The existence and uniqueness of global mild solutions are proven for a class of semilinear nonautonomous evolution equations. Moreover, it is shown that the system, under
considerations, has a unique steady state. This analysis uses,
essentially, the dissipativity, a subtangential condition, and the
positivity of the related C0-semigroup.
1. Introduction
Several chemical and biochemical processes are typically described by nonlinear coupled
partial differential equations “PDE” and hence by distributed
parameter models (see [1] and the references within). The source of
nonlinearities is essentially the kinetics of the reactions involved in the
process. For numerical simulation as well as for control design problems, many
authors approximate those distributed parameter systems by lumped parameter
models [1–5]. However, an important number of questions remained
unsolved. In particular, to study the stability of the tubular reactor, the
trajectory must exist on the whole real positive time interval [0,+∞[. In our previous
works [6, 7], we have proven the global state trajectories existence for a class
of nonlinear systems arising from convection-dispersion-reaction systems,
assuming that the inlet concentrations are independent of time. In this paper,
we investigate the question in the case where the involved inlet concentrations
are functions of time t. The considered
class of models correspond to the following chemical reaction: nA+mB→P, whose kinetic is given by r=(−k1CmLn,−k2CmLn)T, where C and L are the
concentrations of the reactants A and B, respectively, k1 and k2 are the kinetic
constants and m,n are the order
of the reaction to A and B, respectively. More precisely,
we study the global existence and uniqueness of the trajectories of the models
which describe the evolution of two reactant concentrations C and L: ∂C∂t=−ν∂C∂ξ+D1∂2C∂ξ2−k1CmLn,∂L∂t=−ν∂L∂ξ+D2∂2L∂ξ2−k2CmLn,for ξ∈]0,l[andt>0, with the following boundary and initial conditions: D1∂C∂ξ(0,t)−νC(0,t)+νCin(t)=0=D1∂C∂ξ(l)∀t>0D2∂L∂ξ(0,t)−νL(0,t)+νLin(t)=0=D2∂L∂ξ(l,t)∀t>0,C(ξ,0)=C0(ξ),L(ξ,0)=L0(ξ)forξ∈]0,l[.
Additionally, the existence and uniqueness of the
corresponding equilibrium profile will be proven.
In the above equations, D1,D2 are the
dispersion coefficients, ν is the
superficial fluid velocity, t,ξ denote the time
and space independent variables, respectively, l is the length
of the reactor, m and n are two
positive integers, Cin and Lin are the inlet
concentration. For further discussion of parameters, we refer to [3].
Comment 1.
(i) The nonlinear
models considered in this paper have been studied in a qualitative manner by
several authors. In the case, ν=0, [8]
established the asymptotic behavior of solutions for the second-order reaction
(i.e., n=m=1). N. Alikakos [9]
established global existence and L∞ bounds of
positive solutions, when m=1 and 1<n<3/2. This latter result has been generalized by [10] for
the case m=1 and n>1.
In practice, the special cases m=n=1,2,3 have been used
as an industrial pulp bleaching model, where the two reactants are chlorine
dioxide (C) and lignin (L). In particular, [3] studied approximate solutions by
using several methods (orthogonal collocation, finite elements, and finite
difference methods), when n=m and D1=D2. The reader can find another model with D1≠D2 in [11], where
the numerical analysis has been done for m=n=1 and D2=4D1, D2=16D1 (see also
[12]).
Recently, the existence of global solutions for
problems such as (1.2)–(1.6) has been
extensively studied in [6, 7] with constant inlet concentrations.
(ii) For technological limitations and economical considerations, the following
saturation conditions are usually fulfilled for all 0≤ξ≤l and for all t≥0: 0≤C≤C¯,0≤L≤L¯,Cin(t)≤C¯,Lin(t)≤L¯, where C¯ and L¯ are positive
constants.
This paper is organized as follows. In Section 2, we will recall briefly some basic results and
preliminary facts from semilinear nonautonomous evolution equations which will
be used throughout Section 4. In Section 3, the problem (1.2)–(1.6) is converted
through some transformations to a homogeneous form where the semigroup theory applies. In Section 4 we establish the main global existence result for system (1.2)–(1.6). We report the
existence and uniqueness of equilibrium profiles results in Section 5. Finally, the main conclusions are outlined in Section 6. The background
of our approach can be found in [13–16].
