Finite Element Methods for the Incompressible Navier-Stokes Equations

Rolf Rannacher

Institute of Applied Mathematics University of Heidelberg INF 293/294, D-69120 Heidelberg, Germany E-Mail: [email protected] URL: http://gaia.iwr.uni-heidelberg.de

Version of August 11, 1999

Abstract These notes are based on lectures given in a Short Course on Theoretical and Numerical Fluid Mechanics in Vancouver, British Columbia, Canada, July 27-28, 1996, and at several other places since then. They provide an introduction to recent developments in the numerical solution of the Navier-Stokes equations by the nite element method. The material is presented in eight sections: 1. Introduction: Computational aspects of laminar ows 2. Models of viscous ow 3. Spatial discretization by nite elements 4. Time discretization and linearization 5. Solution of algebraic systems 6. A review of theoretical analysis 7. Error control and mesh adaptation 8. Extension to weakly compressible ows Theoretical analysis is oered to support the construction of numerical methods, and often computational examples are used to illustrate theoretical results. The variational setting of the nite element Galerkin method provides the theoretical framework. The goal is to guide the development of more ecient and accurate numerical tools for computing viscous ows. A number of open theoretical problems will be formulated, and many references are made to the relevant literature.
The author acknowledges the support by the German Research Association (DFG) through the SFB 359 Reactive Flow, Diusion and Transport at the University of Heidelberg, Im Neuenheimer Feld 294, D-69120 Heidelberg, Germany.

Introduction

In the following sections, we will discuss a computational methodology for simulating viscous incompressible laminar ows. The description of the numerical algorithms will be accompanied by a heoretical analysis so far as it is relevant to understanding the performance of the method. In this sense, these notes are meant as a contribution of Mathematics to CFD (Computational Fluid Dynamics).

Figure 1: Nonstationary ow around CFD for Re = 500 , driven by rotation of

the outer circle and visualized by temperature isolines; from Turek [98].

The established model for viscous Newtonian incompressible ow is the the system of Navier-Stokes equations, t v v + v v + p = f, v = 0, (1)

in some region (0, T ) with appropriate initial and boundary conditions. We concentrate on laminar ows, i.e., on ows with Reynolds number in the range 1 Re 105 , where Re v l/ . The numerical solution of this system involves several typical diculties: Complicated ow structure ne meshes! Re 1 locally rened and anisotropic meshes in boundary layers! Dominant nonlinear eects stability problems! Constraint v = 0 implicit solution! Sensitive quantities solution-adapted meshes! 2

Accurate ow prediction requires the use of large computer power, particularly for the extension from 2D to 3D, from stationary to nonstationary ows, and from qualitative results to quantitatively accurate results. The key goals in the developing tools for computing laminar ows are: fast (nonstationary calculations in minutes or hours), cheap (simulations on workstations), exible (general purpose solver), accurate (adaptive error control). 1.1 Solution method The method of choice in these notes is the Finite Element Method (FEM) for computing the primitive variables v (velocity) and p (pressure). This special Galerkin method is based an a variational formulation of the Navier-Stokes problem in appropriate function spaces, and determines discrete approximations in certain nite dimensional subspaces (trial spaces) consisting of piecewise polynomial functions. By this approach the discretization inherits most of the rich structure of the continuous problem, which, on the one hand provides a high computational exibility and on the other hand facilitates a rigorous mathematical error analysis. These are the main aspects which make the FEM increasingly attractive in CFD. For completeness, we briey comment on the essential features of the main competitors of the FEM: Finite dierence methods (FDM): Approximation of the NavierStokes equations in their strong form by nite dierences: + easy implementation, problems along curved boundaries, dicult stability and convergence analysis, mesh adaptation dicult. Finite volume methods (FVM): Approximation of the Navier-Stokes equations as a system of (cell-wise) conservation equations: + based on physical conservation properties, problems on unstructured meshes, dicult stability and convergence analysis, only heuristic mesh adaptation. Spectral Galerkin methods: Approximation of the Navier-Stokes equations in their variational form by a Galerkin method with highorder polynomial trial functions: + high accuracy, treatment of complex domains dicult, mesh and order (hp)-adaptation dicult. 3

This brief classication must be supercial and is based on personal taste. The details are the subject of much controversial discussion concerning the pros and cons of the various methods and their variants. However, this conict is partially resolved in many cases, as the dierences between the methods, particularly between FEM and FVM, often disappear on general meshes. In fact, some of the FVMs can be interpreted as variants of certain mixed FEMs.

1.2

Examples of computable viscous ows

Below, we give some examples of ows which can be computed by the methods described in these notes. More examples including movies of nonstationary ows can be seen on our homepage: http://gaia.iwr.uni-heidelberg.de/. Some of the computer codes are available for research purposes: FEATFLOW Code (FORTRAN 77) by S. Turek [96], [97]: http://gaia.iwr.uni-heidelberg.de/~featflow/. deal.II Code (C++) by W. Bangerth and G. Kanschat [5]: http://gaia.iwr.uni-heidelberg.de/~deal/. A collection of experimental photographs of such computable ows can be found in Van Dykes book An Album of Fluid Motion [99]. In the following, we present some examples of viscous ows which have been computed by the methodology described in these notes. Most of these results emerged as side products in the course of developing the numerical solvers and testing them for standard benchmark problems. All computations were done on normal work stations. Example 1. Cavity ow: The rst example is stationary and nonstationary ow in a cavity driven by ow along the upper boundary (driven cavity).

