Escolar Documentos
Profissional Documentos
Cultura Documentos
Schnell
Applied Structural
Mechanics
Fundam entals o f Elasticity, Load-Bearing Structures,
Structural O ptim ization
Including Exercises
TA
t
S ? 1 3
/ ?<77
P r e fa c e
H ans E schenauer
Siegen/ GERM ANY
A pril 1996
Niels Olhoff
A alborg/D E N M A R K
W alter Schnell
D arm stadt / G ERM ANY
C o n te n ts
L ist o f s y m b o ls
xra
I n t r o d u c t io n
F u n d a m e n ta ls o f e la s t ic it y
C h a p te r 2 to 7
A .l D e f in it io n s F o r m u la s C o n c e p ts
2.1
2.2
2.3
2.4
2.5
3
3.1
3.2
3.3
3.4
3.5
4
4.1
4.2
4.3
4.4
4.5
5
5.1
5.2
5.3
5.4
T e n s o r a lg e b r a a n d a n a l y s is
Term inology - definitions
Index rules an d sum m ation convention
T ensor of first order ( v e c to r)
Tensors of second a n d higher order
C urvilinear coordinates
S t a te o f s t r e s s
Stress vector
Stress tensor
C oordinate transfo rm atio n - principal axes
Stress deviator
E quilibrium conditions
S t a te o f s t r a in
K inem atics of a deform able body
S train ten so r
S tra in -d isp la c e m e n t relations
T ransform ation of principal axes
C om patibility conditions
C o n s t it u t iv e la w s o f l in e a r ly e la s t ic b o d ie s
B asic concepts
G eneralized H O O K E -D U H A M E L 's law
M aterial law for p lane states
M aterial law for a unidirectional layer ( U D -la y e r) of a
fibre reinforced com posite
5
6
7
10
13
18
18
20
21
24
25
26
26
29
30
31
31
31
31
32
35
37
VIII
6
6.1
6.2
6.3
6.4
6.5
6.6
6.7
7
7.1
7.2
7.3
7.4
7.5
Contents
E n e r g y p r in c ip le s
Basic term inology an d assum ptions
Energy expressions
P rinciple of v irtu al displacem ents ( P v d )
P rinciple of v irtu al forces ( P v f )
R eciprocity theorem s and U n it-L o a d -M ethod
T reatm en t of a variatio n al problem
A pproxim ation m ethods for continua
39
39
40
44
44
46
46
47
P r o b le m f o r m u la t io n s in t h e t h e o r y o f lin e a r
e la s tic ity
48
48
49
49
50
51
A .2 E x e r c is e s
53
A-2-1
A-2-2
A-2-3
53
60
63
A-3-1:
A-3-2:
A-3-3:
A-4-1
A-4-2
A-4-3
A-4-4:
A-5-1:
A-6 -1 :
A-6 - 2 :
A-6-3:
A-7-1:
A-7-2:
66
67
70
71
73
74
76
78
80
82
83
86
89
Contents
P la n e lo a d - b e a r in g str u c tu r e s
IX
93
C h a p te r 8 to 10
B .l D e f in it io n s F o r m u la s C o n c e p ts
93
93
D is k s
D efinitions - A ssum ptions - Basic Equations
A naly tical solutions to th e homogeneous bipotential
equation
8.1
8.2
93
95
9
9.1
9.2
P la te s
D efinitions - A ssum ptions - Basic E quations
A nalytical solutions for s h e a r-rig id plates
99
99
107
10
C o u p le d d is k - p la te p ro b le m s
Isotropic plane structures w ith large displacem ents
L o ad -b earin g structures m ad e of com posite m aterials
113
113
118
10.1
1 0.2
B .2 E x e r c is e s
123
B-8-1:
B- 8 -2 :
123
128
131
133
137
139
145
151
153
155
157
167
170
172
177
179
Contents
183
199
B-9-9:
C u r v e d lo a d - b e a r in g s tr u c tu r e s
188
195
C h a p te r 11 to 14 C .l D e f in it io n s F o r m u la s C o n c e p ts
199
11
199
199
209
213
G e n e r a l f u n d a m e n t a ls o f s h e ll s
11.1
11.2
11.3
12
M e m b r a n e t h e o r y o f s h e ll s
12.1
12.2
12.3
12.4
12.5
12.6
13
B e n d in g t h e o r y o f s h e ll s o f r e v o lu t io n
13.1
13.2
13.3
14
T h e o r y o f s h a llo w s h e ll s
14.1
14.2
14.3
214
214
215
218
220
221
221
222
222
228
233
241
241
242
245
C.2 E x e r c is e s
247
C -ll-l:
247
251
253
Contents
XI
255
258
260
288
290
293
296
D S tr u c tu r a l o p tim iz a tio n
301
264
267
272
276
281
283
C hapter 15 to 18 D .l D e f in it io n s
15
15.1
15.2
15.3
15.4
15.5
15.6
16
16.1
16.2
17
17.1
17.2
17.3
F o r m u la s C o n c e p ts
F u n d a m e n t a ls o f s t r u c t u r a l o p t im iz a t io n
M otivation - aim - developm ent
S e n s i t i v i t y a n a l y s is o f ^ t r u c t u r e s
P urpose of sensitivity analysis
O verall F in ite D ifference ( O F D ) sensitivity analysis
A n alytical an d s e m i-a n a ly tic a l sensitivity analyses
301
301
301
302
303
306
307
309
310
310
314
321
321
322
322
XII
18
18.1
18.2
18.3
Contents
O p t im iz a t io n s t r a t e g ie s
Vector, m ultiobjective or m u lticriteria optim ization P A R E T O -o p tim ality
Shape optim ization
A ugm ented optim ization loop by additional strategies
325
325
329
334
D .2 E x e r c is e s
337
D-15/16-1:
337
342
347
352
355
360
364
370
R efe r e n c e s
375
A
B
C
D
375
376
377
378
S u b jec t in d e x
383
L ist o f sy m b o ls
Note: The following list is restricted to the m ost im portant subscripts, notations
and letters in the book.
Scalar quantities are printed in rom an letters, vectors in boldface, tensors
or m atrices in capital letters and in boldface.
1. I n d i c e s a n d n o t a t i o n s
The classification is limited to the m ost im portant indices and notations. F u rth er
term s are given in the tex t and in corresponding literature, respectively.
i ,j , k , .. .
a , 3, p ,...
Xj
x1
( ii )
'
n a b la-o p erato r
a4
V
differential operator
fl B
intersection of A and B
A c: B
A is a subset of B
for all
point
XIV
2.
List of symbols
L a tin le t t e r s
a 0 ,a a
a3
a ,b
semiaxis of an ellipse
b ap i
, b^
e;
ejjk , e'jk
f ,f
gj,g
gi.g*
g- , g 1J
height of a boiler
h j, h
hc , h|cu , hkl
buckling value
k =
t^
from the m id -p la n e
Pj
p seudo-load m atrix
p[f ( x )j
preference function
p ,p
Px >P,j >P
points
List of symbols
i j , rk
XV
orthogonal vectors
Sj
t , tk
stress vector
t 1; t x , t y , tz
11
u ,v
displacement vector
vq
w;
w*
x 1; x , y , z
x;
y,yj
transform ed variables
complex variable
Zj
Bik
C j, C
C ^1
elasticity m atrix
flexibility m atrix
Dx ,
, DXJy
Djjkl
flexibility tensor
E ,E
^ap-yl
F ,F
objective functionals
F j, F
F(x )
XVI
List of symbols
shear modulus
Gj
penalty function
Gj
m ean curvature
Hk
H , Hk'
HESSIAN matrices
integral function
I j , I 2 , 13
invariants
JACOBIAN m atrix
compression modulus
GAUSSIAN curvature
Kx ,
, Kxj?
K * , K y , K xy , H
differential operators
LAGRANGE -function
boundary moment
m om ent tensor
Mxx ,
, Mx^
M aP
N ^
N
,N
N
Nxx W xt?
N
N
_xx yy xy
Nx^
Pj
polynomials
Qx ,
boundary force
Rj
Rj
Rj , R2
R.
IR
stress tensor
List of symbols
S
XVII
tensor of n - t h order ( n = 0 , 1 , 2 , 3 , 4 . . . )
Tj
U*
volume
V.
Va
weight
W , W*
3.
Greek letters
or
otrp
otap
oo , Pj
LAGRANGE multipliers
$ap
Pj
P'J
th e rm al-elastic tensor
Yjj
strain tensor
^ofi >1
strains, distortions
shear deformation
shear strain
variational symbol
Sj , Sy
S-
, 8
ijk sijk
Eaf:
yy zz
i > <pip e <9<9
XVIII
List of symbols
T);
slack variable
/Cj, k 2
decay factors
K j, k 2
Kctp
tensor of curvatures
Xj
eigenvalue
Xj
X , (i
LAME constants
(i
v , v , \jy
, I;
POISSON's ratio
curvilinear coordinate system, GAUSSIAN surface param eters
m ass density
Pap
xx yy >aZz
I I I 1 ii i
principal stresses
stress vector
oM
tim e
r 1J
stress tensor
Tx y
Tx z Tyz
<p
X, ^
LOVE function
' ^x 'Jy
o ,X
, <J2
Ejjk ,
A,
0 ( ,5)
0^1
th e rm al-elastic tensor
I l e , FL
n * , 11*
n , II*
In tr o d u c tio n
1 Introduction
1 Introduction
context, one should m ention th e large am ount of novel finite com putation
procedures (e.g. F in ite E lem ent M ethods [A.1, C.25]) as well as th e Algo
rith m s of M athem atical Program m ing applied in stru ctu ral m echanics. One
can thus ju stly claim th a t all of th e a b o v e -n a m e d m ore novel fields and
th e ir solution approaches are all based on th e fu ndam entals of elasticity
w ithout w hich th e cu rrently occurring problem s cannot be solved an d evalu
ated. T h e field of S tr u c tu r a l O p tim iz a tio n increasingly moves away from
th e stage of a m ere t r i a l - a n d - e r r o r procedure to enter into th e very des
ign process using m ath em atical algorithm s (Fig. 1.1). T his developm ent
roots back to th e 17th century, a n d is closely connected w ith th e nam e
G ottfried W ilhelm LEIBNIZ (1646 - 1716) as one of th e last universal scho
lars of m odern tim es. His works in th e fields of m athem atics an d n atu ral
sciences m ay be seen as th e foundation of an aly tical working, i.e. of a cohe
ren t th in k in g th a t is a decisive assum ption of stru c tu ra l optim ization.
LEIBNIZ provided th e basis of th e differential calculus, and he also inven
ted th e first m echanical com puter. W ithout his achievem ents, m odern opti
m ization calculations would yet not have been possible on a large scale.
Here, one m ust also nam e one of th e g reatest scientists L eonard EU LER
(1707 - 1783) who extended th e determ ination of extrem al values of given
functions to practical examples. T he search for th e extrem al value of a
function soon led to th e developm ent of th e v ariational calculus w here
entire functions can becom e extrem al. Hence, Jakob BERNOULLI (1655 1705) determ ined th e curve of the sh o rte st fa llin g tim e (B ra c h isto c h ro n e ),
and Issac N EW TO N (1643 - 1727) found the solid body o f revo lu tion w ith
the sm a lle s t resistance. Jean Louis LAGRANGE (1736 - 1813) and Sir W il
liam Row an HAM ILTON (1805 - 1865) set up th e principle of th e sm allest
action an d form ulated an integral principle, a n d thus contributed to the
perfection of th e variatio n al calculus th a t still is th e basis of several types
of optim ization problems. M any publications on engineering applications
over th e previous decades utilize th e variational principle. LAGRANGE,
CLAUSEN an d DE SA IN T -V E N A N T h ad alread y trea ted th e optim al shape
of o n e -d im e n sio n a l beam structures subjected to different load conditions.
T ypical exam ples here a re th e buckling of a column, as well as th e canti
lever beam for which optim al cross-sections could be found using th e vari
ational principle. T his requires th e derivation of o p tim ality criteria as ne
cessary conditions; these are E U L E R 's equations in th e case of unconstrai
ned problems. If constraints are considered, as, e.g., in solution of an isoperim etrical problem, LAGRANGE's m ultiplier m ethod is used.
A Fundamentals of elasticity
A.1 D e fin itio n s - F o rm u la s C o n c e p ts
2
T e n s o r a lg e b r a a n d a n a ly s is
2 .1
T e rm in o lo g y d e fin itio n s
b ) A n a ly tic a l:
a = ax e x + ay e y + az e z
3
or
a = a e1 , +, a 2e , +, a 3e , V
y a 1e ;
i= l
T =
*21
*22
*23
*31
*32
*33
b ) A n a ly tic a l:
or
T =
t ij e l e J >
b* =
<=>
a1 +
b1 =
a2 +
b2 =
a3 +
b3 =
t
T .
=T
<=>
t
'1
= dT
dXj
T
i '2
=d ^
dx2
a = a* e 4 = a 1 ej + a 2 e 2 + a 3 e 3 ,
a = aQ e = a 3 e1 + a 2 e2 ,
T = ty e 1 es* = tjj e 1 e 1 + t 12 e 1 e 2 + . .. + t 33 e 3 e 3 ,
t; - tj - 1; +
12
13
df = f j d x 1 = ^ j - d x
'
dX
,
+ -^ -d x
dx2
+ -M r-d x 3 .
dx3
21 E 6 . Su G- E
1
J
6 i . . .
K
/ (ii)
a i = a i Vg
In
> a f = a i Vg
etc
E xa m p le s o f ( Hi) :
Following expressions are m ean in g less:
ci- 1i ! =
baa cos $ a
= 1
O:
^ + bk = 0 , k p ia = Bp
2.3
T e n so r o f firs t o rd e r (v e c to r)
Base vectors (F ig .
2 .2 )
, e2 , e3 ,
e; =
gj =
M e asu re o r m e tric co m p o n en ts
Si
Sj Si j gji gj gi .
(2.1a)
(g n ) =
S l2
S l3
622
S23
S3 3
D eterm inan t of th e m etric tensor
det ( g ) = |g | = | g - | .
(2.1b)
(2.1c)
x y = gij x1y J
Length of a vector x
(2.2b)
d = | x ] = / g i j xi xJ
Angle ip betw een vectors x an d y
SijX 1 y
cos ifi
(2.2c)
m n /
^gmn y y VgpqXP
Xq
Vg
1
0
for i = i
for i ^ j .
(2.3b)
Si gj = gji -
(2.4a)
g 1 g j = g ji
(M b )
( z -5a)
gi = gjj g J .
( 2-5b)
4 = gi j gj k -
(^c )
l
S
g2 x g 3
[ g i- g 2 . g 3 ]
g 3 x gi
3
[gi.g2 .g3 ] S
gi x g 2
[g1. g2.g3]
,9R)
s' = Pj gJ ,
S k = / Jk g i'
gk = ^ g
(%-7b)
Pl'P-.=t.
(2.8)
10
v 1 = (31 v 1 ,
(2.9a)
v.
v j = p[. v 1' .
(2.9b)
= /3V ,
or
a* = a;
(2.10)
T e n s o rs o f se c o n d a n d h ig h e r o rd e r
D e fin itio n s :
Two vectors x an d y are given in th e EUCLIDEAN space. W ith th at we are
form ing th e new product
T = xy .
(2.11)
(2.12)
T = tlJ g ; gj = tj gi g J = t J gl gj = U g gJ
According to th e position of th e indices one denotes
tjj
t'J
tlj
t;J
1 base e le m e n t,
T (1) = tsgi
31 =
T (2) = ti j g ig j
T (3) = t ijkg ig jg k
2
3
11
T ra n s fo r m a tio n ru les
For a tran sfo rm atio n of a vector base g ; into a new vector base g ;, equa
tions (2.7a) an d (2.7b) are used:
Si = P\ gj-
an d
g i. = p\. gj .
1 g k,
g l, = t ,J g ; gj
(2.13)
follows
or
t~ g 1 g J = tk, v g k
= / /?J t^ j.
or
t , r = ft) fi\, t kl
(2.14)
(2.15)
(2.16)
tJ
Vg( i i) / g ( j j ) >
t *1j =
'
(2.17)
t* J =
t;i
t+ij = tq
/^TiTT
7 (ii) / g (jj)
+ (at >ij
(2.18a)
(2.18b)
(218c)
12
ijk
~= V &
u i = 4 r ' l
a
(* *
+ 1
- 1
for
i k } cyclic
{ i , j , k } anticyclic ,
{ i >j >k } acyclic
(2.20a)
e 12 = +
,
(2.20b)
(*1)
or
(t* - X 6*) a- = 0 .
(2.22a)
t2 - A
(2.22b)
(2.22c)
13
2.5
h = t| ,
(2.23a)
i2 = i ( t W - t j t o ,
(2.23b)
I 3 = det ( t*) .
(2.23c)
C u r v ilin e a r c o o rd in a te s
* = *(')
(2.24)
B ase vectors
dT
e i = Vd x~' i = r 'i
Si
(2.25)
(2.27a)
dt .
(2.27b)
14
(2.28)
elk
, i = r u 8j .
( 219a)
= ~r L&-
( S9h)
f 2'30)
r m = g kmr ,jk
(Z31)
R u le :
Note:
Tyk = Fyik
2 )
r m n s + r m sn = T r r f '
interchangeability
of th e f i r s t two indices ,
(2.32a)
interchangeability
of th e la st two indices .
(2.32b)
r k = r k
2 s
'
(2.32c)
(2.32d)
a |j = a 'j +
By anology
a^
(2.33)
ak .
(2.34a)
= ai J - . T y a k .
(2.34b)
15
a.*
i ni m
.1l k a m
ij ,k
= A
+ r L
ln jm
..k 3 ,.1 m
*"1
(2.35a)
(2.35b )
+ ' L * " -
v = g rad $ = V $ # |j g J .
G radient of a vector v
(2.36b)
G ra d v = V v = v ^ g * g j .
Divergence of a vector v
div v = V - v = v j |j - --L yg
(2.37a)
( /g vj ) .
Div T = V - T = tkl|k gj
R otation of a vector v
(2.38)
r o t v = V x v = v J|i ( g 1 x g j )
LAPLACE operator
A
k) ,
(2.39)
'
B ipotential operator
AA $ = V 4 $ = V 2 ( V 2 $ ) = $|!j =
(2.40)
GAUSSIAN theorem
J J J H /g
V
d ( 3 = J J v j n jd A .
A
(2.41)
16
x = cos (
-l
-2
x = ( sm (
-3
x f
(2.42a)
Position vector
(2.42b)
= f c o s (2 e l + sin e 2 + e 3 .
r ( 7
62
+ s in e 2 ,
= -^sin ^ ej + ^cos^e2 ,
(2.43)
"*7
e3
C ovariant m etric com ponents according to ( 2.4a)
Sij 6 i
For exam ple:
g 22 = B2 '
62
= 7
) 2 sin 2
+ ( f
cos2
= ( f
(gy) =
(? ?
(2.44)
g^ = gj
O rthogonal base
( )
i
g(ii) = 1
S(ii)
for i f j
17
( g ij)
o U
1)- 2
(2.45)
I gij I
1 )2
(2.46)
T 221 = y { g
2 1 ;2
+ g 12
>2
- g22;1) = - ? r (
A ) = f*
-e
e1
(A p) =
'
1
(2.4 T)
'
( A j2) =
(A ja)-
>
T 22 = glk f 22k =
( - ^ 1) +
-|-0
- 0
= -^1.
(*i> =
0
0
(2.48)
0
(Aj)
1 )- 1
1)"1
0
0
18
3 State of stress
^ ^ gJl^ d +
[ ( / g gU ^ , i ) , i + ( / g g 2 2 ^
= | i [ ( e ^ . 1),i + ( ^ ,
e1
e1^ , n +
^ ,1
^ ,2 2
A $ $ n + ^1 ^ , 1 +
S ta te o f s tr e s s
3.1
S tre s s v e c t o r
2 ),2
, 2 ),2
g J2 ^ ,2
+ gJ 3 # ,3)
+ ( / g g33^ . 3 ),3
+ (e^,3),3] =
+ e1^ 33
^ !^ 2
^ 1 22
^,33
(jg.4s;
19
t=
lim 4M r = 4 r -
aa-o^A
dA
(8.1)
1
'
II
>>
xy
xz J
T
yx
cr
yy
r
yz J
Tzx
(3.2)
ii
<7
zy
L
(7z_z
"positive"
cut planes
"negative"
cut planes
1
cut planes
k c yy
T^vx
positive
stresses
X
M
negative
stresses
Ih
20
3 State of stress
3.2
S tre s s t e n s o r
"
ft x j t y ! t z1
^xx
Txy
L
xz
Tyx
yy
Tzy
Tyz
^zz
Tzx
(3.3)
J
or
t = Sn
(3.4a)
t1 = 7 * 1 1 ,
(3.4b)
I n words-. T his form ula gives th e stress vector t at a given surface or cut
plane in term s of th e stress tensor S an d th e u n it outw ard nor
m al vector n for th e surface or cut.
T he stress vector t acts on th e infin itesim al area dA of th e inclined cut
plane characterized by th e u n it outw ard norm al vector n ( Fig. 3.4 ):
" n
y
, nz.
r t X i
t =
ty
A .
( 3.5a,b)
dA cos a dA n
dA cos B = dA n
H
y
dA,Z = dAcos 71 = dA n ;
dA y
dA- = dA n- .
(3.6)
21
ffx x n x
ty =
r xy n x + < T y y n y +
Tzy * z >
4.
Tx z n x +
^zz n z
+ Tyxny +
Ty z n y
Tz x n z
(3.7)
Note: Shear stresses are pairw ise equal to one another, i.e., the stress ten so r
S is s y m m e t r i c
r xy = r yx
(3.8)
r yz = r zy
Symbolic n otation
S' = B
(3.9)
BT .
22
3 State of stress
3'
3 J
2'
B =
ft
X
(S.10)
P r in c ip a l stresses, p r in c ip a l axes
P rincipal stresses ( see (2.22) and (2.23) )
(3.11)
( Tj a ^ ) n * = 0
C haracteristic equation
cr3 - Ij cr2 + I 2 a - I3 =
(3.12)
w ith th e invariants for anv direction and for th e principal stresses cr.
( i = I , I I , I I I ):
G-T,
yy
Txy
Txy
ayy
yz
C j C jj +
<7n CTjjj +
( 3.13a)
c rm
<7j ,
Ti
(3.13b)
o
h =
XX
T*y
T xz
r xy
yy
Tyz
a .ZZ
xz
(3.13c)
yz
o zz = r xz = r yz =
(3.14)
Stress tensor
a
S =
r xy
xy
a yy
(3.15)
23
/3j = cos ( +
a ) = - sin q
/?2
= cos a .
( CTxx + yy ) + \
( <Ty y - f f x x ) s i n 2 Q
(3.17)
T zxy
v COs 2 q '
P rincipal stresses
ai
1
ii J
+ T.y
= - ( crx x + a y y
2 l
(3.18)
2 t . ...
xy
a x x - a yy
(3.19)
a* and
a* + it or a* and a * +
7T
= i/(s
D irection
yy
a** = a*
+ r xl y.
= ^
~ U
(3.20)
24
3 State of stress
(a
+ T = ( 7 - 70 f + T2 =
(3.21)
2 '
xx
(3.22a)
+1 Oyy
4- txy
3 .4
= r max
(3.22b )
S tre s s d e v i a t o r
D efin itio n :
'r* = t* - a M <5
(3.23)
( x x
y y + a zz ) ~
( a l + Cjj + <7n I ) 3
( 3 .2 4 )
P h ysica l in te r p re ta tio n :
T he stress deviator 'r j expresses th e deviation of th e state of stress from the
m ean norm al stress.
Since 'Ij =
from
0,
( 3 .1 2 )
v 3 + 'I 2 V - ' I
w ith
I 2 I2
3 crM ,
Ij
I 2 <7m +
(3.25)
|
I3
<tm .
(3.26)
25
E q u ilibrium conditions
xx+
3x
dX
dT
d ryx
d
a..
dT
dX
d r zx
dZ
d T.
dX
d
d
(3.27a)
d z ^ + fy = 0
dz
+ f =
o .
(3.21b )
) C urvilinear coordinates
- j li +
or
Div S + f =
(3.28a)
0
0
(3.28b)
3 ) C ylindrical coordinates ( * t ,
d Tyr
d
1 d r ,
dr
+ J - J ! +
r d ip
dZ
f d cr
r
dip
= ip , ( 3 z ) ( Fig. 2.5 )
1
+
r (Va r r - a
dr
9
+ -V 542- + r
r
__zz_
i z
+ fv =
f
rz
) + fr =
tptp'
=
z
(3.29a)
26
4 State of strain
4. 1 (
+ ~r Tdip
7n + T (
ffW
) +
fr =
0 >
(3.29b)
d T
JL +
c)r
1 d<r
dip
' r ^r
dI
3-----r Tr^COt ^ +
+
( 2 a r r - <r,,
- cripip ') + fr
j- v
i/I?
dr
dT
dr
d i?
r sin i?
ip
+ - T
(3.30)
, +
+ y c o t 0 (<rw - crw ) + {* =
dT
jr<_
dI
dT Q
+ -r - Vdl>^ +
iff. .
-----
r s in v
d ip
_|--------- _
r
_|------- T
rip
S ta te o f s tr a in
4.1
K in e m a tic s of a d efo rm ab le b o d y
QOt
ip v
_(_ f
q .
ip
27
(4.1)
(4.2)
(4.3a)
dr =
(4.3b)
of
d^1 = r (i d* = g; d f 1 .
(4.4a)
Si = P \ gj -
(4.4b )
28
4 State of strain
(4.5a)
( 4.9a )
dr = F d f .
(4.9b)
(4.10a)
4.2
Strain tensor
29
d v = Grad v dr .
dv =
du
dv
dw
d U
dU
du
d X
d V
d V
dZ
dV
dW
dz
dW
^y
dZ
d X
d W
dX
4. 2
dx
(4.10c)
dy
dz
S tra in te n so r
(* U a )
(AUb)
(4.12a)
i)
(4.12b)
30
4 .3
4 State of strain
S t r a i n d i s p l a c e m e n t r e l a t i o n s
- C artesian coordinates
From the tensor of th e displacem ent derivatives V follows as the sym m etric
p a rt the lin e a r stra in tensor. According to th e rules (2.18), it becomes in
C artesian co o rd in ates:
J_ 7
-y
2 ' xz
2 ' xy
J_
VS= | ( V + v T)
'yz
(4.1S)
II
_ dU
xx ~ d x
dV
dU
yy ~
7xz
<3U
zz ~~ ~d~Z
d Z
+
+
dV
<3X
dw
(4.14)
dX
<3w
<3V
7yz = <3z +
>y
Special c a s e s :
- C ylindrical coordinates ( Fig. 2.5 ) ( r , u ; (p, v; z , w )
<3u
<3r
_ 1 <3U
r d(p
iu
7rz _ ~d~z
<3W
<3Z
v>z
<3V
dr
3w
~dV
_ <3V
<3z
1 <3W
r <3y
p ,w )
du
dr
1
d-W
a ,
U S in 17 + V COS 17 I ,
>
1 d V
~d
1
dU
r sini?
dip
+
'
dr
w
r
1 iu
, dV
7ri> r <3i?
dr
1 ( dv
<3w .
0\
T qI
h ~ FTsin V ~ W COS V 1 .
in v V d tp
dv
/
V
( i = x , y , z ) are
31
A2 + I 2 A - I 3 = 0 .
( 4.17a)
j .
(4.17b)
T he lin ear stra in - displacem ent relations ( 4.12b) form a system of six
coupled, p a rtia l differential equations for th e th ree com ponents Vj of th e
displacem ents for given values of th e stra in tensor. Thus, th e system is
kinem atically redundant. In order th a t there will exist a displacem ent
vector Vj subject to given values of th e six m utually independent com po
nents of th e s tra in tensor, it is necessary th at th e th re e com ponents of the
displacem ent
vector satisfy th e following com patibility conditions ( DE
SAINT V E N A N T ):
Tijlkl + Tki|ij - 7ii|kj - 7kjLi = Tijlki ^ n ,km = 0 .
(4.18)
M echanical in te rp re ta tio n :
T h e interior coherence of th e body has to be preserved after th e defor
m ation, i.e. m aterial gaps or overlaps m ust not occur.
For a two - dim ensional state of stress or stra in th e com patibility condition
in C artesian coordinates reads as follow s:
d2e
d2e
d2 7
f - + ----------------------- = o
(4.19)
ay2
5
5.1
In th e following we a re going to deal w ith bodies for w hich th ere exist reversibly unique relations betw een th e com ponents of th e stra in tensor and
th e stress tensor, an d we furtherm ore assum e th a t these relations are tim e
in d epend en t. T h e behaviour of th e bodies is denoted as elastic, i.e. th ere are
no perm an en t strain s ep] after rem oving th e load of th e body ( Fig. 5 .1 ). T he
bodies considered shall furtherm ore, as it is usual in th e classical elasticity
theory, be m ad e of a lin e a r ly elastic m aterial such th a t th e ir constitutive
law expresses lin ear relationship betw een th e com ponents of th e stress ten
sor a n d th e s tra in tensor ( range 0 - A - in Fig. 5.1). Such bodies are usually
called H O O K E A N bodies.
32
A
B
C
D
D- E
F
epi
G eneralized
H O O K E -D U H A M E L 's
elastic, isotropic m aterials
law
for
7 xy
Tx y
'x z
therm o
C a r t e s i a n co o r di na t es
i r IK XxX - ^ K v + O l + a T e
XX
zz H=c If o zz - v i Va
xx
+ a y y ')I1 + a
T (9
xz
(5.1)
G Ty%
with
YOUNG's modulus ,
POISSON's ratio ,
E
2(1 + v)
sh ear m odulus ,
one dim ensional th e rm a l expansion coefficient ,
difference betw een fin al an d in itial te m p e ra tu re .
= T, - T0
Symbolic notation
.
V .with
~l~ 1/
1/
g
S -^ g -sI+ a TI
(5.2)
33
i/
i + v ( exx + i _ 2 v e /
a xx -
_ r.
Txy ~~
l + v Veyy + l - 2 v e /
a yy ~
l - 2v Qt
1 - 2v T
_
T ~
l + i/ ( zz + 1 - 2 v e ) _ 1 - 2 v Q t &
7x y
7x*
(5.S)
Tyz = G 7 yz
S = 2G
w ith
-
V
T
------ e l 1 - 2v
1 +
1
v
2
Clrr 0 1
(5.4)
e = volum e d ilatatio n .
Tkl - " E kl T i +
a T Ski = ^ i j k l T
+ a T Ski
( 5.5a)
2E
^ k Sjl 7
S;i Sjk )
Sij Ski
(5.5b)
Ta = Cykl7 k l - l8
/-< /
G (g
ik _jl
il
jk
g + g gJ + x
2V
ij k l .
-2 - g J g )
(5.6b)
E a,T
1 - 2 //
ij
( 5.6c)
Tv = c*iJkl(7kl - aTgkl)
w ith
ik J 1 ,
(5.7a)
il J k '
C*Jk' = A6 gk l + M ( g ' V + e V , )
(5.7b)
34
tfj,
/i = G =
(3
,v
A( 1 - 2 v )
2v
(l +
f j, (E - 2f i )
Ev
(1 +
v) { \ - 2v)
E
(l +
+ 2 /i)
+ /i
i/)
A
(A+
i/)(l-2 i/)A
fi)
V
E
- E
,v
A
A
v =
fi, E
E
E - 2 /i
2 /i
T'k
m
Tk
rm
, m
= y e \ + 7k ,
i 1 rm , m
= y s \ + Tk
^
\I
)
(5.8)
s-
e=-^(JM T 3 aT
w ith
K
=
= s /3
- -
(5.9b)
35
'M
a ) S tate o f p la n e stress
- C artesian coordinates
D e fin itio n :
= T
(5.11)
0 .
yz
V ^vv ) +
yy
- v o
a T
) + a (9
xx'
( 5.12)
yy
yz
a
yy
o-- =
w ith
E
- v
E
l-i.
0
K x
2~ Lyy +
V yy - ( !
0 ]
t xy
(5.13)
T
= Tj
x ,y
= G 71'xy
( 1 + v )aT &1
^ xx
(x ,y )
V )aT
yz
= 0
- T0 .
(5.14)
36
w ith
yy
rxy J
t/
0
0
1
- v
Cq -- CXrp &
yy
L 7'xy
X J
C urvilinear coordinates
T he equations (5.12) and (5.13) read in index notation
I-hvfi-v/T
a p y 6 T
1
7 exp
a p
^ a p -y S
(5.15a)
+ Ctrj, g a p
>.
(5.15b)
( 7 7 - Q T g7 & )
= E
w ith the plane elasticity and th e plane flexibility tensors of fourth order
otfi'yS
1 + vi/ (
2E
E
f
+7 v] )( s
ap 'y6
b)
-l-
^*7 ^/38
oc'y 08
2 ( 1h
if
oc6 P'y
+g
if
,
2 is
+ t ^ ; s
ap
8 s.
D efin itio n s:
(5.16)
e 1 + v
yy
E
U 1 " K
*/ 0 y y ] +
[(1 - * 0 % , - ^
(!
xx] + (1
J ' W
+ v ) a T
> 7 x y = - G ^ xy
, 7 XZ = 0 ,
(5.17)
-yl y z 0 .
Stress - stra in relations
a xx = - ( T + T X r ^ ^ l [ ( 1 " " ) xX + " yy - ( 1 + " )T 0<>]
^yy = ~(I + y f ( l -
2 t/ ) [ ( 1
~ ^ )yy + v e xx - ( ! + " K
0] -
Ev
(1 + v ) ( l - 2 v Y
Txy
^ 7 xy
Txz
Tyz
**
0
(5.18)
37
M a t e r i a l l a w fo r a u n i d i r e c t i o n a l l a y e r ( U D l a y e r ) o f a
fib re rein fo rced c o m p o site
=E
(5.19)
ly s
a ."(3* 7 ' 6
(E 1
) =
2' 2 '
E 1#1<
e 2'2 2 ' 2'
V\'2' E 2'
- vy y vy i.
0
vy i . Er
- vy y vy y
e 2.
1
- V y 2, V p y
0
, 2' - co
(5.20)
<*vr .
2-
38
v 2.y
Gr 2 ,
= V1'2' ^ 2 '/ V
= / $ / / $ / E - r *v
(S S I)
( )
A -
cos a
sin a
- sin a
cos a
(5.22)
E
g l ' 2 ' 22'
i'i'
E11
E1' 2'
E1'3'
E 2'1'
E2' 2'
E 2'3'
E 3'1'
E 3' 2'
E 3'3'
(5 1 3 )
,2 2
-3 3
- ,1 ' 1 '
=E
,2 '2 '
=E
_ 3 '3 '
= E
- ,2 '2 '
cos a + E
4
_ 1 '1 '
cos a + E
,
1 . 2 '
+ -rA
4
. 4
sin a
. 4
sin a
+
,
.1 '
2_
.1 '
2 .
A sm
+ A sin
2a ,
2a ,
(5.24a)
. 2_
sm 2 a ,
39
E 12 == E 21
E 13 == E 31
II
BM
(O
= 4 l-
A 1' -= E2'1' +
A2' sm
* 22 q ,
A
E1'1' + A1' + A
2'
sin a ] sin
(5.24b)
. 2' . 2 I
E 2'2' - A 1' - A sin a j sin 2 a
E 23 == E 32
w ith
0.01
E3
'3'
E n e r g y p r in c ip le s
6.1
40
6 Energy principles
E nergy expressions
( 6.1)
6 W = F(5u .
Here, we use the differential 6 for the changes of state, e.g. deform ation dif
ferentials, strain differentials. For these quantities it is assum ed th a t they
are virtual ( not existing in reality ), infinitesim ally sm all and geom etrically
com patible. Eq. (6.1) illustrates the area of a th in strip w ith the w idth 6 u
and th e height F in a force-deform ation diagram (Fig. 6.1), where term s of
higher order have been neglected. T he total work of th e single force results
from an integration over th e deform ation differentials
U
W =
( 6.2)
F<5u
iup
,r
~ i ->
dV
dx
a)
Fig.
6.1 :
41
(6.3)
or
F
(6.4)
F= 0
= W * = | f - u .
(6.5)
( 6.6)
(6.7)
( 6.8)
T he relatio n betw een th e stress a and the stra in e is given by a non - linear
curve sim ilar to th e one shown in Fig. 6.1a. If a lin ear a, e - curve exists, by
analogy to (6.5) we obtain th e following for the specific deform ation energy
and the specific com plem entary energy
U = U* = - i- a e
(6.9)
42
6 Energy principles
( 6.10)
w ith
-T
P T = ( P x . Py , P z )
FT = ( F ^ ,F j ,...,F f )
o
V =
,o
w
T r a n s itio n to isotropic, lin e a rly elastic body
Specific deform ation energy and com plem entary energy
TT
TT>k
I
t
U = U*
= yI f f t e = -^e
a.
( 6.11)
Introduction of H O O K Es law
U =
l eT C e ,
(6.12a)
U* = y trT D t r
(6.12b)
6.2
Energy expressions
43
( x x y y zz > ^ x y T y z ^ z x )
( x x y y z z *^ x y ^ y z T z x )
1-1/
l-i/
E
( l + (/ ) ( 1 - 2 . )
0
I
|
I
II 1 " ^2 "
" i "
= C
(6.13a)
- 2v
2
- V
1
1
1
1
0
*0
II
D = i
1,
1
11
1*
(6.13b)
2 ( 1 + 1/ )
0
0
0
2(1+1/)
0
0
0
2(l+i/)
= U * = - ^ T ij 7ij
(6.14)
(6.15a)
U + = l D i j kl r 1Jr k I .
(6.15b)
(6.16a)
U* =
(6.16b)
j T1J( 7 ij + a T g y 6 ) = i - D ij k lr u r kl + a T r \ .
44
6 .3
6 Energy principles
P rin c ip le o f v ir tu a l d is p la c e m e n ts ( P v d )
Jt *6 7^ dV
S W = <SU =
(6.17)
v
W ith th e stra in energy U =
U
J l l d V = IT
(6.18a)
(6.18b)
n = n;+ ne.
( 6.19)
n e) = 0 .
(6.20)
( 6.21a)
( 6.21b)
P rin c ip le o f v ir tu a l fo rc e s (P v f)
T he com plem entary v irtu al work 6 W * of the external forces is equal to the
increase of th e com plem entary v irtu al energy <5U*:
6 W * = 6U* = J ^ ij <5rij dV .
(6.22)
n* = nf + n*
( 6.23)
45
U* = JlJ*d V = II1*
v
(6.24a)
-n*
(6.24b)
P rinciple of sta tio n a rity of th e v irtu a l to tal com plem entary potential
(6.25)
( 6.26a)
( 6.26b)
v- = ^ H ( F )
/ s.27a)
SF1
T heorem by M ENABREA
= 0 ,
(6.28)
c) ( F )
w here th e index R refers to th e reaction forces.
Second theorem by CASTIGLIANO
i
f iU f v .)
F =
(6.27b)
+ v j . ) - 7 ij]} d V -
(6.29)
J P q Vjd S + J ( v i0 - Vj) p1dS
(6.30)
46
6 .5
6 Energy principles
R e c i p r o c i t y t h e o r e m s a n d U n i t L o a d m e t h o d
T heo rem by B E T T I
6 .6 T r e a t m e n t o f a v a r i a t i o n a l p r o b l e m
*
<51 = 6 J F ( x , y , y ' , . . . , y ( n ) ) d x = 0
(6.34)
<51 = J ( 1 F _ d d F
J Vdy
dx 5 y '
xi >
47
j d ^ d _ \ g dx
dx2 ?>y" >
-
==> E U L E R s eq u atio n
(6.S5)
+
==>
( dF
iF
I 2
y w ~ t e W T) 6y\ +
'Y J
resid u al ( physical )
b o u n d a r y c o n d itio n s
J
dF
^ y
j-
6y
= 0.
1
1
,. , / '
. \
e sse n tia l ( g eo m etric )
boundary conditions
6.7
C o n tin u o u s a p p ro x im a tio n m e th o d s
1 = J F ( x ,y , y ' ,.. . , y ^ )d x
E x tre m u m .
xi
y * ( x ) = 2 * n y n* ( x ) >
n= 0
( 6-S6)
J L (y * )y ^ d x = 0 ,
(n = 1 ,2 ,3 ,.,.,N )
xi
w ith L ( y* ) as th e differential equation for th e problem ( see (6.35) ).
(6.38)
48
7 Problem formulations
7 P r o b le m f o r m u la t io n s in t h e t h e o r y o f lin e a r
e la stic ity
7.1
B a s i c e q u a t i o n s a n d b o u n d a r y v a l u e p r o b l e m s
(7.2a)
- M ixed b o u n d a ry -v a lu e problem
On one p a rt St of th e surface S of th e body B, th e tractions are given, and
on th e rem ain in g p a rt Sd of the surface th e displacem ents are given ( Fig.
7.1c ). T he boundary conditions th en read
49
< k ) = <TS n i ) s , .
vi<Sd> = ( vi>sd 7.2
( >
S o lu tio n s o f b a s ic e q u a tio n s
i .
A v + e
A + fi
|i
3A +
2n
--------------
A + fj.
A + fi
f = 0
(7.3)
^ - s | ij
(7.4)
*
g* A s = 0
1+ v
S p e c ia l e q u a tio n s fo r th r e e - d im e n s io n a l p ro b le m s
(7.5)
T his bip o ten tial equation has an in fin ite num ber of solutions, e.g. feasible
solutions in cylindrical coordinates are for th e axisym m etric case [ A.9 ]
X = r 2 , In r , r2 In r ; z , z , z3 ; z In r , R , i , In ^
K
H
w ith
, z In ( z + R )
(7.6)
R = / r2 + z2 .
50
7 Solution procedures
o <p<p =
1 - 2i/
2 -1 /
1 1
1 -
7.4
( 7.8a)
2 v dZ
1 -
2i/
i/
2
d
dZ
(7.8b)
(7.8c)
dZ
d
* - V dz*
y -
(7.8d)
j j L
1 - " a z2
d r
S p e c i a l e q u a t i o n s fo r p l a n e p r o b l e m s
a ) S tate o f p la n e stress
- Solved w ith regard to displacem ents
N A V IER's equation
a \/3
1 + 1/ J 3 \ a
1 + 1/
^ila , 1 (O
M ^ + T ^ r ^ l / 3 2T ^ r a T 0 I
g~
=
r.
(7.9)
(7.10)
(7.11)
w here S' is called th e th erm o - ela stic d isp la cem en t p o te n tia l [ A.13].
- Solving w ith regard to stresses
a il
B v _
761
ap-y6T Imi/ + a T
0 _ |6
(7.12)
y6
= e
fir x|
, ir
e $ \OT + V g7
(7.13)
or
b)
51
(7.11)
AA $ = - E a T A - (1 - u ) A Y
S ta te o f p la n e s tr a in
- -j +- ~ a T e
1 -
(7.15)
7.5
1- 2u
1- v
E
1 - i/
AA$ =
C o m p a ris o n
s tra in
of s ta te
o f p la n e
(7.16)
s tre s s a n d
s ta te
o f p la n e
dr
dx
dr
d X
dy
+ fx = 0 ,
x
do
+ V- + f = 0 .
d y
y
(7.17)
xy
sin a yIS
)L = t xO ,1
(7.18a)
Ur- = u
Vr. -= v .
(7.18b)
52
7 Problem formulations
yy
g T*y
i/2 )<7
- i / ( 1 + v ) 0 xx\ ,
(7.19)
Q Tx y
7xy
A (' a X X + a V V ') =
(7.21)
df
-(1 +/)(
dX
= +
_i
(
1 - u )) 1V a x
_____X
^>y )
d y '1
+ V
J,J'
ax'
+v
r xy
^
a2 $
axay
(7.22)
^a x
= -f
^V
b^ y = - f y
AA<2> + 4 A V = 0
(7.23)
If no volume forces are present, th e biharm onic equations (7.23) take the
sam e form for both states
AA< =
bA $ + 2 _ 4 *
2
2
^x4
a x <ay
0 .
^y4
(7.24)
A .2 E x e r c ise s
E x e r c i s e A -2 -1 :
An oblique base gj ( 1 = 1 , 2 , 3 ) expressed by th e orthonorm alized base
vectors e i ( i = l , 2 , 3 ; | e i | = 1) is given:
51 = e i -
5 2 = el + e2 i
63 =
e i +
e 2
+ e 3
,y = - e j
- e2
y , | x | , and th e angle
+ 4e3
ip betw een th e vectors (Fig.
-1 -1
0 -1
0
1
( i ' = 1,2,3),
54
S o lu tio n :
a ) We first determine the required quantities in the orthonormalized b a se :
x = 2 e j + 2 e 2 + e 3 = x1 ej
x1 = 2
y 1 = - 1! ,
y = - e, - e 2 + 4 e 3 = y ej
x2 = 2
x3 = 1 ,
y 2 = - 11 ,
y3 = 4A .
+ x2y2 + x3 y3 = 2 ( - l ) + 2 ( - l ) + l - 4 = 0 .
