of the variances does not
include one, therefore the variances Of the two torque testers are not equal. The multiple
Box-and Whisker Plot in Figure 4 shows graphically the difference in medians represented
by a visible separation of the center lines of each box. The presence of no overlap
between each of the boxes reveals that there is a difference in variances between each Of
the torque testers. The box representing the SecurePak torque tester is much larger
(larger variance) than the box for the Sure Torque tester (less variance). The test for
normality or the Kurtosis for each of the measurement devices was also calculated and
was -0.460 for the SecurePak and was -0.096 for the Sure Torque. A Kurtosis = 0 is
representative of a "normal" distribution. The Kurtosis for the Sure Torque is closer to
zero compared to the Kurtosis for the Secure Pak. This reflects that the Sure Torque
approximates a normal distribution better than the Secure Pak. The Kurtosis value for
each measurement device was negative, which shows that these distributions are flatter

and have shorter tails than does a normal distribution.
4.3 Conclusions

1. Operator variability - the tests for Operator variability show that the SecurePak
removal torque tester has greater operator variability than the Sure Torque removal torque
tester. There is more operator interaction associated with measuring removal torque by
the hand method. This can be attributed to individual operators applying different vertical
and lateral forces to the closure system while rotating the cap during the measuring
process. The Sure Torque provides a fully automated method of measuring removal
torque and significantly reduces the effects of operator variability. The presence of a

constant velocity head and a constant downward pressure applied to the closure system on

REMOVAL TORQUE

20

18

18

14

12

10

TORQUE TESTER VRRIRBILITY

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FIGURE 4

OPVRHRND.COHBINED
OPVfiSURE.COHBINED

TORQUE TESTER

 

 

20

the Sure Torque, virtually eliminates the effect of vertical and lateral forces introduced by
individual operators.

2. Torque tester variability - the tests for torque tester variability Show that the
Sure Torque provides a better method for measuring removal torques than does the
SecurePak. The precision of the two torque testers is reflected in the standard deviations
of each of the processes. The Sure Torque (standard deviation 1.4 TIP) has a lower
standard deviation compared to the SecurePak (standard deviation 2.0 TIP) and is
therefore more precise. The mean removal torque on the Sure Torque (1 1.9 TIP) is 2.9
TIP lower than the mean removal torque for the SecurePak (14.8 TIP). Further research
should be conducted to reveal which measurement device produces a reading that
approximates the "true" removal torque.

3. An evaluation of the measurement device methods for measuring removal
torque shows the Sure Torque to be more consistent. The Sure Torque uses a fully
automated method for measuring removal torque. A bottle is placed against a rubber
guide block, by the operator, and then the Sure Torque is activated. A second rubber
block then closes against the bottle, securing it in place and the base plate raises the bottle
up to locate the cap in the cap chuck. The chuck clamps onto the cap and begins to
rotate. A constant velocity rotating head removes the cap at a constant rate while
measuring removal torque. The rotating head continues to measure removal torque until it
senses a dramatic drop in removal torque force. Once this point is reached, the
measurement process is complete.

The Secure Pak method is not automated and requires more Operator interaction.
A bottle is clamped between four rubber covered posts and the removal torque is then
measured by an operator. The Operator " grips" the cap and removes it by rotating
counterclockwise. Each operator will perform this task differently and introduce various
degrees of vertical and lateral forces. This results in greater variation in the measurements

of removal torque. This is further reinforced by the test for "normality" or the Kurtosis

21

for the Sure Torque (-0.096) being closer to zero than the Kurtosis for the SecurePak (-

0.460). A Kurtosis value of zero reflects that a given process is normally distributed.

The Multiple Box-and-Whisker plots for Operator Variability Secure Pak (Figure
2) and Torque Tester Variability (Figure 4) are Similar since they both Show that there is
no close relationship between the two population groups (no overlapping of the "Boxes").
The median value or the centerline of the individual "Boxes", in both cases, are separated
from each other. This shows that there is a difference in medians which is reinforced by
the t-test discussed earlier. The Multiple Box-and-Whisker plot for Operator Variability
Sure Torque (Figure 3) shows a closer relationship between the two population groups
(the "Boxes" overlap). The median value or the centerline of the individual "Boxes" are
more closely related to each other. This shows that there is not a significant difference in

medians which is reinforced by the Heat discussed earlier.

22

5. REDUCING MACHINE SET-UP VARIABILITY

The Resina capping machine is an effective tool for applying a cap to a bottle as
long as it is set-up in the correct manner. There is a large number of different machine
variables that influence application torque. An evaluation of which machine variable has
the most influence on application torque was conducted. A two-tO-the-three factorial
experiment was conducted to evaluate three factors (variables), each at two levels. The
three variables tested were: roller speed, belt speed, and roller separation. The roller
speed refers to the speed of four sets of rubber rollers that apply torque to the CRC cap.
The belt speed refers to the speed of the two belts that hold the bottle in place as it travels
through the Resina capper. The tests revealed that roller speed and separation between
rollers in a pair are the two most critical parameters that need to be targeted. Belt speed

had negligible effect on the process.

Roller speed is relatively easy to control and is adjusted by simply turning a crank
that alters the position of the motor in relation to the drive belt. The optimum roller speed
is obtained at slow speed setting (3.3 RPM). The Resina should be set-up to run at this
speed during production runs. Roller separation has the most influence on application
torque and is more difficult to control than roller speed. The Resina has no dials or
marked gauges that allow for a consistent roller pair set-up, so a set-up gauge was

designed to enable a better setting of the distance between rollers in a pair.

Another variable that was found to be critical to the cap application process was
the alignment of the roller pairs in the direction of travel (Figure 5). The way the Resina
capper is designed makes it very difficult to set the roller separation correctly and
consistently across all production runs. A high speed video of the process revealed that

bottles would become laterally stressed if the roller pairs were not aligned. The objective *

FflHRED RDLLER SET—UR

TIIP VIEV LII RIILLER PAIRS TLIP VIEV [IE RDLLER PAIRS

V

 

V
L

O
0 0 Q Q

 

 

V

CURRECT DIRECTILIN [IT INCURRECT DIRECTILIN LII
TRAVEL FIIR BLITTLES TRAVEL FLIR BLITTLES

FIGURE 5

24

is to have both rollers of a pair strike the cap at exactly the same time. If the rollers do
not strike at the same time the closure system will become stressed on one side and will
cause the bottle and cap to move laterally within the belts. This will result in application

torque variation as well as the possibility of broken bottles during-production causing

unnecessary down time.

5.1 Designing a set-up gauge

The current system for setting up the Resina is inconsistent and allows excessive
variation to be introduced into the process. A single bottle is "jogged" into the machine
until it is aligned with the center of the shaft on the first set of rollers. The separation
between these rollers is then adjusted relative to the bottle/cap position by turning set
screws. The bottle is then "jogged" onto the next pair of rollers. This is done until all four
pairs of rollers are set. In this procedure there is no assurance that bottles will follow a
straight line through the machine. Each individual operator will set the machine up
differently every time. This variation should be reduced.

To eliminate machine set-up variability, a set-up gauge was designed to make
setting up the Resina easier and more consistent. The gauge itself is comprised of a series
of molded epoxy bottles that are attached to each other by two metal rods that run
through the center of each bottle (Figure 6). The epoxy bottles were molded from glass
bottles that are used in production and provide accurate facsimiles of the original glass
bottles. The bottles are spaced 4 1/2" apart which is the distance between the centers of
the spindle shafts on“ the Resina capper.

During a gauge set-up procedure, the gauge is "jogged" into the Resina so that the
center of each of the bottles is directly aligned with the center of the spindle shafts. Once

the gauge is in the machine each of the consecutive pairs of rollers are individually

RESINA SET—UR GAUGE

SIDE VIEW 4W
\/\/\/fl\/

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

CE I D]

METAL RDD

 

 

 

 

 

TDR VIEW L
wmmmmmm

FIGURE 6

26

positioned according to the gauge. The gauge also provides the operator with the correct
straight line path of travel that a bottle will take through the machine.
An in depth analysis of how the gauge performs on actual production runs was

conducted and this analysis follows in the latter part of the thesis.