Figure 2: Stationary driven cavity ow in 2D for Re = 1, 1000, 9000 (from left to right); for Re > 10000 the ow becomes nonstationary; from Turek [98].

Figure 3: Simulation of nonstationary 3D driven cavity ow for Re = 100 ; from Oswald [66]. Example 2. Vortex shedding: The second example is nonstationary ow around a circular cylinder (von K arm an vortex street)

Figure 4: Von K arm an vortex street; experiment with Re = 105 (left; from Van
Dyke [99]), and 2D computation with Re = 100 (right; from Turek [98]).

Figure 5: 3D simulation of vortex shedding behind a cylinder for Re = 100 , coarse

grid and ow visualized by particle tracing; from Oswald [66].

Example 3. Leapfrogging of vortex rings: Two successive pus of uid are injected through a hole and develop into vortex rings dancing around each other. The second ring travels faster in the induced wake of the rst and slips through it. Then the rst ring slips through the second, and so on.

Figure 6: Leapfrogging of two vortex rings; experiment for Re 1600 (left; from Van Dyke [99]) and 2D computation for Re 500 (right; from [46]). 1.2.1 Extensions beyond standard Navier-Stokes flow

The numerical methodology described in these notes has primarily been developed for computing viscous incompressible Newtonian ows. However, extensions are possible in several directions. These include ows in regions with moving boundaries, as for example pipe ow driven by rotating propellers, and ow of non-Newtonian uids modelled by a simple power-law rule. The extension to certain low-speed compressible ows will be discussed in more detail below in Section 8.

1) Flow regions with moving boundaries

Figure 7: Velocity plot of 2D ow in a box driven by a rotating cross, computed by

a virtual boundary technique; from Turek [97].

2) Flow of a non-Newtonian uid

Figure 8: Computation of the ow of a non-Newtonian uid around a circular

obstacle in a 2D channel (power-law = (1 + |D(v )|)1 ): stationary ow in the Newtonian case (left) and nonstationary ow in the non-Newtonian case (right), both for the same Reynolds number Re = 20; from Turek [98].

3) Low-Mach-number compressible ow

Figure 9: Computation of compressible ow in a chemical ow reactor: ow conguration and mass fraction of excited H2 from Waguet [103] and [10].
( =1)

computed on a locally rened mesh;

Models of viscous ow

The mathematical model for describing viscous (Newtonian) ows is the system of NavierStokes equations, which are the equations of conservation of mass, momentum and energy: = 0, t (v ) + v v [v + = f , t (cp T ) + cp v T [T ] = h . t + [v ] 1 vI ] + ptot 3 (1) (2) (3)

Here, v is the velocity, the density, ptot the (total) pressure, and T the temperature of the uid occupying a two- or three-dimensional region . The parameters > 0 (dynamic viscosity), cp > 0 (heat capacity) and > 0 (heat conductivity) characterize the properties of the uid. The volume force f and the heat source h are explicitly given. Since we only consider lowspeed ows, the inuence of stress and hydrodynamic pressure in the energy equation can be neglected. In temperature-driven ows, h may implicitly depend on the temperature and further quantities describing heat release, as for example by chemical reactions. Such weakly compressible ows will be briey considered at the end of these notes in Section 8. Here, the uid is assumed to be incompressible and the density to be homogeneous, 0 = const. , so that (1) reduces to the incompressibility constraint v = 0 . (4)

In this model, we consider as the primal unknowns the velocity v , the pressure p = ptot , and the temperature T . For most parts of the discussion, the ow is assumed to be isothermal, so that the energy equation decouples from the momentum and continuity equations, and the temperature only enters through the viscosity parameter. The system is closed by imposing appropriate initial and boundary conditions for the ow variables v|t=0 = v 0 , v|rigid = 0 , v|in = v in , (n v + pn)|out = 0 , (5)

and corresponding ones for the temperature, where rigid , in , out are the rigid part, the inow part and the outow part of the boundary , respectively. The role of the natural outow boundary condition on out will be explained in more detail below. In the isothermal case, assuming for simplicity that 0 = 1 , the NavierStokes system can be written in short as t v + v v v + p = f , v = 0 , (6)

with the kinematic viscosity parameter = /0 . In this formulation the domain may be taken two- or three-dimensional according to the particular requirements of the simulation. In our examples, we shall mostly refer to the 8

two-dimensional case. Through a scaling transformation this problem is made dimensionless, with the Reynolds Number Re = UL/ as the characteristic parameter, where U is the reference velocity (e.g., U max |v in | ) and L the characteristic length (e.g., L diam() ), of the problem. As common in the mathematical literature, we assume that U = 1 and L = 1 and consider := 1/Re as a dimensionless parameter characterizing in some sense the complexity of the ow problem. Then, the length of the time interval over which the solution develops its characteristic features is T 1/ , and the relevant scale of its spatial structures is x (width of boundary layers). This has to be kept in mind when the right spatial mesh size h and the time step k is chosen for a numerical simulation which is supposed to resolve all structures of the ow; for a more detailed discussion of the issue of reliable transient ow calculation see [54]. 2.1 Variational formulation

