MPE Mathematical Problems in Engineering 1563-5147 1024-123X Hindawi Publishing Corporation 181348 10.1155/2013/181348 181348 Research Article An Adaptive Finite Element Method for Stationary Incompressible Thermal Flow Based on Projection Error Estimation Guo Zhili 1 Su Jian 1 Chen Hao 2 Liu Xiaomin 2 Ishak Anuar 1 Center for Computational Geosciences School of Mathematics and Statistics Xi’an Jiaotong University Xi’an 710049 China xjtu.edu.cn 2 School of Energy and Power Engineering Xi’an Jiaotong University Xi’an 710049 China xjtu.edu.cn 2013 5 12 2013 2013 06 06 2013 09 10 2013 2013 Copyright © 2013 Zhili Guo et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

An adaptive finite element method is presented for the stationary incompressible thermal flow problems. A reliable a posteriori error estimator based on a projection operator is proposed and it can be computed easily and implemented in parallel. Finally, three numerical examples are given to illustrate the efficiency of the adaptive finite element method. We also show that the adaptive strategy is effective to detect local singularities in the physical model of square cavity stationary flow in the third example.

1. Introduction

As we all know the stationary incompressible thermal flow occurs in many fields of the industrial sectors and natural world. The mathematical model of this flow can be described by a set of coupled equations, which consist of the conservation of mass, momentum, and energy equations. The stationary incompressible thermal flow problems constitute an important system of equations in atmospheric dynamics and dissipative nonlinear system of equations. The main difficulties for the numerical simulation of the stationary incompressible thermal flow problem include not only the incompressibility and strong nonlinearity, but also the the coupling between the energy equation and the equations governing the fluid motion. And the numerical methods for solving these problems have attracted the attention of many researchers. Many authors have worked hard to study for a great variety of efficient numerical schemes for these equations . In , Si et al. gave a coupled Newton iterative mixed finite element method for stationary conduction convection problems. In , a nonconforming mixed finite element method for the stationary conduction convection problems was shown by Shi and Ren. In , a new finite element variational multiscale (VMS) method based on two local Gauss integrations was proposed and analyzed for the stationary conduction convection problems by Jiang et al. Furthermore, there are also some scholars devoted to the nonstationary incompressible thermal flow problems . In , a reduced mixed finite element formulation based on proper orthogonal decomposition for the nonstationary conduction convection problems is given by Luo and his coworkers. In , Li et al. gave a fully discrete FVM formulation for the nonstationary conduction convection problems and derived the error analysis between the fully discrete FVM solution and the accurate solution.

In the numerical simulation of stationary incompressible thermal flow problems, still a big challenge is how to increase the accuracy of the numerical approximations for the solutions. The overall accuracy of numerical approximations often deteriorates due to local singularities or large variations in small scales. Motivated by this phenomenon, a direct strategy is that the grids near the critical regions are refined adaptively to improve the quality of the approximate solutions. In 1978, Babuska and Rheinboldt developed a mathematical theory for a class of a posteriori error estimator of finite element solutions [12, 13]. Based on a posteriori error estimator, the adaptive finite element methods have been designed to generate optimal or near optimal meshes and compute solutions more exactly to the problems with boundary layers. Many researchers are dedicated to deriving such a posteriori error estimators and have obtained many good results in recent decades . There are a lot of work devoted to the development of the a posteriori analysis . In 1996, Verfürth built a basic foundation for a posteriori error estimation for nonlinear equations , based on which Zhang et al.  derived a residual a posteriori error estimator for finite element approximations of stationary incompressible thermal flow problems, in 2011. In this paper, a posteriori error estimator is presented based on a projection operator. Its actions can use only standard nodal data structures, so it can be computed locally at the element level and implemented in parallel. The global upper bound for the error of the finite element discretization is derived following some given assumptions.

This paper is organized as follows. In Section 2, we introduce the governing equations, the notations and some preliminary results used for the stationary incompressible thermal flow problems. The a posteriori error estimator based on local projection operator is presented in Section 3. In Section 4, three numerical results and the numerical analysis to validate the effectiveness of the adaptive method are laid out. The first two examples are known solution examples, and the last example is a physical model of square cavity stationary flow. Finally, we give a short conclusion in Section 5.

2. Governing Equations and the Functional Setting

In this section, we consider the stationary incompressible thermal flow problems in two-dimensions. The viscous incompressible flow and heat transfer for the fluid satisfy the incompressible Navier-Stokes equations coupled with the energy conservation equation under the Boussinesq hypothesis is as follows.

Find u=(u1,u2),p,T such that (1)-νΔu+(u·)u+p=λjT,xΩ,divu=0,xΩ,-ΔT+λ(u·)T=0,xΩ,u=0,T=T0,xΓ, where Ω is a bounded domain in R2 assumed to have Lipschitz continuous boundary Γ. u=(u1,u2) represents the velocity vector, p the pressure, T the temperature, λ>0 the Groshoff number, j=(0,1) the unit vector, and ν>0 the viscosity.

