:

DOCTORA L T H E S I S

Simulation Driven Product Development

Applied to Car Body Design

Nicklas Bylund

Luleå University of Technology

Department of Applied Physics and Mechanical Engineering, Division of Computer Aided Design

:|: -|: - -- ⁄ -- 

!"# $" % &
 ' ' ( ' ' ' )
' %* ' ' +* ,% & ( ' ( '
) - ' ' (
+ ,% -
' &' ' ( * ' ' * ' . ' '
' & -/&
0 ' ' ' 1 ' 2 %3 '
( 4 ' ' ' ' 56 ) 4
% ' ' ' ' & 6-
/ ' & 6 ' 6 ' ' . '
7' ' ' ' % ' '( '
' &' ( ' ' '( - / & 6 ' 6 ' ' )
% ' ) ' 3' 8 '
9' : ' ' ' ) ' ' ) '
- /& ' 6 ' 6 ' ; < ' ' ' ' '
) % ' '
* - ' ( ' / & 6 ' 6 ' '
% ' ' ' & 6 -
' ' & 2 ' ' ' '
: ; +!! ,% */::8* . '( ' '
+= , ' * ' ' ' ' ' 6 & -
: ; ' ' ' ' ' ' ( & ' ' '( '
& 6' & ' -

> ( %

: 6' )
 ' & 0 ' ' %
' %' ' ' ( -
' ( ' ' & ' ' ' '
' -/ ' ( '
' ' ' ' ' '' - ' ( '
' ' ' - )' '(
% ' ' & 6 ' ( % 0
' - ' ' ( ' %
' ' ' 4 ' ( -/ ' ( &
' ( & 4 '
' '% ' & ' ' % (
' ' ' '' & '
' % ' ' ' ' 0 -
' ' '
' ' ' ' ' % ' ( -

' ( ' ( ' % ( ' '

' - ' '
' ? ' * -

% % ' % 6 & '(

% &' % ' ( %
 ' ' ' & ' @

)A : % :- B ;/7 8:% 3-

! "#$ % &' ( ) $ $ $* %

)A : % :-% 2; ;/ 8:% 9- B 983. 8:% >-

% & % + %, % %
$ - , % % /
% # #" 3' % ( 6% ' '-

)A : % :-
.' / & , && &0 % ,
% % - ( ' 1 '
' / ' / %1/ -

)A : % :-
* % % && & )
' =" C% / ' ' )
% /) =% " ! 8 ( =% ('% 1' ' -

"

)A : % :- B 2; ;/ 8:% 9-
* , & 1 && & 2 ,%
' =" $ / ' ' )
% /) =% " ! 8 ( =% ('% 1' ' -

8. D 3 % )-% )A : % :- B 983. 8:% >-

&, & , %
( 1 ' . ( ' -
 #

)A : % :-% / 7 8:%- 7 98;/% *- B ; 8:% -

, % %% 3 - - ,
, - / % #$ 3'
% ( 6% ' '-

)A : % :- B 7 ; 8:% -
/ $ & &- & %
- ( 1 ' . ( ' -

8 ( ' ( ' @

)A : % :-% )A ;E3% 1- B . ; 8:% -

. 4 5 6 , % %
/ ' / ' '
% C ## #% % ' -

>; : % - B )A : % :-
& , &&% & && %, % , %
/ % # #" 3' % ( 6%
' '-

)A : % :- B >; : % -
%, % &&% +
/ :8; />: # #F % %: &' -

; 8:% -% >; : % - B )A : % :-
7 1 ' 8' % , ' % &
. %,
/ . " / ' ' ; ' G 0 H
3' ' ' H% #C #F (
=% 3' % ' -

)A : % :-% >; : % - B 8. D 3 -
& , & %,
/ / ' ' %
/ =% #! # =% 6 % & -
 % &'( ) &%( *****************************************************************************************************
#-# ) 7>;8 : #
#- 38 /* /8: #
#-= */ /8: #
'# # ' + ( &#,& *******************************************************************************************!

! !

" &+# ' ( -*******************************************************************************************************!

. (/ 0# 1# (2 % *************************************************************************************3
-# :>/: ;/:> />: ; ; 9 F
- /3 /8: $
" 1, 8 9
" 1, 8 9
" ! 1, 8
-= /3 /8: 82 3 9 :/ . ;28;3 : 82 ; )8 / /:> 9 2/:/
3 : 3 98 #
"! 7 *
"! / *
"!! * &-
"!" % &* !
- .;8 * 8.3 : #=
"" % !
"" % !
""! % % "
""" % "
"": % %, % :
""# :
4 '# # ' + '( + ************************************************************************************** 4
C-# 2; 3 I 8;7 #C
; ; 9 3 98 #F
C- #F
% ;
: ;
% % . %, ;
: ;
/ - < & & % &% ;
: ! ;
C-= ; ; 9J /8: #$
:! ' % (
 :! % % % (
3 '# # ' + '# )0& ******************************************************************************************* 5
F-# .;8 * 8.3 : % .8 : / /3.;8* 3 : @ ;/ ;/ #!
F- A 82 * 8.3 : 82 ; )8 / @ ;/. /8: / #!
# , 9
# ,
F-= .;8.8 .;8 : : A / 88 @ .; ;/. /8:
#! &- , % % %
< - "
#! &- & :
#!! + , ;
#!" & % (
F- /: ;8 /8: /: /: ;A : * / /8: @ ;/. /8: // !
#" ' &, % &- !
#" . - , % , %% % !
6 #,# )&%7# )22 '- ($ # # #' *****************************************""
"-# . . ; =
; !"
; . , !"
; ! . !:
"- . . ; ) =C
!:
; !:
; . , !#
; ! . !#
"-= . . ; # =F
;! !#
;! . , !;
;!! . !;
"- . . ; ="
;" !;
;" . , !(
;"! . !(
"-C . . ; = =$
;: !(
;: . , !(
;:! . !(
"-F . . ; =$
;# !9
;# . , !9
;#! . !9
"-" . . ; =!
;; !9
 ;; . , "
;;! . "
"-$ . . ; 2
;( "
;( . , "
; (! . "
5 ( 0) %( *****************************************************************************************************.

8 $)&)'# / (' ******************************************************************************************************."

9 '#$#'# # **********************************************************************************************************..

: $
 %

*
' '6 %' ' ( '
' ' % KI & ' ' 6! L-
' ' ( ' '
' - % ' ' %
' '6 ' ( 6 ' '6
' % ' ' ' % KI & ' ' 6! L- /
' % ' ' & 6& ( % '
0 ' ' ' & - / ('
% ' ( ' & '6
& '
K. ! L- I ' ' ' ' '(
( ' ' ' ' '
' -

*! 2
' 0 ' ' ' ' ' ' '
( ' % ' 4 ( '6 & % %
' ' - ' ' ' '
' ( ' ' -
( ' = ' 0 ' ' ' '&
' ' ' ' ' ' ' ' 0 -
& % ' ' ' ' ' ' ( '6 '
% & ' ' ' ' 6
- ' % & (
( & ' ' -

*" 7
' ' ( 0 ' '
' ' ' ( - ' ' '
' ' ( & -
' % 0 ' ' &' -
' & 4 ' -
' & ' ' ' ' ' -
' G & 4
( ' ' '' ' 6
4 ' & - ' 4 '
' ' ' ' - 7 & ' ' '
 ' ? ' ' '( ' ' '
' ' ' -

! ' ; <
& 6 ' ( '( ' & -
' ' ( ' ' . & * ' '
. ' -
' ' ' ' ' ( '
* ' >' ( % & - & & 6 &' '
* 3 ' % *3 ' ' % ' '&
( ' = &' - ' ' ' ( ' &
' 2 3 ' - ' & '
' ( ' ' 2 % > ' - ' ' & 6 &'
' ( ' &2 2 '- '
& &' ' ' ' ' ' ' (
4 &' 2 + 6 ' ' 2 & 6 ,
M K 2 L &' - ' ' ( ' ( : ; +&
; ' ', ' % & ' ' '(
' ' % ' ' - ' ' ( '
' '
% & ' % ' ' ' & ' '
& 6 ' ' '-

" &;
1, %, , & + , 2 %
1, & $ , % &, % & - , $- , ,
, % , - &%

' ' - '

( ' ' ' '- ' (
'( = ' % ' '
' 0 % & ' & ' '
KI' 6% 1 ' ; ! %' ) ' > ' L% 2 #
' 2 - ' ( ' ( '
'% ' (' '- ' & G ' '
' ' ' ' 0 ' ( % ' '
' ' (' % ' & ' ' ' '(
' ' ( ' ' '
' - ' ( % & & %' ' '
' '% - - & ' & ' ' ' ' '
' ' - ' ( ' ' ' ' % % ' %
' % ' ' - ' ( N ' ' & ' '
 ' 0 ' C OP & ' ( ' '
' ' ' ' ( ' ' -
' % & ' ' ' ' % 2 =-
' ' ' ' ( ' ( ' -
' % - - ' P '6 ( & 0' &
' ' % ' ' ' ' ' '
' ' ' ( ' -
' < ' 0 ' '
# % C% % + ,- ' ' '
( ' F " + ,- ( '
' ' ' ' ' ' ' &
' ' - ' ' '
' ' & ' ' %
' ' '
4 & ' ' ' -9 ' ' '
' - ( ' ' ( ' ' & '
' ' ' ' 0
( ' % (' <% % ' & '
( ' ' - ' ( '
' ' ( ' ' & & @' & ' '
- 8 ' ' ( ' ' % ' ' & % ' & %
% % ( % ' - ' (
' ' '( ' % % '
' '' %' ( % % % ' ' - ' (
' ' ' ' & ( ' (
' - ' ( ' ( ' '
' ' % ' '
4 ' ' ' '' % ' ' ' ' ' - 2 ' %
%& % ' ' % % & ' ' ' ' ' '
' ( - ' '?
' ' ' ' 4 ' ' '
K (Q 6 L- / ' ' ' ' 6
' - ( '( % '
' ' ' ' ( ' - '
' ( & (' ' ' %' '
( ( ( ' ' %' '
% ( ' ' -
' ' ' ' ' ( ' 0 ' # %+ ,% '
' ' ' 0 -2 %'
' ( % ' ' '
' ( - %
' ' ( (' ' '
 ' ( -2 0' %( ' ' ( '
&' ' '% ' ' ' ( (' -
' % ' ' ( ' ' '
' ( -9 (' ' 6
( ' ( ' ' '
' P 0 ' % ' '
' - ' ' ' 6% ' (' ( ' ' '
' ' ' ' ' (' -2 %' '
' ' 4 ' & '
( ' ' ( ' ' - / ' ' '
( % ' ' ' ' ( ' ' ' '
( ( ' 4 ' ' ' -
' ' ' ' %
' ' ' ' 6% ' & ' ' ;B ' (
' - .' ' ' 0 ' '6
(' % ' ' ' ' '
' ' ;B & ' ' -

* % $- , & && %
* + - & % $ + ! , %

* ! 6 , & 2 % % , + + ;
.
' 6 & ' ' '
' -

.* # ' ;
' ' ' ( ' ' & ( )
K) ' !$L@

1, & %, < - $
, -, %, % , %, % & % %% & %

' ' ' ' ' '

' ' ' ' & ' ' ' K; '
' 2' % 3' 4 ' ' .' =L- ' ' ' '
' ? ' ' ' ? ( & ' ' ''
K ' ' =% ' ' ' ' =% (L-

2 ' ' '( %

( ' ' ' ( ' ' ' &
' ( & ' ' ' ' % K3 ' !#L- '
' - % % '
( & ' % % % '
' ' - 9 ' ' ( '
6 & ' ' - ( &
' ' ( . ' -
R 3/ % ' ( 9 ' K9 ' =L '
&' ' ' - % % ' ' ' (
'& ' ' ' ' ' & ' %
& ' ' ( ' ' % '
&' -

/ '( % ' & 0 @ (? ? ' %& ' '

' ' ' ( - '
? ' ' ' ' % (
' ' ' %' ' ' '
' ' G ' ' ' ' - 3 ' ' 4 '
' ' ' K (Q 6 L- : &
4 % ' % ' ' ' ' '
' '( % ' ' ' ' ' & - '(
' ' ' ' 4 %' &
4 ' ' ' ( -
> '' % 4 0 P 0'
' ) ' ' K ) %
CL ' ' ' ' ' ' ' ' ' @

, % & $ % $ %
' % < % =& >$ -, %, , % % %
, % % % % %
0% , & &, % ,
, & 0% % $ , $ $ % % $

1, % & % % % & ,
& - & 8 & % $ & $
& , , $ & &
%&% $ % & $ &
% $ % % $ % % $
%
* , $ % & %% $ %, % %&% $
& $ $ , %$ , % % %

( ' ' % ' ' ' 0

' ' - &' % 6 &
' ' ' ' ' -)
' ' % ( ' ' ' (
' ' - 9 & % ' ' (?
' ( @ & ' '( ' ' ' K.' !FL%
(? & & ' ' '
' ( -

/ ' ' & ' ' ' (

% ( % - - ' ' % - -
( ' ' & ' ' @ ' ' ' (
% - - .. 8 % ' K L- 3 '
K ' ) 6 !$L ' ( '
& 0 ' ' ' ? -

3 ( ' ' ' '( % ' ( (' '

( ' ' - 9' ' ' ( (?
'' ' ' ' 0 % ' & ' ' ' & ' '
' ' & '6 % 6 4 '
' ( 4 -
.*!
' ' ' ' ' 6 & (
' ' ' ( - : ' 6'
K: ' 6' $"L ' ' @

& & = % , > , %

$ %$ % , , &,

( ' ' ' (

' ' % ' ( ' - ' '
' ' ' ' ( '
' -I ' ' ( ' ' ( % (
%' ' ' ' ( - & '
( ' % ' ' ' '
2 ' + 2 ,% 3 ( ' +3) , K ' #L% '
&' ' K7 ' #L% ) ' ' +) 3, ' '
' ' ' 2 3 K)'
!FL- ' % ' ' ' ' ' * % '
@

• ' ' 8
8 ' ' 4 ' ' ' -
• = ' '
' ' ' (' '' '6 % '
' ' - ' ' 0 0 C O
% ' ' -
• & ' ' '
' ' ' -
• #
' ' ' ' ' '( ( ' ? +- -
' ? ' ,
• : ' ' ' '

' ' +:*9% ' % % ' %

' % 2 ,% ' ' ' ' ' (
& %( ' 2 ' ' '
' ' ' ' -
 * " & % 2 & % , - ,
+ , && , + +

' ' ' '

2 - ' % 0 ' ' ' ' '( % '
' ( 0 ' ' ' ' &
' ' ' ' ( - 9 ' 0 '
( ' ' ' ' -9 & '
( ' ' ( % ' '% ' '
' -

.*!* &;
2 ( ' ' '
' ' -I 0 '( '
' ' ' ' - '
' ' ( ' 4 '
' ' '-/ ' ' '
' ' ( ' ' ' % '
' ' - / ' ' ' (
' ' ' ( ' (' & ' -

.*!*! &;
' ' ( ' ' ' &' - /
' ' ' '' & ' '
P ' ' % - - K) ' #L- 2
' ' ' ' ( ' &' ' (
' -2 ' ' ' ' ' '
( & ' ' ' ' ( ' '
+ ' ,% ' ' ' - ' ' %&
0 ' ' ' 'P
 ' ' ' K8
L ' ' ( - 2 ' % '
' ' ( & ' ' ' ' < ' &
( ' ' &' ' ( 4 ' @ ' & 6 ' %' '
- ' % 4 ' %( & ' '
' % ' '( & 6 ' -
( & ' ' ' ( (
% ( ( ' 0 '
4 ' ( 0- 2 0' % ( ' '
' ' ( % ' & '
' ' ' ( K; !#L% ' ' K2 =-=-#L-

.*!*" &;
' ' ( ' ' @ ' % ' ' ' ' -
. '( ' 0 ' 6 ' ' '
& & ( - ) 0 ' %
' ' ' ( -I '
' ( ' ( ' %' ' ' ' ( ' (
' ' ' -I ' ' ( ' 4 ' '
' % - - ' ' % ' ' ' '
& ( ' ' ' 6 ' ' '
' ' :*9 +: *(' ' 9' ,' ' ' -/ '
' % ' ' ' ( ( ' '
' ' '( ' ' -

.*" ; ; $
# 2 ;
I - -# - -=- % ' ' % - - 0
% ' ' ' ' ( ' %
' ( - ' & ' ( ' '
' ( ' ' ( - 2 3 % 2 3 '
& ' ' ' ' ( & (
' ' ' ' ' ( ' % 0 '
' %' & ' 0 ' -2 ' ( % & 2 3'
' ' ' ( P ' ( '
( -

.*"* 0 $#2
I ' ' ' 0 %
' ' ' % - - ' ' '
( ' % ( ' '- '
 & ( & ' ' 0 ' K3L% ' + ' , K L% K7L
' ' 2+, ' ( % 4 ' #-
2 ' 0 % ' 0 ' ( <
' % - - % & 4 ( ' ( '
K)' !F #""L- S ( ' -

Eq 1 : M ⋅ X + C ⋅ X + K ⋅ X = F (t )

4 ' # ( ' ' ' ' ' ( ' P ' '

& % - 4 ' ' ( ( '
( ' ' ' ( ' ' ' -
'' % ' ( ' & ' %
- - ' - ' ' ' ' '
+ ' 4 ,% 4 ' # ' ( %
4 ' - ' 4 ' ' '
' ' ( ' ' % K L-

Eq 2 : M ⋅ X + K ⋅ X = 0

4 ' = ( ' ' ' ' ( ' & 2 ' -

4 ' '( ' ' &
& ' -

Eq 3 : K ⋅ X = F

S ' ' ' < P

' '
6 & ' ' (
'' ' ( ' ( - '
' ( ' ( '
' K)' !F% $#" $#$L ' ' % '
.' K ' 2L ' 4 ' -
 ' 0 K 7L
2 ' ( ' ' @
K mm K ms Dm F
Eq 4 : T ⋅ = m
K ms K ss Ds Fs

K m = Reduced stiffness matrix

Dm = Master d.o.f.

Ds = Internal d.o.f.

Fm = Forces on the master nodes.

Fs = Forces on internal nodes, here equal to zero.

Reduction gives:

−1K T
K m = K mm − K ms K ss ms

.*"*! $#2
/ 0 ' 'N - - ' <
G' ' ' ' ' + ' , % (
'- ' ' ' ' %
' ' ' ' % - - ' ' '
( ' - ' ' ' ' ( K)' !F $FL-
/ ' ' ' ' ( % ' ' ' % ' ' '
' -

.*"*" $#
( ( -=-# ' -=- % 2 ' -
' ' ' ( -2 ' ' ( (
-=-#% ' ' (' >' ' K)' !F
F!FL- & ' ' &' ' ' ' ' (
K: ; : L% & ' ' * -2 ' ( % -=- % '
' ' ' - 0' &' ' ( '
K3 ; ' ; /8 L% & ; /8 ' * ' 3 ; '
-
.*"*. $#2=
% 2 ' ' % ' ( ' ' %
' ' % %' ' - '
( ' ' &' &
' ' ' -2 & ' '
' & % ' ' &' ' '
-

.*.
. ( K ' !CL ' @

1, &% $ - , , % & <

, % $ $ & %

( ' ' & ( ' - ' '

' ' ( K; < ( ' 6 !CL-
& 6 ' ' ( ' -

.*.*
2 ( % ' (
' ' ' ' ' - ' K ' !"L ( '
%& ' ' &
@ ' % ' ' ' ' ' - ' '
' - ' ( ' 6
' ' ' ' %
- - && % ' %' ' ' ( ' ' -
' ' ' ' ' -
; ' ' ' ' ( '
' ' - &' ' *
* $C P K.' %
C-C- L

.*.*!
I ' & ( ' ' ' (
( ' ( - ' ' ' '
% ' ' ' P? & '
'' % %' %K !!L@

% % & % , % & & $ %,

% -, %, & -, %, , , && ,
? -
 ' ' ' ' ' ' '
-

.*.*"
I ' ' % ' ( '
4 ' ( - ' - 3'
0 ' ' ' 0 (' % ' '
' ' ' 0% K/ '6 % ' ' ! L ' K. ! L-
& 4 ' ' ' %'
( K ' !CL- ; < ( ' 6 '
' 0 ' ' ' ' K; < ( '
6 !CL- ' ' ' ' '
? ' ' ' ' ' (
' ' ( ' ' ' ' ' '
' ' ' ( %K !!L-

.*.*.
8 ' G (' ' ' ( ' ' ' '
' 0 ' ' K '
!CL- ' &' '' -

' % ' ( ' &'

' % ' (' ' -K !!L-
' % 9 (6' '% K9 (6'% ' ' $$L ' -
' ' % ' ' '
- 8 ' ' ' ' % 0 &
' -/ & % ' '
% ' ' ' ( - ' '
' ' % & ' ' ' '
( -/ % ' ' ' (
' ( -

' ' ' % '

' ' ( = ' K. %/
' / L- ' ( ' % 6
( & - ' ' ('
' ' ' ' ' ( %
% ' % :*9% ' ' -
' (' ' ' ' (' ' ' (
2 ' ' - * 0 ' ( ' ' &
 %( ' '6 ' ' ' ' &
'6 -

.*.*4 ;
' ( & ' K9 (6'% ' ' $$L 6
K ' !CL ' - ' &
' ' ' ' ( -K '
!CL ' ' ' (
' ' - ' ' '
' ' ( ' ' ' -
' ' ' K) ' > ' '
> ' ' ) L-

.*.*3
' ' ( '
- I '0 ' '( ' '
% ' ' ' ' - '
' ' '6 ' ? ' ' -
2 ' & 4
' '6 ' ' -
( ' % &' ' '
% - - / %/ -

4 ' ; ;
' , % , %, & - < %, , , , , 0%

4* $
' ' ' ' %)
' ' ' % ' ' & 6 ' %
K) % '6 '( ' I' ' !$L- ' @
'% /% . ' //- 2 C-
 * : . %, & - <$
' ' ' P ( ' ' % 'P
' % P ' ' ' '
& ' ' % ''- %
' % ' ( ' % ' & '
' ' '( '- ' ' ' '
' ' 4 ' '( ( ' ' 6
' % ' ' ' K 6 % '6 ' ' L-
' ( @ ? %' &'
// ' ( - ' & 6 2 C ' P
' ' ( '6 ' ' (
' ' & 6-

4*! ' ; ;
' ' ' '
( &
' % % % ' ' ' %
( ' ( - ' ' ' ' ' '
& & ' ' ' K) L- .
' ' ' (' 6 '
' ' - ' '
' ( ' ' &'
? K ' #' 2L ( ( ' + ,%
+ ' ' ,% + 0 '
,% +' ' '( ' ' &' , '
' ' + ' ' ' ' / , ' ' ( -
' ' '
' K) !$L- 6 &
 + ' ,% ( ' ' ' '
( - ' ( & ' ' 0( &'
? & ' & 6 2 C-

4*!*
) ' '
' - %' ' ' ' '
- & -# ' ' ' '
' '% ' ' - .' ' +. ,
K) ( ' !=L ' ' ( & 4 '
+ & 6& ( ' , ' ' ' & ' '
' ' - 6 ' % ' . ( '6 '
' ' ' ' 6 & ' '
' ( ' % ' G &' - / ' '
% ' . ' ' ( ' '
% - - & & 6% ' ' &'
' ' ( ' ' * - ' ' (
'' ' ' % 2 C-

4*!*! ' ;
' ' 0 ' ' ' %
& ' ' &' % ' .' '
; ' ' (%. ; ' '
KI ' & !$L- '6
+ , & ' ( ' (
& &' ' ( ' ' ' ' '
&' ' ( - % ' %
& % &' % ' ' 0 % ' ' '
'& ' ( ' % K ' 2L- ' (
' ' ' '6 %
' ' '( ' ( &' - '(
' & %' ' ' ' ' ' % ' ' ' '
' K ' 2L% ' 2 C-

4*!*"
' ' ' ' '
& % - ' %
' ' ' ' ' & '
' - ' ' & % ' ' 0 '
' ' ' ' N ' 6 & -
& 6 ' K) ' ! %) L ' ( ( P &
6 ( & ' '% & ' (
 ( ' % ' ' ' '( ' ' ' '
' '( % ' & ' ' ' ? %( ' '
( ( ' ' 6% 2 ##- I ' ' % '
' ( &' ( KA ! L & ' ' -
//% 2 C% ' ( ' ( ' ' ('
' - / ' K2L & 6 ' 2 ##
' ( ' ' ' ' ' &'
' * ' ' ' '
( F- - -

4*" ' ;>

9 & ' ( &
' ' ' T

4*"* %
( ' ' ' ' ' ' ' '
% & ' '
0 ' - ' '
' ' & ' ' ' % ' (
0 ' ' - ' ' ' ' '
' ' -. &
' ' ' ' ( '
-

4*"*!
/ ' ' 6 & ' (
' '( & ' ' ( ' ' &
' 6 & ' ' ' ' - '
' ( ' ' 6 ' U %
& ' ' ' '-2 ' % 0 '
( ' & ' ' &
' ' ' N -

U ' ( & ' ' ' '

' % - - ' ' ' ' ' -

3 ' ;
' , % , &, - < $ , % +
1, % % %% , & - < * :$ % $ % '$
% % ''
3* :
&, % - % , &, 0%@ , , %
, % < -

0 ' 2@
. ( ' ' ' K
! L- . ' ( ' ' ' & 6 ( '
+- - ' ', ' ' KI ' 6 ' ! L- %
' ' ' ' ' ' -
' ' ( +( ' & %(
% & & %' & , ' - ' ' '
' ' ' 4 ' '( ( ' '
6 ' % ' ' ' K 6 % '6 '
' L-

3*! %
- , , ? $ , % , & - , < - ,
@ % $, - $, % ,

' ( % & ' 0 ' ' %

' ' ' - ' ' ? %
' ' (' ' ' 4 ' ( '
' K =L% ( % % ' & % '( %
:*9 ' & - 3' ' ' 4 ' '6
' - ' ' ' ' 4 '
' ' -; ' 4 %
0 ' ' '6 %' ' ' ( 0
' ' ' '( ' ( '6 ' &
' ' 4 -/ ( ' ' ' ' ' '
4 - 4 ' 0 ( ' '6
' ' ' ' & ' ' ( &
4 - ' 4 ' 0 (
4 ' ' % ' % ' ' '
' -

3*!* # = ;
' 4 ' (
' ' ( - ' %' % ' ' 4
' ' ( 4 ' ' & ' ( 6 ( ' ' - '
' 4 ' ' ' 4 ' % ' & '
% ' ' & &' '6' -
2 ' (? 4 ' ' 4
( & 6 6 % & ( ' ( ?
( & ' ' - ' (' ' %
0 4 & ' & ' ( -
.' 6' 4 ' ' ' ' % ' ( (
' ' ' ' ' & '
'- ' & ' ( % ' '
' % ' ( ' 6 &
4 - ' % % ? '
' % ' % - -# ' K.' L- * '
' ' '
-

3 ' ' & 6 &

+ , % ' (' ' 23 +2'
3 ' ,- '& ' ' ' ''
' ( '( - ' ' (
' ' ( ( % ( '( 4 ' ' '' 0 -
: ( '6 6 0' ' ' - . ' 0
( %K ' L- / ' ' (
& ' 6 & ' '
0 % K.' L- ( ' ' '6
6 0' ' . ' 0 ' ' ('
' ' -2 ' ' ' ' &
' ( ' &
' ( ' ' ( ' - % ' % '
' ' ' '' ( ' ' '
&' ' - ' % %
' ' ' ' ' '6
' -8 ' % ' ' ( ' & &
' - ' %& ' & 0 +F
'' 23,% 0 ' ' '
' 2 3 ' 2 % K) ' =L- / %
' ' ? ( ( ' ' '
&' & 6 ' %K ' L-

3*!*! ;
I ' ' ' ( ? %' ( 0 '
' ' ' ' ( '
' ( -/ ' ( ' '
' ' &' ' 2 3' ' &' @ ' * % / %
: ; :' ; /8 % ' ' 2 %/ %: ; :
' ; /8 - ' ( ' '
' (' & - ) '6 ' ' ' ( ' %
( 4 ' - I
' & 4 ' ( &
- (' ' ' 4 ' & %(
(' & 4 ' % ' '
' % .' )-

; ' ' % ' & ' (

' % ' ' ( ' &
0 - 4
% % '( %
' & ' ' ' - & ' '
' ( ' & ' 6
' ' ' ' 4 K.' L- % '
& ' (' 6 ' ' ' K ' ! L% ' ' 6 &
' ' ' &' ' : ; : ' ; /8 -
2 % ' & ( 6 &
4 ( ' ' ' ' '- '
' ( N ' ' ( ' 4 &
' ' ' '( ' ( ' '
' ' ' ' - ' ' ' (
2 F% ' - & 6&
' 4 ' ' '( ' ' - ' '
& ' ( ' ( ' ' 6 ' '
-
 Global mechanical
requirements

Component
design engineer

Assembled complete shell model

FEM analysis of complete shell model ,

Check if requirements are fulfilled.

Ok Not ok, change design

* # 1 %

& ' ( -

• ' ( ' ' %

' -
• > ' % 0 -
• / ' % - - ' ' ' ' '
' %' ' -
• 3 ' ' 4 ' ( 6 & '
' ' ' ' - '6 ' ' '
' % & 6' ' (' 4 '
' ( - ' ' '
' -
• ' ' 4 ' 0
' ' ' -
• 2 ' ( ( ' ' ' 0 -
• 7 & ' ( ( '6 4 ' ' -
• ' ' ' '
&' -

3*"
' , % % , - ,
1, % % & % & ,
% & % #
2 " & ' & 6 ' (? ' -/ ' (
' ' ' ? ' 4 + ' % '
% -, K2 !FL- . )' 3 +.)3,% K.' L% (
' %' ' ' ' ' ' (
- .)3 ' G ( K9 (6' ' $$L%
& 4 ' ' ' - ( '
( ' ( ' % ? ' ' % K.' L- / ( ' ( '6 &
' % .)3 & ' N % K ' L
' ' ' ' ' - ' ? ' & ( % ' '
0 6 & (' 6- ' (' 6 ' 0 P 6 &
' ' ' ' '
' ' ' ? K.' )L- '
' ' ' .)3 & 6 ('
4 % - - (' %& % ' & % - ' '
' ' ' ' ' ' - 2 0' % ' &
' ? &' ' ' & '
& 6 & ' '0 ' -
' ' (' ' ' -
.)3 ' ( 4 ' ' ' 4 ' ' ' -
'6 & ' ' '& %'
' ' ' ' ' ( ' '
' ' -

.)3 ' 4 ' ( -

' ' -
' ' % & 4 '
' '- ' ( ' ' ' % K.'
#% ' 2L 6 G ' '%
2 $- / ' % G ' ' ' '
6 & (' 6 - ' ' 4 ' ' '
' ( (? ( ' - ' ' (
' ' 4 + ' ' & , '
4 ' ' % ' ' ( - 2 0' % '
' ' '6 '& ' '? & ' ?
6 ' ? ' % ' ' '( -

) 6 .)3% ' '' & ' ' ( %

' ' ' (' 4 ' ( ' ' -/ %
0' % ' ' ' ' ' ' '0 ' '
' ' ' - 2 ' ''
 & ( ' ' ( ' '( ' '
' 4 % ' K.' 2L-

* ; * - <

3*"* ;
=
0 ( .)3 ' ' ' %
' '' &
( ' ' % ' 2 KD ' L
; K ' ' ' . #L- ( '
' ' & 0(
& ? (' '
' ( K2 L- 8 '
.)3 '
' % ' ' ' - ' ' ' '
' & 4 ' ' '
' ' K3 ' ' #L- ' ' ' 2 ' 2
% ' ' * - 2 & '6
'' ' P ' ( '
' ' &' ' : ; : ' ; /8 % ' ' ' * -
2 ' ' ' &
' ' ( ' ' ' -/ ( ( '6 & ('
4 ' 4 2 % & ;/ :
' 3/ -
 Global mechanical
requirements
If the requirements in
the PBM can not be
Fulfilled, rebalancing of
PBM (Requirements requirements are needed
at sub-system level)

Preliminary analysis by
design engineers
using ADRIAN and DAMIDA
Component
design engineer

Assembled complete shell model

FEM analysis of complete shell model,

check if requirements are fulfilled.

Ok Not ok, change design

3*"*! =
' % ( '( ' '
' G '
' ' & '
4 .)3%
2 $- 2 % ' ' ' ' '( & ' '
' ( 4 ' % - -' ' ' ' -I
&' ( & % ' '
' ' ' ' ' ( ' & P ' ' '
( &- / ' K # ' 2L% ' '
' ' &' ' ' '( P ( & & ' '
% ' 2 !' 2 # -

!"# $%
' ' '6 %( ' ( 4
' ' &' - ' ' &' % ' (
'6 ' ( - ' ' ' ' '
' ' ' ' ' ' ' - /
;/ : ' 3/ ? % & &
' ' 0 -
 • ' @ %( ' ' ' '

• / %( ' ' 0 '

• '& ' '

' ' ( 4 ' &

% & ' '6 % ' (
&' % ' C- -# ' #-#-#-

!"" % &'
' ' ' ' ' '
' - ' ' ' ' '
& &' % ' '
' ' ( &' % ' ' ' ( % '
' ' '-/ ' ' ( &'
' '6 ' ' &' - '6 &'
'% ' ' ' ' /> .% ' ' ' ' 6
K: ; : L ( - ' ' %' '
' ' 6 -

!"! (
I ' % '
V ' '
( - ' ' ( '
& ' ' ( - ' % +;8/, (
- ;8/ ' ' (
K) ' =L- &' ? ' (
( % ' ' ;8/- / ' % ' &'
? ' ( ' 0
' ( K ' ' ' . #L- ( 4 '
-

!") *' % & &' % $ ' %'

I ' ' &' ' ' '
' % ' ( & &' ' 6 & &
' -9 &' & ( ' ' ' -J
' ' ' ( '
- &' ' ' (
( ' % ' ' ' ' ' '
( - ' &' ' '
' '( ' ( ' ' ' -
 !"+ %
) '6 & 4 ' ' ' ( - '
' ' ' ' % ' ( ' '
' ' ' ' &' ' ( ' ' ? -8 ' ' (
'6 ' ' ' 4 ' ' %& &
' %& ' :*9-

* 9 A - & &-

3*"*" #< ;
I ' ' ? % ' ' - - ( '6 % '
' '( - / F-=% 4 ( '6 & (' '
' ' ' ' - /
? & 0 ' ' ( - ' (
' (' ' ' ( '
' ' % K.' =L- ' '
% 0' % ' ' & '
' ( ' ' ( & - ' '
( '&' ( * - ' ( ( '6
N? - 0 ( ' ' ' '
' ' ' ( - 0 ' '6 ( ' '
( ' %' ' & ( ' -
 '' ? ( ' '
' % K.' L- / 0 ' ' *
' & ( ' 0 ' 6 ' ' - ( ' '
' ' ' '
' ( ' ' ' -

3*"*.
• ' ( ' ' ' ( '
(' +.)3, ' ( ' ' '
-
• ' & ' ' '6 ( ' 6
% - - N ' ' ( 4 ' -
• ( '6 & 4 ' '
' 4 ' '-
• ( ' ' ' 6
' 4 - ' '
' ' ( % ' ' ' '' G
' -
• ' '( ' %
'' & ' % ( ' ' ( ' '
' (' 4 ' ( ' ' '
' ' - '6 ' ' '
' '( - / ' ( ' ' ( ' '
' 4 %' ' ' ' -
• )' 6 ' ( ' % ' ' '
' ' -
• I ' ' ' &' % ' '? '6
'6 ' % ' ( -
• 3' ' ' & ( ' ' -
• 0 ' ' ' ' (
( '6 -
 2% ) 3 '% 18/:

2
' / ' /

3/ 0 ? ;/ :

)
3 ' 0 3/ & 3 :
(
' ; /8
2 ' : ; :

&
%

; 9 3 ' '
; 9 3 ' '
:/3 8;
:/3 8;

' & ' ' ' .)3

6 ' ' ' ' .