2. Preliminaries
Let X be a real
Banach space with norm ∥⋅∥,J=[a,b[(a<b≤+∞), and let {𝒯(t);t≥0} be a linear
contraction C0-semigroup on X generated by 𝒜. Let ℬ be a nonlinear
continuous operator form Ω into X, where Ω is a subset of J×X. I and 𝕀 denote,
respectively, the identity operator of X and the
function identically equal to 1.
This section is devoted to investigate sufficient
conditions for the existence and uniqueness of global mild solutions to the
following abstract Cauchy problem: x˙(t)=𝒜x(t)+ℬ(t,x(t)),τ<t<b,x(τ)=xτ∈Ω(τ), where Ω(τ) denote the
section of Ω at τ∈J, given by Ω(t)={x∈X;(t,x)∈Ω}. Assume that Ω(t)≠∅ for all t∈J. Moreover, recall that d(x;𝒟)=inf{∥x−y∥,y∈𝒟}, for x∈X and 𝒟 is a subset of X.
The semilinear nonautonomous evolution equations have
been treated by a number of authors [14, 15, 17–21]. However, one may find that in most cases Ω is cylindrical,
that is, Ω=J×𝒟 [14, 22]. More
generally, the cylindrical case of Ω will not be
convenient for the study of evolution system satisfying time-dependent
constraints, that is, x(t)∈Ω(t) on J (see our
problem in Section 3). A noncylindrical Ω case was
studied in [16, 19].
The following result gives sufficient conditions for
the existence and uniqueness of global mild solutions to the semilinear
equations of type (2.1). It is a
particular version of [16, Theorem 8.1], when the nonlinear ℬ(t,⋅) is lℬ-dissipative
[16].
Theorem 2.1 (see [16]).
Suppose that the following conditions are fulfilled:
Ω is closed from
the left, that is, if (tn,xn)∈Ω,tn↑t in J, and xn→x in X as n→∞, then (t,x)∈Ω;
for all(t,x)∈Ω,liminfh↓0(1/h)d(𝒯(h)x+hℬ(t,x),Ω(t+h))=0;
ℬ is continuous on Ω and there exists lℬ∈ℝ+ such that the operator (ℬ(t,⋅)−lℬI) is dissipative on Ω(t) for all t∈J.
If Ω is a connected
subset of J×X such that for
all t∈J,Ω(t)≠∅, then, for each (τ,xτ)∈Ω, (2.1) has a unique mild solution on J.
Comment 2.
It is shown in [16] that the
“subtangential condition” (ii) is a necessary
condition for the existence of the mild solutions of (2.1). For more
details on the conditions of Theorem 2.1, we refer to [16].
In the particular case when Ω(t) is 𝒯(s)-invariant,
that is, 𝒯(s)(Ω(t))⊂Ω(t) for all t,s≥0, we have the following lemma.
Lemma 2.2.
Let ℬ:Ω→X be continuous and let Ω be closed from the left. If Ω(t) is 𝒯(s)-invariant for all t,s≥0, then the
following subtangential condition limh↓0inf1hd(x+hℬ(t,x);Ω(t+h))=0∀(t,x)∈Ω implies the
condition limh↓0inf1hd(𝒯(h)x+hℬ(t,x),Ω(t+h))=0,∀(t,x)∈Ω.
Proof.
Let (t,x)∈Ω, given ϵ>0, from condition (2.2) it follows, by [23, Lemma 3] (see also [24, Lemma 1]), that there is h∈(0,ϵ] and y∈Ω(t+h) such that ∥y−x−hℬ(t,x)∥≤hϵ. Let now u=y−x−hℬ(t,x) and v=(1/h)u. We get ∥v∥≤ϵ such that y=x+h(B(t,x)+v)∈Ω(t+h). By the
invariance properties of Ω(t), we have 𝒯(h)y∈Ω(t+h). Consequently, d(𝒯(h)x+hℬ(t,x);Ω(t+h))≤∥𝒯(h)x+hℬ(t,x)−𝒯(h)y∥,≤∥hℬ(t,x)−h𝒯(h)ℬ(t,x)−h𝒯(h)v∥,≤h∥𝒯(h)ℬ(t,x)−ℬ(t,x)∥+h∥𝒯(h)v∥,≤h∥𝒯(h)ℬ(t,x)−ℬ(t,x)∥+hϵ. By using the continuity of C0-semigroup (𝒯(t))t≥0, the desired result (2.3) is obtained.