8ij
1 1
2 2
x y + x y
cos <p =
/ x 1xJ Sy / y k y 's kl
3 3
+ x y
= 0
<P= 2
3-/l8
i.e.
y 1 x .
e 3 = 83 S 2
- g 2 + g 3 = x 8i
-J
y = - 5 g 2 + 4 g 3 = y gj
= 0
y1= 0
x2 = 1
y3=
= -5
8i ' gj gy
(g ij) =
Sn
Si 2
g 13
*12
S22
S 23
g 13
g 23
g 33
= el el = 1 - 1 = 1
gj2 = g i g 2 = e i ( e i + e 2 ) = 1 >
813
= 1,
4
Exercise A-2-1
622 = g 2
82
= ( e i
+ e 2) =
+ e 2 ) ( e i
623 = 82 83
=( e l e 2 ) ( e l + e 2 + e 3 ) = ^
g 33 = g 3
83
=( el
reads
(g ij) =
1 1
1 2
1
2
+e 2 + e3
1 +
55
1 =
2 >
e3)
= 1+ 1 + 1 = 3
^= ^ >
) (e i+
e2 +
(1 )
y =
-1 -j
g;jx y =
1 j
_2_j
_3_j
gjj* y+g2jx y +g3j x y =
_ 1_ 1
_1_2
_1_3
gnx y+g12x y +gi3x y +
21
_2_2
_2_3
+ g2i x y+ g22x y + g23x y +
+ gji^3y1+ g32x 3y2 + g33x 3y3 =
0 - 0 + 1- o - ( -
+ 1
0 - 4 +
+ 1 1 - 0 + 2 1 - ( - 5 ) + 2 -1 4 +
+ 1 1 0 + 2 1 (-5) + 3 1 - 4 =
- 1 0 + 8 - 1 0 + 12 = 0 .
L e n g th of vector x in the covariant base
I2
XI =
j _________ _1
8 l l X X + gl2 X X +
_3_1
_3_2
+ g 3J X X + g 32 X X
0 + 0+ 0 + 0 + 2 / 9
l-
g l 3 X X + g21 X
-3 -3
+ g 33 X X =
+ g22X X
l + 2- l - l + 0 + 2- l -
-2
+ g23X
l +3- l -
l = 9;
-J
gijx y
cos <p =
m n
/ g ki x k x l
,y
3 /1 8
According to (2 .5a )
g 1 = g 1Jgj
and to ( 2.5b)
g. =
g.. g J .
56
i2
8
i3
Sk2
Sk3
and further
.1
11
k 8 Ski
2
k
12
13
Sk2
Sk3 >
21
22
,2 3
8 Ski ^ 8
Sk2 ^ 8 Sk3
_3
31
,3 2
, 33
k 8 Ski + 8
Sk2 + 8 Sk3
g
g
g
11
12
12
22
13
13
23
23
33
8 11
Si 2
813
812
822
823
813
S23
833
0
1
0
( s ij) = ( S i j T 1
1
2
3
-1
( I-1
1
(8 ii)
=
2 2
2 3
1 1
2 3
1 1
2 2
1 2
1 3
1 1
1 3
1 1
1 2
1 2
1 2
1 1
1 2
1 1
1 2
The determinant is
8 = 8 , J
4 -3
= 6 + 2 + 2
(g ij) = (g ijr 1 = |
0
-1
1
-1
2
-1
g gj
= g
11
12
13
g i + g g 2 + 6 g3 = 2 g i
12
, 22
, 23
S2
S g t + g g 2 + g g 3 = "gi + 2 g 2 - g 3 ,
3
g =g
13
23
g l + g
33
g2 + g
g3= - g 2 + g 3 -
0
0
1
Exercise A-2-1
57
C heck:
According to ( 2.3a)
gj gJ =
with ( 4 ) yields
g1 = ei ( ei - e2) = 1
El
E 1 = ( g 1 + g 2 + g 3 ) ( 2 E i - E 2 ) = 2 S i S2 = 2 ~ 1 = 1
El
g 2 = e j ( e 2 - e 3 ) = ( g 1 + g 2 + g >) ( - g 1 + 2 g 2 - g 3 ) =
- S j + 2g| - S3 = - 1 + 2 - 1 = 0 .
Second w a y :
The contravariant base vectors can be calculated by forming the vector products
according to (2.6):
E
82
X8 3
2
7
S i >S 2 E 3 ]
[ E 1. E 2183J
1
1
= S2 x S3 =
83 x gj
8 ] x E2
[ 811 E 21E 3 ]
[ S i j E 21E 3
= 1
ei
1
e2
1
e3
0
= gj - ( g 2 - g j ) = 2g j - g 2 ,
58
= S3 x Si =
e l e2 e3
1 1 1 = e2 1 0 0
= g 2 - Si - ( g 3 - s 2) =
= - g t + 2g2 - g 3 ,
g = *1 x S 2 =
el e 2 e3
1 0 0 =
1 1 0
e3
= g3 _ g2
g = K K = ( 2 g j - g 2 ) ( 2 g l - g 2 ) = 4 g u - 4 g 12 + g22 = 4 - 4 + 2 = 2 ,
12
1
2 /.
w
X
g = g ' g = ( 2 g t - g 2) ( - g j + 2 g 2 - g 3) =
= ~ 2 gu + 4 g12 - 2 g13 + g12 - 2 g22 + g23 =
= - 2 - 1 + 4- 1 - 2 - 1 + 1 - 2- 2 + 2 = -1 ,
etc.
c ) The vector a in the covariant base
a = g x + 2g2 + g 3 = a'gj
( &1 1 ,
2 , a =1)
ai = 8 n a + gi2 a + gi3 a
1
aj = l - l + l - 2 + l - l = 4 ,
a2 = 8 i 2 a + g22 a + g23 a
a = 1 1 + 2 - 2 + 2 - 1 = 7 ,
a3 = Ei 3 a + g23 a + g33 a
a3 = l - l + 2 - 2 + 3 - l = 8 .
_
+
3
7g+ 8g .
1/
= a Vg(ii)
or
*
/ (i i)
a* = a iVg
Exercise A-2-1
59
a* = 4 / g i i = 4 / 2 ,
a*2 = 2 A 7 2 = 2 A
a* = r y g 22 = 7 A
**3 = l l / l h 3 = A
a* = 8 A
= 8
= 8A
p f .
1
1
1
-1
0
0
-1
-1
1
i' = 1 , 2 , 3
new base
gj > P jr g i + Pi / g z + Pi* g 3
1
gi
g2
g3
el
e 2
e 3
g 2' = P2 - g l + P2 - g 2 + P2 - g 3 = g l - g 3 = e 2 ~ e 3 >
3' = ^3 ' 1
^3' g
2 b
^3 ' 3 = 1 b 3 = ^ e l b
e2 + e 3
1
1
"+ g2' + "o~g 3
g3 "2"2' b"2"g3'
The transformation matrix then follows to be
k'
V2
V2
-V 2
V2
-l
-1
1
0
1
V2
V2
0 -V 2
V2
1
0
0
0
1
0
0'
0 = S:.
1
>
60
E x e r c i s e A -2 -2 :
In order to d eterm ine th e state of stress in a th in p arallelogram disk ( Fig.
A -3), a tran sfo rm atio n in an oblique coordinate system
( a = 1 , 2 ) is ad
vantageously carried out for establishing th e required equations.
a ) D eterm ine th e base vectors, th e co- an d th e contravariant m etric tensor,
and th e CH RISTO FFEL symbols.
b ) How does th e bipotential equation
= 0 (<? = A IR Y 's stress func
tion ) read in th e oblique coordinate system , and how do we determ ine
th e associated stresses from it ?
S o lu tio n :
a ) According to Fig. A-4 , we express the relation between C artesian and oblique
coordinates as follow s:
x 1 = Ic 1 +, Ic 2 cos a ,
2
x = !; sin a
, ( a = const ) .
Base vectors
According to ( 2 .26 ) , for the coordinates in the plane the relation between a base
g Q ( 5 ) and the orthonorm alized base e Q is
_ dx
ea '
Exercise A-2-2
61
dx
dx
df;1
1 ^2
^1
df;1
dx1
dx2
ga
gp
812
^21
= gl
g 2 = el
( c o s a e j + s i n a e 2 ) = cos a
e2 =
cos a
cos a
(2 )
(3)
- cos a
-cos a
(4 )
CHRISTOFFEL symbols of the first and second kind from (2.30) and (2.31)
0)07
8 q^3 = const .
r c*/3 - 0
(5 )
/r
.p
62
AA=gV
AA = g
11
12
+ g
11
+ 8
11
lm + g g
. 1211 + 8
22
11 _
im + g g
2111 + 8
,1212 + g
f2 U2 + g
12
22
12
1221+g
g
22
g 2121 + 6 6
21 ,
22
21
21
22
,2212 + g
1121 + g
11
21
21
21
12
12
22
g ,2211 + g
11
12
21
11
+ g
12
12
11
21
(6 )
22
,2221 + 8
1122
. 12 22 +
,2122 +
22 ^
,2222
By using the sym m etry g ^ = g ^ and SCHW ARZ's perm utation equation we
can further sim plify the fully expanded expression in ( 6 ) :
*
AA = g
11 11 .
, . n 12
, . 12 12
g l l u + 4g g >1112 + 4g g >1212 +
11 22 .
g
+ 2g
12 22 ^
,1122 + 4 8
22 22
,1222 + g
,2222
Application of ( 4 ) yields
A A ffi =
4 c s a |1222 + ,2222 J . 4
yo
6r ^ I
S IIo r
= e
yo
/g
+1
/g
-1
cor
1
and with ( 3 )
sT<T =
sm a
(8 )
0
W ith ( 5 ), the covariant derivative ( 2 .34b) changes into a p artial derivative
I c jt
= 1 , o ) ! = , o r - ,p r o/ rJ = , o r
v
J \t
Exercise A-2-3
63
Each component of the stress tensor can then be obtained from (7.13)
11
T
It _ I
= E
x
12
T
22
>
12 12
12
fiT
12
21
12 11 I
E
fU
| 21 +
11 12,1
E
|l2 +
12 1 2 , 1
E
l 22
,22
12 21 _
+
21
,22 . 2
sm a
11 21
E E
|u +
E E
. 2
sin a
11 22 _
ffi^ j
12 22 _
^
22
,21
26 2t ^
21 21 _
22 21
21 22 ^
22 22
= E E ,t = E E ,11 + E E
,21 + E E
,12 + E E
,22
^
=
1 2t
=
11
11 11,1
| iT =
22
X
21 21
= E
1
(1 1
sm a
,U
E x e r c i s e A -2 -3 :
An infin ite strip under constant tension crx has a crack which can b e pre
sented as an elliptical hole ( see Fig. A -5). T he geom etry is described by an
elliptical - hyperbolical coordinate system *, 2. T he relationship betw een
C artesian and elliptical - hyperbolical coordinates is given b y :
x = c cosh cos
x 2 = c sinh
sin 2
< 0 0
; 0 < < 2 rr .
D eterm ine
a ) the covariant base vectors, th e covariant m etric tensor, its determ inant,
and th e co ntravariant m etric ten so r ;
b ) the CH RISTO FFEL sym bols of th e first an d second kind ;
c ) the physical com ponents of a vector v .
64
S o lu tio n :
a ) For determining the covariant metric components we esta b lish the base vec
tors according to ( 2 .26)
dxh
a?
e/3
dx
dx
df;
df;
g j = j- e j -|
dx
S2
j -I
df;*
1 2
1 2
j- e 2 = c sinh \ cos \
+ c cosh \ sin !; e 2 ,
( 1)
dx
df;
_ ^
d x v dx
z ap - e a
~^p
~ *
. 1 . 1 . 2 . 2
O
X dx , OX dx
2 , . .2 ^.1 2-2
,2 -1 . 2 -2.
g n = j- j + T r = c (sinh cos g + cosh g sin Z ) =
dl
dZ
dZ
dl;
2, . . 2 J , ,
. 2 2.
. ,2 J , . 2 2.
2 , . .2 1
. 2J,
= c (sin h Z (1 sin Z ) + (1 + sinh Z ) sin Z ) = c (sin h Z + sin Z ) >
. 1 . 1
. 2 . 2
dx dx
dx dx
2.
-2 1 . 2 _2
. . 2 1 2 2.
822 = T^2 T 3 + TI2" T72 = c ( cosh 5 sm r + sinh I cos g ) =
df;
di;
df;
df;
2 . 2 1
. 22
= c (sin h I + sin \ ) = gu ,
% 2 d>x 2
dx
dx1 dx1
gl2
d^1 d?2 + d
2
df
(2 )
(gop) =
w ith
g ll
811
0
(3 )
a/3
g g/?T
a
7*
Exercise A-2-3
65
the non - vanishing contravariant metric components are inverses of the elements
of the covariant metric components
g
(aa)
=
. T
g(aa)
for
a = 1,2 ,
(4)
(s* ) =
0
b)
i//g
The CHRISTOFFEL symbols of the first kind can be calculated from (2.30)
S a / ^ 7 ))
6 7 a , (3
rOt
==>
1 d8 u
2 d?1
1 d8 u
2 d?2
A . - K1
r1 211r121 >
^221
2 ^g 21,2 +
8 l 2,2
822 ,1 )
1,
T = 2:
r1 212 r1 122
821,2
812,2)
= C S ln? COS? .
2
99
= c cosh 5 sinh I
~ ^ T
222
2 .
-S
1
^ ( g 22,i +
~~ COSllf> sinh ?
4 ^g u
8 u >2)
r !22 =
2 ^g 22,2 + 822,2
822, 2)
2 . 2
C Sm
C S
^822
'
The CHRISTOFFEL symbols of the second kind follow from (2.31) with ( 4 )
r 6 - </7 r
8
a/3
8 = 1:
rn= 8
^2
17
a /3 7
T-
__ ^
r ii7 = e
11
= e 17 r12T = g
rm + g
r121 +
12
t. t
r U2 = - ^ c sm hc cosh^ ,
r122 =
c* sin f cos r ,
y8
r121
1 r2
1 12
T-1
22
1T
^227
11 ~
221
12
^222
1 2 . , 1
, 1
y~~ c sin^ 5 cosh ^
66
8=
3 State of stress
9
27
rn = g r
ri2- g
27 _
=g
21 _
22
1 2 . ..2
_2
r u] + g r n2 = - ~ = c * i cos? ,
127 8
yg
21 _
r i21 + g
1 2 . , _1
,_ 1
r i22 ~ r= c sinhl; cosh^ ,
22 _
yg
121 112 }
^2
22
27
rL227 g
21
^221
+ 8
22
r 222
c2
c sin 5 cos \
ft
1
sin 2 ^
=
------r -25- = B
2 sinh2 I 1 + sin2 i f
r2
a@ -
-A
-B
(5)
v* = vQ/ g ^
u = v! V g
11
= vi g
v* = v = v2 y g 22 = v2 g
-V 4
'v 4
V1 = g
v2 = g
V4
V4
11
(6)
v
E x e r c i s e A -3 -1 :
For th e sta te of plane stress
crx x = 60 M Pa , aj j = - 2 0 M Pa , r
= 30 M Pa ,
Exercise A-3-2
67
with respect to the x -a x is . From the drawn circle the results can be read for the
given task :
T
T,'m ax
2a** 127 .
E x e rc is e A -3 -2 :
T h e stress com ponents of a th re e -d im e n s io n a l sta te of stress
are given.
D e te rm in e :
68
State o f stress
S o lu tio n :
With the invariants (3.13)
Ij = 100 + 60 + 10 = 170 [MPa
L
O
CM
100
l
O
CM
^2
60
60 0
0 10
10 0
+
0 100
100
l
O
CM
20/3^
60
10
= 48000 [MPa]
Or
= 80 y 6400 - 4800 =
40 MPa .
b ) By substituting the eigenvalues into the homogeneous equation system (3.11)
the orientations of the principal stresses follow. With Cj = 120 MPa this yields:
2 0 -fz n*
= 0,
= 0,
(100 - 120) n* +
(10 - 120) n* = 0 .
From this, it follows that
n* = 0
n* = C
n* = / F i
1
0
Exercise A-3-2
69
yields
II
-1 /2
0
V3/2'
, and in analogy one obtains n*j = / 3 / 2
1 /2
nn*lII _ 0
1
0
0
As the components of the normal vectors correspond to the cosine directions rela
tive to the initial system, we can draw the principal axes in the x , y , z - system
(Fig. A-7).
In order to check, we can carry out the transformation from the initial system to
the system of principal axes. With the rotation m atrix according to (3.10)
^
B =
-T
~i i
S' = B S B
100
0
0
2 0 / 3^
120
40
10
2 0 / 3^
60
0
3,
0
0
0
0
JII
With (3.24)
M = - j ( 100 + 60 + 1) = " X
MPa .
2 0 /iT
0
2 0 / 3
130
3
6 0 -* p
1 0 -f>
20/3^
2 0 /3
10
3
140
3 .
70
d)
State o f stress
_ 130
1 3
10
3
140
3
0,
9700 r
\2
'I 2 = 6400
[ M Pu | ,
170
+ 2 .
1330000
[ MP a f
27
In order to check, we directly calculate the invariants of the deviator from 'S.
Thus, we get for example
'i 3 = r s
140
3
^
20
2 0 /3 -
y^T
1330000
[ MP a f
~ 27
E x e r c i s e A -3 -3 :
For th e in fin ite strip w ith an elliptical hole ( see A -2 -3 ), establish th e equi
librium conditions in elliptical - hyperbolical coordinates.
S o lu tio n :
The equilibrium equations follow from ( 3.28a) :
<%
+ ( =
x ^
+ f1 = 0 ,
a. - 2:
x%
+ f2 = 0 .
Poc\
Poe a -yP U3 ary
\P~ T ,p+ TPyT + V l T
- = l:
+ x21|2 + f1 = 0
X11,! +
x 21, 2
.1
f
+ 3 A xn + 4 B X12 - A X22 + f1 = 0 .
Exercise A-4-1
71
Using the abbreviations A and B for the components of the m atrix of the
CHRISTOFFEL symbols from A - 2 - 3 , one obtains
( 3 sinh 2 t 1' +
+ 4 s i n 2 V 2 - sin h 2 5 1 t 22) + f* = 0 .
In analogy, one obtains for a. = 2:
E x e r c i s e A -4 -1 :
For a disk, which is clam ped along th e boundaries x = 0 , y = 0 as shown
in Fig. A-8 , and has th e opposite edges free, th e strain s have been deter
m ined from stra in gauge m easurem ents during th e action of an external
load. T h e results can be approxim ately described by
( k = constant factor)
x
a
Xy
72
4 State of strain
S o lu tio n :
a ) For the given strains the following strain-displacem ent relations (4-14) are
valid:
_ d \l
= k ( ^ x 2y + a y 2) ,
Exx ~ "dx
dv
, a 2
y y = d ^ = k b y x '
(i)
Integration of ( 1 ) yields
x y + ay x
a 3
/
\
, ai _1_ 3
, .
v ( x , y ) = k-r- y x + g ( x ) .
b 3
(2)
u(0,y) = 0
u ( x , 0) = 0
f(y) = o ,
2)
v (0,y) = 0
v(x,0)=0
g(x) = 0 .
(3)
" (M i
-i)
' a3 b + b2 a )
v(* L ) = v 1 y X
Va b /
33 b3
w a
with
u0 = v0 = k a b
(4b)
The displacements u and v at the free boundaries are presented in Fig. A-9.
u
u
v
1
v
E xercise A-4-2
73
d^ +
1 , 7
d3T = y k U
3
X
a 3 ,
+ 6 a x y + y )
(5 )
3 b ' a3
ab
ajjS/
N2
Ex x
dy2
Ey y
dx2
Txy
^xdy
'
I 1
dx dy
/ h a
a a\ I
- 2ak = 0 .
E x e r c i s e A -4 -2 :
A stra in gauge rosette as shown in Fig. A-10 has m easured th e following
strain s in th e directions 1 to 3
e2 ~ ~
e3 ~
eo
~ o
74
4 State of strain
S o lu tio n :
Suppose th a t the direction 1 lies under the yet unknown angle a against the prin
cipal axis I. Applying the transform ation form ula for stresses ( 3 .17), which is
analogous for strains, and taking into consideration th a t no shear strains occur
between the principal directions, the following is valid for the directions 1 and 3 :
El = y ( EI + EI l) + ^ ( EI eu ) c o s 2 -
( la)
e3
(lb)
= - i - ( I + n ) - i ( EI - EI l ) c o s 2 (x .
>
( EI EH ) cos
= h -H
(2)
S2 =
t ~ tt
E, -j- Eo
E1
En
tan 2 <x
2 2 - 1 - e3
t a n 2 a = ------------------- .
------> a = 30 .
From
EI + EII = 0
( EI ~ EII) cos 60 = 2 Eo
= 2 eo
eii
= - 2 eo -
E x e r c is e A -4 -3 :
In C a rte s ia n co o rd in a tes a d isp la c e m e n t vecto r is given as
X
u
V
y + 4z
4 / 2 x + 3z
D e te rm in e
a ) th e s tr a in te n so r X >
b ) th e p rin c ip a l s tra in s w ith re g a rd to m a g n itu d e a n d d ire c tio n ,
c ) th e v o lu m e d ila ta tio n .
Exercise A-4-3
75
S o lu tio n :
a ) The strain tensor \ follows from (4-13)
1_/ dji , dv \
2 dy
dx '
dv
dy
du
dx
V = k
sym.
dw\
dx
dv/\
dy
2 /2
2
3
V = k
2 /2
1 / du .
2 dz
1 / dv .
2 dz
dw
dz
v1x y = 0
yy = k
ezz = 3 k
Txz = 4 / 2 k ,
Y
1y z
= 4k
det(V s - X I) =
Ex x ~ X
2 Y'xy
Y
2 'x
Y
2 ' xy
Syy-X
Y
2 'y*
i-Y
Y
2 'y*
E -X
zz
2 /2 k
k-X
k-X
2k
2k
3k-X
2 /2
= 0.
X2 3 = (2 / l 3 " ) k ,
Ejj (2 -|- \ 13 ) k
Ejjj = (2 /13 ) k .
76
4 State of strain
0 a lx + 0 a ly + 2 -^2 k a Jz _ 0
a,lz = 0
0 ' a lx +
0 a ly +
2 k a lz = 0
2 / 2 k a lx + 2 k a ly +
2 k a jz = 0
a lx = C
arb itrary ,
-^2 a^x /2 C .
ly
Thus, it yields the eigenvector
1
aj = C
w ith arbitrary C 6 R
0
By analogy, the reader may calculate the two other eigenvectors and obtain
1
a2 - C
,
L
72
( l
a3 - C
1
'
k + (2 /is " ) k = 5 k .
E x e r c i s e A -4 -4 :
T he stra in - displacem ent relation for th e norm al stra in e22 and th e m ateri
al law of an isotropic, in fin ite disk w ith an elliptical hole are to be deter
m ined by using th e elliptical - hyperbolical coordinates from E x e rc ise
A -2-3.
S o lu tio n :
T he physical com ponents of the strain tensor ( tensor of second o rd e r) are first
calculated according to ( 2 .17) :
la p = y ap l/g(Q)g(/J/J) .
Inserting ( 5 ) and ( 6 ) from A - 2-3 yields the following for e22 :
Exercise A-4-4
= v2|2 ^
(1 )
*?
77
/?
, . ,2,1
. 2,2.1/2 1
c ( sinh c + sin t ) 'vv/)
1 . _ ,2
, . ,2 1
1 .
"2 "S
. 2 2.-1/2
- sin 2 \ c ( sinh c, + sm t, )
,1 , . , 2 , 1 , . 2,2 .-1/2
c(sm h !; + sin t, )
u
v
c2 ( sinh2 (} + sin2 f;2 )
. .2 ,1
. 2 ,2 ,-1 /2 1
= (sinh !; + sin !; )
1
, . ,2 ,1
. 2 2,1/2 d V
j- +
,5
1 . , _ , 1, . ,2,1
. 2,2,-1/2
+ sinh 2 5 ( sinh i; + sin \ )
u 1
,2 / . ,2 ,1
. 2,2 ,-1/2
sin 2 !; ( sinh i; + sin !; )
v
c ( sinh2 51 + sin2 2)
1
[/ sinh2 51 +
c (sinh2
+ sin2 !;2)
sin2 1;2
sinh 2
dv
sinh 2 I;1
^(sinh2!;1 + sin2 2)
(2 )
The material law for the state of plane stress without consideration of tempera
ture terms reads according to (5.5)
76
a /3 ~
with
^aP'yS
( 3a )
1 + V,
2 E ^go"f gp6 ^
^p
E goP
Insertion yields
76
1ap =
2 E ^gfv'yg/?^
gafig/3^)
E 8a^g7 JT
^ 22
^27625
&26 g 27 ) _ r&22&-yS
78
5 Constitutive laws
finally follows
f
22 =
822611
[~
1 + V,
2,1
I 22
2E (8 + g ) - 8 j x
1
4,
-ggt
V6
e22
22
22
and t
2 , 2.2 22
= p c ( slnh * + sin ? )
^22
11
22
- VI
11
11
(4 )
= i ( o 22 - vou ) .
22 -
E x e r c i s e A -5 -1 :
A square - shaped steel ingot is em bedded in a rigid concrete base and both
are s tre s s - f r e e at room tem p eratu re (Fig. A -ll). T he upper end of the in
got is free. T he ingot only is now subjected to a constant tem p eratu re in
crease 0 (coefficient of th erm al expansion a T ) .
a)
( l a )
- impeded strains
xx = yy = 0 ,
(lb)
Fig. A - l l : Steel ingot
Exercise A-5-1
79
( lc )
According to (5.1), the relations for the strains in component notation read
Exx = i ( x x
- v (yy + )1 + a T 0
( 2a )
Eyy = k [oyy
~ v(<5 + 0* * )]+ T
( 2b)
Ezz
V(xx
( 2C)
+ 0yy)l+ T
Inserting the conditions ( la,b ) into ( 2a,b) then yields the two equations
x x VOy y
" V 0 XX +
E a T 0
yy =
- E aT 0
( 3a )
( 3b )
ox x = ay y
--------
.
l v
( 4 )'
Ezz = i K x + yy) + T 0
=>
Ezz =
aT 0
(5 )
Exx
-^ yy
Ey y = %
Ezz
+ T 0
+ T 0 .
= -^ y y + T 0
Furthermore, no strain s
3yy
>
( 6a )
( 6b)
( 6c)
E Of.y 0
and from ( 6a ) or ( 6c )
Exx = (1 + v)otT 0 = szz .
The strain in the z - direction is now smaller than in a ).
80
Energy principles
E x e r c i s e A -6 -1 :
A B E R N O U L L I-E U L E R b eam w ith one end clam ped and th e o ther sim ply
supported, is subjected to a tran sv erse load q ( x ) p er u n it length of th e axis
of th e beam ( plan e beam b en d in g ) ( Fig. A-12).
T he differential equation and th e b o undary conditions for th e plane beam
are to be determ ined
a ) by m eans of th e principle of v irtu al displacem ents,
b ) by solution of th e corresponding v ariatio n al p rin cip le.
w(x)
8w(x)
F ig . A -1 2 : Clam ped, sim ply supported b eam und er uniform ly distributed
load
S o lu tio n :
a ) At first, we have to state the strain energy of the beam with d/dx ( ) :
U = - | j E I y w xx2dx = n i .
(la)
(lb)
(2)
Exercise A-6-1
81
1
811 = 8 { J [ 4 E I y w.xx2 - q ( x ) w ] dx J = o
0
1
(3)
8 1 1 = JI
(4)
E I y w , x x 8 w >xx - q^( vx ) )8 w
dx = 0 .
( E l y w ixx) x 8w
x= 0
11
+
E I y W, x x 8 w ,x x=0
(5)
= 0.
( 6 )
The boundary co n d itio n s follow from the boundary terms. In order that these
vanish, we must have at both x = 0 and x = I,
either
and
either
8w = 0
or
(' E I y w , x x )A x = 0^ z = 0 , 1
(7a)
8w
or
E Iyw ixx = - M y = 0 .
( 7b )
=0
x = 0
>
w = 0
= 0 .
(8a)
x = I
w = 0
(8b)
My ~ w .XX = 0 .
2i
(9)
E I yy W , x x 2 - q ( X ) W
A.
aw
( E I y w .xx).xx
- q(x ) = 0
( 10)
82
Energy principles
\ x2
/N L _/_dP_N
V dw
,xj
)x1
[ ( E I y W, x x ) , x S w ] x=0 = 0
[ E I y W, x x S w ,
x=0
= 0
( 11 )
= 0
[ ( E I y W, x x ) , x S w
= 0
E I y W, x x S w , x j x =; = 0 '
In ( 11 ) the boundary conditions ( 7a,b ) as well as ( 8a,b ) are included ( see dis
cussion in Section 6.6 ).
E x e rc is e A -6 -2 :
A linearly elastic body is subjected to a statio n ary tem p eratu re field 0 ( ) .
Set up th e variatio n al functional by HELLINGER and REISSNER for sm all
strains an d show th a t one obtains th e basic equations of linear th e rm o elasticity as necessary conditions from th e corresponding v ariational pro
blem. Volume, surface an d in e rtia forces are to be neglected.
S o lu tio n :
Substitution of the expression for strain energy U from ( 6.16a) into the variatio
nal functional ( 6.29 ) yields
nR =
PT l ! 0 + ^
( 4
+ Vjli) - 1 . j ] } dV
This functional contains altogether 15 unknown functions Yy . r'J , Vj. From the
problem of variation
8 n R = 8 J fiR dV = 0
V
A ilp = o
>
- = ( v | + v.l. )
^nR
2.
-z z - =
ij
ijkl
x = C
ij
Tk l - 13g
'ij
3-
| ^ f l | s= 0
1^ = 0
E xercise A-6-3
83
E x e r c i s e A -6 -3 :
S tructural com ponents in m echanical and civil engineering often have a
com plex shape. F u rtherm ore, they are subjected to m ultiple loads and th e ir
boundary conditions can be quite com plicated. S tarting from the energy
principles of th e theory of elasticity th e F in ite E le m e n t M ethod (FEM) was
developed for determ ining strain s and stresses of such com ponents, and
taking th e principle of v irtu al displacem ents as a basis, displacem ents are
introduced as unknow ns and can be derived from th e equilibrium condi
tions. We proceed in such a way th a t th e supporting fram ew ork is divided
into so -c a lle d fin ite elem ents ( bar, beam , disk, plate elem ents e tc .) w ith
approxim ate expressions chosen for th e displacem ent fields in each elem ent
( for details, see e.g. [ A .l, A.2, A.21 ]).
W ith the exam ple of a disk u n d er tem perature loads th e basic relations for
the m ethod should be determ ined. Therefore, as shown in Fig. A-13, a tria n
g u lar elem ent is taken out of th e dom ain of th e disk a n d is described in a
C artesian coordinate system .
At each of th e th ree nodes 1, 2, 3 of th e tria n g u la r disk elem ent two nodal
displacem ents are adm issible an d can be assem bled in a row vector as fol
lows :
dT = ( u 1, u 2 , u 3 , v 1, v 2 , v 3)T .
84
Energy principles
Solution:
a) The strains in the element - inherent system (upper index e ) are given for the
state of plane stress according to ( 4-12b):
(T ^ )e = y ( v Q|^ + v^|a )e .
In matrix form this relation reads
C = ( D v ) e ,
(la)
where
d
dx
0
yy
d
<^y
xy J
o
(lb)
<^y
_d_
dx
If we introduce the assumed linear displacement field with the six free values
a1, . . . , afi, we can write the following vector in column notation for the displace
ments :
II
with
If
(2a)
1 x y 0 0 0
0 0 0 1 x y
"5
b
Further, we obtain the vector of the nodal displacements as
(3a)
de = ( O a f
with
Cf
1
1
1
X1
X2
*1
X3
y3
y2
(3b)
i
Vi
X1
x2
X3
y3
y2
Exercise A-6-3
85
We now insert the relations ( 2a ) and ( 3 a ) into ( l a ) and obtain the following re
lation between strains and nodal displacem ents:
(4)
-po/M f
= E
[T7 5 - T a7 0
(5)
with
1- v
Bg o(fj> 0
The equilibrium conditions can now be derived from the principle of virtual dis
placements. Thus, according to ( 6.20 ), a body is in static equilibrium, if the virtu
al total potential is equal to z e ro :
s
n .
Here, ITj and II are the internal and external potential, respectively.
W ith the m aterial law ( 5 ) it follows from ( 6.12a) that the internal potential for a
therm ally loaded disk is given by
T
8 Ee-28
E e fl d A .
(6)
n> =
with
. T . e
T . e . e
y ( d ) K d + ( d ) k0
"2
= .(
EVdA
E e0 d A
(7)
(8)
86
7 Problem formulations
_ ,e -e
= K d
kg
- P )e = o
_e
+ ke .
After determ ination of th e nodal forces, we can now establish the equilibrium for
the entire domain. Therefore, the variation of th e virtual energy for the entire
structure has to vanish. This energy is obtained by sum m ation over the elements
( see [ A .l, A.2 , A.21 , C.2 5 ] ).
Rem ark-. I t should be emphasized th a t this example only serves to illustrate th e
way of developing a triangular finite element. Such an elem ent with
constant strain is only a very sim ple one. M ost softw are system s con
tain more accurate and improved elements.
E x e r c i s e A -7-1:
D eterm ine th e stresses in a hollow sphere ( outer radius a, inner rad iu s b )
un der constant internal pressure p from th e basic equations of th e theory of
elasticity.
S o lu tio n :
A spherical - sym m etrical state of stress is given, in which
dcp
= V= w = 0 ,
0 iptp
( 1)
0t/v
QQ ,1 iptp
^<pip
(2)
( 3 )
Exercise A-7-1
r r
E I rr
^iptp
1
E I ^iptp
87
2 V v>v> 1
(4 '
^ ( ^rr ^tp<p ^
^i9t? '
rr
2 \i a
tptp
( 5 )/
+ f ( rr "
) =0
( r2 r r ) , r = 2 l 0 v v '
or
( 6 )
With (5 ) and ( 6 ) we now have two differential equations for err and ov v . If we
replace ov v in ( 5 ) by means of ( 6 ), we obtain one differential equation for er r :
( 2 r r ) , r r - 2Orr = 0 -
( 7a )
( 70 )
the general solution of which can be determined with the aid of a power approach
orr = A rn
(8 )
n2 = - 3 .
. .
Ao
rr = Al +
(9)
> rr ( r = a ) = 0 '
( 10 )
88
7 Problem formulations
- P
Cv v
c&0
2 /a\3
+2
(H)
They are presented in Fig. A-14 for the case of a / b = 2. The maximum tensile
stress occurs at the inner surface and is given by
+2
( 12)
W max
rm
1 the approximations
(r)3
It now follows from (11) that for a thin - walled sphere the tangential stresses
are constant over the thickness, and from ( 12) we obtain
< p v 0 i9i9
Prm
21
Because of its importance in boiler design this relation is called the boiler f o r
mula.
Exercise A-7-2
89
E x e r c i s e A -7 -2 :
A concentrated force F acts upon a n elastic h a lf - s p a c e as shown in Fig.
A-15.
D eterm ine th e strain s and stresses w ith th e help of L O V E 's displacem ent
function x ( r ,z ).
S o lu tio n :
We first seek a LOVE's displacement function x ( r ,z ), which
a ) satisfies the bipotential equation (7.5)
AAx = 0 ,
(l)
( 2a )
i rz ( r , z = 0 ) = 0 .
( 2b )
= C j-J t 2 + z2 + C2 z In ( z + / r 2 +
) .
(3)
90
Problem formulations
1
1 - 2v
(C 1 + C2 ) i J - C I 5 ( _
(4)
1
1 - 2v
( 3 - 4 v ) C j + 2( 1 -
v)c2
(C 1 + C 2 ) z2
1 - 2v
R3
R(z+R) +
I - 2J ) r 3
1 - 32v ^ C1 + C2) R
2G L
[v( C1
^ - T1 -^ 2v R( z2+ R) J
1 +' C2
2 ' )R3
= - 2 G[( c. -
t^
- )
+ t4
Tr,
= - 2 G [ ( C1 - J Z 2 ^ )|T3 +
(5 )
z
t ( c . + c ! ) R5
r z2
. * . ( C1 + C2) R5
- l ~ 2v c
2v Cl -
- -r
zz
3G
z3
(6 )
1 v ( l - 2 v ) Rs '
r= 0
( 7)
r dr
( r 2 + z2 f t 2
r= 0
2
_ v( 1 - 2 v)
1 ~~
2ttG
which with the use of ( 4 ) and ( 5 ) leads to the final expressions for the displace
ments and stresses:
Exercise A-7-2
(,
2 tc ^
=
o ^
1
R(z+R)
_ 2v)
R3
3 z r2
Rs
91
(9a)
(9b)
R ( z + R)
_3F
2 n RS
( 9c )
_ J r rz
rz ~
2 tt R 5 '
( 9d )
( 10)
Thus, according to Fig. A-16, the resu lta n t stress at points of cut planes
const has the magnitude
/2
=
, 2
V z z + Trz
3F
3F
271 ( r 2 + z 2f
2 tt r 2 + z 2
cos p
Z =
(ii)
92
Problem formulations
Thus, according to (11 ), in the intersection points between any plane z = const
and the spherical surface, the magnitude of the resultant stress is the same and
given by [ A.16]
a
3F
2nd
(1 2 a)
w ( r , z = 0) =
(12b)
F .
Thus, the theory yields singular values for the displacements u and w at the point
of force application, and according to ( 9 ) similarly for the stresses. These singu
larities vanish if the force is distributed over a small area of the surface.
D is k s
8.1
d/dx = (),x
,
yy
(8.1)
d/dy = () , y .
LAPLACE operator due to ( 2.S9) ,
= $ (x ,y )
0 = 0 ( x , y )
V = V(x,y)
94
D isks
CTxx
* yy +
CTyy
xx
Txy
( 8.2)
$ yxy
eyy = V.y
(8.S)
7xy = U,y + V,
V(Jy y )
a T
(8.4)
ey y = i r K y ~ I/7*x) + T O 0 >
"Y
'xy
----- T
*y
or due to ( 5.1S)
E
1-
V [ eyy +
(8.5)
v e xx ~ t 1 + ^ ) T 0 ]
^ 7 xy
Txy
* * *
* rrrr + J
) . t
( $ ,T ~ 2 ^/Ttptp ) +
j .
Stresses by (7.13)
a u = r3- ^ '<pv
W + 7r $ ,r + V ( r . V ) ,
r <p
,T T
J_<
2
,ip
( r lP )
_ J_<f
r
,rv>
+ $ ,VVVV) =
- ( i - ) ( y + 7 v , r + ) v
).<p
~ j g ( * ,rr ~ 2 * , r V V ) +
- - E T ( 0 , r + T O0.r + ^ O 0 W ) -
% V >
, d/ dt p
= - ( JL #
\ r
,V ) , t
( 8.6)
8.2
A nalytical solutions
95
e<pV = Y v,v> + 7
( 8.8)
= T U'<p +
e<p<p -
1 ,
E ^
o.
(8.9)
1
G TrV
or
E
[
1-u2 [
avv -
E
[
1 I/2
Trip
8.2
( 8.10)
II
O
r*
s
a rr =
A n a ly tica l
e q u a tio n
so lu tio n s
to
the
hom ogeneous
b ip o te n tia l
a ) C a r te s i a n c o o r d i n a t e s
-
i k
# = ?i ?k aikX y
( 8.11)
96
8 Disks
$(x,y)
2
yy
Tx y
com m ent
ao2 y
2 a 02
an xy
2
a20X
-a u
2 a20
ao3 y3
6 a 03y
a3 0 X
a3 0 X
i + k > 4
P eriodic fu n c tio n s
A load is given as a periodic function along a boundary, or it varies peri
odically.
FO U R IE R expansion of a boundary load q ( x ) , 0 < x < /
q ( x ) = a 0 + 2 an cosQ Hx + 2 bn sinQ n x
n
n
w ith
= ~ ~ f~
1 r
a0 = T J q ( x ) d x
o
(812a)
(n = 1 , 2 , 3 , . . . ) ,
,
2 r
an = j - J q ( x ) c o s n x d x ,
* o
(8.12b)
2 rl
b = T J q ( x ) s i n a n x d x .
o
E xpansion of a stress function in FO U R IE R series in case of an odd func
tion
* ( x , y ) = 2 ^ n ( y ) sinanx
n
( 81S)
- ( d/dy = ,y ) .
(8.14)
97
Solutions to (8.14) :
*n(y)=
or
m=l
* n ( y ) = - ^ - ( A n c o s h a n y + Bn a n y c o s h a n y +
+ Cn sin h n y + Dn a n y sin h a n y ) .
(8.15)
N o n -p e r io d ic fu n c tio n s
Load described by th e FO U R IE R integral form ula
^ oo
q(x ) =
+ oo
J [ cos a x J q ( ( )
^ 0
-o o
-f-oo
oo
(8.16)
+ v / [ sinax
w 0
cos a ( d ] d a +
sin a d ] d a .
-o o
y = y i ( z - z ) ,
and
(8.17)
follow th e derivatives
4.
dx
dz
a
ay
.( d
HVdazz
-V
dz
_ _ a!_
ax2
a2
O ' " A_ A~
Sz2
dzdz
'
a2
d"z2
(8.18)
d \
a z ^/
d2 _
ay2
d2 . a2
az2 T \
s2
az2
A = - ^ + - ^
ax2
ay2
4
azaz
'
} (8.19)
98
D isks
w ith
= f , v
l + i/
= 3 - 4v
b)
- A x i s y m m e t r i c a l states of stress $ = ^ ( r )
D ifferential equation from (8.6) w ith
,rr rr
S olution:
+
#
f
,r r r
<? = C0 +
d/dr = (
+ 4 r 3^
j.2
rr
),r
, V = 0 , <9 = 0
=0.
(8.20)7
'
r 2 + C2 In ^ + C3 r 2 In
(8.21)
,T
a(fitfi
= ^
f TT
- R a d i u s - i n d e p e n d e n t states of stress
D ifferential equation from (8.6) w ith
<
Solution :
+ 4
=Ttfi0 .
(8 .22)'
'
$ = $( <p)
d / d ip = ( ),v
= 0 .
(8.23)
(8.24)
Stresses
=
-
* vv =
Stress function
TtV = - ( t
* ' V ).t -
(8
r iifi = 0
^ = r g ( y >) + h ( r ) ,
( 8.26a)
( 8-26b)
9.1
99
vv
=0
(8.28)
T lip = 0
iy = re '
2it?
2 G ( u + i v ) = ( - z tp' - ip + ni p) e
(8.29)
-it?
P la tes
9.1
direc
A plate is a stru ctu re like a disk w ith sm all thickness t in com parison w ith
o ther dim ensions. T h e plane w hich halves th e plate thickness is called the
m id -p la n e . As shown in Fig. 9.1 a ), th e plate is subjected to surface loads p
perpendicular to th e m id -p la n e . A n a rb itra ry load is resolved vertically
an d parallel to th e surface. T he in - p la n e forces can th e n be dealt w ith by
m eans of th e disk theory ( Ch. 8 ). T he in terest in this chapter is restricted
to th e influence of th e tran sv erse loading on th e plate. T h e thickness of the
plate is assum ed constant in th e following.
100
Plates
a ) P l a t e s i n C a r t e s i a n c o o r d in a t e s
- S h e a r -e la s tic , isotropic p la te
D isplacem ents of an arb itra ry point P at a d istance z from th e m id -p la n e
( cross - sections rem ain plane, see Fig. 9 .2):
u ( x , y , z ) = zi px ( x , y ) ,
v ( x ,y ,z) = z^y(x,y) ,
(9.1)
w ( x , y , z ) w ( x, y , z = 0)
w ith th e bending angles ipx an d ipy
S train - displacem ent
a / d y - ( )-y)
relations
from
(4.14)
w ith
XX= Z ^X,X>
eVyy
= zil>
,
Xy,y
V
r v . v l,
-7tlyz
vz
ez z = 0 ,
n/'zx vr #x + *x .
(9 .1 )
(d/dx=
( ) , x,
= il>
+ w,y
,
Yr y
v
.V
=J
M xx =
yy =
xxzdz
J* cryy z dz
Mxy = Myx
J r *y z d z
torsional m om ents
(9.3 a)
-t/2
+t/2
Qx
bending m om ents
-t/2
-t/2
+t/2
T XZ
dZ .
Q y =
-t/2
dZ
-t/2
T he sign convention consistent w ith ( 9.3a) for th e stress resu ltan ts is show n
in Fig. 9.1 b ).
D efinitions ( 9.3a) w ith th e m aterial law (5.12) an d (9.2) lead to th e stress
resu ltan t - deform ation relations
K (^x.x +
M yy = K v(ip
Yy <y + v ib
x ,x )/ >
l-i/
Mxy = My x = ^ - K ( ^ x + ^ r y) ,
(9.3b)
Qx = Gts(^x + W,x) Qy
= G t 8 (V<y +
Wi y )
K =
E t3
12 ( 1 - v l )
, th e sh ea r m o d u lu s G , a n d th e
101
E quilibrium conditions
Qx ,x + Qy,y + P
M xx,x +
= 0
M xy,y Q x =
My*,* + M yy^y
Qy
(9.4)
0 >
0 .
.X
+1 ih
TV.V
= lb
^ytx ib ,y
(9.5)
w.y = ^ y + C T r K -y T
KA< = P
- Aw =
A! ? - KS V =
(9.7)
l-i/ K
2
Gt
102
9 Plates
(9.8)
- A !? = 0
Vy = - W,y .
by
(9.10)
K K
X +
" W y y
( !
^ ) T l 0 ] .