* * - %, & &-

3*. % %%
1, % % , , &, &,
% , &-

' ' ( & '

' * - ' ' ( ' 6 '
 ' ' ' '
' - / & ' ( ' ' ' ( '
KA ! L ' ( ' ' ' '' -
.' ' ' ( '6
&' K) ( ' !=L- ' ' (
' % - - ( ' (
' ' ' ( ' ' & ( ' N
-

3*.* % ;

2 $& &' % 3/ '

;/ :% ' (
( ' ' ' -
' ( 6 ' ( & ( - ' ' (
' ' ' ( ' ' '
' - 3' ' ' ' 0 ' ( (
' ' - ' (
' ' ' ( ' - ( '6 & '
' ' 6 & (' 6 % - - ( 6 ' '
'( ( ' ' ' ' ' ' ' ' ' -
' ( 6% ? ( 6 % &
& ? % ' K.' L-

' '' ' ' '

&' 2 - ' '&' ( '
' ' ' 4 ' % ' - '
&' ' ' ' ' - '
' ' % & ' '
( % 2 # % ' ( ' ' %' ( ( &-

3*.*! ' ; ;
1, %% % $ # $ - , , 1,
&& &, % % - 1, + , % < &
* ,

' ' ' ' ' ' ( % - -

% ' ' ' (
' ' % ' ( ' ' - / F-#
' % ' ' & 6
2 " '
2 $& ' ' ' ' ( - ' '
( ( 6 & ' 2 ##- ( 6 & '' '
' ' ' '' ( C- -=% ' 6 '
'' ' ' 6 & ' ( &- ' ( %
, ' ' &' % ' ' ' P
( ' & ' - 0
' @& ' & -I ( '6 & ' % *
F% ' 0 ' ( & ' ' '
' '( 0& 6 - / '
' ' ' ( ' ' < % & '6 ' ' & 6
( -I ' % ' 0
2 ' ' ' ' '( ' ' ( ' &
& 3/ - '6 '(
- ' < '? ' ' ' ' <'
' ( ' & ' 2 ' ' # #C & ;/ :-
. ' ? ' ' ;/ : '6
' 0 ' # -2 % ' ' ' '
' P 2 "' !% .' 2- ' ' ' ' ' < %
- - ' <% ' & ' ( ' -
&' % ( ' ' ' '
( ' % ( ( ' ' ? -
: ' ' ( ' ' ( ' ' %
' ' ? ' ( ' ' < & ' ' -I &' %
' ( & ' ' ' '6 ' '
- ( 0 ' ' ' -)
&' . ' * % & ' ( &
' ' ' ' - ' ' '
' ' ( ' ( ' ' ( -1 ' (
' & ' ' ' '- 2 ##% &
& ( ' &' '
' ' ( -/ ' ' ( '
' 0 ' ' &' ' ' ' &
' - '( ( '
' ' ' ( ' ( ' ' %
2 $- ' 4 ' ' &
P ' '& ' % '
&' -
 )
' ' ( ' ( -
-

W
W

2 &
( ' ' -

3 ' '
' ' < +'
( ,-

2' ' ' +'

( ,-

G ' '(
' ' ( -

## ' $& & &

 '( ( ' ' ' ' ( '
( ' ' % 2 $-

• I ' ' %
' ' ' ( ' % ' (
• I ' ' ' %' ' ' (
'' -
• ' ' ' '& ' ' ' '
' '
• 2 ' ' ' ('
4 ' ( 6 & ' ( -

6 #<
1, % % & , % & %, ,
, ,

' % % )% #% % =% % ' 2 ' ( '

- ' %
$ ' ' & (' '
' ( - ' % % & % +
%, % % $ - , % % %
( . )' 3 +.)3, %&
0' %' ' ' (
- ' % .' / & , && &
0 % , % % ( ;/ :% '
' ' &' ( ( %' '
' - ' ' %* % % &&
& ) * , & 1 && &
2,% $ ( 0 ' ' ?
' 0 ' % &, &
, % $ ( '
' ' '
- ' % , % %%
3 - - , , $ '
' 7 & '( & ' -
' % / $ & &- &
% - '? & 6 ' (
' ' ' ' & ' '
-
6*
B7 / $ / / .'3 A/$

! "#$ % ' ( ) $ $* %

6**
/ ' ' ' ' %' ' '
' ' ' % ( '
' ' - ( 6 '
& ' ' .)3
'
' ( ' - / .)3% ' ' (' '
(' 4 - ' '
' % - - ? % ( ' ' % '
' ' +2 , ' % ( '
' - / ' ' ' %
' .)3 ' 4 ' -
% ' ' ' ' ' ' ( '
' ' & ' ( .)3-
'6 ' ' % ( '
- ' '
( & ' ' ' 0 -

' ' ' ' ' ' ( %

' ' ' ' ' @ '
' ' 2 3 & 6 ( '
' - & % ' (' 6' ' '
' ' % ' &( & 6 - ' ' '
' ' 6
' ' ' ' 4 ' ( .)3% ( '
' ' ' - % ' '
( ' '
-) ' .)3% ' ( ' '4 6 (
% & ' ' -

6**! ' ;
.' ' ' . )' 3 % .)3 ( '6 &
4 - .)3 ' '( ' ' -
' .)3 ( 6 & 4 ' ' '6
( '6 ' ' ' (
-
6**" '
' ' ' ' '
( & ' ' ' ' %' ' ' '
' &' ' ' '( - ' % ' ' %&
% ' ' ' ( - )
( '6 & 4 4 ( '
' ' ' ' ' ' ( -

6*!
)A : % :-% 2; ;/ 8:% 9- : 983. 8:% >-
% & % + %, % %
$- , % % ' , % &, & %$
" ; $ <$

6*!*
0 ' ' (? '
' ' & ' 4 ' 4 ' ' % 4 ' ' %
(? ' (? - 0' ' ' ( %&
' ' ' & ' '( 0 ' ( '& -
9 & % 0' ' ' ( ' - (
' ' ' & 0 ' -)
' ' (' ' ' (' '
' %' (? ' ' ' ( ' P ( '(
& - '' ' ' ' '
' % ' ' ( U-

' ( '6 & ( ' ' '6

' ' & 4 ' ' ' - '
( '6 & ' ( + ( , ' ' ' '
' ' ' ' - 4 ' '
' ' 4 ' ' ' ? P '(
( - J ' ' ' ' (
-

' ' ' ' ' &

( - / & ' ' '
' ' % ' (' ' ( ' (
' - &' '6 ' -

U ' ' @ / ' & @

, - %, % %,
 / ' &' ' ' ( ' ' ' '
( % * F% ' ' ' -
&' % ' % ( ' ' ' ' ' (
%
* $- 9 '' ' ' ( % '
' ' ( -

6*!*! ' ;
' % & ' % ' ' 0 & 6
? ' - ' & 6 * " &
' ' ' ' ( & ' ' 0' -

6*!*" '
.' ) & 4 ' (
& '' - '@ (? % (? %
4 ' ' ' 4 ' ' & 0 '
- ' ( & ' ' %(
.)3 4 ' ' ' ? -
' ( ' -
( ' -

6*"
B7 / $ /
.' / & , && &0 % , %
%
( ' 1 ' / ' /
%1/ -

6*"*
' ' ' ' ( ' '
( & ' ' ' -/ ' ( & ' '
' % '6
( & ' ' ' & '
- ' ' )% (' '
. )' 3 % .)3- ' ( ' ' '
( ? & .)3 6 ' ' '
.)3V ' G ( % ? - ( (
' ' -

3 ' ' ' & +- -

' ' , ' ' -
0 & ' ' ( & ' '
' ' ' ' '6 ' ' ' ' -
 ( ' ' ?
% K.' L ' - ' &
' ? '6 ' 4 - 2 %
' ? ' ' 0 ' ' '
4 ' ' ( '6 V? -

? ' ' 9 3 ' ' &

' 4 %' & :/3 8; & & '6' '
-2 % ' ' ' 9 3 ' ' '
' -

6*"*! ' ;
.' # ' ' ' ' (
' ' ' )- ' '
&' % ' 2-

6*"*" '
/ ( ' ' &' ( ( &
6 & ' ' ( ( &' '
' ' ' ( ' -
' ( ' P ' ?
'( & '6' ' -

6*. !
)A : % :-
* % % && & )
' =" C% / ' ' )
% /) =% " ! 8 ( =% ('% 1' ' -

6*.*
' ( ' ? ( & ( ' ( ' ' ' '
(' -I ' % (
0 ' ' ? ( ( ' ' -
'( ' ' ' ( 0 ' ? ( & '
' ' ' '- I '
? % ' ' 0 0 ( ( ' -
% ' ' ' ( ' ' ( ' - '
? ' % %( '
' ' ' & - 2 % '
' ( ' ( 2 3& ' ' 0 -
6*.*! ' ;
0 ' ( ' ' ' '
&' ' #- ' ' 6 ( ' 6
&' ' )-

6*.*" '
/ & ' ( '' ? & ' ' '
' - ' ( ( ? * ' '
' ' ' -

6*4 "
)A : % :- B 2; ;/ 8:% 9-
* , & 1 && & 2 ,%
' =" $ / ' ' )
% /) =% " ! 8 ( =% ('% 1' ' -

6*4*
' ( & ' (' ' ' '
6 - ' ' %' '
' '( 4 - ' ' ' ' '
' ' & ' (' & '( ' 4 ' ''
' ' - ' ' ' '
' ' ' ( ' ( ' ' %
' ' N '0 % ' <
( '6 - ' ' & ' 4
' ' ' '( ' & ' 0
'( & - 8 &' (' ' ( - (
' ' ' ' ' ' ' ' ( -

6*4*! ' ;
' ' ' ' ( '6 &' ' )-

6*4*" '
' ' '
' ' ' '
' '( ' - ' ( '
& -

6*3
8. D 3 % )-% )A : % :- B 983. 8:% >-
&, & , %
( 1 ' ; ' -
6*3*
(? ' ' &' '
' ' 0 ' ' % * '
' +* ,% ( ( ( ' '
% ( & ' ' -
'' 4 ' ' ' ' '
' * %& & ' & ' ' ' -

6*3*! ' ;
' ' & ' & 6
' )% ' ' -3
' ( ' ' ' - '
& ' ' '
' ' ' -

6*3*" '
& ' ' '
' ' ' 0( &' ' ' ' ' ' -
2 % < '
' ' % & ( -
' ' ( & &
' - ' ' (
' ' ' '' ' '( &
% ' ' -

6*6 #
)A : % :-% / 7 8:%- 7 98;/% *- B ; 8:% -
, % %% 3 - - ,
,
/ % #$ 3' % ( 6%
' '-

6*6*
(? ' & 7 & '( %7 %
& ( & ' '
% ' ' '
- ' & ' ' % - - & 6
% ' ' 7 ' ( '
+ ,( ' ' '0 ' ' -
/ & 6% ' ' ' ' '
' ' % & ' ' % ' * ' ' %
' ' ' % ' * ' % ?
' ' %( ' & -

( 7 ' ' '

' ' -. ' ( ' ' ' ' '
P ' ' +- - ' ( '
' ' ,' ' ' 0 ' 0 '
' 6% ( 7 ' -

6*6*! ' ;
.' 7 & '( % 7 %
' ' ' % ' ' ' ( ' #
' ' ' ' ' ' ' ) ' -
' &' & ' ? ' ' ' ' %
' ( & 6% !-

6*6*" '
' ' ' ' ( ( ' %
' ' ' ( ' & ' - '
' ' & ' &' ( ' '
' (
' ' % (' ' ( '
' - ' %, % $ , %$ % % %, &
7 -

6*5 $
)A : % :- B 7 ; 8:% -
/ $ & &- & % -

6*5*
' & ' ' ' '
' ' % ' ' '( &
-9 & % ( '
& ' ( 0 0 -
' (
&' ' ' ' & -
' ' ' ' ' ' *
' ' %* -- & ' % ' '
' ' % ' &' '
' % ' ( 4 ' ' % '
-
 ' & ' & ' ' ' - '
( ' ' -
' ' ' ' -I ' ' '
( ( 0 ' ' '6 ' (
G ' ( (' 6 '
' ' -/ ' ( ' ' ' ' (
G ' ' ( & '6 '
' % ' & & ' ' ( ' ' K ' L-
G ' '6 ' ' ' % &
& ' ( @ ;/ : ' 3/ %
' ' ' ( ? ' ' ( ( ' - ' (
' ' ' ' ' '
' ( ' * ' ' ( '
& 6 ' -

6*5*! ' ;
' ' '
' % .)3 ' ' ' ' )' ' #-
' ' & 6' ' '
? ' ' ' ' -

6*5*" '
/ & ' ' ' ' ' 4 ( '6 & '
& ' &' 4 ' (
'6 ' ' ' ' '
% ' ' ' ' ' -
'6 ' ' & &' ' ( ' '
( ' % ( ' ' ' ( ' '' (
- ' ( '6 &
& -

5
(? ' ' ( ' (
& ' ' ' % C-=-

' * ' & ' ( ' ' '

' ' ' &
' -/ ' ( ' & ' '
' ' ' P ' ' ' ' ' '
' ' '- > ' % ' ' ( '
 & & ' -
' ' ' '' ( ' '
' ' ' &' ' - 2 %
' '
' ' '6 ' -

? ' ' '

( & ' ' ' ' %' ' ' '
' &' ' ' '( - ' % ' ' &
% ' ' ' ( -
J ' ' 4 ' ' ' '
' ( '6 & ( 6 & ' '
%' ' ' ' -
' ' ' ' ' '' ' ' % ' '
' & ' ' -
' % ' % ' ( $ ' ' & 6 '
4 ( '6 & ' & ' ' ' ( '
&' ? - &' ' '
- (' +.)3,% ' ( ' ( '6
& 4 ( ' ' '
- ' ' ( ' ' '
' ' 4 ' ( @ ;/ : ? '
3/ ( ' - ' ' '6 '
' ' 2 ' ' ( '' '
- ' & ( ( ' ' 0 % (
' & ( ' ' '
' ( ( % 2 $ ' # - / &' %
' ( ' ' ' ' ' (
' ' % ' ( -

&
0 ' ' ( % ' ('
' ' % ( '6 '
- 0 ' ' ''
? ' '( ( '6 ' ' &
;/ :-

' ' (
& ' ( ' * -/
' ' ' ' '
'6 ' / ' % ' ' ' '
' ' ' ' ( -
 ' & ' ' ' (
& -. ' ( ' ' 6 &
& ' ' ' ' < 6-
6 & ' ( ( ' '
' ' ' ' ' - 3/ '
' ;/ : ' ( ( 6 % ' '
- 4 ' '
& ( P ' '
& ' ' ' &' -

? ' * ' & '

&' ' '( ' ' ' ' ' '
' & &' ' ' ' ' ( -
' ' ( ' '( ' '? ' ( ' & -/
' % ' ' ' ( '6 F
& 6 & ' - ( &
' ' ' ' ' ( ' ' % ' (
- 3' ' ' * ' ' ' '
' & 6' ' ' (
' ' ? - '( ' %
& ' ' ' '
&' ;/ : ' 3/ & ' ' -

8 $
, - ,
- 1, & &
% $ C D 1, %, , &
0 , - ,2

6 ' & @

' ' - ' ( ' ' '

' ' ' & ' '
0 6 & & ' ? -/ ' '
0( % ' ' ' ' '' & ' '
' ' ( - ( ' ' 4
' ' ' -

% ' ' ' - ' ' '

' ( ' ( ' ' ' -
 4 ' ' ' ' & 6 ' ' '
'' ' & 6- ' & (
( ' < ' -8 % 6 & ' (
( X Y $ O 0 & 6 ' ? -I
? ' ' ' ' O G
' ' ' 4 ' -/
' ' ' ' ' %
( ' ' ' % '
' ' - ' ( ' ' '
' ' ' ' ' ' '( ' + % ' '%
' ' % ' ',- ) ( 7 ' '
& ( ' ' ' ' '
-

' ' ' ' ' ' - & '

' ' 4 % '
0 ' ' ' ' %
' %. 3' ' -: & ' ' & ' '
' -3 & 6 ( ' '
' X Y ' 0 ' -
& ' 6 ' ' ' ' -

9'
) -% & %% - : & A 6@ ' ) '
' %/ % -

)' %7' 1Z -% * % %. 9' % : & 1 % #!!F-

) % - -3-% '6 '(' % - ' I ' ' % 7-3-%

%,% / G ; 3 I 6 % 8 %
7% #!! -

) % - -3-% '6 '(' % - ' I ' ' 7-3-% - & %

. E . %, , % CF " % F , < %% &
% % @2' 6 ( % -' )' 6 ' (% .-% * ' % #!!$-
CF

) % -% 6 , , , % . %,G$ / .
/ ' ' /;. ' % #F #$ 3' %9 7 -
) ( % 1-% > ' % 1-% 3 % - : & I ' % .-% , , %* ,
1, . / @ 9 ;% - : : 3/87 % -% -% .' '
%. ' .' - #!!=% 9 ' @ - (' ' % - # = #CC

) % :-% ) R % 1- B . % -% . 4 5 6 , %
%
/ ' / ' '
% C ## % # % ' -

) % :- B > ' % -$ %, % &&% +

/ :8; />: # #F % %: &' -

) % :-% > ' % -B < 3 '-% & , & %,

/ / ' ' %/ =% #! #
=% 6 % & -

/ % @GG -= - G# - - % -

' ' % -)- ' . % 3-% 1, % & < - %,

& % ' &' =
+ #, ! = !# % #-

6 % -% '6 % .-1- ' ' % 3-% 1, 7 &, . %,% /

% #$ # 3' % ( 6% ' '-

2 % 1-% 4 < &2 , % % 3 .% % #!!F-

2 % 1-% 1 %, 1 & & ' $1 '% 833 :/ /8: 82

9 3% ( #!!FG* - =!% : - !-

2 % 9-% 1 A 5 &2 , % % $ , $ 1,
% -. % 6R % & -

> ' % - B ) % :-$ &, &&% & && %, % ,

%
/ % # #" 3' % ( 6% ' '-

9 (6'% *-% ' % 3- ' % I -@ %% % %

) & % % 7% #!$$-

9 ' % )-% * & % % . 9' % : & 1 %

% =- F-

/ % @GG&&&- - G% -
/ '6 % - >-% ' % 7-1- ' % -1-% 1 +& % % '
. ( > % % #!! -

7 ' % .-% ' % , % %

& '% &&% % ' $. % '
. % 3' ' ' % '
% & % #-

'@ ' % -% R % .-% 7' % -% 3'( ?% -% % -% ' % -% '

R % ) -% 1 ' * - <& % %
/ / =% #! # 3' =% 6 % & -

(@ ' % -% > ' % - ' ) % :-% / ' '( ' @

'% , ' % & . %,
/ . " / ' ' ; ' G 0 H
3' ' ' H% #C #F ( =%
3' % ' -

(Q 6% 3-% , .H ' % & * I %?

&2 %. % -

3 ; % @GG&&&- &' - G G [ ' - T./\#% -

3 '% :- ' -% & % 4 1 % % 1 %<8 1 %

$ ' 2 # # "!=% #-

3 ' % - -% 1, % % & 1 %, % . 8 % &E

$ . % ' 3 ' ' % '
% % % #!!#-

: ; :% @GG&&&- &' - G G [ ' - T./\" % -

: ' 6' % 2-% %1 I B % #!$"-

8 ( % 3-% / &&% $* *
& % %. % . ' 3 ' ' %
. % % & % -

.' % *-% IJ% ? % ' ' %; % #$!F-

. ' ' % )-% % * 8' % % A 5 %

* #% : & 1 @. 9' . ;% #!!F-

. % -% 1 8' , & %% & % % I 6 ' @

I % #!! -
. % @GG&&&- - G' G G G 'G #[ -? T \ #B
&\ B6 & \= % -

; /8 % @GG&&&- ' - G G ' - % -

; % 1-% % & ,% ' I'

( % ' ' <> ' % ' ' % #!!#-

; < ( % :-2-3- ' 6 % 1-% % 8* , %1 I

' % % ' % #!!C-

2 - @GG #- 0 - ( - G G - % 2 % > (9%

* ' ' C% #==CC% ) %> ' -

% ; ' ) 6% .-% % - , % + % . 9' %

#!!$-

% >-% ' . , , % . '

. ( % 7% #!!!-

' % -> $ 1, %, % % % % 3 > '& 9 % #!!"-

% 7- - ' % - -% % % % /&
3 > '& 9 #!!C

A % ;-% . %, , $ ' . ( ' % '

8'6 % #!! -

I' & % A-% #!!$@ 6 , % % . %,G ; ' / ' '%

.' - ' '( @ @GG&&&- - -' G G G' G' G
&' & !$-

I & % - -' ' 6% 7- )-% . 5 % $K 7

$ &&% % K % 2 . % #!! -

I ' 6% 1-.-% 1 % - -' ; % -% #!! -% 1, %, , %, , - 81,

&7 % - : & A 6@ ;'& ' P % 3' ' % ' ' '%
: & A 6% 80 % ' % @ 3'0& 3' ' / ' '-

D % 9- ' -% & * A/ 1 * - , %
/ ' ' ) % 3 ' % 8 ( = C% -
' ' # " F
 Paper A
Simulation Driven Car Body Development Using Property Based Models
In the proceedings of the International Body Engineering Conference, IBEC 2001, 8-12 of
July 2002, Paris, France.

Simulation Driven Car Body Development Using
Property Based Models
Nicklas Bylund*♣ and Magnus Eriksson*
*Department of Mechanical Engineering, Luleå University of Technology
ABSTRACT
A method for the development of car
bodies, from conceptual to detailed
design, is presented. The conceptual
design is broken down to a numerical
property-based model (PBM)
representing the mechanical behavior of
the concept. In the PBM, the local
properties are balanced to fulfil the global
stiffness requirements. The main
topology is defined and the structural
components i.e. joints, beams and sheets
are connected in predefined nodes and
represented in a finite element (FE)
model as super elements, beam
elements and thin shell elements. In the
realisation of the car structure, the
performance of the PBM components are
used as requirements in the detailed
design. Different technologies, materials
and manufacturing processes can be
considered as long as the properties of
the component agree with the ones
stated by the PBM. The detailed design
of each component is made by design
engineers, supported by single purpose
tools. The design engineers iterate the
design until only a small difference
between target and component
performance exists.
INTRODUCTION
The design of modern vehicle structures
is driven by many competing objectives
such as improved safety and fuel
efficiency, lower cost, high performance
1
2001-01-3046
♣
Volvo Car Corporation
Copyright © 2000 Society of Automotive Engineers, Inc
and increased style flexibility. In addition,
the introduction of new manufacturing
processes and materials enables more
design options. Still lead time has to be
reduced continuously. In order to speed
up the design process tools are needed
to support the design engineer when
evaluating component performance. This
should be done before the CAD-parts are
assembled to a complete vehicle model.
Traditionally, the design engineer has
limited possibility to evaluate his design
with respect to the structural
requirements. Although computer tools
are being extensively used, they are still
advanced and analysis experts are
needed. Furthermore these tools are
mainly used in the conceptual design
phase and in the final analysis of the
complete shell model. Between these
phases, in the detailed design of
components where most development
time is spent, there is currently only a
limited amount of easy to use mechanical
simulation support.
RELATED WORK
In the field of car development a number
of efforts have been made to develop
tools for simplified simulations.
One example is AURORA later named
SFE-concept which is a commercial
software for development of simplified
parametric shell models. It started as a
project at the technical university of Berlin
in the late eighties[1]. The models
developed by SFE-concept have shown
good agreement with traditionally
developed FEM-models[2]. Although
made for simplified simulation SFE-
concept needs a user experienced with
FEM analysis. With SFE-concept a
simplified shell model can be made and
analysed quickly, however a detailed
CAD model has still to be done.
SimMod is another tool made for
supporting car body design and is
developed by Ford Motor Company in
collaboration with the University of
Virginia[3]. SimMod is similar to SFE-
concept but is company specific. Also
SimMod is recommended to be used by
engineers experienced with FEM.
Both programs use Nastran for linear
elastic analysis. SFE-concept also
supports plastic analysis using LS-
Dyna[4].
In addition to these programs there also
exist programs for analysis at component
level such as Crashcad [5] and DAMIDA
[6]. These programs can be used by
design engineers without experience of
FEM. They give quick approximate
results valuable for a first evaluation at
component level.
In the field of optimisation of car bodies a
lot of work has been done. Both
topological and shape optimisation
methods have been developed and
introduced as described by [7, 8].
EIx,EIy,EIz
and proposed
section
Sheet
thickness
Figure1. The PBM with requirements for a sheet a beam and a joint.
2
THE PROPERTY BASED MODEL
The PBM can be seen as a drawing of
the mechanical and topological
requirements, see figure 1. The PBM can
be made using different FE-simulation
tools as for example the earlier cited
SFE-concept[2]. The element properties
are balanced, using optimisation [9] and
experience until the PBM fulfils the global
requirements. Global requirements are
thereby broken down to component level.
Together with design, manufacturing and
other requirements the PBM properties
will act as a guide for component design.
By updating the PBM continuously during
the detail design of the shell model, the
PBM will reflect the structural behavior
and will be an essential tool to maintain
the design intent. The updated PBM will
also perform as a quick and efficient tool
for evaluating the effects of late design
changes in the development process.
This method will also enable efficient
evaluation of different design concepts
independently of level of detail or
manufacturing technology.
Stiffness
matrix
Elements in the PBM superelement technique [11], see figures
2 and 3. In the superelement definition of
• Beams are represented by beam the joints, a minimum length of the legs
elements needs to be included. This minimum
length is chosen to ensure that beam
• Joints are represented by super theory is valid through the interface to the
elements connecting the beams. This connecting beams [12].
representation is necessary due to the
elastic properties of the joints, which The stiffness matrix from the FE-model of the joint
component can be partitioned as:
have a significant influence on the
global structural stiffness as pointed K mm K ms
⋅
Dm
=
Fm
T
out in [10]. K ms K ss Ds Fs

K m = Reduced stiffness matrix

• Panels are represented by shell
elements Dm = Master d.o.f, here 6 d.o.f at the end of each leg.

Ds = Internal d.o.f.
APPROACH FOR EVALUATION OF
MECHANICAL PROPERTIES Fm = Forces on the master nodes.

The General equation for comparing the Fs = Forces on internal nodes, here equal to zero.
required and achieved stiffness. Reduction gives :

K requirement − K design = K difference −1K T

K m = K mm − K ms K ss ms

K m = K design
On a component level the Kdifference is
the difference between the required
stiffness, derived from the PBM, and the Figure 2. Superelement reduction.
calculated stiffness of the actual detailed
design.

ELEMENT DEFINITIONS

Beams

Depending on geometry and shape the

Euler-Bernoulli, or Timoschenko beam
theory is used. The stiffness of the beam Reduction
is described by the parameters EIy and
EIz, which are the principal moments of Kdesign
inertia times the young’s modulus of the
material. Torsion stiffness is descried by
the parameter GKx, which is the shear
modulus times the polar moment of
inertia. Figure 3. The FEM-model is
reduced to a compact, 18x18
Panels stiffness matrix.
Joints
For the panels the parameters are the
Different techniques to represent joint shell thickness and elastic modulus. For
stiffness, as for example by using elastic linear analyses these elements
springs, have been developed and tested can be large and still represent load
as described in [12, 13]. However, due to carrying panels such as the roof, floor or
the complexity of the joints, their windshields.
stiffness, K is here described by the

3
 TOOLS • When the complete model is
assembled the level of detail is high.
In order to use the method proposed
some single purpose simulations tools • It takes a long time from the design of
are needed. These should be easy and a component to results from analysis.
quick to use helping the design engineer
to fulfill the requirements from the PBM. • The assembly of the complete shell
Table 1 shows which tools that are model requires that all components
available and which that are under are ready at the same time and have
development. A linear dynamic tool is the same level of refinement.
considered only for sheets in order to
control bulls eye modes. For beams and SIMULATION DRIVEN DEVELOPMENT
joints no linear dynamic tools are needed, USING PBM
because the PBM will not specify
dynamic behavior for each part. When the PBM fulfills the global
requirements the performance of the
Component Beams Joints Sheets PBM-elements are used as requirements
loading for detailed design. To evaluate the
Linear static Ok, see Under Future design and compare with component
ref [5,6] developme work requirements the design engineer is
nt supported by single purpose simulation
Nonlinear Ok, see Future work Future
(crash) ref [5,6] work
tools, see figure 5.
Linear Future
dynamic work Requirements If the requirement in
the PBM can not be
Table 1. Single purpose tools. fulfilled. Re-balancing
is needed.
METHOD PBM

In the traditional process for car body Single

design, the mechanical requirements are purpose
broken down to design guidelines, see simulation
tools
figure 4. Component
designer
designer
design engineer
Requirements

Assembled complete shell model

Component
designer
designer
design engineer FEM analysis of complete shell model,
check if requirements are fulfilled.

Assembled complete shell mode

Ok Not ok, change design

FEM analysis of complete shell model, Figure 5. Development using PBM.

check if requirements are fulfilled.
• Simple analyses of the components
are here made by the design
engineers who may not be
Ok Not ok, change design experienced with FEM. This will be
facilitated by single purpose programs
Figure 4. A traditional process for car body [5,6]. This reduces the design
design with respect to mechanical properties.
4
 iterations for the complete shell vehicle structure, see figures 6 and 7.
model. Here consisting of the a-pillar, the upper
part of the b-pillar, the cantrail and the
• To a certain level, each component joint between these elements.
can be analysed and evaluated
independently of other design
activities.

• Besides the mechanical requirements,

the design engineers have many other
different aspects to take into account,
as for example geometry,
manufacturing, corrosion, etc. To not
overload the design engineers these
single purpose simulation tools have
Figure 6. The portion of the PBM, which
to be intuitive, quick and easy to use. will be a basis for detail design.
Based on the requirement and geometry
• Short time between component from the PBM a detailed design proposal
design and results from simplified is made in a CAD system.
analyses gives the design engineer
possibilities to evaluate more design
alternatives.

Description of the single purpose

simulation tools

For evaluation of beam components the

cross-sectional stiffness parameters such
as GKx, EIy, EIz and load capacity are
evaluated and compared. This is
facilitated by special purpose programs Figure 7. Detailed design proposal.
such as DAMIDA [6] and Crashcad [5]. The performance of the joint and the
connected beams are now evaluated
The evaluation of joints will be facilitated separately. The beams are evaluated
by the use of single purpose programs with respect to their cross section
using superelement technique as parameters, GKx, EIy and EIz see figure
described previously. Strategic stiffness 8 and table 2.
directions such as inward-outward and
rearward-forward bending of the joints
are prioritised [12] and evaluated.

2
Beam GKx EIy [Nmm ] Eiz
2 2
parameters [Nmm ] [Nmm ]
Required 8.0E+9 4.83E+10 2.31E+10
stiffness
Stiffness 8.32E+9 5.73E+10 2.75E+10
from design Figure 8. Proposed beam cross-section.
proposal
From the design proposal of the
Table 2. Beam performances
assembly the joint is extracted and
DEVELOPMENT EXAMPLE modeled as a FEM-model, see figure 9.
This is then reduced to a superelement
The proposed method is here exemplified
on the detailed design of a portion of a
5
and evaluated with respect to
strategic stiffness parameters[12].
Figure 9. FEM-model of proposed joint
geometry.
Both the beam and the joint fulfills the
stiffness requirements as shown in tables
1&2 above. In a more complete analysis
of the design proposal, the stiffness in
more of the joint’s d.o.f have to be
evaluated.
Joint stiffness inward-outward rearward-
[N/mm] forward
[N/mm]
Required 2.4E+3 1.7E+4
stiffness
Stiffness of 2.59E+3 1.9E+4
design
proposal
Table 3. Comparison of some joint performances.
CONCLUSIONS
The new idea here is that computational
tools should directly support the design
phase. Earlier tools for evaluation of
mechanical properties have mostly been
developed for the concept phase and the
final verification phase. Thus only for
specialists in solid mechanics and FEM,
working before or after the design
engineer in the development chain. The
design engineers, mostly with a
background as draftsmen, have until now
been left without tools to check their
results. The advantage of the proposed
method is that the design engineers
themselves, supported by simple and
quick tools, by standard procedure are
checking their design against the
mechanical requirements stated by the
the PBM. Therefore this method has the
potential to reduce the number of costly
and time consuming simulations of the
detailed vehicle model. By continuously
updating the PBM it can be used as a
quick test bench in the design process.
And thus show the impact of late
changes.
FUTURE WORK
In order to support the presented method
a number of development efforts are still
to be done. For linear elastic analysis a
simple joint analysis program using super
element technique is under development.
In the future the PBM and the single
purpose programs will incorporate
dynamic and plastic properties. See
Table 1.
ACKNOWLEDGMENTS
The gratefully acknowledged financial
support has been provided by Volvo Car
Corporation and the Foundation for
Strategic Research through the ENDREA
national graduate program.
REFERENCES
[1] M.Bauer and A.Hänschke
“Konzeptmodel mit
Knotensubstruktur im
Eintwicklungssystem AURORA”
VDI Berichte NR 816,1990,
pp.437-446.
[2] H.Zimmer et al “Use of SFE
concept in Developing FEA
Models without CAD” Paper 2001-
01-2706, IBEC conference Detroit,
Michigan October 3-5 2000
[3] W.D.Pilkey and L.Kitis “The
BEAMSTRESS Project Annual
Report 1997” Ford report, not
published.
[4] A.K.Volz “Car Body design in the
concept stage of vehicle
development” Second European
LS-Dyna Users Conference,
1999,pp.B29-B48.
6
[5] W.Abramowiez and Thomaz Gestaltparameter auf deren
Wierzbicki Steifigheit
“Crashcad analytical beam und auf das Verhalten der
evaluation program”, Gesamtstruktur im Fahrzeug-
Impact Design Entwurfssysstem AURORA “
Diplomarbeit Nr 11/89, Technische
[6] D.Lundgren, M.Johansson and Universität Berlin, Institut für
D.Adin “Damida, beam evaluation Fahrzeugtechnik, 1989,pp.67-69.
program” Master thesis 2001,
Chalmers University of [13] Y. M. Moon, T.H. Jee, Y. P. Park,
Technology, Gothenburg Sweden. “Development of an automotive
joint model using an analytically
[7] C.J.Chen and M.Usman “Design based formulation” Journal of
optimisation for automotive Sound and Vibration (1999)
applications”Int.J.Vehicle Design, 220(4), pp625-640.
vol 25 Nos1/2 (special issue),
2001, pp.126-141.
CONTACT
[8] M.E.Bendsoe, A.Ben-Tal and
J.Zowe “Optimization methods for PhD student Nicklas Bylund:
truss geometry and topology nbylund@volvocars.com
design” Structural optimization,
vol.7, 1994, pp.141-159.
Advanced body department, dept 93420,
[9] A.Klarbring, H.Fredricson and PV4B, SE-40508 Gothenburg, Sweden
Joakim Petersson (2000). (private
communications), Linköping PhD student Magnus Eriksson:
University. magnus@mt.luth.se

[10] J.L.Lubkin “The flexibility of a Luleå University of Technology,

Tubular Welded Joint in a Vehicle Department of Mechanical Engineering,
Frame” Paper 740340 SAE Division of Solid Mechanics, SE-97187
pp.1518-1522.
Luleå, Sweden
[11] MSC/Nastran “Handbook for
superelement analysis.
MSC/Nastran version 61,pp 2.2-3.
[12] M.Bauer “Darstellung der
Geometrie flexibler
Balkenverbindungen (joints), sowie
Auswirkung variabler

7
 Paper B
A design process for complex mechanical structures using Property Based
Models, with application to car bodies.
In the proceedings of Design 2002 Conference, 14-17 of May 2002, Dubrovnik, Croatia.
INTERNATIONAL DESIGN CONFERENCE - DESIGN 2002
Dubrovnik, May 14 - 17, 2002.

A design process for complex mechanical structures

using Property Based Models, with application to car
bodies.

N. Bylund, H. Fredricson and G. Thompson

Design method, concept selection, optimisation, structural design,

mechanical structures, multi-objective

1. Introduction

The objective of the paper is to present an effective process for the design of complex mechanical
structures. A multi-objective process for the design of complex mechanical structures is described.
The process utilises certain particular mechanical properties that have been identified as central to the
success of a design project. A conceptual, mathematical model of a vehicle body is constructed from
these salient mechanical properties and the model is then used to generate optimal solutions. In this
way, design variants can be explored.

The design process for complex mechanical structures is considered from the conceptual phase to
detail design. The requirements for the design are multi-objective and take the form of weight,
stiffness, manufacturing, etc., but also the requirements are not fixed and may change. Therefore the
design process must be flexible to allow for such changes. A car body is the subject of this paper and
its design encompasses all the above considerations. There is a well-established history of car design
and manufacture and traditional methods have a strong influence on current practices. The design
process described in the paper aims to reduce lead times and not exclude innovative solutions.
Shortened lead times may well be achieved by reducing iterative changes during detail design. The
development of the proposed design process has been accomplished using the framework for
engineering design research presented by Blessing, [Blessing 1998].

The paper is organised as follows. The backround to the problem is described and an overview of the
proposed design process given. The proposed process is then presented in a step by step manner
using an automotive body part as a design example. At the beginning of each section icons are used to
visualise the described steps in the design process, see Figure 1, the focused areas of the process are
visualised by solid lines and the areas not concerned have dashed lines. At the end conclusion and
future work are stated.

2. Description
The following shortcomings can be identified in the design process currently used.
− Innovative concepts, e.g. ones that use radically different materials or configurations, are often
ruled out early in the concept selection process.

1
− The results from concept studies are not used in an efficient way during detail design, leading to
many costly redesigns during the detail design phase.
− Knowledge gained by benchmarking is not used quantitatively.
− Late changes in requirements lead to expensive redesign activities, during the detail design phase.

3. Prescription
In order to pursue a meaningful concept development and selection process, it is necessary to
− identify the principal assessment criteria.
− measure the performance of the concept model with respect to criteria.
− assess the ‘value’ (not necessarily in monetary terms) that each performance measurement
contributes to the total quality of the concept.
− improve the concept with respect to the value judgement to obtain an optimal solution.
− revise the concept model in parallel with the detail design activities.

Requirements

Competitor

Concept idea
CONCEPT

Tools / Methods / Bank

Build up & Break down

PBM
Evaluation

Selected
solution
PROJECT

Detailed
design

Figure 1. PBM based process for design of complex mechanical structures

Figure 1 shows the design process that is the subject of the present research. It can be seen that the
starting point for all projects is a set of requirements (mass, structural integrity, etc.) [Fenton 1996]. A
Property Based Model (PBM) is built up for each concept and represents the mechanical and spatial
properties of the body concept. The PBM is constructed from organs, [Hubka et al 1988], which
represent requirements at a local level. The chosen organs for the car body are beams, joints and
panels. In the build up and break down activity, the PBM models are generated for new designs and
existing designs from competitors by in-house design teams. A library of organs is used to generate
PBM models efficiently; typical elements are beam cross-sections and joint properties. Each project
generates new organs and an extensive knowledge bank is created. Such a library is a resource of
expertise and so knowledge is readily transferred and the design process is not dependent of particular
individuals and their subjective value judgements.

2
The optimisation procedure normalises the PBM models with respect to key global requirements, e.g.
global stiffness, weight, crash worthiness, etc [Hidekazy 2001]. The results of the procedure are used
as the basis for comparison of the concepts. The PBM that best fits the quantitative and qualitative
criteria is selected. This is an aspect of the design process and is not left to the outcome of any
particular design evaluation method. The selection involves the decision-making processes with the
company as a whole, and it is important that all relevant parties contribute to, and accept, the outcome
of evaluation.

