A10 German Rodiak
A10 German Rodiak
A10 German Rodiak
Abstract. In this paper we study the existence and uniqueness of the weak solution of a
mathematical model that describes the nonlinear oscillations of a suspension bridge. This
model is given by a system of partial differential equations with damping terms. The main
tool used to show this is the C0 -semigroup theory extending the results of Aassila [1].
1. Introduction
Since the collapse of the Tacoma Narrows Bridge on November 7, 1940, several
mathematical models have been proposed to study the oscillations of the bridge. Lazer
and McKenna proposed a model governed by a coupled system of PDEs which takes into
account the coupling provided by the stays (ties) connecting the suspension (main) cable to
the deck of the road bed. In this model the coupling is nonlinear (for more details see [12]).
In [3] Ahmed and Harbi used the model proposed by Lazer and McKenna to do a detailed
study of various types of damping. Also, they presented an abstract approach which allows
the study of the regularity of solutions of these models.
The model of suspension bridges is given by the system of partial differential equations
mb ztt + αzxxxx − F (y − z) = f1 (zt ), x ∈ Ω, t > 0,
mc ytt − βyxx + F (y − z) = f2 (yt ), x ∈ Ω, t > 0,
z(0, t) = z(l, t) = 0, z (0, t) = z (l, t) = 0,
x x
(1.1)
y(0, t) = y(l, t) = 0,
z(x, 0) = z1 (x), zt (x, 0) = z2 (x), x ∈ Ω,
y(x, 0) = y (x), y (x, 0) = y (x), x ∈ Ω.
1 t 2
Here we denote by Ω the interval (0, l). See [3] and [12] for the physical interpretations of the
parameters α, β, the variables y, z and the boundary conditions respectively. As described in
[3] the function F represents the restraining force experienced by both the road bed and the
suspension cable as transmitted through the tie lines (stays), thereby producing the coupling
between these two.
The functions f1 and f2 represent external forces as well as non-conservative forces, which
generally depend on time, the constants mb , mc , α, β are positive and F : R → R is a
function with F (0) = 0 (F can be linear or not), see [3]. The interested reader is also
refereed to the works of Drábek et. al [6, 7] and Holubová [11] where other models for the
oscillations of the bridge are studied.
2010 Mathematics Subject Classification. 47D06, 35L20.
Key words and phrases. Suspension bridges, semigroup theory, weak solution.
∗
Corresponding author.
EJQTDE, 2013 No. 51, p. 1
The aim of this work is to study the existence and uniqueness of weak solutions for (1.1).
To do this we make use of the semigroup theory. This allows us to do it in a much simpler
way without using maximal monotone operators theory as in [10, Theorem 1] or the Galerkin
approach as in [3, Theorem 4.4].
In Section 2, we study the existence and uniqueness of weak solutions of the linear model
of suspension bridges, i.e., when F (ξ) = kξ and f1 = f2 = 0 (see, for instance, [3]). The
case F (ξ) = kξ and f1 6= f2 6= 0 was considered by Aassila in [1]. We consider the nonlinear
model in Section 3.
Here, F (ξ) = kξ, where k denotes the stiffness coefficient of the cables connecting the bridge
to the bed and suspended cable.
2.1. Existence and uniqueness of solution. Let us denote for H = L2 (Ω) × L2 (Ω),
V = H02 (Ω) × H01 (Ω), and W = (H 4 (Ω) ∩ H02 (Ω)) × (H 2 (Ω) ∩ H01 (Ω)) the Hilbert spaces
endowed with scalar products
Z
h(φ1 , ψ1 ), (φ2 , ψ2 )iH := (mb φ1 φ2 + mc ψ1 ψ2 )dx,
Ω
Z
h(φ1 , ψ1 ), (φ2 , ψ2 )iV := (α∆φ1 ∆φ2 + β∇ψ1 ∇ψ2 + k(ψ1 − φ1 )(ψ2 − φ2 ))dx,
ZΩ
h(φ1 , ψ1 ), (φ2 , ψ2 )iW := (ζ∆2 φ1 ∆2 φ2 + θ∆∇φ1 ∆∇φ2 + ξ∆ψ1 ∆ψ2 )dx
Ω
+ h(φ1 , ψ1 ), (φ2 , ψ2 )iV , ζ, θ, ξ > 0;
EJQTDE, 2013 No. 51, p. 2
and with their respective norms
Z
k(φ, ψ)kH := (mb |φ|2 + mc |ψ|2 )dx,
2
ZΩ
k(φ, ψ)k2V := (α|∆φ|2 + β|∇ψ|2 + k|ψ − φ|2 )dx,
ZΩ
k(φ, ψ)k2W := (ζ|∆2 φ|2 + θ|∆∇φ|2 + ξ|∆ψ|2 )dx + k(φ, ψ)k2V .
