![Page 1: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/1.jpg)
On Kelvin-Voigt model and its generalizations
Josef Malek
Mathematical institute of Charles University in Prague, Faculty of Mathematics and PhysicsSokolovska 83, 186 75 Prague 8, Czech Republic
In memory of Professor Jindrich Necas
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 1 / 30
![Page 2: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/2.jpg)
Contents
1 System of PDEs and its properties
2 Kelvin-Voigt solid and its generalizations
3 Proof of the existence result
4 Final remarks
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 2 / 30
![Page 3: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/3.jpg)
Part #1
System of PDEs and its properties
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 3 / 30
![Page 4: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/4.jpg)
Problem description
For given
Ω ⊂ Rd with ∂Ω = ΓD ∪ ΓN ; T > 0
u0, v0 (initial data); uD , g (boundary data)
%0 : Ω 7→ R density
r , q ∈ (1,∞)
to find u : (0,T )× Ω→ Rd
%0u,tt − div T = 0 in Ω× (0,T )
u(0, ·) = u0 in Ω
u,t(0, ·) = v0 in Ω
u = uD on ΓD × (0,T )
Tn = g on ΓN × (0,T )
T := µ∗(1 + |D(u)|2)(q−2)/2D(u) + ν∗(1 + |D(u,t)|2)(r−2)/2D(u,t)
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 4 / 30
![Page 5: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/5.jpg)
A Jindrich Necas program
Mathematical analysis of nonlinear (quasilinear) elliptic, parabolic andhyperbolic problems
Concept of (generalized) solution, its existence, uniqueness andregularity
Question: Provided that data are smooth, is the generalized solutionof the problem a classical solution?
∆pz := div(
(1 + |D(z)|2)(p−2)/2D(z))
2D(z) := ∇z + (∇z)T
−∆rv = 0
v,t −∆rv = 0
u,tt −∆qu = 0
u,tt −∆qu−∆ru,t = 0
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 5 / 30
![Page 6: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/6.jpg)
Several results
−∆rv = 0 (1)
v,t −∆rv = 0 (2)
u,tt −∆qu = 0 (3)
u,tt −∆qu−∆ru,t = 0 (4)
existence, uniqueness of weak solution for (1) or (2) - monotoneoperator theorypositive answer to full regularity for d = 2 - C 1,α-regularity of weaksolution - stationary problems (1967-71, with J. Stara), evolutionaryproblem (1991, with V. Sverak), extended to generalized Stokessystems (since 1996 by P. Kaplicky, J. Stara)measure-valued solution for (3) for the scalar case (together withM. Ruzicka and M. Rokyta)existence, uniqueness, C 1,α-regularity for (4) - with A. Friedman(1988), also by T. Roubıcek
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 6 / 30
![Page 7: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/7.jpg)
Several results
−∆rv = 0 (1)
v,t −∆rv = 0 (2)
u,tt −∆qu = 0 (3)
u,tt −∆qu−∆ru,t = 0 (4)
existence, uniqueness of weak solution for (1) or (2) - monotoneoperator theorypositive answer to full regularity for d = 2 - C 1,α-regularity of weaksolution - stationary problems (1967-71, with J. Stara), evolutionaryproblem (1991, with V. Sverak), extended to generalized Stokessystems (since 1996 by P. Kaplicky, J. Stara)measure-valued solution for (3) for the scalar case (together withM. Ruzicka and M. Rokyta)existence, uniqueness, C 1,α-regularity for (4) - with A. Friedman(1988), also by T. Roubıcek
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 6 / 30
![Page 8: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/8.jpg)
Avner Friedman and Jindrich Necas 1988 paper
PACIFIC JOURNAL OF MATHEMATICS
Vol. 135, No. 1, 1988
SYSTEMS OF NONLINEAR WAVE EQUATIONSWITH NONLINEAR VISCOSITY
AVNER FRIEDMAN AND JINDRICH NECAS
An equation of the form
• • A d dW(p) A d dV(q) _ f
j^dxt dp, j^dXt dqt J
where p = Vw, q = VM, U = du/dt9 u = d2u/dt2 represents, for
suitable functions W(p), V(q)9 a nonlinear hyperbolic equation with
nonlinear viscosity and it appears in models of nonlinear elasticity.