The selected PBM contains all the properties of the model at the organ level. These properties are the
guidelines to the detail design engineers. During the detail design phase, the designer is thus provided
with quantified solution requirements to meet as decided by the concept PBM that has been selected.
The detail design engineer is supported by the computational tools used in the break down phase and
can use particular solutions from libraries. Local changes required at the later stages of design can be
considered objectively by the detail designer. The changes can be tested against the solution
requirements (stiffness of the subassemblies, etc.) and if the requirements remain satisfied then the
change can be met.

4. First step; Requirements

The steps in the proposed design process are exemplified by the design of an A-pillar,
see figures 3 and 4. The first step in any development process is to state the
requirements [Pahl and Beitz 1996] on the system to build. The requirements on a car
body are numerous and often contradictory. One requirement affecting the A-pillar, is
the American roof crush legal requirement MVSS 216, see figure 2.

Figure 2. American roof crush legal requirement MVSS 216

4.1 Competitors
In the presented process competitors having properties close to the targeted market segment are
chosen. These competitors give hints and ideas on how to design a car body fitting the chosen
market-segment. The process of examine competitors is called benchmarking, see [Andersen 1996].
Benchmarking can be done in many areas and levels, such as business strategy, marketing strategy as
well as on the product itself. Benchmarking of competitors product was used by Xerox and permitted
them to gain in competitiveness, also in our process product benchmarking is used.

3
 Figure 3. Three different A-pillars from competitors, used for
benchmarking.

4.2 Concepts
Concepts are developed based on the requirements and results from benchmarking, the concepts can
be of different level of detail, from sketches on a napkin to a CAD drawing. Since European car
manufacturers built the first self supporting car body (Uni-body) this type of structure has been the
dominating structure with the few exceptions mainly being sports or luxury cars in small series. The
self supporting car body, where sheet metal stampings are spot welded together to form integrated
beams, joints and panels may be a good way of building cars but it can also act as an fictitious
constraint [Pahl and Beitz 1996 ] impeding designers to think of new solutions. Furthermore concepts
based on self-supporting structures will because of the deeply rooted experience of this technology
have behaviour near their optimum, the design process has therefore to be able to handle this bias.

Figure 4. Three concepts for the A-pillar. From the left-hand side, stamped and spot welded boron steel,
hydroformed steel and finally stamped steel with an interior reinforcement tube made of boron steel.

5. Second step; Break down

The chosen competitors and the concepts developed are broken down into manageable
quantities and objective information. This enhances comparison between different
concepts and competitors. For an A-pillar this information is the shape of the middle
line of the beam and the section. By cutting the competitors A-pillar into slices and
using the in-house software DAMIDA, figure 5, the sectional properties can be found.
The sections of the concept ideas can be analysed in a similar way. The results of these analysises are
put into the knowledge bank. And used for creating the property based model, PBM, see figure 5, and
reference [Bylund 2001].

4
 Figure 5. Break down process. A competitors car body or a concept is sliced and the sections numbered
and analysed with the software DAMIDA. Knowledge of the section capacities is gained and put in the
knowledge bank.

6.Third step; Optimisation

By using the intermediate stage of organs, principal function-carriers are identified
without restricting detail design. The model built up of these organs is called PBM
and it links requirements from global to organ level. This solution independent
representation enhances new designs by limiting fixation, [Pahl and Beitz 1996]. The
generality of these organs is put on a level where they give sufficient information for
carrying out detail design but without biasing it. Using the same metrics for all PBMs,
beam parameters, joint parameters and sheet parameters, enhances the selection of PBM. For an A-
pillar the metrics are sectional properties such as Ix, Iy, Iz and moment capacity.

A tool for optimisation of frame structures has been

developed to handle the optimisation step in the design
process, see figure 6. This tool can handle booth beam
and joint organs and in the future also panel organs.
Multiple load cases can be used and also separate mass
constraints for beams and joints or one mass constraint
for total mass. The input to this tool is the PBM model
defined by
− Topology
− Material
− Masses
− Beam section properties
Figure 6. Optimisation tool
− Joint stiffness

By use of this data an optimisation model can be developed for each concept and competitor or
directly from requirements. As an example, the basic topology for a A-pillar is shown in figure 7a and
F F

Fix

Figure 7a. Basic topology Figure 7b. Optimised structure

5
the optimised beam sections can be seen in figure 7b. In the initial design all beam segments had the
same cross section properties and after optimisation one can see that the allowable material have been
distributed in an optimal way between the different beam segments. The result of this is a structure
with the stiffness increased by 2.7 times with respect to the initial design, the maximum deflection can
also be found. The optimal diameter varies quadratically with the distance, x, for the upper part of the
A-pillar as in figure 8. This is the real objective potential for that 0.7
Optimal diameter change for upper A-pillar

concept based on its topology and mass. The same is done for all 0.6

concepts, competitors and requirement based models. In the A-pillar 0.5

optimisation all these models would have the same mass but dia
met0.4
er

different performance in stiffness. Another possibility is to set 0.3

maximum allowed displacement and minimising mass. This results 0.2

in structures performing equally but with different masses. If the 0.1

structure is build up by beams and joints the program can also be 0

1 2 3 4 5
position
6 7 8 9 10

used to optimise joint stiffness in a structure. This is a good way to

get information of how to build the different beam connections Figur 8. Diameter
existing in the structure. An example of such a structure is the 3D-
console beam seen in figure 9 where beam properties and joint stiffness has been optimised
simultaneously. Two load cases have been used in this optimisation. The size of each sphere is
proportional to the stiffness for that joint.

F2 F1

Figure 9a. Ground structure Figure 9b. Optimised frame structure

By optimising the PBM models for minimum weight and given deformation or maximal stiffness for a
given mass all concepts are normalised and are at the edge of their performances. This enables a fair
selection process between the different PBMs.

7. Fourth step; Build-up

When the PBMs are optimised they need to be built up with a technical solution in
order to get sufficient information for the evaluation. This stage can be seen as the
embodiment design stage in [Pahl and Beitz 1996]. The built up stage is done with the
same tools as the break down stage, furthermore results from break down such as
competitors beam sections can be used as inspiration, this way the break down and
built up works as the knowledge bank discussed previously. In our example three different
technologies where chosen in the build up stage of the A-pillar, see figure 4. A bent boron steel tube
with constant section covered with sheet stampings welded together, two stamped boron steel sheets
spot-welded together to a tube with varying section, and finally a hydro formed steel tube. All these
concepts should respect the same PBM’s quantitative mechanical requirements such as Ix, Iy, Iz and
moment capacity. In the evaluation stage presented later, these concepts are compared mainly with
respect to non-mechanical requirements such as price, manufacturability, aesthetics and
environmental since they have already been normalised with respect to mechanical requirements.
During the build up stage it was found that the alternative with an interior steel tube covered with
stampings could not reach the mechanical requirements therefore this concept was rejected already
during the build up phase.

6
8. Fifth step; Concept selection and Detail design

8.1. Concept selection

The design literature contains many references to specific methods that can be used to choose
between alternatives [Pahl and Beitz 1996, Pugh 1990 and Cross 2000]. These methods are suitable
for use at different stages of design, e.g. ideas, concepts and detail design and for different types of
criteria, e.g. quantitative, qualitative, subjective and objective. It is important to use selection methods
appropriately otherwise choices can be made in error. For example, it is not a wise approach to select
concepts by guessing performance scores with respect to criteria that can only be evaluated when
much more detail is known.

Commonly, design alternatives are considered at the conceptual stage by a general qualitative review
of a broad spectrum of criteria followed by more detailed analysis of a reduced set of options that has
been chosen for further consideration. This process can be unsatisfactory. Good options can be
eliminated because of personal preferences, there may be uncertainty concerning the benefits of
certain options or lack of experience may bias judgement. Such problems can be avoided in concept
selection if the key parameters can be identified that can be used to generate potentially optimal
concepts using high quality information.

The use of property based modelling and optimisation normalises all concepts by putting them on the
edge of their mechanical performance. Only the concepts that fulfil the quantitative mechanical
requirements continue to the concept selection step, this reduces the option choice. In order to make a
final selection, qualitative criteria such as surface treatment, weldability etc. need to be considered.
However, since the number of options has been reduced, each qualitative criterion can be considered
in depth.

Topology, material properties, mass, sectional properties and joint stiffness were used in the
optimisation stage of the process. The task is now to consider the two concepts that fulfil the
mechanical requirements and compare them with respect to a range of qualitative criteria. The
concepts are:
− two stamped boron steel sheets spot-welded together to a tube with varying section
− a hydro formed steel tube

The following criteria were identified as being important in the qualitative evaluation process:
1. Cost
1.1 Investment
1.2 Production cost
2. Production: risks associated with the following operations.
2.1 Welding
2.2 Assembly
2.3 Tolerance
2.4 Surface treatment
2.5 Lead time/ production capacity.
3. Attachments: risks associated with the following
3.1 Windscreen

7
3.2 Interior trim
4. Styling freedom

Paired comparison analysis, in which the concepts are compared to each other with respect to each
criterion is an appropriate method to evaluate these concepts [Pugh 1990]. Engineers were consulted
who were expert in the relevant fields and, from their judgements, it was possible to complete the
concept evaluation table. The concept consisting of two stamped boron steel sheets spot-welded
together to a tube with varying section is used as reference. Table 1 shows the results of the
evaluation. In the table, a ‘+’ sign means that a concept is better than the reference concept with
respect to a particular criterion, a ‘-‘ means worse than and an ‘S’ means no difference (or cannot
judge).

CRITERIA CONCEPTS
Stamped spot welded A hydro formed steel
boron steel sheets tube

1. Cost
1.1 Investment -
1.2 Production cost +
2. Production:
risks associated with
2.1 Welding +
2.2 Assembly S
2.3 Tolerance +
2.4 Surface treatment Reference +
2.5 Lead time/ S
production capacity.
3. Attachments:
risks associated with
3.1 Windscreen -
3.2 Interior trim S
4. Styling freedom +

Σ+ 5
Σ- 2
ΣS 3

Table 1. Paired comparison analysis.

From table 1, it can be seen that the hydro formed steel tube concept emerges as a favoured concept.
The next stage of the process would be to investigate the negative aspects of this concept to ensure
that no insurmountable problems will be encountered and to undertake more detailed work on these
and other aspects of the concept.

8.2. Detail design

When a concept is selected the project phase starts and detail design is launched. Teams of design
engineers do the detail design. They are supported by the break down build up tools and the
knowledge bank. The knowledge bank should be used for tips during detail design and not as an
exhaustive list of what designs are permitted. Seeing the bank as an exhaustive list has a negative
impact on the design process augmenting the fixation, see [Pahl and Beitz 1996] and impeding the
possibility for optimal designs. By using these easy to use tools, e.g. DAMIDA detail designers can
continuously compare their design against requirements at organ level. The knowledge gained during

8
the concept phase will therefore survive and contribute to the success of the final detail design. Costly
loops of complete car body simulations will be reduced.

9. Conclusion
The proposed process has the potential to both speed up the development of complex mechanical
structures while at the same time enhancing innovative solutions. This is done by stimulating a more
solution independent approach where a function structure of organs called PBM represents the
behaviour of the car body.

Several easy to use tools that do not require expertise in CAE-analysis are used to evaluate the
performance of organs. The tools are used both during the concept and detail phases. The tools
generate a bank of knowledge of the salient engineering features of organs which makes the
development process less dependent on the subjective interpretation of particular individuals. The
concept selection process is both efficient and clearly objective. Concepts that do not attain essential
performance requirements are rejected at an early stage. In-depth evaluation of concepts focusses on
thos concepts that have attained an optimium performance for their generic type. Qualitative
evaluation, which involves extensive investigations, is only carried out on concepts that can achieve
the required performance criteria.

Breaking down all proposed concepts to organ level makes comparison easy using parameters of the
same type. The same tools are used to break down design proposals for analysis as are used in
embodiment and detail design, therefore the process is efficient and creates a learning design
environment.

Concept selection processes such as that described in section 8.1. are not without their problems. It is
difficult to ensure that all departments and groups within a company accept the decision and it is not
unknown for concept selection decisions to be re-visited later, which is wasteful of time and effort. In
order to select a concept that is acceptable throughout the organisation, it is important not to let some
poor characteristics be masked by other positive characteristics. It is possible for a multi-ctriteria
concept selection method to identify certain concepts as being very good because a negative with
respect to a particular criterion is compensated by a number of other posisive assessments. E.g. a
selection method that results in a light, strong concept but which also has a very corrosion sensitive
surface will not be welcomed by people at the surface treatment department even though that concept
has the largest number of positive characteristics. Another way to take decisions is required.

10. Future work

A better way could be not to place emphasis on any particular method(s) as the leading approach to
concept selection, but instead to clarify and strengthen the decision-making processes used by a
company. The principal point is that all key departments, from aesthetic styling to manufacture,
should own the decision to adopt a particular concept. For such a sense of ownership to be
established, departments need to have a greater involvement in the concept selection process. This
means more than commenting upon the merits and de-merits of particular concepts at some stage so
that a particular selection process can be used.

Involvement by all is required throughout the concept development process from the first stages. It is
beyond the scope of this paper to pursue this line of thinking in depth, except to comment briefly on
future research activity in this area. The decision making paths need to be understood: at what stage
are decisions made, who makes them, are there powers of veto? These questions require answers
based on realism, not idealism, and must encompass all levels of company decision making. Also,
concerning the integration of diverse departmental interests, there are certain guideline principles that
are well established which might be usefully employed. Multi-disciplinary groups can be very

9
productive when their work is directed using creative problem solving principles. Divergent thinking
with suspended judgement coupled with objective convergent thinking helps to identify potential
problems, to generate solutions and to gain acceptance of the decisions by all interested parties.

Good decision making in concept development relies on the use of clear decision making processes
which include all departments in a meaningful way, supported by objective methods such as property
based models.

Acknowledgements
The gratefully acknowledged financial support has been provided by Volvo Car Corporation and the Foundation
for Strategic Research through the ENDREA and IVS national graduate programs. Furthermore we acknowledge
the stimulating discussions held with M.Sc. Jonas Forssell at Volvo Car Corporation.

References
Andersen,B. and Pettersen,P., “The benchmarking handbook”, Chapman & Hall England 1996.

Blessing, L., Chakrabrit, A. and Wallace, K. “Designers – the Key to Successful Development”,
Springer-Verlag, London, 1998, pp 56-70.

Bylund, N. and Eriksson, M., “Simulation Driven Car Body Development Using Property Based Models
” SAE paper 2001-01-3046. In proceedings of IBEC 2001, conference postponed to
9-11 July, in Palais de congres in Paris, France due to the 11 of September event.

Cross, N., “Engineering Design Methods”, third edition, Wiley 2000.

Fenton, J.,” Handbook of Vehicle Design Analysis”, MEP, London, 1996.

Hidekazy et al., “First Order Analysis – New CAE Tools for Automotive Body Designers” SAE paper
2001-01-0768, SAE 2001 World Congress Detroit, Michigan March 5-8, 2001.

Hubka, V., Andreasen, M., and Eder, W.,“Practical Studies in Systematic Design”, Butterworths,
London, 1988.

Pahl, G. and Beitz, W.,”Engineering Design”, Springer, Berlin 1996.

Pugh, S., “Integrated methods for successful product engineering”, Addison-Wesley 1990.

Nicklas Bylund, M.Sc.

Department of Advanced Body Engineering, 93420, Volvo Car Corporation SE-405 08, Gothenburg, Sweden
and
Division of Computer Aided Design at Luleå University of Technology, SE-97187, Luleå, Sweden
telephone: +46 (0)31 765 4145
email:nbylund@volvocars.com

10
Paper C1
 ADRIAN: A software for computing the stiffness of automotive joints and its
application in the product development process
Nicklas Bylund
Luleå University of Technology
Division of Computer Aided Design
Luleå, Sweden
and
Volvo Car Corporation, Göteborg, Sweden

Keywords: design process, automotive joint, car body, joint stiffness, simulation

Abstract
The development of complex mechanical structures such as a car body is an iterative process,
alternating between design and analysis. Traditionally, these are made in different departments,
making the loops between design and analysis slow and costly. This paper presents a method with
accompanying software for the design engineer/draftsman to do preliminary mechanical analysis
himself, which not only makes design loops shorter but also means they can be made in parallel.
This speeds up the development process, while at the same time allowing exploration of more
alternative solutions.

1. Introduction
The Body In White (BIW) is the main structural part of a vehicle. Traditionally, the BIW is built up
from sheet metal stampings spot-welded together to create a shell structure, see Figure 1. The BIW
has a number of functions, such as the bearing structure for the engine, suspension, sub-frames,
powertrain and seats, as well as being the largest visible surface of the car. The main global
mechanical requirements for a BIW are stiffness, crashworthiness, and noise, vibration and
harshness (NVH); these have to be fulfilled at minimum cost and weight.

B UPPER
A UPPER C UPPER

A MIDDLE

B LOWER

Figure 1. Lateral view of Volvo S80 BIW CATIA, with main joints in colour.

1
Detail design is done with Computer Aided Design (CAD) tools, which are used to describe the
geometry in 3D. The CAD geometry is the basis for different Finite Element Method (FEM)
analyses such as static stiffness, eigenfrequencies, eigenmodes and crash at global car body level.
Today, design and analysis are done in different departments; furthermore, analysis is concentrated
on the complete BIW. This is why it takes a long time before the design engineer/draftsman
becomes aware whether the detail design of his particular area of expertise contributes to the BIW's
purpose, which is to fulfill the global requirements. To address this problem the global
requirements for the car body have to be broken down to local requirements corresponding to the
design areas of the designers/draftsmen. This breakdown can be made by using a concept model in
which local mechanical requirements for beams and joints are balanced so that the resulting global
behavior of the complete model corresponds to global requirements, see Chapman and Pinfold [1],
Bylund et al [2 and 3]. By introducing easy-to-use tools to verify if local designs fulfill the local
requirements, the number of long loops between design and analysis can be reduced.
The design engineers/draftsmen carrying out the detail design do not have the time or knowledge to
use general purpose analysis tools to verify his design. This paper presents an analysis tool for
joints, see figure 1, which after a short introduction will be easy-to-use for the designer/draftsman.
While the analysis department simulates complete vehicle models, the design engineer analyses his
particular area of design. The process and tool presented are not intended to replace today’s
complete body simulations; they are merely to permit the design engineer to make a preliminary
check of his design, leaving the more in-depth complete body analysis to the analysis department.
This way the number of immature designs being transmitted to the analysis departments are reduced,
thereby limiting the number of large iteration loops, see Figure 2 (dotted line). The easy-to-use
analysis tools used at VCC so far are DAMIDA [4] for beams, and ADRIAN for joints, which is the
subject of this paper.

Global requirements hard

to use for component
designers

PBM
Requirements broken down
to organ level and thereby
easier to use for component
designers
Easy-to use tools,
Component ADRIAN and
Component DAMIDA
designer
Component
designer
designer

Assembled complete shell mode

FEM analysis of complete shell model check

if requirements are fulfilled,

OK Not OK, change design

Figure 2. Development process using PBM and easy-to-use analysis tools.

2
2.1 Joint stiffness evaluation
In a self-supporting car body the joints, see figure 1, are of great importance, governing up to 60 %
of the global stiffness, Lubkin [5].
The requirements for spot-welded automotive joints are numerous:
1. Stiffness
2. Crash performance
3. Manufacturability (stamping, factory sequence, weld accessibility)
4. Corrosion protection, liquid escape, etc
5. Interaction with interior panels
6. Feasibility to make with the sheets available in the joint*
7. Acceptable cost
*It should be stressed that the joints in a car body of monoque type are not isolated components but
the result of fabrication of steel sheeting. During detail joint design, the design engineers
continually discuss and weigh how the requirements are to be fulfilled. A deep understanding of all
the above types of requirements, and spatial vision, are needed to be able to create automotive joints.
Due to the complex geometry of a joint and the constraints of the automated manufacturing process
(e.g. heavy investments in case of a change of robot configuration) great efforts can be spent on a
detail such as whether or not to add one extra spot weld. Convenient parameterization can not be
made because although the basic topology of a car body joint is repeatable, the detail design of a
joint is one of a kind. This paper describes how each design engineer can analyze the stiffness of a
joint in parallel with his design work without having to rely solely on the analysis department.
There are some major difficulties in evaluating joint stiffness:
• To define a geometric joint center.
• In the case of static measurement, to fix one or more degrees of freedom in an experimental set-
up and assure zero displacement.
• Automotive joints have no standard shape, thus the experimental set-up is unique for every joint.
As it is hard to define where an automotive joint has its center, it is possible for the joint to have
more than one center, see Figure 3.

Centres
Centres

Figure 3. Realistic shape of automotive joint with no

absolute centre.

When no unique center can be found, the distance X, see Figure 4, is not unique, so evaluating the
joint stiffness with the static method in Figure 4, as presented by Moon et al. [6], leads to
unacceptable errors. Even when a unique joint center can be defined, the set-up is laborious: the
joint has to be permuted, e.g. each leg has to be grounded while one of the others is attached to a

3
force-lever. To deal with the difficulties described above, two independent methods are used, as
presented in the following sections.

X Geometric joint center

F

Rigid wall

Deformation measured

Figure 4. Example of static stiffness evaluation method: suitable only for

geometries where the geometric center can be found.

2.2 Super-element method

The super-element method condenses the stiffness of an arbitrary FE model to any desired number
of nodes. The corresponding FE stiffness matrix (K-matrix), e.g. a sheet metal joint modeled with
plate elements, can be a hundred columns by a hundred rows in size depending on the coarseness of
the mesh. This K matrix can be reduced in size to only the number of legs x 6, see Equation 1,
Moon et al [5] and Nastran Handbook [8].

The stiffness matrix from the FE-model of the joint

component can be partitioned as:
K mm K ms Dm Fm
T ⋅ =
K ms K ss Ds Fs

K m = Reduced stiffness matrix

Dm = Master d.o.f, here 6 d.o.f at the end of each leg.

Ds = Internal d.o.f.

Fm = Forces on the master nodes.

Fs = Forces on internal nodes, here equal to zero.

Reduction gives :

−1K T
K m = K mm − K ms K ss ms

Equation 1. Super-element reduction.

The joint center does not need to be explicitly defined and the boundary conditions can be free-free.
This means that the boundary conditions can easily be automated in a software and do not need to be
specified manually by the user. Only master nodes at each leg end have to be defined, see figure 5.
The condensed matrix contains the stiffness in all directions at the leg ends. To simplify the
interpretation of the joint stiffness, the principal directions of stiffness for each leg can be calculated
and presented, see Figure 6.
4
 Figure 5. Joint leg-end section with a master node connected with
rigid body elements.

2.3 Dynamic joint method

As a complement to the super element method, a method using structural dynamics can be used. An
approach using structural dynamics for evaluating joint properties is given in Wyatt Becker et al.
[7]: lumped masses are put at the ends of the joint. The resulting eigenfrequencies and eigenmodes
can be used as a measure of the joint stiffness. The method presented here is similar, and has the
following main features:
• No levers are defined.
• Free-free set-up is used.
• A heavy mass (m>> joint mass) is cast on each leg; experiments have shown that a 5 kg mass
suits car body joints.
The joint is defined as inscribed in a sphere with radius r=250 mm, a size fitting all joints in a car
body. The center of the sphere is placed in the middle of the various joint centers, see Figure 3, and
the legs are cut where the sphere bisects them. Finally, a heavy mass is cast onto each leg end, see
Figure 7. Modal analysis is then done with the joint hanging on soft springs (eigenfrequency of the
assembly less than one tenth of the first eigenfrequency of the joint), thus creating a free-free set-up.
The frequencies and the corresponding eigenvectors are used to visualize the vibration behavior of
the joint. Since the shape and size of the masses are standardized and much heavier than the joint
itself, different joints can be compared, see [8] for more details. The higher the eigenfrequency, the
higher the stiffness. The advantage of the free-free set-up used is that no expensive, stiff, tooling is
needed. The virtual counterpart of the experimental set-up is easy to reproduce; a CAD geometry is
used as a basis for an FE mesh. The masses can be added at each leg end as lumped masses with
appropriate inertia properties. The final stage is a computational modal analysis, omitting the first
six modes (rigid body modes), giving the eigenfrequencies and the eigenmodes as in the
experimental case. As with the super element method, the boundary conditions can easily be
automated in a software without the need for manual interaction. One could argue that the sphere
used when cutting the legs also needs to be in some kind of joint center. However, the results are
not as sensitive to the position of the center as when a static method is used, see Urzua and
Chouping [9]. The dynamic method has the advantage of being exempt from the numerical
frequency values, also giving a qualitative animation where weak areas, e.g. lack of spot-welds, are
clearly visible, thus giving a hint where there is potential for improvement.

5
 Figure 6. Visual representation of the joint stiffness and information about
sheet gauges and spot welds

3. ADRIAN: the joint evaluation program

This chapter describes briefly the development of the ADRIAN program for automotive joints.

3.1 Requirements for ADRIAN

The mechanical requirements for car joints are always of the same type, even if their numerical
values change from one development project to another. Standardizing the model building, analysis
6
and post-processing according to the mechanical requirements is therefore possible. Using the
analysis capabilities in CATIA does not permit standardizing the boundary conditions and analysis
steps (i.e. the dynamic joint method and the super-element method) which is essential if the analysis
should be done rapidly and be repeatable by users with limited experience of analysis. Furthermore,
the automatic meshing in ANSA, standard at VCC, is better than the one in CATIA. As DAMIDA
[10], a similar type of software for fast beam analysis, has previously been developed at VCC, some
experience exists of how to run the development of a similar program for joints, such as ADRIAN.
The key factors for success are identified as:
− Easy to learn and use even for design engineers with little experience of analysis.
− Detailed geometry exported from CAD program used. Thus no need for rework.
− Automatic meshing with small (5mm) elements used; gives good results and works even when
surfaces in the detailed CAD model are small; gives accurate result for global stiffness and
eigenfrequencies of the joint.
− Necessary user-interaction should be concentrated on the CAD environment as far as possible
because that is the home environment for the design engineer.
− Commercial solver to obtain confidence in results; if possible the same solver as the analysis
department.
− Use in-house software to minimize license costs.
− It should take no more than 15 minutes, maximum, to analyze a joint.
− Results should be presented clearly, permitting easy comparison with requirements.
Documentation and communication of the results should be simple.
− A small team of programmers should develop the program entirely, thereby keeping an
overview of development in order to minimize cost and assure quality.
− Possibility for advanced users to override default parameters and also manually run sections of
the program.
− There should be continuous dialogue with the maintenance department during program
development, in order to ensure that the program is easy to maintain.

Figure 7. Experimental set-up for the dynamic joint method

3.2 Problem formulation for the development of the ADRIAN program

In order to fulfill the requirements described in the previous section, the software must be designed
for the specific platform and users at VCC. The main features of the software are its capability to
import a CAD model from a CATIA joint model via the IGES graphical format; transform it to a
finite-element-model; perform linear static super-element reduction, as in Section 2.2; and modal
analysis according to the dynamic joint method, in Section 2.3, and; finally, visualize the results.
Technical requirements are as follows:

7
1. Unix platform using C language; widespread and cheaper than using matlab licenses
2. Adapted to shell models of joints made in the CAD software CATIA.
Compatibility with the IGES graphical exchange format to permit adaptation to other CAD
programs if necessary. (IGES is a VCC standard.)
3. Automatic geometry meshing in ANSA software. (ANSA is a VCC standard.)
4. Automatic and mesh-independent addition of weld spots based on points defined in CATIA by
the design engineer. With the same weld definition as used by the analysis departments, namely
CHEXA elements connected to the CQUAD4 and CTRI3 in the shell mesh using displacement
interpolation functions. (Functionality of the in-house SPOT code.)
5. Automatic modeling of the master node at the leg end and its connection to the section, see fig 5;
no need for users to define the boundary conditions.
6. Automatic submission to the MSC/Nastran solver.
7. Automatic and pedagogic presentation of both stiffness and mode shapes, using
existing post-processing software and HTML pages.

3.3. The work process using ADRIAN

During detail design, the design engineer has to fulfill the requirements stated in 2.1, including the
stiffness requirement. Several alternative joints can be developed and an important criterion when
choosing between alternatives is stiffness. Without ADRIAN the design engineer has to wait until
the complete car body has been analyzed at the analysis department, and as this can take a long time
only a few design alternatives can therefore be tested. On the other hand, using ADRIAN the design
engineer can calculate the stiffness of the joint and compare it to requirements himself. Relative
comparison, i.e. checking how a change in the design of a joint affects the stiffness, is also valuable.

3.4. Data flow in ADRIAN

The program consists of a pre-processor that automatically converts the IGES model to the Nastran
format via the ANSA mesher, a Nastran launcher and a post-processor that treats the data. The
procedure is described below.
3.4.1. Manual user input needed
The user starts with cutting the joint out in CATIA using planes to define the cut on each leg (3 or
4); the user then defines the thickness of the various metal sheets defines the points where spot
welds should be situated and their diameter. Each type of information is put into a different set by
the user: there are thus weld sets, geometry sets and a planes set in the resulting CAD model that is
finally converted to the IGES format.

3.4.2 Automatic data treatment

When ADRIAN is started it automatically extracts the information from the IGES file and takes
appropriate actions for the set content. The geometry sets are the base for the ANSA meshing; the
weld sets are used to create the welds using the in-house code, SPOT; the plane set is used to put the
master nodes at each leg end and connect them with rigid elements to the nodes around the section,
see Figure 5. Once pre-processing is complete and the mesh created, ADRIAN automatically
submits a job to the MSC/Nastran solver, for linear elastic analysis and super-element reduction.
While Nastran runs, ADRIAN waits for significant files, (.f06, .op2 and .pch) to be returned and
then automatically launches the post-processing sequence (.f06 contains warning and error
messages; .op2 contains eigenmodes; and .pch contains the reduced stiffness matrix). Once post-
processing is complete, the user automatically obtains a HTML page created in Matlab containing
rotational and translational stiffness for the joint legs, see figure 6. These are calculated from the
.pch file and information from the plane set. ADRIAN also automatically launches Animator
through its script language, see figure 8. Animator shows the eigenmodes resulting from the
dynamic joint method.

8
 Figure 8. Screen capture of ANIMATOR, here can the mode shapes
according to the dynamic joint method be visualised. Lumped masses with
inertia are put at each leg end, but are not visible.

3.5. The graphical user interface

The ADRIAN Graphical User Interface (GUI) is programmed in OSF/Motif, see Figure 9. The
program may be run completely, or for advanced users, just by launching separate routines. Each of
these sub-processes are stand-alone and may be executed from specific menus. There is also a help
section available from the interface consisting of user information on all ADRIAN functions.
Processing information is printed in the Output window, see figure 9. ADRIAN gives information,
for example, on IGES file contents, Nastran analysis results and processing errors.

9
 Figure 9. ADRIAN graphical user interface during a run.

3.6. The post-processing interfaces

ADRIAN visualizes the results in two forms: as a HTML page with the principal stiffness calculated
according to the super-element method, see Section 2.2 for each leg figure 6, and as eigenmodes
according to the dynamic joint method, see Section 2.3, figure 8. The HTML page also contains a
reference view and information about sheet gauge and spot-welds, see figure 6.

3.7. ADRIAN robustness test

The first test of the tool was to cut out five joints from the Volvo S80 BIW CATIA model. Figure 1
shows the global position of the analyzed joints. The program showed good capability for
transforming the CAD models to FEM models. Two faces were not connected in one of the models,
possible resulting problems are discussed in Section 3.8. The overall performance was satisfying,
i.e. all joints but one could be analyzed in under 15 minutes. All of the FEM models produced by
ADRIAN passed the standard limits for singularities set by MSC/Nastran.

3.8. Analysis of possible instabilities

Rare instabilities could be identified in the connections between the faces in the CAD model.
ANSA has a tolerance level in order to define which faces are joined together. In one of the
analyzed joints, B_LOWER, ANSA failed to connect two fairly long faces, and this resulted in a
crack in the mesh. When studying the modal animations in ANIMATOR, this crack was easily
identified. It may be wise to investigate the underlying problem causing the errors in ANSA, and, if
possible, draw a better geometric CAD representation. However, it is important to emphasize that
10
only two faces of the hundreds of faces in the tested models had a problem that affected the total
result, and this problem was easily identified and corrected by cutting the long surface into two
shorter ones in CATIA. Another approach to correct errors, for the user familiar with basic
commands in ANSA, is to repair the geometry or repair the mesh using the advanced option in
ADRIAN, permitting manual interaction in ANSA.

4. Conclusions
Breaking down global requirements to local joint stiffness requirements permits analysis and
verification at the local level with the help of ADRIAN.
The combined use of the super-element method and the dynamic joint method presented here
complements earlier methods, especially when the geometry of the joint makes it difficult to find a
unique center. In addition, the dynamic joint method has an experimental counterpart enhancing the
use of quantitative benchmarking of competitors’ joints.

With easy-to-use programs for evaluating his design in comparison to local requirements the detail
designer can reduce the number of lengthy design loops involving the complete car body and,
consequently, development time. The joint stiffness values presented in the HTML page permit
comparison with local requirements and the eigenmodes shown in ANIMATOR shows weak areas
in the design . The time gain when analyzing a couple of different design alternatives for a joint is
considerable; analyses can be completed in a day instead of the one to two weeks needed when
submitting the work to the analysis department and waiting to get the results report back.

5. Ongoing and Future work

ADRIAN has been introduced in a pilot group at VCC’s car body department and is used in ongoing
VCC projects. Furthermore a consultant firm hired by VCC has analyzed a number of joints from
earlier Volvo cars using ADRIAN to give a map of the picture of stiffness levels in different car
bodies. Introduction of the software at VCC will continue on a broader scale, the resulting changes
in the design process will be monitored and results published; the focus will be on time and quality
gain.

Acknowledgements
The author gratefully acknowledges the financial support provided by Volvo Car Corporation and
the Foundation for Strategic Research through the ENDREA national graduate program. The
ADRIAN programming performed by Henrik Sandström and Mattias Shamlo has been invaluable.
Furthermore, thanks to design engineer Hossein Rezaei for help with the CATIA modeling and
suggestions for ADRIAN from a user point of view. I would also like to thank Patric Kennerud and
Mats Johansson at Volvo IT for suggestions on program structure.

References
1. Chapman CB, Pinfold M. The application of a knowledge based engineering approach to rapid
design and analysis of an automotive structure. Advances in Engineering Software 2001; 32:903-
912.

2. Bylund N, Fredricson H, Thompson G. A design process for complex mechanical structures using
Property Based Models, with application to car bodies. INTERNATIONAL DESIGN
CONFERENCE - DESIGN 2002 Dubrovnik,Croatia, May 14 - 17, 2002.

3. Bylund N, Eriksson M. Simulation Driven Car Body Development Using Property Based Models.
SAE paper 2001-01-3046, in proceedings to IBEC 2001.

4. Lubkin JL. The Flexibility of a Tubular Welded Joint in a Vehicle Frame. 1974 SAE
740340:1518-1522.

11
5. Moon YM, Jee. TH, Park YP. Development of an Automotive Joint Model Using an Analytically
Based Formulation. Journal of Sound and Vibration 1999; 220(4):625-640.

6. MSC/Nastran: Handbook for superelement analysis. MSC/Nastran version 61, 2.2-3.

7. Wyatt Becker PJ, Wynn RH-Jr, Berger EJ, Blough JR. Using Rigid-Body Dynamics to Measure
Joint Stiffness. Mechanical systems and Signal Processing 1999; 13(5):789-801.

8. Bylund N, Fast and Economical Stiffness Evaluation of Mechanical Joints, SAE 2003-01-2751,
JSAE 20037025, International Body Engineering Conference, (IBEC03), Chiba, Japan, October 27-
29, 2003.

9. Urzua P, Chouping L. Parametric study of the T-joint stiffness of an aluminium car body. M.Sc
Thesis, Dept of Structural Mechanics, Chalmers University of Technology, 1999.

10. Lundgren D, Johansson M. Development of Sectional Capacity Software. MSc Thesis, Dept of
Structural Mechanics, Chalmers University of Technology, 2001.

Author: Nicklas Bylund, 93710, PV2A2, Volvo Car Corporation, 405 31 Gothenburg, Sweden.
Email: nbylund@volvocars.com
Telephone: +46 (0)31 325 41 45
Fax: +46 (0) 59 9990

12
 Paper C2
Fast and Economic Stiffness Evaluation of Automotive Joints
In the proceedings of the International Body Engineering Conference, IBEC 2003,
27-29 of October 2003, Shiba, Japan.