Myy = - K[ w yy + v w xx + (1 + i / J o j , 1] ,
M xy =
K ( 1 -
(9.11)
" H x y
- (1 +
)aTK
(9.13)
103
B o u n d a ry co n d itio n s at a b o undary x = c o n s t:
-
Free boundary
Mxx = 0
or
Qx = Qx +
w (xx + v w yy + ( l + i / ) a T 10 = O ,
M xy,y =
( 9.14a)
1
1
r
W, x x x +
w here
( 2 -
" ) W,y y x +
( 1 +
" ) T
Q x = - K [ w xxx + ( 2 - v ) w yyx
>
x =
0 -
(9M b)
+ (1 + v ) a T l0 x ]
(9.14c)
or
w = 0
Mxx = 0
( 9.15a)
w =0
Aw = 0.
( 9.15b)
w x = 0 .
( 9.16)
C orner force
A = 2 Mxy = - 2 K ( l - i / ) w Ky
(9.17)
xxm ax
^2
o-
yymax
= 6 -^ ^
j2
xymax
= 6 ^f^.
|2
(9.18)
(9.20)
104
9 Plates
(9.21)
For an overall sim ply supported plate as exam ple, the n atu ral angular fre
quencies wmn are calculated from
,4
2
MWmn
K
R efer to th e s h e a r-rig id plate (9.14) to (9.16) for th e b o u n d a ry co n d itio n s
to (9.21).
-
S h e a r - r ig id , o rth o tro p ic p la te
M aterial law - stress resu ltan t - deform ation equations according to (9.11)
M XX = - K X( Vw , x x + I /yW , y y )/
M yy = - K y( vw yy - f i /x w , x x )/ ,
(9.22a)
Mxy = - 2 Kx y w #xy
w ith th e stiffnesses
K*
I 2 ( i - i / x i/y )
Ky
12(l - vx vY )
K*y
12
(g m )
rxxxx
+ 2 II w ,xxyy + K yw
,y y y y
= pF '
(9.23)
(9.24)/
b ) P la te s in p o la r c o o r d in a te s
- S h e a r-r ig id , iso tro p ic c irc u la r plates
(hr
1
r
w,x
i
,
Hf 2- w ,ipip )>
^ ^ HP
'
Wfi
= -K
Y w ,r +
( 1 -
Vr
>
^ 2 W,V V )
i/) K ( w
105
' Tv
f'W.rr +
"7
j.2
,<p 1
( 1 +
v ) T 1 0
(9.26)
v ) T ^
( A w ),r +
( 7 W,rv - ^2 W, v \ v +
(9.27a)
+ (1 + v ) a T ( 10 r cos ip - 10 v - ^
- k
(9.27b)
+ (1 + i / ) a T ( 10 r s i n v> - lG 'V E ^ V )] .
-
D ifferential equation for th e tim e - independent vibration m ode w ( r , tp) according to (9.21) [ B.8 , B.9 ]
( W,rr +
W, r + ^ W | W )
A4 W
A4 =
W ith
pLW
K
(9.28)
(9.29)
i
r
dR
dr
/ 2
VU
n^
r2
)/ Rn = 0
n
^7 ) ) Rn = 0
n2
(9.30a)
(9.30b)
P l a t e s in c u r v i l i n e a r c o o r d i n a t e s
E quilibrium conditions
= 0,
M ^ L - Q = 0 .
QL
+p
(9.31)
106
9 Plates
M aterial law
M0* = - K E*aPyS ( w L + a T a 4 !0 )
(9.32)
with
K
a.ap , a
1
CX
& (( )
tem p eratu re g ra d ie n t,
E *aPl/6 =
) + v a^ a?6
D ifferential equation
(9.13)
K A A w = p - a T (1 4- v ) K A *0
ni =
JJ {\ K[WL
+ K a T ( l + i / ) ( w xx + Wy y ) 1 j d x d y
(9.33)
ni
= J J { i K[ ( W,xx +
W,yy f -
+ K q t (1 + ^ ) ( w jXX + Wy y) 1 j d x d y .
Polar coordinates
K
W rr +
= ( ),r ,
2 t/ ( w W f W
d/ d<p
- w i I T W fr )
+ l1w 2
-
(d/dr
(9.34)
-f
w .Ifiifi w ,r -f 2irw r
2 /')|J J| rd co
dr .
r
^
E i = J J ( y KE*^
wL wU + O y K a ^ E ^ w ^ e l ^ d ^
107
A n a ly tic a l s o lu tio n s fo r s h e a r - r i g i d p la te s
S im p ly su p p o rted p la te s tr ip (
d/dy = 0
, d/dx
( )
AAw = w xxxx = ^
)
.
Pn
. n 7Tx
s in ------= y
^(nTr)4
Solution:
(9.36)
n 7TX
sin ( x ) = 2 P n si
for
w ith
pn = j -
J p ( x ) sin
dx
( n = 1,3,5,...) .
p (x,y)
A A w = w , x x x x +' 2 w #xxyy + w ,y y y y
V V
mwx
nw y
( m-n =
. Q
r
\
W
m7rx . n w y
p f x . y j ^ ^ Z j PmnSln
a sin b
'
m
E xpansion coefficients
(9.37)
Pmn
(9.38a)
21
2c
2d
fcr
p0(x>y)
2d
JC
F ig . 9.3: P late under a uniform ly distributed load over a rectan g u lar subdom ain
108
9 Plates
p(y) =
V
niry
Z P n s m b~
n
w ith
b
v+d
2 f
niry ,
2 f
niry ,
Pn = Y J Pos l n - b
dy = b"
Pos l n - b ^ dy =
n
vv-d
-d
2 b
(
n7r y \
= - b 7 T po l - cos b i
iV~fd
iV+d
,
Iv - d
P0 0 . n 7r v . n 7r d
= 2 -----2 sin r s i n r
nw
b
b
(
to o
\
(n = 1 , 2 , 3 , . . . ) .
v
...
w ith
T h e calculation yields
. m w u . mwc
j s m - sm
m ni
a
a
Po
Pmn =
16
n7rv . n 7rd
- r D
D
a n - j - s m
(m ,n = 1,2,3,...
H erew ith, we obtain w ith ( 9.38b)
\ \ '
Pmn
. m7rx . m r y
w = > > -------------------------------- =-s i n --------- s i n =-=*-
K K
)]
Tw o special cases:
F ull load
----- >
It follows t h a t :
^ Po W
w = t;
c = u = a /2 ,
pmn = 16
_
m
m ni
^
d = v
= b /2
an d
.
m 7r x .
n 7r y
(m ,n = 1
109
a \
I6 P0 a ,
K tt6 I 4
1
300
1 +
1
300
9 324
)
' /
s in -------- sin j
4 . ,
. m7ru . n w v
a
b
p
= t- 4 c d p . s in -------- sin r ------------------------ t------mn
ab
0
a
b
m ire n r d
a
b
si n/ cc
c - o
an d
= 1
sin/id
l i m ----- = 1
d -.o
KC
lim 4 c d p0 = F ,
c > 0
d 0
...
we obtain
-
4 F . m 7 r u . n 7 r v .
mn
= ts in -------- sin r
ab
a
b
i o q
(rn,n = 1 ,2,3,...)
v
Plates w ith two parallel, sim p ly supported boundaries and other bound
aries a rb itra ry
( 9M )
Wn ( y ),yyyy 2 ( ^
Homogeneous so lu tio n :
^ * ( Y ) , + ( ) ' wn ( y ) = ^
g ( Y) -
(Ml)
110
Plates
An overall sim ply supported p late w ith a yet unknow n m om ent d istri
bution MXXj
> solution W j.
2 An overall sim ply supported plate w ith a yet unknow n m om ent d istri
----- *solution w2 .
bution M
F rom th e geom etric boundary conditions - th e bending angles have to van
ish at th e supported boundaries in th e superposition, i.e.,
,
b
X. Y :
W 0 ,x +
W l,x +
W 2 ,x =
0 *
(MS)
vb,y + w i,y + w 2,y = 0 -
111
_ \
mi
m2
_ m 7T / H
a ] / K,
_ m 7r 4 / Kx
a ]/ Ky
m3
m 7 r 4 / Kx
a ] / Ky
'
T his type occurs w ith a crosswise reinforced concrete plate as shown in Fig.
9.5b.
112
9 Plates
2
i , 2 ,3,4
=
a
J i [ K
V
y Ky
= i 1
a 2 r
y Ky
P o la r co o rd in a tes
- A x is y m m e tr ic a l load case
T h e loads an d boundary conditions a re independent of <p * p = p ( r ),
w = w (r).
From (9.25) we obtain E U L E R 's differential equation
w ,rrrr + \
rrr ~ \
W , + ~
W ,r =
IY
(9.46)
Homogeneous solution:
wh = C0 + q r 2 + C2 ln { - + C3 r 2 l n ^
(9.47)
p ( r ,v3) g ( r ) ^ P n c o s n ip
( n in te g e r) .
(9.48)
w ( r ,</?) = ^ w n ( r ) c o s nip .
(9.49)
( 95 )
n = l:
(9.51a)
n>2:
(9.51b)
n =
10
C o u p l e d d i s k p l a te p r o b l e m s
10.1
I s o tr o p ic , p l a n e s t r u c t u r e s w ith la rg e d is p la c e m e n ts
113
- B a sic eq u a tio n s
In th e previous chapters we considered elastic structures w ith sm all dis
placem ents. T his sim plifying assum ption is not always fulfilled; especially
in cases of th in - w a lle d structures subjected to larger com pressive loads, th e
deform ations m ay becom e large com pared w ith th e thickness. T h e equilibri
um conditions m ust th e n be form ulated for th e deform ed state of th e stru c
tu re and term s of higher order m ust be taken into account in th e s tra in deform ation relations. T his corresponds to th e g eo m etrica l n o n -lin e a r ity .
Here, th e m aterial law is considered to be linear. F urtherm ore, th e lem m a
of m ass conservation ( g dV = g d V ) is assum ed to rem ain valid as w eh as
equality of th e volum e forces in th e deform ed an d undeform ed sta te ( f =
f ) . T h e s tr e s s - f r e e initial state (LA G R A N G E fo rm ulation) is tak en as a
basis. W ith these assum ptions, th e equilibrium conditions read as follows
[ B.l, B.2, B.4 ]:
[K
v i |k ) T j k ] | j
+ fi = 0 .
(10.1)
= 1 ( 4
+ 4
+ v k l i v k lj)
( l0 )
F ig. 10.1 : P la n e lo a d -b e a rin g stru ctu re under tem p eratu re and surface loads
114
(io.s)
0 U a ,C) = 0 ( O 4
with
e ( C ) = \ [ 0xu a ) + 2 U a ) ] ,
1&{ C ) = Y [ e i ( 0 - e a ( 0 ]
and an ex tern al surface load p ( ( a ). At its boundaries, it m ust satisfy pre
scribed bo undary conditions. In th e following, we w ill re strict ourselves to
th e s h e a r - r ig id p lane structure, but assum e large deform ations.
T reatm en t of th e problem by m eans of th e H E L L IN G E R -R E ISSN E R energy
functional:
-
+ -2
N /3 = J TPd (
Ma/3 = J r a p ( d ( .
_t
_t
( 10A )
V J / 3
Vp L
W ,a
W.p
'
( 10-5a)
= - wL
(10.5b)
+ ( \p
115
aj3
( 10-5c)
M aterial law
s - a T a7, <9) ,
M a/? = g - EaPyS ( \
(10.6a)
6 - a T a 7, *0 ) ,
(10.6b)
+ T gije ) ] d v
(*-7a)
= J {^
IccP ~ \
( 10'n )
N0^ N74 +
( 10.8 )
+ ^
Da/374M ^ M 74 + 2 a T N 0 + 2 a T M 1 ] ] d A .
116
W ith
^ap-yS = E ^ap-yt =
^^
n K = J J { y eCT/?T* L ( vJ p + vp L + w ,a w ,p ) +
+ f
C*^
B * * * w | ^ w |7, + K a T &ap E
|7/ 00
w |7,
(10.9a)
n n = J F ( ^ ; va . vJ ^ ; w . w , a . w U ; ^ U ) dA -
(10M)
( 10.10)
77e = - W = - J p w d A .
n = nK+ ne = nK- w .
( 10.11)
n = JJ { 1[
( v -y + T wi } + # -yy( u - + Y w- } "
* x y ( U,y +
V, x
W, x W y ) ]
+ K a TTv( 1 + i / /) (Vw , x x + w, yy )/ *0
4 ' $ ,2x x + S#yy
'2 - 2 v $ , x x $ , y y - 2** V
( 1* +1 1"/ )) $ 2
2E
4 q t
( ^ ,x x
^ ,y y ) 0@ } d x d y
JJ
p w d x dy
10.1
117
From th e sta tio n a rity condition 6 II = 0 ( see (6.20) ) in anology w ith ( 6.S5)
now, th e equilibrium condition (10.1Sa), th e com patibility condition (10.13b)
an d th e bo u n d ary conditions follow as :
VO N K A R M A N ' s d ifferen tial equations
K A A w = p - K a T ( l + i / ) A 10 + t 0 4 ( w , # )
AA $
(10.13a)
( 10.13b)
= - E a TA 0 - ^ - O 4(w ,w )
K wG = p - KM 1 + " )
*15
= - E a T e \l
le\l + 1^ wL * U -
- y a>1w ^ w |ap .
(10Ma)
( 10.14b)
'
f,xy,xy
g + 1 f ,yy,xx
g .
(10.15)
1
w = 0 ,
yy
0.
( 10.16a)
w x = 0 or
= 0
=0 .
(10.16b)
F ree boundary
M xx = 0
or
or
M vv x=x 0
M yy = 0
,* N x x w #x + N x y w /V +1 Q
=0
~x
_
, N xyw ,x + N
x'yy w,y + Q
v*y = 0 .
(10.16c)
118
/yy
(10.17)'
(10.18)
10.2
10.2
119
fibre orientation
For a plane stru c tu re m ade u p of several layers, we assum e a lin ear stre ssstra in behaviour as was done for th e previous isotropic plane structures. In
case of a com posite structure, however, th e stress curve exhibits c e rtain dis
continuities at th e boundaries betw een th e single la y e rs; here, th e stress rem id-plane
120
su ltan ts in th e single layers rem ain constant. For th e lam in ate itself they
depend on th e thickness coordinate
because th e com ponents of th e tensor
of elasticity differ from one layer to another. T h e stress resu ltan ts of the
lam in ate follow from th e equilibrium conditions by m eans of sum m ation of
th e stress resu ltan ts of all single layers [ B .ll ]:
N0"3 = 2 kNa/3 - A^ v
+ B a^
+ K'
\ v - N gf
- M g ,
(10.19)
,a3/33
w ith
> a.@\xv
k
yCcPfiv _ ^
aPiii/
= 2kB '
ga3/33 _ ^
ga3/33
( 10.20)
k
a /3
M'
For a physical in terp retatio n of th e relations betw een stress resultants and
strains we w rite (10.19) w ith (10.20) in an appropriate symbolic n o ta tio n :
A
B
0
' N '
M
Q
B
K
0
0
0
S
N,e
7
( 10.21)
M,
w ith
A 13
A 12
13
^22
-^23
33
K,
12
K.13
^22
^23
K.33
12
B 13
22
23
B 33
<A = Ai i) .
K
sym.
11
sym.
10.2
121
^11 ^12
sym. S.22
(S.. = S..),
v u
ji '
m t
= [ m u , m 22, m 12]
Q t = [ Q i .Q2]
0T
1 T
vector of strain s,
I ell >e22>e12J
r
vector of distortions,
7 ~~[ KU>K22K12J
7 T = [ 7 1, 7 2]
JE
w ith
COS CL,
sin a.
sin 2 a.
. 2
sin a,.
cos a.
- sin 2 c l
sin2ak
- sin 2 a,.
Er
1
U1 '2 'U2'1'
V Y 2' ^2*
1 - Vy 2 , V2, v
cos 2 cl
2 ' 1
(10.22b)
Ex,
1 vi'2'vr v
E lasticity m a trix of
a UD - layer (5.20) .
0
T he m aterial param eters can be determ ined by m eans of th e relations by
TSAI and HAHN [ B.10, B .ll ].
As (10.21) shows, disk and plate actions occur coupled in a plane stru ctu re
m ade of com posite m aterial. In addition, as a resu lt of th e transform ation,
th e single com ponents of th e stiffness m atrix depend on th e fibre angle c*k.
T he com ponents of th e subm atrices in (10.21) th erefore are based on th e
lam in ate design an d th e fibre orientation.
122
'l'a/3 =
(10.2S)
^ ^a P
w ith th e strain s
\ p
+ ^L )
an d th e disto rtio n s
\ af3 = \ ( i p a \p
+ ipp \a )
(10.24 )
1
(10.25)
w ith
an d
7 a = ipa + w | q
va
ip
bending angles ,
N0^
From (9.31)
Q| +
= 0
(10.26)
=o
(10.27)
B.2 E x er cises
E x e r c i s e B -8 -1 :
A sim p ly supported disk (le n g th I, height h, thickn ess t ) is subjected to a
constant load q per unit length as show n Fig. B -l.
a)
S o lu tio n :
a ) Based on ( 8.11) we consider the following power series assum ption for the
AIRY stress function:
fl> = a. x y
- constant shear
+ a20X
+ W
+ a2i x y
+ a23x2 y3
~ X X ( x 2 > y ) - y y ( y 3 ) ' Xx y ( X , y 2 )
+ a05y
x x ( y 3
(1 )
124
b)
Disks
0 XX <D*yy
( 2a)
( 2b )
- a n - 2 a21x - 6 a 23x y 2
( 2c )
yy
, XX
xxy = -
xy
yy
^ 3,20
= ~T :
yy( x - ^ ) = 0 :
,1
r
xy
( 3a )
= 0
( 3b)
6a03y+ 4 a23*2y
= ,
23
J +
T Z2ao5y3
U d 05
(3c)
( x , - -jj-) = 0 :
2
h
= :
= ~ jf
oa0 3 y ^ Y
= 0:
V ( X 2 )
"4 a23 F
2 a 20 a2 i h - i a23h3
xx( 2 y )
a2i b
I
3
2
> - a n - 2 a 21x - - a 23h x
= 0 .
(3d)
/xxyy
>
Boundary condition
a 23 =
au
( 4 )/
_ 5 a 05
a 20 =
From ( 3 a ) + ( 3 b )
From (3 d )
= 0
.
3
2
and a 21 a 23 h .
From ( 3a ) - ( 3b ) + ( 3d ) a 23 = - 9 - .
h t
From ( 3 a )
a 2i = _ | j L .
From( 4 )
Since ( 3 c ) cannot be fulfilled for all y, it is dem anded approxim ately th a t the re
su ltan t force F vanishes:
x h/2
Fx =
J 0xx( 4 , y ) t d y = .
-h/2
This condition is fulfilled for all arbitrary a 03, a 23 and a05.
(5)
Exercise B-8-1
125
(6)
= JD( 4 ,y)tydy=
- h/2
Y2
2
3
2 2
( 6 a 03y2 + - a23I2 y2 + 20a ogy4 ) t d y =
h/2
= t ( 2 a 03y3 + Y a 23/ 2 y3 + 4 a osy5 )
-h /2
*1 1
V3 , 1
72 V3 I
1 1 2 a 03
" * " 8 a 23 * "
+
03
h3t V 10
V5
4 a os a
= 0
4 J '
h2 1012
_ i , 2 L U - 4f y f
4
I2 ' h
' h '
f y ) a- l L - l .
Vh )
L*y
2 h
2 h
(7b)
\2
q/ t
f _
h V h / 2 h Z
(7a)
_ f i l i a l
hZVh/'
(7c)
Fig. B-2 shows the approximate stress distribution for some cuts of a disk with
1/ h = 1 and \ x/Z, tj = y / h .
126
c)
8 Disks
Checking the E q u ilib r iu m at Large
h/2
f Xx y ( | y ) dy - 1
- h /2
h/2
f xx y ( _ 4 y ) dy = 2 t
- h /2
f Xx y ( 4 y ) dy = qZ
- h /2
h/2
h/2
2t Ji^[lh2z3Zy2]dy=iH !h2zy_2Zy3)
- h /2
-h /2
d ) The strains can be calculated from the strain - displacement relations (8.3) and
from the m aterial law ( 8.4) for the state of plane stre s s:
xx = u ,x = 4 - v(o xx - v oyy') ,*
(v8a )J
( 8b )
yy
=v
,y
=4-(o
E V yy
f Xy = U, y+ VfX =
- v o ),
xx 7
( 8C )
^ t xy .
I2
/ h2
E^h [5 H i o
u,x
u = i k { ! 6( T j 4
2)
h3 ) ]
+ x J y " 4y v ( 2y - 2 h y - y J l
'
+ T ) ) , - 4 >3
(9a)
- v ( 2 y 3 - - | h 2y - y ) ] x + f ( y ) } ,
Eh3 1
,y
3,2 2
h
(9b)
_ v [ 3( i k ~
+ *2) y2 - y4] + e ( x )}
ik lK w 4
+) -
< *
-14]*+ ^ 1
Exercise B-8-1
127
From ( 8c ) follows :
~ T + t ) - 1 2 y 2x - v ( 6 y 2 - | h 2 ) x + ^ )
.
_ 6 v x y
dg(x)l
2(l+v)
q [ 3,2
6 x y
( 10 )
32
3 ,2
dg(x) df(y)
- - v h x - I x + v + ^
2
2
d x d y
= 0 .
, x
( 11 )
v /
Relation ( 11) can only be fulfilled if the term depending on y equals a co n stan t:
^
= Cl
f ( y ) = C l y + C2 .
g(x)
x4 + - | h 2 x 2 + - | v h 2 x 2 + - | / 2 x 2 - Cj X + C3 .
Eh3t U
h
10
I
x
-4r T
+ 3 ) y - 4 y' 3 -
(l2a)
/
3
3 .2
- v ( 2y - j l i y - i J l x
q J 1 4
3
V = Eh3 t l 2 y _ 4
2 2
- y x4 +
h3
[ / h2
- 2 y - Vl 3 U h 2 x2 +
+ Cjy+C,),
I2
2\ 2
41
T +X Jy - y j ( 12b )
v h2 x 2 + | l 2 x 2 - C 1x + C3 } .
T he following boundary conditions and conditions of sym m etry are employed for
determ ining the constants C ( i = 1 , 2 , 3 ) :
u ( 0 ,y )
= 0 ,
/ I
h ,
v( 2 - - 2)
= 0
t 1
v ( - 2 - 2 ^ = 0
From ( 13b,c )
>
Cj = 0 .
(1 3 a )
(1 3 b )
(1 3 c)
128
8 Disks
r
1
12
z 2
-4
L
3 . 2 2
h3
- 4h y T y " v
4 , / 6 , 3
~ 2 X + U
/ 3
3 2
h3 'I 1
2 \
3
+ X J y - 4y - v ( 2 y - y h y - T J xJ
\,2
2 ,3
+ 4V h X +
2/
y
h2
l l o
,2 2
z 2
-
,
+
, 4/ 3
2\
41
x ) - y 1
1
X _ h ^ 32 + 80 v ) ~
5 ,4
~32l
2 .2 / 3
" h 1 U
{l 4b )
\\
+ I6 V i '
E x e r c i s e B -8 -2 :
A circular, an n u lar disk as shown in Fig. B-3 is assum ed to be supported
frictionless at th e outer boundary an d is subjected to a constant, axisym m etric tem p eratu re distribution over th e thickness t
Exercise B-8-2
129
S o lu tio n :
We proceed from POISSON's differential equation (7.11):
A ' F = ( l + v ) a T0(r).
Due to the axisym m etric shape, boundary conditions and tem perature distribution
of the disk, the derivatives w ith respect to <p vanish. After m ultiplication by r 2 the
p artial differential equation is transform ed into an ordinary differential equation
of second order of EU LER type w ith variable coefficients,
+ v K ^ .
(1)
Yh = Ai lni + V
After choosing an assum ption for the right - hand - side type, the p articular solusolu
tion reads
ur _ 1 + v o l!.
P36
T a a4
(2b)
f = i , l n f + A! + i ^ T 0 , ^ .
(3)
By means of the m aterial law and the strain - displacement relations for the plane
case, the stresses are calculated as
E
1 d
1 + v r dr
E d2T
1 + v dr2
(4a)
1 + V
7 + | < 1 + ) T e . ( T ) T
(4b)
rr(b ) = 0 -
( 5a,b )
130
8 Disks
As ( 4a,b ) shows, only one constant value of Aj will occur in (5 ). Thus, one
boundary condition can be fulfilled a t th e outer or a t the inner boundary only. We
obtain an additional condition if we - by analogy to statics - superpose a 1
system onto the 0 - system (F ig . B-4). If we apply boundary condition ( 5 a ) , an
equal tensile stress p occurs a t the outer boundary ( 0 -sy ste m ). We now subject
the outer boundary of the disk to a com pression - p ( 1 - sy stem ) which has to
act opposite to + p in the 0 - system so th a t this boundary is not subjected to a
stress any longer.
According to (3 a )
According to A-7-1
J 1) _ o r
rr
l+ v
i C2
2 C 1 +
( 6a )
~T
+ (l+ v )
At the boundary r = a
At the boundary r = a
arr ^ ( a ) = const = p .
( 6b )
1) ( a ) = - P
m ust be valid.
In order to fulfill the boundary conditions ( 5 ) we superpose the radial stresses
and obtain
J ) +, 0J1)
_ -_
with
l+ v
A* = Aj - 1 ^ V C2
0r* +
+ (1+v)
C* = 2 C
(7)
l+ v
E
rr(b ) =
r r 0 ) ( b ) +
(8a)
(a) = p - p = 0 ,
rr ( a ) = !r } ( a ) +
^ ' ( h )
( 8b )
= 0 '
Eqs. ( 8a,b ) provide the equations for determining the constants A* und C*
i t - C* =
A* =
_ c* = - (1 + v )
with
T0a( 4 )
4 4
c* = (l+v)
t M
1 + P2 ) b2
+ p 2 + p4 )
p =: b / a .
The following expressions are obtained for the radial and the tangential stresses:
l E T e a {i + (p2 + p 4 ) [ i - ( 4 ) - ( 4 } ,
t 0 ^
1 + ( p 2 + P4 ) [ i + ( 7 )
] - 5 ( 4
(9a)
(9b)
Exercise B-8-3
131
Fig. B-5 presents the non-dim ensional stresses in dependence on the radius ratio
p . Without the need for further integration, the expansion u ( r ) can be determi
ned from (9a,b ), the second strain - displacement equation (8.8) for e , and the
material law (8.9):
u ( r )
v o rr) + r T 0
The reader is asked to check the results ( 9a,b ) by means of the solutions to the
bipotential equation (8.6) . For further examples of thermal stress problems refer
to [ A.13 , B.8J.
E x e r c i s e B -8 -3 :
A circular disk w ith constant thickness t rotates
w ith a constant an g u lar velocity w ( Fig. B - 6 ).
D eterm ine th e location and th e m agnitude of th e
m axim u m stresses for a full disk ( 0 4 r = a )
and for an a n n u la r disk ( b ^ r ^ a ) .
132
8 Disks
S o lu tio n :
During rotation a centrifugal force at a distance r
fr = p r o 2
( la )
Since
t, = ----
r
dr
( lb )
with
( 2a )
2 ( 1 - v ) p u 2 r4
( 2b )
------>
(3)
1*
rr
= T ,r +
^ , , C2 , 1 - V
V =
2 C 1 +
2C + C,
= 2 Cl + p
P U
+ v
g
= .ix+
8 P<J
V = 2Cl ~ ~ ? +
1 + 3v
C2
= 2 Ci
7 2
-------^
22
P
<4,)
r ,
22
3(1 v)
w
22
2 + 5
1
~
2 2
2P
22
r
(4b)
pu r
a ) In case of a so lid d isk the constant C2 m ust equal zero because of th e n o n existing singularity a t the origin.
From
orr ( a ) = 2 Cj - 3 ^ Vp <j2 a2 = 0
follows
2 C. = - ^ V p ij2 a 2
and thus
crr = b -V p <j2 ( a2 - r 2 ) ,
<PH> =
3 + v
2 21 + 3 v
g P a
g
( 5a )
22
P <J r .
/ _, ,
( 5b )
Exercise B-8-4
133
W m ax
/ r- \
p u a .
( 5c )
b ) For the case of an a n n u la r disk, the constants Cj und C2 follow from the
boundary conditions
rr(b ) = 0
>
rr(a ) =
0 -
( 6 )
These conditions yield a system of two linear equations for Cj and C2 Having
determined these, we obtain the stresses
3 + v
p w 2( a 2 + b2 8
- r2 ) ,
3 + v
2( 2 , ^2 , a2 b2
- p U (a + b
8
vv
1 + 3v 2^
r j.
( 7a )
{>
7u \
(7b)
a.
Wm*x
, S
( 7c )
r r ,i -
as
rrm ax
- p u2( 2
8
-
---- 5 "
rr
r-
- 2r ) = 0
> r* = / T b
/* \
3 + v
2/
,
r* ) = p u a - b ) .
>
E x e r c i s e B -8 -4 :
A q u a rte r - circle an n u la r disk (o u te r radius
a, in n e r radius b, thickness t ) according to
Fig. B-7 is clam ped a t A-A. A force F acts
in th e ra d ia l direction a t th e free end B-B
of th e disk.
D eterm ine th e stresses w ith in th e disk by
assum ing th a t th e norm al stresses in the
tangential direction vary proportionally
w ith s i n (p (a s is th e case in th e theory of a
curved b eam ). Discuss th e results.
F ig. B -7 : C lam ped q u a rte r-c irc le
a n n u la r disk subjected
to a single load
134
8 Disks
S o lu tio n :
T h e given q u a rte r-c irc le a n n u la r disk is subjected to a n o n - ax isy rn m etric sta te of
stress .
T h e b ip o te n tia l equation ( 8 . 6 ) in p o la r coordinates r , 9 ( 0 = 0 , V = 0 ) rea d s :
) +
(1 )
4 i_3 (' . r - 2
,tt p tp
) 4 44 (tv 4 itp
4 Itfilfiip
) = 0
> '1 ^4
Iip ip
w ''
iipipifiip
ip >
sin
= f ( r ) sin 9 .
9 , the following a s
(2)
A A = ( f 4 f f rrr -
+ ^ f r - ^ f ) sin
= 0 ,
+ 2 r 3 f ,r r r - 3 r" f + 3 r f r - 3 f = 0 .
(3 )
nj
^3
* ^4 I *
( 4a )
( 4b )
2C2
(V
+ r 1) sin 9
r3
o
,rr
c4\ .
2C2
( 6 C j r + i ++ -T.
) s 1isin
n 99 ,
r3
2C2
XTV>
r3
C4 )
Icos cp
t J
(5a)
( 5b)
(5c)
T h e b o u n d a r y c o n d i t i o n s a re :
rr ( a *V )
>
r r ( b >? ) =
( a > ) = 0
xrV> ( b >f ) = 0 -
( 6a )
( 6b)
Exercise B-8-4
, (
r > 0
t d r
135
b
We can now determine the constants C j , C2 and C4 by m eans of ( 6a,b ) and ( 7).
Substitution of ( 6a ) into ( 5a ) yields
C2
C4
2 C,l a - 2 -4- +
a = 0 ,
c
(8a)
(8b)
2 G, b - 2 i + -7- = 0 .
1
b3
b
From ( 8 a ,b ) we determine
l _
C,
2T2
a b
(9)
Jk v ( r ) tdr = vf _ ( 2Ci r ~
^ r
r
^ i )cos_otdr =
= - t ^ Cj r + 2"
, 2
Cj ( a
2
a2 - b2
b ) -j- C2tl~
2 a 2 ib 2
= F
r')
, a _ F
C* In.
4 b
( 10)
2 ,2
^C
9
"2
a o o 4T- C
^9
2 ,2
"2 a 92 _,29
a b
a b
C4 =
- ^C4 In
t)
_
t
a2 - b2
2C
2 a2 b2
( 11 )
2 C 2 (a2 - b2)
2,2
+
1
a b
ln b
2 , 2 r,
a b
C2 =
and, further.
C4 =
(12a)
2 - b2 - ( a2 + b2 ) l n ^
(a 2 + b 2 ) F
|a 2 - b 2 - (a 2 + b2 ) l n ^ j t
(12b)
136
8 Disks
ta
(13a)
sm cp ,
^2
a2
F
ta
3 - - ^
a
a2
a2
sm cp ,
(1 3 b )
cos cp
(13c)
( 1 + a 2) b l b
ta
i - 1 - ( 1 + T 2 ) lnF
For presenting the results, the stresses are normalized as F / ( b - a ) t . This deli
vers the following stress distributions where the abbreviation p = b / a is used :
o (a-b)t
=
a +P2( f )
=
P= - 1 + ( 1 + P = , , P
10
0.25
0.5
0.7 5
Fig. B-8 : Circumferential and radial stresses in the quarter - circle disk
Exercise B-8-5
137
V ( - b)'
V ( a_b)1
= ------- p------=
T he maximum stresses a
rrm
m axim um a t <p = 0 .
ax
,,
or c,
W m ttx
+p2( ? ) - ( 1 + p ) f
(? . 1+(1+(?)ln(>
TV
has its
Fig. B-8 presents the distribution of the norm al stresses over the cross section for
different ratios of the dimensionless radii p . If p increases,
approaches a
linear distribution corresponding to th a t of a straight beam.
E x e r c i s e B -8 -5 :
A sem i - in fin ite disk y > 0 , - o o < x < + o o ( thickness t ) is subjected to a
concentrated m om ent M at th e origin 0 as shown in Fig. B -9 a .
a ) D eterm ine th e stress function for th is load case by using a force couple
as shown in Fig. B-9b an d by applying th e stress function (ra d iatin g
stress sta te ) $ = ( F /x t ) r ip cos (p for a concentrated force F as
form ulated in [E T 2,A .16].
b ) W hich stresses occur in th e sem i - in fin ite disk?
__
y
F ig . B - 9 : Semi - in fin ite disk u n d er a concentrated m om ent M
Solution:
a ) In order to determine the stress function , we substitute the prescribed mo
m ent by a force couple as shown in Fig. B-9b. Then the stress function for F in
Oj can be w ritten as :
-<D(x - s , y ) .
138
8 Disks
If we superpose the above stress function by the corresponding stress function for
F in 0, i.e. ( x , y ) , we obtain
, = - ( x - s , y ) + ( x , y ) .
(l)
According to the rules of the differential calculus the partial derivative is defined
lim
-* 0
- ( x - E , y ) + ( x , y ) _ d ^
E
d' x
(2)
(3a)
1 = S , X
After transform ation into polar coordinates (c)/c)r ( ) r , d/ d
difference reads :
= e ( r cosq> - 9 V ^ - )
( ) ^ ) the
(3b)
r <p cos tp .
(4)
= J f * cos * F
\
v = ^ r ( cos v ~ v sm 9 '
yield by substitution into ( 3b ) the following stress function :
_
j =
With F e
Fe r
. 2
sm
cp
Fe r
cos
cp
(5 )
Exercise B-8-6
b)
139
rr = T l , r + ^ l , w
(6)
arr = i 2
1 ,<p<p
By
( cos2<p sin2 (P ) ] = ^ [ 1 - cos 2 ^ ] - i,w = ^ 2 s in 2 <p ,
1 ,p
2M
an = ^ s i n 2 cp
we obtain
i.e.,
or r > 0
tension
,
,
^
j
= zrifi = 0 ,
< <p <
t3r r < 0
compression
Fig. B-10 shows the distribution of the radial stresses in this semi - infinite disk.
E x e r c is e B -8 -6 :
A circular, a n n u la r disk m ade of carbon fibre reinforced plastic ( C FR P ) is
subjected to an outer tractio n po, an inner pressure p j , th e centrifugal forces
fr = pru>2, an d a tem p eratu re field 0 ( r ) . T he lam in ate is an a n g le -p ly
com posite consisting of two layers w ith a fibre angle a and th e to tal
thickness t ( t 1 = t 2 = t / 2 ) according to Fig. B -ll.
T he following is to be d e te rm in e d :
a ) the EU LER differential equation for th e corresponding disk problem ;
b ) th e distribution of th e ra d ia l a n d tan g en tial stresses in dependence of
th e fibre angle a a n d th e rad ial expansion of th e disk ;
c ) T he resu lts obtained shall be discussed on th e basis of th e following nu
m erical v a lu e s :
G eom etry:
O uter radiu s a = 0.25 m , in n er radius b = 0.0625 m , thickness t = 2 10 m
M a teria l c h a ra c te ristic s o f the la yer ( fibre T300, m a trix 914C)
( S tandards see Section 5.4 ) :
E j, = 132 680 M Pa ,
vw
>
E2, = 9059 M Pa ,
=
6 kg / mm,3 ,
G r 2 , = 4268 M Pa ,
<
p = 0.60 ,
140
8 Disks
w ith
= 1 C .
Solution :
a ) According to (3.29b), the equilibrium conditions for a circular disk can be ex
pressed as follows in polar coordinates:
r r + r t 5n - , r +
r V , r + tW
+ 2V
(la)
+ F fr =
+ rf*
= 0.
(lb)
All loads are axisymmetric so that both the stresses and the displacements are
independent of the circumferential coordinate <p . Thus, all derivatives with respect
to tp vanish in ( la,b ). Furthermore,
y IV>, and v are equal zero. The simpli
fied equilibrium condition ( l a ) then reads :
(2 )
rr+ r0n , r ev v + r fr = 0
In a similar manner, the strain - displacement equations (8.8) for the axisymmetrical load case can be written as :
zwp
( 3a,b )
r '
Finally, we have to determine the material law for an a n g le-p ly lam inate. This
is achieved by rotating a unidirectional layer ( UD - layer ) ( see Section 5.4 ) . For
pure disk action, we have according to ( 10.21) :
Nr r '
^11
=
-^12
A22
sym.
^13
A 23
^3 3
a Tr
' Er r '
E ifiip
- f r(C> -
Tv-
TrV -
( 4a )
Exercise B-8-6
141
The elasticity matrix in ( 4a ) is fully occupied and its components depend on the
angle of rotation. By special choice of the fibre orientations in a m ulti-layered
composite ( see Fig. B -ll) and constant uniform layer thicknesses symmetrical
with respect to the m id-plane, one obtains a simplified material law for the state
of plane stress. We can prove for a k = a that A13 = A23 = 0 , and thus
( 4 a ) becomes with the assumptions made,
0
rr
__
E 11 ( a )
Ei2 ( a )
E 2l ( a )
rr
( O
Tr
(4b)
*TV
E v>v>
D11 (
rr
Tr
+
D 2l ( a )
(p<p
D12 (
D 22( a )
ipffi
E u (a)
D22 ( )
A()
( 4c )
tv
( y , 8 = 1 , 2 ) . These
Ei2 ()
E 12 ( a ) ~ E 21 A ( )
>
4- r
U,r = d ? ( r ^ ) = 'tptp
r %>>*
( 0 21
0 12 ) 0 ^ + ( Tv. - Tr ) 0 +
D n ) rr + ( 0 22
( 5a)
+ r ( Di2 rr,r
We substitute from ( 2 ) into ( 5a )
142
8 Disks
3 ^22 r rr,r +
rr ,rr +
( ^ 2 2 ~ 11 ) rr =
(5b)
2
(r),r = ~ j ~ (
)3
r r , r r + 3 b )2 2 r 0 r r ,r + ( ^ 2 2 ~ ^11 ) rr =
(5c)
=
( a T r _ 5 a T ifi ) 0 ( "b" )
"*" ( 12 ~ 3 ^ 2 2 ) P
10
(6)
rrh = C r " .
which yields the characteristic equation
X + 2X + (1 with the roots
Dr
Xj 1 + -|/ ^
X2 1
22
pD11
is the degree of anisotropy.
Thus, the homogeneous solution is
rrh = V * 1 + C2rA2
(7 )
b a Tv>) 0
r \4
(*>
(^12 3 ^22 ) p r2 o
( 9 D22 D jj )
(8 )
Exercise B-8-6
143
( 10a )
rr ( a ) = Po '
( 10b )
IY
( 9 D22D22 ~ Dll
D1:1))
(
b - - ( 2( 255 1D22^ -D
d uj )
( 0 12 3 0 22 ) P w
( 9 D ,, - D )
VP ~
C2 = A { - [ P i - 4 ^
( a Tr - 3 i y ) 0 ^ )
(25 D - D )
3 D ^ )pt02 b2 - ( ^
( 9 0 22 Du )
, [
( 0 12 ~ 3 0 22 ) P 10
+ l P
w here
( 9 D22- D u )
>
5 gT ^ ) e o | aA1 +
(25D22
2
T? '
Du )
( a Tr ~ 3 a TV ) 0 a4 ]
a " ( 2 5 n B - D 11)
? J a 1
A =
bll>>A2[ ( F ) Al- ( t ) 1
Introducing the abbreviation p0 = a / b, the complete solution for the radial
stresses reads :
2
( 0 12 ~ 3 0 22 ) P 10 / r \ 2
( 9D2 2 - Di i )
(H )
( a Tr _ 3 a T ^ ) 0 ( 1
(25D22- D u )
V^
>2
Bi = ^ r 7 r l p i p0
Po Po
(,
A2 )f
-P o + U -P o
)[
+
D _
*1
, (,
2
(D 1 2 - 3 D 2 2 )p io
( 99 D
D 2_
,, - D ,, )
V 22
11
( a Tr - 5 Tb ) i
(25 D22 fDne )r *
U ) l]'
A i ^ f ( D1 2 - 3 D 2 2 ) P
2 _ PoAl- P o 2 ^ Pi _ P o + ^ 1 _ P ^
( 9 D 22- D u )( T r ~ 5 g Ty> ) 0
+ (25D3J-DU)
2 ,
" +
(
a, \ 4 1 1
U ^ J ] -
A fter d eterm in atio n of orr , th e rad ial stresses can b e calculated from ( 2 ) :
o tptfi =
r r + r 0 rr ,r + P r 10 '
( 12 )
144
8 Disks
u(r= a)
[mm]
0.030
0.025
0.020
u
a JV d2
2
30 0.470
II
45 1.000
IV
V
oOo
III 60 2.129
3.829
Exercise B-8-7
145
Fig. B-12 shows that the radial deflection depends on the angle a ; the smallest
value occurs at a* = 60". The circumferential stresses in dependence of the de
gree of anisotropy -J Dn / D22 are presented in Fig. B-13. Here, it is obvious that
the circumferential stresses are approximately constant for -J Dn / D22 ~ 2,0 so
that the high stresses at the inner and the outer boundary respectively can be re
duced together.
E x e r c is e B -8 -7 :
A n in fin ite disk possesses an elliptical hole. T h e disk is subjected to a uni
ax ia l tensile stress cr0 , see Fig. B-14. No ex tern al loads ac t a t th e boundary
of th e ellipse.
C alculate th e stresses in th e disk by th e com plex solution m ethod. For this
purpose, use th e conform al m apping
z = f(() =
...
w ith
a +b
c = -
2
C(c +
,
a -b
m = p
a+ b
0 <m < 1
F ig. B -1 4 : In fin ite disk w ith an elliptical hole u n d er u n iax ial tension a 0
S o lu tio n :
C om m ents on conform al mapping
The equation z = f ( ) assigns any point z = x + iy = r e 1?9of the complex z domain to a point ^ = { + ii) = R e 1 0 of the complex
domain (see Fig. B-15).
For further details on conformal mapping refer to, e.g., [A.8, B.6 ].
146
8 Disks
( 1)
maps the area outside the u n it- 1 -circle in the - domain into the area outside
an ellipse in the z-dom ain. Thus, the circle with radius 1 is transformed into an
ellipse.
The equation for the circle with radius 1 in the
C = e, e
domain reads :
( R = 1) .
(2 )
i
i -i
/.
\ e e
+ i ( l - m ) -------------
= c[ ( 1 + m ) cos 0 + i ( l m ) sin 0 ]
or
(3 )
x = c ( l + m) c os 0 ,
(4a)
y = c(l - m )sin0.
(4b)
2
Eq.
( 3 )and (
4 ) are
the
For
0 = 0 we
obtain
the
Exercise B-8-7
147
(5)
yy = >yy
0
yy - a
xx
+1 2 i x x y = , x x - , yy
2 G( u + iv) = - z
where x =
(6a)
+ <P ) .
2
(<p + + ' )
(6b)
( 6c )
1+v
Boundary conditions
a ) Boundary of the hole
The boundary conditions at the boundary of the hole result from the fact that the
load vector must equal the stress vector t at the boundary.
In accordance with Fig. B-16 we formulate the equilibrium conditions considering
the + /- sign of dx :
x-direction :
*
^x
l ^ d s t = x x ^ 1 +
dy
-rds -
dx
dx
dy
yx
ds = ....
'yy -rds + *y
y-direction :
tv d s t = Oyy( - d x ) t + Txyd
yt
xy
v =
dx
dy
dx
_
dy
.
a
i - t = - ------
yy ds
*y ds
-xx ds
-xy ds
148
8 Disks
t = tX + i t y = f ( C,. y - i X) =
=>
Since the choice of constants is not yet limited, we can set them zero :
<y
,x
.7 d
\ dz
d
oz
i( -
d
dz
d \
r = )
dz /
. d
dz
- 2 l ^ =
(7)
Using the chain rule we transform the formulas for the stresses and displacements
into functions of :
z = f(C)
<p ( z ) = <p
<p(z)
p[ [ff(( C)]
c)]
=>
=<P1 ( C ) = <Pi.