27

6. CRITICAL PARAMETER DESIGN

6.1 Evaluating The Closure System

To effectively reduce variation in a process, one needs to know what Specific
parameters control that process. In this case the process that needs to be controlled is that
which produces application and removal torques. The variable of application torque is
very difficult to measure in a production setting. Current capping equipment does not
have the capability to provide an in-process application torque readout. Providing such a
feature is cost prohibitive and may not be accurate at high speed production rates. As a
result, removal torque during production runs was used as a surrogate for application
torque. This evaluation of removal torque is specific to the type of closure, the size of the
closure, the liner material, and the materials of the bottle and cap. The packaging system
studied contains a liquid product.

The type of closure evaluated is a 28 mm Child Resistant Closure (CRC). The
inner shell is made from polypropylene resin and the outer shell is made from high density
polyethylene. The bottle that the CRC is applied to is made from type NP or better amber
glass. The liner is a three layer laminated structure consisting of:

a) Backing - made from pulp board with thickness of 0.035 inches.

b) Facing - made from polyethylene that is extrusion coated to 30 pound
white paper.

c) Coating - made from paraffin wax approximately 0.001 inches thick.

The manufacturer of the CRC cap suggests that application torques Should range
from a minimum of 13 TIP to a maximum Of 19 TIP, for this particular closure system.
The supplier also suggests that a minimum removal torque of 3 TIP is required for the

closure system to be engaged and effective as a child resistant closure.

6.2 Determining Removal Torques: Target, Minimum, and Maximum

To determine the target, minimum, and maximum removal torques for this closure
system an experiment was designed to evaluate a number of critical characteristics. The
difference between application torque and immediate removal torque was studied, as was
the effect that time has on removal torque.

The most critical parameter for the closure system is that of maintaining the
removal torque above the three torque inch pounds that is required to maintain the Child
Resistant feature. A safety factor will be incorporated into the evaluation to ensure that
any experimental or material errors can be accounted for.

The experiment itself was comprised of applying a series of bottles and caps at
three different torque levels (12 TIP, 15 TIP, and 18 TTP). Forty eight bottles were used
at each level for a total of one hundred and forty four bottles and caps. The caps were
applied to the water filled bottles utilizing the Sure Torque torque tester. The bottles
could not be filled with the product itself due to Quality Control issues and cost
constraints. Removal torques were then measured at different intervals to evaluate the
effect of time. The intervals used were immediate removal torque (15 minutes or 0.01
Days), and removal torques after Day 1, Day 4, and Day 7. Each removal torque data
point represents an average of twelve closures. The percent retention in Table 4 is the
removal torque as a percentage of the application torque. The raw data is listed in Table 4

in Appendix A. A summary of the data is listed in Table 4:

29

Table 4 - Removal Torque Study vs. Time Showing Removal Torque Decay

 

 

 

 

 

 

 

 

 

 

 

 

 

12 TIP APPLIC. DAY 0.01 * DAY 1 DAY 4 DAY 7
AVERAGE 7.3 TIP 5.8 TIP 5.7 TIP 5.6 TIP
STD. DEV. 0.53 TIP 0.56 TIP 0.41 TIP 0.35 TIP
% RETENTION 60.7%

15 TIP APPLIC. DAY 0.01“ DAY 1 DAY 4 DAY 7
AVERAGE 8.8 TIP 7.1 TIP 6.6 TIP 6.5 TIP
STD. DEV. 0.67 TIP 0.73 TIP 0.46 TIP 0.36 TIP
% RETENTION 58.7%

18 TIP APPLIC. DAY 0.01 "‘ DAY 1 DAY 4 DAY 7
AVERAGE 10.1 TIP 8.2 TIP 8.0 TIP 7.9 TIP
STD. DEV. 0.50 TIP 0.54 TIP 0.63 TIP 0.63 TIP
% RETENTION 56.2%

 

 

 

 

 

 

 

* DAY 0.01 represents immediate removal torque.

Immediate removal torques are significantly lower than application torques. As
the application torque increases, the percent retention between immediate removal and
application torque decreases. The average percent retention within the test parameters is
58.5%, or immediate removal torque is 0.585 * Application Torque (for this particular
closure system). The effect of time is also very significant. The removal torques degrade
over time but it should be noted that the first one to two day period results in the largest
amount of removal torque decay. After the initial drop off, the removal torque

degradation levels off to a very slow rate.

30

6.3 Predicting the Target Value (Application and Removal)

To fully evaluate the effect of decay over time, a regression analysis was
conducted to try to predict removal torque after a given amount of time. The amount of
time is determined by the shelf-life of the product that is being packaged. The shelf-life of
this particular product has previously been determined (by stability testing) to be two years
or seven hundred and thirty days. A multiplicative (logarithmic) model was used to
predict the removal torques after seven hundred and thirty days, at each of the three
application levels. Twelve closure systems were evaluated at each application level, for
each time interval contained in the model. Previous removal torque studies conducted at
other pharmaceutical companies have shown that, at room temperatures, the extent of
relaxation of the closure system has taken full effect after seven to fourteen days. After
the initial relaxation, removal torque decay subsides and the closure system assumes a
constant removal torque value. Although there is a large error associated with making a
prediction of removal torque after two years based on seven days, one can safely assume
that once a constant removal torque is achieved, there will be little variation around the
constant removal torque value. The prediction of removal after two years is therefore

reasonably accurate. These values are tabulated below:

Table 5 - Predicted Removal Torques

 

APPLICATION TORQUE REMOVAL TORQUE TORQUE ABOVE

 

 

AFTER 730 DAYS MINIMUM (3 T'IP)
12 TIP 4.6 TIP 1.6 TIP
15 TIP 5.2 TIP 2.2 TIP

 

 

 

 

 

18 TIP 6.5 TIP 3.5 TIP

 

 

31

The predictions for all of the application torques are above the three torque inch
pound minimum after two years. The desired safety factor to be built into the system
would be two torque inch pounds above the minimum i.e. five torque inch pounds. This
safety factor will provide adequate protection from defective materials, process
inconsistencies, and harsh environmental conditions. The effects of vibration were
Studied and Show that removal torques may decrease as much as two torque inch pounds
after a vibration test. The current process has a safety factor that is considerably lower
and inadequate. Although the 12 TIP prediction (R-squared 64.55%) is 1.6 TIP above
the minimum of 3 TIP (after two years), this value is below the desired two torque inch
pound safety factor. If a target application torque was aimed at twelve torque inch
pounds, the process (assuming normality) will not guarantee that loose caps will not be
found in the field before two years has elapsed. In fact, based on the model closures
would have degraded to five torque inch pounds after only 12.4 weeks. After two years
the predicted removal torques would be 4.6 TIP. The first seven days of removal torque

degradation are represented in Figure 7 - 12 TIP Application.

The 15 TIP prediction (R-squared 71.90%) is 2.2 T'IP above the minimum after
two years. This value is also above the value that incorporates the two torque inch pound
safety factor. Based on the model, it would take 4.75 years to reach five torque inch
pounds. The two year shelf life of the product is well within this time frame. After two
years the predicted removal torques would be 5.2 TIP. The first seven days of removal

torque degradation are represented in Figure 8 - 15 TIP Application.

The 18 TIP prediction (R-squared 68.76%) is 3.5 TIP above the minimum (after
two years). Based on the model, in practical terms, closures would not reach five torque
inch pounds. Theoretically it would take almost 1500 years before these closures would

degrade to five torque inch pounds. After two years the predicted removal torques would

32

be 6.5 TIP. The first seven days of removal torque degradation are represented in Figure
9 - 18 TIP Application.

12 TIP H’PLICATIW

PREDICTED CURVE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TIPEIN DAYS

FIGIRE?

REHOVAL TORQUE

15 TIP APPLICATION

PREDICTED CURVE

 

 

18.8

 

9.6 rs; ..............................

 

 

8.8 i \

 

 

 

 

.s.....cn....e.........-.--..nu...- .....................