Fast and economic stiffness evaluation of mechanical
joints
ABSTRACT
Car body structures and the joints
between beam members have a great
impact on global vehicle stiffness. With
the method presented in this paper it is
possible to experimentally assess the
stiffness of joints by a robust and
economic means.The stiffness of a beam
can easily be found experimentally just by
cutting it in two and using the cross-
sections to calculate the polar moment of
inertia. When it comes to a joint, there are
no formulae or expressions describing its
behavior. Therefore, measurement of its
mechanical behavior has to be made.
The dynamic joint method presented here
does not need levers or a costly, rigid set-
up, but an economical free-free set-up
and cast-on weights. Furthermore, the
same method can be emulated by FEM
when a digital model exists.
INTRODUCTION
During the design of space-frame
automotive structures the joints are of
great importance, governing up to 60% of
the global stiffness [1]. In order to design
a stiff and light auto body, a lot of effort is
spent on benchmarking global behavior:
measuring torsion stiffness and bending
stiffness. These measurements illustrate
the state of the art within the car body
field. But in order to identify why a certain
automotive body has a good behavior,
measuring the behavior of its parts, such
as beams and joints is needed. The
behavior of a beam can easily be found
just by cutting it in two and using the
cross-section to calculate the polar
moment of inertia. When it comes to
joints there are no explicit formulae or
1
JSAE Paper Number 20037025
SAE Paper Number 2003-01-2751
Nicklas Bylund
Volvo Car Corporation
expressions describing their behavior.
Therefore measurement of their
mechanical behavior has to be done.
The dynamic joint method presented here
does not need levers or a costly, rigid set-
up, but an economical free-free set up
and cast-on weights. Furthermore, the
same method can be emulated by FEM
when a digital model exists.
1. RELATED WORK
The impact of joint stiffness on a
structure’s overall stiffness has been
known for a long time. In the seventies,
Lubkin [1] showed that the overall
stiffness of frame structures is very
sensitive to the stiffness of the joints. He
chose to represent beams as 1D beam
elements and joints as Super elements,
using FEM (Finite Element Method) to
calculate the eigenfrequency of the
frame. Lubkin compared two frame
structures, one with the joints represented
as ideally stiff and one where the joints
were modeled as having a finite stiffness.
His results show a difference of 60% in
stiffness. He verified his results by
experimentally measuring the
eigenfrequency of a prototype. Later,
Moon et al, [2] evaluated three- legged
joints in the plane. They represented the
joint analytically with three rigid levers
representing the three legs. The levers
were connected to a common center and
torsional springs were put between the
three levers, creating a diagonal stiffness
matrix of size 3x3 with 3 independent
spring coefficients. Finally the joint was
measured statically in order to identify the
stiffness of the different springs. The
physical joint was made from straight,
prismatic, hollow beams welded something that we will see is beneficial to
perpendicular to each other. Furthermore, the cost of the experimental set-up.
the cross section of the beam was in the
order of 10x10 mm, thus very far from the 2.1. DIFFICULTIES WHEN MEASURING
size of automotive joints for which cross A JOINT
sections of 100x100 mm are not unusual.
As mentioned in the introduction, there is
A dynamic approach for evaluating joint no explicit formula to describe the
properties can be found in Wyatt et al’s stiffness of a joint. Some characteristics
paper, [3]. Two-legged, small-sized joints, make it difficult, furthermore, to measure
where also the movements out of the a joint’s stiffness. The two major
plane are considered; as in [2] a model of difficulties refers to:
the joint is made by rigid levers to a
common center and springs. But due to 1. The definition of the legs of the joint
the three-dimensional set-up the resulting 2. The experimental set-up
stiffness matrix has the size 6x6 with 6
independent coefficients. Furthermore, F
L
lumped masses much heavier than the Joint center
beam itself are attached to both ends of X
the joint. These masses are analytically
modeled as having a mass and a moment
of inertia creating a square, mass matrix
of size 6x6. By dynamically measuring
Welded plates on
the eigenvalues of the physical joint, with each leg ends
the two masses added and with free-free Deformation
measured
boundary conditions, the coefficients in Rigid floor
the stiffness matrix are found by inverse
modeling. Figure 1. Joint with unique center.
Considering an automotive joint, it is hard
2. THE FAST AND ECONOMICAL to evaluate where the center lies; it is
DYNAMIC JOINT METHOD even possible that the joint has more than
one center, see figure 2. In order to use
The methods presented in related works
the approach of Moon [2] and Wyatt [3],
all have their virtues and drawbacks. In
levers must be defined. The levers are
Moon et al’s method, [2] each leg of the
rigid elements all attached at a common
joint has to be rigidly attached (grounded)
center by a frictionless coupling, see
one at a time, and a lever attached to one
figure 3. The joint stiffness is then
of the other legs, see figure 1. This
represented by springs between these
procedure has to be repeated for each
legs, see figure 3.
leg, so for a three-legged joint this means
three permutations. This is cumbersome
and the need for a rigid set-up makes it
costly. Furthermore, because the plates Center
are welded, only the outer skin of the joint
is attached, and reinforcement within the
sections is not attached: thereby
contributing less to the stiffness than they
actually do when included in the complete
car body.

The method applied by Wyatt et al uses

structural dynamics to evaluate the joint:

Figure 2. Shape of a realistic automotive joint

2 with no absolute center.

Figure 3. Joint modeled with rigid
elements and springs.
When the axis lines of all the beams do
not cross each other at one point, as in
figure 2, no unambiguous center can
defined. This makes it impossible to
represent the joint with levers and springs
as in figure 3. In these cases, it is thus
impossible to use Moon’s static method.
2.2. DESCRIPTION OF THE
DYNAMIC JOINT METHOD
2.2.1. Features of the method presented
In order to avoid the problems stated in
the previous section, the method
presented has the following features:
• No explicit levers are defined.
• Heavy masses are cast at the end of
each leg.
• A free-free boundary condition is used
The two last points are common to Wyatt
et al’s method. By casting the masses
onto the legs, any eventual reinforcement
will also contribute to the stiffness, see
figure 4. The free-free boundary condition
means that no rigid plates or grounding
are needed. To cut out the joint, a sphere
of 250 mm diam. is laid over it with the
center of the sphere at the approximate
mid-point of the various joint centers, see
figure 2. The legs are then cut where the
surface of the sphere intersects them.
The robustness and accuracy of this
method has been shown to be good [4].
2.2.2. Principle of the mass lumping
In the method presented in this paper, as
well as in Wyatt et al’s method, a special
form of mass lumping is used. The idea is
that in order to access the global modes,
3
i.e. where the joint moves globally, and
not only the sheet metal vibrates, heavy
masses have to be added. For a
prismatic beam, for example, there exist
six global modes: torsional, longitudinal,
first bending in the plane, s-bending in the
plane, first bending, first bending out of
the plane and s-bending out of the plane.
The form of all the global modes can also
be made statically by bending the beam
or joint at it’s ends; local modes, on the
other hand, can not be made by statically
bending the beam or joint at it’s ends. In
fact the number of global modes equals
the number of degrees of freedom minus
the six rigid body modes. So for a three-
legged joint this means a total of twelve
global modes. For car body design, the
first three or four global modes are the
most interesting, as these are similar to
the joint’s movements in the car body,
such as forward-backward and inward-
outward movement of the legs. The mass
matrix of the joint itself is unknown and
hard to find. If heavy masses with known
mass and known moment of inertia are
added at each joint end they will
contribute only to the mass matrix’s
diagonal. Therefore, if the added masses
are heavy enough, the contribution of the
non-diagonal elements will be negligible,
i.e. the joint’s own mass does not
influence the dynamic behavior much.
2.2.3. Dynamic equation governing the
joints
The relation between stiffness and mass
can be seen in equation 1. The solution of
equation 5 gives the eigenmodes and the
eigenfrequencies under the form of
equation 2. The unknown in our case is
the stiffness matrix K; the dominating
diagonal masses of matrix M are known.
Therefore by measuring the
eigenfrequencies and the eigenmodes, a
picture of the joint’s stiffness is obtained.
The lowest frequency and its
corresponding eigenmodes show the
weakest direction of the joint. The next
frequency and mode show the second
weakest direction and so on. As
mentioned earlier, it is the first 3-4 modes
that are the most interesting. The
measurements are done by adding
accelerometers at several locations on
the joint and the masses, and exciting the
structure with a hammer equipped with a
force transducer. From the resulting
transfer functions (H=F/a) the
eigenfrequencies and eigenmodes can be
found. It should be remembered that from
the knowledge of the mass matrix,
eigenfrequencies and eigenmodes alone:
although theoretically possible to
calculate the stiffness matrix it is not
practically feasible. The problem is an
inverse one that is very ill-conditioned;
see Berman, [5].
Figure 4. An automotive joint set-up according to
the method presented.
4
eq 1. M ⋅ X + D ⋅ X + K ⋅ X = 0
M = mass matrix,
D = damping matrix
K = stiffness matrix
X = vector of deg rees of freedom
substitutions :
eq 2. X = R ⋅ sin(ω ⋅ t )
X = ω ⋅ R ⋅ cos(ω ⋅ t )
eq 3. X = −ω 2 ⋅ R ⋅ sin(ω ⋅ t )
R = eigenvectors
ω = eigenfrequencies
eq 4. λ = ω 2
2 , 3 and 4 in equation 1 leads to :
eq 5. det( K − λM ) = 0
Which can be solved if K and M are
known.
2.2.4. Examples of use of the method
The method can be used on different
occasions: to test different joint
reinforcement, for example. Volvo Car
Corporation (VCC), is interested in the
influence on stiffness when introducing
bulkheads in joints. Two joints were
made, one without bulkheads and one
with; both were tested with the dynamic
joint method, see figure 5a, 5b. A Finite
Element Model (FEM) was also built and
subjected to the same free-free boundary
conditions, see figures 5c and 5d. The
frequency difference between the
experiments and the FE-models in this
case was 7%. Other comparisons
between FE-results and experiments
have shown differences of 7%-15%. It
should be pointed out that it is much
easier to see the mode shape when the
results (see figure 5,a and 5 b) are
animated rather than as frozen images.
2.3 USE OF THE METHOD IN THE
BENCHMARKING PROCESS

The increased use of aluminum in car

body design has made joint design even
more important. In such designs, more
than ever before, the joints are used as
elements to connect beams. At VCC
joints from the aluminum bodies already
on the market have been analyzed giving Figure 5a. Joint without bulkheads,
some valuable input on joint stiffness in eigenfrequency 85Hz
cast aluminum joints. The alternative:
using the dynamic method would have
required extensive rigid tooling and raised
the almost impossible task ofmeasuring
joints with no geometrical center.

CONCLUSIONS

The proposed dynamical joint method is a

fast and economical means to assess the
stiffness of joints. The method permits all
type of joints to be measured, regardless
Figure 5b. Joint with bulkheads,
of whether an unambiguous joint center Eigenfrequency 185 Hz
can be found or not. Internal
reinforcement contributes to stiffness, and
because of this cast-on weights are used
instead of welded plates. Different joints
can be compared if the same masses and
the same cutting lengths are used. A
drawback of the method is that the static
stiffness cannot be explicitly found
because of the ill-conditioned inverse
problem it presents.

Picture 5c. Joint without bulkheads

Picture 5d. Joint with bulkheads

5
ACKNOWLEDGMENTS to Measure Joint Stiffness.”
Mechanical systems and Signal
The dynamic measurements performed Processing (1999) 13(5), pp 789-801.
by K.G. Johansson at VCC are gratefully
acknowledged. The discussions of the
4. Paulo Urzua and Lou Chouping
experimental set up with Jonas Forssell, “Parametric Study of the T-joint
also at VCC, are gratefully Stiffness of an Aluminium Car Body”
acknowledged. Master Thesis 1999, Chalmers
University of Technology, Gothenburg,
REFERENCES Sweden.

1. James L. Lubkin “The Flexibility of a 5. Beerman, A. System identification of

Tubular Welded Joint in a Vehicle Structural Dynamic Models –
Frame” SAE 740340, pp1518-1522. Theoretical and practical Bounds 84 -
0929
2. Moon Y.-M, Jee T-H and Park Y.-P
“Development of an Automotive Joint
Model Using an Analytically Based CONTACT
Formulation”
Journal of Sound and Vibration (1999) Nicklas Bylund, Tech lic and MSc
220(4) pp625-640.
Email: nbylund@volvocars.com
3. Patricia J. Wyatt Becker, Robert H.
Wynn, Jr.,
Edward J. Berger and Jason R. Volvo Car Corporation
Blough “Using Rigid-Body Dynamics

6
 Paper C3
Field Method for Torsional Stiffness Measurements of Complete Vehicles
In the proceedings of the International Body Engineering Conference, IBEC 2003,
27-29 of October 2003, Shiba, Japan.

Field Method for Torsion Stiffness Measurement of
Complete Vehicles
ABSTRACT
The following paper describes how to
measure the global torsional stiffness of a
complete car under field- like conditions.
All that’s needed are lifting devices, two
stands of equal height, three glide planes
or equivalent, three scales and two
inclinometers, a spirit level, some pieces
of aluminum and a glue gun. The results
from four measured cars are presented
and a comparison is made with values
obtained with laboratory equipment and
data from manufacturers. The method is
a fast and economic means to find the
most interesting cars that then can be
selected for measurement by traditional
methods, giving the stiffness as a
function of the vehicles long axis, and
thus minimizes the cost of benchmarking.
Time for measuring one car with all
equipment readily available and with
personnel having some experience of the
method is about two hours. Only the sway
bars have to be disconnected. Absolutely
no damage to the measured car means
that rented cars can be used.
1. INTRODUCTION
1.1. BACKGROUND
The experience of different benchmarking
studies has shown that finding realistic
values from other vehicles is tricky. This
experience provided the inspiration for
developing this measurement method:
making it possible to find the global
stiffness of a car body in an economic
and easy way. Furthermore, the method
can be used when there is a need to
1
JSAE Paper Number 20037028
SAE Paper Number 2003-01-2754
Bylund, N., Fredricson, H.
Volvo Car Corporation
monitor a car' s eventual change in
torsional stiffness, e.g. if a glued car
body’s stiffness changes with time and
use.
1.2. METHOD
The object of the method is to make it so
easy that it can be done anywhere there
is a vehicle lift available. The motivation
has been to make it possible to take
measurements without any damage to
the vehicle. The equipment is easy to
move to any location where there is a
vehicle lift.
1.4. LIMITATION
During the measurement, the measured
angles are in tenths of a degree so for
good accuracy the inclinometers must
have a precision within thousandths of a
degree, thus limiting this source of error
to a few percent. The scales should be
able to measure weight within in a few
kilos. The installation of the measurement
equipment should be done with care to
avoid unwanted errors due to bad
mounting. From measurements done on
a Volvo S80, good agreement with earlier
values has been obtained with an error of
approximately 4 %.
2. MEASUREMENT DESCRIPTION
This section explains the equipment,
vehicle setup and mounting of
measurement instruments.
2.1. EQUIPMENT used to read the twist angle directly. As
only the relative angle between the front
The major advantage of using this and the rear of the vehicle is measured in
measurement method is the little amount this method, only the global torsional
of equipment used. stiffness of the vehicle is obtained. The
method could therefore be used to
All the needed equipment is shown in the compare vehicles in the early
following photos. benchmarking process, as the stiffness
as a function of the vehicle' s long axis is
not measured.

2.1.2. Glide plane (Fig. 2.3)

During loading of the vehicle the wheels

move upward, sideways and forward or
Fig. 2.1. Transformer and Fig. 2.2. Inclinometer backward. This motion has to be free to
Voltmeter used with the
Inclinometer
avoid parasite loading into the vehicle, i.e.
we do not want any side forces into the
wheels.

2.1.3. Scales (Fig. 2.4)

There has to be scales to measure the

load under at least three wheels, to cater
Fig. 2.3. Glide plane Fig. 2.4. Scales
for front-heavy and rear-heavy vehicles.
This load is later used to calculate the
torque applied on the vehicle.

2.1.4. Stands (Fig. 2.5)

During the setup two diagonally opposed

wheels will be placed on fixed stands.
See the vehicle setup in section 2.2. The
stands should be of equal height.

2.1.5. Forklift (Fig. 2.6)

Fig. 2.5. Stand Fig. 2.6. Forklift
During the setup, the forklift'
s forks will be
placed under one wheel: either a front
wheel or rear wheel depending on the
weight distribution of the vehicle. See the
vehicle setup in section 2.2. If a forklift is
not available, an adjustable stand could
be used.

2.1.6. Vehicle lift (Fig. 2.7)

Fig. 2.7. Vehicle lift
2.1.1. Inclinometer (Fig. 2.1; 2.2)
Torsional stiffness is the relation between
applied torque and amount of twist the
vehicle undergoes. The inclinometer is
2
To be able to locate all the equipment
under the vehicle a regular vehicle lift can
be used. Important: the vehicle lift should
NOT be of the type that directly lifts the
car’s wheels, but the type that lifts the car
at its jacking points! This also makes it
possible to study the body structure and
use the information in the analysis of the In the rear, some problems can occur;
structure. interference with the suspension has to
be avoided. As the suspension moves a
2.2. VEHICLE SETUP lot it has to be checked during the first
loading that there is no such interference
The car is lifted and then placed upon the between the suspension and the
two diagonally opposed stands. Glide inclinometer attachment. It is best to
planes have to be placed on the stands attach the inclinometers to the rear side
because the wheels translate upward, beams, as close to the spring towers as
sideways and forward/backward when the possible, but if they are not easily
suspension is compressed. accessible and the sub-frame is rigidly
attached to the side beams the latter can
Scales have to be put under at least three be used.
of the wheels: If the car is front-heavy
scales have to be put under each front
wheel; if the car is rear-heavy scales
have to be put under each rear wheel. It
is a good idea to measure the balance of Support
Beam
the car before placing the scales. The
Hotmelt
end of the car that has scales under both Glue
left and right wheel should be absolutely
horizontal. This is done by using an
adjustable stand or placing the scales on
the forks of a forklift under the wheel that
is on the same side as the fixed stand at
the other end of the car. (See figures 2.8 Fig. 2.10. Front-end attachment Fig. 2.11. Light beam with
and 2.9 below, of a front-heavy car). to suspension tower. inclinometer attached mounted
on front-end attachments.

Fig. 2.8. Front wheels. Front
right: Fixed stand with scales
and glide plane. Front left:
forklift with scales.
Fig. 2.9. Rear wheels. Rear left:
Fixed stand with, Scales and
glide plane. Rear right: In the
air. (See the twist!)
Fig. 2.12. Rear attachment up Fig. 2.13. Rear attachment to
to rear side beams. rigidly mounted sub-frame.

2.3. INSTALLATION OF
INCLINOMETERS
3. MEASUREMENT PROCEDURE
To obtain good readings; the
inclinometers should be attached to light When the stands have been placed
beams which are then put onto short under the wheels and the inclinometers
vertical supports attached to the are mounted, the following steps have to
suspension towers in front, and to the be carried out.
rear side beams in the rear, see 2.10-
2.13.To mount the attachments at the 1. Mount the supports on the front
front end is usually easy; the suspension suspension towers; see fig. 2.10 and
towers are easily accessible. fig. 2.11.
2. Put the lift under the car. Make sure
that the lift supports the car at

3
 symmetrical locations and at the same
time, to avoid parasite torques. Lift the
car.
3. Put the four scales on the floor and
measure whether the car is front-
heavy or rear-heavy. Lift the car
again.
4. Mount the supports at the rear side
beams, making sure that they are not
too close to the moving parts of the
suspension. See fig 2.12 and fig 2.13.
Observe that soft rust protection must
be taken away otherwise the supports
may slide!
5. Place the fixed stands with scales and
glide planes diagonally under the left
rear and right front wheels. See fig
2.5. Move the forklift to the heavy end
of the car and lower the forks below
the level of the fixed stand. See fig 2.6
6. Lower the car slowly and adjust the
forklift so that the end of the car rests
horizontally (use a spirit level) on the
fixed stand and forklift forks. Finally,
the car should be supported only on
the stands and forklift. The car may
roll a few cm but if the lift is adjusted
so that the car is horizontal it will not
roll much. Fig 2.8 and 2.9.
7. Check so that no part of the rear
suspension touches the inclinometer
support at the rear. Look also for
interactions between the lift and the
floor.
8. Lift the car again so that it is only
supported on the lift. Zero the scales
and the inclinometers.
9. Now it is time to start measuring.
Lower the car slowly down on the
stands and the forklift. Check that the
stands are placed exactly under the
wheels. Check: readings on the
scales, angles, and level. Repeat
several times. The first readings
should be ignored, as the car has to
be set the first time.
10. Test with sway bars mounted and
demounted. A sway bar attached on
the overhang side reduces the
stiffness while a sway bar attached
towards the middle of the cars
augments it.
4
11. Eventually test with open and closed
doors.
4. DATA ANALYSIS
In order to calculate the stiffness value
from the measured values the following
procedure is used.
Torque calculation:
The resultant moment on the car has to
be zero because the car is at rest.
On the side that is just supported at one
point we have the torque:
b1
M 1 = m1 ⋅ 9.81 ⋅
2
m1 = mass on the scale
b1 = track width on side with one scale
On the side that has both wheels
supported, we have the torque:
b2
M 2 = ( m 2 − m3 ) ⋅
2
m2 = mass on left scale
m3 = mass on right scale
Important condition; both torques should
be equal. Anyhow, a difference of 2-3 %
can occur depending on the precision of
the scales and accuracy of the
measurement of the track width.
absM 1 + absM 2
M =
2
M = mean value of torque
Calculation of stiffness, K:
M
K=
α
α = measured torsion angle
5. METHOD CALIBRATION The values from three other vehicles
measured can be seen in the appendix.
To check how the method complies with
values from measurements done by 6. CONCLUSIONS
Volvo Car Corporation (VCC) we have
used the Volvo S80 as a reference With the presented method the global
vehicle. A torsional value of 18.6 torsional stiffness of a vehicle can be
kNm/deg has been measured at VCC, measured under field-like conditions
with an advanced method, and this using only a vehicle lift and some easily
serves as a reference value to our transportable equipment. The time
measurements. From the measurements needed for measuring one vehicle is
done with the field method, we achieved about two hours. The difference between
a torsional stiffness value of 19.2 the field method and the advanced
kNm/deg without sway bars and 23.7 method used at VCC lies within 4%. The
kNm/deg with sway bars. As mentioned difference between the field method and
above in the report the sway bars have a non-verified results from other sources is
great influence on the torsional stiffness. about 10%. Furthermore, the method can
The values from measurements done be used when there is a need to monitor
without sway bars are closer to the a car' s eventual change in torsional
reference measurements, as they are stiffness, e.g. if a glued car body’s
done without any driveline. From these stiffness changes with time and usage.
values it can be seen that the new The speed of the field method, together
method is within 4 % of the reference with the small amount of equipment and
value for the Volvo S80. See fig 5.1. The the non-destructive character, makes it a
maximum error from measurement of good complement to more advanced
three other cars is hard to evaluate, as no methods when making a scan of several
measurements with the advanced vehicles on the market.
method could be made on these vehicles.
In the appendix, some comparisons
between other sources of data outside
VCC and results from the field method
are made: the differences are around
10%.

Torsional stiffness Volvo S80

30000
Values from measurements with no
swaybar 3.3 % higher then ref. test 1
25000
test 2
Torsional stiffness (Nm/deg)

test 3
20000 test 4
Series5
15000 test 5
test 6
test 7
10000
test 8
Series11
5000
reference

0
Swaybar No swaybar Ref.

Figure 5.1

The method also shows great

repeatability, within +/- 100 Nm/deg for
values up to 10000 Nm/deg.

This result is satisfactory, as the method

is so rapid and economic to apply.

5
ACKNOWLEDGMENTS APPENDIX

We acknowledge the help from Shaylor A1. CHRYSLER SEBRING

Duncan at VMCC (Volvo Monitoring and
Concept Center) for the design and
manufacture of the necessary stands,
supports and beams for the
measurements.

REFERENCES

[1] Ford: Static Torsion and Bending

Test – Body and Vehicle,
Engineering Standards & Systems
Engineering (ESSE), EATP:01.01- Chrysler Sebring
16.01

[2] Forssell Jonas: Mätning av

vridstyvhet – komplett bil, Volvo
Values from measurements
Car Corporation, dept. Painted Stifness values from measurements with no swaybar appr. 12 %
lower then Benjamin Yilma.

Body Engineering, 1995, LM – 10000

Ref.
132930
9000
8000
test 1
7000
test 2

[3] Mikael Holmsbo: 6000

5000
Swaybar Benjamin
Yilma.
test 3
test 4
No swaybar
Metodutvecklingsprojekt – mätning 4000
Series5
test 5

av vridstyvhet, kompl. Bil, Volvo

3000
test 6
2000 test 7

Car Corporation, 1995, LM – 1000 test 8

Series10
132618
0
Benjamin Yilma
Car, Chrysler seebring
ref. From manufacturer

CONTACT

Bylund, N., Licentiate of Engineering Average values from measurements of

Department of Advanced Body Engineering, Chrysler Seebring:
93710, Volvo Car Corporation
SE-40508 Gothenburg • Authors, with sway bar: 4780 Nm/deg
Tel: 46-31-325 4145 • Authors, without sway bar: 3898 Nm/deg
E-mail: nbylund@volvocars.com • Value received from Ford: 4389 Nm/deg
Fredricson, H., Licentiate of Engineering
• Value received from the manufacturer:
Department of Advanced Body Engineering, 8800 Nm/deg
93710, Volvo Car Corporation
SE-40508 Gothenburg
Tel: 46-31-325 9707
E-mail: hfredri1@volvocars.com A2. CHEVROLET CORVETTE

Chevrolet Corvette

6
 Values from measurements Values from measurements
Stiffness values from measurements with no Stiffness values from measurement with no swaybar appr. 6 %
swaybar appr.15 % higher then higher then Benjamin Yilma
ref from manufacturer
9000
Ref. No swaybar
12000 No swaybar 8000
Swaybar Swaybar Benjamin Yilma.
test 1
7000 test 1
10000 test 2
test 2
test 3 6000
8000 test 3
test 4 Series4

kNm/deg
5000
Series5 test 4
6000
4000 test 5
test 5
test 6
4000 test 6 3000
Series8
test 7 Benjamin Yilma
2000
2000
test 8
1000
0 Series10
Car, Chevrolet,corvette ref. manufacturer 0
Car,S2000

Average values from measurements of Average values from measurements of

Chevrolet Corvette: Honda S2000:
• Authors, with sway bar: 10468 Nm/deg • Authors, with sway bar: 6649 Nm/deg
• Authors, without sway bar: 11051 Nm/deg • Authors, without sway bar: 7525 Nm/deg
• Value received from the manufacturer: 12570 • Value received from Ford benchmarking: 7121
Nm/deg Nm/deg
• Value received from the manufacturer: not
available
A3. HONDA S2000

Honda S2000

7
 Paper D
A study of the use of Engineering Design methods in an automotive company
A study of the use of design methods in an automotive
company

!" ! # " $ % &'() * +

) ,- $. /.0 "1 , 2' 34 , " 565

Abstract
! "#$% #&' ( ")% % (* + &+,# + -$+ % ,&% . ,&% ! & ,$ % + "# +
!/&," $ %) "+* "$ ( )--, ". # $(0," ' $-"# + &+,# "$ ) % + "& % "#
) & % !/&," $- % ! "#$% "# ,$ " 1" $- & / ,-, ,$!/& * 2$3'$ &+
$+/$+&" $ 42 5 (* % ,+ ( "# ( #&' $)+ $- + + 3&" $ "$ ! "#$%
"# + (*/+$' % #" "$ #$6 + &+,# + ,& )//$+" % + . # %&"& /+ " %
"# + )3" $-7)&3"&" ' + &+,# ,&++ % $)" %)+ -$)+ * &+ &" 2 6# + "6$ $-"#
&)"#$+ 6 + 3$,&" % & + &+,# + . "# ,$ , /" /+ " % # + " #$/ % "#&"
+ &+,# + ,& ( "" + ) % + "& % "# '&3) $- % ' 3$/ % )//$+" " ,# 7)
% )//$+" "$$3 & % "# /+$, $-&%$/" $ $-! "#$% (* %) "+*.
6 7 8 " 8 " # 8 8#" 8

1. Introduction
" &! 6 "# & +& $- / ,&3 % 8 $63% &+ ,$!!$ 3* -$+! % "$ ) % +"&8
% /+$%)," % ' 3$/! ". (* " &! '$3' &," ' " ),# & "+&" *
% ' 3$/! " * " ! $3)" $ % ' 3$/! " ,$ , /" 3," $ /+$(3! + $3)" $ ",.
4 9/ : & & % #$!/ $ ;;<5 $! $-6# ,# &+ ") " ' 3* /+&," %. $!/&
+&33* - 3 ,)+ &($)" "# --, ,* $-"# + ") " ' % /+$, %)+ / ,&33*
"# &+3* "& $-% 6# + "# 3' 3$-) , +"& "* & % ,$ 7) , $-% , $
&+ # #.
# )! +$) % ! "#$% $-"# /& " - 6 % ,&% +&" % (* &,&% ! & #&' "#
/$" " &3 "$ )+ --, ,* " &! % = #$6 ' + "# + !/&," $ %) "+*
,$ % + % )--, " 4 +8#$- + " &3. ;; = 33& % + ;; = $#&" 8* ;; 5.

2. State-of-the-art in use of design methods

() %& " 3" +&")+ "# >;? & % @;? #$6 % "#&" "# ( " / +-$+! !& )-&,")+
A&/& ,$!/& ,$)3% ,+ % " /&+" $- "# + ),, "$ "# + &"" " $ "$ %
$/ +&" $ 4B# " * @>>= B$!&,8 " &3. @@;5. 3"#$) # &"" !/" "$ !/+$' %
$/ +&" $ !) " #&' 1 " % , / $/3 "&+" % "$ % , ")+ & $ "# "$/ , #&
( + &+,# % "# 3& " % ,&% 4 3 ;; 5. 2&+ $) - 3% #&' ,$ "+ ()" % "#
+&" $ $-! "#$% 6 "# & -$,) $ % . .% , , 7)&3"* !& & ! "
,+ &" ' "* ",. # &)"#$+ $-"# /&/ + %$ + &+,# "# % , , - 3%.
, , ! "#$% /3&% -$+ +&" $ &3 &" $ & % * " !&" &" $ % . "#$%
/ +-$+! % -- + " "& 8 "# % -- + " "& $- % 6 + ,$33," % 4 ) # @@ =
A$ @@ = &#3& % ": @@C5. "# 3&" @;? "# 3" +&")+ + 6 &($)" 6#* %
, , ! "#$% 6 + ( $+ % (* %) "+*4 % + @@>= D+$ " @@@5 6# + & A&/&
! "#$% 6 + ) %. ")% $ "# ) $-% ! "#$% %) "+*/+$3- +&" %. +&)0 $
" &3. 4 @@C5 ! & )+ % "# '&3) $-! "#$% & & / +, "& $-,$!/& 6#$ / +, '
"#&" & ! "#$% "+$ 3* ,$ "+ ()" "$ /+$%)," 7)&3"*. $! &)"#$+ % % "# +&3
( 3 -"#&" -$,) $ % $/ +&" $ )&+& " "# () ),, $-& ,$!/& *.
& "&! & 4 @@@5 -$+ 1&!/3 "&" "#&" E% )//$+" " ,# 7) %$ $" 3&% "$
% " +! " , & % /+ % ,"&(3 -- ," , "# ( -" "# * % 3' + "$ "# -+!
,$ % " $ % (*& +&"# + ,$!/31 "6$+8 $- " +&," $ ( "6 & " F. + ,$!! %
'&3)&" " ,# 7) (* ,$ % + "# + -+! / ,-, & + &" $ . $+ & / ," "$ "#
),, -)3 !/3! "&" $ $- ! "#$% 1 " "#& "# -) ," $ &3"* $- ),# " ,# 7) .
3 " &3. 4 @@>5 /+$/$ % & &//+$&,# "$ % &36 "# "# ,$!/31 "* /+ % ," "#
-- ," $- % + &+,# (* "&(3 # ,&) -- ," + 3&" $ # / ( "6 ),,
,+ " + & . . () + 3&" % ! & )+ %$6 "$ ! & )+&(3 ,+ " + & ,3$ "$ "# /+$%),"
% ' 3$/! " &," ' "*. & 3$ ,&3,#& "$ #$6 ,&) & % -- ," &3$ % ,+ ( %
4 @@<5 & & +&3&," $ + &+,# "+&" *.
' +* ,#& & ,$!/& * ,)!( + $! & % !),# --$+" #& "$ ( ' " % -$+ &
,#& "$ $,,)+ & % ( ) "& &(3. "# - 3% $- !& & ! " + &+,# #& (
,$ %)," % "$ ) % + "& % "# /# $! $ $- + "& , "$ ,#& 4D$+% D$+% & %
, &!&+& ;; = $"" + @@G= &)+ + @@C5. $6 ' + !$ " &)"#$+ 6 "# "# %
, , - 3% &+,# -$+ 6#* "# + & 3&,8 $- !/&," $-% , , ! "#$% %$ $"
,$ , "+&" $ +&3+ & $ &($)" 6#* ,#& % --,)3" "$ !/3! " ()" $ "#
"+ ,+ & $ 6 "# "# % ! "#$%$3$ * - 3% "#&" !&8 % ! "#$% % --,)3"
"$ !/3! ". 3 4 ;; 5 -$) % "#&" E!& * $- "# + &+,# + ! "$ 6$+8
H $3&" $ ? $" ' " &" &,")&3 %) "+ &3 % F 6# 3 " !/-3 & % &%8 ,#&)(
4 ;; 5 + !&+8 % "#&" E"# $+* () 3% & % + &+,# ,$ %)," % ) % + "# $+!&" ' "+&
#& $-" 3," % "$ 3$$8 &" 6#&" / $/3 &,")&33*%$ I !/3*/+ ,+ ( & ! "#$%$3$ *
!&* $" ! " "# % $-% + H$)" "# + ?F.

3. Objective of the paper

# $(0," ' $-"# + &+,# "$ ) % + "& % "# ) & % !/&," $-% ! "#$% "#
,$ " 1" $-& / ,-,,$!/& * 2$3'$ &+ $+/$+&" $ 42 5 (* % ,+ ( "# ( #&' $)+
$- + + 3&" $ "$ ! "#$% & % "# + (* /+$' % #" "$ #$6 + &+,# +
,& )//$+" % + . "# ,$ , /" /+ " % # + " #$/ % "#&" + &+,# + ,&
( "" + ) % + "& % "# '&3) $- % ' 3$/ % )//$+" " ,# 7) % )//$+"
"$$3 & % "# /+$, $-! "#$% &%$/" $ (* %) "+*.

4. Research method
$ ) % + "& % "# ) $-! "#$% & ,$!/& * " , &+*"$ ")%*"# /+$(3! -+$!
% -- + " / + / ," ' . +"& 3* + ,& ( & 8 % &($)" "# ! "#$% "# * ) & %
"# + ' 6 $ ! "#$% /+$' % " + " !&" + &3-$+ + &+,# + . $6 ' + & + &"
&!$) " $-"# + 8 $63% "&," & % % --,)3" "$ 1/3& . $+ $' + % + 4 @@>5 ) "
"#&" % + #&' & '$,&()3&+* + 3&" $ "$ "# % &," ' "* "#&" $" ,3$
&3 ! " 6 "# "# &( "+&," $ & % '$,&()3&+* $-+ &+,# + . # ,& + )3" % +
&3!$ " $" ( &(3 "$ + ,$ "# % ,+ /" $ $-! "#$% &( "+&," " +! & & "
"# ! "#$% & % &," ' " 6# ,# "# * + /+ " 4D+$ " @@@5. ( +'&" $ $- %
/+&," , & % & &3* $- %$,)! " % ! "#$% ) % ,& /+$' % 6 #" "#&" "#
+ &+,# + 6$)3% $" - % -$ 3* + 3* $ /+ -$+!)3&" % 7) " $ (& % )/$ "#
"# $+ " ,&3-+&! 6$+8 4 )($ & % &%% ;; 5. $6 ' + "# "*/ $-+ &+,# ,& &3$
( 3! " , 6#&" #&// & /&+" ,)3&+ ,& !&* ( $- ,% "&3,#&+&," + $+
,$++ 3&" % "$ ,$!/& * -&,"$+ &" & # # + 3' 3 & % ,& $" ( +& / % (* & ! +
$( +'&" $ $+ %$,)! " & &3* . , & " $-,$!/3! "&+* + &+,# ! "#$% &+
,$!( % "# /&/ + & ,$ %)," % (* "6$ + &+,# + 3$,&" % %) "+* #&+ "#
&! /#* ,&3&"!$ /# + ,)3")+ & % '$,&()3&+* $- + 4 3$!( + " &3. @@ 5
& % $ )/ +' $+ 3$,&" % &" & ) ' + "*. + & )3&" % ' % , &3$ & + & $ "$
,$!( + &+,# &//+$&,# & ) " % (* 4 @@<5.
# -+ " (3$,8 $- + &+,# 6& & " +' 6 ")%* ,$ %)," % (* "6$ $- "# &)"#$+
( "6 $' !( + ;; & % D (+)&+* ;; 6 "# "6$ !& $(0," ' J "$ ) % + "& % "#
$'&" $ /+$, &" "# /+$0,"? ! 3 ) 3' 3& % "$ ) % + "& % "# ) $-! "#$%$3$ *
&" "# /+$(3! $3' 3' 3. # ,$ % /&+" + /$+" % "# &+" ,3 "#$) # "# -+ "
/&+" /+$' % " &3 -$+!&" $ &($)" "# ,$ " 1" 6# ,# "# ! "#$% &+ ) %. D$+
-)+"# + -$+!&" $ &($)" $'&" $ /&+" "# + &% + + 7) " % "$ + - + "$ 9/ : &
4 ;; 5 & % 9/ : & & % #$!/ $ 4 ;;<5. -" + & ,$ % (3$,8 $-+ &+,# "6$
$"# + &)"#$+ $-"# /&/ + -$33$6 % 6 "# & $( +'&"$+* & % & " +' 6 ")%* $-"#
/+$, $- ,$ , /" 3 ," $ -$+ &($)" 1 !$ "# ;; & % ;; . &+" "#+ $- "#
+ &+,# % ,) "# & &3* $-& )!( + $- 1&!/3 $-%$,)! " % ! "#$% ) %
"# ,$!/& * -$+ ,$ , /" 3," $ &"# + % $' + ' +&3* &+ (* $ $- "# &)"#$+ .
# "#+ / , $- + &+,# &+ ,$!/3! "&+* ( ,&) "# * ,$!( " +' 6
$( +'&" $ & % %$,)! "&" $ & &3* . "# -+ " (3$,8 $-+ &+,# "# + &+,# +
7) + % &($)" % ! "#$% ) % (* + -+$! & '&+ "* $- % /&+"! " . #
,$ % / , $- + &+,# ,$ , "+&" $ "# ($%* % /&+"! " & % $ "# /+$, $-
3," $ $-,$ , /" 4"# 6$+% ! "#$% 6& $" ! " $ %5. # "# +% / , $-+ &+,#
,$ , "+&" $ ! "#$% ) % -$+ ,$ , /" 3," $ & % & &3* $-"# + )3" $("& % (*
% " &! 6 "# "# ! "#$% . "6 ;; & % ;;< "# &)"#$+ #&' #&% &,, "$
" + &3 "+)," $ ,$ " )$) %) "+ &3- % (&,8 -$+ "# + % & & % #&' #&+ % "#
/#* ,&3& % ,)3")+&3 ' +$ ! " 6 "# "# + 4 3$!( + " &3. @@ 5. , "#
+ &+,# ! "#$% -$+ "# 6$+8 ,& ( )!!&+ % & & ,$!( % " $-+ &+,# ! "#$%
,3)% " +' 6 /3& % $( +'&" $ /$ "& $) $( +'&" $ & % %$,)! "
&+,# -+$! % "# ,$!/& * /+$' % & /+$ + ' ) % + "& % $-"# /+$(3!.