4j ( z ) = * [ f ( 0 ]
dip
dz
d2 tp
dz2
d<Pj dC
<
t ffl
- d (
~ dO
dip
dz
<Pl ( 0
dz
' Vi'}
. f' J dz
f l3
~ {'
= 2 '
( ! )
(te)
y y - x x + 2 i x xy =
2G (u + iv)
= -
- (Jjj + x
( 8c )
Pi + J TVi +
+1
= 0
( 9a)
f=c(c + ^ )
f' = c ( l -
( 9b )
,
f" = 2 c m ^ .
( 10)
D e t e r m i n a t i o n o f th e s t r e s s f u n c t i o n b y m e a n s o f L A U R E N T - s e r i e s
In the following, power series shall be established for the functions <Pj und i]q
[B.6 ]. Since the stresses in equations ( 8a ) and ( 8b ) m ust converge at infinity, no
term with positive exponents are allowed to occur in the series. F irst, the follow
ing L A U R E N T -series are set up for the derivatives :
Exercise B-8-7
oo
oo
n= 0
n= 0
149
( 11)
o B
<K = ? 0 C + B1l n C + ^ - Z ^ T
(12b)
+ B-
n= 2
(12c)
n 0
The imaginary part of A Q only represents a rigid body displacement of the disk
and is therefore set zero, i.e.
- 2i V 1
2 i
n= 1
2 -4i
+ m e
3 6i
+ m e
150
8 Disks
.
cpj
A0
Aj
-i
-2i
A2 e
2 -,
+ A3 e
-3i
+ . -. ,
2 \
-i
-3i
Pi =
-i
A0 e
- 1 A , 0 - A2 e
^ B o e '+ iB jO -B .e
A3
--x -e
2i
A4
-2i
3i
~ 5 e - - >
4
- e
- 3 i
Now, all of the above equations are substituted into ( 9b ) and rearranged with re
spect to the e - functions. After that we get
0
B, = 0
B2 = - ^ ( 1 - m + m2 )>
= 0
B;2 n
cop
= 2 ^ ( l + m2 ) ( m - l ) ( 2 n - l ) m 2
2
Bo2 n
+ l
for
for
We thus obtain
n > 3 ,
for
n>2
n > 0 .
(1 3 a)
l + (m -2)-U ,
P i
K =
+ ( l - m + m2 ) - ^ +
n 2
+ (1 + 4
n=2
. I - I B .
rr + ^ Re ( ff> )
^ Re
xx + yy
_
-
0Re
C2 + m - 2
_
r2
/ Ty2
2i
- P r (R e
0
I 'D 2
(Re
0R e
E ----
K1- ^ )
R2 e2i + m - 2
t >2 2i
Ri
m
n W yy2
2i
+ m - 2)(R e
- m) _
- j 2 2 i
~
- m j \ (z R
e
- m)\
2i
(13b)
Exercise B-8-8
4
2 -2i
__i T > R + m R e
2 2i
- mR e
- m + 2m
t>2 2i .
R - m Rr >2e - 2i - m R
e
+ m2
= n
2 -2i
- 2R e
151
^4
R4 + 2 m - m 2 - 2 R2 cos 2 0
i ------------------------- 5-----
R - 2 m cos 2 0 + m
At the boundary of the hole R = 1, we have atT 0 and therefore the stress
becomes
1 + 2 m - m2 - 2 cos 2 0
1 - 2 m cos 2 0 + m 2
In the special case of a circular hole we have a = b and, correspondingly, m = 0 .
For 0 = n / 2 we obtain the maximum stress
m ax = 3 0 0 -
E x e r c i s e B -8 -8 :
An in fin ite disk subjected to a u n iax ial tensile stress <x0 possesses a crack
w ith th e len g th 2a (Fig. B-17). T h e state of stress w ith in th e disk is de
scribed by th e com plex stress functions
a o z + <7o a
f )
- 1
fP =
y,
w ith z = x + i y .
Check w hether th e corresponding stresses ful
fill all bou n d ary conditions
k -2 a !
152
8 Disks
S o lu tio n :
In order to apply equations (8.19) , we calculate the derivatives
,
0 , 0
=T
<P
*
= ----2a
(la)
- 1
(lb)
T/2
KtT
(1c)
3/2 '
m
With (la ,b ,c ), the stresses follow from (8.19)
ax x +1 ayy
= - o + 0
(2a)
M
_Z_
_ Z _
a
yy -
x x + 2 i T xy =
0 +
3/2
(2b)
i m - '
In order to calculate the stresses themselves, the rig h t-h an d sides must be separ
ated into their real and imaginary parts. For checking the boundary conditions the
following considerations of the limit values are sufficient.
1) For large z und , i.e. the outer boundaries, holds
o x x +1 ayy
(3a)
yy xx + 2 i t xy = 0 >
(3b)
Xx =
0 > yy = o
> Txy = 0
yy -
- 0 +
(4a)
2 0
Ml ) - 1
yy -
xx +
2 i T xy =
(4b)
= 0 -
The above is valid both for the non - loaded crack surfaces I -5-1 < 1 and along the
axis I IT I > L
Exercise B-9-1
153
b)
< 1 ,
_x
a
M - i - i
The stress cxx along the crack remains constant. At the crack tips x = a, a sin
gularity occurs, and for large x 0XX approaches zero (lo a d -fre e outer boundary).
)
yy =
_x
a
for
i-
<i
>1.
Thus, the stresses derivable from the functions given in the problem formulation
fulfill all boundary conditions.
E x e r c i s e B -9 -1 :
An isotropic, s h e a r - r ig id rectan g u lar plate (dim ensions a , b , thickness t ) is
sim ply supported a t x = 0 an d x = a , clam ped at y = 0 , and free a t y = b
(Fig. B-18). T h e plate is subjected to a transverse, tria n g u lar load
p ( y ) = P0 ( 1 _ f - )
D eterm ine th e com plete solution to th e plate equation, and form ulate the
boundary conditions.
Po
154
9 Plates
S olu tion :
The differential equation (9.13) ( *0 = 0 )
K M w = Po( l - [ )
(1)
n+ B*
a i ^ c o s h ^a
+ D*n ^ a
( see(9.42) )
, .
(2 )
sinh ^ a) s' i n a^ .
This solution fulfills both the homogeneous differential equation and the boundary
conditions at the boundaries x = const:
w ( 0 , y ) = Mxx ( 0 , y ) = w ( a , y ) = Mxx ( a , y ) = 0 .
In order to determine the particular solution, the load in z-direction is expressed
as a FOURIER series as discussed in the beginning of Section
9.2,
P ( x , y ) = p0 ( l - ^ ) = ( l - | - ) ^ p n s i n ^
(3a)
with
2 f
. nitx ,
4P 0
Pn = a J po s l n ^ - dx = i n r
0
1 c
* n = l,3,5...
(3b)
<*>
( ) * . ( - f ) - & ( - f )
' . - < -
<*>
Wh + Wp = - ^ 2 [ ( l - f ) ( ^ ) 5 + An COSh^ + Bn ^
cosh
+ Cn sinh I ?
+ Dn ^
sinh ^
. n 7TX
sin -
(6 )
Owing to the sine-expansion, the complete solution fulfills the boundary condi
tions at x = const, too.
For the determination of the 4 x n unknowns An, . . . , D , we have two boundary
conditions for each of the boundaries y = const., and these apply for each value
of n = >
for
y = 0 acc. to (9.16)
for
>
M
w( x, 0) = 0,
w (x, 0) = 0,
(7a)
( x, b) = 0,
Q ( x, b) = 0.
(7b)
Exercise B-9-2
155
The stress resultants are obtained from (9.11) according to ( 7b ) or, alternatively,
in analogy with (9.14c) with d/dx = ( ) x , d/dy ( ) y:
M y y = ~ K ( w, y y +
v w ,xx) >
Qy = - K ( w yyy + ( 2 - v ) w xxy) .
Exercise B -9 -2 :
A sim ply supported sem i - in fin ite
plate strip ( width a, thickness t )
as show n in Fig. B-19 is subjected
to a load in form of a uniform ly
distributed m om ent along th e
boundary y = 0. T h e m om ent per
u n it len g th of th e b o undary is
denoted by M 0.
D eterm ine th e influence of th e
plate sh ear stiffness on th e deflec
tion.
F ig. B -1 9 : Semi - inf in ite plate strip subjected
to boundary m om ent M0
S o lu tio n :
We now deal with a shear -e la s tic plate, and refer to the basic equations (9.7).
Since no surface load acts, the basic equations reduce to
K A <D = 0
A w = - <D ,
Y - AT = 0 ,
(1 )
( 2a )
V -
( 2b )
2 g n ( y ) cos otn x ,
w = Jh
n
(y)sina x ,
( 2c )
11
where a n = n it/a . These approximations take into account the simple support at
the boundaries x = const. The circular functions are now separated from ( 1 ) , and
for each n one obtains a coupled system of ordinary differential equations with
constant coefficients, and with the abbreviations d / d x = ( ) x , d/d y = ( ) it
follows in this case
156
9 Plates
fn , y y - n fn
= <
b n ,yy n^n
(3a)
(3b)
>
(3c)
(4a)
fn = Cn e
(4b)
h = ( A + ^ L y ) e- ay
n
' n
2a
with
(4c)
= XB + n >
where only the decaying parts of the semi - infinite plate strip are of interest.
We now expand the boundary moment M0 as a FOURIER series
Mn
v =
sin
n a x .
(v5 ')
" w . y - v w ,xx + ^ ( . y y
v .x x ) +
+ ^ ( ^ x y - v T xy) =
= K { [ - n An + V n An +
+ VV ) ] e' nX +
(6a)
V n ) Cn e_any + ^ Xn n ( 1 V) Dn e ^
} sin a
K { - 2 w (Xy + K. [ 2 <D(Xy + I z v
(v T xx - Y - y y )' J J =
(6b)
An = 0 ,
Mn
K
(7a)
Cn [ 1 + c l > ( l - v ) ] +
+ ^ Xn ( l - ' ' ) D n ,
(7b)
E xercise B-9-3
Cn
_K
-2^C n+ ^ ( - a ^ - X ^ ) D n =0.
2 + Gt,
Mx y ( x , 0 ) = 0
157
( 7c)
and
_1_
1-v
E t3
12 (1 - v )
2 (l + v) 6 _ t 2
5t
10
yields
C =
M n A
= TT n
K
1+ 2:
(8)
i 7 7 2 - [ ( 1 - v K X + f t2 X
n + ^n 1
(9)
2K n
_ n _
lo t
10
10
t* + V
V lot )
*2
woh
1
0.98
1.02 .
Thus, the effect of taking into account the finite shear stiffness in our plate ex
am ple w ith a = 10 7r t , is th a t the deflection increases by only 2% in com parison
w ith th a t of the sh e a r-rig id plate.
Hence, from an engineering point of view, K IR CH H O FF's plate theory generally
provides sufficiently precise solutions for thin plates ( in the present exam ple t / a
= 1 /lO n 1 / 3 0 ) .
T he influence of the fast decaying p a rt of the solution ( Xn a n ) on the resultant
forces is discussed comprehensively in [B.5].
E xercise B -9 -3 :
A re c ta n g u la r p la te o f c o n s ta n t th ic k n e ss t ( le n g th a, w id th 2b, v = 0,3) is
sim p ly s u p p o rte d a t th e b o u n d a rie s x = 0 , a . A s sh o w n in Fig. B-20, tw o
tu b e s w ith th e s a m e b en d in g stiffn ess E Iy a re w elded to th e p la te alo n g th e
b o u n d a rie s y = b . T h e p la te is su b je cted to a te m p e r a tu re g ra d ie n t
th ro u g h th e p la te th ic k n ess. T h e te m p e r a tu re g ra d ie n t d ep en d s on y, a n d is
g iven b y:
1 ( y ) = 0 1 ( l - m | - ) ,
w h ere
T - T
j = h__ 1. = const
and
0 < m < 1 .
158
9 Plates
T h e tem peratu res of both tubes are equal to th e tem p eratu re of th e plate
m id -p la n e which is everyw here equal to ( T j + T 2 )/2. ( T] and T 2 denote
th e tem peratures of th e plate surfaces along th e x - a x i s ).
a ) Derive th e differential equation for th e deflection of th e plate under the
given tem p eratu re field and boundary conditions by m eans of an energy
principle. T h e tubes can be assum ed to be rigid w ith respect to torsion.
b ) Set up th e general solution of th e differential equations and th e system
of equations for determ ining th e integration constants.
c ) Discuss th e influence of th e elastic su p p o rt from th e two tubes ( p a ra
m eter X = ( E I y ) / ( a K ) , 0 < A < o o ) on th e deflection and th e stresses
of a plate w ith th e ratio a / b = 1 and l6 = 6 X = const ( m = 0 ) .
Vt
b
Section at y = 0 :
F ig. B -2 0 : R ectangular plate w ith two sim ply supported and two elastically
supported boundaries subjected to a tem p eratu re grad ien t 10 ( y )
through its thickness
S o lu tio n :
a)
Exercise B-9-3
159
+ b
U pl
J
J [v
y = -b x = 0
/XX
w, y y - w
,xy
2 J) +1
(i)
+ 2 ( 1 + v ) a T 10 ( w (XX + w>yy) ] d x d y
with the notations d/ dx = ( ) x , d/dy = ( )
2E
JL(,]
x= o
Considering My ( x ) = - E I , w ( x , b ) xx yields:
E IV
T1 = __ ?- J w ( x , b ) x dx
(2)
J
r =
j
- b
X =
+ 2 ( 1 + V) 0T 30 ( w xx + w>yy )] dx dy +
a
Ely
H
T
J w(x-b)l dx
J w(x,-b)*
dx
x = o
(3)
with
F(x,y) =
f f w 'XX +1 w ryy /^ - 2 (v l - v ) 'V
fw
2 1 ''
.xx
w , y y - w , xvy 2 /) +
+ 2 ( 1 + v ) cCp *0 ( W(XX + w yy )] ,
f ( x , b ) =
w ( x , b ) xx
160
9 Plates
We formulate the EULER equation in analogy with (6.35) as the necessary condi
tion :
dF ( J L . )
+
V dvr y A,
dw Vdw
dw.,
.y
(4 )
(
dF . .)
+ ( . dF ) + ( dF )
=0 .
dw.xx \ vx
\ dw,Xy 'tXy
' dw.vV
vv
yy \yy
dw
dF
= 0
dw.x
= 0
dF
= 0
dw v
we obtain
dF
dw. XX
dF
dW,Xy
K
2 4 (1 - ^ ) WfXy *
dF
dw.
**
2 ( W ,xx +
W,yy)
2 (1
V ) W ,yy +
2 ( 1 +
V) a T ^
K
- 2
2 ( W ,xx +
W,yy) -
2 ( 1 -
'O 'A x x +
2 0
V) a T ^
Substitution into ( 4 ) yields the differential equation that describes the problem
w ,xxxx +
2 w , x x y y + W ,y y y y +
or, abbreviated,
C1 +
v ) T ( ^ .x x
^ .y y ) = 0
A A w + ( 1 + v ) a T A *0 = 0 .
(5)
- B o undary co n d itio n s
The variation additionally yields the boundary conditions in analogy with (6.35).
Boundary x = 0 , a :
dF
1
8w
d F
)
K dw,xx
-|x
Vdw,xy J,y J
= 0
(6)
Swx ^ - = 0 .
,X dw.xx
Owing to 8w = 0 ; Sw x ^ 0 (sim ply supported at the boundary x = 0 , a ) ,
boundary conditions at x = 0, a a re :
the
w = 0 ,
( 7a )
w,xx + v w,yy + ( l + v ) a T 10 = O.
(7b)
Boundary y = b
8w - (
)
Vdw.xx A
dF
+ dw..
xy ,x
(8)
8 W y
y dw.vu
= 0 .
Exercise B-9-3
161
( 9a )
EI
w ,xxxx -
W ,y yy ~ ( 2
~ V ) W,xxy
'
(* +
V) T
0 ,y =
( 9b )
( 10 )
(11)
w i,xxxx =
* wi = A + B x + C x + D x
wt ( a ) = 0 , ( 12a )
+ ( ! + v ) otj. Xe = 0 and wt
( a ) + ( 1 + v ) otT *0 = 0 .
( 12b )
) .
(13)
P artial solution w2 :
LEVY's approximation (9.40) :
w2 = ^ Yn ( y ) sin n ^ X .
(14 )
w2
(
> y
a n d
w2
( a
> y
>
W2 , x x ( . y ) + ' ' w 2 , y y ( - y ) = 0
and
15a )
^ 15b J
162
9 Plates
(1
v ) n r 8 1( A
, m
h ! S
B S S I . i n i S S L
+ C ..M. S I i + D . i 5 I c c h ^ )
w ith the yet unknown constants An , . . , Dn .
Now we can form the total solution w = Wj + w2 as :
w = ( l + v)ccT e i | y x ( a - x ) ( l - m ^ ) +
( A OTh i S
+ B l S , i n lli a
( 1? )
n 1, 2
, _ . , niry , _ n n y
. n ^ y ^ . nTi xl
+ n sln^
n a"^ cosh
J sin | .
We wish to express all term s in (1 7 ) by means of one sum m ation sign, i.e. we
w rite a FO U R IER series for the first term ( see [ E T 2 | 8.2.4] ) :
n
The FO U R IE R coefficients are then calculated as follows :
a
2 l r / \ nr r x ,
an J ( x )sln a
1 1 /
an =
x ( a -
n7TX
x ) sin - d x =
n TCX J
x sin - d x
n7TX
sin - dx .
2a
3
(1
v ) a T 0 1
00
Ian ( 1
\
-
) +
An cosh~
n - 1,2
n7iy
n 7 ty
n rc y
, _
n rc y
_ nrcy
. n7TX
sin -
(19)
( 20a )
( 20b )
Exercise B-9-3
163
= (1 + v ) a T 0 1( - ^ )
in a FOURIER series :
= 2 y
sin a. x .
a a n (v - 1 ')n - 1
b
b
n1=
=l1,2
,2
Inserting into boundary condition ( 9b ) now yields:
y = b:
a
f a_ ( 1 - m ) + A_ cosh a b + B_ a b sinh a b +
a nL nV
/
n ' n n
n
'
( 20c)
7b a a n4
y = -b :
(_l)n _1]} = ;
El
^ a a n | an ( 1 + m ) + An cosh a n b + Bn a n b sinh a n b Ka
Cn sinh a n b - D_n a n b cosh
{[
( 20d )
( - 1 )" - i ] } = o .
Special case o f m 0
( 21a)
164
9 Plates
( 22a )
( 22b )
X a a n a n + X a a n Bn
+1 a n b sinh a n b
2 B_ sinh a b
Bn \( 1
o l
sinhan b
X aan
b coth a n b /) .
N um erical evaluation:
- D eflection fu n c tio n at y = 0:
OO
w = (1 + v ) o T 0 1 ^
( a n + An ) sin
ub =
n7r ub = n 7T
cl
m = 0
a3 = 0.0048 a
a2 =
l ) X = 0 ( E I y = 0):
Bn = 0
An = 0
w ( 1 + v ) a T 0 j a2 ( 0.12 91 sin
0.1291 a
2 ) X = 1: Bj =
+ 0.0048 sin - )
= 0.0058 a ,
cosh k + jr + sinhn
A3 = - Bj ( 1 +
71 coth 7 1 )
b3
a3
a4 = 0
= - 0.0241 a2 ,
Exercise B-9-3
3 7TX
( 1 + v ) a T Gj ( ay + Aj ) s i n ^ - + a3
* 291 a = 0.00878 a2 ,
COSh 7T+ 7T
Aj = - Bj (1 + 7t coth n ) = - 0.0364 a2 ,
B3 0
>
A3 ~ o
w( x , 0 )
( l + v ) a T 0 3a2
at y = 0
M om ents Afxx at y 0:
M xx = K [ W, xx +
V W , yy +
(!
+ V ) T 0 l] .
PQ
w ,xx(x >) = - ( !
+ v ) t 0 1 2
( an +
2
siniLi
n=l n n
5 k x
3 ttx
7t:x
i
4 /I SI
s .m
1------
7t v
a
3
+ ' ( ) T fl1 1 +
\2
An ( ^ r )
ii - 1
sin^
0.05
0.10
0.15
w(x.0)
165
166
9 Plates
and with
00
^,yy(x,) = - ( 1 + v ) a T 0 1 J
(An + B j (
\2
) sin^
Mxx = K ( ! + ' ) ) a T e i ^
1/
(1 - v ) A n - 2 v B n
\2
) s in ^ ^ .
n= 1
Here, we apply the sam e assum ptions as in the case of th e deflection functions,
i.e.,
1 ) X = 0:
2= 0
(l-v )A 1-2vB1 \
a
MXX = 0 .
2 ) X = 1:
( l - v ) A
M
3 ) X = 00 :
2 v B j
0.202 sin
0.202
K ( 1 + v ) 0^ Gj .
( l - v ) A j - 2 v B j -A- = -0 .3 1 4
- 0.314 sin
K ( 1 + v ) a T 0j .
K (l + v )a T 0 ,
at y = 0 are presented in Fig. B-22.
W ith vanishing stiffness of the tubes ( X = 0 ) at the boundaries the plate de
forms w ithout stresses.
a /4
3 a /4
a /2
1=0
0.1
0 .2 0 2 ^ -^
0.2
0 . 3 1 4 ^ < <' 1 = \
0 .3
?u=
0 .4
'
Exercise B-9-4
167
E x e r c is e B -9 -4 :
A thin, rectangular plate with all boundaries clamped (dim ensions a, b,
thickness t ) is subjected to a uniformly distributed load with the intensity
P = P0 ( Fi& B' 23 )Determine the deflection function by means of the RITZ method, using the
trigonometrical double series as an approximation function
*
V1
w = Z
m= l n=l
1 ' 108
2m7rx \ ( .
2n7ry \
j ^ l - c o s ----
S o lu tio n :
For this task, we first write the internal potential ( strain en erg y ) of a rectangu
la r plate according to (9.33), expressed in term s of the approxim ation function
w*:
ba
n i = i Kj J ] ( W*XX + W*yy)2
00
(0
t3
with the plate stiffness K = ----------- and the notation d/d x ( ) . d/d y
( S.
1 2 (1 - v )
v ''x
v Ay'
ba
IIe = - J J p0 w* dx dy .
0 0
(2)
W hen applying the R IT Z method, the approxim ation function has to fulfill at
least the geometrical boundary conditions. These a r e :
w* ( 0 , y ) = w* ( 0 , y ) = 0
w* ( a , y ) = w* ( a , y ) = 0 ;
( x , 0 ) = w*y ( x , 0 ) = 0
w * ( x , b ) = w*y ( x , b ) = 0 .
(3 )
W'
168
9 Plates
an
_ d(ni + ne) ^ o
^amn
^amn
(x.y) =
m= 1 n=l
oo
00
00
. 2iti7t:x
22ni t y \
*
X X a ______
2mn; c in
2m7t x/I/ I. r n c _____
W?x = Z Z amn
sln- S I 1 - COS~ b ~ j >
m = 1 n1
X
*
V1 V1
/
W*y = Z
amn v
m = 1n = 1
oo
oo
2 2
V NT
4m ti
2m7tx /
w*r X X _______ ____
Z ____
Z amn
2 COS a V
m
n=ln=l
a
oo
oo
oo
o o
V V1
f
Z
A j amn V
m= 1n= 1
2n7iy \
K / 1
2iti7t:x \ 4n k
2nn:y
COS a / 72 COS K *
COS
*
V NT4mnTi
. 2m7tx . 2niry
w ,xy /z_i Z/i a-------mn
a bi sm ;a Sln b
m In r 1
/r \
( 5a)
/
\
( 5b )
. x
( 5c )
0 0
2 t t x \ 4 7r 2
+ an ( 1 - cos 1 . v cos
b2
2 i t y l2
J 2 167I4
2i t x /
2nx \
27ty/,
27ty \
h 1 - v ) I an ^ 2 - ^ 2 cos
A1 - cos- i r - y cosb- V1 COS_b_ J
2
16 7t4 .
_ a u ~ 2T 2 sm
2/
2 7t X \ .
2/
2 7t y \ 1 1 ,
b a
J J po 3- ! ^ 1 - cos^ ^ ) ( 1 - c o s - ^ - ) d x d y .
0 0
^ r ) s m { b- AJ j dx dy
Exercise B-9-4
169
^ ) - p0 a n a b
b
a b
(6)
an
4 7r4 K a b a n ( - T + ~TT2 + A ) Poab = 0
11 ^ a4 ' a2b'
^a ll
11
, 4V( 3 , 2 , 3 V
KVa4 + a2,2
b + b,4
4n
Po
w ( x y ) = 7 7 ^ 7 3
2 ir x \ ( ,
2 rr y \
- cos- b ~ ) -
) ( 1 _ cos^
a
)
a
'
*/ a
W ma x
a \
Po ^
( ~2 ~ 2 ^ ~
8 k 4K
1
_
^ W '
w0
Poa4
it K
one-term
four-term
1 ,2
1 ,2
WL x / W0
0.125
0.12205
This exam ple clearly shows a relatively fast convergence of the R IT Z method.
170
9 Plates
E x e r c i s e B -9 -5 :
D eterm ine th e m axim um deflection due to a uniform ly distributed load p0
by m eans of th e GALERKIN m ethod for a rectan g u lar plate w ith two clam
ped and two sim ply supported edges (Fig. B-24).
sim ply supported
S o lu tio n :
The GALERKIN method requires fulfillm ent of all boundary conditions. In order
to comply with this demand, we form ulate beam solutions for b oth coordinate di
rections. These solutions enable us to fulfill the corresponding boundary condi
tions of the plate. For this purpose, the eigenfunctions of the vibrating beam or of
the beam subjected to buckling prove to be useful, since they also comply with
desirable properties of orthogonality.
In the present task, we will proceed from the basic equations for a vibrating uni
form beam (bending stiffness E I, mass density p, length a ) [ A.20, B.4]:
d w
,2
p o w
To xT + j l To xT ^ = 0 -
(0
( x ) sin u
m 111
111
(2)
Xm ,xxxx(x ) - ^ T Xm ( X ) = 0 -
w ith con
( 3 )
/
2
4 / p Wm
where Xm = a y
, ( m = 1, 2 , . . . ) denote eigenvalues.
T he solution of ( 3 ) reads :
X m( x ) =
C lm sin Xm | -
C 2m c o s X m | -
C 3m s i n h X m T
C 4m c o s h X m | -
(4)
Exercise B-9-5
171
The eigenvalues are calculated from the boundary conditions for a simply supported-clam ped beam :
w(0) = 0
, w>xx( 0 ) = 0
w(a) = 0
y Xm( 0 ) = 0
w x ( a ) = 0 > Xm( a ) = 0
, XB1#xx( 0 ) = 0 >
(5a)
, Xm x ( a ) = 0 .
(5b)
(v 6 ')
f(71
17 \
v
for
m =/= n ,
for
m ^ n
X mX n d x
x 0
(8 )
a
, xx
,xx
2 , xx dx
Jf x m
x= 0
In th e present example, the same boundary conditions apply for the y -d ire c tio n ,
and we can therefore use th e analogous eigenfunctions Yn ( y ) for this direction.
Thus, the product series approxim ation for the plate deflection function reads as
follows :
w * ( x , y ) = 2 2 WmnXm ( x ) Yn ( y )
(9)
m n
...
with
= 2 2 P m n Xm ( X) Yn ( y )
( 10a)
J J p X m Yn dxdy
= -id, ------------------
m n
m"
,
( 10b )
JJx^Xfdxdy
K2
J J Xi Yk ( A A Xm Yn ) dx dy
- ^ ^ pmJ J X m Yn X; Yk dx dy = 0
mn
P mn 1.3 ^6__________
K ( I j I 6 + 2 I 2 I5 + I 3 I 4 )
(11)
172
9 Plates
where
Il = J X m ,x x x x Xm d x , I2 = J X m ,x x Xm d x , I3 = J x ^ d x
0
K .yyyyX A -
^ =
(d /d x = ( ) , x ),
J\
,yy Y dy ,
J Y* dy
( d/dy - ( )#y ) .
w^
X j Yj = w* (sin X j f
yields w ith
sinh X j |- ) ( s m X ^ -
sinh X , | - )
Xj == 3.9266
Pn a
w* = 0.00198 -Y__ .
Proceeding from w*/X = w*ry = 0 , the maximum deflection is determined as
Pna
w*max = 0.00223
at x = Jy = 0.383 a .
E x e r c is e B -9 -6 :
A circular plate clam ped at th e outer boundary r = a is subjected to a
constant circular line load q a t a radius r = b ( b < a ) from th e centre
point ( Fig. B-25 ).
a ) D eterm ine th e general solutions for th e inn er ( r < b ) and th e outer p a rt
of the plate. How m any free constants are obtained for th e solution of
this problem?
b ) How do th e boundary conditions an d th e tra n sitio n conditions re ad ?
c ) C alculate th e deflection functions for both th e in n e r and th e outer p a rt
of th e plate. How large is th e m axim um deflection of th e plate ?
d ) Prove th a t th e rad ial and th e tan g en tial m om ents in th e inner p a rt a re
equal.
Fig. B -2 5 : C lam ped circular plate w ith a constant circular line load
Exercise B-9-6
173
S o lu tio n :
a)
The circular plate is divided into an inner and an outer part with the fol
lowing axisymmetrical loads corresponding Fig. B-26.
E>
( r
~ i)
M
inner section
Fig. B-26: Free-body diagram for the outer and inner part of the circular plate
According to (9.^7) , the solution for the outer part ( r > b, index o ) is
wo ( r ) =
C0 +
C i r2 +
C 2 l n i : + C 3 f 2 l n -I: *
( ! )
C4 +
(2)
C5 r
wo>r( a ) = 0
.
Q ro( b )
( d/dr ( ),r ) ,
- q -
(3a)
(3b )
(4)
174
c)
-
9 Plates
D eflection f u n c t i o n s
Outer p art
= 2Cl r +
C2
- ZCj - ^
A w o -
w o , r r + T Wo , r = 4 C 1 + 4
-y + C3 ( 2 r l n - A + r )
(6a.)
+ C3 (2ln-- + 3 ) ,
(6b)
C3 ( ln T
(6c)
+ 4)
= 0 = C 0 + Cj
w o ,r ( a ) = 0 = 2Cl a + ^
Qr0 ( M
= - K
= - 1
^ w
(7a)
a2 ,
(7b)
+ C3 a >
(8a)
o ,- 4 K C 3 i ,
M rrc ( b ) = M = - K [ w o rr + w o r ] =
2 C 1( l + v ) - % l - v ) + C 3( 2 ( l + v)ln-jl- + 3 + v )
= -K
2Cl ( !
v ) - ( 1
Eq. ( 8 a ) yields :
- v ) + C a ( 2 ( 1 + ' ) 1 T
C3 =
+ 3 + v ) =
(8b)
( 9a)
C2( ^
+ ^ tS ) + ^ l1
+ v ) l n ^ + 3 + v )1 = k ( i + ^ ) -
By substituting ( 9a ) we obtain :
C = _____ 1_____ {
X.
a2
lK ( l + v)
b2 1
bq
4K
(9b)
' - T T v C 2^ + ' ) l n + 3 + - ) ] }
c, = - | ( J + , )
For C q one obtains from ( 7a )
C2
bq
2 ? 8K '
(9c)
Exercise B-9-6
175
By that, the deflection function of the outer plate can be determined, however, still
in dependence on the yet unknown radial moment M at r = b.
-
Inner p a r t:
Derivatives from ( 2 ) :
w; r = 2 C s r ,
( 1 0 a)
wi>rr = 2 C s .
(1 0 b )
M = - K ( w . rr + ^ w . r ) = - K ( 2 C 5 + 2 v C 5 )
M
2 K ( l + v)
(lla)
w i ( b ) =
C4 +
C 5 b2
Mb
(11b)
^4 - W" aa V~
t- 2 K ( 1 + V ) '
( b ;) +
C 4 =
( 12)
Wi ( r ) = Wa ( b ) - ^ K ^ j ( r 2 - b2) '
In order to write the final expressions for the deflection functions, we determ ine
the radial moment M by means of the transition condition ( 5b ) .
According to ( 6a ) and ( 9c ) :
3,r ( b ) = 2 C 1b + ^
+ C3 ( 2 b l n | + b ) =
(13)
= C,
2( r - ^ ) + 2C3bl n^
..<*>) = ( F - y - r T T T ^ t K ( l + v)
a2
b2 l + v
bjq
1 1 ^ 7 ( 2 ( l + v ) l n | + 3 + v)
4K
+ 2 ^4 K
r ( b ) =
v+
-b
AI M
b2+ a 2 K
k{
i 2
b -a
bq
4
(1 + v)
2(1 + v ) l n | - 3 - v]} + | ^ l n |
(14)
176
Plates
Mb
K (l + v ) '
( 15)
Substitution into the transition condition and some algebra finally lead to :
( 16)
Eqs. ( 9a ~ d ) substituted into ( 1 ) yields the following expressions for the outer
deflection function:
,( 0
a2 b2 ( l + v )
f
M
b2( 1l + v ) 4+ a2
a2((1l - vv)) I K ( l + v )
Ti _ ( 2 ( l + v ) l n ^ + 3 +
bq
4K
v ) ] j ( |- - ^
+
+ lni) +
bq
4KV 2
The maximum deformation wmax for the circular plate is obtained from (18 ):
Vi( 0 )
d)
qa b
4K~
( 19)
= - K [ Wi,rr + T Wi, r] -
wi i r r ( r ) = 2 Cs .
= - K ( 2 C 5 + 2 ^ C 5r )
= - 2 C BK ( l + v ) ,
MW = - K ( 7 2 C 5 r + 2 v C 5) = - 2 C 5 K ( l + v )
Mrr = *LVV = M
The two moments are constant for 0 < r < b and equal to the moment M ( see
(1 1 a )).
Exercise B-9-7
177
E x e r c i s e B -9 -7 :
A circular ring plate m ade of steel ( E , v ) is clam ped at th e in n er boundary
r = b ; th e plate is subjected to th e following harm onically varying line load
at th e outer b o undary ( Fig. B-27 )
t
\
Fi
q ( a > ^ ) = ~ ^ costfi = q i c os v>a ) W rite th e differential equation an d
the b o undary conditions for th e
circular plate.
b ) D eterm ine th e deflection function
for th e circular plate.
S o lu tio n :
a) Based upon the differential equation (9.25) for the shear-rigid, isotropic circu
lar plate
AA = ( 4
v dr
+ i A + - L 4 ) 2. = - E < i i )
T dT
r2 dtp J
K
( 1)
= 0
r ( b, cp) = 0 ;
Q r U . ' P )
= q ( a ,<p) = qjcostp .
( 2a,b )
(2c,d)
(3)
178
9 Plates
This form corresponds to the first term ( n = 1 ) in (9.49), which upon substitu
tion into the plate equation (9.25) transforms this partial differential equation in
to the ordinary differential equation (9.50) with n = 1 for the function Wj, and
the solution Wj for the homogeneous equation is given in ( 9.51a). As ( 1) is ho
mogeneous, its solution w thus takes the form:
Co
3
1
w ( r , c p ) = C j r + + C 3 r + C4 r In -M cos cp .
(4)
(5a)
C O S cP
(5b)
C O S (p
+ ~F+C3r + C4rlnTT
= w jrr + i w r + i w w
(5c)
COS <p ,
8 C
*3, r
+1
2C a
cos <p
( 5d )
= -K
+ ^ T w, r + 7 w, ^ ) ]
I c2
c4
KI 2 - j -f* 6 C3 r -f- h
+ A
cos cP
2 ^ ( l - v ) + 2C 3 r ( 3 + v ) + C4 - p ( l + v) j cos<p.
(6)
C rj
F + ~y
Qr = - K
4--
rr V( ri w,v> t \ 2 w
2C4
+ 8 C3 ~
C 2
+ -j
l-v (
r v
C4
C1
F"
X1
+ C3 r + in -g- ) jc o s
^ i 4^ C
\Jiifi 1J ,
c 2
, p
^3 ~
C4
r
r
b
+ 2(3 + v)C3 - ( V i c
( 7)
b Cj 4 -
( 5a ) in ( 2b )
Cl ~ ^4 +
(6)
in ( 2 c )
1* b C3 0 ,
3 b
C3 +
C4 =
( 8a )
( 8b )
0 >
^ - 5- ^ C 2 + 2 a ( 3 + v ) C 3 + (1 + v ) ^ = 0 ,
fl
3.
( 8 c)
Exercise B-9-8
(7)
179
in ( 2 d ) >
- K |
~ *
C 2 +
i ^
2 ( 3
v ) C 3 -
2 ~V
^ ^ 4
c o s tP =
(l ( a > (p )
<l i c 0 8 1
(8d)
c 2 + 2 ( 3 + v ) C3 _ ( i _ ^ ) C4 = _ | .
2C2
_ C4
C2
2(3 + v ) a C 3
(1 + v ) 4 - C 4
( 1 - v ) TT C 2
2(3 + v )
( - 3 + v ) - i - C4
(9)
_ ^ b ^3
( 1 ~ V) ^ T
- 2 b C3 - C 4 = 0 .
- 3 b2 C, - C.
C3
0 ,
0,
( 10)
= - ^ .
a,
q i ab2 ( 3 + u ) a 2 - ( l + u ) b 2
( 3 -f-v)a4 + ( l - v)b4
_
qi a ( l - u ) b 2 + ( l + u ) a 2
qi a
8K ( 3 + v )a4 + ( 1 - v ) b 4
4K '
Finally, ( 8a ) becomes :
r, _
1_
c2
b2
1.2 ^
3
_ q i a (3 + v) a4 - 2 ( 1 + v ) a 2b2 - ( l - v)b4
_ 8K
(3 + v )a4 + ( l - v )b4
4).
r( a . 0 ) = ( l 6 1 n 4 - - ^ ) ^
b3
1178 \
E x e r c is e B-9-8:
A s h e a r- r ig id circu lar plate (ra d iu s a , thickness t ) as shown in Fig. B-28 is
resting on a linearly elastic foundation (fo u n d atio n stiffness k [ N / m 3 ]), and
is subjected to a constant surface load p0 .
a ) Express th e to tal p o ten tial of th e given sy ste m .
180
9 Plates
Total potential
The total potential energy is the sum of the elastic energy of the plate l i p , the
elastic energy of the foundation l i b , and the potential of the external load IIe :
n = n p + nb + n .
(i)
The elastic energy o f the circu la r plate can be obtained from (9.35). Since we
are dealing with an axisymmetric problem all derivatives ( )iV 0. Thus, we
obtain the simplified form with d /dr ( )>r
n_ = i - Et3
p
W,r W,rr dA
(2)
K
The elastic energy o f the fo u n d a tio n is
2 tt a
2 7T
J w 2 rdr d<p.
n b = -^ J J ( k w )wdrrd<p = i k J
00
y) = 0 r= 0
(3)
Exercise B-9-8
181
2-7Ta
n e = - J J p 0 wrdrdcp .
(4)
After integration over <p, (1 ) with ( 2 ) and ( 4 ) yields the total potential energy
n =
b)
a
71J{K[(w.rr+ T w.r) -
w,r w fr +
kW2 -
2 pQw ] r dr .
(5 )
In the following, we will determine the differential equation and the boundary
conditions of the problem. The calculus of variations is employed to achieve this
goal.
The basic requirement is that the variation of the total potential energy must
vanish, and in view of ( 5 ) and (6.34) this may be expressed a s :
S II = 7 r s J ' F ( r , w , w r , w r r ) d r = 0
(6a)
- 2p0r w .
(6b)
V dW/T Jz
( 7)
Vdw h r
( dF 'I 1Sw
^ dw,rr JT i
dF
Sw
dw.rT
(8a)
= 0 .
(8b)
= 0
182
9 Plates
r ct\
(9'
W,r - T
. . ..
0,
( 10a )
= 0.
( 10b )
W , r r - W,rrr ) Sw
...
...
= Qr
K ( W, r r +
T W , r ) S w ,r
= M
(lla)
The approximation must comply with the essential ( geometrical ) boundary con
ditions, i.e.,
w* ( a ) = 0 ,
( 12a )
w* ( 0 ) =
( 12 b )
( lib )
The derivatives
w * = 2a2 r
w *r = 2 a 2
n = 71:J { K [ ( 2a2 + 2 a 2) 2 ~ 2 ^ r V^ 2 a 2 2 a 2 r ] +
0
+ k a2 ( r4 - 2 r2 a 2 + a4 ) - 2 p0 a2 ( r2 - a2 ) } r dr =
a
= 7 r : J { K a ^ 8 ( l -| -v)r-(- k a 2 ( r 5 - 2 r3 a2 + a 4 r ) o
- 2 p 0a2 ( r - r a ) } d r .
Exercise B-9-9
183
n = 7i;|a2[4Ka2( l + v ) + ^ k a 6] + a2 p0-|-a4 J .
From the extrem um condition (6.37) , i.e.
a2 [ 8 K a ( l -f- v ) -(- k a ] + - ^ p 0 a
= 0
(13)
= 0 follows th at
Po*
a^
16K(1 + v ) + - |k a 4
and thus the approxim ative deflection function is determined as
Poa
' CO
16K (l + v ) + |- k a 4
For the given numerical values we obtain
w j = w* ( 0 ) = 1.296
k '
T he exact solution can be found in [ MARKUS, G.: Theory and Calculation of axisym m etrical Structures ( in G erm an). Diisseldorf: W erner-V erlag, 4th edition
1986] as :
ber o t - i ^ b e i ot
ft
wo = k
with
a =
In (14 ) ,
guments,
and Y j (
1950). A
KELVIN
( b e r a - -L ^bei'ot)ber ot+(bei g -f
d
1 -v ,
( 14)
b er' a )b e i a
1 =
I q ( x 7 T ) = ber x bei x .
T he m axim um deflection calculated by means of (14 ) leads t o :
w0 = 1 .2 3 9 ^ .
Hence, the ratio between the approxim ate and the exact value is
10. _
= 1.046 , i.e. the agreem ent is rather good.
E x e rc is e B -9 -9 :
Consider a thin, circular, rotatio n ally sym m etric s h e a r-r ig id plate w ith ra
dius a ( Fig. B -29). T h e plate has nonuniform thickness described by the
hyperbolic function
lo-atr"*.
184
9 Plates
t(r)
S o lu tio n :
a)
We disregard the fact that the m id-surface of the plate is not plane, and proceed
from the equilibrium conditions for a circular plate subjected to an axisymmetric
load ( see [ E T 2 ] ) :
( r Qr ),r +
P ( r ) r
0 *
( la )
( lb )
( d / d r = ( ) ), the radial
Mrr = K ( r ) ( * fr + v | ) ,
(2a)
MW = K(0 (? +
(2b)
Exercise B-9-9
185
E t> >
K(>- )
Substitution of ( 2a,b ) into ( lb ) leads to the differential equation
4i
+ l ( l +
r v
K (r)-f r U
K( r )
- i f l - K Cr_ k v r ) ^ = M l l
r
K( r )
' V K( r )
(3 )
( '
( >
(5a)
E to ~ = const .
K = ------0 12 (1 - v )
( 5b)
The relations ( 5a,b ) are now substituted into ( 3 ), and we herewith obtain a dif
ferential equation of the form
or, alternately,
rx+2
r < L - | - ( l - x ) r i L - ( l - t - v x ) i p = -^.
( 6b)
t^o a
The above equation is an EULER differential equation for the slope
of the de
flection. On the rig h t-h an d side, the yet unknown transverse force Q r occurs
which consists of the constant surface load p and of a concentrated load F at the
centre of the plate (reaction force from the support).
To determine Q r ( r ) in ( 6b ), we integrate once ( la ) with p - const and obtain
r Qr ( r ) = c i -
- pr)
(7)
r2 * ,rr + ( ! - x M , r - (1 + v x ) * =
-pr)-^
.x+1
_2
1 .1/ iU
p r x+1 a 2
2K,A
a
as a differential equation which describes the given problem.
a
p r x-f-3
x+3\
186
b)
9 Plates
Deflection
(9 )
= 0.
( l O)
|- 1+ v x .
(ll)
= A rX+1
=*
= A (x + l ) r X ,
W'r
W i = A ( x + l ) x r X 1 .
( 12)
A [ ( x + l ) x + (1 - x ) ( x
A x ( l - v) =
+ 1) - (1 + v x ) ] =
^ 5
2 K0 a,x~
2 x ( l - v ) K 0 ax 2
Thus, the first particular solution is :
nrx+1
^P,
=
~oZT\
rl
2 x ( l - v )1is
K 0ax -2 '
^13 )
*M = B r
=*
. ,
( )
...
'
B = - 2 [ 8 + ( 3 - v ) x ] K 0ax
Exercise B-9-9
187
p2
(15)
pr x+l
2 ( l - v ) x K 0 a11 2
pr,x + 3
2 [ 8 + ( 3 - v ) x ] K 0 ax
( 16)
The free constants Cj and C2 are determined from the following boundary con
ditions:
1)
*(0) = 0.
According to (11 ), X2 must always be negative,i.e. the solution will always
take the form C2/ + 2 I , and therefore we must have C2 = 0 .
2)
Mrr ( a ) = 0 ,
vanish.
X C aAl-1 +
( x + 1)p a2
2 (1 -
+1
ia ( c r
(x + 3 )p a 2
Kq
2[8+ (3-
pa
2(l-v)xK0
v )x
]K0
pa
2[8+(3-v)x]K0
)j =
x2 + 2 ( 2 + v ) x + 4 ( l + v )
Cl =
a>i - 3 ( X 1 + v ) K 0 I
( 18)
(8 + (3 -v )x )(l-v )x
The slope of the deflection function can now be calculated by means of (16 ), and
the deflection itself is obtained by integration,
J* = - w,r
w = - J i } j ( r ) d r + C3 .