The nite element discretization of the Navier-Stokes problem (6) is based on its variational formulation. To this end, we use the following sub-spaces of the usual Lebesgue function space L2 () of square-integrable functions on : L2 0 () = H 1 () = 1 H0 (; ) = L2 () : (v, 1) = 0 , v L2 (), i v L2 (), 1 i d , v H 1 (), v| = 0 , , uv dx ,

and the corresponding inner products and norms (u, v ) = v = (v, v )1/2 , v
1

v = (v, v )1/2 ,

+ v

2 1/2

These are all spaces of R-valued functions. Spaces of Rd -valued functions v = (v1 , . . . , vn ) are denoted by boldface-type, but no distinction is made 1 d in the notation of norms and inner products; thus H1 0 (; ) = H0 (; ) 2 1/2 has norm v 1 = ( d , etc. All the other notation is self-evident: i=1 vi 1 ) t u = u/t, i u = u/xi , n v = nv, = v etc., where n and are the normal and tangential unit vectors along the boundary . The pressure p in the Navier-Stokes equations is uniquely (possibly up to a constant) determined by the velocity eld v . This follows from the fact that every bounded functional F () on H1 0 (; ) which vanishes on the subspace J1 (; ) = {v H1 0 (; ), v = 0} can be expressed in the form F () = (p, ) for some p L2 () . Further, there holds the stability estimate (inf-sup stability) inf 2 (q, ) H1 0 (;) sup 9 0 > 0 , (7)

q L ()

where L2 () has to be replaced by L2 0 () in the case = . For proofs of these facts, one may consult the rst parts of the books of Ladyshenskaya [58], Temam [88] and Girault/Raviart [29]. Finally, we introduce the notation a(u, v ) := (u, v ) , and the abbreviations H := H1 0 (; ), L := L2 () L := L2 0 () in the case = , n(u, v, w ) := (uv, w ) , b(p, v ) := (p, v ),

where = in rigid . Then, the variational formulation of the Navier-Stokes problem (6), reads as follows: Find functions v (, t) v in + H and p(, t) L , such that v|t=0 = v 0 , and setting := in rigid , (t v, ) + a(v, ) + n(v, v, ) + b(p, v ) = (f, ) H , (v, ) = 0 L . (8) (9)

It is well known that in two space dimensions the pure Dirichlet problem (8), (9), with out = , possesses a unique solution on any time interval [0, T ] , which is also a classical solution if the data of the problem are smooth enough. For small viscosity, i.e., large Reynolds numbers, this solution may be unstable. In three dimensions, the existence of a unique solution is known only for suciently small data, e.g., v 0 1 , or on suciently short intervals of time, 0 t T , with T . 2.2 Regularity of solution We collect some results concerning the regularity of the variational solution of the Navier-Stokes problem which are relevant for the understanding of its numerical approximation. One obtains quantitative regularity bounds from the following sequence of dierential identities
1 d 2 t 1 d 2 t 1 d 2 t 1 d 2 t 1 d 2 t

...

v 2 + v 2 = (f, v ), v 2 + v 2 = (f, v ) + (v v, v ), t v 2 + t v 2 = (t f, t v ) (t v v, t v ), t v 2 + t v 2 = (t f, t v ) + (t v v, t v ) + ... , 2 2 2 2 2 2 t v 2 + t v 2 = (t f, t v ) (t v v, t v ) ... ,

which are easily derived by standard energy arguments; see [40] and [41]. Assuming a bound on the Dirichlet norm of v , sup
t(0,T ]

v (t) M,

(10)

the above estimates together with the usual elliptic regularity results im for 0 < t T , if all the data and are ply that v is smooth on 10

smooth. However, for the purposes of numerical analysis one needs regularity estimates which hold uniformly for t 0 . To get such information from the above equations requires starting values for all the quantities 2 v , v , t v , t v , t v , etc., at t = 0 . However, there is a problem already with t v (0) , as has been demonstrated in [41]. 2.2.1 Compatibility conditions at t = 0 To investigate this phenomenon, let us assume that the solution {v, p} is uniformly smooth as t 0 . Then, applying the divergence operator to the Navier-Stokes equations and letting t 0 implies: (i) in : (t v + v v ) = ( v p) (ii) on : t v + v v = v p t g|t=0 + v 0 v 0 = v 0 p0 , (v 0 v 0 ) = p0 ,

where g is the boundary data, v 0 the initial velocity and p0 := limt0 p(t) the initial pressure. Hence, in the limit t = 0 , we obtain an overdetermined Neumann problem for the initial pressure: p0 = (v 0 v 0 ) in , p0 | = v 0 t g|t=0 v 0 v 0 , including the compatibility condition p0 | = ( v 0 t g|t=0 v 0 v 0 ), (13) (11) (12)

where is the tangent direction along . If this compatibility is violated, then limt0 { 3 v (t) + t v (t) } = ; see [41]. We emphasize that (13) together with (11) is a global condition which in general cannot be veried for given data. Without (13) being satised the maximum degree of regularity is right in the middle of H2 and H3 ; see [78]. In view of the foregoing discussion, the natural regularity assumption for the nonstationary Navier-Stokes equations (without additional compatibility condition) is v 0 J1 () H2 () sup
t(0,T ]

