Research Journal of Mathematics and Statistics 3(1): 28-33, 2011 ISSN: 2040-7505 © Maxwell Scientific Organization, 2011 Received: September 23, 2010 Accepted: December 30, 2010 Published: February 15, 2011 Existence of Strong Solutions of the Energy Equation in the Axisymmetric Flow 1 Chérif Amrouche, 1,2,3Macaire Batchi and 2Jean Batina Laboratoire de Mathématiques Appliquées CNRS UMR 5142, Université de Pau et des Pays de l'Adour 2 Laboratoire de Thermique Energétique et Procédés, Université de Pau et des Pays de l'Adour, Avenue de l'Université 64000 Pau, France 3 Université Marien NGouabi, B.P.69, Brazzaville,Congo 1 Abstract: This study present the existence of the strong solutions of the energy equation associated with the Navier-Stokes equations with nonhomogeneous boundary conditions in axisymmetric flow case. It’s mainly to see the heat transfer when forced convection. Key words: Forced convection, energy equation, incompressible fluid, Navier-Stokes equations, nonhomogeneous boundary conditions, temperature of wall INTRODUCTION y The object of this study proposed in this paper, primarily aimed at considering the nonhomogeneous boundary conditions associated to the energy equation describing the temperature distribution in the axisymmetric geometry flow when the velocity appearing in the convective term obeys to the time periodic boundary conditions. The solutions of the thermal problem are obtained by takin account of the estimates a priori bearing on the velocity and also of the suitable assumptions of the wall temperature. Let be an open and bounded domain of ú2 of class C1,1 with ' = Γ0 U Γ1 U Γ2 . Furthermore in this study suppose that 'i = 'j = Ø, for i … j as shown in the Fig. 1, where '0 and '1 are the input and output sections of the channel respectively and '2 the sinusoidal wall. According to the Fig. 1, x is the longitudinal variable. Consider the thermal problem, which is described by the equation of energy with the convection ensured by the velocity of the fluid of the system (2). The temperature 2(x, t) of the fluid satisfies the system (1), in which suppose that there is not the external source of heat: ⎧ ∂ θ + (v. ∇ )θ − a∆ θ ⎪ ∂t ⎪θ = θ ∞ ⎪⎪ ⎨θ = θ p ⎪ ∂θ ⎪ ∂t = 0 ⎪ ⎪⎩ θ ( 0) = θ 0 Γ2 Γ0 x Γ2 Fig. 1: Geometry of channel with a>0, T>0 and 24, 2p, 20 given reals with 2p … 24. Moreover, there is the velocity field v solution of the Navier-Stokes equations: ⎧ ∂ v − v∆v + v. ∇v ⎪∂t ⎪ + ∇p = 0 ⎪ ⎨divv = 0 ⎪v = g ⎪ ⎪v( 0) = v 0 ⎩ = 0 in QT = Ω × ]0, T[ on Γ 0 × ]0, T[ on Γ 2 × ]0, T[ Γ1 Ω (1) on Γ 1 × ]0, T[ in QT = Ω × ]0, T[ on QT (2) on ∑ T = Γ × ]0, T[ in Ω where g thanks to the result of Lions and Magenes (1968) and v0 satisfy: in Ω Corresponding Author: Chérif Amrouche, Laboratoire de Mathématiques Appliquées CNRS UMR 5142, Université de Pau et des Pays de l'Adour 28 Res. J. Math. Stat., 3(1): 28-33, 2011 g , H3/2, 3/4 (3J), v0 , H1 (S) The open S being smooth, then the problem (6) has a unique solution 2s such that: Furthermore, it is assumed that: Divv0 = 0 in S, v0.n = 0 on ' θ s ∈ I W 1, p ( Ω) (3) 1≤ p < 2 and g.n = 0 on 3J System (5) becomes our energy problem for which we must find solutions, is