Theorem 2.1 with Lemma 2.2 obviously imply the following.
Corollary 2.3.
Suppose that the following conditions are
fulfilled:
Ω is closed from
the left, that is, if (tn,xn)∈Ω,tn↑t in J, and xn→x in X as n→∞, then (t,x)∈Ω;
Ω(t) is 𝒯(s)-invariant, for
all t,s≥0;
for all(t,x)∈Ω,liminfh↓0(1/h)d(x+hℬ(t,x),Ω(t+h))=0;
ℬ is continuous on Ω and there exists lℬ∈ℝ+ such that the operator (ℬ(t,⋅)−lℬI) is dissipative on Ω(t), for all t∈J.
If Ω is a connected
subset of J×X such that for
all t∈J,Ω(t)≠∅, then, for each (τ,xτ)∈Ω,
(2.1) has a unique mild solution on J.
3. Abstract Semigroup Formulation
Throughout the
sequel, we assume H=L2[0,1]⊕L2[0,1], the Hilbert space with the usual inner product〈(x1,x2),(y1,y2)〉=〈x1,y1〉L2+〈x2,y2〉L2 and the induced
norm ∥(x1,x2)∥=(∥x1∥L22+∥x2∥L22)1/2 for all (x1,x2) and (y1,y2) in H.
Clearly, the Hilbert space H is a real
Banach lattice, where for all given x=(x1,x2)∈H,y=(y1,y2)∈H,x≤yiff x1(z)≤y1(z),x2(z)≤y2(z)fora.e.z∈[0,1]. Recall that for every pair x,y∈H, the set[x,y]={w∈H:x1≤w1≤y1,x2≤w2≤y2}=[x1,y1]×[x2,y2] is called the order interval between x and y. Clearly, [x,y] is nonempty if x≤y (for more
details, see, e.g., [25]). A bounded linear operator 𝒯 on H is said to be
positive if 0≤𝒯x for all 0≤x. Similarly, a
family of bounded linear operators (𝒯(t))t≥0 of H is said to be a
positive C0-semigroup on H if 𝒯(t) is a C0-semigroup on H and 𝒯(t) is a positive
operator for all t≥0.
In the following, we will assume that Cin(t) and Lin(t) are positive 𝒞1([0,∞[)-functions. Let us consider the following state transformation: z=ξl,x1=C−Cin,x2=L−Lin,x01=C0−Cin,x02=L0−Lin. Then, we obtain the new equivalent system for all z∈]0,1[ and t>0: ∂x1∂t=−v∂x1∂z+d1∂2x1∂z2−k1(x1+Cin(t))m(x2+Lin(t))n−Cin.(t),∂x2∂t=−v∂x2∂z+d2∂2x2∂z2−k2(x1+Cin(t))m(x2+Lin(t))n−Lin.(t), with di∂xi∂z(0,t)−vxi(0,t)=0=di∂xi∂z(1,t)∀t>0i=1;2,xi(z,0)=x0i(z)forz∈]0,1[,i=1;2, where d1=D1l2,d2=D2l2,v=νl.
This PDEs describing the reactor dynamics may be
formally written in the abstract form as x˙(t)=Ax(t)+B(t,x(t)),x(0)=x0∈Ω(0), where Ω(t) denote the
section of Ω at t∈ℝ+, which is given
in view of (1.7) by Ω={(t,(x1,x2))T∈ℝ+×H:−Cin(t)≤x1(z)≤C¯−Cin(t),−Lin(t)≤x2(z)≤L¯−Lin(t)a.e. z∈[0,1]}.
The linear operator A is defined by D(A)={x=(x1,x2)T∈H:x,dxdz∈Hare absolutely continuous,d2xdz2∈H,didxidz(0)−υxi(0)=0=didxidz(1);i=1;2},Ax=(d1d2x1dz2−υdx1dz00d2d2x2dz2−υdx2dz)=(A1x100A2x2). The nonlinear
operator B is defined on Ω by B(t,x)=(−k1(x1+Cin(t)𝕀)m(x2+Lin(t)𝕀)n−Cin.(t)𝕀,−k2(x1+Cin(t)𝕀)m(x2+Lin(t)𝕀)n−Lin.(t)𝕀)T. It is shown in
[7] that the linear operator A given by (3.14)
is the infinitesimal generator of contraction semigroup on HT(t)=(T1(t)00T2(t)), where T1(t) and T2(t) are the C0-semigroups generated,
respectively, by A1 and A2.