2 v (t) + t v (t)

< .

(14)

Example: Flow between two concentric spheres (Taylor problem) Let the inner sphere with radius rin be accelerated from rest v 0 = 0 with a constant acceleration , i.e., v|in (n, , )T = rin cos()(0, 0, t)T (in polar coordinates), while at the outer sphere, we set v|out = 0 . Accordingly, the Neumann problem for the initial pressure takes the form p0 = 0 in , 11 n p0 | = 0,

which implies that p0 const . However, this conicts with the compatibility condition (13) which in this case reads p0 |in = t (v 0 n )t=0 = rin cos() = 0 .

In order to describe the natural regularity of the solution {v (t), p(t)} , as t 0 , and as t , in [41] a sequence of time-weighted a priori estimates has been proven under the assumption (10) using the weight functions (t) = min(t, 1) and et , with xed > 0 : (t)2n+m2 and
t n n m t v (t) + m1 t p(t)

K,

(15)

et
0

es (s)2n+m2

n m t v (t)

n + m1 t p(t)

dt K,

(16)

Open problem 2.1: Devise a way to construct for any given initial data v 0 (e.g., tted from experimental data) and any > 0 a smooth initial data v 0 J1 (; ) , such that v 0 v 0 1 , and the resulting solution of the Navier-Stokes equations satises the compatibility condition (13) at t = 0 . 2.3 Outow boundary conditions

for any m 2, n 1 .

Numerical simulation of ow problems usually requires the truncation of a conceptionally unbounded ow region to a bounded computational domain, thereby introducing articial boundaries, along which some kind of boundary conditions are needed. The variational formulation (8), (9) does not contain an explicit reference to any outow boundary condition. Suppose that the solution v v in + H, p L is suciently smooth. Then, integration by parts on the terms (v ) (p, ) =
out

{n v pn} do + ( v + p, ) 12

yields the already mentioned natural condition on the outow boundary n v pn = 0 on out . (17)

This condition has proven to be well suited in modeling (essentially) parallel ows, see, e.g., [46], Turek [93, 97]. It naturally occurs in the variational formulation of the problem if one does not prescribe any boundary condition for the velocity at the outlet suggesting the name do nothing boundary condition. In the following, we present some experiences in choosing the boundary conditions implicitly, through the choice of variational formulations of ow problems used in nite element computations. To x ideas, let us begin by considering a common test problem, that of calculating nonstationary ow past an obstacle (here an inclined ellipse), situated in a rectangular channel.

Figure 10: The eect of the do nothing outow boundary condition shown by pressure isolines for unsteady ow around an inclined ellipse at Re = 500; from [46]. We impose the usual no-slip boundary conditions on the channel walls and on the surface of the ellipse, while a parabolic Poiseuille inow prole is prescribed on the upstream boundary. We denote again by that portion of the boundary on which Dirichlet conditions are imposed. At the downstream boundary S = out , we decide to do nothing and leave the solution and 2 the test space free by choosing H = H1 0 (; ) and L = L () . This results in the free-outow condition (17). The results of computations based on (17) show a truly remarkable transparency of the downstream boundary when it is handled in this way; see Figure 10 where almost no eect of shortening the computational domain is seen. 13

2.3.1

Problems with the do nothing outflow condition

Although, the do nothing outow boundary condition seems to yield very satisfactory results, one should use it with care. For example, if the ow region contains more than one outlet, like in ows through systems of pipes, undesirable eects may occur, since the do nothing condition contains as an additional hidden condition that the mean pressure is zero across the outow boundary. In fact integrating (17) over any component S of the outow boundary (a straight segment) and using the incompressibility constraint v = 0 yields pn do =
S S

n v do =

t v do = v (s2 ) v (s1 ) = 0 .

Here si denote the end points of S at which v (si) = 0 , due to the imposed no-slip condition along . Consequently, the mean pressure over S must be zero: p do = 0 . (18)
S

To illustrate this, let us consider low Reynolds number ow through a junction in a system of pipes, again prescribing a Poiseuille inow upstream. In Figure 11, we show steady streamlines for computations based on the same variational formulations as above, each with the same inow, but with varying lengths of pipe beyond the junction. Obviously, making one leg of the pipe longer signicantly changes the ow pattern. The explanation of this eect is that by the property (18), in Figure 11 the pressure gradient is greater in the shorter of the two outow sections, which explains why there is a greater ow through that section.