( 19)
The boundary condition for determining the remaining free constant C3 reads
w(0) = 0
C3 = 0 .
After integration of (19 ), we obtain the following function for the deflection:
x+2
Ai -j-1
Pr
+
(0
^ Xj + 1
2 ( l - v ) x (x +2) K0 a3
( 20a )
x-2
x+4
__________ P______________
2 [8 + ( 3 - v ) x ] ( x + 4 ) K 0 ax
and, in shorthand notation:
Al + i
x+ 4
( r ) = ( K j r Al + 1 - K , r x + 2 + K, r'
)p .
( 20b)
188
x=0.5
tn=const
x 2+ 2 ( 2 + v ) x + 4 ( l + v)
a*1 3 (Xj + v ) Kq [ 8 + ( 3 - v ) x ] ( l - v ) ( ^ + l )
1
2 ( l - v ) x a x 2 ( x + 2) Kfl
x_____________
3 ~ 2[ 8 + ( 3 - v ) ] ( x + 4 ) K0 ax
Fig. B-30 shows curves for the deflection for the shape parameters x = 0.5 ; 2.0
and 3.0. These curves may be compared with the solution for a plate with central
support and constant wall thickness
Po
64 K
r4 + 2 a2 r2 (
+ 4 In
Vl +v
which is also shown in the figure. This expression cannot be determined as a spe
cial case ( x = 0 ) of ( 20b ), since the characteristic equation (10 ) possesses a
double root X = 1 for x = 0.
E x e r c i s e B -1 0 -1 :
A s h e a r -rig id rectan g u lar plate (d im en sio n s a, b ; thickness t ) w ith sim ply
supported edges and w ith a stiffener along its centre lin e y = 0 (cross sec
tional area As , bending stiffness E I ys) as shown in Fig. B-31, is subjected to
an ax ial load Nx = c o n s t.
a)
Exercise B-10-1
Ix =
const
IWrfl
i
i
CD
J____________
189
(h
A s > EIys
( 1)
with
w ( x f ) = I wm ( j ') silli!T L
m
m = 2, 4 ,6 . . . .
(2)
t nx =
Wm ( y \ y y y y ~ 2 & l W m ( y l y y +
( 1 ^
,( 3a ) reads :
) Wm ( Y ) = -
( 3b )
(4a)
190
(4b)
x 3,4 =
nx ) = i / &i ( nx - i) = i x :
(5a)
( 5b )
(6)
w(x, - ) = 0
(7a)
x lf),yy + v w ( x
) . 1 = 0
Ij
antisym m etrical
b uckling
[
|
and thus
W( x !"),yy = 0
sym m etrical
buckling
Exercise B-10-1
191
~2 \ y y
( 7b )
0 '
With symmetrical buckling, the buckling shape at the stiffener has to have hori
zontal tangents in the y-direction and we therefore have
w ( x , ) fy = ,
i.e. according to ( 2 ) :
wm ( 0 ),y = 0 .
( 7c)
Finally, we have to formulate the transition condition between the single plate
parts and the_stiffener ( see Fig. B-33). Assuming that q = q ( x ) isthe stiffener
loading and Qy. ( i = 1 , 2 ) are KIRCHHOFF's effective transverse forces corre
sponding to (9.14) , we demand th a t:
q = % - Q y 2-
(8)
Q
^y2 = -K
i.yyy
2 ,yyy + ( 2 - v) (<
1 ,xxy
I ,xxy ) ] y _ 0
= q
and
W2,y
w i.y y y
w 2 ,yyy '
(9)
The stiffener loading q resulting from the difference of the effective transverse
shear forces relates to the displacement w via the differential equation of the
bending of a BERNOULLI-beam with an additional axial load [A.19]
F x = Nx
h
d l|=
2
Fig. B -33: Free -body diagram of the transition between plate and stiffener
192
i.e.,
q ( x ) = E I ygWfXXXX + N w
h
with the bending stiffness E Iyg , the cross sectional area As , and the height h of
the stiffener. Comparison with (9 ) yields :
E l ys w.l / x x x x +1 Nx -r^
h w,1 /XX -f1 2 K w,1 /
yyy y =
( 10 )
0 ,
o
8=
bh
we obtain
(* ? ) - n | =
- < s)
(W)
With ( 7a -f- d ) we now have four equations for determining the four constants of
(5b).
+ b<D C
- 2 x2
b
co sx 2^
b
sin x 2^
ii
o
II
a
+ 2 xj B
+ *2
ii
' ( Ua )
xl B
b<D A
2 .
b
x2 sin x2 2
2 cos x2 b2 C
x2
v Xj
b A +. Xj2 sinh
i, Xj
b B
Xj2 cosh
sinh Xj
2 . ,
b
Xj sinh Xj
0
b<D
2x.
* 2 cos>2 2
0
b<D
2 .
b
- x 2sin x 2 ^
= 0 .
( lib )
2 x
The vanishing of the buckling determ inant delivers the so-called buckling condi
tion :
Xj X2 ( Xj - X2 ) +
( Xj tan x2^ - x2 tanh Xj^- ) = 0 .
(12 )
According to ( 4b ) and with the above introduced abbreviations we can formulate:
Exercise B-10-1
193
7r2 K
Since we devote our main interest to the critical load Nx = NXcrjt = k jwrite
^
^ _ rrn _
K ~ * b
with the buckling value k. Therefore, with a = a / b we have
*v)-
<13>
In ( 13) the buckling value k is still contained in the values Xj and x2 . This
equation is therefore solved by tr ia l- a n d - e r r o r .
Fig. B-34 presents the k -v alu es for different stiffness ratios y in dependence on
the dimension ratio a .
For
T = TOT =
8 = bK =
one obtains the garland - shaped curves of the unstiffened plate as already describ
ed in [ET2] with a minimum at k = 4 . For
T= 0
= 0.2
the k -v a lu e drops below 4 which implies that the stiffener under compression
possessing no bending stiffness ( E Iys K 0 ) but a certain cross - section ( As > 0 )
has to be supported by the plate. An arrangement of this type would of course be
unsuitable.
20
16
15
o
J3
"cS
>
Ml
c
12
o
10
3
03
2.85
2
3
a = a /b
Fig. B-34: Buckling value of a plate with one stiffener dependent on the ratio
a = a /b
194
Wm ( ! ) , y y
( 14a)
(14b)
(14c)
>
wi n ' -( 0 )/ , y y = 0 .
( 14d )
( 15)
Nv
Exercise B-10-2
195
E x e r c is e B -1 0 -2 :
A th in circu lar p late (ra d iu s a, thickness t, Y O U N G s m odulus E ) is
clam ped along its outer boundary as show n in Fig. B-37 an d is subjected to
a constant surface load p th a t leads to large deflections.
T his problem is to be treated as a coupled d is k - p la te problem, and th e de
flection is to be d eterm ined by m eans of th e GALERKIN m ethod w ith a
sin g le -te rm approxim ation. T he results are to be com pared w ith those ob
tain ed by lin ear theory.
196
rrrrrij
i'
=*t
S o lu tio n :
We proceed from von KARM AN's equations in operator notation (10.13)
AA w = ^ + ^ 0 4 ( w, ) ,
(ia)
AA = - ^- 04 ( w , w ) .
(lb)
wU
|7
I70
- k + k
_E
wU
I
2
-J
Uj*
I
'tJ' v
(2a)
(2b)
The problem is m ost conveniently solved in polar coordinates for which reason we
transform the covariant derivatives in the equation by means of ( 2.4O)
12 21 I _i
= + J L 12 12 I _ 1
K
K Wl22l l l + Wl21ll2 +
21 12 I . I , 21 21 I _|
+ Wll2 I21
Wl11 122
E 12 12 1
1
12 21 1
1
"2"[ Wl22Wlll + W|21 WI12
(3a)
(3b)
21 12
21 21
+ Wll2Wl21 + w l11 w | 22
T he m etric tensors, their determ inant, and the CHRISTOFFEL symbols required
for the further treatm ent of the problem, have already been determ ined for
cylindrical coordinates ( see 2.5). If we now substitute the perm utation symbols
(2.20b) and the covariant derivatives ( 2.34b) (e.g. w| jj = w ,i|i e^c- ) bito the
r ig h t- h a n d sides, and if we take into consideration th a t the dependence on the
angle coordinate i;2 <p vanishes in the derivatives owing to the axisym m etrical
shape and loads, the equations reduce with i;1 = r and d /d r = ( ) to
{ r [ l ( r W-r), r L [ r = ^ + ^ ( w . r ,r ),r ,
(4a)
(4b)
Exercise B-10-2
197
(5)
Eq. ( 5 ) contains terms with three constants which can be determined from the
following conditions. First, we use the disk stresses according to (8.7)
" T 1 => '
I 6* )
Sv = V
= *-'
Substitution of equation ( 5 ) into ( 6a,b ) yields :
a
o
<p<p
<>>
^ J r ( J ^ L dr ) d r + C1( l + 2 l n i ) + 2 C2 + 5 * f
= -E
7 l r ( J ^ - dr) drl'r + Cl (
3 + 2l n i ; )
2C2 - 7 --
( 7a)
Owing to the condition that the stresses in the middle of the disk ( r = 0 ) cannot
become infinite (fin itn e s s condition), the constants Cj and C3 must vanish.
The remaining constant can be calculated from the boundary condition
u (a ) = 0 .
(8 )
For this purpose we proceed from the material law (8.9) and then replace the ex
pansion <p<p by the radial displacement u, using the corresponding linearized
strain-displacem ent equation of the axisymmetrical stress state in (8.8). (Higher
order terms with respect to the deflection w in the strain - displacement equation
vanish in this case.) It then follows that
u = re w
= i ( w
v 0 rr)
U = r { i l Y j r ( j ^ T ^ d r ) d r ]'r +
(9)
198
c2
5 - 3
, 2 C 2 (1
v
+
T7
v ) ] _
I
.2
5 -3 v
a2 12(1 - v ) '
Using the approximation w* according to (10 ) , we now state with ( 5 ) an approximation for :
4 6
, 8\
2 5 - 3v
0i** = E 2( r _
( 12)
2
r
+
^
4
)
+
E
r
C
' 6(1 - v) '
a2
a
Da
r
V a4
Further calculation is then carried out by means of GALERKIN's equations,
stated according to (6.38) . By means of ( 4a ) we write the operator
L ( w *,
( 13)
2x a
J" L ( w *,
0r= 0
J"
<p=
) w* r dr d<p = 0 .
1 - v
-(f)3
16
3 ( 1 - v 2)
Fig. B-38 presents the numerical evaluation of the calculation with v = 1/3. In the
case of higher loads ( p > 5 ), the nonlinear theory leads to smaller deflections
than the linear theory, owing to the membrane forces in the plate which increase
the stiffness against transverse deflection and thus facilitate a higher load bearing
capacity [B.8].
(!)
Fig. B-38: Comparison of maximum deflections obtained by linear and by non
linear theory (dotted lines denote deviation of the pressure for special
deflection values in the scope of the linear and nonlinear theory)
G en era l fu n d a m e n ta ls o f sh e lls
11.1
11.1.1 R e p r e s e n t a t i o n o f s u r f a c e s
W e assum e th a t th e re exist one-to-one relationships betw een th e curvilinear
coordinates ( GAUSSIAN surface p a ra m e te rs ) ( a = 1 , 2 ) and th e C artesi
an coordinates x1( i = 1 , 2 , 3 ) of th e points P of a surface ( Fig. 11.1). T his
correlation is expressed by
(11.1a)
x
or, in vector n otation w ith th e position vector r
[C .5 ,C .ll] as
r = r ( f ) = x 1( f ) e i
of a point on th e surface
(11.1b)
dX
dX
n 1
d X
,x
n 2
( 11.2 )
200
Surfaces of revolution
Definition 1:
One obtains a surface o f revolution ( Fig. 11.2), if a two-dim ensional curve m positioned in th e x 1, x 3 - plane ( x 2 = 0 )
1
x = r
3/
, x = 0 , x x ( t)
( r , $ ) = r cos $ e j + r sin $ e 2 + x 3 ( r ) e 3
(11.3a)
or according to ( 11.1b)
(11.3b)
Note:
Special case :
- Spherical surface
If th e m eridional curve t) = 0 of a surface of revolution is chosen as a
circle w ith th e rad iu s a an d centre point in th e origin (Fig. 11.3), a sphere
is described. A fter introducing an angle tp, one obtains
1
x = asiny>
x = 0
x = a cosy) .
(11.4 a)
, x3 = a cosy) .
(11.4 c)
Fig. 11.3:
201
P aram eters of a
spherical surface
7T
202
r = r U) = yU1) + 2* (f1) .
(U-5)
an d
z = e3 .
(11.7)
,..D >
/ 11.8 a)
e 3
(11.8b)
203
11.1.2
(11.9b)
( 11.10)
T he base vectors a Q are lying in th e tang en tial plane w hich touches the
surface a t a point w ith th e position vector r ( ) (F ig. 11.7a). F rom the
base vectors ( covariant surface tensors of order o n e ) one obtains, in analo
gy to (2.1 a), th e com ponents of th e covariant m etric tensor or surface te n
sor.
C o m p o n en ts o f the co va ria n t m e tric or surfa ce tensor
F undam en tal quantities of first order - f i r s t fu n d a m e n ta l fo r m of surface
theory &aP = r
T .P
a a
' a p
( 11.11)
204
1=
a ll
a i2
a 21
a 22
( 11.12)
(11.13)
dt
(11.14)
(11.15a)
(11.15b)
dr-da3
OiJfP
= ~bapded(>
w ith th e c u rv a tu re co m p o n en ts
(11.17)
I
ft i 3^ , 3.^ 1
ya
(11.18)
205
bll
b12
b 21
b 22
b = lb<J =
(11.19)
( 11.20)
rk
aP
+ = k - 2Hk + K = 0
a
( 11.21)
Invariants
TT
J0 ! ,
"2
b/ ?= T b
&f3 = K
m ean curvature ,
GAUSSIAN curvature
K = l b l = T
(11.22a)
. (11.22b)
1 Pi ~ 2 a
Special c a s e s :
(11.23a)
^3 =
r3 -u
1 a/3 ap
(11.23b)
ri 33a = ir 33
a ir 33
3 uo
aa,p f ap a7 + bQ/Sa3
a J/3
3# or
b a/3a 3
b a a /3
(11.24)
206
(11.25)
sin ? ? e2 + z s e 3 ,
(11.26)
- r sin 0
a2 =
+ r cos 0 e 2 ,
and
W ith
( dr ) + ( dz ) = ( ds )
or
r 2g + z 2 = 1
th e derivative of z is
1 - r
(11.27)
207
2 Q .
- 2Q.
2 ,
a2 =
2 . 2
2
2.
2
a 22 = a 2 a 2 = r sm v + r cos v = r
Thus, one obtains
1
r2
(11.28)
r2
= r
(11.29)
0
, o
J/ r .
(11.30)
= r,ss c o s tf e , + r ;8S s i n t f e 2 + z e 3 ,
ai,2 = r,tfs
a2 2 = r ^
= - r cos &
- r sin i?e2
bn
r #S B cos i?r
sini?
, SB
zi S I
r 8 cos i?
r sini?
- r sin i?
r cos i?
208
r
1 - r:
Ai
= r ,
r #s
'
r ,sa
2
1 - r .s
1 - r ,s
o
(11.31a)
0
~r / 1 - K
( b p ) = a7 b ^
(lim b )
/ 1 r .:
0
and
. V l - ' r- r
Since a12 = a21 as well as bJ2 = b21 vanish, the coordinate lines are also
curvature lines, i.e. lines with extremal curvature.
The mean curvature results from ( 11.22a ) as
rr
A Z eL 1
/
1- A
+ r
2rJ
's
- 1
(11.33a)
1 rl
IS
^ , 86
(11.33b)
K = b = ~
22
,
a
' **2 A, 2
1
=
d
^
\
a A2,2
, 2.
(r } =
a 22,A)
1
2
11
&
22,1
209
( 0
0
II
r
_S
r
-rr.
r'
r
(11.34a)
A A
/
- r
1 - r:
(UMb)
AA
A A
11.2
B a s i c t h e o r y o f s h e lls
[0 .9 ,0 .1 0 ,0 .1 1 ,0 .1 2 ,0 .1 4 ,0 .1 7 ,0 .1 8 ,0 .1 9 ]
(11.35)
(11.36)
(11JS7)
e 3 = a3
T he three-dim ensional base e is transform ed into the base
of th e midsurface by m eans of th e shell tensor or shell sh ifter fj, introduced by
NAGHDI [ C.17 ].
210
, -l.o
= (^
) p
P
*
= a
(11.38)
~ (b \
-Cb I
l - ( b ]
(11.39a)
211
F p U ^ C ) = rp( f C) + v p U ,C)
with the vector of displacement v p
v p U ,C) = v ( ) + C w ( )
(11.40b)
is described
b)
( 11-41)
212
S tr a in ten so r
CAUCHY-GREEN s stra in tensor according to (4-12b) lim ited to sm all
strains ( lin ear th e o ry ):
( 11.42)
A fter transform ation to th e m id-surface of th e shell we obtain
a ap = T ( v i e
( HAS )
V/? L - 2 K p w )
b6 v p\ot
I + 2 b a.f b op w ')
(11.44 )
A lterations of curvature an d to rs io n ;
'Tsa =
=
(11.45)
+ W,c + ba Ve )
Shear strain s .
S tre ss R e s u lta n ts
In analogy to th e plate problem, th e three-dim ensional shell problem is re
duced to a tw o-dim ensional problem of th e m id-surface of th e shell. R esul
ta n t forces and m om ents are introduced instead of th e stresses w hich are
obtained by integrating th e stresses over th e shell thickness.
+t/2
M em brane forces
is
ppP
g T ae dC
N^ =
(11.46a)
- t/2
+t/2
T ransverse sh ear forces
p. T
oc3 j >
dC ,
(11.46b)
- t/2
+t/2
P ae
M o T CdC
M om ents
(11.46c)
- t/2
N^ and M ^ a re nonsym m etrical because of p P .
In the theo ry of shallow shells one can approxim ate p *
+ t/2
M aP =
TaP( d (
(11.47a)
- t/2
+ V2
a p
p
-
t/2
CtP 1/
dC
(11.47b)
213
~ aB
N
is called th e sym m etrical pseudo tensor of resu ltan t m em brane
forces, an d it is valid th a t
( 11.48 )
= Nq/3 + hP M ae
Q
E q u ilib riu m c o n d itio n s [ E T 2 ]
N\
=
- Q bf + PP = 0
(U.48a)
(11.48b)
+ P = 0
- O' = 0
=
(11.49a)
= G t a %
(11.49b)
(11.49c)
M aP
stra in stiffness
bending stiffness
sh ear stiffness
Gt =
Et
1 - v2
Et
12(l-i/)
(11.49d)
Et
2 ( l + i/)
(11.49e)
S h e a r-rig id s hel ls w i t h w e a k c u r v a t u r e
214
ot
~ - w , a - heot v g
is su bstituted by N
( 11.50a)'
'
, th a t m eans
= N ^ .
(11.50b)
+ w /jL) = - [ WU + | ( bve X + t K O U
Kp
(11.51)
Qia +
M *|
- Q ^ O
(11.52)
= 1 - ( VJ p +
- 2hap w )
(11.53)
Kap = 0 * 0 - - wL
' }
(11.54)
M ^ = K
12 M e m b r a n e th e o r y o f s h e lls
12.1 G e n e r a l b a s i c e q u a t io n s
A ssum ptio n '.
215
Equilibrium conditions
+P^ = 0
N ^ /j + P
=0
( 12.1)
These are three equations for three unknown resultant forces N ^, i.e. the
membrane theory is sta tic a lly determ inate. The resultant forces
can
be calculated from the equilibrium conditions ( 12.1) alone and the defor
mations from (12.2) and (12.3).
- Strain-displacement relations
a *p
= i ( vi e + V/?L - 2 ba^ w )
(***)
- Constitutive equations
N* = D H W %
with
(l j a)
1
= eT
xtY^
-i-v
( 12.8b)
with
^>ap l8
( &ocS
1 2. 2
f t D
(12.4)
Coordinates:
s or
i?
ip
in meridional direction,
in circumferential ( latitudinal) direction.
Derivatives:
d / d s = ( ),s ,
d/di? = ( ),tf .
- Equilibrium conditions
( r N ss),s
( rNs ,) ,B
+
+ N^ , ^ + C0SV,N ^
+ rpB = 0
(12.5a)
+ r Po
( 12-5b)
216
of revolution
Kj = centre point of curvature
Case 1: A xisym m etrical loading
P = 0
IT
'* =
( 12.6)
N
N
ie . +
= p
rl
r2
A fter elim ination an d integration, (12.6) leads to
<p
N (pip
i
r2 sin2 (p
J Fl r2 ( p cos tp -
(12.7a)
tp 0
N vv
,, =
r2
, *
p
N tfiip
(12.7b)
2 p ^ m ( s ) s i nm1,
m= 1
2 P m ( s ) cosmi y
m= 1
oo
p(s, tf)
(12.8)
217
N_
N.d-d
2 NM m ( s ) cosmi:>
m= 1
(12.9)
oo
N.
2 Ns^m ( S ) s i nm1 ,
m = 1
C O S ^ N _m +
r P sm
_
>
( 12.10)
= 0 .
N.
=
Pn
(N
) + (1 H -)cot<z>N
+ m
r. pm c o t y - r. p
v Wm ><v
\
r2 /
w m
r 2 sin ip
1 m
1
(
)
+ 2 cot w N ,
v v v m >.<p
r
( 12.11)
N
H T'Pm = m r .
- r, p ,
sin <p
1 sin p
1
Special shells:
1) Spherical shell ( r j = r2 = a , r = asin<>)
( s i n y ) N v v )iV -I-
(12.12a)
( s i n ^ N ^ )tv
+ cos ip
(12.12b)
+ asin^p^
= 0 ,
Nw + N
= Pa
(12.12c)
Ng
a._,_ =
'w
t
= aJdd
21
(12.13)
218
^xtf ,x + Y
= o ,
f0 + Ptf
N
(12.14b)
(12.14c)
Nxx
Poa
t = 2t
N
t
>
II
e,
Stresses
= pa
(12.14a)
>
e*
B o ile r fo r m u la :
(12.15)
3 ) C ircular conical shell w ith sem i-angle a ( r = s sin a , cos <p = sin a ,
ds = T jdip] see Fig. 13.3)
( s Nss ),s + gjn Q
(s
(tf+ N S0 + s p^
),s
sjn Q
+ s pg
= o
(12.16a)
(12.16b)
= s p ta n a
(12.16c)
1 2 .3 E q u i l i b r i u m c o n d i t i o n s o f t r a n s l a t i o n s h e l l s
Hyperbolical shell
C onsidering th e class of tran slatio n shells we restrict our tre a tm e n t to th e
special case of a hyperbolic shell. T his type of shell has a w ide-spread appli
cation especially in th e design of cooling towers. Due to th e ir negative
(hyperbolic) curvature, th ese shells display a bearing behaviour th a t differs
decisively from th a t of shells w ith a positive (elliptical) cu rv atu re (e.g. spheri
cal shells, elliptical shells of revolution).
219
^ = r i r2 ( p c o s y - p ^ s i n ^ )
,
(12.17)
( r 2 s i n v N v # )<v + r j c o s y - r a NW f#
- - r i r 2( ptf s i n p + p >#) .
U = r 2, sin p~ N tpip
V = r2sin 2p
I
j
(12.18)
r
v , + f 2 s in y U v
(12.19a)
V"f+
= - r i r 2s i n V ( p # s in y + p ^ )
(12.19b)
We now consider only the homogeneous part of the two differential equa
tions ( 12.19). V can be eliminated by differentiating ( 12.19a) w ith respect
to p and (12.19b) with respect to i), and subtracting the equations. We
then have
2
r2 sin 2 p U w + r2 r 2 U w +
sin p ( ^ ~ sin p \ v U v = 0 .
( 12.20a)
W ithin the classification of linear partial differential equations of second
order w e write
A U ,w + C l) ,w + a U v = 0 ,
w here A, C, a
nan t
(12.20b)
D = A C = rjT jS in 2^
eq. ( 12.20b) exhibits a different solution behaviour, w here the decisive fac
tor is w h eth er th e product of th e radii of curvature rt r 2 is positive or ne
gative. T h e following cases sh all be co n sid ered :
a) Tj r 2 > 0
b) Tj r 2 < 0
220
M echanical in te r p r e ta tio n :
A principle way of solving th is problem is to re-transform th e p a rtia l diffe
ren tial equation into two ordin ary differential equations by using a sep ara
tion approach ( see [ C.18, C.19 ] ) :
- S o lu tio n s o f d iffe re n tia l eq u a tio n s o f the ellip tica l type w ith r t r 2 > 0
(sp h eric a l shell, elliptical shells of revolution) a re such th a t discontinu
ities of th e boundary values occuring in th e case of point supports do not
propagate into th e in n er regions but a re confined to a narrow boundary
zone.
- S o lu tio n s o f d iffe re n tia l eq u a tio n s o f th e hyperb o lical type w ith r^ r 2 < 0
display a com pletely different behaviour. T hese solutions a re associated
w ith curves on th e shell surface, so-called ch aracteristics, along w hich dis
continuities of boundary conditions propagate over th e en tire shell [ C.2 ].
T his problem occurs in p articu lar w ith hyperbolical shells w ith single sup
ports. In th is case, th e m em brane theory is not sufficient for determ ining
th e state of stress; bending deform ations an d resu ltan t m om ents m ust th e n
be considered by an extended theory.
12.4
D e f o r m a t io n s o f s h e lls o f r e v o lu t io n
=
( u ' V + w )' ,
^
v
(12.21a)
Hs
= 7 (v
,tf
(12.21b)
TW
= T u .0 +
7 7
- 7 c os ^
(12.21c)
Special sh ells :
1)
Spherical shell
(1212a)
= t ( u -* + w )
=
'V
( ^
v^ + u c o t^ + w )
= T ( - i I ^ u. * + %
- vcoM
(12.22b)
'
(1212c)
(12.23b)
(1213 c)
221
s >
= f
s ' sm a
+ U + W cot Q )
_1_ y
s sin a
(12.24b)
>8
(12.24c)
D [ e*> +
C1 +
^ ) a T <S*]
w ith
= D -v
(12.25)
Et
D =
1 - i/
or
= rrr(N
Et
+ a
(12.26)
1^,0
_ 2(1 + v )
Et
+ P(x)y + Q(x) = 0
c]
(12.27)
U = 2 W t\K o +
tptp ww
tfitfi +
<p&urff
ted )
(12.28 a)
( 12-28b)
222
13
B e n d in g th e o r y o f s h e lls o f r e v o lu tio n
13.1
B a s ic e q u a t io n s fo r a r b it r a r y lo a d s
D erivatives: d/ di p = ,v or d / ds = ,s , d / d f l =
(13.1a)
,0 + r j e o s ^ N ^ + T j S i n ^ Q ^ = - r r l P 0
(13.1b)
= - r r lP
(13.1c)
(v r N v v
+ ri ^ip-S.S - Tj C O S ^ N ^ + r Q v
>.*
( r V
),lp + ri
( rtO
>H> + ri
( rM
>.*
Fig. 13.1:
'' r
ri sin V N W
- Tj COS if M m
= TTi %
(13.Id )
+ ri
+ ri cos V
= r r iQ^
(13.1e)
223
= X,. ,
= - r ^
( ls -2a)
+ ^
X,# + ( r
(1M >)
)>B - 2 ip cos v?
( 13.2cJ
>$
u - w
X = ---------
in m eridional direction ,
( 13.3a)
v sin tp - w ^
ip = -----------------'
( 13.3b)
Special relations for w eakly curved shells of revolution w ith (11.53) [ C.4 ]:
6c <p<p = V
(
) P ,
fj w 'v Ji<
(13Aa)
( 1SAb)
= - 2 ( w,
(13Ac)
w f# )
- M aterial law
Nss
= D [ ess + v e M - ( 1 + v ) a T 0 ] <
(!3.5a)
NW
= D [ M + " ss - ( 1 + " ) a T 0 ] -
( 1S5b)
Ns*
M ss
K K s
(13.5c)
- ( ! + l' ) a T1e>]
Mw = K [ k m + v k m - ( 1 + v ) ^ 1} ,
^
= K ^ - k
Et
w ith D = -------------- ,
1 - i/2
rr
Et3
K = -------- 5
12 (1 - v 2 )
according to (11.50a)
6
1
(ISM)
(13.5e)
224
= Q, = 0
v =
^ =
= 0
(13.6)
~ ti cos V n #o + T%
= -thPv>
( i s -7a)
( r Q v ),v
r N w - ri s i n ^ Nw
= rriP .
( 137b)
- Tj COS ip
( 1S-7c)
),V
( rM W
r r i Q v,
e<
p<
p = 4 ^ %
w )
(13.8a)
(13.8b)
Kss = X,.
or
(13.9a)
Kvtfi = X, ,
cos w
"w =
(13.9b)
D [ ew
=
M
K [L ac<p<p
<>
II
v e M
(1
i / ) a T ]
(13.11a)
(1
v ) a T ]
(13.11b)
(1
i / ) a T 1 ]
(13.11c)
V K
( 1 +
v ) a T * ]
k m
(13.11d)
225
Special sh ells:
1) C ircular cylindrical shell subjected to axisym m etrical loads
D erivative: d / d =
w ith =
3,
- E quilibrium conditions
= 0 -
(13.12a)
Q x .f - Nw + a p = 0 ,
(13.12b)
( 13.12c)
Nxx,f + a Px
= 0
aQ x
(13.13)
w
a
(13.Ua)
(13.14b)
B ending angle
* = T w-f
(13.14c)
- M aterial law
+ Um
- (1 + v ) a T ]
+ " exx
- (1 + v ) a T ]
Mxx = K[xx
+ vkm
- ( 1 + v ) a T 1]
+ ^xx
- (1 + v ) a T * ]
II
^>
= D fe
L xx
Nxx
(13.15)
< 4
+ 4k w =
w ith
a P
/ii
\ 2 l/-\
. i 4
0/^
j / - (1 + i / ) a a T & + 4 k a a T &
k4 = 3 ( l - i / 2) ( - y )
( k = decay factor )
(IS.16a)
( 13.16b)
226
2 k2 K
a D
R cos
K
2/c
N.
2)
M ( cos k
- sin
Qx
R ( cos
( sin
() -
( 13.17a)
je
K^ ,
( lS .l7 b )
) j e
*^,
( 13.17c)
k )
M sin
j e *^
ff5J7d,1
Q ,tptptptp
+ 4 a Q
E ta
K
w ith
- v
( 13.18b)
1. T hen, the
With
+ 4 k , QS = 0
-4
/,
2 \ cot a
K, = 3 (1 - V ) -
(13.19)
227
S p h erica l sh e ll
(lS.20a,b)
2K k
Xc
2K k
R + M
.
A rn =
s
Xoc
R a3 . 2
M a2 .
------- 5- sin w. H---------- 5- sin ip,
2K k
2K k
Yl
Ma
Ra
2
s
i
n
^
i
T
OT
2 K/c
(lS.21a,b)
228
4)
1 3 .2
S h e lls o f r e v o lu t io n w i t h a r b i t r a r y m e r id io n a l s h a p e T r a n s fe r M a trix M e th o d
( r N ^ Xv
th is system of differential
of elasticity (lS .llb ,d ) and
we obtain six differential
quantities:
+ w ) cos y
rr,
(13.22)
+ w )s i n + r Nw - r ^ p
(' r M w )>,<p
x c o s ip + i/ cob tp x iV
= K
+ rriQ v .
= u - r,x
ri
229
,
(13.22)
,
ri
M _
v x cos ip
rM W u - w - *)
(13.23)
(13.24)
-1
rl
2
D cos ip
t ri
\
siny1+ vjcosip
ri
D cosi^sinv?
(h
\
siny> +v)sm<p
rl
1 rl
D r
0 -L.1
r,
2
K cos ip
A =
VCOS,ip
- 1 - v sin ip
r
1
h
- VCOS ip
T
b 24 = D^sin</?
b 36 = K u c o s t p
230
r r i P v . - r r j P , 0, 0, 0, 0
- B i) _1( A j y ^ i + P i )
(13.25a)
231
(13.25b)
T he sta te
vector y ; still contains th e rad iu s r of th e
i-th subelem ent. As
th e rad iu s
m ay differ from one subelem ent to another,
it m ust be elim ina
te d from th e vector w hen using th e tra n sfe r m atrix procedure. In addition,
the load quantities m ust be included into th e vector, and we therefore re
place y ; by a new state vector z; defined as
ZT = ( Nw > Q*,. Mw . u > w * X , i ) ;
(1S.26)
(13.27 a )
T he tra n sfe r m a trix C j shall not be w ritten explicitly here as it can be deriv
ed from ( IS.25 b) .
T he conditions of continuity an d com patibility expressing th a t a t the point of
tran sitio n betw een two elem ents equal forces an d m om ents are tran sferred
and th a t equal deform ations m ust occur, finally yields the tra n sfer procedure
betw een th e boundaries i = 0 and i = n :
zn =
II1 C j Z j ^
= Cz0
(13.28)
T he above m a trix equation represents a set of lin ear equations containing six
equations w ith 2 x 6 unknow n state quantities a t both boundaries. By giving
2 x 3 = 6 boundary conditions at the beginning and a t th e end of the shell,
one obtains a solvable set of equations for th e boundary quantities.
E x te n s io n to sh ells w ith large deflections
If large deform ations are to be tre a te d by a purely linear m ethod, th e sin
gle step procedure proves to be very suitable. Here, the load is applied in
crem entally, an d th e to tal tran sfer m atrix is recalculated a fte r each incre
m ent. W hen using the tra n sfe r m a tr ix procedure, one proceeds from the
equilibrium conditions of th e undeform ed structure, where the position vec
tor r ; for th e i-th shell elem ent is assum ed to be constant, b u t shall be tre
ated as a function of th e displacem ents u an d w . Since th e position vec
tors Tj of th e deform ed stru ctu re cannot be determ ined analytically, m atrix
C can only be calculated for an u n d efo rm ed stru c tu re . Thus, equation
( 13.27a) becom es :
~ 0
,0 , 0 . 0 . ~ 0
z i =
, A p ) z i_j
.
Thus, we obtain as a tra n sfe r rule (Fig. 13.5):
^ k
^k /
k- 1
z i = c i(rj
k- 1
. k x^ k
. A p ) * i_.
(13.27b)
232
w ith
Arj
k/ k
k 'i
= A r ; ( uj , Wi
) .
Step 2:
Step 3:
in ( 2 ) ) .
k
T his procedure is repeated u n til th e sum of th e load steps A p equals the
to ta l load to be applied. T hus, th e n o n lin ear load-deform ation-curve is ap
proxim ated by piecewise lin e a r sections as shown in Fig. 13.6.
Since th e equilibrium is established by th e deform ed structure, a correction
is not carried out, an d thus th is procedure has th e disadvantage th a t the
approxim ated solution deviates from th e exact solution w ith increasing
loading. O n th e oth er hand, th is procedure is characterized by num erical
stability an d by a sim ple realization since th e stru c tu ral analysis program
does not require any m anipulation.
233
w ith =
d / dtf S
N x x,*
= -ap x
: l
= a p*
+
X ( Mxx
+ Mx^
+ M^x j # +
'QQ ) -
= - ap
(1S19)
- R esu lta n t forces - displacem ent relations from (11.53) and (11.54)
Nxx = T [ U' +
XT
+ W)1
D 1- v ,
K( \ - v ),
>
+ w + v n -t> - a
Mxx = - - a2 - [ W, +
(13.30)
" 2 v -* w + " w . )
- 2v,tf - W)]
, 1 v
u + j
, 1+ v
+ j
,
+ i/Wff
a Px
= -------g -
^ ,,,,.1
(13.31a)
1+ v
2
u . ftf
v ,-d-d +
V <&
2 ^ [ ( ^
+ w , r 2 k ( wi W - wrtf + wifftf)
2 v . w l
= -
a2 p
D
( 1SSlb)
234
UU,( + V,t? - 2 k ( v , * w -
+ w +
2
k (
AA
- 2
w -
a p
D
( lS.Slc )
,2
w ith
2 v w
( 1S.SU )
a.2 D
12 a 2
u = v = w= w, = 0 .
- F ree boundary at x = c o n s t:
n
= n x # = m xx =
q x
= o
( 1S.S2 b)
= c o n s t:
( I S M a)
u = v = w = w^ = 0 .
- Free boundary a t d c o n s t:
N
(lSJSSb)
M xl? ,0
(lS.Si )
b)
(1S.S5)
(1S.S6)
=
)
235
W )]
= ( V,tf + w + i / u f )
N*>
= N. x = T J T i i ( u^ + \ t )
(1337)
M xx
Mw
K
- r
a
( W,
>
1
V
1- V
2
+ V
v ,w
u .e*
u ,f
+ v,* +
, 1
+ 1/
v .e* + " w .f
1 - V
v ,ff + w ,*
w + k AA w
(13.38)
= 0
(13.39)
3.
AA w +
t
(13.40 a )
(13.40 b)
236
c)
u =
oo
^ u m c o sm ti > v =
m= l
oo
2 vm sin r n '^ , w =
m =l
wm c o s m d
(13.41)
m= l
vm = V e ^
Wm = W eAf
(1*4*)
k4
- y L + 6 m 2(m 2 - l ) ] A 4
2 m 2 [2 m 4 - ( 4 + i / ) m 2 + ( 2 +
(13.43)
) ] A2 + m 4 ( m 2 - 1 ) 2 = 0 .
T his fo u rth order equation in A has four com plex roots. We thus obtain
solutions for
m > 2 ,
m = 0 and
(1S.440-)
m = 1 .
(13.44b)
(13-45)
( ^ ^ -1- 6 m 4 ) A4 - 4m 6 A2 +
m8 = 0
(IS. 46)
237
T h e o ry fo r fa s t decaying b o u n d a ry d istu rb a n c e s
= 0 >
= 0 ,
i
-
Mx x , - N^
(* /* s
().( .
().*)
= 0 ,
(IS.48)
u ,tf + v ,( = 0
-
= o
(1S.47)
XT
+ W) (13-49)
M
Nxx
NXJ>
Mxx
Owing to ( IS.48) , no m aterial law can be given for Nxx an d Nxtf. T hese two
m em brane forces a re obtained from ( IS.47) .
Substitution of (13.49) into (13.47) yields after some re-calculation the
six th order differential equation
w . + 1 i r ^ w - = 0
( l s -5 )
(13.51)
238
w ith th e roots
^ 1 , 2 ,3,4 = Ki i ^ i
w ith
Ki ==
Ki
jr n
A*i == v/ ~2
/
2 V
~ kk '
(1S.52)
= 0
_1
a
1
a
N*? -
= 0 ,
(13.53)
T he stra in
an d th e shear stra in 7 xj? vanish. T his requires state
m e n t of th e following couplings betw een th e d isp lacem en ts:
v + w = 0 ,
I
u .o +
v ,e
(13.54)
N XX =
(1S..55)
+ W)
N xx
>
N xtf
Nw
Solution th en leads to
1"k
v2
+
w .dddddddd +
2 w ,dddddd +
w ,dddd
(13.56)
239
( m 11
(1S.57)
1 , v A4 + m 8 = 0
(lS.58a)
th e sm all roots a s :
k 2 if i 2 w ith
' 5 , 6 , 7 ,8
k2
1 v
( lS.58b)
v.e
Ux
= -^ -N
Nxx,e
.
a
xx
( 1SJ59)
r= M
Nx*,f = 0
((N ^ W r-fN ^ a
K
3
V,
)
(IS.60)
a
i
ft
i+
Js
240
E Ir
+ 2v
+ V'M )
(1S.61)
a
E quations ( 1S.59) a n d (13.61) suggest assem blage of th e essential field
quantities u , v , Nxx, Nxtf in a state vector, an d to solve th e problem by
m eans of th e tra n sfe r m a trix procedure as described in Section 13.2 [ E T 2 ].
f)
- T
- ( v .#
N.XI?
( u ? + v . ( )
=
K.
r W.
(13.62)
a
K
M.t? t?
M XT?
( v itf + w )
2 W,T?T?
_
xt?
241
14 T h e o r y o f s h a llo w s h e lls
14.1 Characteristics of shallow shells
Shallow shells possess a very large c h a ra c te ristic shell radius or, in other
words, a very sm all, non-vanishing shell curvature. Therefore, a typical beha
viour of such shells also occurs, nam ely th e su p p o rt of transverse loads on
th e m id-surface by m eans of m em brane forces. T his effect has already been
described w ithin th e scope of m em brane th eo ry in C hapter 12.
In addition, th e th eo ry of shallow shells does not neglect com pletely the
tran sv erse forces a n d bending m om ents, but considers them in th e equa
tions of equilibrium of forces p erpendicular to th e m id-surface, as well as
in th e equilibrium of m om ents. Thus, we a re no longer dealing w ith a sta
tically d eterm in ate system , as was th e case in th e m em brane theory, and
th e com putational effort for solving th e shell problem therefore increases.
In th e following, however, it will be shown th a t th e effort does not exceed
an acceptable lim it in com parison w ith a tre a tm e n t by th e com plete shell
theory [ C.7, C .8, C.20 ].
Fig. 14.1:
242
OtJ3
M'
(14.1)
P
~ P
V +
P
W zl
(14.2)
=
+ P
M l. - <f
0
(UJ)
= 0
af3
+ z U wL ) >
(14.4)
(14.5a)
(14.5b)
243
1 2
Fig. 14.2: Projection of a m id-surface onto th e x , x -plane E of th e threedim ensional space
R eduction of th e num ber of basic equations follows in analogy w ith (7.1S)
Nap
and
= ea l / S$\i6 - Pap
(14.6)
(14.7)
PaP\a = p'3 .
,aP ,
K a ^ a 75w |Q/375 - e 7 / 5 z L
# \
Lys -- pP + P'
ap9
lap
= 0
(Hi)
+ E
= o
A A + E t Q 4 ( z , w ) = F2( x , y )
w ith
AA = (
)>xxxx + 2 (
) xxyy + (
) yyyy
bipotential operator,
244
0 (f)g )
a nd
(xx&,yy+ ^.yyS,xx
^^.xy^xy '
F 1( x , y ) = p - z xxJ p x dx - z yyJ p y dy
F2(X>y) =
( j P x d x ),yy +
(J P y dy ),x x
"j
,
(14.10)
~ " ( Px ,x + Py.y) -
Special c a se s:
1) C urvature an d disto rtio n vanish in one direction
-
z,xx = z,xy = 0
>
= p ,
A A $ + E t Ky w xx = 0
px = py = 0:
( 14.11a)
( 14.11b)
z ,xx = z
,yy =
z ,xy = 0
= 0
(14.12a)
.
(14.12b)
245
Q
xe = Q
x +1 M x y . y
^Qy e = Q + M xy,x
} '
<U,S>
(14.14a)
- Simply supported
Nxx = Mxx = v = w - 0
(14.14b)
- Free edge
Nxx = Nxy = Mxx = Q Xe = 0
14.3
(14.14c)
T his tye of shell can often be found in civil engineering applications, e.g.
as a ty p ical roof construction extended over a rectangular base (le n g th 2a,
w idth 2b). T h e m id-surface of th e shell is defined by z = z ( x , y ) , w here
th e following ch aracteristical values are a ssu m e d :
z
z,xx = Kx = const
/yy
= k = const
y
z /xy
v =0
(14.15)
>yy
y $ xx
(14.16a)
= P
A A # + E t ( / c x w yy + Ky w xx ) = 0
(14.16b )
(14.17a)
$ = - E t 0 4 ( z , V ) = - E t ( / c x 7p yy + Ky V>(XX) = 0
(14.17b)
246
( 14-18)
A A A A ip + ~ A * A * ip = ^ K
K
The approaches for the auxiliary functions ip (14-18) are also substituted
into the relations for the stress resultants (14-6) and (l4-5b), and we thus
obtain in Cartesian coordinates
= -E tA * ,S ,
N =
<f.xx = - E t ( - = , l f ' V+ S ^ )
= - E t i * ^ xx ,
= E t A * ^ >xy ,
Mxy = - K ( l - ^ A A ^ x y
(14-20)
m=l n=l
sin v
s i n - n -x
(U-M)
C.2 E x e r c ise s
E x ercise C - ll- 1 :
A circu lar conical surface constitutes a
special case of an elliptic conical sur
face, an d belongs to those conical sur
faces th a t can be described by moving
a gen eratrix (p a ra m e te r) along a di
rectrix y ( j?) (c irc le w ith rad iu s a ) p a
rallel to th e x ^ x 2-plane (see Fig. C -l).