 

7.8

-
\
5' —I
l- d
P -
— v . - - - ‘ z
- o
'- -1
_ d

I
J

 

 

 

-‘-‘

 

5.8

 

 

 

 

 

 

 

FIGURE 8

TIME IN DAYS

REHOVN. TIRQUE

18 TIP FPPLICATION

PREDICTED CURVE

 

 

18.9

 

ll
-'u
i

 

 

 

 

 

9.9 ._..j. ....... .\ ......

 

 

 

 

8.9

 

 

 

 

 

 

 

 

a 2
7.9 . ;\-
b ‘ a g ~;~..“ x
\ : ..‘~. .
\ I . -‘~"‘~-
- \‘ g ? a “-5
\ . s
+- x
‘ :
.. .“.é a
. :".. a
8.9 ; .‘_ c
I I I I I I I I I I I I I

 

 

 

 

TIPE IN DAYS

FIGURE9

 

 

36

Based on the above data an application torque target of between 16 TIP and 18
TIP would ensure that a very small percentage of the torques would degrade to below five
torque inch pounds during the shelf life of the product. Tail area probabilities are
calculated in Chapter 7, which show the probabilities of falling outside specified limits
based on calculated process variations. The five torque inch pound value is still two
torque inch pounds above the minimum of three torque inch pounds required to unscrew
the CRC because of the reverse ratchet mechanism. Any torques that degraded below five
would have a very remote chance of falling anywhere close to three torque inch pounds.
By setting an application target of between 16 TIP and 18 TIP, a target immediate
removal torque of between 9.4 TIP and 10.5 TIP is achieved. These values were

calculated using the following formula:

Immediate Removal Torque = 0.585 * Application Torque (7)

6.4 Predicting the Minimum and Maximum Immediate Removal Torque Values

The minimum removal torque, based on the removal torque study, that can be
tolerated in production is seven torque inch pounds. This would mean that a minimum
application torque of 12 TIP would be allowed (based on the predicted Equation 7). This
value would ensure that no removal torques would decay to below three torque inch
pounds over the two year Shelf life of the product.

The maximum removal torque that can be tolerated in production is 14 TIP. This
would mean that a maximum application torque of 24 TIP would be allowed (based on the
predicted Equation 7). This value is 5 TIP above the recommended supplier maximum.
Previous data and current processes suggest that the closure system can withstand well in
excess of the recommended 19 TIP value. A summary of the recommended process

parameters is listed in Table 6 - Summary of Recommended Process Parameters:

37

Table 6 - Summary of Recommended Process Parameters

 

APPLICATION TORQUE REMOVAL TORQUE

 

TARGET 16-18 TIP 9.4-10.5 TIP

 

MININIUM 12TIP 7TIP

 

 

 

 

MAXIMUM 24TIP 14TIP

 

 

38

7. TAGUCHI PROCESS EVALUATION

To evaluate two or more processes utilizing the Taguchi philosophy, one needs to
know various statistical facts about each process. A Target (T) value has to be set, upper
(USL) and lower (LSL) specification limits must be set, and the standard deviation for
each process has to be calculated. Knowing all of these values, one can then calculate the
Mean Squared Deviation (MSD), the Signal to Noise Ratio (S/N), the Loss, and the
Saving associated with a given process. The Performance Characteristic or Process
Capability Index (Cpk) can also be calculated to evaluate the "capability" and centrality of

a given process mean within a given limits.

7.1 Assumptions

For the purpose of the Taguchi analysis, various assumptions were made regarding
each of the given processes that were studied. Each process was assumed to be normally
distributed. The distributions of each of the processes was shifted so that the average of
each process was centered around the predicted target value. This target value was
calculated to be 10.5 TIP based on the removal torque study discussed earlier. In practical
terms the average value for each of the processes can be shifted by adjusting the torque
roller separation so that average of a given process falls on the target value. To lower a
process average the distance between the torque rollers will increase. To raise a process
average, the torque rollers are moved closer together. This assumes that variation in a
process does not change at different application torque levels.

The costs used in the analysis of cost comparisons between processes are not
exact. This information, for proprietary reasons, cannot be fully disclosed. Therefore the

cost of one unit that is rejected during production will be set at one dollar ($1.00).

39

7.2 The Current Process

The current process being used includes the standard operator set-up procedure on
the Resina capper. This set-up does not include using the set-up gauge discussed
previously. The process also uses the Secure Pak removal torque tester as the
measurement device. The data that was used for the analysis of this procedure was taken
from the production orders of three previous production runs. The data was collected in
groups of five samples every hour over each of the production runs. Each production run
is approximately seven hours in length. The raw data is listed in Table 7 in Appendix A.
A summary of the data collected for this process is listed in Table 7 below.

Table 7 - Current Process Data Summary

 

 

 

 

 

 

 

 

 

 

 

 

 

SAMPLE SIZE 80
AVERAGE 1 5.2 TIP
ADJUSTED AVERAGE 10.5 TIP
TARGET VALUE 10.5 TIP
STD. DEV. (CLASSICAL) 2.21 TIP
STD. DEV. (TAGUCHI) 2.19 TIP
MEAN SQUARED DEVIATION 4.82 TIP *
VARIANCE 4.88 TIP
S/N RATIO -6. 83

Cpk 0.53

 

* M = 0 (Target Value = Average Value)

The current process produces a classical standard deviation of 2.21 TIP and a
Taguchi standard deviation of 2. 19 TIP. The Mean Squared Deviation is 4.82 TIP based
on an adjusted process mean to the target value (M = 0). The Signal-to-Noise Ratio is
negative 6.3 calculated from the MSD. The Process Capability Index is 0.53 based on the
Adjusted Average. The data shows that this process is not very capable of producing the
quality that is demanded by the company. Based on this data, six out of every one
hundred closures applied will fall below 7 TIP and above 14 TIP. The calculated
probability is 0.057.

7.3 Incorporating a Better Measurement Device (Sure Torque)

By incorporating the Sure Torque tester into the process, operator variability on
the measurement device can be reduced significantly. During the data collection, the
Resina capper was set up using the standard procedure (without the gauge). Removal
torques were monitored over a three day period during a large production run. Twelve
closures were evaluated every half hour throughout the run. The raw data is listed in
Table 8 in Appendix A. A summary of the data collected for this process is listed in Table
8 below.

41

Table 8 - Improved Process Data Summary

 

 

 

 

 

 

 

 

 

 

 

 

SAMPLE SIZE 908
AVERAGE 8.7 TIP
ADJUSTED AVERAGE 10.5 TIP
TARGET VALUE 1 0.5 TIP
STD. DEV. (CLASSICAL) 1.13 TIP
STD. DEV. QAGUCHI) 1.13 TIP
MEAN SQUARED DEVIATION 1.28 TIP *
VARIANCE 1 .28 TIP
S/N RATIO -1.06

Cpk 1.03

 

“ M = 0 (Target Value = Average Value)

The improved process produces a classical standard deviation of 1.13 TIP and a
Taguchi standard deviation of 1.13 TIP. The Mean Squared Deviation is 1.28 TIP. This
is based on an adjusted process mean to the target value (M = 0). The Signal-tO-Noise
Ratio is negative 1.06 calculated from the MSD. The Process Capability Index is 1.03
based on the Adjusted Average. The data shows that this process is a vast improvement
over the previous process. Based on this data, one out of every one thousand closures

applied will fall below 7 TIP and above 14 TIP. The calculated probability is 0.001.

 

42

7 .4 Incorporating the Set-up Gauge into the Process

Incorporating the set-up gauge discussed in chapter four into the above process
results in further reduction in variation. The process utilizes the Sure Torque
measurement device and the Resina capper is set up using the set-up gauge. Operator
variability is minimized on both the measurement device and the Resina capper. The Sure
Torque and the gauge contribute to reduced operator variability. A better overall set-up is
achieved with the gauge. The raw data is listed in Table 9 in Appendix A. A summary of
the data collected for this process is listed in Table 9 below.

