Long-Time Behavior of Solution for a Reactor Model

In this paper, we would consider the dynamical behaviors of the chemical model represented by Satnoianu et al. (2001). Using the Kuratowski measure of noncompactness method, we prove the existence of global attractor 
for the weak solution semiflow of system. Finally, several numerical experiments confirm the theoretical results.


Introduction
Satnoianu et al. [1] considered a chemical model which assumed that a precursor  + is present in excess.They required further species  − to be present in the reactor at a concentration similar to that of  + though this species does not take part in the reaction.Assume that  + decays at a constant rate to form the substrate  + via  + →  + , rate =  0  0 , where  0 is the initial concentration of the reservoir species  + .The substrate  + and autocatalyst  + subsequently react by following a linear combination of quadratic and cubic steps according to the scheme  + +  + → 2 + , rate =   ,  + + 2 + → 3 + , rate =    2 ,  + +  + → 2 + , rate =  1 , (1) where   ,   ,  1 are the constants and ,  are the concentrations of  + and  + .This leads to the dimensionless reactor model where , V, ,  are the dimensionless concentrations and time and distance along the reactor, respectively.With the Dirichlet boundary condition  (, ) = V (, ) = 0,  ≥ 0,  ∈ Ω (3) and an initial condition  (0, ) =  0 () , where Ω is the bounded set [0, 1].
Here,  =   /  is the radio of diffusion coefficients of substrate  + and autocatalyst  + . is a parameter denoting the intensity of the applied electric field.The parameter  ∈ [0, 1] measures the strength of the quadratic step against the cubic step.For  = 1, the nonlinearity in the kinetics is purely quadratic, while for  = 0, it is purely cubic.
Satnoianu et al. discussed that diffusion-distributed structures (FDS) are predicted in a wider domain and are more robust than the classical Turing instability patterns.FDS also represent a natural extension of flow-distributed oscillations.Nonlinear bifurcation analysis and numerical simulations in one-dimensional spatial domains show that FDS have much richer solution behavior than Turing structures.Tang and Wang [2] showed a rigorous bifurcation analysis for the model with combination of quadratic and cubic steps by the parameters  ∈ [0, 1] and  = 0.

Advances in Mathematical Physics
For the infinite-dimensional dynamical systems about the chemical model You [3] considered the Brusselator equations; he explored a decomposition technique to show the -contraction of the solution semiflow and proved the existence of a global attractor for the solution semiflow of the Brusselator equation and the Hausdorff dimension and the fractal dimension of the global attractors are finite.You [4] studied the reversible Selkov equations with Dirichlet boundary condition on a bounded domain which is a representative cubic-autocatalytic reaction-diffusion system.He obtained the existence of ( 2 ,  1 ) global attractor and the upper semicontinuity of the global attractors in  1 product space when the reverse reaction rate tends to zero.For more details about the infinite-dimension dynamics systems about reaction-diffusion models, we can refer to [5][6][7][8][9][10][11].
In this paper we will discuss the long-time behavior of the solutions to (2)-( 4) for  = 0; that is, We prove the existence of the global attractor by Kuratowski measure.The paper is organized as follows.In Section 2, we prove the existence the weak solution of the system and absorbing property.In Section 3, we give asymptotic compactness of the semiflow by Kuratowski measure and prove the existence of the global attractor.Finally, in Section 4, we give the simulation of the system.

Absorbing Property
Define the product Hilbert spaces as follows: The norm and inner-product of  or the component space  2 (Ω) will be denoted by ‖ ⋅ ‖ and ⟨⋅⟩, respectively.The norm of   (Ω) will be represented by ‖ ⋅ ‖   if  ̸ = 2.By the poincaré inequality and the homogenous Dirichlet boundary condition there is a constant  > 0 such that We will take ‖∇‖ to the norm ‖‖  of the space .
Using the analytic semigroup generation theorem and Lumer-Phillips theorem, it is easy to check that the densely sectorial operator is the generator of an analytic  0 -semigroup on the Hilbert space.
By the Galerkin method, analogous to the arguments in [3,4], we can show the following conclusion about existence of solution.
Lemma 1.For any initial  0 ∈ , there exists a unique local weak solution ((), V()),  ∈ [0, ] for some  > 0 of the evolutionary equation (10) such that  satisfies In order to prove the existence of global attractor, we will investigate the absorbing property of the solution semiflow of problem (5).Lemma 2. Assume  > 2/27, and then, for any initial data  0 ∈  + , there exists an absorbing set  0 ∈  for the solution semiflow {()} >0 of the Cauchy problem of (5): where  0 is a positive constant independent of initial data.