4. Global Existence
This section is concerned with the existence and the
uniqueness of mild solution for our problem given by (3.6)–(3.9) In order to be
able to apply
Corollary 2.3, we need the following lemmas.
Lemma 4.1.
For each (t,x)∈Ω,limh↓01hd(x+hB(t,x);Ω(t+h))=0.
Proof.
Let (t,x)∈Ω. Observe that Ω(t) is given by Ω(t)=Ω1(t)×Ω2(t), where Ω1(t)=[−Cin(t)𝕀,(C¯−Cin(t))𝕀],Ω2(t)=[−Lin(t)𝕀,(L¯−Lin(t))𝕀]. Denote X1(t)=x1+Cin(t)𝕀,X2(t)=x2+Lin(t)𝕀, we have, for x∈Ω(t),X(t)=(X1(t),X2(t))T∈[0,C¯𝕀]×[0,L¯𝕀]. Let h0>0 be sufficiently
small such that h0k1C¯m−1L¯n≤1.
Let, now, h∈(0,h0), then X1(t)(𝕀−hk1X1m−1(t)X2n(t))∈[0,C¯𝕀]. Hence f1(t,X(t))=X1(t)(𝕀−hk1X1m−1(t)X2n(t))−Cin(t+h)𝕀∈Ω1(t+h). By using the regularity of the inlet function Cin, we get d(x1+hB1(t,x),Ω1(t+h))≤d(X1(t)−hk1X1m(t)X2n(t)−Cin(t+h)𝕀,Ω1(t+h))+hϵ(h)≤d(f1(t,X(t)),Ω1(t+h))+hϵ(h)=hϵ(h), where ϵ(h)→0 as h→0. Whence limh↓01hd(x1+hB1(t,x);Ω1(t+h))=0. By similar
considerations as above, taking into account the regularity of the function Lin, we also get limh↓01hd(x2+hB2(t,x);Ω2(t+h))=0. Observe, now, that d(x+B(t,x),Ω(t+h))≤d(x1+B1(t,x),Ω1(t+h))+d(x2+B2(t,x),Ω2(t+h)),
combining the latter with (4.8)-(4.9) we get the desired result (4.1).
The following lemma is useful to establish the dissipativity property.
Lemma 4.2.
There exists lB∈ℝ+ such that the
operator (B(t,⋅)−lBI) is dissipative
on Ω(t) for each t≥0.
Proof.
Let t≥0 and let x,y be in Ω(t). Denote gi(t,x)=−ki(x1+Cin(t)𝕀)m(x2+Lin(t)𝕀)nfori=1,2, and let also X1(t)=x1+Cin(t)𝕀;X2(t)=x2+Lin(t)𝕀;Y1(t)=y1+Cin(t)𝕀 ,Y2(t)=y2+Lin(t)𝕀. Observe that, for each x,y∈Ω(t),(Xi(t),Yi(t))T∈[0,C¯𝕀]×[0,L¯𝕀] for i=1,2. Hence, by
applying the mean value theorem, for i=1,2, we get∥gi(t,x)−gi(t,y)∥L2≤ki(C¯2m∥X2n(t)−Y2n(t)∥L22+L¯2n∥X1m(t)−Y1m(t)∥L22)1/2≤ki(n2C¯2mL¯2n−2∥x2−y2∥L22+m2L¯2nC¯2m−2∥x1−y1∥L22)1/2≤kiC¯m−1L¯n−1max(nC¯;mL¯)∥x−y∥. Finally, ∥B(t,x)−B(t,y)∥=(∥g1(t,x)−g1(t,y)∥L22+∥g2(t,x)−g2(t,y)∥L22)1/2≤max(k1,k2)C¯m−1L¯n−1max(nC¯;mL¯)∥x−y∥. Consequently, B(t,⋅) is an lB-dissipative
operator on Ω(t) [14, page 245], where
lB=max(k1,k2)C¯m−1L¯n−1max(nC¯;mL¯).
Finally, we state the invariance properties of the state trajectories of the model given by (3.6)–(3.9).
Proposition 4.3.
One has that Ω(t)is𝒯(s)-invariant∀t,s≥0.
Proof.