5. Results
# + )3" #&' ( $+ & % &,,$+% "$ "# "#+ /+ ' $) 3* ! " $ % + &+,#
(3$,8 ' :. % ! "#$% ) % "# /+$, $- ,$ , /" 3 ," $ & % & &3* $-
%$,)! " % ! "#$% ) % -$+ ,$ , /" 3," $ . # + &+,# ! "#$% ) % "$ $("& "#
%&"& &+ 3&($+&" % -)+"# + &,# (3$,8.
5.1 Research block one: Design methods used as reported by
interviewees to an in-company researcher
-" + $ * &+ &" "# ,$!/& * &() %& " -$+!&" $ &($)" "# ) $- ! "#$% 6&
&"# + %. $6 ' + !$ " % & &($)" 6#* ! "#$% #&% $" + &"3* !/&," % "#
,$!/& * ,&! -+$! "# &! + %), % )!( + $- + 6#$ 6 + -&'$)+&(3 "$
) ! "#$% . # * &+ # + + - ++ % "$ & -$+!& " . -$+!& " + ' &3% "#&" "# *
6 + ) &6&+ $- "# # # )!( + $- 1 " &,&% ! , % ! "#$% ( -$+ "#
+ &+,# /+$0," 6& " &" %.
$ &,# ' !$+ $(0," ' "* "# /+$(3! 1/3$+&" $ & " +' 6 ")%*6& ,$ %)," %
6 "# & + /+ "&" ' )!( + $- + -+$! % -- + " % /&+"! " . # $(0," ' 6&
$" "$ &"# + %&"& &($)" &33! "#$% ) % "# ,$!/& * ()" "$ " & ) % + "& % $-
"# "*/ $- ! "#$% ) % "# /+$, (* 6# ,# "# * &+ &//3 % & % "# 3' 3 $-
&" -&," $ $- + ) "# ! "#$% . # % & 6 "# "# " +' 6 ")%* 6& "$
- % & +$)/ $- + 6#$ ,$33&($+&" % ,3$ 3* $ "#&" #&+ % 8 $63% &($)"
! "#$% ,$)3% ( % " ," %. " "# &! " ! " 6& % +&(3 "#&" "# * + /+ " %
% -- + " % /&+"! " $ "#&" "# +$)/ 6& $" 3! " % "$ $ "*/ $-% "& 8. #
" +' 6 "&+" % 6 "# "#+ + -+$! "# ,#& % /&+"! " 6 "# 6#$! "#
+ &+,# + #&% $-" ,$ "&,". # *6 + & 8 % "$ ) " -$+ "# " +' 6 $"# + +
-+$! & * % /&+"! " "# * $+!&33* 6$+8 6 "#. - "# ) " % + 6 + -+$!
,#& "# * 6 + &3$ & 8 % "$ ) " $"# + + "#) )+ "#&" &33 +
6 + "# + -+$! "# ,#& % /&+"! " $+ $"# + % /&+"! " "#&" 6$+8 6 "# ,#&
+ )3&+3*. # )!( + $- " +' 6 + 6 "$ . 336 + % ' %)&3 1, /" -$+ $ 6# +
"6$ / $/3 6 + " +' 6 % !)3"& $) 3*. # " +' 6 6$+8 % &" '&+ $)
% /&+"! " & % % &3" 6 "# % -- + " & / ," $-"# /+$%)," % ' 3$/! " /+$, 4 &(3
5. # " +' 6 6 + + ,$+% % & % "+& ,+ ( %. % "& 3% 1/3& &" $ $- "#
" +' 6 ? "& 8 ' 4 9/ : & & % #$!/ $ ;;<5. + "# "& 8 &+ $ 3*
+ 3&" % "$ "# ! "#$% ) % % ,+ ( % -$33$6 ," $ .

D )+ . /&+"! " $-"# " +' 6 & % )!( + $-* &+ 6$+8 &" 2 4&%&/" %
-+$! 9/ : & & % #$!/ $ ;; 5

&"& & &3* /+ " % 1" % " -* "# ! "#$% ) % /&"" + ) & %
&%$/" $ $- ! "#$% "# ' 6 $- + $ % ! "#$% ) &( 3"* & % "#
!/$+"& , ' "$ ,$ , /" 3," $ .
-$+ "# " +' 6 "$$8 /3&, &,# " +' 6 6& ,$ "&," % ' & !& 3 " 3/#$ $+
/ + $ & % ' & " "&" ' " +' 6 7) " $ 3 " & % & 3 " &! % ! "#$%
"#&" ,$)3% ( ) % %)+ "# " +' 6 "$ # 3/ "# ! + ! !( + "# /$ (3 ! "#$%
"#&" "# *#&% ) %. # 3 " $-! "#$% /+ " % &// % 1 .
<
 # " +' 6 + /$+" % & )!( + $-! "#$% ) %. $! $,,& $ ! "#$% 6 +
% ,+ ( % & /+ ! % "&" % /+$, %)+ ) % (* & +$)/ -$+ & ,$ , "+&" % / + $% $-" !
"$ &,# ' & $&3. # &! ! "#$% ,$)3% + 7) + & )!( + $-,$ , "+&" % / $% $+
$ "$ ( &//3 % ()" "# +&3/3& $ #$6 "$ /+$, % 6$)3% + !& ) "$),# %.
# ! "#$% &+ ,&33% 1/3," ! "#$% "# /&/ +. $"# + $,,& $ ! "#$%
6 + % ,+ ( % & & ,#& $-&," $ $,,)++ &" % -- + " $,,& $ 6 "# & # # " % ,*
"$6&+% $//$+") " , % & % ,+ ( % (* 2 + 4 @@<5 & % % + & % 3
4 ;;<5. # * &+ "*/ ,&33* ) % +"&8 (* % ' %)&3. "# + " ! "# ! "#$% % ,+ ( %
(* "# " +' 6 6 + & "& %&+% % ,$!/& * /+ ,/3 $-6$+8 +&"# + "#& &
" ,# 7) &" "# /+$(3! $3' 3' 3. # $(0," $- ")%* "# (3$,8 $-+ &+,#
1/3," ! "#$% . 1&!/3 $- $//$+") " , " / & % "& %&+% % ,$!/& *
/+ ,/3 $-6$+8 &+ /+ " % &(3 . + ) 1/3," ! "#$% ,$)3% (
% ,+ ( % & ) & /+$, $+ " % &//+$&,# % 6# + & + ) & ,#& $-
" / ,$)3% ( % ,+ ( % & $&3$+ " %.

1&!/3 $-$//$+") " ,,#& $- " / 1&!/3 $- "& %&+% %

,$!/& * /+ ,/3 $-
6$+8
H ,) $ 6 "# ,3' + -+ % ? 7) + ! " +&, &( 3"*
D$+ & /+$0," $ " +' 6 + /$+" % & & +4 5
% ,) "# /+$(3! 6 "# $! ' +* " & +&3 "$$3 -$+
$'&" ' ,$33& ) . % ,% % $" ' " "$+ & % ,$ ,"
"# ! "$ "# (+& "$+! $ ( ,&) + 7) + ! " . # "$$3
"# * 6$)3% % +)/" ". " &% # ! + 3* 6& !/3! " % &" "#
% ,) % "# % & 6 "# "# ! "$ (+$&% # ,$!/& * 3' 3 & % "
' 6 $ "# /+$(3!. # # (+& "$+! % !/3! "&" $ 6&
" 6 "# $"# + ,$33& ) . B# "6$ $+ "#+ &,,$+% 3* "$/ %$6 . .
$3)" $ ! + % -+$! "# (+& "$+! !& & ! " 6& '$3' %
$ # & % & )!( + $-/ $/3 "$ "# % , $ $-
3$$8 &" "# ! & !$+ " ,# ,&3!& +. !/3! " "# "$$3.

&(3 . 1&!/3 $-$//$+") " ,,#& $- " / & % /+ ,/3 $-6$+8

5.1.2 Explicit methods

&(3 3 " "# 1/3," ! "#$% . # " +' 6 + * " !&" ,&33* & 8 % "$ % ,+ ( "#
! "#$% &! % "$ &'$ % /+$(3! $-% -- + , $! ,3&")+ . # + % ,+ /" $ &+
' "# 1" ," $ . &(3 #$6 "# & $+"! " $- ! "#$% ) % (* &,#
" +' 6 .
"$"&3$- C % -- + " "*/ $- 1/3," ! "#$% ,$' + !$+ "#& ; &,&% ! , %
! "#$% 6 + + /$+" % (* "# " +' 6 . # )!( + $- " +' 6 "#&" ) &,#
! "#$% $" ) -)3 # + ( ,&) " % / % " $ "# "& 8 % ' 3$/ % (* &,#
+ & % "# )!( + $- " +' 6 -$+ &,# "& 8 $" ) -$+! 4 &(3 5. $6 ' +
"# )!( + $-! "#$% "#&" &,# +) &! & -)3%&"&J
- 6$ $-"# "6 "* "6$ + ) ' "*/ $-! "#$% . # *% &36 "# ,) "$! +
% 1/3$+&" $ & % "+&" * % ' 3$/! " "# /+$%)," /3& % /&+"! ". D$+
"# "& 8 "# )!( + $-/ $/3 3! " %. "# + % /&+"! " &($)" #" +

G
 #&' ! 3&+ "& 8 . # + -$+ "# * &+ # #3* + /+ "&" ' $-"# + % /&+"! " & %
#&' & + 3&(3 $' +' 6 $- " /+&," , .
- -"# "6$ + ) -' ! "#$% $ 6$+8 -$+ & % /&+"! " !/3! "
% )//$+" "$$3. # $"# + + 6$+8 6 "# " ,# ,&3% ' 3$/! " & % #& &
+ /)"&" $ -$+ /+$!$" "+),")+ % 6&* $-6$+8 &!$ # ,$33& ) .
- + ) "6$ "#+ $+ -$)+ ! "#$% &+ -+$! " ,# ,&3 % ' 3$/! "
+ /+ " $- "# " +' 6 . #" $- "# "# +" " +' 6 )
(+& "$+! & % !&"+ 1 (& % '&3)&" $ ! "#$% & % -$)+ $- "# + !&
"# +" ) (+& "$+! . " 1/3& % "# 1" ," $ "# "6$ "*/ $-
! "#$% &+ ) % (*" &! "# + $6 6&*4& % $" &36&* &" -&,"$+ 3*5.
- + ) $ $+ $ ! "#$% &+ "#+ /)+,#& + $ + -+$! /+$%)," $
46$+8 $ "# ' +* 3& " /#& $-"# ,&+ /+$0,"5 & % & + -+$! " ,# ,&3
% ' 3$/! " "#&" #& & + /)"&" $ -$+ ( H& & "? ! "#$% .

"
"#$% K

! "#$%

# 8
/3$*!

$"&3 $. $-! "#$% ) %

,# 7)
&"+ 1 (& % '&3)&" $
! "#$%

" $$3
' "'
'&3)&" $

* " ! %* &! ,
, &+ $ ! "#$%
M)&3"*D) ," $
! "#$%

"&" " ,&3A&/&

2&3) & &3*
L " +' 6

!
"$+!

1 $!&" ,
"+),")+ %
%D %
# ,83 "

,
! "#$%

& &
*" !
+&

)
D

N
1 1
1 1 1 1 <
1 1 1 1 1 G
< 1 1
G 1 1 1 1 <
C 1 1
N 1 1 1
> 1 1 1 1 1 1 1 N
@ 1 1 1 1 1 1 1 N
; 1 1 1
;
1 1 1
1 1
< 1
G 1 1 1
C 1 1 1
N 1 1
> 1
@ 1
; 1 1 1
;
1 1 1 1 1 G
$"&3 G N G < G
&(3 . 1/3," ! "#$% ) % (*"# " +' 6

C
 # % ,&" "#&" % ' 3$/! " + "*/ ,&33*) O ! "#$% "6$ $-6# ,# " % "$
( (+& "$+! & % !&"+ 1 (& % '&3)&" $ ! "#$% . +$%)," /3& + ) &
(+$&% + & $+"! " $-! "#$% .
+ 6 + &3$ & 8 % -"# * ) "# ! "#$% $-" & % 6#&" "# * ) % "# ! -$+. "
( ,&! &//&+ " "#&" !& * &+ ) % $ 3* $ , . D+$! &(3 "# ! "#$% + )3&+3* & %
) -$+!3*) % "# ,$!/& *&+ J
- D$+ "# ")%*$-,) "$! + % %- % " ,# 7) 38 -$,) +$)/ " +' 6
& % !& &: + &% &+ 6 33 "&(3 # % & 6 33& & * " ! "$ '&3)&" "# ()
$//$+") "*$-,&+ /+$0," .
- + "# 3&" "& $-,&+ /+$0," #&' ( "+& % & % ) D& 3)+ $%
& % -- ," &3* 4D 5& % !(3* D .
- $! ,# ,83 " &+ ) % * " !&" ,&33*-$+ $! / ,-,"& 8 .
, !& * ! "#$% #&' ( "+ % ()" - 6 &+ + )3&+3* & % ) -$+!3* ) % (*
% -- + " " &! . # 7)&3"&" ' + &+,# ' % , "#&" % ! "#$% #&' & % &
, +"& &,, /"& , ()" &+ " 33 $" ) % ,$ " "3*& % &" -&,"$+ 3*.

5.1.3 Patterns in the use and adoption of methods

# " +' 6 ,$!! " % &($)" "# % -- + " 1/3," ! "#$% "# * #&% "+ % 6 "#
"# + % ,+ /" $ $-"# ! "#$% ( /+ " % 1"J
- +& "$+! ! "#$% 42 #&+ " &3. @@@5
# &! ' "$ & /+$, 6# ,# + /+$/$ & % % ,) % & 6 "#$)"
) / % 0)% ! "&3 "# 8 & % ,+ " , !. ,&% ! , (+& "$+! ) " "#&"
0)% ! "&3"# 8 #$)3% ( ) / % %. $6 ' + "# + & &")+&3" % ,* "#
+ &3% &," ' "* "$ ,$ -3," ! + , , (+& "$+! $ &+ "*/ ,&33* ,+$
% ,/3 &+*. &,# + + /+ " & % -- + " &""+ ()" $+ /&+" $-& * " !P/+$%),".
+$ % ,/3 &+* " &! &+ & !(3% "$ "# + -$+ &+3* ,$ -3," + $3)" $ . D$+ "#
+ & $ 0 )% ! "&3"# 8 & "# &")+&36&*. +& "$+! ) % "$ "&+" &
/+$0," $+ $3' /+$(3! . . - % & 6&* "$ + %), "# ,$ " $-& 1, ' 3* 1/ '
/+$0,". $! / $/3 ,$ % + (+& "$+! "$ ( -+) "+&" ( ,&) " "&8 6 8 "$
+ $3' $! /+$(3! . '&+ "* $- (+& "$+! ! "#$% &+ ) % ,3)% (& ,
(+& "$+! & % (+& "$+! 6 "# $ " " $" . " +' 6 ! " $ % "#&"
( -$+ & (+& "$+! $ ; ! )" $- / + $ &3 + -3," $ $-" %$ .
+& "$+! 6 "# $ " " $" ! 1 % 6 "# "# ,3) " + " ,# 7) ' +&33$$/ .
-" + "# % & &+ + /+ " %& %/ % $ "# 6&33 $! $ & 8 % "$ ,3) " + "# !
& , +"& 6&*. # ,3) " + % ,) % & % 6 % & !&* /$/ )/. # ,3) " +
/+$, + / &" % ' +&3" ! ( %$ (* % -- + " " &! /&+" ,/& " . $! / $/3
%$ $" 38 (+& "$+! 6 "# $ " " $" ( ,&) "# !& &%'& "& $- ,3& ,&3
(+& "$+! /$ 3%. % & ,& $" ( ,$ , ' % (* ' +&3 / $/3 6# $!
% ' %)&3 ,+ &" "# ( $-& % & & % $"# + "# %. "# + + &+ ) "#&"
(+& "$+! 6 "# $ " " $" ' "# /$ ( 3"* -$+ !$+ #* % ' %)&3 "$
,$ "+ ()" .
- &"+ 1 (& % '&3)&" $ ! "#$% 4 &#3& % ": @@C5
) # &" Q B #" $+ & + 6 ,$!/&+ $ &+ 1&!/3 $- !&"+ 1 (& %
'&3)&" $ ! "#$% . # ) $- "# ! "#$% 1" % % "# ,#& % /&+"! ".
+ " % "$ ,$!! " "#&" "# 8 % $-! "#$% -" "# + $6 6&* $-"# 8 .
-- + " '&+ & " &+ ) % % / % $ "# / + $ &3/+ - + , $-"# /&+" ,/& " & %
"# /+$(3! &" #& %. ' +&3' + $ $- 1, 3" !/3&" ,+,)3&" &!$ "# + .

N
 " +' 6 ,$!! " % $ $! ,& 6# + "# ! "#$% 6 + & % 6 + $"
&//3 % &" -&,"$+ 3*. + /$+" % "#&" "$ ) ) # *$) #&' "$ ( ,&+ -)3 "#&" "#
&3" + &" ' &+ &" & ! 3&+ 3' 3$-% - " $ . " ) -& + "$ ,$!/&+ & 8 $6 $3)" $
"#&" #& &3+ &%* ( $/" ! % 6 "# & 63* % - % $3)" $ . # + /+ " & /+$(3!
( ,&) "$ !&8 & -& + ,$!/&+ $ &33 $3)" $ #&' "$ ( (+$) #" "$ "# &! 3' 3$-
% - " $ . # /+$(3! #& &3$ ( % &3" 6 "# $"# + % /&+"! " $-"# ,$!/& *.
D$+ -)+"# + -$+!&" $ "# + &% + + 7) " % "$ + - + "$ *3) % " &3. 4 ;; 5. $"# +
" +' 6 ! " $ % "#&" "# !$ " !/$+"& " "# 6 "# "# ) # ! "#$% -$+ "#
% -- + " /&+" ,/& " $-& " &! -+$! ,#& 3,"+ ,&3 & % ($%* "$ -$,) $ "#
" + /+$(3! " &% $-"# + /&+" ,)3&+ &+ &. -" + "# 6#$3 /+$, 1&! "#
$("& % - )+ 6 33 $" ( , &+* ()" "# + 6 33( & +&3,$ ) &($)" 6# +
"# * #$)3% ( # &% . # !&"+ 1 "# $ 3$ + !/$+"& ". " $ 3* + /+ " "#
& %& $-6#&" #& ( %$ "$ ,+ &" ,$ ) & % $ 3* ) -)3"$ 1/3& 6#* &
, +"& % , $ #& ( "&8 " +! $- 6#&" #& ( /& % &"" " $ "$. $"# +
" +' 6 ! " $ % "#&" " &% $-) - )+ $+ /3) & %! ) # " % "$ )
,$3$)+ ( ,&) !& & ! " - % " & + "$ +& /. 3$ # %$ $" ) 6 #" -$+ "#
,+ " + &. " &% # $+ & "# ! +$)/ . . ,$ " +$)/ 3 &3% !& % ",.
"# 8 + #$)3% % ,% "#+$) # % ,) $ 6# "# + $+ $" $ ,+ " + $ '
!$+ !/$+"& , "#& & $"# +. $"# + " +' 6 ! " $ % "#&" /& + 6 ,$!/&+ $
& &3* !/$+"& " "$ # % + % , $ ( "&8 (*%$! & " / $/3.
- D& 3)+ $% & % -- ," &3* 4D 5 ! "#$% 4 #$!/ $ @@@5
D & ,$!!$ ! "#$% ) % "# ,$!/& * "$ % " -* &33/+$(3! + 3&" "$
* " ! . # * &3$ %$ + 8 & &3* 6 "# D (* ,&" $+ "# -- ," $--& 3)+
!$% &" % -- + " 3' 3 $-+ 8.
!(3* D & ! "#$% ) % "# 3&" "& $-,&+ /+$0," "$ /+ % ," 6#&" ,& $
6+$ 6 "# & 6 & !(3* % "& 3& % "$ /+ ' " " -+$! &,")&33* #&// . " &33$6
1&! "# /+$(3! -+$! % -- + " / + / ," ' . . + $ $! , & !(3*7)&3"* ",.
# ! "#$% $" ) % -"# & !(3*% "& 3 $" 6 ( ,&) "# 8 $63% /+$' % %
(* /+ ' $) 1/ + , $) #. !(3* D & 6 33 "&(3 # % ! "#$% "#
,$!/& *. " "&) #" * " !&" ,&33* "$ + 6#$ 6$+8 6 "# & !(3* & % "# + &
)//$+" +$)/ "$ & 6 + /$ (3 7) " $ &($)" "# ) $-"# ! "#$%. + )
"# ! "#$% "&" % "$ ( &" - % 6 "# ".
- # ,83 " 4 #$!/ $ @@@5
# + &+ & )!( + $-,# ,83 " % -- + " &+ & $! $--,&3& % $! ) $--,&3. #
) $--,&3 ,# ,83 " &+ ,+ &" % (* "# + -$+ "# ! 3' $+ "$ &'$ % 3
1/ + , % + "#&" !&* "&8 $' + "# + +$3 $" -$+ "" !/$+"& " ) 6#
% ' 3$/ "# * " !.
1&!/3 $- & $--,&3,# ,83 " )//3 + '&3)&" $ &"+ 1 4 5 . . & ,$+
,# ,83 " -$+ )//3 + '&3)&" $ . " '&3)&" "# / +-$+!& , $- )//3 + & & ,$!/& *.
" ) % "$ )+ "#&" )//3 + 6$+8 6 "# 2 #&' & # # 3' 3$-7)&3"* & 6 33
& ' +* " ! & 6 )//3 + ,$ "&," % $+ & & ,# ,8 ! "#$% 6 "# &3+ &%* 8 $6
)//3 + .
- %D % ,# 7) 4 + &)" " &3. ;; 5
) "$! + &""+ ()" &+ & &3* % (*' +* 1/ + , % + "# ,$!/& * )+
&""+ ()" 1/3$+&" $ 6 "# & % / 8 $63% $-6#&" " ,# $3$ ,& &,# ' . +
,&++* $)" "# "*/ $- 6$+8 ! " $ % & # # )!( + $- " ,# 7) ),# & "+ ,"
,$!/&+&" ' " " %+ ' !& &: + &% " + " %&"& &"# + /&+" ,/&" $

>
,$ - + , % ,) $ 6 "# )//3 + -$,) +$)/ 7) " $ & + * % ,&" % +&3
")% % !$ +&/# , -$+!&" $ &"# + ",.
- ) '&3)&" $
" 6& % ,+ ( % & & -$+!&3! "#$% % ' 3$/ % #$) & % ) % "$ ")%* "# !/&,"
$-,#& % -- + " &""+ ()" $-"# ,&+ $ ,) "$! + &" -&," $ .
- 2&3) & &3* 4 3 @>@5
" ) % "$ ")%* #$6 "# )! $- &33 ,$!/$ " &,# ' "# + 7) + ! " &" "#
,$!/3" ' # ,3 3' 3. $!/$ " &+ "# $/" ! % -$+ ,$ " 6# 3 " 33 ! "
,) "$! + &""+ ()" . # " +' 6 ,$!! " % $ "# ,)++ " + )+ , $- '&3)
& &3* 6# ,# 6& ) % & % ,&% & $.
- *" ! + 4 33' + " &3. @@N5
,# ,&3+ 7) + ! " &+ " -$+ "# /+$0," &" "# "$/ 3' 3 "# ,$!/3" ' # ,3 & % &+
% ,+ ( % -) ," $ &3" +! . # -) ," $ ) % & "# ,&++ + -$+ &33+ 7) + ! "
,3)% ,) "$! + &""+ ()" !& )-&,")+ /+ + 7) " ",. * " ! &+ % " - % "$
-)3-3 "# + 7) + ! " . # * " ! &+ "# % ,$!/$ % "$ ,$!/$ " & % "#
+ 7) + ! " &+ " -$+ &,# ,$!/$ ". # ! "#$% $ ) % + "# &! $- * " !
+ $+ ,& ,&% "# ,$!/& *. + ) "# ! "#$% 6 + &" - %
6 "# " & % ,$!! " % " ( & + , " 6&*$-6$+8 .
- M)&3"*D) ," $ /3$*! " 4MD 5 4 & ),# & % 3&) @@;5
2 % ,% % "$ ) MD , & $"# + ,$!/& * "# * 6 + ,$ "&," 6 "# 6&
),, -)33*) "# ! "#$%. # *+ , ' % $! ' +*% "& 3% 1&!/3 $-"# ! "#$%
) % &" "# $"# + ,$!/& * & % "+ % "$ %$ " 7)&33* % "& 3% ()" " 6& 1"+ ! 3* " !
,$ )! . It required cross-disciplinary groups having to gather abundant amounts of
information. The perceived result was simple project specifications that were initially clear, but
which were highly susceptible to change during the design process. B# / $/3 "#
,$!/& * "&38 &($)" ! "#$% *$) &36&* " "# 1&!/3 $-MD & & 1/3& &" $ $-
6#* "# * &+ $" $ ) -)3. ,,$+% "$ $ " +' 6 MD ! "#$% !&* ( !$+
) -)3 ,$!/& 6# + /+$%)," $ %$ $" '$3' $ !),# ' "! " & % 6# +
"# + & # # + -+ %$! "$ % ' 3$/ "# "$"&33* 6 6&* . 6$ " +' 6
% ,) % "# !/$+"& , $-"# /+ ,/3 6 "# "# MD ! "#$% $-) % + "& % "#
,) "$! + . " 6& ! " $ % "#&" ' -"# ! "#$% $ 3$ + ) MD "# 8 " 33
. $"# + " +' 6 ) " % "#&" MD !&* ( !$+ - & (3 6$+8 6 "# / ,-,
&+ & +&"# + "#& 6 "# "# 6#$3 /+$%),".
- "&" " ,&3A&/& ! "#$% 4B) & % &!&%& ;;;5
# /# 3$ $/#* $- A&/& "&" " ,&3 ! "#$% 6& "+ % &" "# % - % /#& .
A&/& ! "#$% "+* "$ " )/ & " " /3& "$ ! ! "# )!( + $-" " % % -$+ &
+ 3&(3 + )3". ,,$+% 3* /+$%)," /3& + "+ % "$ ")%* #$6 ,) "$! +
% - % " ,# 7) ,$)3% ( $/" ! %. $! &"" !/" 6 "# "# ,$ , /" 6 + %$
()" " 6& -$) % "$ $" ( 6$+"# "# --$+" "#&" " + 7) + %.
- , &+ $ ! "#$% 4'& % + 0
% @@C5
" ) % -$+ "+&" * ,+ &" $ . # " +' 6 ) "# ! "#$% & % "#&" &" "# "+&" *
/#& ,$!/&+ $ ( "6 ,$ , /" ,& $" ( !&% %) "$ ) , +"& "* "# %&"&. #
1/3& 6#* "# ! "#$% ,$ % + % & ' +* / ,)3&" ' /+$, (& % $ /+ ' $)
1/ + , 6& ) %. $ ) "# ! "#$% "# " +' 6 + &% ($$8 &($)" "+&" ,
/3& & % " 6& "# " +' 6 &3$ 6#$ ) % "# ! "#$%.
- 3) $" " &3 $ , + ' +,$! ,$ , + 4 ,5 42 #&+ " &3. @@@5
&"# + "#& ( & ! "#$% ) % (* "# ($$8 " ,$)3% ( - ++ % -+$! "# ,$ ' + &" $
6 "# "# " +' 6 "#&" " ) % & & 6&*$-"# 8 . " 6& % ,+ ( % & & &3* "#
@
/+$ & % ,$ $-%$ "# % -- + " 6&* &" ' +* &+3* "& $-% 6# &
/3&"-$+! $-& ,&+ % ,) %. # " +' 6 % % $" ) "# &! -$+ "# ! "#$%.
+ /$+" % "#&" " # $6 6&* 6$+8 /3&"-$+! % ' 3$/! ". ) "# ! "#$%
( ,&) # - % " ) -)36# "# + # # ) , +"& "* "# + & $ -$+ ,#$$ &
,$ , /".
- "+),")+ % ' " ' # 8 4 5 4 ,8&-) @@N5
# ! "#$% 6& ) % 6 "# & -&,3"&"$+ -+$! & ,$!/& * 6 "# "# &! &)"$!$" '
+$)/ 6# + "# ! "#$% 6& + )3&+3*) %. + &+,# + /&+" ,/&" % "# $ . "
6& ) % "$ +&" % & -$+ & 6 * " ! $3)" $ "#&" + 7) + % & $'&" '
&//+$&,#. # " +' 6 /$ " % $)" "#&" "# ! "#$% & ' +* "+),")+ % 6&*(*6# ,#
!& * % & &+ +&" %. /&+" ,)3&+3* &//+ ,&" % "# -$,) $ ) % + "& % "#
/+$(3!.
- 1 $!&" , 4 &! ;; 5
" 6& ) % "$ "&8 & - &3% , $ &($)" 6# ,# ,$ , /" "$ /)+ ) & 6 $3)" $
% ' 3$/! ". ) # ! "#$% #&% /+ ' $) 3*( ) % & % "6$ $3)" $ 6 + -$) % "$
/ +-$+! 7)&33*6 33. &1 $!&" ,% !&"+ 1 6& ) % "$ % " -*6# ,# $3)" $ #&%
3 ,$)/3 & % "# + -$+ # # + 7)&3"*.
- * " ! * &! , 4D$++ " + @C 5
" & ! "#$% 3 8 &33"# /$ (3 -$+, $- & ,$!!$ /+$, "#&" -3) , "#
),, $-&," $ . # ! "#$% 6& "# /+$, $-( 1/3$+ % (* & % /&+"! " &"
"# " ! $-"# " +' 6 .
- ' & & ! " $$3 4N 5 4 : 8 & % &8& @@;5
% /&+"! " ) " % ) "# " $- "$$3. # * 6 + "&) #" ,$)+ ()" "#
" +' 6 "# 8 "#&" "# * #&' $" ( ) %. ,$!! " % $ "# + "+$%)," $
6 "#$)" ,$!/)" + )//$+". # ' "$$3 &+ &3$ /+$!$" % "# " + &3 "6$+8.
&+% MD D * " ! * &! , & % N "# % , $ "$ !/3! " "#
! "#$% ,&! -+$! !& & ! " $+ -+$! % /&+"! " % ' 3$/ $+ /)+,#&
)//$+" "$$3. # ,& ( ,&33% & "$/ %$6 !/3! "&" $ $- ! "#$% . D$+ "#
+ !& ! "#$% " 6& "# ) + 6#$ % ,% % . . " 6& & ($""$! )/ $,,)++ , . #
" +' 6 6 + & 8 % #$6 "# * ,&! &($)" "$ ) "# ! "#$% . # -$33$6
& 6 + 6 + 1/+ % (* "# + . $! 3 #" !$% -,&" $ &+ !&% $!
& 6 + "$ + % + "# ! ) % + "& %&(3 $)" % "# ,$ " 1" $-"# / ,-, 7) " $ "#&"
6& -$+!)3&" % %)+ "# " +' 6 $+ "$ &'$ % % ,3$ ,$ -% " &3 -$+!&" $ .
3 -.9
: " "8 ;
%9 < = >" "# 8 ?# @, "" 8# '' A
B 8 # # # " C
B% " " " C
B- " " 8 + 8 " + C
B% 8#" 8 "8 " 8 # " 8 " " 8
" 8#" "8 D# " " " " C
B% 8#" " + 8 5 " +
C
B- " " 8#" " 8 C
B- " " # # + 8 5- 8 " 1 C
B- " D# " " 8 C
B- 1 " " "8 "5- " 88"
E #" C

;
 B "8 5- " 8# " + "1 # +8 5 8 8
# +8 " + C5
B4 # # 1 " 8 59 8" "## " 8+ "
"8 C
B 8 " -1 C
B " 8 59 " " D " "
8 8 8 8+ D 8 C
B " " "8 # " " C
D+$! "# & 6 + " + " & / ," $-"# ) + ? % , $ "$ &%$/" & / ,-,! "#$%
& ($""$! )/ &//+$&,# ,& ( - ++ %J
- # " &38 $63% & % &($)" ! "#$% "$ %$ & /&+" ,)3&+ "& 8 -3) , "#
! "#$%$3$ *"#&" 6 33( ) % -$+ & 3$ / + $%.
- # + /)"&" $ $- "# -$+!&" $ $)+, 4 . . ($$8 ),, -)3 ,$!/&
) ' + " 5 "$ 3&+ ! "#$% &-- ," "# &%$/" $ $-! "#$% .
- $! " ! ' ! "#$% "#&" &+ $" -)33* ) % + "$$% &+ ) %. ! "#$% $-"
) % "#&" $ 3* $ $+ "6$ / $/3 "# % " &! ) % + "& %. # + !&
/&+" ,/& " 3&+ "# ! "#$% "#+$) # "# /+$, $-) ".
- "#$% () 3" $ /+ ,/3 "#&" &+ & + "$ +& / & % !&",# + "# 8
#&' & + &" + ,#& , "$ ( ) %.
+"& ! "#$% ) % & ,$!/& * + )3" -+$! & "+&" , ,#$ , $-!& & ! " $+ &
% /&+"! ". $6 ' + "# %$ $" )&+& " " ),, -)3) . *3) % " &3. 4 ;; 5
% ,) "# % " ," $ ( "6 ! "#$% &,, /"& , & % "# + ),, -)3 ) . To
successfully !/3! " ! "#$% %) "+* "6$ (&++ + !) " ( $' +,$! J 4 5 ! "#$%
&,, /"& , & % 4 5 ),, -)3 ) $- ! "#$% . "#$% &,, /"& , !/3 %) "+*
( ,$! " + " % ) & ! "#$%. # $,,)+ 6# "# + !& & ! " $+ +
$+ ($"# % ,% "$ "+* & ! "#$% /+$%)," % ' 3$/! " ( ,&) "# * ( 3 ' "#&" " )
6 33 ( -" ,$!/& * / +-$+!& , . ),, -)3 ) $- ! "#$% ! & "#&" & ,$!/& *
/ +, ' & "+$ ,$ "+ ()" $ $-& ! "#$% "$ /+$%)," 7)&3"* $+ + %), % /+$0," 3&%
" ! & % % ,% "$ ) " -$+ , +"& ")&" $ . # % " ," $ ( "6 ! "#$%
&,, /"& , & % " ),, -)3) !&% ( ,&) "# + &+ ! "#$% "#&" %$ $" $' +,$!
"# + $-"# (&++ + ! "#$% "#&" $' +,$! $ 3* "# -+ " (&++ + & % ! "#$% "#&"
$' +,$! ($"#. 3* "#$ ! "#$% $' +,$! ($"# (&++ + ( ,$! / +!& "3* ) %
& % ,& ( + &+% % & "+& - ++ % -+$! &,&% ! & "$ %) "+*. $6 ' + -+$! $
! "#$% "#&" 6& ) % & % &(& %$ % &-" + $! " ! MD "# " +' 6 ,$!! " %
#$6 "# ! "#$% /+ ,/3 + !& "# + ? 6&* $-"# 8 6 "# /&+" $-"#
! "#$% ( ,$+/$+&" % ,$!/& * ,+ &" % ! "#$% . , ' -! "#$% &+
$" &" -&,"$+ 3* ) % " /$ (3 "# * ,$ "+ ()" "$ )//$+" % + (* /+$' %
'&3)&(3 % /+ ,/3 .
B# "# ) $-! "#$% $+ &" -+$! "# % $-) + 6#$ % ,% "$ &//3* " "#
,#& , $- "" "# ! "#$% * " !&" ,&33* ) % &+ 3$6 +. +&33&,8 $-+ $)+,
4/+$/ + % ,+ /" $ & % 1&!/3 $-"6&+ )//$+" -&,3"&" $ /+$/ + "+& ",5
"# !& ,&) . # ,$!!$ /# $! $ $- + "+* "$ !/3! " %
! "#$% (*"# ! 3' #$6 ' + & % ,&" ' $-"# '&3) $-% ! "#$%$3$ *
+ &+,# -$+ %) "+*.
" +' 6 ,$!! " % "#&" "$ !/3! " ! "#$% & ($""$! )/ !& + "
, &+* "$ 0," "# % & )! +$) /3&, / ,&33* 6 "# E/$6 +-)3 !& &" '
/ $/3 6#$ #&' "# /$6 + "$ -3) , $"# + F. * %$ $ "# $+ & &" $ 6 33
' ")&33* &( $+( "# ! "#$%. $/3 6 33 )%% 3* "&38 &($)" "# ! "#$% & - " 6& &
 "& %&+% "$$3. )" "" "$ "# 3' 3+ 7) + !),# --$+". //$ -$+, &+ $-" ! "
& % $ ,& $" ' )/ $"# +6 "# 6$+8 3$ ".