We introduce the following Hilbert spaces X=H01(Ω)2, V={vX,·v=0}, Y=L02(Ω)={qL2(Ω),Ωq  dx=0}, and W=H1(Ω), W0=H01(Ω). The standard variational formulation of (1) is given by: find (u,p,T)X×Y×W and T|Γ=T0 satisfying (2)νa1(u,v)-d(p,v)+b1(u,u,v)=λ(jT,v)vX,d(q,u)=0qY,a2(T,ψ)+λb2(u,T,ψ)=0ψW0,u=0,T=T0xΓ. Here we used the notations (3)a1(u,v)=(u,v),u,vX,a2(T,ψ)=(T,ψ),TW,ψW0,d(q,v)=(q,divv),vX,qY,b1(u,v,w)=12[Ωi,k=12uivkxiwkdx-i,k=12uiwkxivkdx],u,v,wX,b2(u,T,ψ)=12[Ωi=12uiTxiψ  dx-Ωi=12uiψxiT  dx],uX,TW,ψW0, and (·,·) the inner product in L2(Ω) or in its vector value versions. The norm and seminorm in Hk(Ω)2 are denoted by ·k and |·|k, respectively. H01(Ω) will denote the closure of C0(Ω) with respect to the norm ·1. The space X is equipped with the norm ·0 or its equivalent norm ·1 due to the Poincaré inequality. Throughout this paper, we use the letter C (with or without subscripts) to denote a generic positive constant which may stand for different value at its different occurrences.

For the finite element discretization, let Kh be the regular triangulations of the domain Ω, indexed by a parameter h=maxKKh{hK;hK=diam(K)}. We choose the finite element subspace XhX,YhY,WhW as follows: (4)Xh={vhc(Ω)2;  vh|KP2(K)2,  KKh},Yh={qhc(Ω);qh|KP1(K),  KKh},Wh={ψhc(Ω);  ψh|KP2(K),  KKh}, where Pl(K)  (l=1,2) is the space of piecewise polynomials of degree l on K. We will also need the piecewise constant space (5)R0={lhL2(Ω),  lh|KP0(K),  KKh}. We denote that W0h=WhH01(Ω). It is obvious that the (Xh,Yh) satisfies the discrete LBB condition (6)supvhXhd(vh,qh)vh0β  qh0qhYh.

With the previous notations, the Galerkin finite element discretization of (2) is given by the following: find (uh,ph,Th)Xh×Yh×Wh such that (7)νa1(uh,vh)-d(ph,vh)+b1(uh,uh,vh)=λ(jTh,vh),vhXh,d(qh,uh)=0,qhYh,a2(Th,ψh)+λb2(uh,Th,ψh)=0,ψhW0h.

For problem (1) the following assumptions and results are recalled (see [8, 22, 23]).

There exists a constant C0 which only depends on Ω such that

u0C0u0,u0,4C0u0,uH01(Ω)2,

u0,4C0u1,uH1(Ω)2,

u0,421/2u01/2u01/2,uH01(Ω)2.

Assuming ΩCk,α(k0,α0), then for T0Ck,α(Ω) and a sufficient small ε, there exists an extension in C0k,α(R2) (denoted T0 also), such that T0k,qε,k0,1q, where ε is an arbitrary small positive constant number.

b1(·,·,·) and b2(·,·,·) have the following properties:

for allu,v, wX (or T,ψH01(Ω)), there holds that (8)b1(u,v,v)=0,b1(u,v,w)=-b1(u,w,v),b2(u,T,T)=0,b2(u,T,ψ)=-b2(u,ψ,T),

for all uX, vH1(Ω)2, wX, (or T H1(Ω), and ψH01(Ω)), there holds that (9)|b1(u,v,w)|N1u0v0w0,|b2(u,T,ψ)|N2u0T0ψ0,

where(10)N1=supu,v,w|b1(u,v,w)|(u0v0w0),N2=supu,T,ψ|b2(u,T,ψ)|(u0T0ψ0).

Theorem 1 (see [<xref ref-type="bibr" rid="B9">8</xref>]).

Under the assumption of (A1)~(A3), and letting A=2ν-1λ(3C0+1)T01, B=2T00+2(C0λ)-1A, assuming there exist 0<σ1,σ21 such that ν-1NA1-σ1, σ1-1ν-1C02λ2BN-1-σ2, then there exists a unique weak solution (u,p,T)X×Y×W, for problem (1) and (11)u0A,T0B.

Theorem 2 (see [<xref ref-type="bibr" rid="B9">8</xref>]).

Under the assumption of (A1)~(A3) and 64C02Nλε/3ν2<1, 16C02λ2N-ε/3ν<1, T00ε/4, T00C0ε/4, then (7) has a unique solution (uh,ph,Th)Xh×Yh×Wh such that T|Ω=T0 and (12)uh08C02λ3νε,Th0ε.

