Methodes Numerique
Methodes Numerique
Methodes Numerique
scalar a a
vector ~a ai
2nd order tensor a aij
4th order tensor a aijkl
matrices a
Voigt notation a→a
V(a.b)
Products
. c = ~a.~b c = ai bi
~c = a.~b ci = aij bj
c = a.b cij = aik bkj
: c=a:b c = aij bij
c=a:b cij = aijkl bkl
⊗ c = ~a ⊗ ~b cij = ai bj
c =a⊗b cijkl = aij bkl
[Einstein convention]
~
~ + dX
~u (X ~ ) = ~u (X ~ ) + ∂ u .d X
~ ) + d ~u = ~u (X ~
~
∂X
∂~u
„ «
~ + d ~u =
d ~x = d X 1+ ~
.d X
~
∂X
Transformation gradient
∂~x ∂xi
F = FiI =
~
∂X ∂XI
m0 = m
~u 0 = Q(t).~u
0
T = Q(t).T .Q(t)T
Generalization
0
T(n) = ~u1 ⊗ · · · ⊗ ~un objective if T(n) = ~u10 ⊗ · · · ⊗ ~un0 where ~ui0 = Q.~u1
F is not objective
∂~x 0 ∂~x 0 ∂~x
F0 = = . = Q.F
∂X~ ∂~x ∂ X~
B = F .F T is objective
B 0 = F 0 .F 0T = Q.F .F T .Q T = Q.B.Q T
m0 = m, ~u 0 = ~u , T0 = T
C = F T .F is invariant
C 0 = F 0T .F 0 = F T .Q T .Q.F = C
dδ~x ~
dF .δ X dF ~
= = ~ = Ḟ .F −1 .δ~x = L.δ~x
.δ X = Ḟ .δ X
dt dt dt
L can be separated into symmetric (D) and an skew-symmetric (W ) parts:
L = Ḟ .F −1 = D + W
d “ 1 2” dδ~x 1 dδ~x 2
δ~x .δ~x = .δ~x 2 + δ~x 1 . = (Dij + Wij )δxj1 δxi2 + δxi1 (Dij + Wij )δxj2
dt dt dt
=
h i h
0
Dij δxj1 δxi2 + Dij δxi1 δxj2 + Wij δxj1 δxi2 =+ Wij δxi1 δxj2
i
= 2δ~x 1 .D.δ~x 2
With
D0 = Q.D.Q T
W0 = Q.W .Q T + Q̇.Q T
Only D is objective
C = F T .F ≡ U.U
det C = (det F )2 > 0 and det C 6= 0
Eigen frame for C = P T .C 0 .P
0 1 0√ 1
c1 0 0 c1 0 0
√
C 0 = @ 0 c2 0 A → U0 = @ 0 c2 0 A
√
0 0 c3 0 0 c3
U = P T .U 0 .P
Consequently
R = F .U −1 and R T = U −1 .F T
so that
R.R T = F .U −1 .U −1 .F T = F .C −1 .F T = F .(F −1 .F −T ).F T = 1
idem for V
B = F .F T T
Bij = FiK FKj = FiK FKj
1 1“ 2 ”
Green-Lagrange E= (C − 1) = U −1
2 2
Biot strain tensor E Biot = U − 1
Logarithmic strain tensor E log = log U
3
X 3
X
U= ~i ⊗ N
λi N ~i V = λi ~ni ⊗ ~ni
i=1 i=1
so that:
3
X 3
X
C= ~i ⊗ N
λ2i N ~i B= λ2i ~ni ⊗ ~ni
i=1 i=1
One has:
3
X
V = R.U.R T = ~ i ) ⊗ (R.N
λi (R.N ~i)
i=1
Tension loading
ux = (∆L/L)x, uy = (∆l/l)y , uz = (∆l/l)z
0 1
1 + ∆L
L
0 0
F =@ 0 1 + ∆ll 0 A
B C
∆l
0 0 1+ l
∆L 2 ∆l 2
„ « „ «
C11 = 1+ C22 = C33 = 1 +
L l
! !