Let t,s≥0 and (x,y)T∈Ω(t). We have (−Cin(t)𝕀,−Lin(t)𝕀)T≤(x,y)T≤((C¯−Cin(t))𝕀,(L¯−Lin(t))𝕀)T. Hence, by using the positivity of (T(t))t≥0 [26], we have (−Cin(t)T1(s)𝕀,−Lin(t)T2(s)𝕀)T≤T(s)(x,y)T≤((C¯−Cin(t))T1(s)𝕀,(C¯−Cin(t))T2(s)𝕀)T. Since, Ti(t)𝕀≤𝕀fori=1;2 (see [26]) and
by using the inequalities
(1.8)
(i.e., C¯≥Cin and L¯≥Lin), the invariance
of Ω(t) holds for all t≥0. Thus, (T1(s)x,T2(s)y)T∈Ω(t) for all t,s≥0.
Now, we are in
a position to state and prove our global existence result for problem (3.6)–(3.9).
Theorem 4.4.
Let Cin(t) and Lin(t) be positive C1([0,+∞[)-functions.
Then, for every x0∈Ω(0), the problem (3.6)–(3.9) has a unique
global mild solution.
Proof.
Since B is continuous function in Ω, by Corollary 2.3,
it is sufficient to prove the condition (i) in Corollary 2.3
and to check that the subset Ω is connected
Let us first show that Ω is closed from
the left.
Let tn↗t and xn∈Ω(tn) with xn→x∈H, then there
exists a subsequence of (xn) which is also
denoted by (xn) such that xn(z)→x(z), that is, on [0,1] which implies,
by continuity of Cin and Lin,thatx(z)∈[−Cin(t),C¯−Cin(t)]×[−Lin(t),L¯−Lin(t)], that is, on [0,1], hence x∈Ω(t) for each t≥0.
Let us, now, check that Ω is connected in [0,+∞[×H:
Let K=[0,C¯𝕀]×[0,L¯𝕀] and define G:[0,+∞[×K→Ω such that for
all (t,x)∈[0,+∞[×K,G(t,x)=(t,x1−Cin(t)𝕀,x2−Lin(t)𝕀)T. Since Cin and Lin are continuous
functions in [0,+∞[, it follows that G in [0,+1[×K is also a
continuous function. Observe that G is surjective;
since [0,C¯𝕀]×[0,L¯𝕀] is connected in H, we get that Ω=G([0,+∞[×K) is also
connected in [0,+∞[×H.
Thus the proof of the theorem is complete.
The next section deals with the existence and uniqueness results of equilibrium profile solutions for a nonlinear model given by
(3.6)–(3.9).
5. Equilibrium Profiles
In the steady-state solution analysis, the inlet functions Cin and Lin are independent
of time t, which implies
that the domain Ω(t) is also
independent of t. If we denote by Cin¯ and Lin¯ the values of Cin and Lin which
correspond to the steady-state solutions, the corresponding steady-state system
to the models (3.6)–(3.9) is
given by the following equations:
−vdx1dz=d1d2x1dz2−k1(x1+Cin¯)m(x2+Lin¯)n=0,−vdx2dz=d2d2x2dz2−k2(x1+Cin¯)m(x2+Lin¯)n=0, with didxidz(0)−vxi(0)=0=didxidz(1),i=1;2,Ω(t)=Δ={(x1,x2)T∈H:−Cin¯≤x1(z)≤C¯−Cin¯,−Lin¯≤x2(z)≤L¯−Lin¯for almost allz∈[0,1]}. The following existence result can be proven as in the case where Cin and Lin are independent
of time.
Theorem 5.1 (see [7, 27]).
The tubular reactor modelled by the nonlinear coupled partial differential equations given
by (3.6)–(3.9) has at least
one equilibrium profile in Δ.
The sequel of this paper will deal with the uniqueness
analysis of steady states in the important case where d1=d2=d.
First, since d1=d2=d, we denote 𝒜=d(d2/dz2)−v(d/dz)=Ai with D(𝒜)=D(Ai) for i=1;2.
Now, we derive a positivity lemma, which will play a
fundamental role in the proof of the uniqueness result of steady states.
Lemma 5.2.
Let b(⋅) be a bounded nonnegative
function defined in [0,1]. If u∈L2([0,1]) satisfies the equations 𝒜u=buin]0,1[,u∈D(𝒜), then u=0 in [0,1].
Proof.
Let u be the solution of problem (5.5), then 〈𝒜u,u〉L2=〈bu,u〉L2.