In this paper existence and regularity of solutions for the Cauchy
problem will be established. In particular, if n = 2, or if n > 3 and
the eigenvalues of (d2V/dqjdqj) belong to a "small" interval, then the
solution is classical. These results will actually be established for a
system of equations of the above type.
Introduction. Consider a system of N nonlinear equations
in a cylinder Q x (0, oo), with initial data
(0.2) uk (x, 0) = uk0(x), uk (x, 0) = uki (x)
and boundary conditions
(0.3) u = 0 if x e dn, t > 0;
here Q, is a bounded domain in R",
P = (Pu). Q = i.Qn) and
dui diii . dwPli = dx-r gii=dxl- W = ~dt-
The special case
has been studied by several authors. For n = 1, existence and unique-ness of a classical solution was established in [1], [2], [6], [7]. For
29
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 7 / 30
![Page 9: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/9.jpg)
Assumptions/1 - homogeneous Dirichlet bc’s
For given
Ω ⊂ Rd with ∂Ω = ΓD ∪ ΓN ; T > 0
u0, v0 (initial data); g (boundary data)
%0 : Ω 7→ R density
r , q ∈ (1,∞)
to find u : (0,T )× Ω→ Rd
%0u,tt − div T = 0 in Ω× (0,T )
u(0, ·) = u0 in Ω
u,t(0, ·) = v0 in Ω
u = 0 on ΓD × (0,T )
Tn = g on ΓN × (0,T )
T := E(D(u)) + S(D(u,t))
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 8 / 30
![Page 10: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/10.jpg)
Assumptions/2
(S(D1)− S(D2),D1 −D2) ∼ |D1 −D2|2∫ 1
0(1 + |D1 − s(D1 −D2)|)r−2ds
(E(D1)− E(D2),D1 −D2) ∼ |D1 −D2|2∫ 1
0(1 + |D1 − s(D1 −D2)|)q−2ds
If S(0) = 0 and E(0) = 0 then
S(D) ·D ≥ C1(|D|r − 1) E(D) ·D ≥ C1(|D|q − 1)
|S(D)| ≤ C2(1 + |D|)r−1 |E(D)| ≤ C2(1 + |D|)q−1
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 9 / 30
![Page 11: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/11.jpg)
Definition of solution
g ∈ Lr ′(0,T ; (W 1− 1r,r (ΓN)d)∗)
u0 ∈W 1,qΓD
(Ω)d , v0 ∈ L2(Ω)d
0 < m ≤ %0 ≤ M <∞ a.e. in Ω
Then u is a weak solution to the problem if
u ∈ L∞(0,T ; W 1,qΓD
(Ω)d) ∩W 1,∞(0,T ; L2(Ω)d) ∩W 1,r (0,T ; W 1,r (Ω)d)
%0u,tt ∈ Lmin(r ′,q′)(0,T ; (W1,max(r ,q)ΓD
(Ω)d)∗)∫ T
0〈%0u,tt ,ϕ〉+
(S(D(u,t)) + E(D(u)),D(ϕ)Ω
)dt =
∫ T
0〈g,ϕ〉ΓN
dt
for all ϕ ∈ Lmax(r ,q)(0,T ; W1,max(r ,q)ΓD
(Ω)d)
limt→0+
(‖u,t(t)− v0‖22 + ‖u(t)− u0‖2
1,2) = 0.
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 10 / 30
![Page 12: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/12.jpg)
Theorem 1 (M. Bulıcek, J. Malek, K.R. Rajagopal)
Theorem (Existence and Uniqueness)
Let S and E satisfy the monotone, growth and coercivity conditions with
1 < q ≤ 2 ≤ r <∞
Then, for any set of data Ω, T , %0, u0, v0 and g there is unique weaksolution to the problem.