therefore, written: (4) ⎧ ∂ θ + ( v. ∇)θ ⎪∂t ⎪− a∆θ = −( v. ∇)θ s ⎪ ⎪⎪θ = 0 ⎨ ⎪θ = 0 ⎪∂θ ⎪∂ x = 0 ⎪ ⎪⎩θ ( 0) = θ0 Recall that the system (2) has a unique solution as shown in the results of Theorem (a) (Amrouche et al., 2007; Batchi, 2005; Kaniel and Shinbrot, 1967). Qualitative properties of the problem (2) can be discussed in many books (Ladyzhenskaya, 1969; Temam, 1985). To simplify it boils down to the non dimensional variables when set: 2* = 2-24 / 2p-24 So that the system (1) becomes: ⎧ ∂ θ * + (v. ∇ )θ * − a∆ θ * = 0 in Ω × ]0, T[ ⎪ ∂t ⎪θ * = 0 on Γ 0 × ]0, T[ ⎪⎪ on Γ 2 × ]0, T[ ⎨θ * = 1 ⎪ ∂θ* on Γ 1 × ]0, T[ ⎪ ∂t = 0 ⎪ in Ω ⎪⎩ θ * ( 0) = θ 0* on Γ0 × ]0, T[ on Γ2 × ]0, T[ (7) on Γ1 × ]0, T[ in Ω with 20 = 2* - 2s and to simplify the writing, it was noted 2 instead of 2˜ in (7). Remark (a): Since: θ s ∈ I W 1, p ( Ω) 1≤ p < 2 and v , L2 (0,T; H2 (S)). Then œ 1# p < 2, v.L2s , L2 (0,T; Hp (S)). To solve Problem (7), we will use again the Galerkin method (Girault and Raviart, 1986). Here it defines the space N = {N , H1 (S), N = 0 on '0 U '2}. The following Lemma allows us to obtain a special adjusted basis in which we have the notations (.,.) and ((.,.)) indicate the scalar products in L2(S) and in H01(S), respectively: (5) with 20* = 20-24 / 2p-24 To study problem (5), it boils down to the homogeneous boundary conditions by setting: 2˜ = 2* - 2s (u,v) = IS u(x)v(x)dx, ((u,v)) = IS Lu(x). Lv(x)dx where 2s represents the temperature function at the wall and satisfies: ⎧∆θ s = 0 ⎪θ = 0 ⎪ s ⎨θ = 1 ⎪ s ⎪∂θ* = 0 ⎩ ∂t in QT = Ω × ]0, T[ in Ω on Γ0 on Γ2 with |.| and ||.|| the associated norms. Lemma (a): There is a sequence (Rj)j$1 of N and a sequence of reals (8j) j$1 such that: λ j > 0, lim λ j = +∞ (6) j →∞ œ N , N, ((Rj, N)) = 8j (Rj, N) on Γ1 (Rj, Rk) = *jk, ((Rj, N)) = 8j *jk 29 Res. J. Math. Stat., 3(1): 28-33, 2011 Proof: For f , L2 ((S)) let R , H2(S) the unique solution of: EXISTENCE OF SOLUTION FOR AN AUXILIARY PROBLEM Now let consider the following auxiliary problem: ⎧ ⎪− ∆ψ = f in Ω ⎪ on Γ0 U Γ2 ⎨ψ = 0 ⎪∂ψ on Γ1 ⎪⎩ ∂ t = 0 ⎧ ∂ θ + (v. ∇ )θ − a∆ θ ⎪ ∂t ⎪θ = 0 ⎪ ⎨ ⎪ ∂∂ θx = 0 ⎪ ⎪⎩ θ ( 0) = θ 0 The operator ^: f÷R L2 (S) ÷ H2 (S) ( ) on Γ 0 U Γ 2 × ]0, T[ on Γ 1 × ]0, T[ (8) in Ω where h and 20 are given. is a linear continuous operator from L2(S) into itself, this implies that ^ is compact. Since this operator being selfadjoint, then: Propose to show the following results: Lemma (b): If 20 , H1(S) and 20 = 0 on '0 U '2 and if h , L2(0, T; L2(S)) then the problem (8) has a unique solution satisfying: (Êf1, f2) = (u1, u2) = ∫ Ω Lu1.Lu2 dx = (f1, Êf2) 2 , L2(0, T; L2(S)) 1 L4 (0, T; L1(S)) 2! , L2(0, T; L2(S)) Consequently, L2(S)has an hilbertian basis composed of eigenvectors of such that: ÊRj = :jRj, :j , ú, :j ÷ 0 when j ÷ 4 Proof: C Approximate solutions of (8): Define 2m(t) an approximate solution of the problem (8) by: therefore, have: 2m(t) = 3mj=1 gjm(t) Rj (Ri, Rj) = *ij So that: and for all N , H1 (S) such that N = 0 on '0 U '2, it was then: (ÊRi, N) = :i (Ri, N) i.e., ⎧− ∆ψ = j ⎪ ⎪ ⎨ψ j = 0 ⎪ ∂ψ j ⎪ ∂x = 0 ⎩ = h in Ω × ]0, T[ 1 µj ψ j in Ω on Γ0 U Γ2 (2!m(t), Rj) + " ((2m(t), Rj) + b(v, 2m(t), Rj) = (h, Rj) with (9) ⎧θm ( 0) = θ0m ∈ ψ 1 ,....,ψ m ⎪ ⎪ on H 1 ( Ω) ⎨θ0m → θ0 ⎪θ = 0 on Γ0 ⎪ ⎩ 0 (10) when m ÷ 4 where b is defined as for all u, v, w , H1(S) by: on Γ1 b(u, v, w) = 3mi,j=1 IS ui (Mvj/Mxi) wj dx Note that: C (Rj, Rk) = 8j *jk where 8j = 1/:j Estimate 1: Multiply (9) by gjm(t) and adding on j from 1 to m: 1 d 2 dt and that (Rj)j$1 is an orthogonal basis of the space M. In particular for all N , M there exists Nm , (R1, … , Rm) such that Nm ÷ N in H1(S) when m ÷ 4. 