We have, 〈𝒜u,u〉L2=∫01[dd2udz2(z)−vdudz(z)]u(z)dz,=−∫01d[dudz(z)]2dz+d[dudz(1)u(1)−dudz(0)u(0)]−12v[u2(1)−u2(0)],=−d∥dudz∥L22−12vu2(1)−12vu2(0),≤0. Since b(z) is nonnegative
function in [0,1], then by (5.8) and taking into account (5.6)
〈bu,u〉L2=∫01b(z)u2(z)dz=0.
Which implies,
in view of (5.6)-(5.7), that 〈𝒜u,u〉L2=0=d∥dudz∥L22+12vu2(1)+12vu2(0). Then, we get dudz(z)=0a.e.z∈[0,1],u(0)=0=u(1). Clearly, by
using the Sobolev imbedding theorem, D(𝒜)⊂𝒞([0,1]). Therefore, u=0 since u∈D(𝒜).
Theorem 5.3.
For d1=d2=d, the steady-state problem given by (5.1)–(5.3) has a unique
solution in Δ.
Proof.
Let x=(x1,x2)Tandy=(y1,y2)T be solutions to (5.1)–(5.3) on [0,1]. To obtain the desired result, we will be showing that x=y. Let g(x1,x2)=−(x1+Cin¯𝕀)m(x2+Lin¯𝕀)n,w1=y1−x1∈D(𝒜),w2=x2−y2∈D(𝒜). Then −𝒜w1=k1(g(y1,y2)−g(x1,x2))=k1(y1+Cin¯)m[(x2+Lin¯)n−(y2+Lin¯)n]+k1(x2+Lin¯)n[(x1+Cin¯)m−(y1+Cin¯)m] Hence, by
applying the mean value theorem, we get
−𝒜w1=k1n(y1+Cin¯)mξ2n−1w2−mk1(x2+Lin¯)nξ1m−1w1, where (ξ1,ξ2) are some
intermediate values between (0,0) and (C¯,L¯).
By similar considerations as above, we also get −𝒜w2=−k2(g(y1,y2)−g(x1,x2))=−k2n(y1+Cin¯)mξ2n−1w2+mk2(x2+Lin¯)nξ1m−1w1, for the same ξ1 and ξ2.
Now, we have the following system: −𝒜w1=−a1w1+b1w2,−𝒜w2=a2w1−b2w2, where, for i=1;2,ai(z)=mki(x2(z)+Lin¯)nξ1m−1(z),bi(z)=nki(y1(z)+Cin¯)mξ2n−1(z).
Multiplying
(5.16) by k2 and (5.17) by k1, we get by addition of both equations that 𝒜w=0,w∈D(𝒜), where w=k2w1+k1w2. By Lemma 5.2,
this system has a unique solution w=0 in [0,1]. Now, let
−𝒜w2=a2w1−b2w2 and
substituting the expression w1=−k2−1k1w2 yields 𝒜w2=cw2, where c(z)=a1(z)+b2(z). Observe that,
for i=1;2, 0≤ai(z)≤mkiL¯nC¯m−1,0≤bi(z)≤nkiC¯mL¯n−1.
Let λ=max(mL¯,nC¯)max(k1,k2)C¯m−1L¯n−1, then we have 0≤c(z)≤2λ. By Lemma 5.2 we get w2=0. Thus it follows, by (5.21), that w1=0, which ensures
the desired result, that is, x=y.
6. Conclusion
In this paper, we have studied the existence and uniqueness of the global mild solution for a
class of tubular reactor nonlinear nonautonomous models. It has also been
proven that the trajectories are satisfying time-dependent constraints, that
is, x(t)∈Ω(t). Moreover, the
set of physically meaningful admissible states Ω(t) is invariant
under the dynamics of the reactions. In addition, the existence and uniqueness
results of equilibrium profiles are reported.
An important open question is the stability analysis
of equilibrium profile for system (1.2)–(1.6). This question is under investigation.
Acknowledgments
This paper presents research results of the Moroccan “Programme Thématique d'Appui à la Recherche Scientifique” PROTARS III, initiated by the Moroccan “Centre National de la Recherche Scientifique et Technique” (CNRST). The scientific responsibility rests with its authors. The work
has been partially carried out within the framework of a collaboration
agreement between CESAME (Université Catholique de Louvain, Belgium) and LINMA
of the Faculty of sciences (Univesité Chouaib Doukkali, Morocco), funded by the
Belgian Secretary of the State for Development Cooperation and by the CIUF
(Conseil Interuniversitaire de la Communauté Française, Belgium). The work of B. Aylaj is supported by a research grant
from the Agence Universitaire de la Francophonie.