5.1.4 Why do engineers think design methods have not had the
expected impact?
-" + "# " +' 6 1/3& % 6#&" ! "#$% "# * ) % "# * 6 + & 8 % 6#* ! "#$%
&+ $" ) % !$+ $-" . 1, +/" $-"# " +' 6 1/3& 6#* ! "#$% &+ $" ) %
,& ( -$) % &// % 1 . + & )!!&+* ' 6 "# "# !$ " + 3'& " )
# #3 #" % (*"# " +' 6 J
# !& + & $ -$+ & 3&,8 $-) $-! "#$% %) "+* &,,$+% "$ "# " +' 6
,& ( )!!&+ % & -$33$6 J
- $ " + %$ $" #&' "# " ! "$ 3&+ "$ ) 6 ! "#$% .
- # &'& 3&(3 " ! -$+ /+$0," % ' 3$/! " %$ $" &33$6 -$+ ) ! "#$% .
- # ,$++ ," ) $+ 3," $ $-! "#$% 3&% "$ % &//$ " + )3" .
- $! ! "#$% &+ &( $+( % (& % $ "# + /$/)3&+ "* & % $! " ! "# *%$ $" ) "
"# /+$(3!.
- "#$% &+ 38 -& # $ = "# * &+ ) % -$+ ' +*"# $ * &+ & % % &// &+ "#
-$33$6 * &+.
- # ! "#$% -+$! ,$!/& *!& )&3 &+ % ,+ ( % #&+% "$ + &% -$+!&" .
- $ " + &+ ) &6&+ $-"# &'& 3&(3 ! "#$%
- # + 3&,8 $- ) %& , "# ,$!/& * "# ) $-! "#$%
- $/3 %$ $" 6& " "$ ,#& "# + 6&*$-6$+8
- "#$% &+ 1, ' 3*&,&% ! ,
- &,8 $-,$!/)" + )//$+"
- "#$% + %), -+ %$! "$ "# 8 & % &+ ($+
- + &+ $" "+& % "$ ) ! "#$%
- # '&3) $-"# ! "#$% $ 3* ) % + "$$% 6# *$) #&' $! /+&," ,&3 1/ + ,
6 "# "
- ' +* / + $ % ' 3$/ $! "# #& & ! "#$% ,&3/+$, -$+ ". B#* #$)3% "# *
,#& " -$+ & &,&% ! ,! "#$%R
- ! "#$% $ 3* & ()+ &),+&" , * " ! "$ $--,&33* &//+$' $! "# "#&" #&
&3+ &%*( % ,% % )/$

5.1.5 The importance given to concept selection

6$ " +' 6 ,$!! " % "#&" "# + +&33* $ 3&,8 $- % & = " "# /+$, $-
!/3! " "# ! "#&" --, ". $! $$% % & &+ 3$ " 6 "#$)" $" , . # *
+ &3 "# !/$+"& , $-3$ " % & 6# & ,$!/ " "$+ 3&) ,# & $3)" $ "#&" #&% (
1/3$+ % & )!( + $-* &+ ( -$+ ()" 6& ' + !/3! " % & ,&+. & * +
( 3 ' "# /+$(3! 3 "# ,$ , /" 3," $ /+$, 6# + % & &+ 3$ " %) "$ /$$+
% , $ !&8 . $ , /" 3," $ -&," & )(0," "#&" #& ( 3&+ 3*% ,) % 6 "#
+ $-"# ,$!/& *%)+ "# -$)+ * &+ .
# + & $ -$+ "# !/$+"& , ' "$ ,$ , /" 3," $ "# ,$!/& * 3 "# +
/+$%)," % ' 3$/! " "+&" * & % "# ,#&+&," + $-"# &)"$!$" ' %) "+*. 2$3'$ &+
$+/$+&" $ #& & !/3! " % $--"# # 3- $3)" $ "+&" * 4 9/ : & & %
#$!/ $ ;;<5 . . "# * )+ "# + ) $-% ' 3$/ % " ,# $3$ * & 6 % '&+ "* $-
/+$%)," . 6 $3)" $ -$+ * " ! &+ % ' 3$/ % % / % "3* $-,&+ /+$0," & % &+
 $" !/3! " % & ,&+ ) " 3 "# + - & ( 3"* & % &%'& "& + 3&" $ "$ "# $3%
$3)" $ &+ ,$ " "3* /+$' 6 "# - 6 1, /" $ . 3* "# %$ "# * ( ,$! $--"#
# 3- $3)" $ . . * " ! $3)" $ ( ,$! /&+" $-& "$,8 6 "# 8 $6 / +-$+!& ,
-+$! 6# ,# "$ ,#$$ -$+ ,&+ /+$0," . # $(0," ' $-& ,&+ /+$0," !& 3* "$ 3," "#
,$!( &" $ $-$--"# # 3- $3)" $ -$+ "# * " ! $-& ,&+ -+$! "# H"$,8? "#&" ( "" +
-)3-3 ,) "$! + &""+ ()" . # ,&+ /+$0," /+$, ,& "# + -$+ ( % ,+ ( % & &
,$ " )$) /+$, $- )+ "#&" "# $3)" $ ,#$ -$+ "# ,&+ !$% 3 -)3-3 "#
"&+ " % ,) "$! + % !& % & ( " & /$ (3 &" & ,$ " ) "&(3 "$ & () ,& & %
"#&" "#$ $3)" $ ,& ( % ' 3$/ % 6 "# & 3! " % " ! -+&! . $3)" $ &+
/+$ + ' 3* 3," % -$+ "# % -- + " 3' 3 $- "# ,&+ (+ &8 %$6 /3&"-$+! * " !
,$!/$ ". # %$ (* "" & &" -&,"$+* % + $-, +"& "* "# 8 $63%
&($)" $3)" $ / +-$+!& , )+ "# 3," % $3)" $ $/" !&3 & % ,& (
!/3! " % 6 "# & 3! " % " ! -+&! . , "! $ !/$+"& " "# &)"$!$" '
%) "+* " #& ( % ,+ ( % (* ' +&3 " +' 6 & "# + &3% , $ "&8 +. #* ,&3
& % !)3&" % " " &+ -) %&! "&3! & "$ + %), ) , +"& "* "# &)"$!$" '
,$!/& *. , "# " ! -$+ & % , $ +) $)" & $3)" $ #& "$ ( "&8 6 "# "#
&'& 3&(3 -$+!&" $ .
!/$+"& " "$/ , $-% ,) $ %)+ "# " +' 6 6& "# +$3 $-,$ , /" 3," $
! "#$% "# % , $ !&8 &," ' "*. D+$! "# /$ " $- ' 6 $- $! +
,$ , /" 3," $ ! "#$% ,& $ 3* )//$+" $--,&3% , $ !&8 . % "# + &
+ &3 -$+!&3,$ " )$) /+$, $-% , $ !&8 6# ,# $ !$% 3-" . ,,$+% "$
$ " +' 6 % , $ &+ "&8 $)" % "# ! " . $"# + " +' 6
,$!! " % "#&" $ 3* & - 6 % , $ & ,$!/3" ,&+ /+$0," ,& &,")&33* ( "&8 6 "#
"# ) $-& ! "#$%. D+$! "# /$ " $-' 6 $-$"# + + "# * " !&" , ) $-
! "#$% !/$+"& " ( ,&) &%&/" "$ & % -- + " 6&* $-6$+8 -$+ ' +* /+$0,"
% --,)3". 1/3,"3* ! " $ % (* $! " +' 6 "# ) $- -$+!&3 ! "#$%
% / % $ "# $/ $ &($)" "# ! $-"# !$ " %$! & " / $/3 "# ! " .
" +' 6 ,$!! " % "#&" ! "#$% -$+ ,$ , /" 3," $ &+ 7)&33* $$% (* "# +
#& % $+ ,$!/)" + )//$+" ()" "# * !) " &36&* ( 8 /" &" & !/3 3' 3. # * #$)3%
' + ( $ ,$!/3,&" % "#&" + 3$ - 3 -$+ 6#&" "# * &+ %$ . # %&"&
"+$%), % ! "#$% ) )&33* ' +* ) , +"& . "" $(0," ' + )3" ,$!-$+"
()" ! "#$% #$)3% &36&* 3" + "&8 % , $ . # " +' 6 "&" % E " ' +*
!/$+"& " $" "$ 3" "# ! "#$% 6$+8 -$+ "# ! 3' F.
# 1" "6$ (3$,8 $-+ &+,# % / "# /+$, $-,$ , /" 3," $ . "# ,& $-
" +' 6 ,$!/3! " % 6 "# $( +'&" $ "# $(0," $- ")%* "# /+$, (* 6# ,#
,$ , /" &+ 3," %. "#$% ,$)3% ( % " ," %. "# ,& $-& &3* %$,)! " %
) % ! "#$% "# $(0," $- ")%* "# &//3,&" $ $- &,&% ! , ! "#$% -$+ ,$ , /"
3," $ .

5.2 Research block two: the process of concept selection as reported by

engineers to an in-company researcher and complemented with
observations
# ,$ % (3$,8 $-+ &+,# ,$ " % $- ! "+),")+ % " +' 6 &($)" "# /+$,
$-,$ , /" 3," $ . # " +' 6 6 + 3," % &!$ "% + & &3*
1/ +" & % % ' 3$/! " + 6$+8 6 "# "# % ' 3$/! " $- 6 ,&+ ($% . 33
" +' 6 1, /" $ 6 + ,$ %)," % "# &" ' 3& )& $- "# " +' 6 . #
& 6 + 6 + "+& ,+ ( %. )" % $-"# " +' 6 + 6 + &3$ -$+!&33*& 8 %
7) " $ & % "# + & 6 + 6 + 6+ "" %$6 . # &%'& "& 6 "# ,&",# / $/3 $
"# $ & % "&8 "# $//$+") "* "$ % + ,"3* & 8 7) " $ % / % $ & % &($)" "# +
% ")&" $ $)"6 # "# % &%'& "& $- $" &36&* -$33$6 & "+ ,"
7) " $ & + . & ,$!/3! " "$ "# " +' 6 !&" + &3 $( +'&" $ 6 + !&% &"
"#+ &+ & ! " . # &+ & ! " &+ $ %&* ! " ,&++ % $)" 6 "# ,+$
% ,/3 &+* " &! . 1/ +" -+$! '&+ $) - 3% ),# & "&!/ & !(3* ,$++$ $
/+$" ," $ $ & % ' (+&" $ 4 2 5 ",. &+ &"# + % & " &! $-&//+$1 !&" 3* G ;
/ + $ . # ! " 6 + $( +' % & % $" 6 + "&8 . D)+"# +!$+ ,$/ $-"#
! " ! )" 6 + ,$33," %.
( +'&" $ -+$! ($"# "# " +' 6 & % $( +'&"$+* ")%*&+ 1/3& % 1"J
- !/+ $ &($)" "# ,$ , /" 3," $ /+$, & % "# ( #&' $)+ $- % +
% &+ # #3* % / % " $ "# + /+$(3! $3' "*3. )+ "# " +' 6
")%* % -- + " "*3 $-% + 6 + ,$) " + %. +"$ 4 @@<5 "&" "# /+$(3!
$3' "*3 $-/ $/3 3 ( "6 "6$ 1"+ ! "# $'&" ' & % "# &%&/" '
/+$(3! $3' +. $'&" ' % + " % "$ &+,# -$+ +&% ,&33*% -- + " $3)" $
"#&" #&' "# /$" " &3"$ )+/& "# ,)++ " /&+&% !. &%&/" ' % + " % "$
!&8 !&33& % " ' !/+$' ! " "$ 1 " $3)" $ . $ " / $/3 &+ ")&" %
$! 6# + ( "6 "# "6$ 1"+ ! . 1&!/3 $- & &%&/" ' % +
% ,+ ( "# /+$, $-,$ , /" 3," $ -$33$6 J /+ ' $) 3* ) % $3)" $ &
2$3'$ #$)3% ( " " %= -"#&" $3)" $ %$ $" &" -* "# + 7) + ! " & $3)" $
-+$! & ,$!/ " "$+ #$)3% ( "+ % & % - "# + ! " "# + 7) + ! " & 6
) 8 $6 $3)" $ #$)3% ( % ' 3$/ %. & % E 6 $3)" $ ,& ,+ &" /+$(3!
3&" + $ " &- + "$ $ -$+ /+$' $3)" $ -/$ (3F. 1&!/3 $-& $'&" '
% + & % "#&"J E $) #$)3% "&8 "# ,#& , "$ " 6 ,$ , /" . #
!/+$' ! " / +-$+!& , !&*( $" ( " &33* ()" "# *#&' & + &" /$" " &3
6# + & "# $3% $3)" $ ,& $" ( !/+$' % !),# -)+"# +F. # "6$ " +' 6
6 + &!$ "# 1"+ ! "# " +' 6 ")%* & % #$6 "#&" / + $ &3 "*3 + &"3*
&-- ," "# 6&* $- % & % #$6 "# ,$ , /" 3," $ /+$, 3$$8 38 .
,,$+% "$ & " +' 6 % !& & + 6#&" !/$+"& " "$ !&8 " &! 6 "# & ! 1
$-"# + #" / $/3. # /+$(3! "#&" $! " ! "# + & 3&,8 $-, +"& 8 % $-
/ $/3.
- $"# + -&,"$+ #$6 (&,8 +$) % &-- ," "# $" $ $-,$ , /" 3," $ . -,& "
% -- + , 6 + -$) % ( "6 "# " +' 6 . $ " $- "# " +' 6 % %
+ &" 2 #&' 6$+8 % !& 3* 6 "# +&" %+&6 ) % -- + "
$-"6&+ . # + 1/ + , 3 ,+ &" !$% 3 $- $3)" $ "#&" -)3-3'&+ $)
+ 7) + ! " . # % + "&" % "#&" !& * % , $ 6 + "&8 &-" + +$)/
% ,) $ $+ (* ") " $ & % 1/ + , . + 6 "# & (&,8 +$) % 6 "# D
& &3* $-" + 3&" % "$ % , $ ( (& % )/$ & &3* + )3" . # "6$ -&,"$+
&-- ," "# / +, /" $ $-"# /+$, $-,$ , /" 3," $ /+$(3! $3' "*3 & %
(&,8 +$) % + ' &3"#&" % ' %)&3 +& / -+$! ")&" $ 6#&" "# + ! % /+ /&+ %
-$+ & ) " % (* ? $ $+ & % *!$)+ 4 @@;5.
- )+ "# " +' 6 ")%* % + 6#$ &+ $-" 0
)% % "$ ( 8 3-)3 (* "# +
,$33& ) & % !& & ! " 6 + &,")&33* ) 1/3," ! "#$% ' -"# *
% % $" + - + "$ "# + /+$, %)+ & ! "#$% %)+ "# " +' 6 . # ! & "#&"
/+$, $+ " % + &+ 6 33 + &+% % " &! % . !$ " "# 3'
" +' 6 "#+ ) % 1/3," ! "#$% . . ! "#$% / +- ,"3* ,$!/3* 6 "# "#
% - " $ (* )(8& E! "#$% &+ * " ! $- ! "#$%$3$ ,&3+)3 "#&" % " +!
,3& $-/$ (3 /+$, %)+ & % &," $ "#&" &+ 38 3* "$ 3&% $ & /3& % /&"# "$
<
 "# &,,$!/3 #! " $-& % + % & %F 4 )(8& @>;5. -$)+"# ! "#$% 6& % " ," %
"#+$) # $( +'&" $ & % -$+!&37) " $ "$ /&+" ,/& " . # * &+ /+ " % 1".
# -$)+ ! "#$% #&' ( 3&( 33% & % .

5.2.1 Method A
# -+ " ! "#$% /+ " % # + ) % -$+ 3," &!$ % -- + " ,$ , /" . "
% ,+ /" $ ,&! & & & 6 + "$ "# 7) " $ J EB#&" "# ,$ , /" 3," $ /+$, RF
6 +J EB ) "# /3) I ! ) ! "#$%.F # " +' 6 % / + $ 6& $" &(3 "$ " 33
"# $+ $-"# ! "#$%. $"# + $,,& $ + #&' &3$ + - ++ % "$ "# ! "#$%
& "# /3) ! ) ! "#$% 6 "#$)" -)+"# + % ,+ /" $ . "# ! "#$% 4 &(3 5
" ++ 3&" % % ,+ " + & &+ % ,&" % & % %$,)! " % "# ! "#$%. D$+ 1&!/3 -
,$ , /" $ /+$(3!&" , "$ !& )-&,")+ & ! ) /)" % "# ($1 & % & #$+"
1/3& &" $ ' . D)+"# + ,#& "#&" ,& &3" + "# ! ) "$ & /3) &+ &3$
% ,&" %. # % ,&" % " ++ 3&" $ -- ," &+ &,,$+% "$ "# " +' 6 % % +
,$!!$ . 1&!/3 $- & " ++ 3&" $ J - ,$ , /" "#+ #& & ( &,, #$3 -$+
6 3% " !&*3&% "$ (&% / +-$+!& , "+ "# & % " -- .

( ,&) S . T ( ,&) S
()" -,#& "# T
T ( ,&) $-S ( ,&) S ()" -S

( ,&) S ()" -S

&(3 . H # /3) I ! ) ! "#$%?

# ! ) & % /3) &+ ' + )!! %. B# "# % + 6& & 8 % -!$+ "#&
$ /3) $+ ! ) ,$)3% ( ' "$ & ,$ , /" # & % E $ 6 8 $6 #$6 !/$+"& "
% -- + " ) &+ F. # !&"+ 1 ) % & & ) % %)+ % ,) $ &($)" ,$ , /"
6# ,# & ! "$ + &,# ,$ ) &($)" 6# ,# ,$ , /" "$ ,#$ .

5.2.2 Method B
# ! "#$% !$+ / ,)3&" ' &")+ . " 6& &3$ ,$) " + % %)+ & " +' 6
6# & 8 %J UB#&" "# ,$ , /" 3 ," $ /+$, RU " #& ( +&" - % (* -$+!&3
% ,) $ 6 "# $"# + % + . # $&3$-"# ! "#$% "$ 33& ,$ , /" $+ %
$3)" $ "$ !& & ! " 6# ,# "# % + #&' &3+ &%* % ,% % )/$ & % ,$ % +
$/" !&3. # ! "#$% 7) " !/3. B# & % " &! #& % ,% % " + &33* 6# ,#
,$ , /" ( " "# * ,& "# !&8 & ,$ , /" 6 "# & 3 1/ ' % "#&" %$ $"
-)3-3$+ #&+%3* -)3-3 "# ,+ " + & & % & ,$ , /" "#&" -)3-3 "# ,+ " + & (* & 6 % !&+
& % &""+&," ' -+$! & " ,# ,&3/$ " $-' 6 ()" 1, ' 3* 1/ ' "$ /+$%), .
D &33* "# "#+ ,$ , /" &+ /+ " % "$ !& & ! " & $6 + $ " /+ "&" $ .
& & ! " #& ( $( +' % (* "# + "$ &3!$ " &36&* 3," "# + "#
G
,$ , /" "#&" "$$ U(&%U6 "# + / ," "$ ,+ " + & $+ "# $ 6 "# "$$ # # ,$ ". ,
!& & ! " 3," "# ,$ , /" "# % " &! "# 8 "# !$ " ) "&(3.

5.2.3 Method C
# "# +% ! "#$% "# !$ " 3&($+&" . " 6& % ,+ ( % & & 1/3," ! "#$% (*"#
" +' 6 6# & 8 % "# &! 7) " $ & &($' . " + " 3* "# / ,-,! "#$%
6& % ' 3$/ % (*& !&33,$ , /" % ' 3$/! " " &! $--$)+ / + $ %)+ "# )/ "&+" $-
& /+$0,". # ! "#$% 6& % % "$ ,$/ 6 "# "# /+$(3! $- 3," ( "6 G;
% -- + " ,$ , /")&3,#& "$ & 1 " ,&+ ($%*"$ !/+$' " ! ,#& ,&3/+$/ +"
6# 3 + %), 6 #" &33"# 6 "# & ' +*" #" ()% " -$+ ,#& . # ! "#$% #$6
&(3 <. # "&") ($1 % ,&" -& ,$ , /" " 33$/ . . ' " &" $ &+ $ $
"$ 1/3$+ "# - & ( 3"*$-"# ,$ , /". ),# ' " &" $ ,& ( . . D & &3* $-
"+ "# $+ " -- $+ ($"#. + $ ! & /+ $+ "*& % & -) ," $ $-"# /$ (3 !/&," $-
& ,$ , /". -"# !/&," ( "# H/+ $? # #. 8 ",# $-,$ , /" 3 8 &+ !&% "$
$"# + %$,)! " ,&33% $ /& + ,$ "& ,$ % % -$+!&" $ &($)" ,$ , /"
& &3* %&"& %+&6 "&(3 ",. ' 3$-% --,)3"* #$6 % --,)3" " 0 )% % "$ !&8
"# ,$ , /". & %+ ' + #$6 6#&" "# %+ ' -$+, -$+ !&8 "#&" ,$ , /" . #
&-- ," /3&"-$+! ($1 #$6 -"# ,$ , /" 6 33&-- ," "# % $-"# 6#$3 /3&"-$+! $+ - "
& U($3"U$ ,$ , /". - " &-- ," "# /3&"-$+! !$+ "&8 #$3% + 6 33 % "$ (
,$ ' , % &($)" "# , "*$-"# ,$ , /". % ,$!/ " "$+ -& ! 3&+ ,$ , /"
( ,)++ "3*) % & ,$!/ " "$+ & % # , ( ,#!&+8 /$ (3.

"&") + $P 8 ",# $- ' 3 $- &! & %+ ' + -- ," % ,$!! "

!/&," ,$ , /" % --,)3"* /3&"-$+! ,$!/ " "$+
/ P 8 "$ & * B #" P $ &! $-
,3$ % / ,")+ --,)3" $ " ,$!/ " "$+
% $ % )! $!!$ &3"*
/& ..
% ,+ /" $
$ , /"

$ , /"

$ , /"

&(3 <. "#$% % ' 3$/ % #$)

)+ "# 3," $ /+$, &33,$ , /" &+ )(! "" % "$ 7) " $ &($)" "# -$33$6
) J
J $" " &3P %+ ' + 4+ %), % 6 #" & %P$+ !/+$' % / +-$+!& , 5
J "&!/ - & ( 3"* - 6 /&+" & %P$+ !&" + &3
J A$ - & ( 3"* - 6 /&+" & %P$+ !&" + &3
J +$, ,$ 7) , 4& !(3*-&,"$+* /& " -&,"$+*5
J $!!$ &3"*6 "# $"# + !$% 3 (& % $ "# &! /3&"-$+!
DJ B #" / &3"* - / +-$+!& , !/+$' ! " "# %+ ' +P/ +-$+!& , / &3"* -
6 #" + %)," $ "# %+ ' +
,,$+% "$ "# " +' 6 -& ,$ , /" / +-$+! 1, ' 3* (&%3* 6 "# + / ," "$ $
$-"# 7) " $ " ,& ( % ,&+% %. D)+"# +!$+ "# 7) " $ &+ " " $ &33* $+% + %
& % ,&" %. -"# /$" " &3 # # $ ,$ " ) "$ 7) " $ = -"# "&!/&( 3"* $-"#
,$ , /" $8 "# $ "$ = & % $ $ . # 3$ ,( # % "# $+% + "#&" "$ -& ,$ , /"
3&% "$ 0 $ /+$(3! $ !) " &3+ &%* 8 $6 "# "*/ $- !&" + &3= # , "#

C
 "&!/&( 3"* ,& ( '&3)&" % ( -$+ 0 $ . # &! &//3 "$ "# 7) " $ J $
% "$ 8 $6 "# "*/ $- 0 $ ( -$+ '&3)&" "# -&,"$+* ,$ 7) , . - "#
,$ , /" - & (3 "$ /+$%), -+$! & !& )-&,")+ /$ " $-' 6 ()" " ,$!!$ &3"*
3$6 " ,& " 33 ( + &+% % & /+$(3!&" ,. D &33* & ,$ , /" !) " $" 1, ' 3*
/ &3 $"# + & / ," $- "# % . # 7) " $ &+ ) % & & )//$+" & % "# +
&//3,&" $ -31 (3 ! & "#&" $! 7) " $ !&* ( % ,&+% % 6# $"
&//3,&(3 & % $"# + !&*( &%% % - % %.

5.2.4 Method D
# -$)+"# ! "#$% 6& ,$) " + % 6# 3 !&8 $( +'&" $ &" "# $ ,&33% &+ &
! " . ' +*"6$ 6 8 "# ! " &+ # 3% -$+ "# 1 % &+ & $-"# ,&+ ($%*.
1/ +" "# - 3% &-- ," "# ,&+ ($%* &+ /+ " ),# & "&!/ & !(3* & %
,$++$ $ /+$" ," $ . &,# % + $-"# # " ! "&3/&+" $-"# ,&+ ($%* " +
&,,$+% "$ & /+ % " +! % $+% + & % /+ " "# + ,$!/$ ") & /+$0,"$+.
& $"# + ,+ & ,+ #$" $-"# ,$!/$ " #$6 % & $6 + $ " %$,)! ". #
% + /+ " "# ,$!/$ " & % "# 1/ +" " &! % ,) % -- + " & / ," $-"#
% &,,$+% "$ "# + &+ & $- 1/ +" . +$(3! ,$!! " & % /+$/$ % ,#&
&+ 6+ "" $ "# % $-"# ,+ #$" "# $6 + $ " %$,)! " 4D )+ 5.

CA D PPT
Exponential growth
Important difference
Opportunity to reduce weight

Figure 2. Method D

$!! " + !& $ 6 8 &-" + & $3)" $ #& ( -$) %. # $6 + $ " 3% &+
,$33," % "$ & %$,)! " ,&33% "# $%* A$)+ &3 4 A5. "$"&3 $- 1 0 $)+ &3 &,#
+ /+ " & % &+ &= -+$ " -+$ " -3$$+ + &+ -3$$+ % + % $)" + & % +$$-
1 " -$+ &,# ,&+ /+$0,". # ! "#$% ),, -)3-$+ + - $ $3)" $ %)+ % "& 3
% ()" 6 &8 + -/&+&333,$ , /" &+ % ' 3$/ %. " 6& $( +' % "#&" 6# ' +&3
% -- + " % &3" + &" ' 6 + !)3"& $) 3*/+$/$ % "#$) # $-3! " % ,$!/31 "*
"# #&+ % ' 6 $- "# % 6& 3$ " & % % -- + " /&+" ,/& " "# " &! 7) ,83*

N
( ,&! ,$ -) % & % ! 1 % "# ,$ , /" . !/3* 6+ " "# % -- + " ,$ , /" %$6
( % & #& % 8 ",# #& , % & #&+ % ) % + "& % .
# $%* A$)+ &3 & %$,)! " "#&" ! & " "$ ) % "# % + 6 "# &33 "#
+ ,$!! % % ,#& & % + )3" -+$! "# &+ & ! " . # + & , "+&3/+$0," /$+"&3
6# + "# 3&" " ' + $ $-"# A #$)3% ( &,, (3 "$ &33 "# /+$0,". D &")+ $-"#
/+$0," /$+"&3 ,3)% "# /$ ( 3"* "$ + , ' !& 3 6# & ,#& "$ & %$,)! "
,$ , + & , +"& &+ & + , ' %. " "# " ! $-"# ")%* "# /+$0," /$+"&36& 6
& % " ) 6& $" ( !$ "$+ %.

5.2.5 Characteristics of methods

,#&+&," + " ,$-"# ) $-"# -$)+ ! "#$% "#&" "# * &+ &")+&33* &%&/" % "$ "#
")&" $ 4 % +& % 3 ;;<5. "#$% "# H 33 ? ! "#$% ,& ( + &+% % &
$%% ()" / +! " % , $ "&8 (* "# % + 6 "# #" "$ "# " ,# ,&3% "& 3
& % 6 "# !& & ! " ,# ,8 "# ,$ , /" (+ -3*. # ! "#$% ,$++$($+&" "#
% ,) $ "# + &+,# (3$,8 $ &($)" "# "6$ % $-% , $ !&8 "# $--,&3
% & % "# + &3 ()" -$+!&3/+$, . "#$% & % + !(3 !$+ "# ! "#$%
% ' 3$/ % % , , + &+,#. "#$% ) $! - &")+ -+$! "# ) #
! "#$% 4 &#3& % ": @@C5 ()" %$ "# + 1/3,"3* )! /3) $+ ! ) $+
& 6 #" "$ ,+ " + &. " &% + 3&" ' !/$+"& , $-,+ " + & (& % $ "# /+ ' $)
1/ + , $-"# /&+" ,/&" % + . # 6& &3$ $( +' % & & ,#&+&," + " ,$-#$6
$! " &! 6$+8 % 6 "# !&"+ 1 (& % '&3)&" $ ! "#$% + &+,# (3$,8 $ . "#$%
+ !(3 "# 3) $" " &3 $ , + ' +,$! ,$ , + 4 ,5 ! "#$% ()"
!$+ + - % "# "#&" !$+ -&,"$+ $- "+&" , '&3) -$+ "# ,$!/& * &+
,3)% % & % 7) " $ "$ ) % "# 3," $ &+ "+$%), %. # $ /& + &""&,# % "$
"# ! "#$% ,$ , 3*% ,+ ( "# - &")+ $- &,# ,$ , /". $ 1/3," & 6 + $ 6#&" "#
( " ,$ , /" &// &+ -+$! "# ! "#$% 1, /" -$+ ! "#$% %) "$ " ! + 3* $--,&3
,#&+&," +. "#$% & % )//$+" % + 1/3$+ & % &++& %&"& -+$!
% -- + " ,$ , /" . "#$% &%&/" % "$ "# + - ! " $- $ ,$ , /" "$ % "& 3
% . " + !(3 6#&" 6& /+ ' $) 3* ,&33% & "& %&+% % ,$!/& * /+ ,/3 $-
6$+8 . " & 3' %$,)! " & 8 % $- +&/# ,&3 ! " $" #$6 "#
,$!/$ " % ,&" /+$(3! & % #$6 "$ $3' "# /+$(3! .

5.3 Research block three: analysis of methods used for concept

selection
# " +' 6 ")% + &+,# (3$,8 $ ,+ & % "# % ,) $ &($)" %
! "#$% ( "6 + & % $ $-"# + &+,# + . -" + "# " +' 6 ")%* ( "6
&+,# ;; & % ) ) " ;; -' " +' 6 ,&! (&,8 "$ "# + &+,# + 6 "# 7) " $
&($)" % ! "#$% & )!( + $- " ! . %% " $ &33* $ -$)+ $,,& $ %)+ "#&"
/ + $% + 6 "# 6#$! "# + &+,# + #&% $" #&% /+ ' $) ,$ "&," 7) + % &($)"
! "#$% & + ,$!! % % (* $! $-"# " +' 6 . # & ! $-"# ! " +& %
-+$! % ,) & ! "#$% "$ ( ) % -$+ & / ,-, "& 8 '&3)&" "# + 3&( 3"* $-&
&3+ &%* ) % ! "#$% '&3)&" & ! "#$% ) " % (* & ,$ )3"& " $+ '&3)&" &
$--,&3! "#$% "# ,$!/& *. # + &+,# + #&% ! " $-"# "*/ ; $-6# ,#
6 + &($)" ! "#$% -$+ 3," $ $-,$ , /" . D ' $-"# " ! " 6 + &($)" &3+ &%*
) % ! "#$% . + ,&! 6 "# %$,)! "&" $ $-"# ! "#$% ) % & % /+$' % %
1/3& &" $ $-#$6 "# * #&% ( ) %. # * 6& " % "$ % ,) "# ! "#$% ( ,&) "#
" &! $+ /&+" $-"# " &! 6& $" &" - % 6 "# "# + )3" $("& % 6 "# "# ! "#$%.
>
+ /$+" % + &+,# (3$,8 $ ! "#$% &+ -+ 7) "3* ) % 6 "# &% #$,!$% -,&" $ "$
&%&/" "# ! "#$% "$ "# / ,-, ")&" $ . $% -,&" $ "$ ! "#$% 6 + -$) % "$
$+!&33* /+$%), "6$ "*/ $- -- ," "# !/3! " % ! "#$%J ,+ &" ' &%&/"&" $ $-
! "#$% & % ) + 3&(3 !$% -,&" $ $- ! "#$%. " ,& &3$ #&// "#&" "# &!
!/3! " % ! "#$% "# + &+ ,+ &" ' &%&/"&" $ & % ) + 3&(3 !$% -,&" $ .
+ &" ' &%&/"&" $ $- ! "#$% $,,)+ 6# & ! "#$% &//3 % "$ & ")&" $ 6 "#
!$% -,&" $ "#&" ,+ & " '&3) & % 6 "#$)" + %), " + 3&( 3"*. 1&!/3 $-
,+ &" ' &%&/"&" $ $-! "#$% &+ -$) % + &+,# (3$,8 "6$. D$+ "& , ! "#$%
,& ( & & &%&/"&" $ $-"# ,! "#$%. # /+ ,/3 $-"# ,! "#$%
"$ % /3&* "# &%'& "& & % 3! "&" $ $- ,$ , /" & % "# 6&* "$ $' +,$!
3! "&" $ . ! "#$% "# &%'& "& & % % &%'& "& &+ % " - % "#+$) # &
)!( + $-/+ ,$ , ' % /&+&! " + 43' 3$-% --,)3"* &-- ," /3&"-$+! !& %+ ' + ",.5
& % & )!( + $- 7) " $ /$ % -$+ ) & ,+ /+$, . # !$% -,&" $
,+ & "# '&3) $- "# ! "#$% , "# * ,3)% -) %&! "&3 3," $ -&,"$+
% " - % -$+ "# 8 % $-/+$(3! .
+ 3&(3 !$% -,&" $ $-! "#$% $,,)+ 6# & ! "#$% &//3 % "$ & ) ) "&(3
")&" $ $+ 6 "# !$% -,&" $ "#&" + %), " + 3&( 3"*. -$+") &" !$% -,&" $ $-
)( ! "#$% $,,)+ -$+ & 3&,8 $- ) % + "& % $- "# ! "#$%. )!( + $- -+ 7) "
! "&8 3&% "$ ) + 3&(3 !$% -,&" $ 6 + % " - % "# ) $- ! "#$% -$+
,$ , /" 3," $ . 6$ $-"# ! &+ /+$' % % # + & 1&!/3 J
- # ) # 38 ! "#$% $-D )+ 6& ) % &" 2 "$ 3," & ) / $ .
D )+ + /+ " -$)+ % -- + " $3)" $ ()" "# + &3example contained twelve. It has
been simplified to make it legible, and confidential data has been deleted. # ! "#$% 6&
) % ( ,&) "# $3)" $ / +-$+!& , 6& $" /+ , 3* 8 $6 & % " 6& !$+
(3 "$ '&3)&" "# &3" + &" ' " +! $- ( "" + $+ 6$+ 6 "# + / ," "$ &
+ - + , "#& &( $3)" " +! . $6 ' + + $$ 6& " % "$ "+$%),
6 #" "+* "$ + -3," "# !/$+"& , $-&""+ ()" & % % -- + " 3' 3 $-( "" +
& % 6$+ . # 3% "$ )! + ,&3+ )3" "#&" 6 + % --,)3" "$ +& /. # * ,$)3% $"
) % + "& % 6#&" . . H G? ! & + 3&" $ "$ H G?. # * % % $" 8 $6 - "#$
)!( + + /+ " % & ( % -- + , + &3/ +-$+!& , . This type of misuse of the Pugh
method is very common and completely destroys its goal, i.e. permitting comparison without
detailed quantifiable knowledge of the solutions.
- $"# + ,$!!$ ! "&8 6 "# "# ) # "*/ $-!&"+ 1 "$ ( 3 ' "#&" "# + )3"
$("& % (*,$!/&+ % -- + " ,$ , /" 6 "# & + - + , ,& ( ) % "$ ,$!/&+ "#
,$ , /" / +-$+!& , 6 "# &,# $"# +. -"# + - + , ,$ , /" / +-$+! % -,& "3*
(&% "# ,$!/&+ $ $-"# ,$ , /" 6 "# "# + - + , 6 33 $" # #3 #" "# /$" " &3
% -- + , / +-$+!& , ( "6 ,$ , /" . -&," ) 3 "# + - + , ,$ , /"
/ +-$+!& , 1&,"3* "# ! & / +-$+!& , $-&33,$ , /" ) % 6 "# + / ," "$ &33
,+ " + & $ % -- + , ( "6 ,$ , /" / +-$+!& , 6 33( # #3 #" % + 3&(3*.
$3)" $ "$ "# /+$(3! "# ),, ' 3! &" $ $-,$ , /" . D$+ "& , "#
1&!/3 $-D )+ " ,& ( "&" % "#&" ,$ , /" 6$+ "#& "# + - + , & %
"#&" "# + - + , 6$+ "#& ,$ , /" & % 2. # + -$+ "# + - + ,
& % ,$ , /" ,& ( % ,&+% %. # & $"# + ,$ , /" ,& ( "&8 & & + - + , -$+
,$!/&+ $ 6 "# "# $"# + ,$ , /" . # ,$ , /" &+ /+$ + ' 3* 3! &" % ) " 3
$ 3-".