Table 9 - Optimum Process Data Summary

 

 

 

 

 

 

 

 

 

 

 

 

 

SAMPLE SIZE 120
AVERAGE 7.0 TIP
ADJUSTED AVERAGE 1 0.5 TIP
TARGET VALUE 10.5 TIP
STD. DEV. (CLASSICAL) 0.88 TIP
STD. DEV. (TAGUCHI) 0.88 TIP
MEAN SQUARED DEVIATION 0.78 TIP *
VARIANCE 0.79 TIP
S/N RATIO 1.08

Cpk 1.32

 

* M = 0 (Target Value = Average Value)

43

The optimum process produces a classical standard deviation of 0.88 TIP and a
Taguchi standard deviation of 0.88 TIP. The Mean Squared Deviation is 0.78 TIP based
on an adjusted process mean to the target value (M = 0). The Signal-to-Noise Ratio is
1.06 calculated from the MSD. The Process Capability Index is 1.32 based on the
Adjusted Average. The data shows that this process is the best (least amount of variation)
and is the closest to producing the quality demanded by the company. Based on this data,
four out of every one hundred thousand closure systems will fall below 7 TIP and above

14 TIP. The calculated probability is 0.00004.

The three actual process probability density functions are graphically represented
in Figure 10 and Show the distributions of the removal torque values around their actual
means. The three adjusted process data density functions are graphically represented in
Figure 1 1 and Show the distributions of the removal torque values, for each process,

around the adjusted mean (10.5 TIP).

scrum. PRWESS “ITY FINCTIWS

SUREHITHGHBE

 

 

 

 

 

..........

oooooooooooooo

nnnnnnnnnnnnnnnnnnnnnnnnn

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Eng CHI—Heme!

8.1

15

18

REI'DVALTCRGJE

FIGIREIO

HAND NO GAUGE

SURE NO GAUGE

ADJUSTED PROCESS DENSITY FUNCTIONS

SURE HITH GAUGE

 

 

 

 

..............

 

 

 

 

 

 

 

l

 

 

 

.............................................

 

 

 

 

 

................................................................................

 

 

.................

l

L

1

LI

J

 

 

 

 

 

 

 

 

8.3
8.2

>hHmzmn >PH4Hn¢nO¢m

19

15

11

REMOVAL TORQUE

FIGURE 11

8. ESTIMATING THE LOSS AND SAVINGS GENERATED BY THE
IMPROVED PROCESSES

8.1 Calculating the Loss Constant (K)

The Loss Constants (K) for the processes discussed in Chapter 7 will all have the
same value. For each of the processes, the tolerance is the same and was calculated to be
13.5 TIP. The loss associated with one defect during production was assumed to be one
dollar. Although this is not an exact figure, it closely approximates the cost of a defective
unit in production. This one dollar figure incorporates the costs of: labor, materials, and
product materials. For proprietary reasons the exact costs cannot be disclosed in the
thesis. The one dollar does not include research and development costs for the product
itself, cost of returns, cost of a recall, cost of a lost customer etc. The Loss Constant is

calculated by substituting the appropriate values into the Loss Constant Equation 2:

K - (LOSS/ UNIT) / (TOLERANCE)2
- $1. 00 / (3. 5TIP)2
- $0. 082 / TIP2
. WHERE:
TOLERANCE - (14TIP - 7TIP)/2

The resulting Taguchi loss curve that is produced from the above calculation is
graphically represented in Figure 12 - The Taguchi Loss Function for Removal Torque.

O LOSS

1.8.

1.4

1.2

8
O

8.4

THE TAGUCHI LOSS FWCTION FOR

REMOVAL TORQUE

 

IlllIIIlllIllllIIlIIIIIIIIIIIIILIIIIIIIIIIILLL

8 7 8 9 1811 12 13 14 15

REMOVAL TWQUE (TIP)

FIGlRE 12

48

8.2 Comparing the Current Process to the Improved Process

The average Loss associated with one defective unit, utilizing the current process,
is calculated by substituting the above Loss Constant (K) and the Mean Squared Deviation

for the current process into the Taguchi Loss Function Equation 3:

LOSS - K(MSD)
- $0.082 / TIP2 * 4.82T1P2
- $0.393 PER PART

The average Loss associated with one defective unit, utilizing the improved
process, is calculated by substituting the above Loss Constant (K) and the Mean Squared

Deviation for the improved process into the Taguchi Loss Function Equation 3:

LOSS - K(MSD)
- $0.082/TIP2 *1.28TIP2
- $0.104 PER PART

The Saving or benefits of using the Sure Torque over the Secure Pak are
calculated by comparing the Signal-to-Noise Ratios of each process (Current vs.

Improved) as follows:

Savings _ (1 _ 10(-(S/N,-S/N,)no} ) a. 100
_ (1_ 10H-1-06-(433V10i) #100
- (1 - 0 265) . 100
- 73.5%

The difference in S/N ratios reflects a reduction in variation from the three noise

factors, environment in which the product is used, product deterioration with age, and

49

manufacturing variations. External and Internal noise factors are relatively constant for
both of the processes. As a result, the noise factor that has been reduced the most is that
of Unit-tO-Unit variation. Variations in the manufacturing process have been reduced by
incorporating the Sure Torque into the manufacturing process. There is a definite
advantage to using a measurement device that provides more precise removal torque data.
In terms of cost, a seventy three point five percent improvement is realized. This cost
improvement assumes that all out-of-tolerance components are rejected. In reality those
components that are out-of-tolerance above the specified maximum (14 TIP) would not
necessarily be rejected. A more realistic cost saving figure would be half of the
calculated cost saving value (36.8%), assuming all components below the Specified
minimum (7 TIP) are rejected.

8.3 Comparing the Improved Process to the Optimum Process

The average Loss associated with one defective unit, utilizing the improved
process, is calculated by substituting the Loss Constant (K) and the Mean Squared

Deviation for the improved process into the Taguchi Loss Function Equation 3:

LOSS - K(MSD)
- $0.082/TIP2 *1.28TIP2
- $0.104 PER PART

The average Loss associated with one defective unit, utilizing the optimum
process, is calculated by substituting the Loss Constant (K) and the Mean Squared

Deviation for the optimum process into the Taguchi Loss Function:

50

LOSS - K(MSD)
- $0.082/TIP2 * 0.781er
- $0.064 PER PART

The Saving or benefits of using the Sure Torque with the set-up gauge over using
the Sure Torque alone are calculated by comparing the Signal-tO-Noise Ratios of each

process (Improved vs. Optimum) as follows:

Saving _ (1 _ lot-(SINz-Sthflloi)cloo
_ (1-101m4-mmoir 1m
- (1-0.611)"'100
- 38.9%

The Saving analyses Show a definite advantage to using the set-up gauge which
provides more consistency to the application process. Again, the noise factor that has
been reduced the most is that of Unit-to-Unit variation. Variations in the manufacturing
process have been reduced by incorporating the gauge into the manufacturing process that
utilizes the Sure Torque. In terms of cost, a thirty eight point nine percent improvement is
realized. Again, this cost improvement assumes that all out-of-tolerance components are
rejected. A more realistic cost saving figure would be half of the calculated cost saving
value (19.5%), assuming all components below the specified minimum (7 TIP) are
rejected.

8.4 Comparing the Current Process to the Optimum Process

The average Loss associated with one defective unit, utilizing the current process,

is $0.393 and the average Loss associated with one defective unit, utilizing the Optimum

51

process, is $0.064 (calculated above). The Savings or benefits of using the Sure Torque
with the set-up gauge over using the Secure Pak alone are calculated by comparing the

Signal-to-Noise Ratios of each process (Current vs. Optimum) as follows:

Saving _ (1 _ loi-(S/Nz -S/Ni)/1°l ) a 100
_ (1 __ loi-(l-m-(w)/10})c100
- (1- 0.162) *100
- 83.8%

There is a definite advantage to using the Sure Torque and the set-up gauge
compared to using the Secure Pak alone. The Sure Torque provides more precise data
and the set-up gauge improves process consistency. In terms of cost, an eighty three point
eight percent improvement is realized. Again, this cost improvement assumes that all out-
of-tolerance components are rejected. A more realistic cost saving figure would be half
of the calculated cost savings value (41.9%), assuming all components below the Specified

minimum (7 TIP) are rejected.