2 θ m (t ) + α θ m (t ) 2 ( = h, θ m ( t ) ) ≤ C h( t ) θ m ( t ) 30 Res. J. Math. Stat., 3(1): 28-33, 2011 so that: 1 d 2 dt θm ( t ) + α θm ( t ) 2 d dt 2 ≤ h( t ) 1 αC 2 2 (11) Integrating (11) from 0 to t, where t , ]0, T] get: θ m (t ) ( 2 + α ∆ θ m (t ) 2 L∞ ( Ω ) ≤ 3 4α v ≤ 3 4α h( t ) + 2 ) 2 + α 4 ∆ θ m (t ) 2 θ m (t ) α 4 2 ∆ θ m (t ) 2 and θm ( t ) ≤ θ + 2 2 1 aC 2 h 2 L2 0,T ; L2 ( Ω ) ( ) d dt ≤ and deduce that: 2m , bounded of L4 (0, T; H2(S)) C L4 (0, T; L2(S)) 1 L2 (0, T; H1(S)) 2 θ m′ (t ) + (12) 2 ( α d 2 dt ( ≤ v( t ) ) ( ) 2 hence + α ∆ θ m (t ) 2 ) ( ∫ ) t 0 ≤ v ≤ 3 4α L∞ ( Ω ) v 2 (13) ) θ m (t ) t 0 t 0 2 + ) ( ) θ m ( t ) θ m′ ( t ) + h(t ) θ m′ ( t ) α 4 θ m (t ) d dt 2 L∞ ( Ω ) 2 θ m (t ) 2 ∫ + C∫ + θ m (t ) ∆ θ m (t ) 2 L∞ ( Ω ) h( t ) 2 2 θ m′ ( s) ds + α θ m (t ) ≤C but ( 3 2α + C h( t ) 2 Integrating between 0 and t, obtain: ≤ b v , θ m ( t ), ∆ θ m ( t ) + h , ∆ θ m ( t ) b v , θ m ( t ), ∆ θ m ( t ) θ m (t ) L∞ ( Ω ) θ m′ (t ) + α ≤ C v( t ) ( + therefore, 2 b v ,θm ( t ) , ∆θm ( t ) = h, ∆θm ( t ) 2 2 ≤ b v(t ), θ m (t ), ∆ θ m (t ) + b h, θ m′ (t ) θm ( t ) + α ∆θm ( t ) + θ m (t ) θ m (t ) Estimate 3: Multiplying (9) by g’jm (t) and adding on j = 1 to m, it was: Estimate 2: Multiply (9) by 8jgjm (t) then adding on j and through the choice of the basis (Rj) set to Lemma (a), it was: 1 d 2 dt 2 L∞ ( Ω ) 2 1 Finally, one has 2m , bounded of 1 d 2 dt v ) 2m , bounded of L2 (0, T; H2(S)) 1 L4 (0, T; H1(S)) 2m , bounded of L (0, T; H (S)) C 3 2α ( + α ∆ θ m (t ) Using the lemma of Gronwall, then integrating between 0 and T, it is deduced that Then, again integrating (11) between 0 and T, it comes: 2 2 θ m (t ) ∆ θ m (t ) v( t ) 2 L∞ ( Ω ) 2 h( t ) ds + α θ 0 2 2 θ m ( s) ds 2 It follows then that: 2 2m , bounded of L2 (0, T; L2(S)) C and as a result, obtain: 31 (14) Passage to the limit: Here use the following compactness results (Temam 1985; Lions, 1969): Res. J. Math. Stat., 3(1): 28-33, 2011 Moreover, Theorem (a): Let B0, B, B1 be three Banach spaces such that: B0 d B d B1, B0 and B1 being reflexive and suppose that the embedding of B0 into B is compact, Let: W = {v , L p0 (0, T; B0), v’ = dv/dt , L p1 M2k/Mt – ")2k = hk – (v.L)2k It is clear, according to Lemma (b) that: (0, T; B1)} with 1 < pi < +4, i = 0,1 i.e., p Then the embedding of w in L 0 (0, T; B) is compact. Now apply this theorem by setting: 2k , L2(0, T; H2(S)) 1 L4(0, T; H1(S)) v.L2k , L2(0, T; L2(S)) in particular: W = {2 , L2(0, T; H2(S)), 2’ = d2/dt , L2(0, T; L2(S))} The embedding w d L2(0, T; H1(S)) being compact, so there is a subsequence 2m, recorded again 2m, as m ÷ 4, the following convergence take place: 2m ÷ 2 in L2(0, T; H2(S)) weakly 2m ÷ 2 in L4(0, T; H1(S)) weakly