Figure 11: The eect of the do nothing outow boundary condition shown by
streamline plots for ow through a bifurcating channel for Re = 20 ; from [46].

2.3.2

Modification of transport or diffusion model

The foregoing example suggests that one might consider formulating problems more generally, e.g., in terms of prescribed pressure drops or prescribed uxes; we refer to [46] for a thorough development of the corresponding variational formulations. Both choices of boundary conditions lead to well posed formulations of the problem. However, the situation is less satisfactory than in the case of pure Dirichlet boundary conditions. Although the variational problem 14

looks well set in this situation, surprisingly there is a problem with its well posedness. The related Dirichlet problem of the Navier-Stokes equations, stationary as well as nonstationary, is well known to possess weak solutions (not necessarily unique or stable) for any Reynolds number. The standard argument for this result is based upon the conservation property (v v, v ) = 0 of the nonlinear term, which is obtained by integration by parts and using v = 0 . In the case of a free boundary this relation is replaced by
1 (nv, v 2 ) , (v v, v ) = 2

= i i ,

(19)

which generally does not allow to bound the energy in the system without a priori knowledge of what is an inow and what is an outow boundary. As a consequence, in [46] the existence of a unique solution could be shown, even in two space dimensions, only for suciently small data. Kracmar/Neustupa [57] have treated the case of general data by formulating the problem as a variational inequality including the energy bound as a constraint. This still leaves the question open whether one can expect existence of solutions for the original formulation with general data. A positive answer is suggested by numerical tests which do not show any unexpected instability with the discrete analogues of the formulation (8), (9) in the case of higher Reynolds numbers. One may suspect that this theoretical diculty can be avoided simply by changing the variational formulation of the problem, i.e., using other variational representations of the transport or diusion terms. It has been suggested to replace in the momentum equation (8) the Dirichlet form (v, ) by (D [v ], D []) , with D [v ] = 1 ( v + j vi )d i,j =1 being the deformation tensor. 2 i j This change has no eect in the case of pure Dirichlet boundary conditions as then the two forms coincide. But in using the do nothing approach this modication leads to the outow boundary condition nD [v ] pn = 0 on out , which may result in a non-physical behavior of the ow. In the case of simple Poiseuille ow the streamlines are bent outward as shown in Figure 12. Another possible modication is to enforce the conservation property on the transport terms. Using the identity ( 1 |v |2) = v (v )T , the transport 2 term can be written in the form
1 |v |2 . v v = v v v (v )T + 2

This leads to a variational formulation in which (v v, ) is replaced by b(v, v, ) := (v v, ) (v, v ), while the term
1 |v |2 2

(20)

is absorbed into the pressure. An alternative form is (21)

1 (v v, ) 2 (v , v ), b(v, v, ) := 1 2

15

which is legitimate because b(v, v, ) = (v v, ) for v J1 () . Notice that in both cases b(w, , ) = 0 for any w . The corresponding natural outow boundary conditions are, for (20): n v pn = 0, |v |2 , and for (21): with the so-called Bernoulli pressure p = p + 1 2
1 n v 2 |nv |2 n pn = 0.

(22)

(23)

Both modications again result in a non-physical behavior across the outow boundary; streamlines bent inward as shown in Figure 12. Hence, for physical reasons, it seems to be necessary to stay with the original formulation (8), (9). For a detailed discussion of the boundary conditions, and for an extensive list of references, we refer the reader to Gresho [34] and Gresho and Sani [35].

Figure 12: The eect of using the deformation tensor formulation (top) or the symmetrized transport formulations (middle) together with the do nothing outow boundary condition compared to the correct Poiseuille ow (bottom); from [46].

Open Problem 2.2: Prove the existence of global smooth solutions (in 2D) for the original variational Navier-Stokes equations with the do nothing outow boundary condition for general data.

16

Spatial discretization by nite elements

In this section, we recall some basics about the spatial discretization of the incompressible NavierStokes equations by the nite element method. The emphasis will be on those types of nite elements which are used in our codes for solving two and three dimensional ow problems, stationary as well as nonstationary. For a general discussion of nite element methods for ow problems, see to Girault/Raviart [29], Pironneau [68], and Gresho/Sani [35]. 3.1 Basics of nite element discretization

We begin by a brief introduction to the basics of nite element discretization of elliptic problems, e.g., the Poisson equation in a bounded domain Rd (d = 2 or 3) with a polyhedral boundary , u = f in . We assume homogeneous Dirichlet and Neumann boundary conditions, u|D = 0, n u|N = 0, (2) (1)

along disjoint components D and N of , where = D N . The starting point is the variational formulation of this problem in the natural 1 solution space H := H0 (D ; ) : Find u H satisfying a(u, ) := (u, ) = (f, ) H. (3) To discretize this problem, we introduce decompositions, named Th , of into (closed) cells K (triangles or quadrilaterals in 2D, and tetrahedra or hexahedra in 3D) such that the usual regularity conditions are satised: = {K Th } . Any two cells K, K only intersect in common faces, edges or vertices. The decomposition Th matches the decomposition = D N . In the following, we will also allow decompositions with hanging nodes in order to ease local mesh renement. To each of the decompositions Th , there corresponds a mesh-size function h = h(x) which is piecewise constant such that h|K =: hK . We set hK := diam(K ) and denote by K the radius of the ball of maximal size contained in K . We will also use the notation h := maxK Th hK . The family of decompositions {Th }h is said to be (uniformly) shape regular, if chK K hK , (4) and (uniformly) quasi-uniform, if
K Th

max hK c min hK ,
K Th

(5)