T he position vector r of a point P on
the surface reads in param etric pre
sentation;
r= r(s,$) =
s sin a cos
-(- s c o s a e 3
w ith
s ,$
a = const
D eterm ine
a ) th e fundam en tal quantities of first an d second order ,
b)
S o lu tio n :
a)
(,#)=
s sin a cos $
s sin a sin &
s cos a
(1)
It then follows
a ,1 = r fS =
sin a cos 8
sin a sin &
cos a
a 2 ~ r ,tf
( 2a,b )
248
By means of ( 2a,b ) and according to (11J1), the covariant components of the sur
face tensor ( f i r s t fu n d a m e n ta l fo r m fo r the surface ) are calculated as :
a a /3
all = al ' al = 1
s* sin* a
0
(3a)
s2 sin2 a
a = \ &ap\ = s2 sin 2
The diagonal form of ( 3a ) ( a 12 = 0 ) implies that the parametric lines are mutu
ally perpendicular (orthogonal mesh). The contravariant surface tensor can be
obtained by forming the reciprocal values of the elements of the principal
diagonal, i.e.
(4)
( * * ) = ( a p ) _1 =
/ f ---------
b =
with the derivatives
1.1
3a 1
3s
(0 \
0
2,2
33
\0 J
1.2 _ 2,1
3s
f - sin a sin 3 ^
sin a cos 3
V
One obtains the components of the curvature tensor by formulating the scalar
triple products :
0
sin a cos 3
- s sin a sin <
0
sin a sin 3
s sin a cos 3
0
cos a
0
0 ,
Exercise C-ll-1
sin a sin &
cos a
- s sin a sin d
s sin a cos
- s sin a cos
b 22
- s sin a sin
b, =
sin a cos A
- s sin a sin &
sin a sin $
cos a
249
s sin a cos a ,
( ha p )
s sin a cos a
(5a)
(5b)
The form of the fundamental quantities allows us to draw the following con
clusions :
-
&
i f
Jll
1
1
a ^
OB
R ,
1
R
1
2
hi
= o
(6)
_^L = _ J _ c o t a
3>oo
22
The two invariants describe the curvature properties of a surface ( see ( 11.22a,b) ):
H = 4 a a/3'
a/0
K =
mean curvature ,
GAUSSIAN curvature .
This yields
H = K = 0 .
s cot a ,
( 7a )
(7b)
250
+ P - 0
+ N21|2 + p1 = 0 ,
(8a)
shall be written in expanded form. The resultant normal forces N 11, N21 are ten
sors of the second order, and their covariant derivatives are to be formed accord
ing to ( 2.35b) :
+ P* = 0
( 8b )
( a e/0'7 +1 7 e>P
a fl7,p
)
/
0 7 -e
(rie) =
0
0
0 s sin a
( r^ ) =
0
1/s
1/s
0
( 9a,b )
( 10)
(11a)
(lib)
(11c)
88
sin a
d N 8$
d "d
-
= 0
( 12)
= 0
Exercise C-12-1
251
* N 22 s sin a cos a + p = 0 .
N&&
- p s t an a .
E x e r c i s e C-12-1:
A shell of revolution w ith an elliptic m eridional shape (Fig. C-2) is subject
ed to a constant in tern al overpressure p0 .
D eterm ine th e m em brane forces in th e shell.
Fig. C - 2 :
S o lu tio n :
We take from analytical geometry the radius of curvature Tj for a point P of the
ellipse
1
a 2 b2
( a 2 sin 2 9 + b 2 cos 2 <p) 3^2
Assuming that
N.
vv
( a 2 sin 2 9 + b 2 cos 2 9
) 1//2
a 4 ,2
b
2 P0cos
( a 2 sin 2 9 + b 2 cos 2 9 )'
v?= 0
By means of the substitution
2
b2
sin
9 = z , 2 sin 9 cos 9 d 9 = d z ,
- b2
the integral can be transformed into a basic integral.
d9
252
and by assuming that r ( 9 ) = r 2 sin 9 , one obtains the membrane force in the
meridional direction
N
= P2 12
99
and the membrane force in the latitudinal direction by ( 12.7b)
w
_
r2Po
r2 Pp r2 _
/,
Tj 2 ~~P r2'
99
"
with rt = r 2 = a-g2 - th a t
P0__
a2
N
r2 \
2Tj
2b
For a > 72 b , i.e. in cases of more shallow shells, a compressive stress occurs in
the circumferential direction at the equator. An elliptic shell bottom reduces its
diameter when subjected to overpressure. In the special case of a spherical shell
with Tj = r 2 = a = b , the boiler formula (12.13) is verified in the form:
n
tfitp
Pna
n
y_
tW
2
Exercise C-12-2
253
E x ercise C-12-2:
A spherical boiler ( radius a , w all
thickness t) subjected to in tern al
overpressure p0 is supp o rted in b ear
ings a t its top a n d bottom points (Fig.
C-4). T h e boiler ro tates around th e
v ertical axis A-A w ith a constant an
g u la r velocity w.
D eterm ine th e ro tatio n al speed for
unset of yielding, assum ing th a t only
a m em brane sta te of stress exists an d
th a t th e deadw eight can be neglected.
Fig. C -4: Spherical boiler
N um erical values:
-3
a = 1m
t = 2 10 m
S o lu tio n
Besides the internal overpressure, a centrifugal load occurs in this problem. With
r = a sin 9 , the resulting load components in the meridional and the normal di
rection become:
2 . o
(la)
p = p 0 + ptco a s m 9 ,
p^ = p t o a sin
cos
( l b)
N 99 = - -2
I ( p cos
sin 9 __J
- p sin
*
) sin
d 9 ~ %
sin 9
9=0
I p 0 cos
sin
dip .
9=0
Fig. C - 5 :
254
( _ p c^ 2 9 + c )
N.
vv
(2)
sin2 <p
only depends
>
(p = 0
C =
vv
(3a)
The resultant forces in the latitudinal direction are calculated by means of equa
tion ( 12.12c) and by superposing the two load cases:
N m > = ^ f - + p t u 2a 2 sin2 <p .
(3b)
o v v = Poa
- ^
The maximum stress occurs at iz / 2 . Following the von MISES hypothesis, the
maximum stress can be expressed as follows :
3r = / o 2 +
02
10 2
<
0 j.
(4 )
With ui = 7rn , relation ( 4 ) allows us to calculate the rotational speed n for un30
se t of yield in g :
26.4 rev/sec
Exercise C-12-3
255
E x ercise C-12-3:
C alculate th e m em brane forces in a spherical shell (radius a) subjected to a
w ind pressure described by th e approxim ate distribution
p = - p 0 sin ip cos $ .
T an g en tial frictional forces occur in p ractice but will be neglected here.
side view
top view
2cos <p
)f<, - cos
9 Nw
+ ( Nw )'#
0,
= 0 ,
(1 )
= - p 0a s i n 9 cosf> .
Nw + N ^
By a product approach according to ( 12.9)
Nw
= ( 9 )cos
N ^
= Y ( 9 )Bin , Nw = 0 ( 9 ) c o s ,
(2)
9 0 ' + cos 9 0 +
sin 9 Y 1+2 cos 9 V - 0
sin
V cos
90
= 0,
=
0 + 0
( 3a )
0,
- p 0a sin 9
(3b)
.
( 3c )
By eliminating from ( 3c )
= - O - p0a sin
9,
we obtain
sin
9 O'
s in 9 + '
2cos 9O +
+ + p 0a sin
9 cos 9 = 0
+ 2 c o s 9 V + O + p 0 a sin 9
,
(4)
256
The form of ( 4 ) suggests introduction of the sum and the difference of the un
known functions as new functions :
F1 = + ' r , F
= - T .
(5 )
Xj 2 = 2 cot <p ^
implies
Pj
= p 0 a ( cos tp 1 ) ,
(7)
implies - .
The ordinary inhomogeneous differential equations of the first order with variable
coefficients ( 6 ) have the following solutions according to ( 1Z.Z7) :
F ..:. =
(8 )
J1
Ai dw
= e
2 In sin id + In tan w / 2
. 2
- sin
(p
tan
-fAjdv
e J
COt "2
= ---- 5
sin tp
<p
f>2dV
, eJ
-2
. <P
-f>2dv
= sin tp cot , e J
2
tan ~2
= .9
sin 2 (p
C, 1
cot
p . a ( cos tp + 1) sin 2 <p tan
d<p .
.
J u
I
sin <p
By means of
1 -f cos tp = 2 cos 2 ^
sin 2 tp = 4 sin 2 ^
co s2
>
dip = J li cos 3
sin 3 ^ d t p =
<Pd<p =
1
- COS tp
COS3 tp
If we substitute
o
2 V.
(9)
Exercise C-12-3
257
we o b tain fro m ( 9 )
^ .
/
1
i \ 1 1 + cos m
Cl1 + 0Pna ( cos V ~ To cos <P) J----sin
J3-----.
9
an d analogously
F2 = [C2 -
,
1 i
P0 a ( cos<P ~ T COS
1-
cos 9
sin39
'
= i ( F 1 + F2) =
= y
D j -(- D 2 cos 9 + 2 p Qa c o s 9 ( c o s 9 - y c o s 3 9
Y = - y ( Fj - F 2 ) = y
>
( 10)
D2 + Dj cos 9 + 2 p 0 a ( c o s 9 - y c o s ^ ) ]
( 11)
10
9= 0
a n d for th e second derivativ e
/
o
3 \
D j cos 9 -|- 2 p 0 a ( - cos 9 - 2 cos 9 sin 9 + c o s 9 J
= 0
J <p = 0
9 = 0,
Dj = 0
a n d thus, according to ( 11 ) ,
D2 = - y P o a
W ith ( 1 0 ) a n d ( 2 ) th e follow ing ex p ressio n s for thie
e m e m b ran e forces a re ob
tain ed :
\ COS <P
cos3 v )
258
E x e r c i s e C -1 2 -4 :
A hanging conical shell ( height h, conical sem i-angle a ) supported as de
picted in Fig. C-8 is filled w ith liquid of m ass density g.
D eterm ine expressions for th e m em brane forces in th e ranges I and II
show n in Fig. C-8. T h e deadw eight of th e shell can be disregarded.
p = 0 ,
pg = 0
Range II:
p = p g z = p g ( h 1 - s cos a ) ,
ps = 0
,
.
(la)
(lb)
Exercise C-12-4
259
The expressions for the membrane forces can be determined by means of the
equilibrium conditions (12.16) for the axisymmetrical load case
^ - ( s N88) = Nw
(2a)
(2b)
= p s tan a
Range I:
N88 =
N* * =
(2c)
N.
, 2 sin a
Tb - P g h 1l cosz5--a
(3)
We determine the constant C from the boundary conditions for s = Sj = hj / cos a
with ( 3 ) as follows :
2 sin a
cos2 a
C = l - Psh ;
Substitution into ( 2c ) then yields the following expression for the membrane force
Nss in range I :
N 88 = 46- Pw g6 -
h 31
s3a
sin a
(4)
s p g ( h j - s cos a ) ta n a
i a + C ) tan a .
(5a)
260
C = 0
Thus, we determine the following expression for the membrane force Ngs in range
II:
(5b)
Ngg = ^
( 3 hj tan a - 2 s sin a ) .
E xercise C-12-5:
A section of a casing has th e shape of a circu lar toroidal shell as show n in
Fig. C-10 ( rad iu s of th e circu lar section a, rad iu s from centre point r0 , w all
thickness t) .
At th e boun d ary tp tp0 th e shell is subjected to a uniform ly distributed
boundary load N0 acting in th e tangential direction.
S tate th e basic equations for determ ining th e displacem ents u and w for
th e section of th e casing.
Solution :
a)
We proceed from the equilibrium conditions for shells of revolution with arbi
+ r rj p v = 0 ,
U-V7
= 0):
(la)
Exercise C-12-5
261
With the angle tp relative to the axis of rotational symmetry, the radius of curva
ture rj = a , the distance r = a sin tp + rQ from the centre line, and the auxiliary
radius r2 = a + rQ/ sin tp resulting from the projection onto the centre line, the
following system of equations is obtained:
+
[ (a s in < p
N
a
r j N
^ J ^
rQ + a sin
tp
- a cos
tp N * ,
0,
( 2a )
( 2b )
a cos <p
rQ -f- a sin <p ( N w - N w ) = 0 ,
(3a)
tv ( a sin
a sin tp
(3b)
tp
N..
v>v>
a cos
Nt p t p ,t p +
* tp (\ i +
iin
a sin tp
v>tp =
a cos tp
------- (- cot tp N
= 0 .
j
r + a sin tp
T] w
P(<P)
i(fi + P ( <p )
N tptp = Ce"^P( v ) dv
Evaluation of the integral leads t o :
JP(<p)d<p =
rp + T s i n y
(4)
or
N tptp = C
Boundary condition :
4-
sin tp
N
sin tp
C * _____
1_________
sin tp (r Q + a sin tp )
( tp = tp0 ) = - N0
(5)
(6)
262
Thus we obtain
s i n <P0 ( r p +
v v
and
by
s i n <p ( r 0 +
in c lu d in g (
3b )
. V
a s i n ' tp
T h e s tr e s s e s a r e g iv e n
^
0
a v>v>
a s in y 0)
a s i n (p )
( )
by
N* ,
ow = - p -
= _p>
and
(9 )
--------------s.
M. = --------------------------r
C , ,
Y^
=0
( 10a )
u cos ip + w sin
r
a sin
<p -f, rQ
v( 10b )
( 10c )
ew = t < Nw - v N >
( lla )
ew = i t ( Nw - v N w )
( llb )
Solution of
(10 )
( 10a ) >
w s
a - u
( 10b ) >
w=
( a sin *P + ro ) ~ u cos *P
sin <p
By comparing we obtain
a sin <p - n iV>sin <p =
u v - u c o t < p = sw a - s ( a + -i ^
( 12)
Exercise C-12-5
263
u v -ucotq. = ^ T[(Nw - v N w ) - ( N w - v N w ) ( +
E t
1X
( a + v a + v ) - N ( a + v a + r-2 ) ]
sin tp y
sin tp / J
vv\
-)] =
(13)
'
'
Nn [
si n<P0 ( ro + asin(P0) U ( ! + v ) + v - ^
( 15 )
_ ginsintp
sin tp
_ sjn(p
j Q ( < p ) e / P M dv,d<p =
_
N /
" o lo fI a (vl + v ) + v
sin
tp ,
E t J ( rQ
a sin tp) sin 2tp ^
/ fa(l-t-v) +
^ 0*
0I v
sin0 tp
Et J
a sin 2tp
jj
( 16)
264
E x e r c i s e C -1 2 -6 :
A thin-w alled circu lar cylindrical shell w ith one end clam ped as show n in
Fig. C -ll is subjected to a sinusoidal distrib u tio n of tan gential m em brane
forces at its free end w ith th e shown vertical force F r as resultant.
c)
S o lu tio n :
a ) We assume that the vertical force FR at the free end of the shell stems from
the following sinusoidal distribution (see Fig. C-12):
- Nxj? = k sin R
Then
F*
7t /2
J -
tt /2
(1)
lI s
must hold. From (1 ) follows that k = f ^ / 7ra, and according to (12.14) with
= px = p = 0 we obtain the resultant forces as follows
Nv, v, = 0
F
sin R
N x t ,f = -----7 t
a
F
Nx x = Ja - ^7 t xa c o s f t + Cl . v( )'
0) =
(2)
Exercise C-12-6
265
This corresponds to the solution that would be obtained by the elementary beam
theory. By defining the moment of inertia for a thin-walled circular section as Iy
= 7ra t and the bending moment of the cantilever beam My = - FR x , the nor
mal stress at sections x = const is
Mv
FR*
7 ra3
a cos
x 4" cos $ =
t
ca^
N XX
t
= E F ( Nxx - v N *>)
Et ( n w - v n
+ W
4- V
+ v ,x
(3b)
(3c)
^ ( 1 + v )
E t
_E
j 2 ^ - c o s S + C2 ( ) ]
7ra
(4 a )
1
Et
_ 2 ( 1 + v ) 's t ; x s in ^ +
^ 3
1
Et
^ J ( 2 + v)xcos - - ^
! T s in ^ - "ir C2,.?+ C3 ( ^ )]
(4b)
(4c)
^ - c o s + f C2 M - C3(1?]
The two arbitrary functions C2 ( & ) and C3 ( & ) only allow the fulfillment of two
boundary conditions, e.g. u ( / ) = v ( i ) = 0 , instead of the fo u r boundary condi
tions for the clamped boundary u ( / ) = v ( / ) = w ( I ) = w x ( /) = 0 . Thus,
( 4a ) yields
u (I) = 0
( 4b ) yields
v (I) = 0
C3( ^ ) = 2 (
1+
(5a)
C2 ( ) = - ^ 7 2 ^ COS&
V) fn ra * sin^ -
c 3(&) = ^ s i n [ ^
Tta*5
2 (1
sin + -j*
7ta
+ v)-J
jl
6a
J L
2 a 2
4*r-sin$
(5b)
266
L it
cos =(= 0
a,
Sim ilarly,
( / ) ={= 0
1
The membrane theory cannot meet these essential boundary conditions and there
fore only yields an approximate solution as we have already seen in various exam
ples. In order to fulfill the essential boundary conditions w ( / ) = w x ( / )
= 0, a
bending solution has to be superposed onto the approximate solution.
From ( 5c ) we obtain for the displacement of point A :
w ( x = 0 , = 7t ) = wma x=
2(1 + v ) 4 - ]
(6)
When compared to TIMOSHENKO beam theory, the first term represents the
contribution from bending, and the second term the contribution from shear de
formation.
c)
= d U* ( F-*) = a u ( F j )
dFl
d Fl
(7)
V
Equation ( 7 ) applies to a linearly elastic structure. In the present case, the defor
mation energy according to (12.28b) can be employed. For the circular cylindrical
shell x s <p , s o :
U=
t J [ NXX+ N ^ - 2 I / N XXN ^ + 2 ( 1 + v ) N x% ] d A
(8a)
(8b)
A
We then obtain the displacement w by ( 7 ) with ( 8b ) as
x = 0
= 0
N*
2 ( 1 + v ) N -
i5 w
] d 8 'i '
'
<9)
We now substitute into ( 9 ) the resultant forces Nxx and Nxi? from ( 2 ) and their
derivatives:
W= f t
J J
x = 0
[ J t
x2cos2 + 2 ( 1 + v ) J ^ s i n 2 dx d
= 0
E17T L A
3 a 3- +
2 (1 + v ) 4 - l
Exercise C-12-7
267
E x e r c is e C -1 2 -7 :
A type of shell often found in civil and m echanical engineering is a ruled
shell as shown in Fig. C-13. Its mid-surface has the form of a special hy
perbolical paraboloid which is generated by moving a straight line g along
a rectangle ABCD. The rectangle lies in the x 1, x 2-plane and has the side
lengths lv l2 . The straight line moves along the line AD and along the hy
potenuse of the triangle BEC. This so-called skew hyperbolical paraboloid
shell is also termed a h y p a r shell and its parametric description is given
by
r ( x , y ) = x e j + y e 2+ ^
(x =
, y s f
, c = ^ )
b)
P3 = S
S olu tion
a) Equilibrium conditions
First, the fundamental quantities of first and second order as well as the CHRISTOFFEL-symbols have to be determined. Proceeding from the given parameter
description
r ( x , y ) = x e 1 + y e 2+ - ^ e
al = r,x = ei +
2 r , y ~
e2 +
Te3 c
e 3
( la )
(lb)
268
Similarly, the metric tensors are calculated according to (11.11), the determinant
according to (11.12), and the covariant tensor of curvature according to (11.18):
xy
c2
xy
c2
= M
/( a
1+
(2a)
(f)2
= 1+ (i)! + ( i f
(2b)
xy
c2
1 + (t )2
\ _( aQ
( ) \1 1
)
xy
( 2c )
1 + ( i f
C2
(3)
aP
c Va
y
c2a
(C >
0
WV-
( 0
c 2a
c2a
(4)
Nli + n21|2+ p1 - 0
A
+ n212 + P2 = 0
11
19
99
N bn + 2N b 12 + N
(5)
b22 + p = 0 ,
the covariant derivatives of the stress resultants are required. With the relations
( 2.35b) and ( 4 ) they become
11
2y
n 12|1 = n 1 2 1 + - ^ n 22 + - ^ - n 12
'
era
c a
( )
Exercise C-12-7
269
, , , 2l
12 ,
3y
N ,l + N , 2 + 2 _ N
, T12
, t22
, 3x
N ,1 + N , 2 +
, Tn
+~T~N
,,12 ,
T - N
+p
,,2 2
+ - f N
, , 12
-N
=0
+P
(7)
= .
+ p
(8a)
W )
(8b)
Z 5 l"
<)y
a 22
,2
JLiL
c4 a / ^ Van a22
d
dx
L l.
c4 a / ^ Van a22
6
dy
c2 a / a
22
"11
I ___
c2 a / i :
N*22 =
Nyy
N*12 = Nxy
Eqs. ( 7 ) finally yield the equilibrium conditions of the skew hyperbolic parabo
loid shell (d/dx = ,x , d/dy ,y):
270
77 =
2=
r r N + f / r = o , ( * , )
- | - N >y + P* = 0
. (sc)
Eqs. ( 9a,b ) are a system of first order partial differential equations with variable
coefficients. Eq. ( 9c ) yields the membrane shear force
Nxy = - f a p *
Nxy(y = - | p * - f a p *
and by substituting them together with the metric ( 2a ) into ( 9a,b ) , we obtain
after re-formulation two uncoupled differential equations for the two unknown
membrane forces:
p* + f
p*,y - ^ p * 1 .
V* + f a
P*,x ^
V*2
(ia)
( 10b )
First, the physical load components in the global Cartesian coordinate system
x1 ( i = 1 , 2 , 3 ) have to be decomposed into components both in the direction of
the local surface parameters and perpendicular to them.
a 3 is calculated from a j and a 2 (la ,b ) by forming the vector product according
to (11.16):
e l
+ e3 >
(1 1 a)
*2 ^ 2
cfe 3
(11b)
a l
y
a3
X
^
+ e3
( H e )
The above vector equations constitute the transformation between the local base
vectors and the base vectors in the Cartesian coordinate system. The latter vec
tors can be written in abbreviated form as
a,,
= 3 1 .^
Correspondingly, the vector can be written in different bases. The covariant com
ponents of the load vector with (2 3 a ) read, for instance,
With (2 )
t i t i = 81
Exercise C-12-7
X2
i + TCT
(< ) = !
_2
C
2L
c
c_ 2
V2
2
C
271
( 12)
By substituting the load components we obtain with (2.10) the physical compo
nents of the load vectors:
P* = ~ h
cE
/ a 22
a
(1 3 )
P* = - g -
2y
In ( x + -/c2 + x 2 + y 2 ) -f C ( y )
( 15 )
A ^ ( |y
and thus
(1 6 a)
In-
N xx = 2 6 y
/ 1+ ( t ) =
l + VT -
Nyy = 2 g x
In ------------------
/A
(16b)
r + / r
(16c)
TIMOSHENKO [C.24] and other authors have treated the same problem by pro
jecting the forces onto the x,y-plane, and then formulating the equilibrium. Their
results can be transformed, by respective measures ( e.g. Nxx = -/ a n / a 22 N n TIM)
into eq. (16 ). Given the prescribed boundary conditions, the load at the bounda
ries x = 0 and y = 0 only acts via shear. Thus, boundary stiffeners are required, a
fact that leads to incompatibilities between the deformations of the stiffeners and
of the shell boundaries. For this reason, the membrane solution has to be augmen
ted by a solution from bending theory. Further examples are treated in [ C.2 , C.8 ].
272
E x e r c is e C -13-1:
A circular w ater ta n k (ra d iu s a, height h ) h as a linearly varying wall thick
ness ( t 0 = m axim um w all thickness)
t ( x >= y
1" f )
W -M lJ V tF
(*->.
J ( N^ > +
M^ v ) dA j ( p " va + P w ) dA
( X)
Exercise C-13-1
273
J"(E xx exx +
e^,j
+ 2 N xij
uxx+
+ 2 M x^ o x#) d A - J ( Px u + p ^ v + p w ) d A
(2)
An axisymmetrical load case is given in the present problem, and the longitudinal
force Nxx vanishes. Thus, ( 2 ) reduces to
11 = 2" J ( N<WeW + Mxxu x x ) dA - J p w d A
A
(3)
With (13.14)
'
(*)
arid
Nw
= D ( l - v2 ) ^ - = E t ( E ) f
( 4a)
( 4b)
From ( 3 ) follows that
n = n(E,w,wfa)=
2t
h/a
J { y [E t(f) + K(
) ] _ Pw J dA . ( 5 )
+ ^ = o
Et5 - " p = 0
"
( 6 )
Et 2
_ pa4
! T a W~ K
4k 4
=0
w = 0
a 2
or
8w c = 0 ,
(
>
( 7a )
y J
274
= const
= 0
= 0
or
Sw
= 0 .
( 7b)
With dA =
2 ? ia d x =
2 7 t a 2d i
we
o b ta in
h /a
n = 2 7t I -7T E t n ( 1 - T1f= o
- g p h ( l - ^ - 5 ) a 2 w d ^
( 8)
.
The application of the RITZ method (cf. Section 6.7) requires that we choose an
approximation to w with linearly independent coordinate functions in such a way
that the essential, i.e. geometrical, boundary conditions are fulfilled. According to
(6.36) we choose an approximation
N
w* = 2 cn fn ( 0
n= 1
( n = 1, 2 , . . . , N ) ,
(9a)
= 0
, n = 1 , 2 , . . ,,N ,
h /a
{= o
( 10)
Exercise C-13-1
275
h/a
n = 1:
ci j K t 1 - i r ^ fi +
r 5 >3f ? . ] d^ +
=0
h/a
+
c2 J
[ Eto (1
h I ) f2fj +
a2 ( 1
h ^ ) f 2, *1,
dt, =
=0
h/a
= g p h a 2 J (1 -
h/a
n = 2:
= 0
h/a
= gpha"
J (!-F^f2dt.
= 0
0.5
2 w*
[10' m]
Fig. C-15:
[10 N/m]
[10 N ]
276
13
After integration and solution of the linear system of equations we obtain the fol
lowing coefficients:
_ g p (1 - v 2 ) h5
1008
Cl ~~
E t2
468 + 87 X + X2
_
g p ( l - v 2) h 5
E t2
w ith
2 1 (3 6 - X)
468 + 87 X + X2
X = t 1 - V2 ) ( ^ V )
a t 36 - x ) ( i - F * >
E t2
Vh
(1 1 )
468 + 87 X + X2
E xercise C-13-2:
A reinforcing ring 2 (cross-section b 3 1 , b <K 1) is to b e positioned in the
m iddle of a thin-w alled, long pressure tube 1 m ad e of sheet steel ( radius a,
w all thickness t) . F or th is purpose, th e ring is w arm ed up in such a way
th a t it can be slided into its position on th e unloaded tube ( see Fig. C-16).
At a te m p e ra tu re T 2 = 50 C th e ring ju s t fits th e tube in stress-free contact.
Cooling of th e ring to th e tu b e te m p eratu re of
= 20C leads to shrinking
of th e ring.
Exercise C-13-2
277
a = 1 ra , b = 1 m , t = 1.5 10
-2
0.3
E , = E = E = 2.1 10s M P a .
S o lu tio n :
The problem w ill be solved by means of the well-known, so-called M eth o d o f
T h e o r y o f S t r u c tu r e s (Section 13.1.4). For this purpose, we p artitio n the pressu
re tube and the reinforcing ring into three subsystem s ( 0 , 1
and 2 - s y
stem ) according to Fig. C-17. We can now formulate the com patibility conditions:
w() + W(1) + w W
w () +
w ^
(ia)
(2 2 )
(2)
(lb)
AF=
"0" - s y s te m
Z22
Z33
'1" - s y s te m
3 V Z22
M
M
2" - s y s te m
278
13
Tube 1:
( 0)
R a3
(*)
Et
x() = 0
(2) _
1
2K/c3
(l)
R a2
2 K x2
M a2
2K x2
( 2) _ M a
KX
(2a)
( 2b )
Ring 2:
v() = w ( 01) + w(2) = A + P b a " = A +
2
2
+ EA
4E t
v2
(3a)
(l) _ 2 R a 2 _ R a 2
W2 = E A
2 E tb
w<2> = 0
(3b)
xx (0)
2
<(1)
^2
.(2 )
^2
pa2 _
Et
0 -
Ra3
Ma2
2K x3
2 K x2
Ra2
2K x2
Ma
Kx
A +
Ra2
pa
+ 0
(4a)
= 0
(4b)
R = M
a
(5a)
Eq. ( 4b ) leads to
and ( 4a ) correspondingly to
A -
M =
3 p a
4 Et
ax
2K x2
E tb
(5b)
JVL
Et
a T2 ( ^ 2 -
Ps ( a + 2 t )
3Et
( 6a )
^1 ) =
a T2
(6b)
Exercise C-13-2
279
1:
_
3 Et
4
(8 )
a T26
and with s
A
4 a T2
(9 )
P i
<->
E T + a a T2
aK
E tb
2 k2 K
2 /t K
( 10)
Mn
[N ]
[ 1 0 3N /m ]
Fig. C -19:
K^
in the
280
Now we replace R by ( 5 a ) and substitute it into the relation for the circumferen
tial force Nm :
w = pa - ^
M ( cos K i; +sin K 1; ) e- K^
( 11a )
M (here
( lib )
Fig. C-19 depicts the curves of the resultant moments and forces for the given
numerical values.
Finally, we calculate the stresses in the tube and the ring :
- Tube 1
Longitudinal stress :
oxx = ^
xx + P ^
( 12a )
at both ends, since in
- 6 M xx( f ; = 0 )
t2
pa
+ 2t
6M , pa
t2 + 2t
Circumferential stress :
6 MXX1
= j v ^2
pa
=
2/c2
I T
M ( cos K
+ sm
, - Kt
K 5 )e
1 = 0
----- I - 0.35
we obtain
= 0 35 ) !, mlx
Numerical values:
o vv
m 156 M P a ,
max
o,,
v% ax rs 102 M Pa .
- Ring 2
Circumferential stress :
~ Pa , P a . 2 R a _ E g T2 e
^ 2 ~ 3 t + 4 t + 4 b t
4
Numerical value :
Ow2 82 M Pa
pa
41
kM
bt
Exercise C-13-3
281
E xercise C-13-3:
A pressure boiler m ad e of steel
consists of a circu lar cylindrical
shell (ra d iu s a, w all thickness
t ) closed a t each end by two
sem i-spherical shells (Fig. C-20).
T he boiler is subjected to a
constant in te rn a l overpressure p
(th e deadw eight of th e boiler
can be neglected).
D eterm ine th e curves for th e
stress resu ltan ts both in th e cy
lindrical shell a n d th e semispherical shells.
N um erical v a lu e s :
p = 10 M P a
, t = 0.1 m
1
E = 2.1 10 M P a , v = y
S o lu tio n :
Owing to the symmetry we only consider one half of the pressure boiler. As in
the previous exercise C-13-2, we partition the spherical shell from the cylindrical
shell and mark the single loads according to the 0 , 1
and 2 -system s in
Fig. C-21.
(0)
(0 )
r m
"2" - s y s t e m
282
Here, the compatibility conditions for displacements and rotations at the interface
become:
<> + > 4. W!> = <?> +
4 " + 4
(la)
4 + X<> + *?>
(lb)
- 4 > + 4 + 4
We substitute the single deformation values for the boundary loads (see Fig.
C-21); the deformations of the 0 -system are membrane solutions of the cylin
drical and the spherical shell:
pa2 ,
2Et ^
R a 3 , M a2
pa2 ,
2Kx3
2K/c2 _ 2 E t
,
R a3 , M a2
/
) + 2 K k3 + 2 K k2 *
. R a2
Ma
, R a2 , M a
0 + j K ^ - K ^ = 0 + ^ s
+ in r
x
*
/
( 2b)
( 3a )
Owing to the fact that the semi-sphere and the cylindrical shell exhibit the same
deformation behaviour at their boundaries when subjected to boundary forces,
and because no twisting angle x of the boundaries occurs subject to internal
compression, the compatibility of the deformations can be introduced by the
transverse boundary forces alone. Eq. ( 2a ) then leads t o :
R = - W
'
( 3b)
Mxx = f t t * e~ K ( s i n K t
Qx
Nxx =
I) ,
= 8k2 6
1 Sm K Wl
~ ^2
Fig. C-22 shows the behaviour of the stress resultants around the transition
between the cylindrical and the semi-spherical shell.
Exercise C-13-4
283
x[m]
N<p<p lN H [ l O N / mm]
10000
0.1
0.2
Fig. C-22: Resultant forces and moments in cylindrical and semi-spherical shell
E xercise C-13-4:
A thin-w alled circular cylindrical tu b e m ad e of steel (radius a, w all thick
ness t) as shown in Fig. C-23 is horizontally su p p orted betw een two rigid
walls in such a way th a t th e cross-sections a t b o th ends of th e tu b e are
com pletely clam ped.
D eterm ine th e stresses in th e tube due to its specific deadw eight g g , after
rem oval of th e m ounting equipm ent w hich ensures an in itial stress-free
state of th e tube. Use th e following num erical values :
I = 10 m
E = 2.1 10* M P a ,
a = l m
v = 0.3 ,
= 1 1 0 2m
g g = 8 104 N /r n
284
Fig. C-23: C ircular cylindrical tube clam ped horizontally a t both ends
S o lu tio n :
The complete solution is determined by superposition of a membrane solution
(Ch. 12) and the solution of the boundary disturbance problem (Ch. 13).
M embrane solution (denoted by superscript 0)
Using the abbreviated notation y for the
deadweight p g t per unit area of the mid
surface, the following surface loads are act
ing on the shell (see Fig. C-24).
Px = 0
Y cos 8 .
We obtain the following resultant forces by substituting the loads into the equili
brium conditions ( 12.14) and by defining 1; =
(la)
N - . = - y a cos 9 ,
N x i?
- ( 2 Y a ( + Dj ) sin $ ,
NXX = ( f a ^2 + Di + D2 ) cosd
(lb)
( 1c)
= | f ( nxx- v N )
V* + W
= t ( NW
u I ,.
U^ + v-f
... 2 ( l + v ) a
m
v N xx)
XJ?
(2a)
(2b)
(2 c )
Exercise C-13-4
285
u = t [ Y a ( T
2
+
D i T
D2?
+ Ds]0" *
( 3a)
= t [ ^ a (&4 - T ( 4 + 3 v ) ) + Di ( T - 2 ( 1 + v ) 0 + D2 T +
+ D3 I + D4 j sin 9
o
w
a
Et
[ Y
1 )
(3b)
(3c)
+ D2 (
] cos
=0
(4 )
- (2 -v f
(5)
(6 )
.
286
(x - V
) AJ +
+ XJCJ = 0
+ ( - 1 r Lx ) Bi +
cj
= 0
1
4p*
1
Bi = 4p2
Since X2 = -^ 1
- 1 + 2 v pj + i ( 1 -(- 2 v pj )
Cj = (
+ i *2 ) Cj ,
(7 )
^ + i(2 + v ) j C j
= ( p i + i p 2 ) Cj
A2 = ( 0Cj - i a 2 ) C2
B2 = ( P j - i P2 ) C2
(8)
N * ( i ) = 0
und
u ( 2 i ) = 0
with ( lb ) and ( 3a )
The remaining four constants C j, C2 , D2 and D4 result from the four boundary
conditions
u(0)
= u ( 0 ) + u1 ( 0 )
= 0
v (0)
= v ( 0 ) + v1( 0 )
=0
w(0)
= w ( 0 ) + w1( 0 )
=0
w, e ( )
= w ,e ( 0 ) + w ^ ( )
After carrying out the numerical calculation with the given values we obtain the
circumferential force as
= - 8 + ( 47.3 cos 12.91 + 5.02 sin 12.91 ) e 12 9f cos A .
Exercise C-13-4
287
Fig. C-25 shows the membrane forces according to ( 1 ) and the bending moments
Mxx und
acting along the top longitudinal line 9 - 0 of the shell. One can
see how fast the bending disturbance has decayed already at a distance of ~ 0.4
m from the boundary. The stresses are calculated from
Nxx , 6 Mxx
x*=
~ J 2 ~
~T~
( 9)
T 5-
'T '
-4N
[N /m ]
Fig. C-25: Membrane forces and bending moments along the top longitudinal line
of the cylindrical tube under deadweight
288
xxmax= 1.31 ,x
v 0.92 = 2.23 MPa.
o ,,
,
.
The numerical values show that both the longitudinal and the circumferential
stresses due to the boundary disturbances are of similar magnitude as the mem
brane stresses.
E x e r c i s e C -1 3 -5 :
w :
'
_ _x_
a
^yy
P a
^xy ~
f )
At buckling the component Nyy w;M must be included in the equilibrium condi
tion in the radial direction of the deformed shell . Then, u and v can be eliminat
ed, and we obtain in analogy with (13.39)
k A A A A w + ( 1 - v2 ) w
+ - ^ A A w |W = 0
(2)
(m,n = 1,2,3...)
x in the centre
(3)
of
Exercise C-13-5
289
fulfills the boundary conditions of the simply supported shell in the longitudinal
direction
(
27
) =
' (
5 7
) =
(4)
P = TT
P =
(5a)
+2n ^ + ( 1 v2) J 2 ( \ 2 . 2 \2
n
n (X + n )
(5b)
We now have to determine that combination of m and n for which p has the
smallest value. We can immediately see from ( 5b ) that X will attain its smallest
value for m = 1. The shell therefore buckles with one wave in the longitudinal
direction.
Assuming that
a =
>
2
we obtain
p = k n 2( 1 + a ) 2 + (1 - v2)n (l + a)
1-
kX4
(1+oQ2
( 6a)
(l+)2
l - v 2f
kX4
1 ,
3")
+ )
(6b)
1X2
kX4
= 0
1 - v: - ( - A ) + 3 2
a* = X 4
3 (1 v2)
( 7)
p = !(*) = 7 T ? f f /d - ) (1
(8a)
or with v = 0.3
pcri t 0 . 9 2 E f ( ) S/2
(8 b )
290
* = = x v ^ i 5 y
" ^ t i / F
E xercise C-13-6:
D eterm ine th e eigenfrequencies of th e free vibrations of a circular cylindri
cal shell w ith sim ply supported ends as shown in Fig. C-26 (w ith o u t exter
n a l pressure p ) . A ssum e sm all vibration am plitudes (lin e a r th eo ry ) and
solve th e exercise using DONNELL s theory.
Note: T h e coordinate system has in th is case been m oved to th e lower
boundary.
S o lu tio n :
In order to treat the circular cylindrical shell, the basic equations (13.31) are sim
plified in accordance with DONNELL's theory (see (13.38)):
u. + J ^
^
u.*> + i T
+ v, M + ^ v ^
+ v w .f =
(la)
+ w
(ib)
2
Vu f + V + w + k A A w
( lc )
t d2 v
Py = - e ^ r
A d2 w
f
{2)
where p denotes the mass density and r the time ( r is introduced in order to
avoid confusion with the wall-thickness t). This approximate theory neglects the
rotational inertia of the shell; its consideration would result in additional, very
high frequencies, whereas its influence on the lower frequencies considered here is
negligible ( see [ C.21, C.22, C.23 ]).
The eigenfrequencies are determined via separation of variables:
u=u(x,<p)sin(Jt
v = v ( x , <p ) sin u r ,
w = w ( x , <p ) sin g>r .
(3)
Exercise C-13-6
291
u +
~~2 ~
v i( + v w { = 0 ,
v.a + v +
vu +
v e- Xw 4-k A A w
= 0 ,
(4)
= 0
(B)
In ( 4 ) , n 7 V, W are
position functionsu ( x , 9 ) , etc. The following assumption
regarding the form of these functions
U cos n
tt
cos m
x
V V sin n cp sin m a
r~
mx
w = tit
W cos n 9 sm
(6 )
,i
m
tu
a
with m
I
fulfills all_ boundary conditions for a shell with simply supported ends
( w = v = Mx = Nx = 0 for x = 0 and x = l ) . Substitution into ( 4 ) yields a ho
mogeneous system of equations for the unknown amplitudes U, V, W. By setting
the determinant of the coefficients equal to zero, one obtains the characteristic
equation of an eigenvalue problem
X3 - AX2 + BX - C = 0 ,
( 7)
where the ^efficients A, B, and C depend on the dimensions of the shell as well
as on m ^ffd n .
Numerical evaluation of ( 7 ) shows that each pair of values of m and n defines
one lower and two higher eigenvalues which exceed the lower ones by powers of
ten. The technically relevant lower eigenvalue can thus be approximated from ( 7 ):
xr = -
<)
(9 )
( 10)
292
The eigenfrequency equation (1 0 ) consists of two termes, where the first stems
from extensional vibrations and the second from bending vibrations. Fig. C-27
presents a numerical example, where both terms are drawn separately in depen
dence on n . The curves clearly illustrate that for a small number n of circumfe
rential waves extensional vibrations predominantly occur, and for large n ben
ding vibrations, respectively. Close to the minimum, the two terms are approxi
mately equal. Therefore, no further simplifications must be performed for the
simply supported shell. Elimination of the first term in ( 9 ) , for instance, cannot
be admitted since this would correspond to an inextensional vibration. If defor
mations which are incompatible with the assumption of inextensionalbending are
prescribed at the boundary, bending and extension will act jointly, and thus have
to be considered by a complete shell theory.
The minimum of Xj can be calculated by a formula. Eq. (10 ) implies that Xj in
creases with m . It attains its smallest value for m = 1, i.e. the shell vibrates
with one wave in the longitudinal direction. Xj then only depends on n . If the
actually discrete number of waves n is assumed to be continuous, (1 0 ) can be
differentiated with respect to n :
- 7 :; 2dXl 2, 2 = - ( 1 - v 2 ) 7 - # - - 2T4 + k = 0 .
d(m + n )
v
( m 2 + n 2) 4
It follows that
( m 2 + n 2) 2 = m 2 - | / ( l
V )
10
(11)
11
12
13
Exercise C-14-1
293
Substitution yields (the two terms in (10 ) result in the same value)
X,
J M in
= 2 /( 1 - v 2 ) k m 2
' V
'
( t )s
< >
This proves that the lowest frequency depends on all dimensions and on the ma
terial data. From ( 11) we determine the number of waves n assigned to the mini
mum. In practice, the adjacent integer value of n would occur. In our example,
the shell vibrates with nine circumferential waves; using the given numerical
values, ( 11) would give the value n = 9.22.
For a shell with free boundaries, inextensional bending may be assumed as a pos
sible vibration mode. For this case, Lord RAYLEIGH determined, by equalling ki
netic and elastic energy, that
nz + 1
We obtain the same result by defining m > 0 in (10 ) . Here, the differences in
dependence on n are a result of the DONNELL simplifications. For larger n ,
these differences are unimportant (X = k n 4) .
E xercise C-14-1:
A spherical cap ( Fig. C-28) is extended over a circular base (radius r0 ,
r 0 a), and is assum ed to be subjected to a co n stan t surface pressure
load p. T h e height of th e cap is given as f.
294
As r 0
= Ky ~
a)
b)
S o lu tio n :
a)
With the given assumptions, the system of differential equations (14-9) reads
= K
<2>)
AA -
a Aw =0
(v2 b )7
X E tK
'
.
K
(3)
v
'
If the underscored terms in ( 3 ) are equal, one can formulate a differential equa
tion for
F = w + X
(4 )
( 5a )
1 - v2 ) = ik'
( 5b)
(6)
A A F - ik 2 A F - p
K
(7 )
In the present case it is sensible to use polar coordinates owing to the axisymmetry of shell and load. The LAPLACE-operator is then independent of the
angular coordinate S and hence
Exercise C-14-1
295
We can thus determine partial solutions from the two differential equations:
AF
= 0
( 8a)
A F - i k2 F = 0
( 8b )
(9a)
_ ik2 F = 0
( 10)
(9b)
,
(11)
(kr)
+ B2 bei ( k r )
B3 ker ( k r ) +
+ B4 kei ( k r ) + Bs + Bg In r ,
2
<D = -------M
f- Bj bei ( k r )
( 12a )
+ B2 ber ( k r ) - B3 kei ( k r ) +
- *2 )
+ B4 ker( k r ) + B? + Bg l n r j
( 12b )
296
14
(' 1 3 a)>
have to
point
be fin ite .
( 13b )
where
(1 4 )
V = Q + N
z
rr
Fa
/l2 (l-v 2)
2 7t
E t2
/ l 2 ( l - v2 )
2 it
kei ( k r )
( 15a )
E t2
The m axim um deflection occurs at the top point where the load acts. For r = 0
we have k e i ( o ) = - 7t / 4. T his y ie ld s:
wmax = - l / 3 ( l - v 2 )
(1 5 b )
In his fundam ental papers, REISSNER has treated problem s of shallow spherical
shells with a num ber of load cases. F or further details refer to [ C.20 ].
E xercise C-14-2:
T h e eigenfrequencies of a hyp ar shell projected against a rectan g u lar base
(Fig. C -29) sh all be determ ined. T h e distance f betw een base and shell is
assum ed to be sm all.
a)
b)
Exercise C-14-2
297
= V
p tf i t^2
=0
Here, the only difference relative to the equilibrium conditions for the shallow cy
lindrical shell (13.36) is that instead of the circumferential force a component of
the shear force occurs perpendicular to the shell.
The kinematic relations are obtained from (14-4)- By transforming the first equa
tion of ( 14-4) by means of ( 14-2) we write
a /J = i ( Val/J +
vp\-
2 z la/Jw )
P ap
= - WLp
( 2 )
The relations between the resultant forces, moments, and the strains are described
by the material law (14-5)
N^ D H ^ a 7,
m^
= k h W p 74
(3 )
We assume that the boundaries of the hypar shell belong to the linear generatri
ces. The mid-surface can then be described in Cartesian coordinates as
Z=S ( X' f ) ( y - | )
(4)
298
we obtain with
z|xx = z|yy = 0
2f
z|xy - a b
T <3 .t + r cU
+ i b N*j = t , dt2
f
>
. -
Q >
(5 )
= -r- v
yy
b 'V
Ex x = T U , f
xy
(6 )
P
pxx = - i w.