3. A Projection Error Estimation for Stationary Incompressible Thermal Flow Problems

Before deriving the projection estimator, we define the orthogonal projection operator  Π:L2(Ω)R0 acting on the pressure which satisfies the following properties: (13)((I-Π)g,lh)=0,gL2(Ω),lhR0,Πg0Cg0,gL2(Ω),(I-Π)g0Chg1,gH1(Ω). Here, I is the identity operator. In the following discussion, operator Π2×2, Π2×1, which acts on the velocity deformation tensor and the gradient of the temperature Th, are also denoted by Π for simplicity.

Now, based on the residual between the gradient of the finite element solutions velocity component uh; pressure component ph; temperature component Th; and their projections Πuh, Πph, and ΠTh; our projection estimator can be constructed locally as follows: (14)ηΠ,K:=(I-Π)uh0,K+(I-Π)ph0,K+(I-Π)Th0,K.

Then the global error estimator is given by (15)η:=(KKhηΠ,K2)1/2.

Remark 3.

Based on orthogonal projection properties of operator Π, the computation of I-Π can be done at the element level using only standard nodal data structures.

Remark 4.

Our local projection error estimator can be computed more precisely and explicitly based on two local Gauss integrations technique presented in [25, 26]. We give the detailed form of the I-Π operator as folows: (16)(I-Π)qh0,K={K,j(j2)(qh)2dx-K,1(qh)2dx}1/2,qhYh. HereK,lg(x)dx,l=1,j denotes a quadrature formula.

Before giving the global upper bound, we recall a lemma in .

Lemma 5.

There exists a positive constant C such that (17)Chph0(I-Π)ph0phYh.

Remark 6.

This lemma was given in , the proof of this lemma will be rewritten here.

Proof.

We note that ph is continuous and Πph is a constant on each element K from the definition of space Yh and the projection operator Π, so it is obvious that (Πph)|K=0.

Additionally, under some mild assumption on Kh, the following inverse inequalities holds . (18)qh0CIh-1qh0,qhYh. Spaces concerning with vector-valued functions will hold too.

As a result, using the previous inverse inequality (18), (19)h2ph02=KKhh2ph0,K2=KKhh2(I-Π)ph0,K2KKhCI2(I-Π)ph0,K2C(I-Π)ph02.

Then, this completes the proof of Lemma 5.

For uhR12×2, ThR12×1, and spaces consisting of vector-valued functions will also hold for (20)Chuh1(I-Π)uh0,uhXh,ChTh1(I-Π)Th0,ThWh.

Theorem 7.

There is a constant C depending on the smallest angle in the triangulation Kh and the domain Ω, such that, we have the following global upper bounds: (21)(u-uh)0+p-ph0+(T-Th)0Cη, where C is independent of h.

Proof.

We assume that there are two positive constants Cm,Cn, such that (22)Cmp1,ph1,u2,uh1,T2,Th1Cn we can deduce the following results with the discussion of Lemma 5: (23)(u-uh)0+p-ph0+(T-Th)0Chu2+Chp1+ChT2CCnCmhuh1+CCnCmhph1+CCnCmhTh1C{(I-Π)uh0+(I-Π)ph0+(I-Π)Th0}Cη.

Then, the theorem is proved.

4. Numerical Experiments

In this section, we present three examples to illustrate the effectiveness of the adaptive method. In all experiments, the nonlinear systems are solved by the Newton iteration  and implemented in the two-dimensional framework using a public domain finite element software . The first two examples are known solution problems. The last example deals with “lid driven cavity,” which is a benchmark problem for testing numerical schemes.

We make use of the Newton iteration scheme presented in  as follows. Given (uhn-1,Thn-1), find (uhn,phn,Thn)Xh×Yh×Wh and Thn|Γ=T0 such that (24)νa1(uhn,vh)-d(phn,vh)+d(qh,uhn)+b1(uhn,uhn-1,vh)+b1(uhn-1,uhn,vh)=b1(uhn-1,uhn-1,vh)+λ(jThn-1,vh),vhXh,qhYh,a2(Thn,ψh)+λb2(uhn,Thn,ψh)=0,ψhW0h.

For the sake of clarity, the main idea is offered about the refinement strategy presented in  for the algorithm of the discrete problem (7). From the beginning of original triangulation Kh0, a polygonal approximation Ω and a series of refined triangulations Khj were constructed as below. Given Khj, first of all, we compute the error estimator ηΠ,Kj after computing the solution (uj,pj,Tj) from the problem (7); then the new mesh size is given by the following formulation: (25)hj+1=hjfj(ηΠ,Kj), here hj is the previous “meshsize” field, and fj is a user function defined by (26)fj=min(max(ηΠ,Kj(cη-j),1.0000),3), where η-j means the mean of ηΠ,Kj and c is an user coefficient generally close to one and the numbers 1.0000,3 also can be changed by the user according to the requirement.

For simplicity, we choose c=0.8 in all later numerical tests. Certainly, we can also consider other refinement strategy, the one in  for example, whereas there is no much differences between them.