Ω
It is well know that norm k(·, ·)k2V defined in V is equivalent to the usual norm of H 2 (Ω) ×
H (Ω) and, consequently the norm k(·, ·)k2W defined in W is equivalent to the norm of
1
H 4 (Ω)×H 2 (Ω). Therefore, by the Sobolev embeddings in [5, p. 23], we have the embeddings
dense and compact W ⊂ V ⊂ H. Identifying H with its dual H 0 , we obtain W ⊂ V ⊂ H =
H 0 ⊂ V 0 ⊂ W 0 with embeddings dense and compact.
Let the bilinear form a : V × V → R be given by
a(u, ũ) = αh∆u1 , ∆ũ1 iL2 (Ω) + βh∇u2 , ∇ũ2 iL2 (Ω) + khu2 − u1 , ũ2 − ũ1 iL2 (Ω) , (2.2)
where u = (u1 , u2 ), ũ = (ũ1 , ũ2 ) ∈ V . To simplify the notation, we use h·, ·iL2 (Ω) = h·, ·i and
k · kL2 (Ω) = k · k.
Lemma 2.1. The bilinear form a is continuous, symmetric and coercive.
Proof. For u = (u1 , u2 ), v = (v1 , v2 ) ∈ V , we have
|a(u, v)|2 6 α2 k∆u1 k2 k∆v1 k2 + β 2 k∇u2 k2 k∇v2 k2 + k 2 ku2 − u1 k2 kv2 − v1 k2
+ 2αβk∆u1 kk∆v1 kk∇u2 kk∇v2 k + 2kαk∆u1 kk∆v1 kku2 − u1 kkv2 − v1 k
+ 2kβk∇u2 kk∇v2 kku2 − u1 kkv2 − v1 k
6 α2 k∆u1 k2 k∆v1 k2 + β 2 k∇u2 k2 k∇v2 k2 + k 2 ku2 − u1 k2 kv2 − v1 k2
+ αβk∆u1 k2 k∇v2 k2 + αβk∆v1 k2 k∇u2 k2 + kαk∆u1 k2 kv2 − v1 k2 +
+ kαk∆v1 k2 ku2 − u1 k2 + kβk∇u2 k2 kv2 − v1 k2 + kβk∇v2 k2 ku2 − u1 k2
= k(u1 , u2 )k2V k(v1 , v2 )k2V .
Thus, a(u, v) 6 kukV kvkV , and this shows that a is continuous. The symmetry of a is
immediate. For the last, a(u, u) = αk∆u1 k2 + βk∇u2 k2 + kku2 − u1 k2 = kuk2V , for all
u = (u1 , u2 ) ∈ V , and thus we have the coercivity of a.
From Lemma 2.1, there exists a linear operator C ∈ L (V, V 0 ) such that a(u, v) =
hCu, viV 0 ,V , ∀u, v ∈ V .
For u = (z, y), the system (2.1) can be written as
utt + Cu = 0, (x, t) ∈ Ω × (0, ∞)
u = ux = 0, on ∂Ω × (0, ∞) (2.3)
u(x, 0) = u (x), u (x, 0) = v (x), x ∈ Ω,
0 t 0
EJQTDE, 2013 No. 51, p. 3
where Cu = (a∆2 z − p(y − z), −b∆y + q(y − z)), a2 = mαb , b2 = mβc , p = mkb , q = mkc ,
u0 (x) = (z1 (x), y1 (x)), and v0 (x) = (z2 (x), y2 (x)).
With this notation we can see the problem (2.3) as second order ODE in H,
(
utt + Cu = 0, t ∈ [0, ∞),
(2.4)
u(0) = u0 , ut (0) = v0 ,
i.e., c is coercive.
From Lemma 3.1, there exists a linear operator A ∈ L (V, V 0 ) such that c(u, v) =
hAu, viV 0 ,V , for all u, v ∈ V .