The Existence of Global Attractor
In this section we will prove that the solution semiflow {()} >0 of ( 5) has a global attractor.According the existence of criterion in [6,12], we just need to prove that the semiflow {()} >0 is asymptotic compactness.Firstly, the following proposition is about the uniform Gronwall's inequality which is a basic tool in the analysis of asymptotics.
Lemma 3. Let ,  and ℎ be nonnegative function in  1  (0, ∞, ).Assume that  is absolutely continuous on [0, +∞) and the following differential inequality is satisfied: If there is a finite time  1 > 0 and some  > 0, such that for any  >  1 , where , , and  are positive constants then for any  >  1 + .
Then {()} ≥0 is asymptotically compact and there exists a global attractor A in .
Lemma 6 (see [4]).Let {()} ≥0 be a semiflow on , and there exists a global attractor A in  for this semiflow, if and only if the following conditions are satisfied: (1) There exists a bounded absorbing set  0 in  for this semiflow.
Lemma 8. Let  0 be the bounded absorbing set of solution semiflow {()} ≥0 to problem (5), and then, for any given  4 > 0, it hold that where   :  →  2 (Ω) is the orthogonal projection from the product Hilbert space  onto the first component space associated with the -component.
Lemma 10.Let  0 be the bounded absorbing set of solution semiflow {()} ≥0 to problem ( 2)-( 4), and then, for any given  4 > 0, it holds that where  V :  →  2 (Ω) is the orthogonal projection from the product Hilbert space  onto the first component space associated with the V-component.
Proof.Taking the inner-product (5) with −ΔV() over the set Ω 1 , we have and, thus, it follows that for  >  0 .In view of (62) and, applying Lemma 3, there exist  4 > 0,  6 >  4 such that it shows that  V (() 0 ) Ω(|()|< 4 ) is bounded set in  1 0 (Ω) for any  >  6 , so that The proof is completed.Proof.By Lemma 2, we have {()} ≥0 that has a bounded absorbing set  0 in  and by Lemmas 8 and 10 we have that the conditions in Lemma 4 are satisfied.So by Lemma 4 there exists a global attractor A in  for the solution semiflow {()} ≥0 generated by problem (5).The proof is completed.

Simulation
In this section, using the finite difference method to (5), the discrete equation of (5) was obtained as follows: Here ℎ is space stepsize, Δ is time stepsize, and    is a numerical approximation to (  ,   ) at (  ,   ) = (ℎ, Δ).Then we apply Euler's Method with the Dirichlet boundary condition to solve the discrete equation (5).In what follows, we always set space stepsize ℎ = 1 and time stepsize Δ = 0.5 to simulate the problem under different conditions.Now we set the parameters  = 1,  = 0.5,  = 0.1, and let initial data  0 () = cos(/10), V 0 () = cos().The numerical results are shown in Figure 1.These plots indicate that the global attractor in Section 3 is complicated and consist of many period trajectories.In the meanwhile, when we let initial data  0 () = cos(/10)V 0 () = cos(/10), the plots can be shown in Figure 2, and it indicates that the solution is convergent to a stationary solution.

Conclusion
In this paper, it can be seen from the numerical results that the global attractor A of problem (5) exists for certain  and .It includes several types of periodic solutions and quasiperiodic solutions.Besides, the solutions also reduce to stationary solution in some parameter value.All these numerical simulations are consistent with our theoretical conclusions.