DochainD.1994Louvain-la-Neuve, BelgiumUniversite Catholique de LouvainRayW. H.1981Boston, Mass, USAMcGraw-HillButterworths Series in Chemical EngineeringRenouS.2000Montreal, QC, CanadaGénie Chimique, Ecole Polytechnique de MontrealRenouS.PerrierM.DochainD.GendronS.Solution of the convection-dispersion-reaction equation by a sequencing method200327561562910.1016/S0098-1354(02)00095-9WinkinJ. J.DochainD.LigariusP.Dynamical analysis of distributed parameter tubular reactors200036334936110.1016/S0005-1098(99)00170-3MR1828862ZBL0979.93077AchhabM. E.AylajB.LaabissiM.Global existence of state trajectoties for a class of tubular reactor nonlinear modelsProceedings CD-ROM of the 16th International Symposium on the Mathematical Theory of Networks and Systems (MTNS '04)July 2004Leuven, BelgiumAylajB.AchhabM. E.LaabissiM.Asymptotic behaviour of state trajectories for a class of tubular reactor nonlinear models200724216317510.1093/imamci/dnl013MR2333043MartinR. H.Jr.Mathematical models in gas-liquid reactions19804350952710.1016/0362-546X(80)90089-9MR574370ZBL0442.35040AlikakosN. D.Lp Bounds of solutions of reaction-diffusion equations19794882786810.1080/03605307908820113MR537465ZBL0421.35009MasudaK.On the global existence and asymptotic behavior of solutions of reaction-diffusion equations1983123360370MR719974ZBL0529.35037Van ElkE. P.2001Enschede, The NetherlandsUniversity of TwenteDanckwertsP. V.1970New York, NY, USAMcGraw-HillCurtainR. F.ZwartH.199521New York, NY, USASpringerxviii+698Texts in Applied MathematicsMR1351248ZBL0839.93001MartinR. H.Jr.1976New York, NY, USAJohn Wiley & Sonsxi+440Pure and Applied MathematicsMR0492671ZBL0333.47023PazyA.198344New York, NY, USASpringerviii+279Applied Mathematical SciencesMR710486ZBL0516.47023IwamiyaT.Global existence of mild solutions to semilinear differential equations in Banach spaces1986163499530MR867577ZBL0618.34056AulbachB.Van MinhN.Nonlinear semigroups and the existence and stability of solutions of semilinear nonautonomous evolution equations19961435138010.1155/S108533759600019XMR1481548ZBL0934.34051 BarbuV.1976Editura Academiei Republicii Socialiste România, Bucharest352MR0390843ZBL0328.47035BotheD.Flow invariance for perturbed nonlinear evolution equations19961441743310.1155/S1085337596000231MR1481552ZBL0954.34054LuoZ.-H.GuoB.-Z.MorgulO.1999London, UKSpringerxiv+403Communications and Control Engineering SeriesMR1745384ZBL0922.93001PaoC. V.1992New York, NY, USAPlenum Pressxvi+777MR1212084ZBL0777.35001PavelN.Invariant sets for a class of semi-linear equations of evolution19771218719610.1016/0362-546X(77)90009-8MR0637080ZBL0344.45001IwamiyaT.Global existence of solutions to nonautonomous differential equations in Banach spaces19831316581MR693551ZBL0515.34050KenmochiN.TakahashiT.Nonautonomous differential equations in Banach spaces1980461109112110.1016/0362-546X(80)90020-6MR591303ZBL0452.34053ArendtW.GraboschA.GreinerG.19861184Berlin, GermanySpringerx+460Lecture Notes in MathematicsMR839450ZBL0585.47030LaabissiM.AchhabM. E.WinkinJ. J.DochainD.Multiple equilibrium profiles for nonisothermal tubular reactor nonlinear models2004113339352MR2055361ZBL1067.80006AchhabM. E.AylajB.LaabissiM.Equilibrium profiles for a class of tubular reactor nonlinear modelsProceedings of the 13th Mediteranean Conference on Control and Automation (MED '05)June 2005Limassol, Cyprus