The computation adaptive strategy here is to choose a tolerance η*, start from the original triangulation Kh0, and then compute the η by the following two steps.

Step 1.

If ηη*, stop the calculation, then we obtain the final solution; otherwise, go to the Step 2.

Step 2.

Compute the ηΠ,K and their mean value η-, and generate a new mesh size h by the above strategy and then solve the problem (7) and recompute η based on this new triangulation. Go back to Step 1.

For convenience of presentation, we introduce the following notation:

N:= number of elements in Khj,

Eff:=η/err the effective index, that is, the ratio between the error estimator and the true error. Here, η2:=KKhηΠ,K2, err:={e1(u)2+e0(p)2+e1(T)2}1/2,

e1(u):=(u-uh)0, e0(p):=p-ph0, e1(T):=(T-Th)0.

Example 1.

For this problem the analytical solutions were taken to be (27)u1(x,y)=10x2(x-1)2y(y-1)(2y-1),u2(x,y)=-10x(x-1)(2x-1)y2(y-1)2,p=10(2x-1)(2y-1),T=[cosh(a)-cosh(ay)][cosh(a)-1], with the chosen function added to the right-hand side of (1). The problem is solved on the unite square. From Figure 1, we can easily find that the temperature field has large variation near the boundary y=1, when a=50. So we expect grid points are clustered in the region of rapid variation of the temperature field to improve accuracy. In terms of the meshes refined on the left of Figure 2, we notice that the adaptive strategy creates a lot of triangles in the area near the boundary of y=1. We also give the numerical solution for temperature field on the right of Figure 2. It shows very good agreement with our predictions.

Tables 14 present the numerical solutions based on uniform meshes and the adaptive procedure. As shown in Tables 1 and 2, we can obviously observe that the adaptive strategy almost gets the same good results compared with the uniform one when a=10 in which case the temperature field shows smoothly and flat. But uniform refinements do not achieve good approximation when setting a=50, where the temperature field has a severe variation in small scales, yet the adaptive strategy still obtains much better approximation solution. For example, when the global total error decreases to 0.10297, the uniform meshes needed increase to 8712, while only 1049 triangles are used to get a smaller global total error in the adaptive procedure by comparison of Tables 3 and 4.

All in all, our projection a posteriori error estimator can assess the domain of rapid solution variation and refine them to improve the global accuracy with less meshes.

Uniform meshes for ν=1.0; λ=1.0; a=10.

N e 0 (p) e 1 (u) e 1 (T) err η
128 0.0403662 0.0127483 0.118008 0.125371 0.777592
288 0.0179335 0.00576264 0.0553006 0.0584207 0.532955
512 0.0100866 0.00326292 0.0317188 0.0334435 0.403894
1152 0.00448271 0.00145755 0.0143002 0.015057 0.27133
2048 0.00252149 0.000821408 0.00808489 0.00850871 0.204051
2592 0.00199229 0.000649351 0.0063969 0.00673136 0.181512

A series of adapted meshes for ν=1.0; λ=1.0; a=10.

N e 0 (p) e 1 (u) e 1 (T) err η
232 0.0202677 0.00817272 0.0752798 0.0783876 0.589817
496 0.0157563 0.00659604 0.012231 0.0210087 0.331746
906 0.00893535 0.00428475 0.00900528 0.0133901 0.237672
1628 0.0077005 0.00420461 0.00351317 0.00945086 0.176478
2987 0.003767 0.00197233 0.00270253 0.00503825 0.131168
5714 0.00193568 0.000917294 0.00104072 0.00238147 0.0938836

Uniform meshes for ν=1.0,  λ=1.0,  a=50.

N e 0 (p) e 1 (u) e 1 (T) err η
128 0.0404411 0.0127525 2.11996 2.12038 2.84188
288 0.0179432 0.00576368 1.55127 1.55139 2.70273
512 0.0100887 0.00326323 1.12379 1.12384 2.46931
1152 0.00448294 0.00145758 0.631091 0.631109 1.99439
1800 0.00286897 0.000934275 0.437859 0.43787 1.7078
2592 0.00199231 0.000649354 0.31882 0.318827 1.48158
4608 0.00112066 0.000365579 0.188543 0.188546 1.15963
7200 0.000717221 0.000234067 0.123639 0.123641 0.947138
8712 0.000592744 0.000193468 0.102968 0.10297 0.866675

A series of adapted meshes for ν=1.0,  λ=1.0,  a=50.

N e 0 (p) e 1 (u) e 1 (T) err η
232 0.0202895 0.00817464 1.72925 1.72939 2.72944
445 0.0182222 0.007264 0.388509 0.389004 1.63213
1049 0.0182522 0.00735227 0.058678 0.0618895 0.693979
2154 0.0129152 0.00591016 0.0356371 0.0383632 0.449907
3745 0.00694219 0.00345662 0.0187145 0.0202577 0.340617
6457 0.00600729 0.00300182 0.0123216 0.0140328 0.259971

Exact solution for temperature. a=10 (a) and a=50 (b).

