Journal of Mechanics of Materials and Structures vol. 5 (2010 ... - MSP
Journal of Mechanics of Materials and Structures vol. 5 (2010 ... - MSP
Journal of Mechanics of Materials and Structures vol. 5 (2010 ... - MSP
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
PLANAR BEAMS: MIXED VARIATIONAL DERIVATION AND FE SOLUTION 779<br />
In Section 3, D−1 denotes the fourth-order elastic tensor; from now on, we use the same notation to<br />
indicate the corresponding square matrix obtained following engineering notation. Therefore, we have<br />
D −1 = 1<br />
⎡<br />
⎤<br />
1 −ν 0<br />
⎣−ν<br />
1 0 ⎦ .<br />
E<br />
0 0 2(1+ν)<br />
4.2. Formulation <strong>of</strong> the problems. We consider the special case <strong>of</strong> a beam for which<br />
¯s = 0 on A0 = ∂�s, t �= 0 on Al, f = 0 in �, t = 0 on S0 ∪ Sn.<br />
Hence, ∂�t = Al ∪ S0 ∪ Sn; the beam is clamped on the left-h<strong>and</strong> side A0, <strong>and</strong> is subjected to a nonvanishing<br />
traction field on the right-h<strong>and</strong> side Al. Furthermore, we suppose that t|Al can be exactly represented<br />
using the chosen pr<strong>of</strong>iles for σ · n. Recalling (13), <strong>and</strong> noting that n|Al = (1, 0)T , this means that there<br />
exist suitable vectors ˆtx <strong>and</strong> ˆtτ such that<br />
�<br />
pT σx t =<br />
ˆtx<br />
�<br />
. (14)<br />
p T τ ˆtτ<br />
Therefore, the boundary condition σ · n|Al = t may be written as (see (13))<br />
� � � �<br />
ˆσx(l) ˆtx<br />
= . (15)<br />
ˆτ(l) ˆtτ<br />
All these assumptions can be modified to cover more general cases; nonetheless, this simple model is<br />
already adequate to illustrate the method’s capabilities.<br />
4.2.1. HR grad-grad approach. In the notation <strong>of</strong> Section 4.1, the HR grad-grad stationarity (5) becomes<br />
δ J gg<br />
HR =<br />
� ��<br />
d<br />
� dx ET d<br />
1 +<br />
dy ET � � �<br />
T<br />
2 Psδˆs Pσ ˆσ d� + δ ˆσ<br />
�<br />
T P T ��<br />
d<br />
σ dx ET d<br />
1 +<br />
dy ET � �<br />
2 Ps ˆs d�<br />
�<br />
− δ ˆσ T P T σ D−1 �<br />
Pσ ˆσ d� − δˆs T P T<br />
t dy = 0. (16)<br />
Exp<strong>and</strong>ing (16) we obtain<br />
δ J gg<br />
HR =<br />
�<br />
(δˆs ′T P T<br />
�<br />
s E1 Pσ ˆσ + δˆs T P ′T<br />
s E2<br />
�<br />
Pσ ˆσ )d� +<br />
�<br />
�<br />
�<br />
−<br />
Using the Fubini–Tonelli theorem, (17) can be written as<br />
δ J gg<br />
HR =<br />
�<br />
where<br />
∂�t<br />
(δ ˆσ T P T σ ET 1 Ps ˆs ′ + δ ˆσ T P T σ ET 2 P′<br />
�<br />
δˆs T P T<br />
s t dy = 0. (17)<br />
δ ˆσ<br />
�<br />
T P T σ D−1 Pσ ˆσ d� −<br />
(δˆs<br />
l<br />
′T Gsσ ˆσ + δˆs T Hs ′ σ ˆσ + δ ˆσ T Gσ s ˆs ′ + δ ˆσ T Hσ s ′ ˆs − δ ˆσ T Hσ σ ˆσ )dx − δˆs T Tx<br />
Gsσ = G T σ s =<br />
�<br />
P<br />
A<br />
T<br />
s E1<br />
�<br />
Pσ dy, Hσ σ = P<br />
A<br />
T σ D−1 Pσ dy,<br />
Hs ′ σ = H T σ s ′ =<br />
�<br />
A<br />
P ′T<br />
s E2 Pσ dy, Tx =<br />
�<br />
Al<br />
P T<br />
s t dy.<br />
Al<br />
s<br />
s ˆs)d�<br />
�<br />
�<br />
� = 0, (18)<br />
x=l<br />
(19)