^xy
^yy
b 2 ,fw
a b Wf{r}
^xx
^ ( Ex x
v Ey y )
N xy =
D ( ! -
M xx =
K (P x x +
M yx =
K ( 1 - v ) Pxy
v ) x y
* ^yy
^ ^ Ey y
v Ex x )
v Pyy)
> M yy =
K ( P yy +
v Pxx )
Here, the physical components of the originally tensorial quantities are denoted by
the usual indices x , y. The displacements of the mid-surface of the shell are de
noted by u and v .
The kinematic relations ( 6 ) differ from the kinematic relations of the shallow cy
lindrical shell (see (13.37)) by two terms only:
1)
In the case of the cylindrical shell the radial displacement is a part of the
circumferential strain syy , and
2 ) in the case of the hypar shell the deflection w contributes to the shear strain
Exercise C-14-2
299
1 v
U- + ~^ 2 ~ V.(r, +
a v
l 1 + Vn
tab d u
a (
.f
'I
p ta
b \ ur;77 - 4 a W, J = D dx2
I i-vbf,.
..
^ _ ptab
(8)
h2 ( b 2
,)
+ 2 ( 1 - v )-4r- ( a u + b v c - 4 f w ) = p t * b ^
ab
")
D dt
b) The following boundary conditions have to be fulfilled for a shell with all sides
simply supported:
u = w = 0
Nx y
Mx = 0
for
5= 0
and
5= 1
v = w = 0
Nxy = My = 0
for
T) = 0
and
r) = 1
(9 )
( 10)
( m , n integer )
ptab
where the solutions depend on the dimensions and the integers m and n . Numeri
cal evaluation shows that there exist one lower and two substantially higher ei
genvalues for each pair of values of m and n . The numerical values clearly show
that the lowest frequency corresponds to pronounced transverse vibration ( W
U , V ) , and we therefore obtain a good approximation to the smallest eigenvalue
provided that the terms of inertia forces tangential to the mid-surface of the shell
are neglected in ( 8 ) :
p t d ju
dx2
pt
d V
dx2
300
where a = a / b denotes the side aspect ratio. Multiplying (11) by a , and intro>tar
during the scaled frequency o 2 =
<j2 then yields:
D
=
+ v
>2
- 2 ) ( ^ ) 8(
<1 2 >
We now need to determine that combination of m and n for the given dimen
sions which provides the smallest values of tJmn .
If we, at this point, limit our considerations to values a < 1 ( all solutions for
a > 1 can be obtained by suitably exchanging the sides) , we can deduce from
( 12) that n always has to take the value 1. Calculations show that m > 2 is
always valid for the lowest frequencies. We can therefore approximately assume
m 2 oc2n 2 , and we thus obtain from (12 )
+
(1 3 )
mi n
(1 4 )
Thus we have obtained the desired approximate formula for the lowest frequency
in dependence on the dimensions.
D Structural optimization
- C h a p te r 15 to 18 -
D .l D e fin itio n s F o rm u la s - C o n c e p ts
15
F u n d a m e n ta ls o f s tr u c tu r a l o p tim iz a tio n
15.1
S tr u c tu r a l o p tim iz a tio n
C urrent research in optim al stru ctu ral design m ay very roughly be said to
follow two m ain paths. Along th e first, th e research is p rim arily devoted to
studies of f u n d a m e n t a l a s p e c t s o f s t r u c t u r a l o p t i m i z a t i o n . B road conclu
sions m ay be draw n on th e basis of m ath em atical properties of governing
equations for op tim al design. T hese properties a re not only studied analy
tically in order to derive q u alitative results of general validity, but are also
often investigated num erically via exam ple problems. Along th e o th e r m ain
p ath of research, th e em phasis is laid on th e d e v e l o p m e n t o f e f f e c t i v e n u
m e r i c a l s o l u t i o n p r o c e d u r e s for optim ization of com plex practical stru ctu
res [ D .3, D.12, D.21, D .22, D.30 ].
T he constan t flow of general reviews, surveys of subfields, conference pro
ceedings, and new textbooks on o ptim al stru c tu ra l design testify th e strong
activity, recen t advances, an d increasing im p o rtan ce of th e field. A selec
tion of such recent publications is listed as references at th e end of this
book.
302
(1 )
303
or dom ain of th e elem ent, an d such problem s th en fall into category (1),
w hile problem s of optim izing inh eren tly discrete structures like trusses,
grillages, fram es, or com plex practical structures belong to category ( 2 ).
15.3
D e s ig n
O ptim ization problem s can best be classified in term s of th eir design vari
ables. Based upon th e exam ple of a truss-like stru c tu re according to L.A.
SCHM IT/ R.H. M ALLET [D.41], an d N. O L H O F F /J.E . TAYLOR [D.29]
possible design variables can be divided into five (o r six, respectively) dif
ferent classes (Fig. 15.1). In th e following, th ese groups are briefly descri
bed in term s of th e degree of com plexity w ith w hich they en ter into th e
design process:
a) C o n s tr u c tiv e la y o u t
304
f) C ross-sectio n s
T his class of design variables has been used m ost frequently in optim iza
tion tasks (cross-sectional areas, m om ents of in e rtia ).
W hen considering stru ctu ral param eters, we distinguish betw een indepen
dent an d dependent variables, as well as pre-assigned (c o n sta n t) p aram e
ters, w here a stru c tu re is uniquely characterized by statin g th e values of
its independent variables, th e so-called design variables. M ost often, cer
ta in cross-sectional ch aracteristics (thicknesses, d iam eters) a re em ployed
as independent variables, from which all o th er values can be calculated.
As previously m entioned, consideration of th e constant param eters and th e
determ in atio n of th e dependent variables are m ade in algorithm s for a n a
lysis of th e stru c tu ra l design.
T h e i-th design variable w ill be denoted by Xj, an d all n design variables
are com posed in a vector x which lies in th e design space, an n-dim ensional EUCLIDEAN space:
(15.1)
[ x i , x 2 , . . . , Xj, . . . , x j .
a)
Construction
b)
Topology
M aterial
properties
d)
G eom etry
shape
e)
Supports
loadings
i)
Steel
A lum inium
Com posite
A"
Cross-section
305
gj(x)=<0
(i
l.---,q).
(15.2a)
(j = 1,..., p ) .
( 15.2b)
306
a c t iv e c o n s tr a in ts
g(x) = 0
f e a s i b l e in itia l d e s ig n
f(*o )
is o - c u r v e f ( x ) = c o n s t
fo r t h e o b j e c t iv e f u n c t io n
in f e a s ib le s e t
g(x) >
Fig. 15.2: Concepts of stru ctu ral optim ization
T he objective function is in m ost cases a scalar function f of th e design
variables x defined as follow s:
f:=f(x)
15.4
(15.3)
T h e usual problem of o ptim al stru ctu ral design consists in determ ining the
values of th e design variables xi ( i = 1 , . . . , n ) such th a t th e objective func
tion a tta in s an extrem e value w hile sim ultaneously a ll constraints are sa
tisfied. M inim ization of th e objective function, i.e. M in f , is considered in
th e conventional m athem atical form ulation. If an objective function f is to
be m a x im iz e d , one sim ply substitutes f by - f in th e form ulation, since
M ax f < > M in ( - f ).
M athem atical fo r m u la tio n :
(15.4)
w ith
Rn
x
f(x )
g(x)
h(x)
F easible d om ain:
X : = j x e R n| h ( x ) = 0
, g (x ) < 0
(15.5)
307
D e f i n i t io n s in M a t h e m a t ic a l O p t im iz a t io n
2)
x e X .
(15.6a)
x e X n U (x*),
(15.6b)
N ecessary condition
V f(x*)= (^-,
v '
2)
v d Xj
-i!
d x2
*1)
d xn v
=0
'(15.7)
Sufficient condition
d2 f
(15.8)
.d X; d Xj
jSj ( x )
( 159)
j=i
V L ( x * ) = V f ( x * ) + 2 Q * v h i ( x *) + 2 / ?f V S j ( x *) = 0
i=i
j=i
( 1510a)
308
(i = 1
q)
(15.10b)
gj(x*)<0
(j = l , . . . , p )
(15.10c)
pf>
(j =
i,...,p)
(m od)
Sufficient conditions
- V f ( x * ) = 0 * V g l ( x * ) + (3* V g 3 ( x * )
(15.11a)
T he negative g rad ien t of th e objective function does not lie w ithin th e sub
set defined by th e g rad ien ts of th e constraint functions, i.e. (15.10d) is vio
lated by eith er /3* or /5* . x ^ is not a m inim um point as th e function v a
lue f ( x * ) can be reduced in th e feasible set.
b) Point B :
- V f ( x * ) = (3* V g 2 ( x * ) + (3% V g 3 ( x * )
(15.11b)
T he considered point xjjj is a local m inim um point w ith a ll KUHNTU C K ER conditions satisfied. Note th a t for poin t B th ere is no direction
in th e feasible set in w hich th e function value f ( x * ) can be reduced.
309
310
j
I
d e c isio n
m aker
I
1
d ata
~~1
o p tim al
d esig n
transform ed
variable
o p tim izatio n
alg o rith m
optimization model
p.g.h
r)p
r)h
Sxdxdx
op tim izatio n
strateg ics
stru c tu ra l m o d e l
stru ctu ral a n a ly sis
u = u (y )
ev alu atio n
m o d el
m m rnz
se n sitiv ity
ainalvsis
16
A lg o r ith m s o f M a th e m a tic a l P r o g r a m m in g (M P )
In the following, solution algorithm s for optim ization problem s cast in the
sta n d ard M P form (15.4) will be considered. O ne distinguishes betw een
o p tim iza tio n a lg o rith m s o f zeroth, f ir s t and second order depending w he
th e r th e solution algorithm only requires th e function values, or also th e
first an d second derivatives of th e functions. It w ill be assum ed in this
C hapter th a t in general th e functions f, h j , gj in (15.4) are continuous and
a t least tw ice continuously differentiable.
T he m ajority of solution algorithm s is of an iterative character, i.e. sta rtin g
from an in itia l vector x 0 one obtains im proved vectors
, x 2 , . . . , by suc
cessive application of th e algorithm . Iterative solution procedures are neces
sary since p ractical problem s of stru ctu ral optim ization are generally highly
nonlinear.
16.1 P r o b le m s w i t h o u t c o n s tr a in ts
(a )
( i = 1 ,2 , 3 , . . . )
(16.1a)
311
(16.1b)
a straig h t line occurs in th e design space, and we have the following form
of th e objective fu n c tio n :
f [ x ( a ) ] = f ( Xi +
QSj ) = f ( a )
(16.2)
312
or
|P2( S ) - f ( a ) |
< e
(16.3b)
P2( )
( b ) F irst PO W E LL m ethod o f conjugate directions - m ethod of 0th order
In order to calculate an iteration point x i+1, inform ation on th e previous
points X j ,..., Xj is used. F or th is purpose, we need th e notion of conjugate
directions. T h e vectors Sj and ^ are called conjugate if they satisfy the
condition [D.37]
SJ H s k = 0
(j ^ k ) *
( 16A)
(i = 0 ,1.2,...,n) .
(16.1c)
313
(16.1d)
w hich leads to point x n+2. For th e following search cycle, s t is elim inated,
th e index of th e rem ain in g directions 8 2 , . . . , 8 + 1 is decreased by one, and
th e described procedure is th e n applied to all subsequent search directions
(Fig. 16.3).
In m any cases th e convergence behaviour of th is procedure is insufficient
due to generation of alm ost linearly dependent search directions. C ertain
m odifications, however, lead to im provem ents [ D.18, D.38 ].
(c)
(16.5)
(i 1,2,3...)
(16.6)
F LETC H ER-REEVES-M ethod o f conjugate gradients [D .19,D .23] m eth o d o f 1st order
T h e first one-dim ensional m inim ization step is carried out in the direction
of th e steepest descent according to (16.5)
si = ~ v f ( x i)
an d we thus reach point x 2 (Fig. 16.5).
ta n g e n tia l
h y p e r p la n e
V f ( x 1+1)
f = co n st
- V f ( x i+1)
314
x,
(e)
(16.8)
w ith
1^ = I ,
16.2
16.2.1
(16.9b)
P r o b l e m s w it h c o n s t r a i n t s
R eduction to u nconstrained problem s
315
^ ( x . R J ^ f U j + R i^G fg^x)]
j= i
(i = 1,2 ,3,...)
(16.10)
(16.11a)
( max[ . g j ( x )])2
( i = 1 >2 3 - -)
(16.11b)
j= i
-3
R; ( e.g. R t = 10 ,
R 2 = 102, R 3 = K f 1, . . . ) .
Owing to th e inequality constraints gj ( x ) < 0 ( j = 1 , 2 , 3 , . . . ) in ( 16.11a)
we o b ta in :
G [ gj ( x ) ] = 0
G [ g j ( x )] = gj ( x )2 > 0
(16.1Sa)
316
0 2( x , R 2 ) = M in
0 2(x ,R 2 ) = M in
g ito
x2
g2(x ) = 0
317
(i = 1 , 2 , 3 , . . . ) .
(16.12b)
f(xk) + V f ( x k)A x,
(16.1Sa)
hj( xk + A x )
(16.1Sb)
6j( x k + A x )
(16.1Sc)
318
X;
(16.14)
(i = 1,2,..,,n) ,
n o n lin e a r
in itia l p r o b le m
1.
a p p r o x im a tio n
2.
a p p r o x im a tio n
3.
a p p r o x im a tio n
319
i = l,2,...,n.
(16.15)
T h e linearized problem th e n r e a d s :
M in
| cT y J =
cT y
w ith
c = V f xk ) ,
(16.16)
(16.17)
w ith
pi = i [ M * ) ] 2 .
i=l
p2=
P3
Z[gj(*)]2
i eC
- ca = {j|/?.>o},
Z [ g j ( x ) ] 2 . Cb = { j | ^ j > 0
jecb
and gj > 0 } -
320
[D.38]
A
by
in
by
[D .l]
procedure
c o n s is tin g
of
SQ P-
and
G R G -m e th o d s
(Q P R L T )
[D.1,D.35]
In a first step, a search direction is d eterm ined by m eans of th e SQP-algorithm ; then, th e optim al step length is calculated by th e G RG -algorithm .
T hus, th e advantages of SQP a n d GRG are com bined in one single algo
rithm .
321
17
S e n s it iv it y a n a ly s is o f s tr u c tu r e s
17.1
P u r p o s e o f s e n s itiv ity a n a ly s is
x e Rn
(17.1)
322
[aij]mxll
w here A =
(YlJSa)
ip
d x
ip{
dxj
m x n
d ip
d Vi
dll
dll
auk
d X
m x nu
dUk
a Xj
(17S b )
( i
* ------------------------- a j ; -------------------
l , . . . , m )
o = i
>
<m >
w ith
Xj =
Axj
323
(17.5)
F = F (x )
u = u (x )
r = r (x )
load vector .
d ll
d Xj
du
' d Xj
dXj
'
(m )
Pj
w here Pj is th e so-called pseudo load vector associated w ith th e design va
riable Xj.
W ith r know n an d u obtained by solution of (17.5) , com putation of the
pseudo load vectors Pj ( j = l , . . . , n ) in (17.6) only requires th a t th e de
sign sensitivities dr/ dXj an d dF /d X j of th e load and the system m atrix
are known. Note here th a t th e form er sensitivities vanish if th e load is de
sign independent.
If in (17.6) th e design sensitivities of th e global system m atrix d F / dxj are
d eterm in ed an aly tically before th e ir num erical evaluation, th e approach is
called th e m ethod of a n a ly tic a l sensitivity analysis, an d if th ey are deter
m ined by num erical differentiation, cf. (17-4), th e label se m i-a n a ly tic a l
sensitivity analysis is used. W hile th e an aly tical m ethod is expedient for
problem s w ith cross-sectional design variables ( see Section 15.3), it is usu
ally a form idable task to im plem ent th e m ethod w hen shape design vari
ables (se e Sections 15.3 a n d 18.2) are encountered. Thus, a large am ount of
an a ly tica l work an d program m ing m ay be required in order to develop
an aly tical expressions for derivatives of, for instance, various finite elem ent
stiffness m atrices w ith respect to a large num ber of possible shape p aram e
ters. For problem s involving shape design variables, it is m uch m ore a ttra c
tive to apply th e sem i-analytical m ethod because it is easier to im plem ent
as it trea ts different types of fin ite elem ents an d design variables in a uni
fied way.
324
<5u
dx.
( hr
<5F
= ( ^^ xj
r - ^ r^ xj
" ) = P i
<m >
l ^ F
- 1( i = l , . . . , m ) .
(17.8a)
fTji = ( ^ ) T
for A: th a t is to be solved for th e m
).
\
..., m
<m h )
right-hand sides
<5u
( i = 1,
18
325
O p tim iz a tio n s tr a te g ie s
In order to tre a t different types of optim ization tasks like shape an d topo
logy optim ization problem s as well as m u lticriteria or m ultilevel optim iza
tion tasks, specific optim ization strategies have to be integrated into th e
optim ization loop according to Fig. 15.4. T hese strategies are sub-parts of
optim ization m odeling, an d they tra n sfe r a rb itra ry optim ization problem s
into so-called su b s titu te pro b lem s by way of tran sform ation or decom posi
tion so th a t th e given task s can be solved by u sual scalarized p aram e ter
optim ization procedures. In th e following, two of these strategies w ill be
briefly treated , nam ely
- vector, m u lticriteria or m ultiobjective optim ization, an d
shape optim ization,
w here a tran sfo rm atio n into p ara m e te r optim ization problem s for both
strategies is carried out.
18.1
V e c to r, m u ltio b je c tiv e o r m u l t i c r i t e r i a
P A R E T O - o p tim a lity [ D.14, D .34, D .43, D.44 ]
o p tim iz a tio n
g(x)<o]
1
(18.1a)
326
18 Optimization strategies
f (x )
\ U X)
for all
je{l,...,m }
fj ( x ) < fj ( x* )
an d
j {1,..., m } .
(m
Design space
327
p[f(x)l
(18.S)
i e R n
p[f(x)]
n ii
x e R
(18.5a)
P Zj(x)
a)
I
lmin
f,
b)
a)
b)
Trade-off-form ulation
M in-m ax-form ulation
Z,
x2
328
- Sum
18 Optimization strategies
o f d is ta n c e fu n c tio n s
m
p [ f ( x ) ] : = ( 2 | f j ( x ) - yj r ) Vr
xeR"
(18)
j=i
w ith th e vector y designating given goal values or dem and levels for crite
ria fj ( j = l , . . . , m ) . Here, th e values of th e com ponents of y an d th e ex
ponent r ( l < r < o o ) are a t th e choice of th e designer.
- C o n s tr a in t-o r ie n te d tr a n s fo r m a tio n (T r a d e -o ff m e th o d )
P [ f ( x )] = fi ( x )
j =
x e Rn
(18.7)
p [ f ( x ) ] := Max[z.(x)]
w ith
z. ( x )
f. ( x ) - f.
J- ,
= -----=
fj
x Rn
_
f. > 0
j =l,...,m
(18.8)
p [f(x )] := Max[wj-f.(x)]
j
3
x e Rn
(18.9)
(18.10)
'
(3
subject to
w. f ( x ) < /3 .
(18.11)
33
329
In d ir e c t m eth o d s [ D .4, D .5, D .7, D .9, D.21, D .22, D .27, D .28, A .3]
2)
J f ( ( a , Rj , Rj Q ) dA
(j
1,2,3) ,
(18.12a)
w here
J f k( f
,Rj ,Rj, J d A = 0
(k = l , . . . , m ) ,
(18.12b)
J f z ( f , R j , Rj rQ) d A < 0
(Z = 1
r)
(18.12c)
330
18 Optimization strategies
J f z ( , Rj , RjfJ d A + ? = 0
(1 =
(18.13)
m
* = f ( f ,Rj ,Rj, J + 2 \ f k ( f
k=l
(18.14a)
+ 2 V z U .Rj .Rj, J .
Z= 1
we obtain th e L A G R A N G E -functional as
I = f ^ ( f , R j,R\a , \ , ^ ) d A + 2
A
>7?
(UMb)
Z=1
(1815)
r- dR1
fn
1 = 0 ,
V d R \ J ia
<5Rj d r = 0 ,
(18.16)
dR \a
(I = 1
r)
(18.17)
(I = 1 , . . . , r ) .
(18.18)
D ire c t m eth o d s
331
O ne determ ines an o ptim al shape function R* for which the objective func
tional F a tta in s a m inim um
MinF(R)
Rer2
----- F ( R * )
(18.19)
Hi R = ^ i
(i = i , - - , q) >
G j R < Xj
(j = 1 , , ? )
T he unknow n functions R are approxim ated by suitable functions R , socalled shape approxim ation functions.
In recent years, th e progress of CA D -techniques in th e design an d con
struction d ep artm en ts has substantially increased th e im portance of ge
o m e tric a l m od elin g also in application to stru c tu ra l optim ization. Basically,geom etrical m odeling deals w ith com puter-based design an d m an ip u la
tion of geom etrical shapes [ D .8, D.30 ].
T he choice of suitable approxim ation functions for optim al geom etries is
problem -dependent. T he chosen function is to approxim ate th e course to be
followed as precisely as possible, a dem and th a t leads to a large am ount of
shape p aram eters and thus to increased com putational effort. A reduction
of this effort can be achieved by decreasing th e num ber of param eters,
which, however, requires some a-priori knowledge and experience concer
ning th e choice of a given type of approxim ation. If this inform ation does
not exist, optim ization should proceed w ith sim ple approxim ations to be re
fined w ith increasing level of knowledge.
In th e following, we w ill introduce som e of th e m ost im p o rta n t approxim a
tion functions for geom etric m odeling of shapes of co m p o n en ts:
l)
(18.21)
332
18 Optimization strategies
( 18-22)
Tj = 2
1 > Ta = 2T1 - T 1 - T 0 ,
M
= ^Tj Tk l - Tk 2
Z P i Bik ( 0
( 18-ZSa)
i= 0
w ith
Pj
.
1
2
.
vector of th e i-th control point in th e given x ,x coordinate
system ,
B jk( ( )
m ixed function .
333
Pi
n=8
BikU )
li+k-l
li
( 18-zsh)
Bik_1( 0 + t ti+k; /
Bi+lk_1( 0
i+k li +1
(18.23c)
w ith
the
description
of the
2 P i B i k( 0
i=0
Pj
Bik ( )
(im h )
10.0
B: Kj = 6,0
k2 =
2.0
C: K, = 2.0
k2 =
5.0
D: Kj = 2.0
k2 =
2.0
E: Kj = 1.5
k2 =
1.5
k2 =
2.0
G: Kj = 1.0
II
<N
*
ii
k2 =
A: K, = 10.0
o
'ui
18 Optimization strategies
n
r,
334
(fP + (fPa /
(18.25a)
vb
x 2 = b ( cos ip) ^ 2
ip.
( 18.25b)
18.3
A u g m e n te d o p tim iz a tio n lo o p b y a d d i t i o n a l s t r a te g ie s
[ D.3 , D.12, D.40]
335
M in f [R(S ,x)]
hj(x) = 0 , i = l,...,q
gj(x) < 0 , j = l,...,p
Xj < X < xu
336
18 Optimization strategies
data input
determination
OS
decision maker
y0
initial design
optimal design
stmctural
parameters
yn -> *o
y = const
OS/SOP
Idesign model |
discretization
rule \
X
optimization
algorithm
P,f,g
dp df dg
dx dx dx
optimization model
'shape algorithm"
shape function
Eft)
OS/VOP
preference
model
1H'
structural model
stmctural
analysis
f x,u(x)f ,g[x,u(x)j
evaluatio n m odel
state variable
u(x)
sensitivity
analysis
8u
dx
OS : optimization strategy
SOP: shape optimization
VOP: vector optimization
D. 2 E x e r c ise s
E x e r c is e D -1 5 /1 6 -1 :
An uncon strain ed optim ization problem is given by th e objective function
f ( Xj , X j ) = 12
+ 4
- 12 Xj x 2 + 2 Xj
----- M in
Xj , x 2 e R.".
S o lu tio n :
a ) We are confronted with an unconstrained optimization problem with a conti
nuously differentiable objective function possessing an exact solution.
According to (15.7) the candidate minimum point is obtained from the necessary
conditions
df
dXj
24 Xj - 12 x2 + 2 = 0 ,
^ - = 8 x2 - 12 x1
2 X1
>
*
X1 =
H (x ) =
24
-1 2
-1 2
= 48
This proves positive definiteness, i.e. a minimum solution has been found.
b)
= ( - 1 >- 2 )
f0 = 2
338
F irst cycle
X
( -1 + a
+ a
( 1)
-2
da
= 2 4 ( - l + a ) + 24 + 2 = 0
12
which yields
_
il
12
and
f ( x j ) = 1.9167
-2
ic2 = Xj + a si = I
12
12 I + aj
(2)
-2 + a
Substitution of ( 2 ) into the given function:
f(a) = 1 2 ( - 1 |) 2 + 4 ( - 2 + a)2 - 1 2 ( - g - ) ( - 2 + a) + 2 ( - g )
da
= 8 ( - 2 + a ) + 13=0
'
'
and
_ 13
8
f ( x 2 ) = 1.3542
.1 3
' 8 J
j_ \
12
3^
V 8 /
Then it follows
(
1 \
(3 )
339
1 4
) - ( 1
- *
'
f ( x 3 ) = 1.319728
and
The second cycle also proceeds from the search direction sQ = ( 1 , 0 ) . We get
( - 147
I +
^
_ 83
49
x4 = x 3 + a 80
2 (-
157
147
(5)
bf
- = 2 4 f - 4 ^ + ) + 12
+ 2 = 0
V 147
/
49
ba
a = 0.1377552
f ( x 4 ) = 1.092008
-1.6938775
We then formulate
f -0.0833333 \
( -0.9302720 'I
x 5 = x 4 + ocs2
(6)
+
V -1.6938775
I,
0.375
dec
a = 0.438567
340
From ( 6 ) follows
( - 0.9668191
and
f ( x ) = 0.89566164
-1.5294147
The course of the optimization process clearly shows a very slow convergence to
wards the solution point. We therefore stop the treatment at this point and pro
ceed to c) .
c)
/ 2 4Xj - 12XJ +
V f ( x 0) = V f ( x )
l0
8x2 ~ 12xi
= - V f ( x o) = (
I )
and thus
f ( 0 ) =
12( - 1 - 2 0 )2 + 4 ( - 2 + 4 o )2 - 12(-1 - 2 a 0) ( - 2 + 4 a 0) + 2 ( - l - 2 a 0) ,
jr
3H
nr0 = 0
a nU= 0.048077
( -1.0961 \
and
V f ( Xj ) =
V - 1-8077 )
V - 1-3084 /
|v fM
_ / 2.6140 \
Bl
V 1.3084 J +
( 2.6140 \
( -2.6140 \
(-2.6140 f+ (-1.3084f ( ~ 2
( - 2 ) 2 + 42
/ - 2 \
+ 0.4272
V 1.3084)
V 4 )
^ 4
( 1.7596 \
V 3-0172 J
341
V -1-8077 )
( 1.7596 ^
+ ai
V 3.0172 )
yields
otj = 0.4334
One obtains
( - 0.3334 \
= ( -05 j
( 0 \
"d
V'(Xl) = ( o J '
/ 24 -1 2 \ ( 1.7596 \
- 2 , 4)1
0
V - 12
8 / \ 3.0172 )
is fulfilled.
Fig. D-l illustrates the single search steps for the FLETCHER-REEVES-method.
It is obvious that this method converges much faster than the POWELL-method.
By suitable modifications, however, convergence of the latter method can be im
proved.
342
E x e rc is e D - 1 5 /1 6 - 2 :
T h e tru ss stru ctu re shown in Fig. D-2 consists of 13 steel bars w ith th e
cross-sectional areas A; ( i = 1 , . . . , 13 ) an d 10 nodal points. A v ertical force
F = 100 kN acts a t node 3.
D eterm ine th e cross-sectional areas in such a way th a t th e w eight of th e
stru c tu re is m inim ized. T h e stresses in th e single bars m ust n o t exceed an
adm issible tensile stress of cr.
= + 150 M P a , and an adm issible com*adm
pressive stress of a
c adm
= -1 0 0 M P a .
x,u
343
S olution:
a ) The relation between the nodal forces and nodal displacements is established
by means of the displacement method. This will be demonstrated for the forces F3
and F4 acting at node 2 and the corresponding displacements u2 and v2 E q u ilib riu m co n d itio n s :
Equilibrium conditions give the external forces
in terms of the b ar forces at node 2:
F3 = S6 + i / 2 ' S7 - i / 2
S5 ,
(1)
F4 = 1 / 2 S7 + 1 / 2 S5
E lasticity la w :
AI Z S
6 EAg 6
AI
- EA7
J7
J 21
> A L = EAg
(2)
Kinem atics:
AZ6 = 0 ,
AZ7 = T
-J* V2
( 3 )
F4 J
22
2J
-Ag + Ay
Ag + Ay
Ag + Ay
E
2 / 2 7 K 22
( 4 )
V = (u1,V1,U2,V2,U3,V3 ,U4,V4)U5,Vg)T
344
2V21
*11 =
CO
with
12
13
*14 ;
K 23
^ 4 i K 25
K 32
K 33
k 34 j
K 42
K 43
K 44 : K 45
5 1;
K52
K 53
S 4
A2 1 A4 1 A 8
/ 2 A 3 + A4 + As
A 4 - A5
A4 - AS
A4 + A5
A jj + 2 /2 ^ A 12 + A 13
A11
A13
A jj
A13
A11 + A13
A10 + A11
- Aj0 + Ajj
A10 + A11 J
- A.
K 13 = K 31
4J
'10
K,51
"1 0
~A5
*^ 3 2
As
K 54 = K 45
10
A 5
- A 11
5J
Ml
K 55
A io
+ A4 + Ag -f A10
K 33 =
35
4 i
K 22 = see ( 4 ) ,
*^ 23
' A2 + A4 + Ag - Ajo
15
K 22
2V2 Aj + A^ + A4 +
+ Ag + A10
K,55
(5b)
345
(6)
Thus, the displacements of the end-point of each single bar is established, and we
can now, on the basis of the element stiffness matrices, determine the stresses
within the bars by means of the matrix relation between stresses and displace
ments :
a = Rv
( 7 )
Here, R is a (13 x 10)matrix of the form
_i _
" 2
0
" 2
0
0
J_
2
0
J_
2
0
J_
_1_
2
1
' 2
0
2
0
2
0
_1 _
_1 _
_1 _
~2
~2
2
0
13 x 10
(S )
Substitution of (6 ) into ( 7 ) then yields the relation required for calculating the
stresses:
a = R K 1f
(9 )
The equations for the structural model that is required for the optimization have
now beeen established.
b ) In the following, the equations of the optimization problem shall be set up. In
accordance with the problem formulation, the cross-sectional areas of the bars
shall serve as design variables, i.e. we define
x := A
According to ( 5b ), K depends on the design variables, i.e. K = K ( x ) . Given the
same bar material, weight minimization is equal to volume minimization; the ob
jective fu n ctio n of the structural volume is thus a linear function with respect to
the design variables:
f ( x ) : = V ( x ) = I T . x = f i iXi
i 1
( 10)
346
i ( x ) tadm
( i = 1 , . . . , 13 ) ,
- 8 c j ( x ) : = cadm ~ j ( x ) ^ 0
(j = 1 , , 1 3 )
(11a)
.
(lib)
for all
i = 1 , . . . , 13
( 12 )
(x)])2
(i = 1 , 2 , 3 . . . )
( 13)
Here, the choice of a suitable initial value for the penalty parameter Rj is crucial;
for the present task we choose:
Rx = l ( f 5
347
E x ercise D -1 5 /1 6 -3 :
Fig. D-4 shows a section of a circu lar cylindrical shell C w ith a ring stiffe
n er S. T h e considered p a rt of a boiler is subjected to a constant in tern al
pressure p an d h as a constant in n er tem p eratu re 0 ic.
T he tem p eratu re distribution w ith in boiler an d stiffener has been determ i
ned by m easurem ents; for th e cylindrical section C we assum e a lin ear tem
p erature distrib u tio n over th e thickness w ith th e grad ien t 10 c = const in
the longitudinal direction
c ( z ) e c + z r
iC +
w ith
0C
c =
oc
iC
w ith
( r - 1) ( r + 1 - 2 a ; ) /
(i - w)2
0 oS
e.c )\
= a/ b
.X .
-2h
U ---------
348
N u m erica l v a lu e s:
G e o m e try :
b = 0.5 m
a = 0.65 m
L o a d s:
&\c = 170 c
- 0 os = 50 c
l 0 c = 1 c / m m
. P = 2 M Pa
M a te ria l:
a TC = a TS = cxT = 1.11 10~ S/C , Ec = Es = E = 2.1 10s M P a
&c = Ps ~ 6 0.785 104 k g / rr? , v 0.3 , cQ s
= 200 M P a .
S olu tion :
a ) - Structural model and structural analysis
The stress state of the given stiffened boiler structure can be most conveniently
calculated by applying the compatibility between the single parts. Since the re
spective steps for establishing the structural equations have already been described
in detail (see C.13.1/2), only the most im portant aspects will be treated here.
In a first step, we separate the two semi-infinite cylindrical shells from the ring
stiffener. Owing to the different deformations of boiler and stiffener at the interfa
ce point, the required compatibility is induced by yet unknown boundary forces R
and boundary moments M. Each of the substructures shows deformations caused
by temperature and pressure loads (state 0 ), and by the forces R acting at the
boundaries (state 1 ), and the moments M (state 2 ). In the present case, the
parts of the boiler can be idealized as circular cylindrical shells subject to axisymmetric loads (pressure, temperature, boundary force R, and boundary moment
M), and the ring stiffener can be treated as a circular disk subject to internal
pressure, temperature and the radial force R. The boundary moment M of the cir
cular cylindrical shell does not effect the stiffener.
The deformations are calculated from the basic equations for the circular cylindri
cal vessel and for the circular disk ( see C-13-2). After determination of the defor
mations at the points of the substructures, we formulate the compatibility condi
tions
wc = wc 1 + ' o 11 + " c =
Xc =
'
0> +
io ,
11 = s -
( la)
(lb)
where wc and Ug denote the expansions of the vessel and the radial displace
ments of the stiffener, respectively; xc are the corresponding angles of rotation.
Conditions ( la,b ) constitute a linear system of equations for determining the un
known boundary quantities R and M. After some calculation one obtains:
M = " i ^ R - Kc ( 1 + ' O T C l 0 C
349
( 2b)
with
iC J
12 L
(1 - w)
fe)l
%
e l<
Kc =
E c t'
C 1 2 ( 1 - v 2)
k4
= 3 ( 1 - v2 ) ( | - ) 2
'
Refer to C-13-2 for further details of determining the curves of stress resultants
and deformations.
- S tresses w ith in the p a rts o f the boiler
Cylindrical shell C
- Longitudinal stresses
x x (x ) = ^ { [
Rsin* x + ( m + (1 +
(3a)
( sin
Kx +
cos k x )] e KX - (1 +
- Circumferential
al stresses
V > ( x ) = 2 2 K c { R C S * X + [ M + C1 + v ) T C K C l 0 C
(3b)
( cos k
orCn = Jy o2
2ipxx +1 a ip
Ring stiffener S ( disk )
- Radial stresses
axx a
tptp
>
(3c)
V
350
- Circumferential stresses
2
r
+ ^ ) + ES oTS0 i c [ ^ ( 1 ) +
a<p<p(r ) =
(4 b )
- Reference stresses
2
rr 1
<p<p
/0 + 0
(4c)
- 0 rr 0<p<p
The reference stresses provide the basis for defining the constraints for the
optimization.
b ) Definition of the O ptim ization model
In order to illustrate the design domain, only two design variables are considered
in the following: the half thickness of the stiffener ring Xj = : h and the shell
thickness x2 : t, both of which are combined in the design variable vector-.
T
x = (Xi,X2 )
( 5 )
,
for x2 b
We now consider the c o n stra in ts that at each point x or r of the two boiler
parts, the reference stresses Cr have to be smaller than the maximum admissible
stress values:
Cylinder C
Cadm
- 1< 0
(7a)
Stiffener S
(7b)
351
0 < x2 < 40 m m
(8 )
Now, the following structural optimization problem (15.4) with the scalar objective
function ( 6 ) and the inequality constraints ( 7a,b ) shall be solved:
Min { f ( x ) | g ( x ) < o }
xeR "
In order to solve this constrained problem, an algorithm can be chosen from MPalgorithms of zeroth, first, and second order. In the case of the actual non-convex
problem (Fig. D-5), the algorithm should perform as simply and robust as possi
ble; here, one of the penalty function methods (e.g. internal penalty function) or
the COMPLEX algorithm by BOX are very suitable zero-order methods (see
[D.24]).
Since only two design variables are considered, the determination of the optimal
design of the current problem can be carried out analytically. As shown in Fig.
D-5, the feasible domain X of the design domain is determined by the active con
straints of the reference stresses in the vessel and the stiffener ring ( 7a,b ), and
by bounding the wall-thicknesses of vessel and stiffener ( 8 ). In addition, the iso
lines of the objective function f ( x ) ( total weight W of the considered parts of
the boiler) are depicted in the diagram.
Fig. D-5 displays the optimal values for the design variables as
Xcpt = ( 4-2 - 11-8 )T .
W
11000 [N]
10000
9000
8000
7000
6000
5000
4000
3000
2000
20
h [mm]
352
18 Optimization strategies
.20 )T [ mm ] ,
E x e r c i s e D -1 8 -1 :
Perform a mapping into the criteria space for a vector of the two objective
functions (criteria)
' w
.
a)
b)
'
f2 ( x ) .
x2 - 4 x + 5
1 2 _
, 29
_T x - 5 x +
S o lu tio n :
a ) Presentation of the objective functions in the design space:
Fig. D-6 : Objective functions and domain of the functional-efficient set of solutions
Exercise D-18-1
353
Fig. D-6 shows that the curves have slopes of opposite signs in the dotted area;
according to Def. 1 in Section 18.1 there exist functional-efficient (or PARETOoptim al) solutions of the two functions.
b ) Functional-efficient set of solutions in the criteria space
The Vector Optimization Problem (18J ) can be transformed into a scalar, con
strained optimization problem by minimizing only one of the objective functions,
for instance f j ( x ) , and by imposing upper bounds on the remaining ones (18.7),
e.g.
fj ( x ) Min
V x 6 X ,
subject to
( )
fj ( x )
yj
L ( x , p ) : = fx( x ) + 2 P j [ fj ( x ) ~ yj]
(2)
j= 2
(i = l , . . . , n )
dx
dL
dpj
fj ( * )
(3a)
(3b)
yj
f(x) =
x - 4x + 5
fl ( x )
. f2 ( x )
1-2
c i 29
~2 X - 5 x + T -
f2 ( x ) = Vj
(j = 2 ,---,6)
Min
354
18 Optimization strategies
4^- = 2 x - 4 + / ? ; ( x - 5 ) = 0
dx
J
d/?j
= 4 -x 2 - 5 x +
- y = 0
yl
J 1-2
* x ?
Ji,2
2xj 1,0 - 4
UJl____
( 4)
5 - x *1(2
= 5
( 5)
y2
'1,2
y3 = 6.5
x*
3 1.2
r5 2 73
1
/ 42.78/1.22 \
f - 3.773
, 3*
={
3 1*3 \- 0 .2.27
( 37.0/1.0 \
{ -4 .0
= 5 3
8.0
6,
-5.0
y4 = 4.o
11,2
y5
2,0
y6
1.0
3* = oo
k?
6 1,2
= 5 i 72
, no real solution
'
This proves that only the constraint level of 2 < y. < 6.5 leads to unique functio
nal-efficient solutions. Fig. D-7 presents the 3*-values belonging to the different
constraint levels yj in the criteria space. The efficient boundary dY* (solid line)
of Y is valid for non-negative values of 3*-
c r ite r ia s p a c e
y 2 = 8 .0
RZ =-3.73
y3 =6.5
(33i =-4.00
y4 = 4-0
y5=2.0
y6 = 1 0
12 16 20 24 28 32 36 40 44 48
fj
Exercise D-18-2
355
E x ercise D -1 8 -2 :
A simply supported column as shown in Fig.
D-8 has variable, circular cross-sections ( radius
r ( x ) ) and is subjected to buckling. The length I
and the total volume V 0 are given.
Ill
V0= const
A (x)
V = j u [ r ( x ) ]2 dx = V0 = const
o
F_
w rX X +
Differential equation
for column buckling
With
I ( x ) K[ r (x )]
4
follows
4 F C
-|------ ^
'xx 1 h E
w(0) = 0
(1)
(2 )
( 3)
w ( I) = 0
(4)
r ( x )4
( 5)
w.xx + V-
r ( x)
with
2
4F
P = tcE
= o
[ r ( x ) ]4 = - p
2 W
(6 )
356
18 Optimization strategies
Tip
Vo
- 1
< >
x 0
----- >
Min.
(8a)
,xx
b)
,xx
( 8 b )
In accordance with the rules of the calculus of variation one obtains EULER's
differential equation as the necessary condition:
I
d v i
With
dF \
=0
w /X X / # x x
JJ
dF
w
and
2 (V
1 (
2
I
dF
wf X X
(9 )
1/2
/X X
w
W
1/2
/XX
w :
Multiplication by w leads to
- F * z
,xx
>xx
Eq. ( 10b ) constitutes the optimality criterion for the present problem.
<10* )
Exercise D-18-2
357
v ,xxw
w , xx
0
--------*
( v ,xw -
v w ,x),x =
( U )
(w ).x
( 12)
(13a )
W , x3 x
Since w ( x ) can only be determined up to a multiplying factor ( w ( x ) eigenmode ), one can choose c = 1, and thus
w = -
w , X3X
( 13b)
w , xx = ~ w
or
( w ,x
),x =
3 ( w2 /3 ),x
2 w , x w ,xx =
>
-I/3
" 2 w ,xw
( 14)
w, x = J J J a2 - w2/3
( 15)
j
-fz /
dw
a^T T T
(16a )
Now, introducing
w = u
dw = 3 u du
,----------- + C
7/ aa 22 - u- 2
arcsin
^ 7 a 2 - u^ | + C
( 16b )
358
18 Optimization strategies
w = 0
>
x = 1:
w= 0
>
C = 0
2
a2 =
n
w,xx
According to ( 6 ) we have
r 4 = - p2 W
w =
W
,xx
==>
=[L
or
= - w 1^ 3
(i ,18, !)
w '
( 19)
Substitution of ( 19 ) into (17 ) then yields the implicit form of the equation for the
optimal cross-sectional radius function
x = / arcs in ------ / 1 ~
7C |
rQ r0 *
\r 0J
1
j
with
r4 =
0
3U3
^
E
. (20)
v
r = r0 f ( x )
( 21)
( 22)
'
S i* 5
(23)
( 24)
Exercise D-18-2
359
16 r
(25)
( * / f 2( x ) d x )
It is demanded that the volume V0 be identical for both the column with constant
and with variable cross-section. Thus, VQ = * r^ I be valid for the column with
constant cross-section.
T
n
4
The area moment of inertia for rk = const
1y = T rk
Thus, we can write
Vo = 4 * 1 /
( 2 6)
.2
Ely It'
= w F_c r i t ^
( t Jf2(x)dx)
crit c o n s t
c n tc o n s t
1.0 Fc n tc o n s t
L 3 6 F c ritc o n St
M a x im u m
b u c k lin g lo a d
>i<
' ~r ~'
A
Fig. D-9: Comparison of buckling loads for simply supported columns with the
same volume and circular cross-section
360
18 Optimization strategies
E x e r c is e D -1 8 -3 :
T h e essential com ponents of a conveyor b e lt d ru m a re th e belt, th e suppor
ting rollers, as well as th e drive an d guide d ru m (Fig. D -10a,b). T h e single
drum s consist of a d ru m casing ( 1 ) and a d ru m bottom ( 2 ). For th e pre
sent ty p e of construction, th e bottom is connected w ith th e sh a ft ( 4 ) v ia a
tension pulley ( 3 ).
) = p0 + px cos -d + p2 cos 2
7T/2
F d = ra I
j
o
p ( ) cos
di?
(la)
Exercise D-18-3
361
and from th e condition th a t th e load for th e rem aining a rea a tta in s a m ini
m um via a root m ean square form ulation.