17

with some constants c independent of h ; see Girault/Raviart [29] or Brenner/Scott [19] for more details of these properties. In the following, we will generally assume shape-regularity (unless something else is said). Quasi-uniformity is usually not required. Examples of admissible meshes are shown below.
K  h @ @ @ @ @ @

@ @ @ @ @ @

K h


Figure 13: Regular nite element meshes (triangular and quadrilateral) On the decompositions Th , we consider nite element spaces Hh H dened by Hh := {vh H, vh|K P (K ), K Th }, where P (K ) are certain spaces of elementary functions on the cells K . In the simplest case, P (K ) are polynomial spaces, P (K ) = Pr (K ) for some degree r 1 . On general quadrilateral or hexahedral cells, we have to work with parametric elements, i.e., the local shape functions are constructed by K between the physical cell K and a using transformations K : K ) . This construction is xed reference unit-cell K by vh|K (K ()) Pr (K necessary in general in order to preserve conformity (i.e., global continuity) of the cell-wise dened functions vh Hh . For example, the use of bilinear shape functions span{1, x1 , x2 , x1 x2 } on a quadrilateral mesh in 2D employs K . We will see more examples of likewise bilinear transformations K : K concrete nite element spaces below. In a nite element discretization, consistency is expressed in terms of local approximation properties of the shape functions used. For example, in the case of a second-order approximation using linear or d-linear shape functions, there holds locally on each cell K : v Ih v
K

+ hK (v Ih v )

2 cI h2 K v

K,

(6)

and on each cell surface K : v Ih v

K

+ hK n (v Ih v )

cI hK

3/2

2 v

K,

(7)

where Ih v Hh is the natural nodal interpolation of a function v H H 2 () , i.e., Ih v coincides with v with respect to certain nodal functionals (e.g., point values at vertices, mean values on edges or faces, etc.). The interpolation constant is usually of size cI 0.1 1 , depending on the shape of the cell K . 18

With the foregoing notation the discrete scheme reads as follows: Find uh Hh satisfying a(uh , h ) = (f, h ) h Hh . (8) Combining the two equations (3) and (8) yields the relation a(u uh , h ) = 0, h Hh , (9)

which means that the error e := u uh is orthogonal to the subspace Hh with respect to the bilinear form a(, ) . This essential feature of the nite element Galerkin scheme immediately implies the best approximation property e = min (u h ) . (10)
h Hh

In virtue of the interpolation estimate (6), we obtain the (global) a priori convergence estimate e c I h 2 u c I c S h f , (11)

provided that the solution is suciently regular, i.e., u H 2 (), satisfying the a priori bound 2 v c S f . (12) In the above model, this is the case if the polygonal domain is convex. In case of reduced regularity of u due to reentrant corners, the order in the estimate is correspondingly reduced. In the case of approximation by higherorder polynomials, r 2 , and higher order of regularity of u , the estimate (11) shows a correspondingly increased power of h. The order of h in the energy-error estimate (11) can be improved by shifting to the L2 -norm. This is done by employing a duality argument (Aubin-Nitsche trick); see, e.g., Brenner/Scott [19]. Let z H be the solution of the auxiliary problem z = e
1

e in ,

z = 0 on ,

(13)

satisfying an a priori bound 2 z cS . Then, there holds e = (e, z ) = a(e, z ) = a(e, z Ih z ) cI cS h e , and we conclude the improved a priori L2 -error estimate
2 2 e c2 I cS h f .

(14)

(15)

In order to convert the problems (8) into a form which is amenable to pracN tical computation, we introduce the nodal basis {1 h , . . . , h } , N = dim Hh , of the space Hh , dened by i h (aj ) = ij , i, j = 1, . . . , N , where aj are the nodal points (e.g., the vertices) of the mesh. Then, setting
N

uh =
i=1

xi i h, 19

problem (8) is equivalent to the linear algebraic system Ax = b , (16)

for the nodal value vector x = (xi )N i=1 . Here, the stiness matrix A and the load vector b are dened by
j A := a(i h , h ) N , i,j =1

b := (f, i h)

N i=1

In the case of variable coecients and force the integrals have to be computed by using integration formulas; in our implementations usually Gaussian formulas are used. For the pure diusion problem the stiness matrix A is symmetric and positive denite. Its condition number behaves like (A) = O (h2) , where the exponent -2 is determined by the order of the dierential operator (it is independent of the spatial dimension and the polynomial degree of the nite elements used). Below, we show a sequence of hexahedral 3D meshes used for computing the pu-pu ow mentioned in the Introduction; observe the successively rened approximation of the curved boundary.