@
D )+ . ) # 38 ! "#$% & ) % &" 2 4D+$! & &+3 + ")%*(*"6$ $-"# &)"#$+ 5

6. Conclusions
# /&/ + #& ( & 0
$)+ * "#+$) # & 3&+ &!$) " $- '&+ % %&"&. + "
,$ ' " "$ /+$' % -,& , "$ "# & !(3& $-"# -$+!&" $ (* 1/3& "#
!& ) & % "# + !/3,&" $ .
% " ," $ #& ( $( +' % ( "6 1/3," ! "#$% /+$, %)+ &//3 % &,,$+%
"$ /3& -$+ & ,$ , "+&" % / + $% $-" ! "$ &,# ' & $&3 & % $//$+") " , ,#& $-
&," $ . . &," $ "&8 %)+ % -- + " $,,& $ (*!&8 "# !$ " $-"# !/+$!/")
$//$+") " "#&" -$+! /&+" $-& +&3 $&3 & &3+ &%* ) " % (* 2 + 4 @@<5 & %
% + & % 3 4 ;;<5. # ) $- 1/3," ! "#$% -+ 7) "3* /+$!$" % (*
/+$, $+ " % / $/3 " &! % . !/3! " % 1/3," ! "#$% $-" + !(3
"#$ % ,+ ( % (* &,&% ! & , % , , ! "#$% % + ' -+$! $( +'&" $
%) "+* ()" "# * &+ &//3 % 6 "# !$% -,&" $ "$ &%&/" "$ "# /+$(3! &" #& %. "#
/&/ + "# -$,) #& ( $ ,$ , /" 3," $ ! "#$% . " ,$)3% ( "#&" $! "*/ $-
! "#$% &+ !$+ 3&(3 "#& $"# + "$ ( !$% - % 6# &//3 %. //3 % ! "#$%
+ !(3 &,&% ! ,! "#$% "# "#&" "# * ,$ "& !& * $-"# + /+ ,/3 . " #&
&3$ ( $( +' % "#&" !$% -,&" $ $! " ! 3&% "$ ! "#$% 6 "# & # # + '&3) ()"
$"# + "$ ) + 3&(3 + )3" . # /# $! & #&' ( ,&33% ,+ &" ' &%&/"&" $ $-
! "#$% & % 6+$ 3," $ $- )( ! "#$% "# /&/ +.
" #& ( $( +' % "#&" $! + + &% ($$8 & % /&/ + &($)" % ! "#$%
&+ ,$ "&," 6 "# ) ' + " % ' 3$/ ! "#$% & % &+ " + " % 8 $6 6#&"
"# /+$, %)+ $-$"# + ,$!/& &+ . # * !&8 ) $-"# 8 $63% "# * &,7) +
"#+$) # "# $)+, % -- + " 6&* . $! + -$33$6 ! "#$% (* "# ($$8
$"# + ) "# /+ ,/3 $+ /&+" $-"# ! "#$% "$ ,+ &" "# + $6 ! "#$% & % $"# +
) ! "#$% & & 6&* $-"# 8 . D$33$6 ! "#$% (* "# ($$8 + -) % (* !& *
+ 6#$ 6& " "$ #&' "# -+ %$! "$ "# 8. # &() %& , $-% -- + " $/ $
;
&($)" #$6 * " !&" ,% #$)3% ( %) "+* + !(3 "#&" $-&,&% ! &. -- + "
6&* $-6$+8 ) " % -- + " / $/3. )+ $) 3* "# + & " +' 6 ")%*
6#$ 6 + 6 33+ &+% % /+$- $ &33* (* "# + ,$33& ) & % !& & ! " &+ /+$,
$+ " % / $/3. 6$ $-"# -$)+ ! "#$% % ,+ ( % 6 + 1&!/3 $-,+ &" ' &%&/"&" $
$-! "#$% .
# + &+ ! "#$% "#&" ( ,$! !/3! " % ( ,&) $-& !& & ! " % , $ . $)+
&+ "# $+ & % & % "$$3 ( ,$! &'& 3&(3 "#&" !& * + &+ ,$)+& % "$ ) .
"#$% !&* " 33 ' ")&33* ( &(& %$ % ()" $! "# $- "# ! + !& "#
,$!/& *J "# + /+ ,/3 & % "# ) % + "& % $- 6#* "# ! "#$% % % $" ) " "#
,$!/& *. # ) % + "& % & + -3," $ &($)" #$6 "# *6$+8 + 3&" $ "$ $&3 & %
/+$' % "# ! 6 "# #" "$ #$6 "# * #$)3% 6$+8.
! "#$% + &+,# #& "# + -$+ & !/$+"& " ,$ "+ ()" $ "$ !&8 )//$+"
% + . ! "#$% + &+,# #$)3% ,$' + #$6 "$ &'$ % 6+$ 3," $ $- )(
! "#$% #$6 "$ /+$!$" ,+ &" ' &%&/"&" $ $-! "#$% & % #$6 "$ $3' "# &/
% ! "#$%$3$ *. +&3 "# ,$ "+ ()" $ #$)3% /+$' % "# ! & "#&" &33$6
/+&," " $ + "$ ) % + "& % & % + -3," )/$ "# + $6 % &," ' "*.
References
+&)0
$ % ""$ "$ &!/ 33$ + D. B+ #" 4 @@C5 # )" 3:&" $ $-
/+$%)," % ' 3$/! " ! "#$% J & )+' * $- %) "+*. A$)+ &3$- + N // CG
NN
% + 3 4 ;;<5 "# )/ + $+ "* $- $//$+") " ,% "+&" %)+ &+3*
!($% ! " % . J &+0 & $', 4 %5 +$, % $- ;;< )(+$' 8 &* N ;.
D&,)3"*$- ,#& ,&3& % &'&3 +,# " ,")+ & + (= $, "* 3& $6 // N
+8#$- + % !& 3( + + 4 ;; 5 +$%)," % ' 3$/! " & & "+),")+ % & %
" +&," ' "6$+8 $-8 $63% I & + '$3)" $ &+*&//+$&,#. J )33* A )--* , &#$
B&33&, 4 % 5 +$, % $- ; 3& $6 ) ) " . +$- $ &3 +
)(3 # // <GN <C<
3 4 ;; 5 B#&" "# "# ,&33% &+,#R J +$, % $- ;; "?3
! &+ $ $ C > &* // C.
3 4 ;; 5 B#&" "# "# ,&33% % + &+,#R * $" /&/ + J D$38 $ &+V
$+ 33 33 + 4 % 5 /+$, % $- ; "$,8#$3! ) ) " @
3 #&8+&(&+" B&33&, 4 @@>5 + I "# * "$ ),, -)3 ' 3$/! ".
/+ + 2 +3& $ %$
3$!( + A &,$! A $ # + 6 "$ B&33 4 @@ 5 "# $ +&/# ,D 3% "#$% & % # +
3&" $ "$ . J #)3+ &! $8& 4 % 5 &+" ,/&"$+* + ,/3 & % +&," , . .
+3(&)! $,&" 33%&3 // GG
*3) % D+ %+ , $ #$!/ $ 4 ;; 5 % /+$, -$+ ,$!/31 ! ,#& ,&3 "+),")+
) +$/ +"* & % $% 3 6 "# &//3,&" $ "$ ,&+ ($% . J &+0 & $' , 4 %5 +$, % $-
;; $ - + , < N $- &* ;; )(+$' 8 +$&" &. D&,)3"*$- ,#& ,&3& % &'&3
+,# " ,")+ & + (= $, "* 3& $6 // C C ;
*3) % +& " 9/ : & 4 ;; 5 &( 3"* %) "+* $-! "#$% -+$! % + &+,#.
J D$38 $ &+V $+ 33 33 + 4 % 5 /+$, % $- ; "$,8#$3!
) ) " @
& "&! & 4 @@@5 ( " /+&," , ,&/&( 3" & % / +-$+!& , . A$)+ &3$- +
; // ;G >
)($ &%% 4 ;; 5 * " !&" , ,$!( J & &(%)," ' &//+$&,# "$ ,& + &+,#.
A$)+ &3$- ) &+,# 2$3. GG ;; // GG GC;
% + B 4 @@>5 !$% 3 I % , , &//+$&,# 4& % 6#*%$ %) "+* $" ) "R5.
A$)+ &3$- + @ // GG N
+ &)" !! &33 #&! 4 % 5 4 ;; 5 M)&3"&" ' !&+8 " + &+,# J /+ ,/3 & %
/+&," , . & $ %$
D$+% A D$+% B , &!&+& 4 ;; 5 "& , & % "# (&,8 +$) % ,$ ' + &" $ "$ ,#& .
A$)+ &3$- + & :&" $ &3,#& 2$3. G ) // ;G
D$++ " + A 4 @C 5 %) "+ &3 * &! , . & &!(+ %
D+$ " 4 @@@5 B#* %$ %) "+* $+ % , , R A$)+ &3$- + ; //
; ;<
&+' 4 @>N5 $!/ " $ "# #" ! $ $- M)&3"*. &+'&+% ) ' 6
$' !( + , !( + / ;
&) + A 3&) 4 @>>5 # $) $-M)&3"*. &+'&+% ) ' 6 &* A) /C
 $$8 4 @CC5 2 " 8&/3 & -$+ W8. + !& "$,8#$3!
)(8& 2 4 @>;5 +! $3$ * $- "# , , $- + C & )& . X+ ,#
6 ": +3& %.
A$ A 4 @@ % %5 ! "#$% . A$# B 3* Q $ 6 $+8 # ,# " + B # !
+ (& &/$+ $+$ "$
33& % + A 4 ;; 5 B#* % ! "#$%$3$ &+ % --,)3" "$ !/3! ". ". A. ,# $3$ *
& & ! " 2$3. $ . P< // N NC
+"$ A 4 @@<5 %&/"$+ & % $'&"$+ . "*3 $-,+ &" ' "* & % /+$(3! $3' . $)"3%
$ %$
$"" + A 4 @@G5 &% ,#& J 6#*"+& -$+!&" $ &3 --$+" -& 3. &+'&+% ) ' 6 2$3.
N ) // G@ CN
9/ : & #$!/ $ 4 ;; 5 # &//3,&" $ $- "# < !$% 3 "$ "# !& & ! " $-
,+ &" ' "* Q $'&" $ /+$%)," % ' 3$/! ". J @"# " + &" $ &3 +$%)," ' 3$/! "
& & ! " $ - + , N > &* $/# & " /$3 . Y,$3 % % &+ & % )+$/ &
" ")" -$+ %'& , % ")% & & ! "4 5 &+ // G>N C;
$/ : & #$!/ $ 4 ;;<5 ")%*$-& $--"# # 3- $3)" $ !$% 3$- $'&" $ . &/ +
&,, /" % -$+ ,$ - + , /3& % -$+ A)3* < ;;< & % + ,# %)3% ) " 3 &+,# ;;G
&)+ + 4 @@C5 + "& , "$ () 3% )//$+" -$+ ,#& . A$)+ &3$-M)&3"* Q &+" ,/&" $
2$3. @ ) // GC C
3 4 @>@ +% %5 ,# 7) $-2&3) &3* & % + . 3& $+ 3 B&38 +
&! 4 ;; 5 1 $!&" ,% J &%'& , & % &//3,&" $ . 1-$+% ' + "* + 6 $+8
1-$+%.
? $ $+ A *!$)+ A 4 @@;5 "+$%), )+$ ) " , +$ +&!! J *,#$3$ ,&3 8 33 -$+
% + "& % & % -3) , $/3. #$+ $ $ %$
3' + 33# + & A 4A+5 4 @@N5 + $!/31 * " ! 6 "# !$% 3 & % $(0," .
, +&6 33 6 $+8 & D+& , ,$ B& # "$ )83& % $ $"& &+&,& ($ &%+ %
1 ,$ "* 3& $ "+ &3 6 3# & A)& &/$+ *% * $8*$ $+$ "$
:8 &8& 4 @@;5 & %($$8 $-7)&3"* "$$3. # A&/& &//+$&,#. +$%)," '"* +
&!(+ %
&#3 ": 4 @@C5 + % J & * " !&" ,&//+$&,#. /+ + + &" + "&
) # 4 @@ 5 $"&3 J " +&" % "#$% -$+ ),, -)3 +$%)," + . % $
B 3* B$8 #&!
$#&" 8* 4 ;; 5 & $ "# &/ ( "6 ! "#$%$3$ * $- + % & %
%) "+ &3/+&," , . J )33* A )--* , &#$ . B&33&, 4 % 5 +$, % $- ;
3& $6 ) ) " . . .J +$- $ &3 + )(3 # // < G
,8&-) 4 @@N5 - % "+),")+ % ' "' # 8 J $6 "$ ' ". " 33,8 +$ 3
,# & .
" !/3 A &%8 ,#&)( 4 ;; 5 # 8 % " &! & & &3* $-" &! ,$!!) ,&" $ .
")% 2$3. ;; // <N <@C
& ),# 3&) 4 @@;5 $() " M)&3"*. &+'&+% ) ' 6 A& )&+* D (+)&+* @@;
// CG NG
#$!/ $ 4 @@@5 !/+$' & "& &( 3"* & % 3&( 3"* "#+$) # . +$- $ &3
+ )(3 # .
2& % + 0
% 4 @@C5 , &+ $ J # +" $- "+&" , $ ' + &" $ . A$# B 3* Q $
# ,# " +
2 #&+ A 33+ D+ " 4 @@@5 + &" '"* ) ($) %. "+$%)," $ "$ + &" ' +$(3!
$3' . $'&" $ * " ! +$)/ 6 $+8
2 + B 4 @@<5 + & &" $ $- % &," '" J $//$+") " , 6 "# # +&+,# ,&3 / $% .
" +&," 6 "# $!/)" + 2$3. C $. // G >
B# " * 4 @@>5 & )-&,")+ (*% . &+'&+% ) ' 6 CC A)3*P ) ) " // < >
B$!&,8 A A$ $$ 4 @@;5 # &,# "#&" ,#& % "# 6$+3%J # "$+* $- &
+$%)," $ . &16 33 &,! 33& " + &" $ &3 6 $+8 &6 $ $,&" $+$ "$ $33 +
&,! 33& & &%& 6 $+8 1-$+% &/$+ *% *J.
B) DA &!&%& . 4 ;; 5 1/ + ! " J /3& & &3* & % /&+&! " + % $/" ! :&" $ .
B 3* 6 $+8 # ,# " +
4 @@<5 & ")%* &+,#. & % "#$% . & )(3,&" $ #$) & % &8

<
Appendix 1
" $-! "#$% " "$ "# " +' 6 J
#3 #" ' + (+& "$+!
-- "*% & +&! ' "&" $ &3 " !
)3" -&," / ,8 )/ # 3&%% + $-&( "+&," $
1 "# 8 #&" B$+% %& ,
3&" $ % & +&! + $ &3& &3$ *
&+% $+" !& &+*(+& "$+!
& + % ,$!/&+ $ & &3* 4 5 &) & % -- ," % & +&!
B #" % $(0," ' "+ 4$+ - #($ ,#&+"5
" ' "*& % , +"& "*& &3* GB 4$+ "# 1 ) ' + &37) " $ 5
) # ! "#$% +& "$+!
&3* +&/# $- 33/ +& "$+! 6 "# $ " "
,# + 8 &++&* +& 6+ " /$$3
3," $ ,#&+" &+% ,+,)3&"
)!/" $ !& # + ," & &3$ *
$!/&" ( 3"*!&"+ 1 ,")+ -$3% + (+& 6+ "
" 3"*-) ," $ 4$+ '&3) -. *!($3,& &3$ *
$+ % +&( 3"*-.5 & %$! /)"
+-$+!& , /&+&! " + I2&3) ,&3 #&+" 2 )&3,$ ," $
M)&3"*D) ," $ /3$*! " 4MD 5
$%)3&+ D) ," $ /3$*! " 4 D 5
&"+ 1 & +&! + &" ' +$(3! $3' # ,83 "
,# ,$ $! ,+&" ,#&+"
'&3)&" $ ,#&+" ""+ ()" 3" 4$+ &""+ ()"
!$% -* 5
+$-3 ,$ " , % &"$$ 3) /+ "
&+&! " + /+$-3 !&"+ 1 * " !&" ,%$)("
+&( 3"*-) ," $ $/" ! &" $ % !&/
' 63 " $+/#$3$ ,&3,#&+"
D& 3)+ !$% & % !& "& &( 3"* & &3* 4$+ !$+/#$3$ ,&3!&"+ 15
D& 3)+ !$% & % -- ," & &3* 4D 5 3& -,&" $ ,# !&
D&)3" + &3* 4D 5 4$+ ,#&+&," + " , ,3& -,&" $ 5
&:&+% & % $/ +&( 3"*4 5 $ "+&% ," $ "&(3 4 5
8& ! " ,&"&3$ )
M)&3"* ,#!&+8 /3$*! " 4M 5 D) ," $ & &3*
# ,83 " &,86&+% " /
$ "+$3,#&+" 4$+ ,$ ' + " "#$) #"5
&+ +&/# (0," ' "+
+&/# 4$+ +) ,#&+"5 +-$+!& , / ,-,&" $
,&"" + % & +&! C G
&+ "$ % & +&! $") 3$ $! ,# 7)
"$ +&! &33+*
+$, % , $ /+$ +&! ,#&+" 3/#
2&3) + D$+6&+% " / 4$+ % ' + " "#$) #"5
3) /$" " &3 ,$ , + $' +,$! "$+*($&+%
,$ , + 4 ,5 "# + R

G
Appendix 2
1, +/" $-"# " +' 6 1/3& 6#*! "#$% &+ $" ) % !$+ $-" J
B- 1 " " # +8 " " " 5. # + 8 "
#" 8 " D# 8 " " B
8 "8 C5 8 " " 8 8
" " 8 " 8 # +"+ 8"
8 " 8 " " "## # " 8 " " 8 + "
8 " + " CF B% 7-E #1 " "
+ " )' " " -" 1 8 B:" #" #" #
8 " ;C" " " B< " " + "
" +" C - " B/ G" " ;C
B< 8" 8" " 8 "# " " " " D "
" " C5% "-" " # +"+ 555 " G " CF 5B<
G 8 " " 5< 8 " +
8 " 8" G 8" " F D
" " C
B-G "" 8 " # # # " " + " +
+ CF B< 8 C
B/ " # # " " 5< 1 1 " " " G 8# 8
8 C
BH " " + "" 8 + " 8 8 " CF B, " 5
H " 8"1 "I 1 5- " " 8" " I F
- G 1 5- 1 " + " "+ 8 5< " " "1
" 8 # 5J 8 8# " " 8"1 " "F
" " -8 " " + 8 + " G + 8 5/
" "- G 1 " "8 85- +
" " 8" + E "1 8 8 8 + 8 " 8
" " " "8 E F J D"8# # E -G8 1
E - 1 E " 8# " "1 + F G
8" CF B< 8 " + 8 "
8 5 " 1 " 1 8 " " "
" 8 5 "## 8 " 1
" 8 " " K " "G " + " " +
" G " 8 " " "- 8 "
1 F - " " G - " G 1 "
1 8 5C
B- " + " 8 CF B- " G 8 -" G +
" " - " 8 CF B- 1 8 8" " 8 5
% 8" + " 8" + G " " " # # " #"
8#" "" CF B- + " 8 G " D# 8
G " " 1 "+ " " "+ ";CF B- 1 8"
# # 1 G 8 1 " 1 8"1
" CF B< 8"1 " G F " 8 " " D#
" " " "8 + " " "
" "1 8 8 D# " 8 "G "
+"" I " "5% # # CF

C
B- + " " 8# # "88 5= - " 8" 8 "
#" 8#" 8" D5-G "+ D 8" DCF B< " 8" " "
1 " "# # " " " # 5- " "1 " 5% G
+ " " " 8# G 5 " " I"
" 8 " # 8 "8 " " 8 " "
" C
B- 1 " " 8 "+ 8 " + G" # CF B-
1 "8 1 " 8 + 8" " 5- "8" 8 " "
1 " 8" + CF B% " # " "8+ " "+ 8 "
8" CF B, 8 " 8# " CF B- 1 " "
" 8 " " "## " D " 5% 8 8
D 8"CF B- 1 " 3J4 + 8 " 5 " " 8 "
#" " "## " " I 5, " + " " " 1 " G " " +D
1 " C
B " 8 " 8 C
B " " " 8# " " 8+ " G
1 "G 8 " " F - 1 " " 8
8# " CF BH G " "8 # C
B9 8" - G " " " 8 + " D# - " "
1 " ## 1 " " " 5/ " " "1 " 8 " - 1
8 " ## " +" I " #
8 " # 8 # + #" " " 8 " 5%
- 1 8 8# " 5- G " " 8 "
" " D"8# + " 8 8 8 " "
+ 8" " " 8 " " 8 K L "
# # " " " " " " # G5 G8 "I
" # + # " ## 5H " " ## "G " "+
" # #" 8 + "+ # "" " 8 " 5 "
1 5 -1 " + " + " " 1
" " " " " " " " " " 8
## + 8" # + 8 " # + ## # 8
CF B9 - G 1 59 " 8 " " #8 " # E 5/
- 1 " " " " F 8 " F
< " # 1 " " 8 5< " - " - 1+" 1 " " " "
" " " " " " " " - 15/ G " "
"8 " E " 1 CF B- 1 " + 8
" " 8 CF B 8" # + 8 - 1 " " "# " "+ "
" # 5% " 8 "- " # + 8" # " 8" "
" # # 5 " " # "8 #
" 5- 1 CF BJ 8 " " "G # + 8
8 " 8 " 8 -1 " # 8" DC
B- 1 " 8 " " " # #"
" 8 5 - 1 8 8 + 8 + 5, " +
8 " + " 8" + G " 5- G 1 5/ - 1 "
# #" " 8 C
B$ # " 8 8 " 8 " " "
8# " 8" + " 8 1 " 1" 5/ " " "
8 " " 8 8 " " 5 G" # #
+ " " -CF B- 1# # " 8 1

N
 " " 8 1 " " 8 1 " 8#
8 5H +" 1 " #" #" 15. "
D# " 8 + " " " " 1 5- " " "1 8 5CF B G"
# # " "8 " " G " " "5CF
BH " 1" # # " " 1 # " 1 #
" 5/ # # " I " " " 5CF B%
" 8" # # 8" " 8 "8 " # # G +
5 "G 8 # 5C
B- G 1 8 " 8 5- # 8 C
B< - " G 8 8 " "" " " 5 " " 1
"" 5. 8 + 8 " 8 " " " "
8 8 " + 5C
B. 8#" " " + " " 8 5 1 +
"8 8 + G 8 1+ " G
8" + G 8 " " + 5C
B, " + " G " " 8 " C
B% " "" " + # " ''' # + +
" " 1 " " 5% " + " # " 8" #
" 8# " D" " " 15% "#
" " " " # " 5-G "
" " # " 8" " # " + " 1 5<
" " 1 #8 #" 8 5. 8 " #8
# E # 5% # "# "+ 5 1 "8 " 5 " G
" # + 85 "# + 8 8" # E " # "88 "
" 1 " " " 1 8 5% " 1 8
" 8# G I 5 "G " + # "
+ 8 " " " + " " " "# " #
" # E " " 5CF B- " "1 8 + "1 1
8 CF B " + # "" "+ C
B-G " I 5- 1 G8 " " " 1 8"
8 ; " " " " 8 "
" " " " E " 8# 5- "
" " #"# " " 8 + # 8" 5-
1 " " 8# " 8# " - 1 " " "
" 8" " 1+ " " 15C
B< " " " " " " " " 5 8" 8 "
8" " 5 8 " 8" " 8 " " C
E $/3 %$ $" #&' !),# " ! "$ 3&+ 6 6&* $-"# 8 . "# + 6$+8 "# * &+ $
"+ %. $! / $/3 !&* #&' % --,)3" "$ ) % + "& % & 6 ! "#$%. B 33 "$ ,#&
*$)+ 6&* $- "# 8 & #&+% 6$+8. $/3 &+ &36&* & ( " 3&:*FS E - *$) #&'
$! "# "#&" 6$+8 7) " 6 33 " "&8 & 3$" "$ ,$ ' , "#&" )%% 3* *$) #$)3% $
%$6 "$ : +$ "$ "# &-" +6&+% $ "$ & # # + 3' 3. " & + 8 *$) "&8 & % "#
' "! " " ! *$) 6& " "$ ( )+ "#&" "? ( "" + -)")+ &-" +6&+% F.

>
 Paper E
Enhanced Engineering Design Practice Using
Knowledge Enabled Engineering with Simulation Methods
In the proceedings of Design 2004 Conference, 18-20 of May 2004, Dubrovnik, Croatia.
INTERNATIONAL DESIGN CONFERENCE - DESIGN 2004
Dubrovnik, May 18 - 21, 2004.

ENHANCED ENGINEERING DESIGN PRACTICE USING

KNOWLEDGE ENABLED ENGINEERING WITH
SIMULATION METHODS

N. Bylund1,3, O. Isaksson2, V. Kalhori3 and T. Larsson3

1
Volvo Car Corporation: nbylund@volvocars.com
2
Volvo Aero Corporation: ola.isaksson@volvo.com
3
Luleå University of Technology: vahid.kahlori@ltu.se, tobias.c.larsson@ltu.se

Engineering design, knowledge enabled engineering, simulation

1. Introduction

The objective of this paper is to discuss how Knowledge Enabled Engineering, when combined with
simulation methods is a development step for product development processes, engineering design
methods and evaluation support systems. The paper opens the discussion on how these approaches,
i.e. work methods, simulation support and Knowledge Enabled Engineering (KEE) methods affects
best practice in engineering design (ED) by adding synthesis support to the already existing analysis
support. In the presented work the authors discuss the actual state of industrial applications, with
challenges and opportunities, at Volvo Car Corporation, automotive manufacturer, and Volvo Aero
Corporation, jet engine component manufacturer, both operating in Sweden.

1.1 Engineering Design

The company specific product development (PD) process is the process beginning with the
perception of a market opportunity and leading to the delivery of the product. The generic view of
the mechanical engineering part of the PD process is the engineering design (ED) process, a
decision-making process [Hazelrigg 1997] that transforms needs and requirements into verified
solutions. Most frequently, limitations in lead time and cost restricts the time available for
engineering work. Two areas of technology that are used to enhance engineering work are
highlighted here. One area is Knowledge Enabled Engineering, which represents a methodology
used to capture and reuse this knowledge in computer aided design systems. The other area is
computer simulations where knowledge of properties and behaviour of forthcoming products may
be predicted. KEE includes the more traditional term KBE and similar knowledge rich strategies.

Knowledge Based Engineering was developed during the 80’s, and gained some industrial
acceptance primarily during the 90’s within several organisations. One definition of KBE is:

“The use of advanced software techniques to reduce lead-time to capture and re-use product and
process knowledge in an integrated way” [Stokes 2001]

1
The main objective is to reduce lead-time by capturing product and process knowledge with a
product model [Andreasen and Hein 1987] as the core of the system. The key concepts are that the
logics of the design object (artefact) and the actual design process is described in a way that allows
generation of design solutions (i.e. geometries and more). A KEE system is needed which provides
a language for defining an engineering design process and a user interface that allows the activation
of the design process definition and the subsequent creation of a design [Rosenfeld 1995]. Using
KEE, the advantages are quite appealing; lead time for standard work activities can be dramatically
decreased in combination with an increased and controllable quality. Standard solutions can be
generated, evaluated and reported repeatedly at a low cost for every iteration. Engineers can
concentrate on the more intellectual parts of engineering work rather than spending time doing
routine work. In this way, the design team can afford to investigate more design alternatives on a
more detailed and controlled level than what is possible using interactive approaches.

Simulation technologies are extensively used as an integral part in the design process to increase the
number of iterative synthesis-analysis loops and to decrease the total lead-time. Virtual prototyping
has proven to give insight and good support in both product development and choice of process
parameters. It is less expensive than testing, at the same time mistakes are cheap and can be
discovered and addressed earlier. Research in computational engineering has traditionally been
focused on efficient numerical solution strategies and more accurate models. The user-friendliness
has improved mostly due to effort from software providers this have made simulation tools more
available for designers. Anyhow especially the pre-processing in traditional simulation software is
time-consuming and demands deep insights in boundary conditions and material modelling, making
simulation difficult to use continuously to drive early decision-making processes.

2. Methods
The systems used as examples in this paper are made in close collaboration between researchers
from Luleå University of Technology (LTU) and Volvo Aero Corporation (VAC) and Volvo Car
Corporation (VCC). The results of this research are now in use in systems used in the product
development process at both companies.

3. Combining KEE and simulation for the benefit of Engineering Design

Combining KEE methods and simulation technologies can improve the design evaluation process.
Positive effects are the possibility of early standard analysis of design concepts; shorter analysis
cycles (i.e. creating the possibility for optimisation and more iteration) and the fact that experienced
simulation experts can spend less time on routine tasks that are done by the KEE system users
instead. Efforts in combining Finite Element Methods with KEE has been made by [Pinfold &
Chapman 2001, Isaksson 2003] where Pinfold et al. propose a rule base method concerning the
creation of first geometry of the vehicle structure and then from this a simplified model for mesh
generation and then finally the generation of the FE mesh itself. Isaksson proposes a similar
approach and emphasises the opportunity given to study a wider set of design variations than what is
traditionally possible using most parametric strategies.

3.1 Examples from auto industry and aero engine industry

Two industrial examples are presented. First it is presented how simulation technology can be
organised and tailored to contribute to the engineering design work, and in the second it is presented
an example where KEE integrates Finite Element techniques to evaluate the rules.

3.1.1 Simulation support for car design

Combining simulation of mechanical properties with Rule Based techniques has been made for car
body design at VCC, with the software DAMIDA and ADRIAN, see Figure 1. The effort to develop
2
simulation tools was earlier concentrated only to the later simulation phases where very high
accuracy is the ultimate goal. Some years ago it was realised that considerable gain could be made if
the simulation with reasonable accuracy of common assemblies in the car body, such as beams and
joints could be made faster and during the actual design activity. The geometrical design at VCC is
made by design engineers striving for a design that fulfils numerous, and often, contradictory
requirements, such as low weight, high strength and stiffness, joining possibilities, corrosion
resistance, and commonality with similar vehicles etcetera. The mechanical requirements are often
of similar nature between different cars even though numerical values vary. This makes it possible
to reuse the boundary conditions and analysis type from project to project. By standardising the
analysis procedures for some often recurrent assemblies, analysis can be made fast and with
reasonable accuracy by design engineers without deep experience of engineering analysis. The goal
of the software have been their user friendliness and that they should be safe to use, the standard
user in this case do not alter the software, but more experienced users can run advanced modes
where more user interaction is allowed.

Figure 1. DAMIDA and ADRIAN implementation at VCC.

This software consists of a main program with scripts that starts and manages commercial software
already used within the company. In this way licence costs are minimised. Furthermore by using the
software already in standard use, it is easier to gain acceptance and confidence within the company.
The software used in DAMIDA and ADRIAN are seen in Figure 1. CATIA is the CAD software
used at VCC, meshing is done internally in DAMIDA in the beam case and in the commercial
ANSA software in the case of ADRIAN for joints. The non-linear analysis for beams the beams in
DAMIDA is done with Radioss, a FEM program for explicit analysis. The linear analysis for the
joints in ADRIAN is done in Nastran, a FEM program for implicit linear analysis. The results are
presented both on a webpage and for ADRIAN also in the post processor ANIMATOR.
3
Maintenance of the software is important because the commercial software used changes version
regularly.

ADRIAN and DAMIDA changes the way the development is being done, the effect of a design
change can be seen within hours and days instead of weeks and months. Simulation can be made
more in parallel because more people are able to simulate than before.

3.1.2 Simulation support for jet engine component design

Volvo Aero has a jet engine component specialisation strategy, which enables engineering design
systems can be tailored to generate and evaluate conceptual product models in a repetitive manner
and with a short lead time. The conceptual models need to be represented as 3D CAD models,
including manufacturing and materials information and evaluation of the structural response require
analysis using Finite Element Analysis.

In a product development situation, the lead time and quality of preparing analysis models for
numerical evaluation require a significant effort. Using KEE technology, the lead time for tedious
analysis model generation can be nearly eliminated and thus allows more concepts to be studied.
KEE supported by simulation technology thus allows more design iterations which contributes to an
improved engineering design process.

Figure 2. FEA supported alternative design evaluations.

Using this system, a conceptual engineering design study for a jet engine component was conducted.
The conducted design study required Finite Element simulation for evaluation and enabled concepts
with significant configuration differences to be generated and evaluated [Isaksson, 2003], Figure 2.
The lead time for each design iteration was reduced by at least 90%, taking iterations from weeks
down to hours.

One challenge for successful deployment of simulation supported KEE, in this case, lies in the
technical contradiction of model generality and flexibility verses strict model quality and control of
the associated simulation models. Another challenge resides in the fact that a significant part of

4
conceptual design now takes place as a design systems development effort rather than the actual
product development work.

3.2 Challenges and opportunities

In the new work environment, as presented in the examples, challenges and opportunities appear.
First, there are technically oriented challenges. It is technically possible to already at an early stage
of design define product models with a high level of detail due if making use of pre-existing know-
how. It is still technically challenging to define generative models that capture the wider design
space that automatically can be automated using simulation tools. This is due to that the simulation
techniques impose additional modelling restraints on the design model. Examples of areas
undergoing rapid technical development are
Distribution and collaboration technologies
Integration of design and simulation techniques
Knowledge acquisition and maintenance support

In these areas there are many up-coming vendor solutions but few standards and tool independent
and neutral solutions.

Secondly, there are methodological challenges. The main shift is that all logical product solutions
and their combination must be defined upfront and coded into a computer application. This requires
a systems development- and maintenance work which is traditionally separated from interactive
engineering work. Often, the Knowledge models can be developed to be “good enough” for 80% of
the expected work up-front of a project. Once entering the product development work an additional
20% of systems development/up-dating is needed due to additional situation dependent
requirements. A challenge is to define and develop these engineering systems so that users still have
the necessary control and not become “black boxes”. It is then crucial to have an efficient up-dating
and re-design methodology, since such work tend to be carried out in a severely restrained time
frame. There is a need to develop simulation models adapted to the actual stage of design and
available information. By doing this the combined KEE and simulation environment will become a
design support system rather than a design verification system.

Third, there are cultural and social challenges. The new generation of engineering support systems
increasingly integrates techniques, methods and experiences from disciplines that are normally
represented among different users, such as CAD, PDM and Simulation users. Although it is a
technical reality that the systems mergers, new roles and situations appear amongst the users.
Challenges are found in that new roles are defined, such as “Knowledge Engineers”. More work is
spent by users into actually defining the design systems compared to “simply” using a pre-existing
tool from a vendor.

4. Conclusions
Systems that combine synthesis and analysis continue to be developed and increasingly deployed.
As the analysis phase can be supported by simulation, the entire design - evaluation loop can be
supported allowing iterative design. The presented industrial applications show a direction going
towards bridging simulation and Engineering Design and combining the result into design support
systems applications, based on product models containing both process and product information.

The challenges of introducing these systems can be seen on three different levels;
Technically oriented challenges
Methodological challenges
5
Cultural and social challenges

The challenge of introducing simulation technology together with KEE in Engineering Design is
rather on the methodological level than on the strict technical level. Best practice is yet to be seen,
and a new way of work is the probable result of introducing simulation supported KEE in industry.

Acknowledgement
The authors would like to acknowledge the financial support from VINNOVA via the Polhem
Laboratory, and The Foundation for Strategic Research via the ProViking research programme.
Volvo Car Corporation and Volvo Aero Corporation are also acknowledged for the access to
company specific data.

References
Andreasen, M.M. and Hein, L., "Integrated Product Development", Springer Verlag, 1987.

Isaksson, O., “A generative modeling approach to engineering design”, International Conference

on Engineering Design, ICED’03, Stockholm, August 19-21, 2003.

Hazelrigg, G.A., ”On Irrationality in Engineering Design”, Journal of Mechanical Design, vol.
119, Transactions of the ASME, 1997, pp.194-196.

Pinfold, M., Chapman, C., “The Application of KBE techniques to the FE model creation of an
automotive body structure”, Computers in Industry. vol. 44, 2001, pp 1-10.

Stokes, M., Ed. “Managing Engineering Knowledge – MOKA: Methodology for Knowledge Based
Engineering Applications”, ASME Press, 2001.

Rosenfeld, L.W., “Solid Modeling and Knowledge-Based Engineering” in “Handbook of Solid

Modeling”, LaCourse, D.E. (Ed), McGraw Hill, 1995.

Nicklas Bylund
Volvo Car Corporation 93710
PV2A2
SE-405 31 Göteborg, Sweden
Tel: +46 (0)31-325 4145, nbylund@volvocars.com

6
 Paper F
Needs, development and implementation of software for simulation driven
car body design.
Needs, development and implementation of software for simulation-driven car body design

Nicklas Bylund1,2
Lennart Karlsson2
1
Volvo Car Corporation and 2Luleå University of Technology
Key words:
Simulation based design by designers, car body, stiffness, crash, lead time, engineering design

Abstract
Companies must perform their product development with increasing efficiency to keep up with
shortening lead times and higher demands on product performance. By making the product
development simulation-driven, the properties of the products can be checked faster and at lower
cost. This paper presents and discusses the development, implementation and impact of simulation
driven-design by designers. A product development process coupled to analysis software is
presented. It is shown that developing and implementing analysis software accessible to the design
engineer improves product performances. The proposed product development process has the
potential to reduce lead time when implemented among a majority of the design engineers/ drafters.
Compared to today's development process: design in the design department and analysis in the
analysis department, simulation-driven design by designers that is checked by the analysis
department, can revolutionize product development in terms of quality and lead time.

1. Introduction
Simulation is widespread in industry as a means to reduce product development costs and lead time,
and to learn more about the product without the need for much testing. However, the possibilities of
systems that support simulation throughout the whole product development process have not been
explored to the full extent. There is still room for improving the product development performance
[King, Jones and Simner 2003] by striving towards a simulation-driven design instead of a
simulation-verified design. Product development performance is becoming increasingly important
[Anderson 94]. Production has been rationalised and the work spent on building a product (e.g. a
car) has decreased [Womack, Jones and Roos 90]. Thus, product development rationalisation is a
natural development. This paper evaluates the success of the increase in simulation support at
Volvo Car Corporation, VCC, to improve product development performance. The whole chain,
from identifying needs, proposing an alternative development process, development of simulation
software, and its implementation and subsequent measurement of its impact is presented and
discussed.
The lead time when designing a new car is several years. The design of the car body is one of the
processes that need the longest lead time. During the design process there are long loops of design
and analysis. When analysis is done only at complete body level by expert analysts it takes a lot of
time before the design engineer / drafter of the sub-system get the feedback on the performance of
his particular design. It has been suggested that supporting preliminary analysis by the design
engineers/drafters at sub-system level will make verification of solutions faster, leading to fewer
slow and costly complete car body analyses [Bylund and Eriksson 02]. To permit the design
engineers/drafters to make preliminary analysis, two methods with corresponding support tools
have been developed: ADRIAN [Bylund 04] and DAMIDA, for analysis of car body joints and car
body beams respectively. The objective of this paper is to describe the development of these tools
and to measure the impact that these tools have had in the product development of car bodies at
Volvo Car Corporation (VCC). The tools are tested in a pilot study that preliminarily measures the
impact of the tools. Further implementation is ongoing and has been used to add more details to the
impact measurement.