The operator A : D(A) ⊂ H → H is the realization of the operator
Au = (a2 ∆2 u1 , −b2 ∆u2 ) (3.5)
with the boundary condition given in (1.1) and the domain given by
D(A) = {(u1 , u2 ) ∈ H : ∆2 u1 , ∆u2 ∈ L2 (Ω), u1 = ∇u1 = u2 = 0 in ∂Ω}.
Let t > 0 be and consider u(t) = (u1 (t), u2 (t)) = (z(t, .), y(t, .)) where the components are
functions defined in Ω. Also, consider the operator Fe : R+ 0 × H → H given by
Fe(t, u) = F1 (t, ., u1 (t, .), u2 (t, .)), F2 (t, ., u1 (t, .), u2 (t, .)) . (3.6)
Thus, we can write the system (3.1) as the following abstract second order ODE in the
Hilbert space H
(
utt + Au = Fe(t, u), t > 0,
(3.7)
u(0) = u0 , ut (0) = v0 ,
where U = (u, ut ) = (u, v), U0 = (u0 , v0 ), A : D(A) ⊂ H → H is given by A(u, v) = (−v, Au)
with
D(A) = {U = (u, v) ∈ V × H : (−v, Au) ∈ V × H}
= {U = (u, v) ∈ V × V : Au ∈ H}
h−AU, U
e iH = (v, −Au), (e
u, ve) H
e) + h−Au, veiH
= c(v, u
= c((zt , yt ), (ez , ye)) + (−a2 ∆2 z, b2 ∆y), (e zt , yet ) H
Z Z Z Z
2 2 2 2 2
=a ∆zt ∆e z dx + b ∇yt ∇ey dx − a ∆ ze zt dx + b ∆ye
yt dx
Ω Ω Ω Ω
Z Z Z Z
2 2 2 2 2
=a zt ∆ zedx − b y dx − a
yt ∆e ∆z∆e zt dx − b ∇y∇eyt dx
Ω Ω Ω Ω
= v, Ae
u H
+ c(u, −e
v ) = (u, v), (−e u)iH = hU, AU
v , Ae e iH .
∗
From this we have (−A) = A, and analogously we have h−AU, U iH = 0. Thus, −A and
(−A)∗ are dissipative. Now, from Corollary 4.4 [13, p. 15] the result follows.
EJQTDE, 2013 No. 51, p. 8
Theorem 3.5. Assume that F , f1 and f2 satisfy
(i) F , f1 and f2 are of class C 1 with F (0) = 0, f1 (0) = 0 and f2 (0) = 0.
(ii) F , f1 and f2 are locally Lipschitz continuous with constants M , c1 and c2 , respectively.
(iii) |F (s)|2 6 1 + N |s|2 , |f1 (s)|2 6 1 + c3 |s|2 and |f2 (s)|2 6 1 + c4 |s|2 , for all s ∈ R and
some positive constants N , c3 and c4 .
Then, for each (z1 , y1 , z2 , y1 ) ∈ V × H the problem (1.1) has a unique weak solution
(z, y) ∈ C([0, +∞), V ) ∩ C 1 ([0, +∞), H).
Proof. If U ∈ Br = {η ∈ H : kηkH 6 r} then c(u, u) + kvk2H 6 r2 . Since c is coercive and
V ,→ H it follows that kzkL2 (Ω) , kykL2 (Ω) , kzt kL2 (Ω) , kyt kL2 (Ω) 6 r. Similarly, if Ũ ∈ Br we
obtain the same estimates. Thus, for U = (u, v), Ũ = (ũ, ṽ) ∈ H we have
kF(t, U ) − F(t, Ũ )k2H = c(0, 0) + kFe(t, φ) − Fe(t, ψ)k2H
6 2(m−2 −2 2
b + mc )kF (y − z) − F (ỹ − z̃)kL2 (Ω)
+ 2m−2 2 −2 2
b kf1 (zt ) − f1 (z̃t )kL2 (Ω) + 2mc kf2 (yt ) − f2 (ỹt )kL2 (Ω)
6 4(m−2 −2 2 2 2
b + mc )M ky − ỹkL2 (Ω) + kz − z̃kL2 (Ω)
+ 2m−2 2 2 −2 2 2
b c1 kzt − z̃t kL2 (Ω) + 2mc c2 kyt − ỹt kL2 (Ω)
6 δ0 ku − ũk2H + δ1 kv − ṽk2H
6 δ0 d−1 2
0 c(u − ũ, u − ũ) + δ1 kv − ṽkH
6 Λ2 kU − Ũ k2H ,
where we used the hypothesis (ii), δ0 = 4(m−2 −2 2 −2 2 −2 2
b + mc )M , δ1 = max{2mb c1 , 2mc c2 } and
−1
Λ2 = max{δ0 d0 , δ1 }. Therefore, F is locally Lipschitz with respect to the second variable.