Figure 14: Sequence of successively rened hexahedral meshes for computing the
pu-pu ow in 3D.

3.2

Stokes elements

We consider the stationary Navier-Stokes problem as specied in Section 2. In setting up a nite element model of the Navier-Stokes problem, one starts from the variational formulation of the problem: Find v v in + H and p L , such that a(v, ) + n(v, v, ) + b(p, ) = (f, ) H, b(, v ) = 0 L. (17) (18)

The choice of the function spaces H H1 ()) and L L2 () depends on the specic boundary conditions imposed in the problem to be solved. On a nite element mesh Th on with cell width h , one denes spaces of discrete trial and test functions, Hh H, 20 Lh L.

in The discrete analogues of (19), (20) then read as follows: Find vh vh + Hh and ph Lh , such that

ah (vh , h ) + nh (vh , vh , h ) + bh (ph , h ) = (f, h ) h Hh , bh (h , vh ) = 0 h Lh ,

(19) (20)

in where vh is a suitable approximation of the inow data v in . The notation Hh H indicates that in this discretization the spaces Hh may be nonconforming, i.e., the discrete velocities vh are continuous across the interelement boundaries and zero along the rigid boundaries only in an approximate sense; in this case the discrete forms ah (, ) , bh (, ) , nh (, , ) and the discrete energy norm h are dened in the piecewise sense,

ah (, ) :=
K Th

(, )K , (, )K ,

bh (, ) :=
K Th

(, )K ,
2 K 1/2

nh (, , ) :=
K Th

:=
K Th

In order that (19), (20) is a stable approximation to (17), (18), as h 0 , it is crucial that the spaces HhLh satisfy a compatibility condition, the so-called infsup or Babuska-Brezzi condition, inf sup
wh Hh

qh Lh

bh (qh , wh ) qh wh h

> 0.

(21)

Here, the constant is required to be independent of h . This ensures that the problems (19), (20) possess solutions which are uniquely determined in Hh Lh and stable. Further, for the errors ev := v vh and ep := p ph , there hold a priori estimates of the form ev
h

+ ep ch

2 v + p

(22)

A rigorous convergence analysis of spatial discretization of the Navier-Stokes problem can be found in Girault/Raviart [29] and in [41, 43]. 3.2.1 Examples of Stokes elements

Many stable pairs of nite element spaces {Hh , Lh } have been proposed in the literature (see, e.g., Girault/Raviart [29], Hughes et al. [49] and [77]). Below, two particularly simple examples of quadrilateral elements will be described which have satisfactory approximation properties and are applicable in two as well as in three space dimensions. They can be made robust against mesh degeneration (large aspect ratios) and they admit the application of ecient multigrid solvers. We note that, from the point of view of accuracy, in our context quadrilateral (hexahedral) elements are to be preferred over triangular (tetrahedral) elements because of their superior approximation properties. 21

Both types of elements may be used in the spatial discretization underlying the discussions in the following sections. 1 /P0 Stokes element 1) The nonconforming rotated d-linear Q The rst example is the natural quadrilateral analogue of the well-known triangular nonconforming nite element of Crouzeix/Raviart (see [29]). It was introduced and analyzed in [77] and its two- as well as three-dimensional versions have been implemented in state-of-the-art Navier-Stokes codes (see Turek [93, 96], Schreiber/Turek [83], and Oswald [66]. In two space dimensions, this nonconforming element uses piecewise rotated bi-linear (reference) shape functions for the velocities, spanned by {1, x, y, x2 y 2 } , and piecewise constant pressures. As nodal values one may take the mean values of the velocity vector over the element edges (or, alternatively, its point values at the midpoints of sides) and the mean values of the pressure over the elements. For the precise denition of this element we introduce the set Th of all (d 1)-faces S of the elements K Th . We set
1 2 Q1 (K ) = q T : q span{1, x1 , x2 i xi+1 , i = 1, ..., d} .

The corresponding nite element spaces are Hh := Lh := vh L2 ()d : vh|K Q1 (K )d , K Th , FS (vh|K ) = FS (vh|K ), S K K , FS (vh ) = 0, S ,

qh L : qh|K P0 (K ), K Th , FS (vh ) = |S |1

with the nodal functionals vh do ,

S

FK (ph ) = |K |1
r r

ph dx .
K

d d
1 |K | K

ph dx

d D D D 1 D || D D De

1 |K |

ph dx

v ds h

r r

r r

1 ||

vh ds

Clearly, the spaces Hh are non-conforming, Hh H1 ()d . For the pair {Hh , Lh } the discrete inf-sup stability condition (21) is known to be satised on fairly general meshes; see [77] and [13]. For illustration, we recall from [13] the essential steps of the argument. Proof of the inf-sup stability estimate (21): Using the continuous inf-sup estimate (7), we conclude for an arbitrary ph Lh that 0 ph sup r h |bh (ph , rh )| |bh (ph , )| sup sup . r h rh Hh H H 22 (23)

where rh Hh is an approximation to H satisfying bh (h , rh ) = 0 h Lh , r h c . (24)

These properties are realized for the Q1 /P0 Stokes element by the natural nodal interpolation dened by rh do =
S S

do S T.