Novel aspects
r > 2 (nonlinearity in ∇u,t)
variable density %0
no potential structure
Friedman, Necas - nonlinear S of the potential structure with r = 2and d = 2 or eigenvalues of ∂S
∂D lie in small interval
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 11 / 30
![Page 13: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/13.jpg)
Theorem 2 (M. Bulıcek, J. Malek, K.R. Rajagopal)
Theorem (Regularity)
(1) If in addition %0 ∈ C0,1(Ω) and v0 ∈W 1,2loc (Ω)d then
(1 + |D(v)|) r−22 D(∇v) ∈ L2(0,T ; L2
loc(Ω)d×d)
v ∈ L∞(0,T ; W 1,2loc (Ω)d)
v,t ∈ Lr ′(0,T ; Lr ′loc(Ω)d)
(2) If %0 ∈ C0,1(Ω), v0 ∈W 1,2(Ω)d , g ∈W 1,r (0,T ; W 1+ 1r,r (ΓN)) and
Sij(D) = ∂DijUS(D) then
v,t ∈ L2(0,T ; L2(Ω)d) v ∈ L∞(0,T ; W 1,r (Ω)d)
Elliptic regularity then implies (for smooth Ω) that
(1 + |D(v)|) r−22 D(∇v) ∈ L2(0,T ; L2(Ω)d×d)
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 12 / 30
![Page 14: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/14.jpg)
Part #2
Kelvin-Voigt solid and its generalizations
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 13 / 30
![Page 15: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/15.jpg)
Kelvin-Voigt solid - mechanical analog
Spring
Dashpot
l
σS = EεS and σD = νεD
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 14 / 30
![Page 16: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/16.jpg)
Kelvin-Voigt solid - mechanical analog
Spring
Dashpot
l
σS = EεS and σD = νεD
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 14 / 30
![Page 17: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/17.jpg)
Kelvin-Voigt solid - mechanical analog
Spring
Dashpot
l
Spring
Dashpot
l + ∆
F−F
F = FS + FD and ∆ = ∆S = ∆D
σ = σS + σD and ε = εS = εD =⇒ = ε = εS = εD
σS = EεS and σD = νεD
σ = EεS + νεD
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 15 / 30
![Page 18: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/18.jpg)
Kelvin-Voigt solid - mechanical analog
Spring
Dashpot
l
Spring
Dashpot
l + ∆
F−F
F = FS + FD and ∆ = ∆S = ∆D
σ = σS + σD and ε = εS = εD =⇒ = ε = εS = εD
σS = EεS and σD = νεD
σ = EεS + νεDJ. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 15 / 30
![Page 19: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/19.jpg)
Kelvin-Voigt solid and its linearization
Response corresponds to the mixture of solid and fluid components withneo-Hookean solid and Newtonian fluid
T = Te + Tv Te = µ∗B Tv = ν∗D
where B = FFT (right Cauchy-Green stretch tensor) andD = D(v) := 1
2
(∇v + (∇v)T
). Then
T = µB + ηD
Linearized elastic solid ε := 12
(∇u + (∇u)T
)T = µε + ηD = µε + ηε
Question: Is the model that is non-linear in the linearized strainjustifable?Yes, but one has to take a different point of view.
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 16 / 30
![Page 20: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/20.jpg)
Implicit constitutive theory (K.R. Rajagopal, 2003)
General implicit relations of the form
T = Te + Tv f(Te ,B) = 0 g(Tv ,D) = 0
include two explicit constitutive theories:
Te = f1(B) and Tv = g1(D)
andB = f2(Te) and D = g2(Tv )
Linearization of the second set leads to
ε = f2(Te) and D = g2(Tv )
If this relations are invertible we obtain the model under considerationsthat can shear thin or shear thicken, and creep (as a viscoelastic material).
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 17 / 30
![Page 21: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/21.jpg)
Implicit constitutive theory (K.R. Rajagopal, 2003)
General implicit relations of the form
T = Te + Tv f(Te ,B) = 0 g(Tv ,D) = 0
include two explicit constitutive theories:
Te = f1(B) and Tv = g1(D)
andB = f2(Te) and D = g2(Tv )
Linearization of the second set leads to
ε = f2(Te) and D = g2(Tv )
If this relations are invertible we obtain the model under considerationsthat can shear thin or shear thicken, and creep (as a viscoelastic material).