∆L 2
„ « „ «2
1 ∆L 1 ∆l
E11 = 1+ −1 ≈ E22 = E33 = 1+ −1
2 L L 2 l
Biot ∆L Biot Biot ∆L
E11 = , E22 = E33 =
L L
1 + γ2
0 1 0 1 0 1
1 γ 0 1 γ 0 γ 0
F = @0 1 0A C = @γ γ2 +1 0A B=@ γ 1 0A = R.C.R T
0 0 1 0 0 1 0 0 1
det F = det C = 1
Eigenvalues of C
1 2 1 1 2 1
q q
1 1+ γ + 4γ 2 + γ 4 1+ γ − 4γ 2 + γ 4
2 2 2 2
R has the form 0 1
cos θ − sin θ 0
R = @ sin θ cos θ 0A
0 0 1
τ = Jσ τij = Jσij
1 “ ”
E = F .F T − 1
2
1 “ ”
Ė = Ḟ .F T + F .Ḟ T
2
1 “ ”
or F −1 .Ė.F −T = F −1 .Ḟ + Ḟ T .F −T
2
1“ ”
F −1 .Ė.F −T L + LT = D Ė = F .D.F T
2
Second Piola-Kirchhoff stress tensor
−T
σij Dij dΩ = σij FiK ĖKL FLj−1 JdΩ0
−1 −T
= Jσij FKi FjL ĖKL Ω0
−1
= JFKi σij FjL−T ĖKL Ω0
= SKL ĖKL Ω0
with
1
S = JF −1 .σ.F −T or σ= F .S.F T
J
The Second Piola-Kirchhoff stress, S, is the work conjugate of the Green-Lagrange strain
tensor, E.
The First Piola-Kirchhoff stress, Π, is the work conjugate of the transformation gradient F
1
Π = Jσ.F −T or σ= Π.F T
J
Π is not symmetric
Finally
Z Z Z Z
σ : D dΩ = τ : D dΩ0 = S : Ė dΩ0 = Π : Ḟ dΩ0
Ω Ω0 Ω0 Ω0
Kirchhoff stress
2
τ = Jσ τ = Fk F⊥ σ τ =σ for incompressible materials
Second Piola-Kirchhoff
2
F⊥ 1
S = JF −1 .σ.F −T S= σ for incompressible materials S = σ
Fk Fk2
Q̇ = W 0 .Q − Q.W
Q̇ T = −Q T .W 0 + W .Q T
~u˙ 0 = Q̇.~u + Q.~u˙ = (W 0 .Q − Q.W ).~u + Q.~u˙ = W 0 .~u 0 − Q.W .~u + Q.~u˙
So that:
~u˙ 0 − W 0 .~u 0 = Q.(~u˙ − W .~u )
This allows to define an objective derivative of vectors (Jaumann rate):
~u J = ~u˙ − W .~u
An objective derivative (Jaumann derivative) is then obtained for second order tensors:
T J = Ṫ − W .T + T .W
σ R = R.σ.R T or σ = R T .σ R .R
Following the same methodology as for the Truesdell rate, one gets:
σ = R T .σ̇ R .R
I1 = TrC
1“ ”
I2 = (TrC)2 − TrC.C
2
I3 = det C for incompressible materials: I3 = 1
J = det F I3 = J 2
∂W ∂W
S= =2
∂E ∂C
Mooney-Rivlin law
W = C1 (I1 − 3) + C2 (I2 − 3)
Ogden
N
X µp “ αp α α
”
W (λ1 , λ2 , λ3 ) = λ1 + λ2 p + λ3 p − 3
αp
p=1
1
F →F = F such that det F = 1
(det F )1/3
then
C = F T .F
and
I1 = TrC
I2 = TrC.C
det C = 1
W = C1 (I 1 − 3) + C2 (I 2 − 3)
The constitutive equations are written in a rate form relating any objective stress rate to the
deformation rate D:
◦
σ J , σ, σ, · · · = Λ : D
These constitutive equations may be path dependent . . . not physical
F = F e .F p
Express Lp = D p + W p
~ s ⊗ ~ns ))
γ̇s = γ̇s (T : (m
Isotropic von Mises plasticity
3 T0
Dp = ṗ 0 Wp = 0
2 Teq
T rotated stress (various possibilities)
σ R = R.σ.R T
σ Q = Q.σ.Q T
where Q is obtained so that the instantaneous rotation rate of the medium wih respect to the
frame is zero:
W 0 = Q̇.Q T + Q.W .Q T = 0
so that
Q̇ = −Q.W
The corresponding strain tensor is:
Z
εQ = Q.D.Q T dt
t