52

9. RECOMMENDATIONS

9.1 Measurement Device

The Secure Pak does not provide a consistent tool for measuring removal torque.
The Sure Torque is more precise and more consistent than the Secure Pak. The cost
benefit of utilizing the Sure Torque over the Secure Pak is evident in the Taguchi cost
saving analysis. It would be beneficial for the Company to implement these Sure Torque
removal torque devices on current packaging lines. The Sure Torque tester also lowers

the probability of a recall occurring by reducing operator variability.

9.2 Resina Set-up Gauge

The set-up gauge that was designed to be used in conjunction with the Resina
capper resulted in definite process improvements. Process variation was reduced and
improved consistency of the capper set-up was realized. The probability of a recall
occurring will be reduced by incorporating the set-up gauges on current packaging lines.
The set-up gauges are inexpensive and can be made for each bottle style currently being

[111'].

9.3 Critical Controls

The current process controls were determined by deduction rather than by
conducting tests. The recommended critical controls contained in the thesis are based on
testing rather than deduction. The current and recommended process controls are
summarized in Table 10 - Summary of Process Controls. If these recommended process

controls are implemented, the probability of a recall is reduced. The current process has

53

only one critical control (minimum removal torque), which is too low to ensure that a
recall does not occur. For a process to be effective, a target value, a minimum value, and

a maximum value should be set to keep it under control.

Table 10 - Summary of Process Controls

 

 

 

 

 

 

 

CONTROL CURRENT RECOMMENDED
MINIMUM REMOVAL 5 TIP 7 TIP

TORQUE

TARGET REMOVAL NONE 10.5 TIP
TORQUE

MAXIMUM REMOVAL NONE 14 TIP

TORQUE

 

9.4 Implement an Aggressive Statistical Process Control Program (SPC)

The current Quality Control program utiliwd on the packaging lines can be
improved by implementing an SPC program. The current quality control program does
not allow for effective in-process monitoring of quality. Statistical checks are made but
nothing is done with that data during production. AS a result, any defects that are found
after production require a one hundred percent inspection of the lot to locate a specific
problem area. Reworking is not cost and time effective. An SPC program will reduce
process costs and manufacturing time. This is another component of K that is not

accounted for. If this figure is calculated the value of K would be greater. SPC allows

 

54

one to evaluate a process while it is in progress, so that defects can be identified and

process adjustments can be made during production.
9.5 Purchase a New Capper?

Purchase of a new chuck style capper is currently under consideration. Based on
the data contained in the thesis, there is no need to purchase a new capping machine. The
Resina capper can produce the quality that is required by the company. The Company can
save the cost of a new machine if the Sure Torque and the set-up gauges are incorporated
into the process. All packaging materials will also have to be converted to the same
components that are described in the materials section. This money can contribute better
process improvements in other areas. If the Resina is replaced, the overall output of the
packaging line will not increase. It would be better to apply this money to replace

equipment that currently limits the process output (e. g. Liquid Filler).
9.6 Future Research

Future research areas include:

1. An investigation of other bottle/closure combinations. This research Should
focus on closure sizes different from the 28 mm closure already studied. Different
materials combinations (e.g. metal caps on plastic bottles) should be evaluated. Not all
closure systems behave in the same manner and different specification limits and target
values will have to be determined for individual closure systems.

2. A more in depth investigation of removal torque decay over a longer period of
time. Due to time constraints, the thesis did not allow for a longer test period. This will

allow for more accurate predictions of removal torque after a specified time period.

55

3. A test of the effect of placing a shrink band around the closure system should
be conducted to evaluate how the band reduces removal torque decay.

4. The effects of vibration on removal torque decay should be evaluated.

56

10. SUMMARY OF MATERIALS USED

10.1 Materials Tested

1. Product - Liquid (water for non-production testing and product for production
testing).

2. Glass Bottle - 28 NHVI finish, 120 ML volume, and NP or better amber glass

3. Cap - 28 MM Child Resistant Closure with Outer (High Density Polyethylene)
and Inner (Polypropylene).

4. Liner - Backing (0.035" pulp board), Facing (Polyethylene extrusion coated to
30 pound white paper), and Coating (0.001" paraffin wax).

10.2 Equipment Used

1. Sure Torque Tester (Model NEBT33).

2. Secure Pak Tester (ET 87503), Serial Number: (100 115D).

3. Resina Capper (U40 805), Serial Number: (81 6174).

4. Resina Set-up Gauge - manufactured in company maintenance shops according

to specifications laid out by Christopher A. 1. Clarke.

10.3 Testing Conditions

All testing was conducted at the Company's ambient conditions: 70-75°F and less

than40% RH.

10.4 Analytical Tools
1. Statgraphics Plus Version 6.0 Software
2. Qualitek-4 Software
3. IBM 386 Computer

APPENDIX

57

APPENDD( A - RAW DATA TABLES

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 1
OPERATOR VARIABILITY sruov T SECURE PAK (HAND)
SAMPLE OPERATOR ONE OPERATOR TWO
1 14.7 13.0
2 14.3 11.3
3 17.1 ‘ 12.3
4 12.1 12.6
5 19.1 13.1
6 13.0 12.2
7 14.8 13.5
8 14.4 12.2
9 14.1 13.5
10 15.0 14.5
11 16.7 11.4
12 18.6 12.6
13 17.6 10.3
14 15.6 12.2
15 16.7 14.0
16 15.8 12.9
17 17.4 12.8
18 17.0 12.6
19 15.4 13.9
20 15.6 13.1
21 15.6 13.4
22 14.6 13.3
23 15.5 14.0
24 17.2 14.2
25 16.5 14.8
26 15.5 13.0
27 16.4 14.4
28 16.6 12.8
29 16.3 14.8
30 19.7 14.4
31 14.3 14.8
32 14.4 10.4
33 15.1 15.5
34 16.1 13.3
35 17.5 14.1
36 14.6 12.9
37 18.2 14.3
38 16.0 12.3
39 17.8 14.1
40 16.4 13.8
41 16.5 13.9
42 15.7 12.9
43 18.5 14.8

 

 

 

 

 

 

 

59

 

OPERATOR VARIABILITY STUDY - SECURE PAK (HAND) - CONTINUED

 

 

 

 

 

 

 

 

 

 

 

 

 

 

44 15.6 14.5
45 17.3 13.3
46 17.8 14.1
47 17.5 12.3
48 17.3 14.2
49 17.9 15.4
50 17.8 12.7
* ALL VALUES IN TORQUE INCH POUNDS (TIP)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 2
OPERATOR VARIABIILITY - SHURE TORQUE TESTER
l
SAMPLE OPERATOR ONE OPERATOR TWO
1 14.0 10.1
2 12.1 10.9
3 11.7 11.0
4 10.9 11.5
5 8.2 10.9
6 13.2 10.9
7 11.1 12.8
8 10.5 10.0
9 9.9 11.0
10 9.3 9.7
11 10.5 13.3
12 12.9 11.8
13 N/A 14.5
14 12.0 11.3
15 12.7 11.0
16 12.2 10.6
17 11.7 12.0
18 13.8 11.3
19 9.9 12.0
20 9.2 13.5
21 N/A 12.1
22 11.6 12.0
23 13.8 11.7
24 13.2 11.8
25 12.4 11.6
26 12.0 12.4
27 12.1 10.4
28 13.4 10.3
29 11.7 9.5
30 13.4 11.3
31 13.5 11.0
32 14.2 12.1
33 13.0 11.3
34 8.7 10.5
35 13.5 10.0
36 14.0 11.2
37 12.5 11.7
38 12.1 10.0
39 11.6 10.5
40 13.0 11.1
41 15.6 11.6
42 12.3 11.7
43 12.7 12.2

 

 

 

 

 

 

 

 

 

61

 

 

 

 

 

 

 

 

 

OPERATOR VARIABILITY - SHURE TORQUE TESTER - CONTINUED
44 13.2 10.7
45 13.2 14.5
46 14.1 11.8
47 13.8 10.2
48 12.3 11.9
49 14.4 12.4
50 12.4 12.3