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 17 / 30
![Page 22: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/22.jpg)
Implicit constitutive theory (K.R. Rajagopal, 2003)
General implicit relations of the form
T = Te + Tv f(Te ,B) = 0 g(Tv ,D) = 0
include two explicit constitutive theories:
Te = f1(B) and Tv = g1(D)
andB = f2(Te) and D = g2(Tv )
Linearization of the second set leads to
ε = f2(Te) and D = g2(Tv )
If this relations are invertible we obtain the model under considerationsthat can shear thin or shear thicken, and creep (as a viscoelastic material).
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 17 / 30
![Page 23: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/23.jpg)
Implicit constitutive theory (K.R. Rajagopal, 2003)
General implicit relations of the form
T = Te + Tv f(Te ,B) = 0 g(Tv ,D) = 0
include two explicit constitutive theories:
Te = f1(B) and Tv = g1(D)
andB = f2(Te) and D = g2(Tv )
Linearization of the second set leads to
ε = f2(Te) and D = g2(Tv )
If this relations are invertible we obtain the model under considerationsthat can shear thin or shear thicken, and creep (as a viscoelastic material).
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 17 / 30
![Page 24: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/24.jpg)
Implicit constitutive theory - Example
Spring
Wire
σ
ε
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 18 / 30
![Page 25: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/25.jpg)
Special deformation applied to full nonlinear model
Consider the special shearing motion given by
x = X + u(Y , t) , y = Y , z = Z
and apply it to the generalized Kelvin-Voigt solid
T = Te + Tv Te = f(B) Tv = g(D)
Then one obtains the scalar equation of the above form without restrictingto small gradients of the displacement.
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 19 / 30
![Page 26: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/26.jpg)
Part #3
Proof of the existence result
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 20 / 30
![Page 27: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/27.jpg)
Definition of solution
S, E are uniformly monotone (growth, coercivness assumptions)1 < q ≤ 2 ≤ r < +∞g ∈ Lr ′(0,T ; (W 1− 1
r,r (ΓN)d)∗)
u0 ∈W 1,qΓD
(Ω)d , v0 ∈ L2(Ω)d
0 < m ≤ %0 ≤ M <∞ a.e. in Ω
Then u is a weak solution to the problem if
u ∈ L∞(0,T ; W 1,qΓD
(Ω)d) ∩W 1,∞(0,T ; L2(Ω)d) ∩W 1,r (0,T ; W 1,r (Ω)d)
%0u,tt ∈ Lr ′(0,T ; (W 1,rΓD
(Ω)d)∗)∫ T
0〈%0u,tt ,ϕ〉+
(S(D(u,t)) + E(D(u)),D(ϕ))Ω dt =
∫ T
0〈g,ϕ〉ΓN
dt
for all ϕ ∈ Lr (0,T ; W 1,rΓD
(Ω)d)
limt→0+
(‖u,t(t)− v0‖22 + ‖u(t)− u0‖2
1,2) = 0.