star 2m ÷ 2 in L2(0, T; H1(S)) strongly 2’m ÷ 2’ in L2(0, T; L2(S)) weakly (19) and 2k , L2(0, T; H1(S)) hk – (v.L) 2k , L2(0, T; Lp(S)) So that by using the results on the heat equation (Lions, 1969) and the convergence of hk towards –v.L2s in L2(0, T; Lp(S)), it deduces that for all 1#p<2: (15) (16) (17) (18) 2k , bounded of L2 (0, T; W2,p(S)) spend to the limit in (9), one obtains: (20) and therefore, œR , M, (2’(t), R) + "((2(t), R)) + b(v, 2(t), R) = (h, R) 2’k , bounded of L2 (0, T; Lp(S)) Moreover, for all t , [0, T] (21) finally pass to the limit to show that the problem of energy (7) admits a unique solution 2 such that: 2m(t) ÷ 2(t) in H1(S) weakly 2 , L2 (0, T; W2,p(S)) 2’ , L2 (0, T; Lp(S)) and in particular: 20m = 2m(0) ÷ 2(0) in H1(S) weakly Then, formulate the: Thus 20 = 2(0). It is clear that eventually through (15)-(18), check the boundary conditions in (8). Theorem (b): Let θ s ∈ I W 1, p ( Ω) 1≤ p < 2 Remark (b): It can be verified easily uniqueness of the solution in a similar manner through the proof to be made for the Navier-Stokes solutions (Batchi, 2005). 20 , H1(S) and v , L2 (0, T; L2(S)). Then Problem (7) has a unique solution 2 satisfying, for all 1 # p < 2: RESULT OF THE ENERGY EQUATION 2 , L2 (0, T; W2,p(S)), 2’ , L2 (0, T; Lp(S)) From remark (a), note that v.L2s , L (0, T; L (S)) for all 1#p<2. There exists hk , D(QT) such that: 2 (22) (23) p From this results, one deduces the: Corollary: Let 20 , H1(S) and v , L2 (0, T; H2(S)) Thenthe problem (1) has a unique solution 2 satisfying, for all 1 # p < 2: hk ÷ -v.L2s in L2(0, T; Lp(S)) Thus there exists for each k an unique: 2 , L2 (0, T; W2,p(S)), 2’ , L2 (0, T; Lp(S)) 2k , L2(0, T; H2(S)) 1 L4(0, T; H1(S)), 2’k , L2(0, T; L2(S)) Remark (c): If v , L4 (0, T; H2(S)), show that: solution of the first equation of (8) with right hand side hk and initial data 2k(0) = 20. 2 , L2 (0, T; W2,p(S)), 2’ , L2 (0, T; Lp(S)) 32 Res. J. Math. Stat., 3(1): 28-33, 2011 Girault, V. and P.A. Raviart, 1986. Finite Element Methods for Navier-Stokes Equations, Springer Series SCM. Kaniel, S. and M. Shinbrot, 1967. A Reproductive Property of the Navier-Stokes Equations. Arch. Rat. Mech. Anal., 24: 363-369. Ladyzhenskaya, O.A., 1969. The Mathematical Theory of Viscous Incompressible Flow. Gordon and Breach, NY. Lions, J.L., 1969. Some Methods to Solve the Nonlinear Boundary Value Problems. Dunod, Paris. Lions, J.L. and E. Magenes, 1968. Nonhomogeneous Boundary Value Problems and Applications. Dunod, Paris. Temam, R., 1985. Navier-Stokes Equations, Theory and Analysis. North-Holland, Amsterdam. ACKNOWLEDGMENT The authors are grateful to the three institutions for assistance of the study. (mention institutions) REFERENCES Amrouche, C., M. Batchi and J. Batina, 2007. Reproductive Srong Solutions of Navier-Stokes Equations with Nonhomogeneous Boundary Conditions. CCSd/HAL, France. Retrieved from: http://hal.archives-ouvertes.fr/docs/00/14/20/28/ PDF/ArticleSolutionReproductive_3eng.pdf. Batchi, M., 2005. Mathematical and numerical study of the thermal phenomena transfer to the non stationnary flows in axisymmetric geometry. Ph.D. Thesis, University of Pau and Pays de l'Adour. 33