1
2. Method
This paper originates from research done at the VCC automotive manufacturer during four years.
The paper puts into context and elaborates on earlier publications, [Bylund and Eriksson 02,
Bylund, Fredricson and Thompson 02, Bylund 03, Bylund 04]. The work presented has been done
while Bylund was present continuously in the product development process at VCC; this has made
it possible to regularly check the needs and impact of changing the use of simulation in the product
development process [Blomberg et al 93]. Interviews and both planned and unplanned observations
[Yin 94] have been used as a means to collect data throughout the study. Participatory design has
been used to involve the stakeholders of the developed software [Blomberg et al 93]. The study
takes place in the ongoing product development of real designs, i.e. the artefacts designed have been
introduced in cars manufactured by the company and will be introduced in the company's future
products. To measure the impact of introducing the methods supported by the software the notion of
measurable criteria is introduced [Blessing 02]. A measurable criterion is one that can be measured
in an unambiguous way. The success criterion or criteria is on a higher level and may not be
directly measurable in a study limited in time. By building a logical chain of evidence between the
measurable criteria and the success criteria, success or failure of a project can be measured. The
success criteria in this study are better designs (better crashworthiness, better stiffness, lower
weight, and lower cost) in shorter lead time. The network of influencing factors [Yin 94 and
Blessing 02] can be seen in figure 1.

Shorter development Better

lead time for car body designed car
development bodies
+
+

Fewer loops of
complete body analysis

More alternative
designs analyzed
(at sub-system

Faster analysis (at

sub-system level)

Design engineers/ drafters able to use

analysis tools for sub-systems

Figure 1. Network of influencing factors.

2
Plusses at both ends of a line indicate that more of a criterion has a positive effect on the criterion at
the other end. Minus means a negative effect. In figure 1, the criteria are, from the bottom to the
top: how many design engineers that are able to use the tools and how many that use them; the
speed of analyses, both in absolute numbers and in relation to development without these tools;
more solutions are analysed; fewer analysis loops are done at complete car body level; and finally
the already mentioned success criteria. The results of the measurable criteria are presented further
on in this paper. There are also criteria related to success that are not quantifiable but nevertheless
important and closely linked to success or failure, such as the opinion and attitude of the users of
introduced software [Eckert, Clarkson and Stacey 04]. There are also criteria that have been
discovered during the study due to increased knowledge.
To be able to compare the situation without and with a requirement breakdown and methods and
tools for design engineers/drafters, the two situations are described in figure 2 and figure 3
respectively.

Requirements

Component
designer
designer
design engineer

Assembled complete shell mode

FEM analysis of complete shell model,

to check if requirements are fulfilled.

Ok Not ok, change design

Figure 2. A traditional process for car body design with

respect to mechanical properties. [Bylund and Eriksson 02]

This figure (2) describes the development process without analysis software for the design
engineers. The global mechanical requirements are not broken down. The design engineers/drafters
generate the geometry of the components with CAD tools. At predetermined moments they release
their components, i.e. they make their component(s) available to be assembled with the components
designed by their fellow engineers into a complete car body. Finally the complete model is used as a
basis for FE-analysis. If the car body does not fulfil the global requirements, some changes have to
be made by the design engineers and the loop starts all over again. The time from release to the
design engineers getting some feedback is about 4-6 weeks.

3
 Requirements If the requirement in the
PBM can not be
fulfilled, re-balancing is
PBM needed.

Single
purpose
simulation
Component tools
designer
designer
design engineer

Assembled complete shell model

FEM analysis of complete shell model,

to check if requirements are fulfilled.

Ok Not ok, change design

Figure 3. Development using requirement breakdown and corresponding methods

and tools. [Bylund and Eriksson 02]

Figure 3, describes product development when breaking down global mechanical requirements to
sub-system level and providing design engineers/ drafters with simulation tools adapted to sub-
system analysis. In this product development process each designer checks how their sub-system
fulfils the local requirements before the release of their components and subsequent assembly and
analysis at global level. The designers also make relative comparisons, i.e. analyse different
solutions and compare their performance. If the analysis at sub-system level indicates that the
prescribed local requirements can not be fulfilled this is brought up and rebalancing of requirements
is done.

3. Simulation support in PD
At an early stage in product development, feasibility studies are made. Their purpose is to see if the
chosen strategy to solve the design problem and achieve the global requirements is possible. In the
automotive industry old, altered, computer models as well as old, altered, real cars are used in order
to make these feasibility studies. The reason for choosing an earlier model is that making a physical
prototype from scratch would be very costly and imply a high risk at this early stage when little is
known, and it would also take a much longer time than altering an existing vehicle. Traditionally
the detail design of a car body from this stage on is made with CAD support; design
engineers/drafters' meticulously design each component to a detail level showing the end shape
within tenths of a millimetre. When all parts are designed they are assembled, the geometry is
"cleaned" and subsequent meshing and analysis is done. The results from these analyses at global
level are finally transmitted back to the design engineers/drafters.

After the feasibility study, basic solution paths to global requirements already exist in order to
benefit most in later phases from the early studies. Requirements should be broken down to
adequate levels and support tools provided to the design engineers. This is however not done on a
regular basis. To draw the most from the early feasibility studies and concept studies, a
requirement's breakdown strategy with a corresponding simulation strategy that will seamlessly
carry and refine the information is suggested [Bylund and Eriksson 02]. The global mechanical
4
requirements on a car body such as weight, stiffness and crashworthiness have to be broken down
into entities that match the sub-systems that the individual design engineer/ drafter designs: Beams
and joints comprise the sub-systems for the breakdown process, see [Bylund, Fredricson and
Thompson 02].

By using simulation, expensive and time consuming testing can be reduced. Building a car body
prototype implies considerable cost, although special stamping tools exist for prototype stampings
in small series. Computer aided simulation based on the finite element method (FEM) has been used
to verify the performances of car bodies for quite some time now. By tradition, the simulation takes
place in especially dedicated departments situated distantly from the design department; the
distance reduces daily contact and information exchange [Larsson et al 03, Kiesler and Cummings
02]. A FE-simulation department takes care basically of three steps: Pre-processing i.e.
transforming the CAD model to a computational model that can be sent to the analysis software;
this procedure consists of cleaning the CAD-model of excess details, applying a FE-mesh to it and
applying boundary conditions. Sending the computational model to an analysis server and waiting
and checking the quality of the result. Post-process i.e. visualise the result as tables, coloured stress
or strain images, or animations using post-processing softwares to treat the result files imported
from the analysis software. The most time consuming stages are the cleaning and meshing of the
CAD geometry, and time spent on assembling the hundreds of components designed by the
numerous design engineers/drafters. Roughly, four to six weeks are spent from the time a design
engineer/drafter releases the component drawing to the time the analysis is ready and post-
processed. During this time the drawing work continues. Thus, when the results are available they
may not reflect the current state of the design of the component. By inspecting the types of
simulation that are often made as well as examining how CAD is being used, some possible routine
simulations can be identified. In car body design beam sections are used as definitions of how the
basic car body should look, and they govern to great extent the stiffness and crash performances of
the complete car, see Figure 4. Furthermore, the main joints are defined early on with respect to the
factory assembly sequence; they are of great importance to the overall stiffness of the car body, see
Figure 4. Thus it would be of great value to develop software that is adapted to these two sub-
systems, beams and joints [Bylund and Eriksson 02]. Simulations of these sub-systems can be made
accessible to users with less experience of analysis, especially design engineers/ drafters, by
automatising the laborious file transfer between analysis related software. In addition, by
standardising the boundary conditions the analysis can be made safer, faster and more repeatable.

3.1 Traceability and late changes

Strength of requirement breakdown to sub-systems/organs is that traceability between global and
local requirements is created. This works both top - down and bottom - up. In the top – down case,
changes in global requirements due to changes in the market can be broken down to local level and
necessary design changes can be quickly anticipated by the designer using adapted analysis tools.
In the bottom - up case, if for some reason a local requirement is shown not to be possible to fulfil
by the design engineer using software developed for fast analysis of joint and beams, the effect of
this can be traced back to the global requirements.

5
 The upper A- The upper D-
joint joint

The B-pillar

Figure 4. Car body in white, (BIW) without floor pan and roof, for clarity, and with indication
of the areas mentioned in the pilot study.

3.2 Development of simulation-support tools

In a company developing advanced mechanical structures, such as an automotive company, several
commercial analysis-related software exists, including pre-processors, analysis codes and post-
processors, each adapted to specific simulation needs. As described earlier, there are analysis
departments dedicated to convert the CAD models into analysis models and run them appropriately.
For some tasks great skill and deep knowledge of how to adapt the software to the actual simulation
situation is needed, in other cases the boundary conditions can be standardised and then it is merely
a question of transferring files between different software in an appropriate way. The possibility to
automate tasks depends on the design knowledge: the more design knowledge the more automation
is possible, see [McMahon and Draper 02] for a discussion of this topic.
Using analysis-related software already in use at the company as building blocks for in-house, easy-
to-use simulation software, minimises license costs and furthermore increases trust in the results.
Another benefit is the possibility to override the automatic steps for the user familiar with the
ingoing software. Connecting commercial softwares can be done using API (Application
Programming Interface) scripts that start the programs, run them in batch mode, and organise
transfer of input and output files. For this to be possible the ingoing software must have an API and
files must be compatible or at least possible to convert so the next program can read them.
Fortunately many software have API's, and the files are in ASCII text so conversion is pretty
straightforward, e.g. IGES.

3.3 Software developed at VCC: ADRIAN and DAMIDA

To make the use of the FE analysis more accessible to the design engineer, the pre-processing,
analysis and post-processing of the results in the case of analysing beams and joints has been
standardised with DAMIDA and ADRIAN software, see Figure 5 and Figures 6-9 in Appendix.
Both softwares use a master program to couple commercial software for automating the workflow
used at the analysis department.
The softwares have been developed with respect to different stakeholders. With the design engineer
foremost, the main users have been involved regularly during the development, using principles of
participatory design [Blomberg et al 93]. Design engineers that had shown an interest in developing
their skills were chosen to participate in the project. Meetings have been held and demo versions
have been used to trigger suggestions from the users. The future software maintenance department
has also been involved to ensure that future maintenance can be made easier. The program structure
can affect the long-term cost with more than 50% [Horowitz 93]. Carefully selected teams of
masters' students have done the programming as a part of their masters' theses. This has made it

6
possible to produce the software at very reasonable cost. Listed below are all the teams,
departments and disciplines involved in some way in the development of the software.

1. The development teams (MSc students, Bylund the industrial supervisor, and their academic
supervisor.)
2. VCC intellectual property department.
3. VCC Advanced Engineering.
4. VCC Internal IT-department.
5. Volvo IT, ltd.
6. VCC design engineers.
7. VCC concept engineers.
8. VCC analysis engineers.

3.3.1 DAMIDA for beam analysis

A car body consists of numerous sheet metal parts that are welded together to form an integrated
structure consisting of beams, joints and panels, see figure 4. Schematically the beams can be
divided into two categories, the ones that should deform in a controlled manner in the case of a
crash situation, and the ones that should act as backup and hence not deform. The beams that should
deform in a controlled manner are the ones in the crash zones of the car, i.e. front, rear and side.
The ability of the crash zones to deform in a controllable way decides how severe the deceleration
and how deformed the car body will be. To obtain a safe car body, the deceleration should not be
too high, and neither should the intrusion be too great in the passenger area. These are global
requirements: deceleration curves and intrusions at different locations that can be broken down to
requirements at beam level. There are three principal types of load that a car body beam needs to
withstand. Torsional load, axial load and bending load. Because closed sections are used the
torsional load is seldom an issue in the case of car body design. The axial load on the other hand
can be very high in the event of a crash and the deformation can be of three kinds: elastic, global
buckling, or Euler buckling, and local buckling.
Elastic axial deformation occurs if the beam section is big enough, the metal is thick enough (or
homogenous), and the beam is short; the backup structure should ideally have this behaviour. Euler
buckling occurs when the beam is slender, i.e. the length of the beam is considerable in comparison
to the size of the section. The strength in a structure after Euler buckling has started is only a
fraction of the initial strength, leading most often to large deformations far from where the load is
applied. This is not desirable behaviour and if it occurs in the backup structure the result may be
fatal. If it occurs in the crash zone the force level in the beam with respect to deformation will be
low, hence the energy absorption of the crash structure will be low leading to a bad deceleration
profile. Because the beams in modern car bodies are seldom very slender, Euler buckling is not
likely to occur but its severity makes it a danger that has to be checked. The third form of
deformation, local buckling, occurs when the beam is not slender and when the wall thickness of the
beam is thin compared with the overall size of the beam section. Local axial buckling is the axial
deformation type that has the potential to absorb most energy. If designed in an appropriate manner,
deformation will start where the load is applied and the resisting force will quickly build up and
stay at a relatively stable level as the beam is being successively compressed. For the two first type
of deformation, elastic axial and global buckling formulas exist in various manuals. Local axial
buckling on the other hand is not as straightforward to predict; there are formulas [Rhodes 91] but
they are difficult to apply because the shape of the section and the strength of the metal greatly
influence the crash behaviour. These formulas are the outcome of a mix of experiments and analysis
and their validity and generalisation is restricted. For bending loads the picture is somewhat similar,
there is a domain where there is a linear relation between stiffness of the section, the load and the
deformation. There are explicit formulas for calculating the stiffness of many different section
shapes, for complex shapes the analysis is cumbersome with hand calculation. With increased loads

7
the beam will deform plastically where the bending moment has its maximum, as in the axial case
there exists formulas [Rhodes 91] but these are limited to a restricted number of sectional shapes
and are difficult to use.

To be able to analyse any type of beam in both local axial buckling and plastic bending, the FE
method is more general. To analyse local buckling and plastic bending, explicit non-linear FE
analysis is used [Bathe 96 a]. As described earlier, the analysis department uses the FE method
extensively to analyse the behaviour of the complete car body. To conveniently analysing beams
DAMIDA has been developed. As described in section 3 the FE-method implies the use of several
softwares to pre-process, analyse and post-process the results. DAMIDA has been developed to
automate cumbersome file transfers, providing default values for analysis parameters such as
element size and time step, and providing guidance for boundary conditions, see sections 3.2 and
3.3. DAMIDA furthermore provides the results in a standardised format, see Figures 6 and 7, in
Appendix.

3.2.2. ADRIAN for joint analysis

The different beams in a car body are connected in joints, see Figure 4. The joints in a steel car
body are not isolated components but are made up from the same sheet metal components as the
beams. The stiffness of the joints affects the stiffness of the complete car body to a large extent
[Lubkin 74]. The global car body stiffness impacts the NVH (Noise, Vibration and Harshness) of
the car. Low stiffness means that relative movements between the car body and the interior panels
and even doors and hatches become important, as these movements cause noise and make the car
feel less solid. In an estate wagon this phenomenon can be especially noticeable. Hence, the large,
boxy luggage compartment of such a car needs great attention to become stiff enough not to cause
noise exceeding that in a sedan car. The global requirements for designing a stiff car body are
global torsional- and global bending-stiffness and their distribution through the length of the car.
The global stiffness can be broken down to stiffness requirements for the individual joints. The
principal loads that a joint in a car body is submitted to are torsional and bending in different
directions. The deformations that are caused by the bending and torsion of a car body are small and
within the elastic domain. Due to the complex shape of a car joint there are no formulas to calculate
the stiffness. With the FE method the stiffness for a joint of complex shape can be analysed using
static condensation [Bathe 96 b]: not even explicit boundary conditions have to be defined. A
complementing dynamic method to visualise the stiffness properties of the joint has been
developed, which is described in [Bylund 03] see also [Whyatt Becker et al 99]. The software
ADRIAN includes both static condensation and the dynamic joint method and permits the
presentation of the results in a standard format, see Figures 8 and 9, in Appendix.

8
 Flow scheme of Easy-to-use analysis tools

DAMIDA for BEAMS ADRIAN for JOINTS

M
Detailed shell model in CATIA A Detailed shell model in CATIA
N
U
A
Designer put section into DAMIDA L Designer put joint into ADRIAN

A
U
Meshing and extrusion in DAMIDA T Meshing in ANSA
O
M
Analysis in RADIOSS Analysis in NASTRAN
A
T
I
C

Results on HTML-page and Results on HTML-page and

in ANIMATOR in ANIMATOR

Compare results with local target values from PBM

or check relative change in performances alternative designs.

Figure 5. Schematic flow scheme for DAMIDA and ADRIAN

4.1. Getting fast ROI using small projects

Analysis software for design engineers/ drafters with a considerable impact can be produced at
limited cost and within a short time. In fact many of the successful software projects of this kind are
those that have been done in less than six months [Chapman and Pinfold 01]. The development of
the two softwares has been done on a tight budget; this has made the return on investment fast and
thus the support for implementation easier [Bylund, Grante and Lopez-Mesa 03].

9
4.2. Maintenance and ownership of simulation-support tools
Developing simulation-support tools adapted to a company's specific needs implies that there is a
coupling between the tool and the know-how of the company. The software will thus be of strategic
importance to the company, therefore questions regarding intellectual property have to be made
clear from the beginning. Although software development may not be the core business of the
company, the strategic nature of a software may imply that it should be made and kept in-house.
Sharing the software with other companies can lead to an unfavourable sharing of business
advantages.

4.2.1 Maintenance
Software needs maintenance, and software made by using scripts to interconnect commercial
software already in use at a company is sensitive to version changes. This means, for example, that
if the software that meshes (ANSA at VCC) changes version, the scripts within the developed
software may have to be changed in order for the program to continue working. It has to be clear
who is responsible for making these changes so that the software is always working. During the
development phase it has to be assured that parts in the software that will probably need changes
will be easily accessible. Such issues should be discussed with the maintenance department. In
short, the software has to be serviceable as with any product.

5. Implementation of simulation support

Implementation is done stepwise, for a number of reasons. The software needs to be tested on a
restricted number of users so that possible bugs are found early on. If a broad introduction is made
immediately and numerous bugs come up, the software may get a bad reputation that is hard to be
get rid of even when the program is fixed. Furthermore, the impact on the product development
process must be checked. Eventual risks associated with the introduction are less if introduction is
made in steps. Also, as in this case, when a limited number of licences and limited computational
resources are used, as in the analysis department, implementation must be made carefully not to
disturb the overall analysis capabilities of the company. Finally, the teaching resources can be a
limiting factor. Respecting the need for stepwise implementation, a pilot study has been made, and
further implementation has been done and is currently ongoing.

5.1 The pilot study

ADRIAN and DAMIDA are mainly intended for use by design engineers with limited analysis
experience. The pilot study was introduced by involving some of the design engineers that had
previously been participating during the development of the software. Hand-outs with instructions
were also provided. Support was handled informally by the development team and others
knowledgeable with the ongoing software.

5.1.1 Examples from the pilot study of the use of ADRIAN and DAMIDA in the
PD process
The first three examples come from the pilot study, in which five design engineers at the car body
department were informed and trained in using the tools. All five work with component design,
using the CATIA V4 CAD tool. Their expertise is in designing components to fulfil the complex set
of various balanced non-mechanical requirements such as manufacturability (stamping, factory
sequence, and weld accessibility), corrosion protection, liquid escape and interaction with interior
panels. This has to be done at an acceptable cost, as well as fulfilling numerous mechanical
requirements such as stiffness, crash performance and life length. The fulfilment of the above
mentioned non-mechanical requirements is judged in collaboration with experts of the various areas
in so-called area meetings. To check the fulfilment of the mechanical requirements the analysis
department has to be involved, see figure 2. The idea of the pilot study is to provide the design

10
engineers with the easy-to-use simulation tools presented, and see if a design process of the type in
figure 3 can be established, so that preliminary analysis can be done directly by the design
engineers.

The first example was designed as a test of the ADRIAN software at the outset of the pilot study. A
relative analysis of the difference between using a cover plate where the roof header beam goes into
the upper a-joint, see figure 4, or leaving the section open was chosen to test whether ADRIAN was
user-friendly and worked in a correct manner. The first design alternative, cutting the joint out of
the BIW model, modelling the weld spots and assigning the sheet thickness, took one and a half
hours. To create the design alternative by taking away the cover plate and changing some spot
welds took another half an hour. On both occasions the analysis in ADRIAN took less than ten
minutes. So the total time for modelling and analysing the two alternatives was less than three
hours. It should be pointed out that this was a real case without any simplifications: taken directly
from one of the Body in White (BIW) in development at VCC. The results showed that the two
design alternatives for the upper a-joint had a stiffness difference of 50 % in the most important
direction of interest. The design engineer said that the HTML page was a great means of
communicating the analysis results within the company; see Figure 9, in Appendix. In addition, the
frequency animations were a great means of seeing weak areas of the design, e.g. where welds
could be added to increase stiffness, see Figure 8, in Appendix. A considerable amount of time was
saved in comparison to waiting for the results from complete body analysis.

Another example comes from the design of an upper d-joint, figure 4. This example was revealed
when interviewing one of the participants in the pilot study. The design engineer used ADRIAN as
a means to analyse the stiffness of the d-joint. The design, that due to secrecy reasons can not be
revealed, did not provide enough stiffness. One of the legs of the joint was twisting due to low
torsional stiffness, meaning that the overall torsional stiffness of the rear of the car body would also
be at stake. By using the functionality to animate the eigenmodes in ADRIAN the design engineer
could see how excessive movement took place around one of the spot welds, leading to both risk of
fatigue and low stiffness. With this in mind the design engineer added one more spot weld in a
strategic location and the torsional stiffness of the joint improved by 50%. Excessive movement
around one single spot weld was also avoided. Later testing indeed showed that the first design was
leading to fatigue cracks around the overloaded spot weld. The cost of slightly redesigning part of
the joint to be able to put in one more spot weld is much less costly than the alternatives, which
would have been either adding glue, laser welding, or adding a reinforcement panel. Adding a
reinforcement panel would also increase weight, especially, which a spot weld does not. The design
engineer said that the fact of having analysis results in an easy-to-understand format, see Figure 7,
in Appendix, was a major advantage when explaining his alternative solution to stakeholders from
stamping and assembly. The generation of the alternative design took a few days and the
preparation for analysis and subsequent analysis, as in the first example, took only a few hours.

A third example regarding DAMIDA comes from interviewing one of the most frequent users. The
design engineer works with the b-pillar as a sub-system: this is the pillar between the front door and
the rear door in a four-door car, see Figure 4. The b-pillar is of outmost importance when it comes
to side impact crashworthiness. The section size and shape of the pillar varies with the height so that
its deformation in case of a crash follows a pre-determined pattern, minimising injuries to the
passengers. In the traditional development process, see figure 2, all the crash analysis is made at the
analysis department. In the development process in Figure 3, the design engineer himself makes
preliminary analysis of his design then "lending" the design to the analysis department for an
analysis at complete vehicle level. In this example, the design engineer in fact received requests
from the analysis department to make analyses of b-pillar sections. The analysis department is often
over-loaded with work and in this way they could transfer some of their work load back to the
design department. This remarkable situation proves that great trust is put in the DAMIDA

11
software, even by experienced analysts from the analysis department. Furthermore, this has created
improved contact and understanding between the design department and the analysis department.
Some design engineers have also become increasingly interested in learning more about analysis in
general.

5.1.2. Conclusions from the pilot study

Without these easy-to-use simulation tools and the breaking down process to sub-system level, the
impact of the cover plate and the relocation of the spot weld would not have been detected until
analysis of the complete car body a lot of time would have been wasted. The time waiting for the
analysis results could be used to refine the design instead, and to move forward to the design of
other components. Analysing the sub-systems with general-purpose FEM programs instead of easy-
to-use tools like ADRIAN and DAMIDA is possible but time-consuming. As seen in the third
example, the analysis department voluntarily sent an analysis request to the design department.
Thus it can be seen that by providing the design engineers with simulation software adapted to their
needs for preliminary analysis, the analysis department can be relieved of some of the analysis.
Hence, because the design has already undergone preliminary analysis during the design stage, the
analysis department can use their resources and expertise for more complex analysis at complete
vehicle level instead.

5.2. Implementation on a larger scale

With the positive results from the pilot study, management supported a larger-scale implementation
at the car body department. A side effect not realized at the outset of the development of these
softwares, was the empowerment experienced by the drafters/designers. Designers can not only
make a preliminary check that their design is fulfilling the requirements received from the
breakdown process, but by making analyses themselves they can also iteratively improve the design
and show that their ideas improve the part/s they are working on. The empowerment effect makes
for greater interest among the design engineers in learning to use the software. Design engineers
interested in learning the software were found and contacted through their department managers.
The number of interested design engineers lead to several courses being run.

5.2.1. Course Development

The goal of the course is to provide the users with basic knowledge in preparing the model for the
program and basics in analysis with the FE-method.
The basic knowledge in model building consists of how the CAD model should be made to fit the
software, e.g. how spot welds and material should be defined, also what should be avoided such as
extremely small surfaces, extremely short lines and sharp angles. The softwares are made so that the
application of boundary conditions are restricted to a number of standard cases and default values
exist for mesh size and other parameters. Basics FE principles are taught as what happens if the
default values in the software, such as element size are changed. The courses are organized by one
of the authors and held together with the design engineers from the pilot study. The idea is that the
design engineers from the pilot study are experienced with the CAD environment and are most
suitable in explaining how to adapt the CAD models to fit the analysis software.

5.2.2. User support

One of the authors provides user support and acts as a contact between design department and the
software maintenance department. Personal interest by one design engineer from the pilot study
together with management support has made him responsible for user support concerning the use of
the software.

12
5.3. Result from the introduction of simulation software for the designer
on a larger scale in the alternative development process
In the method part of this paper, the criteria for success of the implementation of the working
method in figure 3 were shorter lead time and better solutions. These criteria have been broken
down according to figure 1. The broken down criteria are easier to measure than the final success
criteria and are linked logically to the success criteria and are marked with italics below. The
criterion at the bottom, the number of design engineers that can use the software, is easy to
measure; this number is currently increasing with each course held. The next criterion in the chain:
faster analysis is as follows. Without using the breakdown strategy, see figure 2, approximate time
between releases of component design until analysis results are received is about four to six weeks.
If a design engineer insists on going to the analysis department to get his sub-system analyzed, it
would take at least one week before he gets the results. With the tools presented in this paper, non-
linear explicit FE analysis of the plastic performance of a beam section can be done in a few hours
with DAMIDA. Cutting the section out from the vehicle model takes about fifteen minutes.
Analyzing the stiffness of a joint in all directions and a visualization of the stiffness can be made
with linear FE analysis in 10-15 minutes with ADRIAN. Preparing the CAD model of the joint for
the analysis in ADRIAN takes approximately 1-2 hours. Furthermore, the results are presented in
standard format pages; see Figure 7 and 9, in Appendix. The analyses are standardized, e.g. time
step and mesh size, as well as the boundary conditions. The possibility of faster analysis has led to
more design alternatives being tested for both beams and joints than before. Numerous alternative
beam sections can be launched and run overnight, different alternative joint designs can be analyzed
within a day. Without the software, the time delay between design and analysis makes it difficult to
actively test various solutions. A better exploration of the solution space leads to better designs. By
using the software in Product Development at VCC, weight has been reduced with maintained or
increased stiffness and strength. The enhanced iteration has led to more material combinations in
beam sections having been tested. Joints have been made stiffer without having to add more
material. As seen in figure 1, the new development process with the described simulation software
is supposed to lead to shorter lead time for car body development. Indications of this have been
seen and are expected to increase as courses in the software are regularly held and the new
development process comes into place. Considerable time could be saved if one analysis large loop
could be eliminated due to better design in each small loop, see figure 3. An important qualitative
success criterion is the perceived empowerment of the users; it creates a will for learning more and
increases the successful use of the software.

6. Conclusion
Driving product development with simulation instead of verifying designs a posteori can be made if
the actual design engineer can perform more simulation himself without having to wait for the
analysis department. A breakdown process adapted to the product that provides the design engineer
with requirements corresponding to his design area increases product quality and has the potential
to reduce lead time. This is achieved by providing software adapted to the breakdown process and
the skills and needs of the design engineers. To shorten lead time the majority of the design
engineers have to be taught and to use the software, and the breakdown process must be in place.
These findings correlate with the findings in [King, Jones and Simner 03], that introducing basic
CAE analysis has great impact on quality and lead time when tightly integrated in the development
process.

In this paper, it has been shown that by adapting analysis software to be used by the design
engineer, the time between design and a standardized analysis result is reduced to hours instead of
four to six weeks, as is normal in the traditional development process. The reduced time from
design to analysis increases the number of iterations at sub-system level, leading to better solutions.

13
The lead time is likely to be affected considerably when more courses have been held and the
majority of the design engineers use the software.

The tight coupling between the development process, the product and the software make the
software a strategic asset. The ownership must be clear, and in-house development can be the best
alternative, though software development may not be a main activity of the company at all. The
development of the two softwares in this paper has been done on a restricted budget within a
stringent time frame; this has made the return on investment fast. By actively identifying
stakeholders, from users to software maintenance engineers, and involving them in the development
of software it becomes easy to use and at the same time professionally made and maintainable.

7. Future development
The development of softwares for engineering design could aim at providing the possibility for:
"Simulation Driven Design by Designers", the motto of the Polhem Laboratory, Luleå University of
Technology. There are issues to solve, regarding both the simulation software and the role of the
engineering designer.

7.1. Integration of process simulation in the design process

In order to stay competitive in the market, the design of products has to be increasingly optimized.
For example, having lower weight while yet being stronger and less expensive. Simulation has to be
used more extensively and proactively, and routines automated and made more efficient [Anderson
94]. When more elaborate materials are used together with more complex manufacturing processes,
the influence of the manufacturing process on the behaviour of the product increases in importance.
Simulation of manufacturing processes and its coupling to final behaviour of the product is an
active field of research [Hilding 02] and is still not used extensively in industry. Nevertheless,
augmenting the use of coupled simulation in industry is necessary to be competitive: hence the
possibility to make the bulk of simulations of this type should be given to the design engineer
working closely with the design [Sandberg, Kalhori and Larsson 04]. The analysis experts should
work with extending the range of the simulation and create possibilities for the design engineer to
make standard simulations. Complex cases of analysis will still be dealt with by analysis experts.

7.2. Changing the role of the design engineer

To generate the complex shapes of the stamped components that make the car body, skills in
handling CAD software are needed. Various and sometimes contradictory requirements have to be
respected when designing the components [Bylund 04]. The fulfilment of some of the requirements
is made by inspection and by discussions with experts, e.g. sealing possibility, corrosion and
corrosion protection issues. Others are simulated, such as collision detection, stamping possibility,
stiffness and crashworthiness. Getting the development process more simulation-driven means that
more of the simulation has to be done by the actual design engineer. This is possible by making the
simulation tools more user-friendly and adapted to the problems at hand, as shown in this paper.
Also the view of the design engineer/drafter as a person that only generates geometry has to be
changed. The work has to be recognized as the process of going from requirements to a
computationally verified solution. The demands on the designer/drafter will change, the ability to
design complex shapes will remain and the ability to run integrated analysis programs or even to
alter them to better fit the situation will increase [Bylund et al 04].

14
Acknowledgements

The participation of the design engineers Hossein Rezaei, Issa Rezaei, Håkan Runius and others
from the car body development department at VCC througout the development of the analysis
softwares and their implementation is acknowledged. We acknowledge the programming efforts
performed by Mattias Shamlo, Henrik Sandström, Maria Andersson and Fredric Bohman as well as
input from Volvo It and the in-house It department at VCC regarding program structure.

References

Anderson, J, D, A., "The Role of knowledge-based engineering systems in concurrent engineering",

Concurrent Engineering: Concepts-Implementation and Practice, Chapman and Hall 1994.

Bathe, Klaus-Jürgen, a, "Finite Element Procedures", Prentice Hall, New Jersey ,1996. pp 824-825.

Bathe, Klaus-Jürgen, b, "Finite Element Procedures", Prentice Hall, New Jersey ,1996. pp 817-818.

Blessing, L., “What is this thing called Design Research?”, Proceedings of 2002 Int’l CIRP Design
Seminar, Hong Kong, 16-18 May 2002, pp. 1-6.

Blomberg J., Giacomi J., Mosher A. and Swenton-Wall P., “Ethnographic Field Methods and Their
Relation to Design ” in Shuler, D. and Namioka, A., “Participatory Design, Principles and
Practice”, L. Erlbaum Associates, Hillsdale, 1993, pp. 123-155.

Bylund, N. and Eriksson, M. Simulation Driven Car Body Development Using Property Based
Models SAE paper 2001-01-3046, in the proceedings of the International Body Engineering
Conference, IBEC 2001, 8-12 of July 2002, Paris, France.

Bylund, N., Fredricson, H. and Thompson, G.

A design process for complex mechanical structures using Property Based
Models, with application to car bodies. In the proceedings of Design 2002 Conference, 14-17 of
May 2002, Dubrovnik, Croatia.

Bylund, N., Grante, C. & López-Mesa.,"Usability in industry of methods from design research", In
the proceedings of the International Conference of Engineering Design, ICED03, 19-21 of August,
2003, Stockholm, Sweden.

Bylund, N. "Fast and Economic Stiffness Evaluation of Automotive Joints"

SAE paper20037025, in the proceedings of the International Body Engineering Conference, IBEC
2003, 27-29 of October 2003, Shiba, Japan.

Bylund, N., Isaksson, O., Kalhori, V., and Larsson, T., " Enhanced Engineering Design Practice
Using Knowledge Enabled Engineering (KEE) With Simulation Methods"
In the proceedings of Design 2002 Conference, 18-21 of May 2004, Dubrovnik, Croatia.

Bylund, N. ADRIAN a program for evaluating the stiffness of joints and its application in the
product development process. Submitted to Journal of Computing And Information Science In
Engineering JCICE. 2004

15
Chapman, C.B and Pinfold, M., "The application of a knowledge based engineering approach to
rapid design and analysis of an automotive structure." Advances in Engineering Software 32 (2001)
pp. 903-912, Elsivier, 2001.

Eckert, C., Clarkson, P, J. and Stacey, M.," The Lure of the Measurable in Design Research", In the
proceedings of Design 2004 Conference, 18-21 of May 2004, Dubrovnik, Croatia.

Hilding, D., "Influence of forming parameters on crash properties", Nordic LS-DYNA Users’
Conference 2002, September 16 & 17, 2002, Gothenburg, Sweden

Hinds, K., and Kiesler, S., Eds. "Distributed Work", The MIT Press, Cambridge, MA, 2002, pp57-
80, pp167-186.

Horowitz, B, M., "Strategic Buying for the Future", Libey Publishing, 1993.

Larsson, A., Törlind, P., Karlsson, L., Mabogunje, A., Leifer, L., Larsson, T., and Elfström, B-0.,
"Distributed Team Innovation- A Framework for Distributed Product Development", In procedings
of ICED03, 19-21 of May 2003, Stockholm, Sweden.

Lubkin JL. "The Flexibility of a Tubular Welded Joint in a Vehicle Frame." 1974 SAE
740340:1518-1522.

McMahon, C.A., and Draper, C., "Patterns of Design and Development in Adaptive Design: How
do We Match Design Methods to Design Circumstance?", Third International Seminar and
Workshop, EDIPROD02, Zielóna-Lagow, Poland, 2002.

Rhodes, J., "Structural Design for Crash", Lectures on Ultimate Strength of Thin-Walled members,
Carl-Cranz Gesselshaft, course material, 1991.

Sandberg, S., Kalhori, V., and Larsson, T., "FEM Simulation Supported KEE in High Strenght Steel
Car Body Development" submitted to IMECE04, 2004 ASME, International Mechanical
Engineerig Congress, 13-20 of November 2004, Anaheim, California USA.

Whomack, J.P, Jones, D.T, and Roos, D., "The Machine that Changed the World.", Macmillan
Publishing Company, Canada, 1990.

Whyatt Becker, P.J., Wynn, R. H., Berger, E.J., and Blough, J.R, "Using Rigid-Body Dynamics to
Measure Joint stiffness.", Mechanical systems and Signal Processing (1999) 13(5), pp789-801.

Yinn R., “Case Study Research. Design and Methods”, Sage Publications, Thousand Oaks, 1994.

16
APPENDIX
The appendix contains four pages with screen captures from DAMIDA and ADRIAN, showing the analysis processes
in more detail as well as explaining the result pages. The geometries are not simplified but taken directly from the
product development process at VCC.

17
Complete car Body in Sub-system in CATIA: Sub-system in CATIA: Pillar
CATIA B-pillar section

Start page Step 1: Import section Step 2: Launch 2D meshing

Step 3: Choose material Step 4: Choose load Step 5: Launch 3D mesh and
launch analysis

HTML result page with Figure 6.

Visualisation of plastic
elastic and plastic properties. Detailed flow scheme of DAMIDA
bending. run.
18
The analysed Information about material Information about Internal energy and
section. grades and welds. analysis parameters. hourglass energy.

The bending moment in Sectional

relation to bending angle. constants.

Figure 7. Result page from DAMIDA.

19
 Complete car body in Catia Sub-system in CATIA: ADRIAN: Start of ADRIAN
Joint cut out, sheet gauge which manages pre-
and welds defined processing, analysis and post
processing

ANSA is started by ADRIAN ADRIAN automatically ADRIAN launches eigenmode

and auto-meshes sheet by assembles the meshed sheets and analysis and static condensation
sheet. prepares the model for analysis in in NASTRAN.
NASTRAN.

Eigenmodes showed in HTML result page Figure 8. Detailed flowsheme of

ANIMATOR showing static stiffnesses ADRIAN run

20
 The analysed Information about sheets and
joint. welds.

Stiffness related to each leg of the joint. Red shows the direction of max
principal stiffness, green the direction of minimum stiffness.

Figure 9. Result page from ADRIAN.

21