Thus, th e d ru m force F d = 650 kN leads to th e load:
p ( t f ) = (0.2117 + 0.3 3 2 6 co stf + 0.1411 c o s 2 tf) [ M P a ]
(lb)
* ( > . * 3 ** 4 ) =
t ( ^ , t 4)
( t x - t2) ^
(*3
<P3
<PX <<P<<P2 ,
~ *4 )
P2
5 - uf'<P2 <<P<<P 3 ,
= t4
<
p 3 <<p < y
(2)
(3a)
(3b)
362
18 Optimization strategies
a)
b)
S o lu tio n :
a) The objective functionals read as follows:
F i ( f , ) = / Pgd V W ,
(4a)
F 2 (<f>) = ma x [ o r (<f>,)]
( 4b )
where the reference stress is calculated by means of the VON MISES hypothesis :
(5 )
i? i? - <p v 0 *, , 4 x.tp d
max
[r ( x , 9
<
p^ <. tp < 7t: / 2
0 < i? < *
(6a)
,
,0
(6b)
)]
g2 ( x ) =
g3 ( x ) =
max
[or ( x ) , 9 , ]
Cradm < 0
[or( x ) , 9 ,] -
oradm < 0
9, < V < 92
max
[0r( x ) , 9 =<P3,a] -
(7)
^radm 0
with or adm = 30 M Pa
In the structural analysis, the drum bottom is treated as an uncoupled disk-plate
problem, and the drum casing is considered as a circular cylindrical shell. For this
purpose, a special transfer matrix procedure has been used according to Section
13.2. The results were additionally verified by control computations by means of
an FE-software system [ A.21 ].
Exercise D-18-3
363
b) The present shape optimization problem has been solved by means of the opti
mization algorithms SLP and LPNLP (see 16.2.2 a,b ). Fig. D-12a,b illustrates the
efficiency of the above algorithms when using the constraint-oriented transforma
tion as optimization strategy for the treatment of multicriteria optimization pro
blems. Sequential linearization shows fast convergence; if active constraint limits
are imposed, about six to ten linearization steps are necessary, where the gradient
evaluations require the highest computational effort. Fig. D-12b shows that the
rate of convergence of the LPNLP-algorithm is lower than that of the SLP-algorithm.
Fig. D-12:
E x e r c is e D -1 8 -4 :
C om ponent optim ization plays an im p o rta n t role especially in space tech
nology. As a typical exam ple, a satellite th a t is to be brought into its orbi
ta l position should have an extrem e lightw eight design for saving tran sp o r
tatio n costs; even sm all weight savings for single com ponents resu lt in a
su b stantial cost reduction. O ne of th ese com ponents is th e fu e l ta n k of th e
satellite w hich stores th e fuel for th e position control rockets over th e en t
ire life-tim e of th e satellite.
In th e present exam ple, th e calculation an d optim ization of a thin-w alled,
satellite ta n k subjected to constant in tern al pressure shall be dealt w ith ( a
q u a rte r section of th e com ponents can b e considered for reasons of sym m e
try (Fig. D-14) ).
Exercise D-18-4
x2'
'
365
500
400
300
200
100
100
200
300
400
500
x 'f m m ]
g = 4.5 103 k g / m
YOUNG's m odulus
E = 1.1 10 s M P a ,
POISSON s ratio n
v = 0.3 ,
B reaking stre n g th
c B = 1080 M P a
366
-
18 Optimization strategies
T he stren g th verification is perform ed depending on th e sign of th e
p rincip al stresses in m eridional an d circum ferential direction an d in ac
cordance w ith th e following stress hypotheses of th e state of plane
s tre s s :
1)
2)
r s o
S o lu tio n :
S tru ctu ra l A n a lysis
In the following, some general remarks shall be made concerning the stru c tu ra l
analysis. The minimum volume of the tank already occupies more than 80% of
the given construction space. This fact demands tank contours that smoothly fol
low their boundaries both at the poles and at the equator. At the equator, the
shape of the tank approaches a cylindrical shell curved in the circumferential di
rection; at the pole, the radii of curvature increase to such a degree during optimi
zation that a very shallow shell emerges. It appears that linear calculations pro
duce large displacements in proximity of the pole, exceeding the wall-thickness by
fax. As the displacements do not occur constantly over the arc length, the radii of
curvature of the deformed structure change substantially. According to the above
theory, the pole area shows a decisively larger curvature in the state of deforma
tion, which results in a violation of the conditions of equilibrium that were origi
nally formulated for the undeformed element. Thus, we used an augmented ap
proach for the structural analysis according to Section 13.2.
Shape O ptim iza tio n
The following shape optimization requires a mathematical description of the tank
shape as a function of free parameters by means of shape functions. The descrip
tion should be characterized in such a way that a large number of admissible
shapes can be achieved with a relatively small amount of parameters. The shape
functions have to comply with the following requirements:
Exercise D-18-4
367
The tank shape should not exceed the specified fitting space.
The meridional contour of the curve must be determined in such a way that
the given minimum volume is attained.
or
Xj = Xj
x2 = x2
x3 = t
x,. = t-
The thicknesses t of the shell elements are used as additional design variables in
the transfer matrix procedure. Since the computational effort increases substanti
ally with the number of design variables (> 200), only the geometry variables are
368
18 Optimization strategies
x2 = x2
V
gj = 1 < 0
vo
volume constraint
,
,
f , = m a s s m [k g ]
Exercise D-18-4
Fig. D-16:
369
370
18 Optimization strategies
E x e r c i s e D -1 8 -5 :
Fig. D-17 shows a spatial sketch of a parabolic radiotelescope reflector w ith
circular aperture. T h e reflector is assem bled from single panels w ith sand
wich stru ctu re, each of which consists of an alum inium honeycom b core and
top layers m ade of Carbon F ibre R einforced P lastics C F R P (see Fig. D -18).
Exercise D-18-5
371
with
'
( X)
The design variable vector x here defines the fibre angle oc , the ply thicknesses
t j , the core height hc , and the sides ratio b / a :
x
= ( a i > - >a n i
i - > ! hc ; )
(2)
The panel weight is the sum of the layers and of the core:
f j ( x ) : = W = g a b { h c pc + i I iti [pF.<pF, + pM. ( l - <PF i)]}
(3 )
with pp M c defining the density of fibres, matrix, and core, respectively, and <pF.
denoting the fibre volume fraction.
The optimization modeling is also augmented by a number of inequality con
straints like a fibre breakage criterion, a fibre bonding criterion for the single
layers as well as a shear failure criterion for the core [B.9,B.10]. In addition, the
design variables are bounded by the following upper and lower constraint values:
0" < ock < 90'
^ min
^ max
(4 )
h,,
^ ll,, ^ h r
^ mill
c.
^,
S tru ctu ra l ana lysis
The maximum displacement wmax as the secondary objective fu n c tio n is here
determined from the following system of equations, using the FE-program system
ANSYS [A.21]:
372
18 Optimization strategies
( 5)
K(x)u = r
with
" max
K(x)
vector of displacements ,
load vector .
[mm]
( * ) =- 0.39
- ----- I( - at=0
i-1-MM-uuumii-i*=22.6mm
0.035 t=0.2 i.m
mm
(b\* - 0.450.030
\a )
0.040
0.025
t?=o.7:
mn
0.020
0.015
0.010
0.005
0.000
20
40
60
80
100
120
140
160
W p^
W [N]
Fig. D-20: Optimal fibre angle
as a function of weight W
Exercise D-18-5
373
b /a
[-]
4 - p o in t su p p o r t
- 5 - p o in t su p p o r t
------6 - p o in t su p jo r t
r~ ------- r ----------- r
'
40
80
120
160
W [N]
Fig. D-21: Optimal ratio b / a as a function of weight W
R e su lts
Some optimization results shall be presented and interpreted in the following. Fig.
D-19 shows the functional-efficient solutions of a panel supported at six points,
including different fibre orientations as well as layer - and core thicknesses. Fig.
D-20 illustrates a panel with one core thickness only, where the optimal fibre an
gles depend on the weight. It can be shown that, for a weight > 80 N, the fibre
angle is nearly constant for all types of support. The fibre angle is equal to about
45 in the case of four or six supporting points, whereas it is 30 for five sup
ports. According to Fig. D-20, a substantial change of the fibre angle occurs at the
panels with five and six point-supports and with a weight less than 60 N and
80 N, respectively. The dependence of the optimal ratio b/a on the weight is shown
in Fig. D-21. For each optimal weight, the panel with four supporting points dis
plays an almost constant ratio of w 1, while b /a ranges between 0.4 and 0.6 for
the other cases of support of the panel. The above results illustrate the importan
ce of the optimization investigations as they present an important decision tool to
the engineer for choosing a best possible design.
R e fe r e n c e s
The present work is based on the contents of the following two volumes ( in
German ):
[ETl] ESCHENAUER, H.; SCHNELL, W.: Elastizitatstheorie I - Grundlagen,
Scheiben und Platten. 2. Auflage, Mannheim, Wien, Zurich: Bl-W issenschaftsverlag 1986, 277 pages.
ESCHENAUER, H.; SCHNELL, W.: Elastizitatstheorie II - Schalen. Mann
heim, Wien, Zurich: BI-W issenschaftsverlag 1983, 269 pages.
[ET2] ESCHENAUER, H.; SCHNELL, W.: Elastizitatstheorie - Grundlagen, Flachentragwerke, Strukturoptimierung. 3. vollstandig iiberarbeitete und erweiterte Auflage, Mannheim, Leipzig. Wien, Zurich: Bl-W issenschaftsverlag
1993, 491 pages.
ESCHENAUER, H.; SCHNELL, W.: Elastizitatstheorie - Form el- und Aufgabensammlung, Mannheim, Leipzig. Wien, Zurich: BI-Wissenschaftsverlag
1994, 279 pages.
F u n d a m e n ta ls o f e la s t ic it y
C h a p te r 2 t o 7
[A.l]
[A.2]
[A.3]
[A.4]
[A.5]
[A.6]
[A.7]
[A.8]
GREEN, A.E.; ZERNA, W.: Theoretical Elasticity. 2nd ed. Oxford: Claren
don Press 1975
[A.9]
376
References
P la n e lo a d - b e a r in g str u c tu r e s
C h apter 8 to 10
[B.l]
[B.2]
GREEN, A.E.; ZERNA, W.: Theoretical Elasticity. 2nd ed. Oxford: Claren
don Press 1975
[B.3]
JAHNKE, E.; EMDE, F.; LOSCH, F.: Tafeln hoherer Funktionen. 7. Aufl.
Stuttgart: Teubner 1966
[B.4]
[B.5]
[B.6]
[B.7]
[B.8]
References
[B.9]
377
C u r v e d lo a d - b e a r in g str u c tu r e s
C h a p te r 11 to 14
[C.l]
[C.2]
[C.3]
DONNELL, L.H.: Beams, Plates, and Shells. New York: M cGraw-Hill 1976
[C.4]
[C.5]
[C.6]
[C.7]
GECKELER, J.W.: Zur Theorie der Elastizitat flacher rotationssymmetrischer Schalen ( in German ). Ing.-Arch. 1 (1930 ) 255 - 270
[C.8]
GIBSON, J.E.: Thin Shells, Computing and Theory. Oxford, London, New
York, Paris: Pergamon Press 1980
[C.9]
[C.10] GOULD, P.L.: Analysis of Shells and Plates. Berlin, Heidelbeig, New York:
Springer 1988
[C.ll] GREEN, E.; ZERNA, W.: Theoretical Elasticity. 2nd ed. Oxford: Clarendon
Press 1975
[C.12] KOITER, W.; MIKHAILOV, K.G.: Theory of Shells. Amsterdam, New
York, Oxford: Proc. IUTAM Symp., North-Holi. Publ. Comp. 1980
[C.13] KRATZIG, W.B.: Thermodynamics of Deformations and Shell Theory.
R uhr-U ni Bochum, Inst. f. Konstr. Ing. Bau (1971) Techn. Wiss. Mitt.
7 1 -3
[C.14] KRATZIG, W.B.: Introduction to General Shell Theory. In: Thin Shell
Theory - New Trends and Applications, ed. by W. OLSZAK. Wien, New
York: Springer 1980, 3 -6 1
[C.15] MARGUERRE, K.: Zur Theorie der gekriimmten Platte grofier Formanderung ( in German ). Proc. 5th Int. Congr. Appl. Mech. ( 1939 ) 93 - 101
[C.16] MORLEY, L.S.: An Improvement on DONNELL's Approximation for
Thin-W alled Circular Cylinders. Quart. Journ. Mech. and Appl. Math. XII
(1959 ) 89 - 99
378
[C.17]
References
NAGHDI, P.M.: Foundations of Elastic Shell Theory. In: Progress in Solid
Mechanics, Vol. IV ed. by SNEDDON/HILL. Amsterdam: North-Holi.
Publ. Comp. 1963
[C.18] NIORDSON, F.I.: Shell Theory. Amsterdam, New York, Oxford: North Holland Series in Appl. Math, and Mech. 1985
[C.19] NOVOZHILOV, V.V.: The Theory of Thin Shells. Groningen: Noordhoff
1970
[C.20] REISSNER, E.: Stresses and Small Displacements of Shallow
Shells. J. Math. Phys. 25 (1946 ) 80 - 85, 279 - 300; 27 (1948),
(1959 ), 16 - 35
Sherical
240; 38
S tr u c tu ra l o p tim iz a tio n
C h a p te r 15 t o 18
[D.l]
[D.2]
[D.5]
[D.6]
References
379
[D.7]
[D.8]
[D.9]
[D.10] CHENEY, E.W.; GOLDSTEIN, A.A.: NEW TON-Method and Convex Pro
gramming and CHEBYSCHEV Approximation. J. Num. Math. 1 (1959)
253 - 268
[D.ll] DAVIDON, W.C.: Variable Metric Method for Minimization.
Research and Development Report ANL-5990, 1959,
A.E.C.
[D.12] ESCHENAUER, H.: The Three C olum ns for Treating Problems in Opti
mum Structural Design. In: BERGMANN, H.W. ( e d .): Optimization: Me
thods and Applications, Possibilities and Limitations. Berlin, Heidelberg,
New York: Springer 1989, 1 - 2 1
[D.13] ESCHENAUER, H.: Shape Optimization of Satellite Tanks for Minimum
Weight and Maximum Storage Capacity. J. Structural Optimization 1
(1989 ) 171 -180
[D.14] ESCHENAUER, H.A.; KOSKI, J., OSYCZKA, A.: Multicriteria Design Opti
mization. Berlin, Heidelberg, New York: Springer 1990
[D.15] ESCHENAUER, H.A.; SCHUHMACHER, G.; HARTZHEIM, W.: Multidisci
plinary Design of Composite Aircraft Structures by LAGRANGE. In: Com
puters & Structures, 44, 4 (1992 ), 877 - 893
[D.16] ESCHENAUER, H.; WEINERT, M.: Structural Optimization Techniques as
a Mathematical Tool For Finding Optimal Shapes of Complex Shell Struc
tures. In: GIANESSI, F. (Ed.): Nonsmooth Optimization Methods and Ap
plications. Switzerland, Australia, Belgium: Gordon and Breach 1992,
135 -153
[D.17] FIACCO, A.V.; McCORMICK, G.P.: Computational Algorithm for the Se
quential Unconstrained Minimization Technique for Nonlinear Program
ming. J. Management Sci. 10 (1964 ) 601 - 617
[D.18] FLETCHER, R.; POWELL, M.J.D.: A Rapidly Convergent DescentMethod
for Minimization. Computer J. 6 (1963 ) 163 -168
[D.19] FLETCHER, R.; REEVES, C.M.: Function Minimization by Conjugate Gra
dients. Computer J. 7 ( 1964 ) 149 -154
[D.20] GRIFFITH, R.E.; STEWART, R.A.: A Nonlinear Programing Technique for
the Optimization of Continuous Processing Systems. J. Management Sci. 7
(1961) 379 - 392
[D.21] HAFTKA, R.T.; GURDAL, Z.; KAMAT, M.P.: Elements of Structural Opti
mization. Dordrecht, Boston, London: Kluwer Academic Publishers, 2nd ed.
1990
[D.22] HAUG, E.J.; ARORA, J.S.: Applied Optimal Design. New York, Toronto:
John Wiley & Sons 1979
[D.23] HESTENES, M.R.; STIEFEL, E.: Methods of Conjugate Gradients for Sol
ving Linear Systems. J. Res. Nat. Bur. Std. B49 (1952 ) 409
380
References
Applied
Nonlinear
Programming.
New
York:
[D.25] KIEFER, J.: Sequential Minimax Search for a Maximum. Proc. Am. Math.
Soc. 4 (1953 ) 105 -108
[D.26] KUHN, H.W.; TUCKER, A.W.: Nonlinear Programming. In: NEYMAN, J.
( Ed . ): Proceedings of the 2nd Berkeley Symposium on Mathematical Stati
stics and Probability. University of California, Berkeley 1951, 481 - 492
[D.27] LUKASIEWICZ, S.: Local Loads in Plates and Shells. Alphen aan den Rijn:
Sijthoff and Noordhoff 1979
[D.28] MROZ, Z.; DEMS, K.: On Optimal Shape Design of Elastic Structures. In:
ESCHENAUER, H.; OLHOFF, N. ( ed s.): Optimization Methods in Structu
ral Design. Mannheim, BI-W issenschaftsverlag 1983, 224 - 232
[D.29] OLHOFF, N.; TAYLOR, J.E.: On Structural Optimization. J. Appl. Mecha
nics, 50 (1983 ) 1139 -1151
[D.30] OLHOFF, N.; LUND, E.; RASMUSSEN, J.: Concurrent Engineering Design
Optimization in a CAD Environment. In: HAUG, E.J. ( e d .): Concurrent
Engineering Tools and Technologies for Mechanical System Design. New
York: Springer-Verlag 1993, 523 - 586
[D.31] OLHOFF, N.; LUND, E.: Finite Element Based Engineering Design Sensiti
vity Analysis and Optimization. In: HERSKOVITS, J. ( ed. ): Advanced in
Structural Optimization. Dordrecht, The Netherlands: Kluwer Academic Pu
blishers 1995, 1 - 4 5
[D.32] OSYCZKA, A.: Multicriterion Optimization in Engineering. New York, To
ronto: John Wiley & Sons 1984
[D.33] PAPALAMBROS, P.Y.; WILDE, D.J.: Principles of Optimal Design. Cam
bridge, New York, Melbourne, Sydney: Cambridge University Press 1988
[D.34] PARETO, V.: Manual of Political Economy. Translation of the French edi
tion by A.S. SCHWIER (1927). London-Basingslohe: The McMillan Press
1972
[D.35] PARKINSON, A.; WILSON, M.: Development of a Hybrid SQP-GRG Al
gorithm for Constrained Nonlinear Programming. Proc. ASME Design En
gineering Technical Conferences, Columbus/Ohio 1986
[D.36] PIERRE, D.A.; LOWE, M.J.: Mathematical Programming via Augmented
Lagrangians. London: Addison Wesley 1975
[D.37] POWELL, M.J.D.: An Efficient Method for Finding the Minimum of a
Function of Several Variables without Calculating Derivations. Computer J.
7 ( 1964 ) 155 -162
[D.38] POWELL, M.J.D.: VMCWD: A FORTRAN Subroutine for Constrained Op
timization. University of Cambridge, Report DANTP 1982/NA4
[D.39] ROZVANY, G.I.N.: Structural Design via Optimality Criteria. Dordrecht,
Boston, London: Kluwer Academic Publishers 1989
[D.40] SCHITTKOWSKI, K.L.; HORNLEIN, H.: Numerical Methods in FE -B ased
Structural Optimization Systems. ISNM-Series, Zurich: Birkhauser 1993
[D.41] SCHMIT, L.A.; MALLET, R.H.: Structural Synthesis and Design Parame
ters. Hierarchy Journal of the Struct. Division, Proceedings of the ASCE,
Vol. 89, No. 4, 1963, 269 - 299
References
381
Subject index
A-conjugate directions
312
AIRY Is stress function
50, 93, 123, 235
Algorithm of conjugate gradients
313, 340
Aluminiumhoneycomb core
370
Analogy disk-plate
51
Antisymmetrical tensors of second order
11
Arc element, length of
204
Area of a surface element
204
Auxiliary variable method
324
Axisymmetrical state of stress
98
Axisymmetrical loads
216, 224
-, circular cylindrical shell
225
spherical
226
-, conical shell
226
B
B-spline-functions
332
Barrier function
315
Base, covariant
8
, contravariant
8
oblique
53
Base vectors, covariant
7, 13, 203
contravariant
9, 204
Basic theory of shells
209
Behavioral constraints
305
BELTRAMI differential equation
87
BELTRAMI-MICHELL s equations
49
Bending angle
214, 223, 225
Bending theory of circular cylindrical shell 233
-, of shells of revolution
222
BERNSTEIN-polynomium
333
BESSEL function
183
BESSEL's differential equations
105
BETTI, theorem
46
B^ZIER-curves
333
BFGS-formula
320
Bipotential equation
50ff, 93ff
Bipotential operator
15
Boiler equation
225
Boiler formula
217, 218
Boiler structure, stiffened
348
Bound method
328
Bound variable
328
Boundary conditions 103, 117, 147, 158, 234,
242, 244
-, elastically supported
158
-, NAVIER s
103
plate with mixed
170
Boundary disturbances of circular cylindrical
shells
228
fast decaying
237
slowly decaying
238
, mixed
-, second
BOUSSINESQ's formulas
BROYDEN
Buckling load, maximizing the
optimal
Buckling modes
147
48
48
48
91
320
355
355
189, 194
c
Carbon Fibre Reinforced Plastics CFRP
37
-, plate made of CFRP
118 ff
-, circular disk made of CFRP
139
Cartesian coordinates
25, 32, 35, 106, 107
isotropic disk
93
-, plates
100
Casing with toroidal shell shape
260
CASSINI-curve
367
CASTIGLIANO, theorems
45, 264
CASTIGLIANO and MENABREA principle 45
CAUCHY s formula
20
CHEBYCHEV-functions
332
CHRISTOFFEL symbols
14, 17
- of the first kind
14
- of the second kind
14
- in surface theory
205
Circular plate on elastic foundation
180
-, centre-supported
184
-, thin
195
Circular toroidal shell
260
Compatibility conditions
31, 73
Complementary energy, specific
41
Complementary work
41
COMPLEX algorithmby BOX
351
Complex solution method
145
Complex stress function
97ff
Composite materials
118
multilayer
119
Compression modulus
34
Conditions for a minimum
307
Conformal mapping
145
Conical shell
218, 221, 226
-, boundary disturbances of
228
Conical surface
247
Constitutive equations
214, 221
-, isotropic shells
213
Constitutive laws of linear elastic bodies
31
Constrained optimization problem
356
Constraint operator
329
Constraint-oriented transformation 328, 352,
363
Constraints
303
-, active
306
384
Subject Index
- , b e h a v io ra l
305
- , g e o m e tric al
314
- , p ro b le m s w ith
310
- , p ro b le m s w ith o u t
8 210
C o n tra v a ria n t b ase
204
C o n tra v a ria n t b a se vectors
9, 204
C o n tra v a ria n t c o m p o n e n ts
332
C o n tro l polygon
C on v ey o r b e lt d ru m
360
C o o rd in a te tra n s fo rm a tio n
21
25, 32, 35, 106, 107
C o o rd in a te s, C a rte s ia n
c u rv ilin e a r
13, 25, 36, 105, 106
c y lin d ric a l
16, 25, 30
e llip tic a l-h y p e rb o lic a l
63, 70, 76
obliqu e
53, 60
p o la r
98, 106, 112
sp h erica l
26, 30
113
C ou p led d isk -p la te p ro b lem
8
C o v a ria n t b ase
14
C o v a ria n t d e riv a tiv e s
9
C o v a ria n t m e tric c o m p o n e n ts
203
C o v a ria n t m e tric ten so r, co m p o n e n ts o f
C rite ria space
326
305
C rite rio n fu n c tio n
C u rv a tu re c o m p o n e n ts
204, 205
C u rv a tu re , G A U SSIA N
205, 208, 210, 249,
- , m ean
205, 208, 249
213
- , s h ea r-rig id sh ell w ith w eak
- , te n so r of
204, 249
C y lin d ric a l s h ell
218, 220, 225, 226, 264
233
- , b e n d in g th e o ry
202
C y lin d ric a l surfaces
C y lin d ric a l tu b e
283
D
D A N T Z IG
319
D E S A IN T V E N A N T
31
D eadw eight,
283
D ecay fa c to r
225, 227
D eflections, p la n e s tru c tu re s w ith large
113,
195
231
, shells w ith la rg e
D e fo rm a tio n energy, sp ecific
41, 215, 221
D e fo rm a tio n g ra d ie n t
28
D e rivatives, c o v a ria n t
14
D esign o p tim iz a tio n p ro b lem s
304
D esign space
304, 326
D esign Space M eth o d
324
D esign v a ria b le s
302 ff
D e te rm in a n t, te n s o r o f c u rv a tu re
205
, m e tric te n so r
8
- , s h ell te n so r
210
- , su rface te n so r
204
D ifferen tial e q u atio n , b o ile r
273
- , e llip tic a l ty p e
219
EULER
183, 210
- , h y p e rb o lic a l ty p e
219
- , V O N K A R M A N 's
117
- , BESSEL a
105
, coupled
243
E ffective in -p la n e s h e a r force
E ffective tra n s v e rse s h e a r force
103, 105,
E ig en freq u en cies
E ig en v alu es o f a s y m m e tric a l te n so r
E IN S T E IN s s u m m a tio n co n v en tio n
E la s tic e n erg y o f fo u n d a tio n
E la s tic e n erg y o f p la te
E la s tic -p la stic sta te
36, 43,
E la s tic ity m a trix
- p ro p e rtie s
33,
- te n so r
E llip se fu n c tio n s w ith v a ria b le e x p o n e n t
234
234
296
F easib le d o m a in
F IA C C O
307
314
12
329
180
180
32
121
34
213
333,
367
E llip tic a l-h y p e rb o lic a l c o o rd in a te s
63, 70, 76
E llip tic a l p a ra b o lo id su rface
241 ff
E n e rg y ex p re ssio n s
40, 106
E n e rg y fu n c tio n a l, H E L L IN G E R -R E IS S N E R
114
E n e rg y p rin c ip le s
39 ff, 80
E q u ilib riu m a t la rg e
123, 252
E q u ilib riu m c o n d itio n s 25, 213, 215, 218, 222,
225, 233, 234, 242, 267
E U C L ID E A N s p ac e
5, 8, 10,
304
E U L E R e q u atio n s
330
E U L E R 's d iffe re n tia l e q u atio n
81, 112, 185
E x c h a n g in g in d ices, ru le o f
9
E x te r n a l p e n a lty fu n c tio n
346
Subject Index
F IB O N A C C I-search
311
F in ite E le m e n t M eth o d (FEM )
83
F ir s t fu n d a m e n ta l fo rm o f surface
248
FLETCH ER
320
F L E T C H E R -R E E V E S
337, 340
F L E T C H E R -R E E V E S -m e th o d
313
F le x ib ility te n so r
33
- m a trix
43
F le x ib le sh ells, th e o ry of
238
F L U E G G E , sh e a r-rig id th e o ry
233
F o rc e -q u a n tity p ro c e d u re
228
F o u n d a tio n , e la stic e n erg y o f th e
180
96, 216
F O U R IE R series e x p a n sio n
F u lly -stre ssed design
368
F uel ta n k o f a sa te llite
364
F u n c tio n a l efficiency
326
F u n c tio n a l m a trix
199
F u n c tio n a l-e ffic ie n t b o u n d a rie s
368
F u n c tio n a l-e ffic ie n t set
352, 353
F u n d a m e n ta l form , first
203
F u n d a m e n ta l q u a n titie s, fir s t o rd e r
203, 247
, second o rd e r
203, 204, 248
385
H O O K E 's law
34
H O O K E 's law, D U H A M E L-N E U M A N N fo rm of
85
H O O K E -D U H A M E L s law
32
H O O K E A N b o d ies
31
H y b rid p ro c e d u re Q P R L T
320
267, 296
H y p a r shell
H y p e rb o lic a l p a ra b o lio d s h ell
241, 242, 267
218
H y p e rb o lic a l sh ell
I
6
In d e x ru le
329
In d ire c t m e th o d s
In flu en c e coefficients
46
101
In flu en c e fa c to r
78
In g o t
In v a ria n ts
13!, 22, 24
93
Is o tro p ic d isc in C a rte s ia n c o o rd in a te s
93, 94
Is o tro p ic d isk
103
Is o tro p ic p late, tra n s v e rse ly v ib ra tin g
233
Is o tro p ic shell, g e n e ra l sh e a r-rig id
213
c o n situ tiv e e q u atio n s
J
G A L E R K IN e q u a tio n s
47, 198
m e th o d
47, 170, 195
G A U S S -W E IN G A R T E N d e riv a tiv e eq u atio n s
205
205, 208
G A U SSIA N c u rv a tu re
p a ra m e te rs
200
249
- c u rv a tu re
323
- e lim in a tio n
210
- m e a su re o f c u rv a tu re
209
- su rface p a ra m e te rs
15
- th e o re m
G E C K E L E R , m e th o d by
226
G e n e ra l b e n d in g th e o ry
233
331
G e n e ra l p o ly n o m ia l fu n c tio n
320, 368
G e n eralize d R ed u ced G ra d ie n ts
G e n e ra trix
201
G e o m e tric a l c o n stra in ts
305
- , m o d elin g
331
G e o m e try o f shells
209
G O L D FA R B
320
GOURSAT
98
G ra d ie n t m e th o d , stee p e st descen t
313
G ra d ie n t, s c a la r fu n c tio n
15
, v ecto r
15
G R E E N -D IR IC H L E T s p rin c ip le
44
G R E E N -L A G R A N G E *s c o m p o n e n ts o f s tra in
29
G R IF F IT H
317
H
H alf-space
H E L L IN G E R - R E IS S N E R fu n c tio n a l
H E R M IT E in te rp o la tio n
H E S SIA N m a trix
89
45. 82,
114
311
307, 312, 320, 337
JA C O B I A N m a trix
199
K
183
K E L V IN fu n c tio n
26
K in e m a tics o f a d e fo rm a b le b o d y
K IR C H H O F F s effective tra n s v e rse s h e a r force
103
114
n o rm a l h y p o th e sis
102, 214, 244
- , p la te th e o ry
98
K O LO SO V
8
K R O N E C K E R s d e lta
307 ff, 319
K U H N -T U C K E R c o n d itio n s
L
L A G R A N G E fo rm u la tio n
- , n o ta tio n
307, 320,
- , -fu n ction
-a u g m e n te d
- , -fu n c tio n a l
311,
- , -in te rp o la tio n
- , -m u ltip lie r-m e th o d (L P N L P )
L A G R A N G E AN a p p ro a c h
307, 330,
m u ltip lie rs
L A M c o n sta n ts
L A M -N A V IE R S e q u atio n s
L A P L A C E o p e ra to r
15, 18, 97,
L A U R E N T -series
L ayout, c o n stru ctiv e
L e a st stiffness
L en g th o f a n a rc e le m e n t
L E V Y s a p p ro a c h
Line elem en t, len g th
Line lo ad , c o n s ta n t c irc u la r
113
26
353
319
330
346
363
40
353
34
49
106
148
303
195
204
109
13
172
386
Subject Index
311
30
230
93
216
216
49, 89
M
M apping, c o n fo rm a l
145 ff
M a te ria l law
115, 2 2 1 ff
- , p la n e states
35
- , U D -la m in a te
118 ff
37
U D -lay er
M a te ria l p ro p e rtie s
303
M a th e m a tic a l P ro g ra m m in g , a lg o rith m s
310
M atrix , fu n c tio n a l
199
JA C O B I AN
199
M ax im u m ru le
6
46
M A X W E L L , th e o re m
314
M cC O R M IC K
M ean c u rv a tu re
205, 208, 249
7
M easure co m p o n e n ts
M E IS S N E R e q u atio n s
226
M e m b ra n e th e o ry o f shells
214
M em b ran e th e o ry , g e n eral e x p ressio n s
221
M E N A B R E A , th e o re m
45
M e rid io n a l curves
200
7
M etric co m p o n e n ts
, c o n tra v a ria n t
9
, c o v a ria n t
9
M etric te n s o r
13
, d e te rm in a n t
8
M in-M ax, e x te n d ed
328
M in -M ax -fo rm u la tio n s
328
M in im a , g lo b al
307
- , lo cal
307
M odeling, g e o m e tric al
331
M o d ifie d ellip se
334, 367
M odulus, sh ea r
94
YOUNGs
94
M O H R 's circle
24, 66
M O IV R E fo rm u la s
146
M oving A sy m p to te s MMA, m e th o d
321
M u ltic rite ria o p tim iz a tio n
325, 367, 371
M u ltila y er c o m p o site
119
M u ltio b je ctiv e o p tim iz a tio n
327
N
N A G H D I-sh ifte r
N A V IE R 's a p p ro a c h
b o u n d a ry c o n d itio n s
- , e q u atio n
N o n -a x isy m m e tric a l s ta te of stress
N o n -sy m m e tric a l lo a d in g
N o rm a l forces, te n s o r of
N o rm a l h y p o th e sis
K IR C H H O F F s
209
107
103
50
99
216
212
213
114
O
O b jectiv e c o n flic t
O b jectiv e fu n c tio n
303, 305,
- fu n ction, v ector
O b jectiv e fu n c tio n a ls
O b liq u e b ase
O n e -d im e n sio n a l m in im iz a tio n ste p s
O p tim a l design, sim p ly s u p p o rte d colu m n s
O p tim a lity c o n d itio n s
329,
O p tim a lity c rite rio n
O p tim iz a tio n a lg o rith m
O p tim iz a tio n lo o p
a u g m e n te d
O p tim iz a tio n m o d el
O p tim iz a tio n p ro b le m , c o n stra in e d
- , c o n tin u o u s
d iscrete
, M u ltic rite ria
325,
M u ltio b jectiv e
n o n -lin e a r (N LOP)
O p tim iz a tio n stra te g ie s
310,
, d ire c t
O p tim iz a tio n , m u ltic rite ria
, m u ltio b jec tiv e
, sh ap e
325,
O rth o tro p ic c y lin d ric a l shells
O rth o tro p ic p la te s
326
317
326
329
53
310
359
353
355
309
310
335
309
356
302
302
367
325
307
325
335
325
325
329
240
104
p
P a n e ls
370
P a ra b o lic ra d io te le sc o p e re fle c to r
370
P a ra b o lo id , e llip tic a l
241
P a ra b o lo id , h y p e rb o lic a l
241 ff
P a ra b o lo id , skew h y p e rb o lic a l
267
P a ra m e te rs , G A U SSIAN
200
P A R E T O -a p p ro a c h
329
P A R E T O -o p tim a l so lu tio n s
326
P A R E T O -o p tim a lity
325
P A R E T O -so lu tio n s
325, 329
P e n a lty fu n c tio n
314
e x te rn a l
346
- m e th o d of e x te rio r
315
- m e th o d of in te rio r
315
P e n a lty -te rm s
319
P e rm u ta tio n sy m b o l
12
P e rm u ta tio n te n so r
12
P h y sic al c o m p o n e n ts
10, 11
P la n e s tra in , s ta te of
51
P la n e stress, s ta te o f
51, 93, 147
P la n e stru c tu re s w ith la rg e d e fle c tio n s
113
P la te b u ckling, b a sic e q u atio n
118
P la te s in C a rte s ia n c o o rd in a te s
99 ff, 110
- in c u rv ilin e a r c o o rd in a te s
105
, in p o la r c o o rd in a te s
104
- s h e a r stiffness
155
- s tn p , s e m i-in fin ite
155
- K IR C H H O F F s th e o ry
102, 244
- w ith m ix e d b o u n d a ry c o n d itio n s
170
Subject Index
c irc u la r o n e la stic fo u n d a tio n
180
circu lar, c e n tre -su p p o rte d
184
c la m p ed c irc u la r
172
c la m p e d re c ta n g u la r
167
e la stic en erg y o f th e
180
e n erg y ex p ressio n
106
re c ta n g u la r stiffened
189
, re c ta n g u la r
158
- , sh e a r-e la stic
155
, s h ea r-ela stic , iso tro p ic
100
- , s h ea r-rig id , iso tro p ic c irc u la r
104
, s h ea r-rig id , o rth o tro p ic
104
th in c irc u la r
195
, tra n sv e rse ly v ib ra tin g c irc u la r
105
P O IS S O N 's e q u atio n
50
P O IS S O N s ra tio
32 ff, 37, 38, 94
P o la r c o o rd in a te s
94, 98, 104, 106, 112
P o te n tia l energy, to ta l
272
- , v o lu m e forces
93
P O W E L L m e th o d
337, 346
P O W E L L m e th o d o f co n ju g ate d ire c tio n s
312
Pow er series e x p a n sio n
95
327
P referen c e fu n c tio n
277
P re ssu re tu b e
P r in c ip a l ax es
21, 22., 31
P r in c ip a l ax es tra n s fo rm a tio n
12
P r in c ip a l s tra in s
74
22
P r in c ip a l stresses
P rin c ip le o f s ta tio n a rity
44, 45
P rin c ip le o f v irtu a l d is p la ce m e n ts
44, 80
P rin c ip le o f v irtu a l forces
44
,
-,
-f
-,
-,
Q
Q u a si-N E W T O N p ro c e d u re SQ N P
314
R
98
98
322
47
46
189
167
311
233 ff,
235
206
272
167, 180, 182, 322
131
15
38
R a d ia tin g s ta te o f s tre ss
R a d iu s-in d e p e n d e n t s ta te o f stress
R A Y L E IG H
R A Y L E IG H -R IT Z s m e th o d
R e c ip ro c ity th e o re m s
R e c ta n g u la r p la te w ith stiffe n e r
R e c ta n g u la r p late, c la m p ed
R eg u la fa lsi
R e s u lta n t fo rce -d isp la c em e n t re la tio n s
R ev o lu tio n , su rfaces of
R I T Z a p p ro a c h
- m e th o d
R o ta tin g c irc u la r d isk
R o ta tio n o f a v ecto r
- o f a U D -layer
R u le d su rface
201
s
S atellite, fuel ta n k
S c a la r fu n ctio n , g ra d ie n t
S c a la r p ro d u c t
364
15
8
387
S em i-B en d in g th e o ry
238 ff
S em i-M em b ran e th e o ry
238
S e n s itiv ity a n a ly sis
302, 310, 321
a n a ly tic a l
322
- , O v e ra ll F in ite D ifference (O FD )
322
, s e m i-a n a ly tic a l
322
S e n s itiv ity m a trix
322
S e p a ra tio n a p p ro a c h
170
S e q u e n tia l L in e a riz a tio n P ro c e d u re SL P 317, 363
S e q u e n tia l Q u a d ra tic P ro g ra m m in g S Q P
320
Series ex p an sio n , F O U R IE R
95 ff
S h a llo w shells, th e o ry o f
242
SHA N N O
320
Shape
303
- fu n c tio n s
329, 331
- o p tim iz a tio n
325, 329 ff, 366, 367
- o f sh allo w shells
242
S h e a r field th e o ry
239
S h e a r force, effective in -p la n e
234
effective tra n s v e rse
105, 234
S h e a r m o d u lu s
32, 38, 94
S h e a r stiffn ess, p la te
155
S h e a r s tra in , te c h n ica l
30, 52
S h e a r-e la stic p la te
100, 155
S h e a r-rig id o rth o tro p ic p la te
104
- p late, a n a ly tic a l so lu tio n s
107
- sh ells w ith w eak c u rv a tu re
213
- F L U E G G E 's th e o ry
233
- is o tro p ic c irc u la r p la te s
104
Shell
- e le m en t
228
- o f re v o lu tio n , e llip tic a l m e rid io n a l 251
- s h ifte r
209
- s tru c tu re s, c o m bined
228
- tensor, d e te rm in a n t of
210
- , c irc u la r co n ic al
220 ff, 226
- , c irc u la r c y lin d ric a l
220 ff, 225,
264
, c irc u la r to ro id a l
260
- , c y lin d ric a l
226
, h y p a r
267, 296
, h y p e rb o lic a l
218
, h y p e rb o lic a l p a ra b o lo id
267
- , ru le d
267
, sh e a r-rig id w ith w eak c u rv a tu re
213
- , so a p -film
241, 242
s p h e ric a l
217, 220, 226, 255
S h ells o f re v o lu tio n w ith a rb itr a r y m e rio d io n a l sh ap e
228
- , b e n d in g th e o ry
222 ff
, d e fo rm a tio n s
220
d e fo rm a tio n e nergy
221
, e q u ilib riu m c o n d itio n s
215
- , w eakly c u rv e d
223
Shells
- , la rg e d e fle c tio n s
231
, b a sic th e o ry
209
, b o u n d a ry d istu rb a n c e s
228
, c h a ra c te ris tic s of shallow
241
- , c o n stitu tiv e e q u atio n
213
- , d e sc rip tio n of
199
388
Subject Index
g e o m e try o f
209
m e m b ra n e th e o ry
214
- , o rth o tro p ic c y lin d ric a l
240
sh allo w
241
217
sp ecial
stiffen ed
239
Shell stru c tu re s, co m b in ed
228
S ide c o n stra in ts
305
Sign c o n v en tio n
19
S IM P L E X -p ro ced u re
319
S im u lta n e o u s s h ap e -th ick n e ss o p tim iz a tio n 367
Single force, to ta l w ork
40
Skew h y p e rb o lic a l p a ra b o lo id
267
Skew h y p e rb o lic a l p a ra b o lo id su rface
201 ff
S lack v a ria b le
329
S lid in g surface
203
Slow ly d ecay in g b o u n d a ry d istu rb a n c es
238
S o a p -film sh ells
241 ff
S o lu tio n m e th o d , com plex
145
Sphere, hollow
86
S p h e ric a l b o ile r
253
S p h e ric a l cap
293
S p h e ric a l co o rd in a te s
26, 30
S p h e ric a l sh ell
217 ff, 226
S p h e ric a l shell, w in d p re ssu re
255
S p h e ric a l su rface
200
S ta te o f p la n e s tra in
36, 51
S ta te o f p la n e stre ss
22, 35, 51 93, 147
S ta te o f s tra in
26
S ta te o f stress
18
a x isy m m e tric a l
98
n o n -a x is y m m e tric a l
99
- , ra d ia tin g
98
- , ra d iu s -in d e p e n d e n t
98
S ta te S pace M eth o d
324
S ta te v ecto r
229, 231
STEW ARD
317
S tiffened b o ile r stru c tu re
348
Stiffness, le a st
195
S tra in gauge ro se tte
73
S tra in te n so r
29, 74
- , lin e a r
30
S tr a in , G R E E N -L A G R A N G E s c o m p o n e n ts 29
S tra in -d is p la c e m e n t re la tio n s
30, 214, 215,
223 ff, 242
S tra in -stre ss re la tio n s
35, 36
S tra in s, p rin c ip a l
74
S tress d e v ia to r
24
Stress fu n ctio n , A IR Y s
50, 93, 123, 235
- , com plex
97, 99
S tress, u ltim a te lim it
32
re s u lta n ts
114
- , te n so r
20
18 ff
v ecto r
S tre s s -stra in re la tio n s
35, 36
S tresses, p rin c ip a l
22
S tr u c tu ra l a n aly sis
302, 309
S tr u c tu ra l m o d el
309
S tr u c tu ra l o p tim iz a tio n
301, 306
S u b s titu te p ro b le m s
325 ff
S u m m a tio n c o n v en tio n , E IN S T E IN 'S
6
SU M T
Surface
314
204
- elem ent, a re a of
199, 209
- p a ra m e te r, G A U SSIA N
247
- te n so r
203
- tensor, c o m p o n e n ts
204
- tensor, d e te rm in a n t
199
- th e o ry
203 ff
- , b ase vectors
247
- , c irc u la r c onical
- , c u rv a tu re in a p o in t o f a curve
205
242
- , e llip tic p a ra b o lo id a l
248
- , first fu n d a m e n ta l fo rm fo r
201
- , ru le d
- , skew h y p e rb o lic a l p a ra b o lo id
201
203
- , slid in g
200
- , sp h erica l
203
- , tra n s la tio n
200, 206
- , rev o lu tio n
202
- , c y lin d ric a l
11
S y m m e tric a l te n so rs o f second o rd e r
T
102
T e m p e ra tu re g ra d ie n t, pla te
128
T e m p e ra tu re field, s ta tio n a ry
114
T e n so r
- , c o v a ria n t m e tric
203, 248
204
- , c u rv a tu re
- , c u rv a tu re , d e te rm in a n t of
205
27, 28
- d is p la ce m e n t d e riv a tiv e s
- , eigenvectors o f a s y m m e tric a l
12
213
- , e la stic ity
5, 7, 10
first o rd e r
10
h ig h er o rd e r
13
m e tric
212
n o rm a l forces
12
- , p e rm u ta tio n
5, 10, 15
- , second o rd e r
- , second order, a n tis y m m e tric a l
11
11
- second order, s y m m e tric a l
15
- second order, div erg en ce of
- second order, p h y sica l c o m p o n e n ts 11
29, 74
s tra in
20
stress
203, 247
- , surface
33
- , th e rm o -e la s tic
5
z ero th o rd e r
T h e o ry o f s tru c tu re s, m e th o d
228, 277, 347
32, 34
T h e rm a l e x p a n sio n coefficient
309
T h re e -C o lu m n s-C o n c e p t
303
T o p o lo g y
272
T o ta l p o te n tia l energy
44
T o ta l p o te n tia l, v irtu a l
40
T o ta l w ork o f single force
328, 352
T ra d e -o ff m e th o d
228 ff
T r a n s fe r m a trix m e th o d
9
T ra n s fo rm a tio n b e h av io u r
22 ff
- , coefficients
- , m a trix
22, 53, 69
12
- , p rin c ip a l axes
Subject Index
rules
9, 11
- of bases
9
- of tensor of first order
10
Translation shells, equilibriumconditions 218
Translation surface
203
Truss structure
342
283
Tube, circular cylindrical
u
37
356
307
46
Water tank
Work, external
272
42
Yield point
YOUNG's modulus
32
32, 37, 94
389