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 21 / 30
![Page 28: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/28.jpg)
Reformulation of the problem
%0v,t − div S(D(v))− div E(D(u)) = 0 in Ω× (0,T )
u,t = v in Ω× (0,T )
u(0, ·) = u0, v(0, ·) = v0 in Ω
u = 0, v = 0 on ΓD × (0,T )
(S(D(v)) + E(D(u)))n = g on ΓN × (0,T )
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 22 / 30
![Page 29: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/29.jpg)
Step 1 - Galerkin Approximations
un(t, x) :=n∑
i=1
ci (t)ωi (x) =⇒ vn := un,t
(%0vn,t ,ωi )Ω + (S(D(vn)),D(ωi ))Ω + (E(D(un)),D(ωi ))Ω = 〈g,ωi 〉ΓN
for all i = 1, 2, . . . , n
un(0, x) = un0(x)
vn(0, x) := un,t(0, x) = vn
0(x)
The Caratheodory theory and uniform estimates give long-timeexistence for un
Uniform estimates
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 23 / 30
![Page 30: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/30.jpg)
Step 2 - Uniform estimates
(%0vn,t , v
n)Ω + (S(D(vn)),D(vn))Ω + (E(D(un)),D(vn))Ω = 〈g, vn〉ΓN
d
dt
‖√%0v
n‖22 + ‖D(un))‖qq
+ 2(S(D(vn)),D(vn))Ω
≤ C + |(E(D(un)),D(vn))Ω|
supt∈(0,T )
(‖vn‖2
2 + ‖un‖21,2
)+
∫ T
0‖vn‖r1,r dt ≤ C
supt∈(0,T )
‖E(D(un))‖q′q′ +
∫ T
0‖S(D(vn))‖r ′r ′ ds ≤ C∫ T
0‖%0v
n,t‖r
′
(W 1,rΓD
(Ω))∗dt ≤ C
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 24 / 30
![Page 31: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/31.jpg)
Step 2 - Uniform estimates
(%0vn,t , v
n)Ω + (S(D(vn)),D(vn))Ω + (E(D(un)),D(vn))Ω = 〈g, vn〉ΓN
d
dt
‖√%0v
n‖22 + ‖D(un))‖qq
+ 2(S(D(vn)),D(vn))Ω
≤ C + |(E(D(un)),D(vn))Ω|
supt∈(0,T )
(‖vn‖2
2 + ‖un‖21,2
)+
∫ T
0‖vn‖r1,r dt ≤ C
supt∈(0,T )
‖E(D(un))‖q′q′ +
∫ T
0‖S(D(vn))‖r ′r ′ ds ≤ C∫ T
0‖%0v
n,t‖r
′
(W 1,rΓD
(Ω))∗dt ≤ C
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 24 / 30
![Page 32: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/32.jpg)
Step 2 - Uniform estimates
(%0vn,t , v
n)Ω + (S(D(vn)),D(vn))Ω + (E(D(un)),D(vn))Ω = 〈g, vn〉ΓN
d
dt
‖√%0v
n‖22 + ‖D(un))‖qq
+ 2(S(D(vn)),D(vn))Ω
≤ C + |(E(D(un)),D(vn))Ω|
supt∈(0,T )
(‖vn‖2
2 + ‖un‖21,2
)+
∫ T
0‖vn‖r1,r dt ≤ C
supt∈(0,T )
‖E(D(un))‖q′q′ +
∫ T
0‖S(D(vn))‖r ′r ′ ds ≤ C∫ T
0‖%0v
n,t‖r
′
(W 1,rΓD
(Ω))∗dt ≤ C
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 24 / 30
![Page 33: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/33.jpg)
Step 2 - Uniform estimates
(%0vn,t , v
n)Ω + (S(D(vn)),D(vn))Ω + (E(D(un)),D(vn))Ω = 〈g, vn〉ΓN
d
dt
‖√%0v
n‖22 + ‖D(un))‖qq
+ 2(S(D(vn)),D(vn))Ω
≤ C + |(E(D(un)),D(vn))Ω|
supt∈(0,T )
(‖vn‖2
2 + ‖un‖21,2
)+
∫ T
0‖vn‖r1,r dt ≤ C
supt∈(0,T )
‖E(D(un))‖q′q′ +
∫ T
0‖S(D(vn))‖r ′r ′ ds ≤ C∫ T
0‖%0v
n,t‖r
′
(W 1,rΓD
(Ω))∗dt ≤ C
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 24 / 30
![Page 34: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/34.jpg)
Step 3 - Limit as n→∞
∫ T
0〈%0v,t ,ω〉+ (S,D(ω))Ω + (E,D(ω))Ω dt =
∫ T
0〈g,ω〉 dt,
for all ω ∈ Lr (0,T ; W 1,rΓ (Ω)d)
v = u,t
Key problem is the identification of E and S. It suffices to to show that(for a subsequences un and vn)
D(vn)→ D(v) and D(un)→ D(u) a.e. in (0,T )× Ω (5)
Operators are strictly monotone
v is admissible test function for the limit equation
Limit eq. is sufficient to establish time-continuity and the attainmentof the initial conditions