 

 

 

 

 

 

 

 

 

 

N/A - CAPS WERE NOT IN THE LOCKED POSITION AND WERE DISCARDED

 

 

62

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 3
TORQUE TESTER VARIABILITY
SAMPLE SECURE PAK (HAND) SHURE TORQUE (AUTO.)
1 14.7 14.0
2 14.3 12.1
3 17.1 11.7
4 12.1 10.9
5 19.1 8.2
6 13.0 13.2
7 14.8 11.1
8 14.4 10.5
9 14.1 9.9
10 15.0 9.3
11 16.7 10.5
12 18.6 12.9
13 17.6 N/A
14 15.6 12.0
15 16.7 12.7
16 15.8 12.2
17 17.4 11.7
18 17.0 13.8
19 15.4 9.9
20 15.6 9.2
21 15.6 N/A
22 14.6 11.6
23 15.5 13.8
24 17.2 13.2
25 16.5 12.4
26 15.5 12.0
27 16.4 12.1
28 16.6 13.4
29 16.3 11.7
30 19.7 13.4
31 14.3 13.5
32 14.4 14.2
33 15.1 13.0
34 16.1 8.7
35 17.5 13.5
36 14.6 14.0
37 18.2 12.5
38 16.0 12.1
39 17.8 11.6
40 16.4 13.0
41 16.5 15.6
42 15.7 12.3
43 18.5 12.7

 

 

 

 

 

63

 

TORQUE TESTER VARIABILITY - CONTINUED

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

44 15.6 13.2
45 17.3 13.2
46 17.8 14.1
47 17.5 13.8
48 17.3 12.3
49 17.9 14.4
50 17.8 12.4
51 13.0 10.1
52 11.3 10.9
53 12.3 11.0
54 12.6 11.5
55 13.1 10.9
56 12.2 10.9
57 13.5 12.8
58 12.2 10.0
59 13.5 11.0
60 14.5 9.7
61 11.4 13.3
62 12.6 11.8
63 10.3 14.5
64 12.2 11.3
65 14.0 11.0
66 12.9 10.6
67 12.8 12.0
68 12.6 11.3
69 13.9 12.0
70 13.1 13.5
71 13.4 12.1
72 13.3 12.0
73 14.0 11.7
74 14.2 11.8
75 14.8 11.6
76 13.0 12.4
77 14.4 10.4
78 12.8 10.3
79 14.8 9.5
80 14.4 11.3
81 14.8 11.0
82 10.4 12.1
83 15.5 11.3
84 13.3 10.5
85 14.1 10.0
86 12.9 11.2
87 14.3 11.7
88 12.3 10.0
89 14.1 10.5
90 13.8 11.1

 

 

 

64

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TORQUE TESTER VARIABILITY - CONTINUED
91 13.9 11.6
92 12.9 11.7
93 14.8 12.2
94 14.5 10.7
95 13.3 14.5
96 14.1 1 1.8
97 12.3 10.2
98 14.2 11.9
99 15.4 12.4
100 12.7 12.3
* ALL VALUES IN TORQUE INCH POUNDS (TIP)

 

 

N/A-CAPS WERE NOT IN THE LOCKED POSITION AND WERE DISCARDED

 

 

 

TABLE 4

 

 

 

 

REMOVAL TORQUE STUDY VS. TIME

SHOWING REMOVAL TORQUE DECAY

 

I

l

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

IMMEDIATE DAY 1
APPLICATION 12 15 18 12 15 18
IMMEDIATE 7.4 8.4 10.2 5.8 7.5 7.8
7.7 9.5 9.9 5.9 8.4 8.5
6.9 8.4 10.2 6.2 7.5 8.8
7.4 8.0 9.3 4.7 6.6 7.7
7.6 8.8 9.8 6.4 7.0 9.1
8.2 8.4 10.8 6.1 6.7 7.4
7.7 10.1 10.4 6.3 5.8 8.7
7.0 9.4 9.5 5.1 6.3 8.6
7.5 8.8 10.8 5.7 7.1 8.0
6.2 8.8 10.7 6.0 8.0 7.6
6.9 7.8 9.9 6.5 7.7 8.0
6.8 9.3 9.8 5.2 7.0 8.1
AVERAGE 7.28 8.81 10.11 5.83 7.13 8.19
STD 0.53 0.67 0.50 0.56 0. 73 0.54
DAY 4 DAY 7
APPLICATION 12 15 18 12 15 18
IMMEDIATE 5.8 6.6 8.9 5.1 6.6 7.0
6.1 6.6 7.6 5.7 6.6 8.1
5.9 6.6 8.3 5.4 6.2 8.4
5.9 5.6 7.7 5.4 7.0 7.7
5.1 6.9 8.6 6.3 6.5 8.8
5.1 6.3 7.8 5.5 6.7 6.9
6.1 7.2 8.8 6.1 5.7 8.1
5.7 6.2 7.5 5.3 6.4 8.0
5.5 6.6 6.9 5.4 6.2 8.2
6.3 6.5 7.5 5.8 6.6 7.5
5.5 7.0 7.7 5.6 6.4 8.6
5.2 7.3 8.6 5.9 7.0 7.1
AVERAGE 5.68 6.62 7.99 5.63 6.49 7.87
STD 0.41 0.46 0.63 0.35 0.36 0.63
ALL VALUES IN TORQUE INCH POUNDS

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 7

CURRENT PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
1 18.9 21 11.2 41 18.4 61 14.0
2 15.6 22 11.1 42 18.7 62 16.0
3 17.1 23 10.2 43 17.9 63 15.0
4 15.9 24 15.9 44 18.2 64 15.0
5 14.7 25 10.4 45 16.3 65 17.0
6 14.6 26 10.9 46 17.5 66 15.0
7 14.9 27 11.5 47 ‘ 15.9 67 14.0
8 14.1 28 13.1 48 16.1 68 16.0
9 15.2 29 12.8 49 16.8 69 15.0
10 15.0 30 11.8 50 17.1 70 13.0
11 12.6 31 16.1 51 11.0 71 15.0
12 13.0 32 16.5 52 15.0 72 15.0
13 15.5 33 14.5 53 15.5 73 14.0
14 16.6 34 17.5 54 13.0 74 16.0
15 14.7 35 17.9 55 15.1 75 14.0
16 15.7 36 17.5 56 15.0 76 14.0
17 13.4 37 19.8 57 16.0 77 14.0
18 13.0 38 18.4 58 15.0 78 15.0
19 21.2 39 17.0 59 15.0 79 15.0
20 15.1 40 19.3 60 17.0 80 14.0

* ALL VALUES IN TORQUE INCH POUNDS

 

 

 

 

 

 

67

 

TABLE 8

 

 

IMPROVED PROCESS RAW DATA

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
1 11 41 8.5 81 8.5 121 7.8
2 9.4 42 9 82 8.9 122 8.9
3 9.9 43 7.8 83 9.7 123 9.9
4 10.2 44 9.4 84 8.3 124 9.3
5 9.1 45 9.1 85 8.3 125 8.6
6 10.1 46 8.3 86 7.7 126 9.9
7 10.3 47 10.3 87 11.8 127 8.4
8 9.9 48 8.9 88 7.8 128 10.4
9 9.3 49 9.7 89 9 129 9.9
10 8.5 50 8.4 90 9.4 130 8.4
11 9.9 51 11 91 10.4 131 7
12 8.1 52 7.8 92 10.5 132 7.3
13 9.8 53 9.5 93 10.6 133 7.6
14 11.3 54 8.7 94 10.1 134 9.1
15 9.9 55 9.3 95 10.3 135 7.5
16 9 56 11 96 9.5 136 8.4
17 8.6 57 11.3 97 9.8 137 8.8
18 10.2 58 7.1 98 11.3 138 8.7
19 10.3 59 7 99 9.4 139 8.8