Final adapted mesh (a) and the numerical solution for temperature (b) for a=50.

Example 2.

We choose the known solutions as follows, Ω=[0,1]×[0,1], and the chosen functions are added to the right-hand side of (1) such that the exact solution of the problem is given by (28)T(x,y)=u1(x,y)+u2(x,y),u(x,y)=(u1(x,y),u2(x,y)),u1(x,y)=(1-cos(2π(er1x-1)er1-1))×sin(2π(er2y-1)er2-1)r22πer2y(er2-1),u2(x,y)=-sin(2π(er1x-1)er1-1)×(1-cos(2π(er2y-1)er2-1))r12πer1x(er1-1),p(x,y)=r1r2sin(2π(er1x-1)er1-1)×sin(2π(er2y-1)er2-1)er1xer2y(er1-1)(er2-1), where r1 and r2 are two strictly positive real parameters. The velocity field of this solution is similar to a counter clockwise vortex in a unite box. we can move the center of this vortex that has coordinates x0=(1/r1)log((er1+1)/2) and y0=(1/r2)log((er2+1)/2) along with the change of the parameters r1 and r2. The center goes rapidly towards the right-hand vertical side when increasing r1, in the same way it approaches the top edge when increasing r2. We can verify clearly this phenomenon from the right of Figures 3 and 4. We can also observe that the pressure has large variation in small scale near the right-hand vertical side and the top edge from the left of Figures 3 and 4. So we expect that the adaptive strategy can be sensitive to detect these critical regions and refine them to avoid the numerical oscillation.

Figure 5 presents a sequence of adaptive meshes for r1=8.5 and r2=0.1. The mesh refinement appears near the right-hand vertical side after a series of adaptive iterations. Also the adaptive mesh refinement for r1=3.5 and r2=9.1 in Figure 6 appears near the corner of the right-top, where the solutions have high gradients. These results are consistent with our analysis of problem's requirement. Finally, we give the numerical solutions after a series of adapted iterations when r1=8.5,r2=0.1, and r1=3.5,r2=9.1 in Figures 7 and 8.

To show the effectiveness of our adaptive method and the adaptive procedures based on the residual posteriori error estimator [21, 30, 31], in which the detailed formulation of the local error estimator is presented.

We present the numerical results for Example 2 in Tables 510. Tables 5 and 6 show the results on uniform meshes and adaptive meshes when r1=8.5 and r2=0.1. The results on adaptive meshes based on residual a posteriori error estimator are presented in Table 7. Also, we report the same numerical results in Tables 810, when r1=3.5 and r2=9.1. From Tables 510 we notice that the effective index Eff confirms the reliability and efficiency of posteriori error indicator η.

Comparing Tables 5 and 6, we observe that the global error of the adaptive procedures decreases much faster than those obtained by the quasiuniform ones. For example, when the error is around 0.18, we need 11858 triangles by uniform procedures, while only 2074 triangles are needed by adaptive ones when r1=8.5 and r2=0.1. For another example, when the error decreases to 0.13428 with 2168 meshes by adaptive procedures; however, the global error by uniform ones only drops to 0.173739 even if the meshes required increases to 12800 from the comparison of Tables 8 and 9, when r1=3.5andr2=9.1. Concerning the meshes, we find that our method cost less memory space than common Galerkin finite element method. In fact, it can nearly save four fifths the memory space of the computer to get almost the same precision.

Uniform meshes for ν=1.0,  λ=1.0,  r1=8.5,  r2=0.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
288 0.528492 5.54052 5.48443 7.81381 10.6499 1.36296 1.688
512 0.388401 2.95819 2.90888 4.16693 10.8996 2.61573 2.844
1152 0.177667 0.990161 0.954468 1.38672 8.21263 5.92234 6.594
2048 0.0908647 0.573599 0.553251 0.802096 6.13502 7.64874 11.688
4608 0.037012 0.304226 0.296276 0.426266 4.04822 9.49692 28.063
8192 0.0202685 0.184336 0.180081 0.258495 3.03164 11.728 50.922
10368 0.0159031 0.148582 0.145263 0.208401 2.69474 12.9306 83.172
11858 0.0138586 0.131136 0.128251 0.183948 2.51991 13.699 79.578
12800 0.0128167 0.122067 0.119403 0.171236 2.42552 14.1648 89.109

A successive sequence of adapted meshes for ν=1.0,  λ=1.0,  r1=8.5,  r2=0.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
238 0.651433 6.32201 6.31274 8.95784 10.1411 1.13209 1.422
427 0.161466 0.588001 0.577591 0.839898 6.03079 7.18038 2.531
906 0.0891423 0.217318 0.20884 0.314305 2.37486 7.55591 5.375
2074 0.0376693 0.13206 0.126842 0.186943 1.45756 7.79682 12.39
3808 0.0143992 0.0623338 0.0619315 0.0890413 1.04981 11.7901 23.109
6994 0.0115494 0.0437032 0.0428615 0.0622934 0.761153 12.2188 44.047