Now, using the hypothesis (iii), we obtain
kF(t, U )k2H 6 2(m−2 −2 2 −2 2 −2 2
b + mc )kF (y − z)kL2 (Ω) + 2mb kf1 (zt )kL2 (Ω) + 2mc kf2 (yt )kL2 (Ω)
6 2(m−2 −2 2 −2 2
b + mc )(|Ω| + N ky − zkL2 (Ω) ) + 2mb (|Ω| + c3 kzt kL2 (Ω) )
+ 2m−2 2
c (|Ω| + c4 kyt kL2 (Ω) )
6 Λ̃2 (1 + kU kH )2 ,
where δ2 = 6|Ω|(m−2 −2 −2 −2 −2 −2
b + mc ), δ3 = 4(mb + mc )N , δ4 = max{2mb c3 , 2mc c4 } and Λ̃ =
2
max{δ2 , δ3 d−1
0 , δ4 }. Thus, F satisfies the sublinear growth.
Finally, as the problem (1.1) is equivalent to (3.9), by Proposition 3.4, Theorem 1.4 [13,
p. 185] and by Theorem 11.3.5 [14, p. 261], we conclude that, for all U0 ∈ H there exists a
unique global solution U ∈ C([0, ∞), H). Thus,
U ∈ C([0, ∞), V × H) ⇒ u ∈ C([0, ∞), V ), ut ∈ C([0, ∞), H)
⇒ (z, y) ∈ C([0, ∞), V ) ∩ C 1 ([0, ∞), H).
EJQTDE, 2013 No. 51, p. 9
Acknowledgements
The authors would like to thank the referee for his/her valuable suggestions. The first
author was partially supported by FAPESP (Brazil) through the research grant 09/08088-9
and the second author was partially supported by FAPESP (Brazil) through the research
grant 09/08435-0.
References
[1] Aassila, M. Stability of dynamic models of suspension bridges. Mathematische Nachrichten, Weinheim,
v. 235, p. 5–15, 2002.
[2] Adams, R. A. Sobolev Spaces. New York: Academic Press, 1975.
[3] Ahmed, N. U.; Harbi, H. Mathematical analysis of dynamic models of suspension bridges. SIAM Journal
on Applied Mathematics, v. 58, n. 3, p. 853–874, 1998.
[4] Brézis, H. Opérateurs maximaux monotones et semi-groupes de contractions dans les espaces de Hilbert.
Amsterdam: North Holland, 1973.
[5] Cholewa, J. W.; Dlotko, T. Global attractors in abstract parabolic problems. Cambridge: Cambridge
University Press, 2000.
[6] Drábek, P.; Leinfelder, H.; Holubová, G. Coupled string-beam equations as a model of suspension bridges.
Appl. Math. 44 (1999), no. 2, 97–142.
[7] Drábek, P.; Holubová, G.; Matas, A.; Necesal, P. Nonlinear models of suspension bridges: discussion of
the results. Mathematical and computer modeling in science and engineering. Appl. Math. 48 (2003),
no. 6, 497–514.
[8] Engel, K. J.; Nagel, R. A short course on operator semigroups. New York: Springer, 2006.
[9] Friedman, A. Partial Differential Equations. New York: Holt, Rinehart and Winston, 1969.
[10] Guesmia, A. Energy decay for a damped nonlinear coupled system, Journal of Mathematical Analysis
and Applications, v. 239, p. 38–48, 1999.
[11] Holubová, G. Mathematical models of suspension bridges. Appl. Math. 42 (1997), no. 6, 451–480.
[12] Lazer, A. C.; McKenna, P. J., Large-amplitude periodic oscillations in suspension bridges: some new
connections with nonlinear analysis. SIAM Rev. 32 (1990), no. 4, 537–578.
[13] Pazy, A. Semigroups of linear operators and applications to partial differential equations. New York:
Springer-Verlag, 1983.
[14] Vrabie, I. I. C0 -semigroups and applications. Amsterdam: Elsevier, 2003.