20 10.9 60 7.3 If!) 10.2 140 8
21 10.4 61 7.9 101 9.1 141 10
22 9.9 62 7.9 102 8.5 142 11.2
23 8.7 63 9.9 103 9.6 143 9.6
24 9.9 64 9.4 104 9.5 144 7.9
25 9.5 65 9.8 105 8 145 8.4
26 9.3 66 9.9 106 10.2 146 8.4
27 9.6 67 8.4 107 13.5 147 9.1
28 8.4 68 9.2 108 10.6 148 9
29 8.1 69 7.7 109 9.5 149 9.7
30 7.8 70 8.7 110 10.1 150 8.4
31 12.6 71 9.5 111 9.3 151 9
32 9.9 72 9.2 1 12 9.6 152 8.7
33 9.3 73 8.6 113 8.6 153 9.1
34 9.9 74 8.8 114 7.9 154 9.6
35 8.9 75 8.8 115 9.3 155 9.2
36 9.7 76 9.1 116 8.1 156 9.8
37 9.6 77 8 1 17 8 157 9.5
38 10.5 78 9 118 8.4 158 9.5
39 8.3 79 10.1 119 9.3 159 8.7
40 8 80 8.9 120 8.4 160 8.9

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 8 - (IZONTINUEEI)

IMPROVED PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
161 7.9 201 8.8 241 7.9 281 7.5
1 62 8.4 202 7 242 9.9 282 7.2
163 9.4 203 11.9 243 8.9 283 10.4
164 9.9 204 9 244 8.2 284 9.7
165 9.1 205 8.9 245 10.1 285 10|
1 66 8.3 206 7 246 1 0.4 286 9.2
167 9.2 207 8.8 247 7.6 287 7.2
168 9. 7 208 7. 7 248 9.6 288 8.8
169 8.9 209 10.9 249 9.6 289 I 7.4
170 7.6 210 8.6 250 9.2 290 8.8
171 8.4 211 9.6 251 8.9 291 9.9
172 9.8 212 8.9 252 9.3 292 8.5
173 9.4 213 9.1 253 10 293 9
174 9.4 214 8.9 254 9.7 294 8.7
175 10.8 215 9.9 255 8.7 295 9.1
176 7.7 216 8 256 8.1 296 9
177 8.8 217 9.1 257 7.3 297 9.3
178 8.4 218 9.6 258 7.6 298 11
179 8.6 219 10.1 259 9.9 299 8.2
180 9.1 220 9.6 260 10 3(1) 8.5
181 9.2 221 8.4 261 8.9 301 9.3
1 82 9 222 6.9 262 8.4 302 9.9
183 11 223 7.6 263 7.6 303 9.5
184 10.1 224 9.8 264 8.9 304 10.4
185 8.5 225 8.7 265 9.1 305 8.7
186 10.2 226 10.2 266 10.6 306 9.9
187 8.1 227 11.4 267 8.1 307 7.7
188 10.1 228 8.4 268 9.2 308 6.8
189 9.1 229 10.3 269 11.3 309 7.8
190 9.8 230 7.7 270 8.8 310 7.9
191 9.8 231 10.2 271 8.6 311 9
192 8.9 232 8.5 272 7 312 8.5
193 8.8 233 9.1 273 9.6 313 8.5
194 10.6 234 8.3 274 9.1 314 8.8
195 8.8 235 9.3 275 9.5 315 8.6
196 10.7 236 10.6 276 9.1 316 10.3
197 8.7 237 8.2 277 8.7 317 10.5
198 9.5 238 8.1 278 10.9 318 10.8
199 9.5 239 9.6 279 1 1.7 319 9.4
200 1 1.3 240 9.9 280 8.6 320 10.2

 

 

 

 

 

 

 

 

 

 

 

 

 

69

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 8 - CONTINUE?

IMPROVED PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
321 8.2 361 8.3 401 7.8 441 9.2
322 9.7 362 7.9 402 8.4 442 8
323 9.3 363 8.9 403 8.9 443 10.2
324 11 364 7.3 404 7.7 444 7.9
325 8.8 365 9.5 405 9.5 445 9.4
326 9.6 366 8.1 406 7.6 446 8.8
327 7.6 367 8.1 407 9.9 447 7. 7
328 8.5 368 7.4 408 9.5 448 8.9
329 9.1 369 7.6 409 9.4 449 7.5
330 10.2 370 9.5 410 9.1 450 7.7
331 9.5 371 9.2 411 8.2 451 8.4
332 7.9 372 8.2 412 9.9 452 8.1
333 7.1 373 8.7 413 8.2 453 9.4
334 10.5 374 9 414 8.2 454 8.8
335 8.7 375 9.3 415 8.8 455 10.2
336 9.4 376 8 416 8 456 7.4
337 8 377 9.8 417 10.7 457 8.4
338 8.5 378 8.4 418 9.9 458 8.8
339 9.5 379 9.1 419 8 459 9.1
340 9 380 9.1 420 10.5 460 8.3
341 12.6 381 9.1 421 9.9 461 8.4
342 8.3 382 7.7 422 9.5 462 8.3
343 9 383 9.7 423 8.8 463 8.3
344 1 1.3 384 8.3 424 8.4 464 10.6
345 10.5 385 8.4 425 9.1 465 8.8
346 7.8 386 8.2 426 8.4 466 8.7
347 8.9 387 10.5 427 7.5 467 8.6
348 8.9 388 1 1.3 428 8 468 7.8
349 8.6 389 10. 1 429 8.8 469 8
350 1 1.2 390 10.8 430 8.8 470 7.8
351 9.5 391 9.9 431 9.8 471 9.2
352 8.3 392 10 432 7. 7 472 8
353 9.3 393 8. 1 433 9. 1 473 7
354 8.1 394 8. 7 434 8.5 474 9
355 8.3 395 10 435 9.1 475 10.2
356 9.6 396 8.4 436 9.7 476 9.8
357 9.6 397 8.2 437 7.8 477 7.3
358 9. 1 398 8 438 7.3 478 9.9
359 6.6 399 9.1 439 7.9 479 7. 7
360 8.8 400 8.8 440 9.1 480 8.8

 

 

 

 

 

 

 

 

 

 

 

 

 

70

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 8 - (IIONTINUEIID

IMPROVED PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
481 7.9 521 9.3 561 10 601 9.4
482 8.3 522 9.4 562 9.6 602 8.6
483 8.7 523 9.5 563 7 603 6.2
484 7.7 524 6.6 564 8.9 604 8.6
485 8.4 525 8.4 565 8.3 605 7.6
486 7.8 526 8.4 566 8.5 606 8.9
487 8.1 527 9.9 567 9 607 8.2
488 1 1 528 9.2 568 8.7 608 7.1
489 7.1 529 9 569 10.2 609 8.9
490 8.3 530 10.3 570 8.2 610 7.4
491 9.5 531 7.1 571 7.6 611 9.1
492 9.5 532 7.6 572 8.7 612 7.2
493 8.5 533 8.4 573 9.5 613 6.4
494 8.8 534 8.1 574 7 614 7.1
495 7.2 535 9.5 575 8. 7 615 9.8
496 8 536 5. 1 576 10.2 616 8.3
497 9.9 537 8.5 577 9.1 617 8.4
498 7.4 538 8.5 578 7.3 618 9.2
499 7. 7 539 9.5 579 8.6 619 8.1
500 7.3 540 8.5 580 9.2 620 9.1
501 9 541 7.6 581 8.4 621 9.1
502 7.2 542 7.8 582 7.8 622 8.2
503 9 543 7.4 583 8 623 7.2
504 7.9 544 8.7 584 6.7 624 8.9
505 10.4 545 9.3 585 6.5 625 7.2
506 8.9 546 7.7 586 9.3 626 8.4
507 1 1 547 9.5 587 9.7 627 9.6
508 10.5 548 8.4 588 9.5 628 8.5
509 9.3 549 8.9 589 11 629 9.2
510 10.1 550 10.3 590 8.3 630 9.4
511 10.3 551 7.6 591 8.8 631 8.4
512 8 552 7.1 592 8.3 632 7.7
513 6.6 553 6.1 593 8.4 633 9.4
514 9.5 554 7 594 8.8 634 8
515 8.2 555 8.1 595 10.2 635 6.7
516 9.9 556 9.5 596 10.1 636 7.3
517 8 557 6.2 597 8.8 637
518 7.5 558 9 598 9.3 638 8
519 10.6 559 9.1 599 8.6 639 9.3
520 8 560 7.9 600 8.1 640 8.8