A successive sequence of adapted meshes based on a residual error estimation for ν=1.0,  λ=1.0,  r1=8.5,  r2=0.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
238 0.651433 6.32201 6.31274 8.95784 74.4933 8.316 1.609
286 0.167818 0.608207 0.594685 0.867023 8.75867 10.102 1.672
669 0.0785628 0.279946 0.274727 0.400021 4.11085 10.2766 3.89
1212 0.0425637 0.238712 0.237269 0.339252 3.31492 9.77129 7.156
2309 0.041893 0.14484 0.138815 0.204947 2.12799 10.3831 13.672
2962 0.0186839 0.0952816 0.0945824 0.135549 1.34394 9.91483 17.984
5533 0.0194194 0.0628666 0.0620324 0.0904286 0.891394 9.85743 33.891
7468 0.00787077 0.045213 0.0443621 0.0638291 0.618652 9.69232 46.922
12045 0.00932973 0.0325655 0.0319025 0.0465331 0.450524 9.68181 77.109
15180 0.00419643 0.022182 0.0221098 0.0315989 0.306211 9.69056 99.094

Uniform meshes for ν=1.0,  λ=1.0,  r1=3.5, and r2=9.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
288 0.970876 5.40546 5.24747 7.59589 8.74174 1.15085 1.641
512 0.663034 3.11391 2.97829 4.35961 9.4165 2.15994 2.875
1152 0.329059 1.10894 1.00909 1.53502 7.4726 4.86808 6.641
2048 0.169734 0.609773 0.548335 0.837439 5.6408 6.73578 11.797
4608 0.0672255 0.313963 0.289187 0.432113 3.72585 8.62239 28.218
8192 0.0364009 0.189996 0.176725 0.262021 2.79027 10.649 51.594
10368 0.0284943 0.153168 0.142818 0.211351 2.48038 11.7358 70.953
11858 0.0248071 0.1352 0.126205 0.186607 2.31957 12.4302 82.297
12800 0.0229315 0.12586 0.117553 0.173739 2.23276 12.8512 86.297

A successive sequence of adaptive meshes for ν=1.0,  λ=1.0,  r1=3.5, and r2=9.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
238 1.06895 6.04125 5.92896 8.53182 8.25774 0.967876 1.5
417 0.132011 0.558502 0.539777 0.787851 5.4622 6.93303 2.453
912 0.0502047 0.167527 0.153125 0.23245 2.01927 8.68689 5.469
2168 0.0301542 0.0959719 0.0889449 0.13428 1.12367 8.3681 13.016
4158 0.0137853 0.0438102 0.0398118 0.0607812 0.783331 12.8877 25.594
7906 0.00823037 0.0287469 0.0265661 0.0399986 0.556596 13.9154 50.734

A successive sequence of adaptive meshes based on a residual error estimation for ν=1.0,  λ=1.0, and r1=3.5, r2=9.1.

N e 0 ( p ) e 1 ( u ) e 1 ( T ) err η Eff CPU
238 1.06895 6.04125 5.92896 8.53182 75.8374 8.8887 1.375
253 0.138697 0.552635 0.537633 0.783385 8.51898 10.8746 1.469
568 0.0835765 0.245759 0.224761 0.343365 3.69854 10.7714 3.234
1045 0.0421669 0.180767 0.17622 0.255945 2.59402 10.1351 6.016
2022 0.0471101 0.135906 0.124863 0.190475 1.9947 10.4723 11.89
2441 0.0217445 0.0960252 0.0947848 0.136667 1.34754 9.86003 14.437
4011 0.0222083 0.0678677 0.0615177 0.0942532 0.946418 10.0412 24.032
5523 0.0109319 0.0484351 0.0486974 0.0695478 0.672826 9.6743 33.859
8373 0.01082 0.0322648 0.0297083 0.0451738 0.45166 9.99828 52.172
11033 0.0052382 0.0226784 0.0225007 0.0323733 0.317636 9.81166 69.656

Exact solution for r1=8.5  and  r2=0.1; pressure (a) and velocity field (b).

Exact solution for r1=3.5  and  r2=9.1; pressure (a) and velocity field (b).

The successive adaptive refined meshes for r1=8.5 and r2=0.1.

The successive adaptive refined meshes for r1=3.5 and r2=9.1.

Numerical solution for r1=8.5 and r2=0.1; pressure (a) and velocity field (b).

Numerical solution for r1=3.5 and r2=9.1; pressure (a) and velocity field (b).

Comparing Tables 6 and 7, we notice that we need 6994 triangles by our adaptive method less than 7468 triangles based on posteriori error estimator when the global error is about 0.062 with r1=8.5,r2=0.1. We also notice that our adaptive iterations needed are less such that we can save more CPU time to obtain nearly the same accuracy. This means that our adaptive method are more sensitive to those critical regions and refine them more efficiently than the adaptive procedures based on the residual a posteriori error estimator. The numerical tests in Tables 9 and 10 also verify this conclusion with the global error of about 0.06 when r1=3.5 and r2=9.1.