un(t)− u(t) =∫ t
0 vn(τ)− v(τ) dτ + un0 − u0
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 25 / 30
![Page 35: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/35.jpg)
Step 4 - Almost everywhere convergence of D(vn)/1
lim supn→∞
∫ t∗
0(S(D(vn)),D(vn))Ω + (E(D(un)),D(vn))Ω dt
≤ −1
2‖√%0v(t∗)‖2
2 +1
2‖√%0v0‖2
2
Next, taking ω := v in the limit Eq., integrating the result over timeinterval (0, t∗) and comparing the result with the above Ineq. (usev(0) = v0)
lim supn→∞
∫ t∗
0(S(D(vn)) + E(D(un)),D(vn))Ω; dt ≤
∫ t∗
0(S + E,D(v))Ω dt
This implies that for all t∗ ∈ [0,T ]
lim supn→∞
∫ T∗
0(S(D(vn))− S(D(v)) + E(D(un))− E(D(u)),D(vn − v))Ω ≤ 0
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 26 / 30
![Page 36: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/36.jpg)
Step 5 - Almost everywhere convergence of D(vn)/2
∫ T1
0(S(D(vn))− S(D(v)),D(vn − v))Ω dt ≤ g(n) (lim sup
n→∞g(n)→ 0)
+
∫ T1
0(E(D(un))− E(D(u)),D(vn − v))Ω dt
Since q ≤ 2 ≤ r and un(t)− u(t) =∫ t
0 vn(τ)− v(τ) dτ + un0 − u0
C∗
∫ T1
0‖D(vn − v)‖2
2 dt ≤ C
∫ T1
0‖D(vn − v)‖2‖D(un − u)‖2 dt + g(n)
≤ C
∫ T1
0‖D(vn − v)‖2
∥∥∫ t
0‖D(un − u) ds
∥∥∥2
dt + g(n)
≤ (CT1 + ε)
∫ T1
0‖D(vn − v)‖2
2 dt + g(n)
If T1 is such that CT1 < C∗ then a.e. convergence follows.
Inductively on [0,T ].
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 27 / 30
![Page 37: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/37.jpg)
Step 5 - Almost everywhere convergence of D(vn)/2
∫ T1
0(S(D(vn))− S(D(v)),D(vn − v))Ω dt ≤ g(n) (lim sup
n→∞g(n)→ 0)
+
∫ T1
0(E(D(un))− E(D(u)),D(vn − v))Ω dt
Since q ≤ 2 ≤ r and un(t)− u(t) =∫ t
0 vn(τ)− v(τ) dτ + un0 − u0
C∗
∫ T1
0‖D(vn − v)‖2
2 dt ≤ C
∫ T1
0‖D(vn − v)‖2‖D(un − u)‖2 dt + g(n)
≤ C
∫ T1
0‖D(vn − v)‖2
∥∥∫ t
0‖D(un − u) ds
∥∥∥2
dt + g(n)
≤ (CT1 + ε)
∫ T1
0‖D(vn − v)‖2
2 dt + g(n)
If T1 is such that CT1 < C∗ then a.e. convergence follows.
Inductively on [0,T ].J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 27 / 30
![Page 38: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/38.jpg)
Concluding notes
Long-time and Large-data analysis for generized Kelvin-Voigt modelsfor solids (interesting for all possible values of r , q) limited to1 < q ≤ 2 ≤ r < +∞
Understanding of the model from the implicit constitute theory (orrather dual explicit constitutive theory when the kinematicalquantities are functions of stresses)
Recall the paper by A. Friedman and J. Necas (WOS - 17 citations,MathSciNet - 13, Google Scholar - 37)
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 28 / 30
![Page 39: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/39.jpg)
Citation of the A. Friedman and J. Necas paper
0
1
2
3
41992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Years
Citations in Each Year
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 29 / 30
![Page 40: On Kelvin-Voigt model and its generalizations · 2012. 7. 18. · On Kelvin-Voigt model and its generalizations Josef M alek Mathematical institute of Charles University in Prague,](https://reader035.vdocuments.mx/reader035/viewer/2022071505/61263d2ec0d3e442cc598e75/html5/thumbnails/40.jpg)
Part #4
Final remarks
J. Malek (MFF UK) On Kelvin-Voigt solid December 14, 2009 30 / 30