 

 

 

 

 

 

 

 

 

 

 

 

 

71

 

TABLE 8 - CONTINUED

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

l I
IMPROVED PROCESS RAW DATA
SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
641 9.5 681 8 721 7.4 761 9.5
642 10.6 682 8.2 722 7.2 762 7.8
643 6.3 683 9 723 7.7 763 8.5
644 6.8 684 9.4 724 8.5 764 9.1
645 8.7 685 9.1 725 9.9 765 8.5
646 8.4 686 9.8 726 8.5 766 8.5
647 9.4 687 8.1 727 8 767 8.3
648 8.8 688 7.8 728 7.3 768 8
649 8.6 689 10.3 729 7.4 769 13.3
650 8.5 690 6.5 730 6.9 770 8.4
651 7.7 691 8.7 731 10 771 10
652 8 692 8.7 732 7.7 772 8.9
653 9.1 693 7.3 733 10.6 773 7.9
654 6.3 694 7 734 8. 1 774 6.9
655 6.9 695 9. 7 735 8.1 775 9.2
656 7.7 696 7.7 736 10.2 776 7.7
657 7.4 697 7.9 737 10.6 777 7.8
658 7.5 698 7.9 738 9.4 778 8.6
659 7.8 699 8.7 739 8.6 779 8.4
660 9.9 700 8.3 740 8.4 780 9.8
661 6.9 701 8.7 741 8.3 781 8.7
662 6.9 702 8.1 742 8.6 782 9.4
663 8.7 703 9.8 743 8.6 783 8.7
664 7.3 704 8.6 744 6.8 784 7.7
665 8.9 705 9.1 745 9.2 785 8.4
666 6.8 706 9.7 746 8.4 786 6.8
667 6.9 707 7.4 747 6.9 787 9.9
668 8.4 708 7.9 748 8 788 8.2
669 8.6 709 7.6 749 7.7 789 8.8
670 7 710 6.3 750 6.7 790 7.2
671 9.6 711 7.5 751 8.4 791 9.6
672 8.8 712 7.1 752 8 792 7.8
673 7.3 713 6.9 753 8.3 793 9.6
674 6.6 714 6.9 754 7.5 794 7.8
675 6.6 71 5 6.5 755 7 795 9.5
676 7.3 716 9.8 756 8.8 796 9.7
677 5.3 71 7 9.3 757 7.6 797 8.2
678 9.2 718 8.5 758 8.5 798 8.3
679 7.7 719 6.7 759 9.4 799 8.4
680 6.6 720 9.7 760 7.6 8(1) 8.4

 

 

 

 

 

 

 

 

 

 

 

 

 

72

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 8 - CONTINUED
I

IMPROVED PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
801 8.5 841 9.2 881 9.9
802 9.8 842 8.4 882 7.5
803 9.5 843 7.9 883 8.7
804 8.4 844 7 884 6.9
805 9.8 845 8 885 7.7
806 7.9 846 8.6 886 6.6
807 8.7 847 8.4 887 8.9
808 7.9 848 5.9 888 7.9
809 7. 7 849 9.1 889 8.6
810 8.8 850 8.8 890 9.1
811 8.4 851 7.5 891 7.1
812 8.6 852 8.8 892 8.8
813 7.1 853 8.2 893 7
814 8.4 854 8.4 894 8.8
815 8 855 7.5 895 7.4
816 7.1 856 5.5 896 8.2
817 7.3 857 7.7 897 8.4
818 8.6 858 7.9 898 7.9
819 8.8 859 7.4 899 9.6
820 8 860 8.2 900 7.7
821 6.7 861 8.7 901 9.1
822 7 862 8.4 902 8.9
823 7.4 863 9.3 903 8
824 7.6 864 9.2 904 6.9
825 8 865 8.9 905 8
826 7.3 866 7.7 906 7.1
827 6.6 867 10 907 13.8
828 8.4 868 7.6 908 8.7
829 8.8 869 6.9
830 6.2 870 7.4
831 7.7 871 8
832 8.4 872 6.5
833 7.8 873 8
834 9.4 874 8.3
835 10 875 8
836 5.7 876 8.8
837 7.1 877 8.8
838 7.5 878 8.5
839 7.4 879 9.3
840 9.2 880 8.2

 

 

 

 

 

 

 

 

 

73

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

TABLE 9

OPTIMUM PROCESS RAW DATA

SAMPLE VALUE SAMPLE VALUE SAMPLE VALUE
1 6.9 41 5.7 81 5.8
2 7.3 42 7.1 82 8.8
3 8.4 43 9.9 83 6.9
4 6.4 44 7.9 84 6.1
5 6.1 45 7.3 85 5.9
6 6.3 46 7.2 86 5.8
7 5.8 47 7.3 87 6.9
8 8 48 6.5 88 6.9
9 5.9 49 6.6 89 6.5
10 6.2 50 7.3 90 6.2
11 6.1 51 8.1 91 6.9
12 6.1 52 5.8 92 7.5
13 6.2 53 6 93 6
14 5.9 54 6.7 94 5.7
15 6.1 55 7.3 95 6.2
16 6.3 56 8.4 96 6
17 7.2 57 8.8 97 6.9
18 7.3 58 8.4 98 6.5
19 7 59 8.7 99 6.9
20 6.6 60 7.7 100 6.6
21 5.7 61 8.2 101 7.1
22 5.7 62 6.7 102 6.9
23 7.5 63 7.3 103 6.7
24 6.2 64 8.3 104 6.6
25 6.9 65 9.1 105 6.2
26 7.9 66 9.4 106 7.7
27 8 67 7.3 107 7
28 7.9 68 8.5 108 6.7
29 8.3 69 7.7 109 7.3
30 6.1 70 7.9 110 6.8
31 7.3 71 8.3 111 7.8
32 7.1 72 8.3 112 7
33 6.7 73 5.8 113 6.6
34 6.6 74 6.5 114 6.4
35 7.3 75 6.7 115 6.8
36 7.1 76 6.8 116 7.5
37 8 77 6.5 117 6.8
38 7.5 78 7.3 118 7.2
39 8.4 79 6.8 119 7.5
40 7.8 80 6.3 120 7.7

* ALL VALUES IN TORQUE INCH POUNDS

 

 

BIBLIOGRAPHY

74

BIBLIOGRAPHY

1. Gilliland, Dennis. W. Dubuque: Kendall/Hunt, 1990.

2. Bhattacharyya, Gouri K., and Richard A. Johnson. Statistical Coggpts and Methods.
New York: John Wiley and Sons, 1977.

3. Box, George E. P., William G. Hunter, and J. Stuart Hunter. Statg' tifi for
Exm’ enters. New York: John Wiley and Sons, 1978.

4. Davies, Owen L.. MW. London: Oliver and
Boyd, 1960.

5. Barker, Thomas 3.. W New York: Marcel Dekker,
Inc., 1985.

. Taguchi, Genichi. W. New York: Kraus
International Publications, 1986.
. Taguchi, Genichi, and Wu, Yuin. WW. Japan:
Meieki Nakamura-Ku Nagaya, 1979.
8. Juran, J. M., and Frank M. Gryna, Jr.. W. New York:
McGraw Hill, Inc., 1970.
9. Manugistics. WM. Maryland: Manugistics, 1992.
10. Ross, Phillip J.. W New York:
McGraw Hill, Inc., 1988.
11. Qualitek-4. MW. Nutek 1991.
12. Oakland, John S., and Followell, Roy F.. W. Heinemann
Newnes Publishing Ltd., 1990.

13. Motgomcry. Douglas C. 12W New York:
John Wiley and Sons.

0

\l

14. Moen, Ronald D., Nolan, Thomas W., and Provost, LLoyd P.. W
W. New York: McGraw Hill. Ino-