Combining all discussions above, we can derive a conclusion that these results suffice to show that our adaptive strategy obtains much better approximation with less meshes and CPU time. That is to say we can save lots of work by our adaptive procedures.

Example 3 (Square cavity stationary thermal flow).

The last example is a physical model of square cavity stationary flow, which is a popular benchmark problem for testing numerical schemes. The side length of the square cavity and the boundary conditions are given in Figure 9. From Figure 9, we can see that T=0 on left and lower boundaries, T/n=0 on upper boundary, and T=4y(1-y) on right boundary of the cavity. In this example, we set λ=1.0 and ν=1.0.

The left part of Figure 10 is the initial mesh. Then the successive adaptive refined meshes are generated on the basis of a posteriori error estimator (14). The first and third adaptive meshes are presented in the right of Figure 10 and the left of the Figure 11, respectively. From these adaptively generated meshes, we observe that the adaptive strategy is capable of recognizing the singularities and the regions with high gradients of the solutions. In addition, we show the numerical solution of (uh,ph,Th) after three levels of adaptive meshes refinement. The streamline of velocity numerical solutions appears on the right of Figure 11. The distributions of the pressure and temperature are provided on the left and right of Figure 12, respectively.

Physical model of the cavity flows.

Initial mesh (a). The first adaptation mesh (b).

The third adaptation mesh (a). The streamline of velocity numerical solutions after three levels of adaptation mesh refinement (b).

Numerical isobar of pressure solution (a) and numerical isotherm of temperature solutions (b) after three levels of adaptation mesh refinement.

In general, we cannot know the exact solutions of the stationary incompressible thermal flow problems. We take the numerical solutions using the P2-P1-P2 finite element pairs on the finer mesh 150×150 as the “exact solutions.” In Figure 13, we give the numerical solutions on the final adaptive meshes 3631 and uniform meshes 45000 concerning the vertical midline for the x component of velocity and the horizontal midline for y component of the velocity. We can see that the adaptive finite element method approach the standard Galerkin method on the much fine mesh.

Vertical midline for x component of velocity (a) and horizontal midline for y component of velocity (b).

5. Conclusion

In this paper, we present an adaptive finite element method based on a posteriori error estimator for the stationary incompressible thermal flow problems. This error estimator is constructed by a projection operator and can be calculated explicitly and precisely by the difference of two Gauss integrations technique. The discussions and the numerical tests indicate that the projection error estimator is effective and efficient. Whereas there are still many questions in further analysis, such as the convergence and optimality of the adaptive finite element methods.

Acknowledgments

This work is supported by the National Natural Science Foundation of China (nos. 11001216, 11371288, 11371289, 11272251, and 11271298) and the China Scholarship Council (no. 201206285018).

Si Z. Zhang T. Wang K. A Newton iterative mixed finite element method for stationary conduction-convection problems International Journal of Computational Fluid Dynamics 2010 24 3-4 135 141 10.1080/10618562.2010.495931 MR2682925 ZBL1267.76066 Si Z. He Y. A coupled Newton iterative mixed fixed element method for stationary conduction-convection problems Computing 2010 89 1-2 1 25 10.1007/s00607-010-0093-0 MR2660670 Si Z. He Y. A defect-correction mixed finite element method for stationary conduction-convection problems Mathematical Problems in Engineering 2011 2011 28 370192 10.1155/2011/370192 MR2764986 ZBL1204.76021 Shi D. Ren J. Nonconforming mixed finite element method for the stationary conduction-convection problem International Journal of Numerical Analysis and Modeling 2009 6 2 293 310 MR2574910 ZBL1165.65080 Shi D. Ren J. A least squares Galerkin-Petrov nonconforming mixed finite element method for the stationary conduction-convection problem Nonlinear Analysis: Theory, Methods & Applications A 2010 72 3-4 1653 1667 10.1016/j.na.2009.09.005 MR2577566 ZBL1189.76345 Jiang Y. Mei L. Wei H. Tian W. Ge J. A finite element variational multiscale method based on two local Gauss integrations for stationary conduction-convection problems Mathematical Problems in Engineering 2012 2012 14 747391 10.1155/2012/747391 MR2999536 ZBL1264.76062 Tang L. Q. Tsang T. T. H. Least-squares finite element method for time-dependent incompressible flows with thermal convection International Journal for Numerical Methods in Fluids 1993 17 4 271 289 2-s2.0-0027643079 Luo Z. D. Mixed Finite Element Foundation and Its Application 2006 Beijing, China Science Press (Chinese) Reddy J. N. Gartling D. K. The Finite Element Method Transfer and Fluid Dunamics 2001 2nd Washington, DC, USA CRC Press MR1882785 Luo Z. Xie Z. Chen J. A reduced MFE formulation based on POD for the non-stationary conduction-convection problems Acta Mathematica Scientia B 2011 31 5 1765 1785 10.1016/S0252-9602(11)60360-3 MR2884949 ZBL1249.65199 Li H. Luo Z. Sun P. An J. A finite volume element formulation and error analysis for the non-stationary conduction-convection problem Journal of Mathematical Analysis and Applications 2012 396 2 864 879 10.1016/j.jmaa.2012.07.046 MR2961277 ZBL1248.35163 Babuska I. Rhernboldt W. C. A posteriori error estimates for the finite element method International Journal for Numerical Methods in Engineering 1978 12 1597 1615 Babuška I. Rheinboldt W. C. Error estimates for adaptive finite element computations SIAM Journal on Numerical Analysis 1978 15 4 736 754 MR0483395 10.1137/0715049 ZBL0398.65069 Ervin V. J. Layton W. J. Maubach J. M. Adaptive defect-correction methods for viscous incompressible flow problems SIAM Journal on Numerical Analysis 2000 37 4 1165 1185 10.1137/S0036142997318164 MR1756420 ZBL1049.76038 Zheng H. Hou Y. Shi F. Adaptive variational multiscale methods for incompressible flow based on two local Gauss integrations Journal of Computational Physics 2010 229 19 7030 7041 10.1016/j.jcp.2010.05.038 MR2677765 ZBL05785961 Zheng H. Hou Y. Shi F. A posteriori error estimates of stabilization of low-order mixed finite elements for incompressible flow SIAM Journal on Scientific Computing 2010 32 3 1346 1361 10.1137/090771508 MR2652081 ZBL05880056 Verfürth R. A posteriori error estimators for the Stokes equations Numerische Mathematik 1989 55 3 309 325 10.1007/BF01390056 MR993474 ZBL0674.65092 Verfürth R. A Review of a Posteriori Error Estimation and Adaptive Mesh-Refinement Techniques 1996 Wiley Teubner Zienkiewicz O. C. Zhu J. Z. The superconvergent patch recovery and a posteriori error estimates. I. The recovery technique International Journal for Numerical Methods in Engineering 1992 33 7 1331 1364 10.1002/nme.1620330702 MR1161557 ZBL0769.73084 Zienkiewicz O. C. Zhu J. Z. The superconvergent patch recovery and a posteriori error estimates. II. Error estimates and adaptivity International Journal for Numerical Methods in Engineering 1992 33 7 1365 1382 10.1002/nme.1620330703 MR1161558 ZBL0769.73085 Zhang Y. Hou Y. Zuo H. A posteriori error estimation and adaptive computation of conduction convection problems Applied Mathematical Modelling 2011 35 5 2336 2347 10.1016/j.apm.2010.11.033 MR2763864 ZBL1217.76040 He Y. Li J. Convergence of three iterative methods based on the finite element discretization for the stationary Navier-Stokes equations Computer Methods in Applied Mechanics and Engineering 2009 198 15-16 1351 1359 10.1016/j.cma.2008.12.001 MR2497612 ZBL1227.76031 Temam R. Navier-Stokes Equation: Theory and Numerical Analysis 1984 3rd New York, NY, USA North-Holland Publishing MR769654 Li J. He Y. A stabilized finite element method based on two local Gauss integrations for the Stokes equations Journal of Computational and Applied Mathematics 2008 214 1 58 65 10.1016/j.cam.2007.02.015 MR2391672 ZBL1132.35436 He Y. Li J. A stabilized finite element method based on local polynomial pressure projection for the stationary Navier-Stokes equations Applied Numerical Mathematics 2008 58 10 1503 1514 10.1016/j.apnum.2007.08.005 MR2454961 ZBL1155.35406 Zheng H. Hou Y. Shi F. Song L. A finite element variational multiscale method for incompressible flows based on two local Gauss integrations Journal of Computational Physics 2009 228 16 5961 5977 10.1016/j.jcp.2009.05.006 MR2542923 ZBL1168.76028 Bochev P. B. Dohrmann C. R. Gunzburger M. D. Stabilization of low-order mixed finite elements for the Stokes equations SIAM Journal on Numerical Analysis 2006 44 1 82 101 10.1137/S0036142905444482 MR2217373 ZBL1145.76015 Ciarlet P. G. The Finite Element Method for Elliptic Problems 2002 40 Philadelphia, Pa, USA SIAM SIAM Classics in Applied Mathematics 10.1137/1.9780898719208 MR1930132 Freefem++ version 2. 17. 1, http://www.freefem.org/ Zhang Y.-Z. Hou Y.-R. Wei H.-B. Adaptive mixed least squares Galerkin/Petrov finite element method for stationary conduction convection problems Applied Mathematics and Mechanics. English Edition 2011 32 10 1269 1286 10.1007/s10483-011-1499-6 MR2883639 ZBL1237.76078 Zhang Y. Hou Y. Du C. A posteriori error estimation for a defect correction method applied to conduction convection problems Numerical Methods for Partial Differential Equations 2013 29 2 496 509 10.1002/num.21717 MR3022896 ZBL06143721