Then, the rst relation in (24) is obvious, and the H1 stability follows from r h
2 h

=
K Th

(rh , n rh )K (rh , rh )K (rh , n rh )K (rh , rh )K .

= (, rh )h +

K Th

The argument becomes particularly simple for parallelogram meshes. In this case n rh | const. and rh |K 0 , such that the last sum vanishes. The general case requires a more involved estimation. Now, the desired inf-sup stability estimate follows with the constant = 0 /c . As discussed in [77], the stability and approximation properties of the Q1 /P0 Stokes element depend very sensitively on the degree of deviation of the cells K from parallelogram shape. Stability and convergence deteriorates with increasing cell aspect ratios. This defect can be cured by using a nonparametric version of the element where the reference space Q1 (K ) := q span{1, , , 2 2 } is dened for each element K independently with respect to the coordinate system (, ) spanned by the directions connecting the midpoints of sides of K . This approximation turns out to be robust with respect to the shape of the elements K , and the convergence estimate (22) remains true. Below, we will relax this requirement even further by allowing the elements to be stretched in one or more (in 3D) directions. 1 /P0 Stokes element (see Finally, we mention an important feature of the Q [93]): It possesses a divergence-free nodal-basis, which allows the elimination of the pressure from the problem resulting in a positive denite algebraic system for the velocity unknowns alone. The reduced algebraic system can be solved by specially adapted multigrid methods; see Turek [93]. 2)The conforming d-linear Q1 /Q1 Stokes element with pressure stabilization The second example uses continuous isoparametric d-linear shape functions for both the velocity and the pressure approximations. The nodal values are just the function values of the velocity and the pressure at the vertices of the mesh, making this approximation particularly attractive in three dimensions. With
1 Q1 (K ) = q T : q span{1, xi , xi xj , i, j = 1, . . . , d} ,

23

the corresponding nite element spaces are dened by Hh = Lh =

d d vh H1 0 (; ) : vh|K Q1 (K ) , K Th

qh H 1 () : qh|K Q1 (K ), K Th

with the nodal functionals ( a vertex of the mesh Th ) Fa (vh ) = vh (a) , Fa (ph ) = ph (a) .
r r r r

vh (a), ph (a)

K
t

t D D vh (a), ph (a) D D D D Dt

K
r r r r

This combination of spaces, however, would be unstable, i.e., it would violate the condition (21), if used together with the variational formulation (19), (20). In order to get a stable discretization, it was proposed by Hughes et al. [49], to add certain least squares terms in the continuity equation (20) (pressure stabilization method), b(h , vh ) + ch (h , ph ) = gh (vh ; h ), where h2 (h , ph )K , K T K h gh (vh ; h ) = h2 (h , f + vh vh vh )K . K T K ch (h , ph ) =
h

(25)

The correction terms on the right hand side have the eect that this modication is fully consistent, since the additional terms cancel out if the exact solution {v, p} of problem (17), (18) is inserted. On regular meshes, one obtains a stable and consistent approximation of the Navier-Stokes problem (17), (18), for which a convergence estimate of form (22) holds true. The argument follows a slightly dierent track than that used above for the nonconforming 1 /P0 element; see [13]. Q Proof of the inf-sup stability estimate (21): From the continuous stability estimate (7) we conclude that 0 ph sup r h | ( p h , r h ) | | ( p h , r h ) | sup + sup , r h H H 24 (26)

rh Hh

where rh Hh is an approximation to H , satisfying

K Th 2 h K rh 2 K 1/2

+ r h c .

(27)

The existence of such an approximation can be shown by employing an averaged nodal interpolation. From this, we obtain ph c sup | ( p h , h ) | + c h h2 K p h
2 K 1/2

h Hh

K Th

which yields the desired stability estimate with the constant = 0 /c . It was shown in Hughes et al. [49], and later on in a series of mathematical papers (see, e.g., Brezzi/Pik aranta [21], and the literature cited therein) in the context of a more general analysis of such stabilization methods, that this kind of discretization is numerically stable and of optimal order convergent for many relevant pairs of spaces Hh Lh . The stabilized Q1 /Q1 Stokes element has several important features: With the same number of degrees of freedom it is more accurate than its triangular analogue (and also slightly more accurate than its nonconforming analogue described above). Furthermore, it has a very simple data structure due to the use of the same type of nodal values for velocities and pressure which allows for an ecient vectorization of solution processes. Thanks to the stabilization term in the continuity equation, standard multigrid techniques can be used for solving the algebraic systems with good eciency (see the discussion in Section 5 below). We note that the triangular analogue of this element is closely related (indeed almost algebraically equivalent) to the inf-sup stable MINIelement (see Brezzi/Fortin [20]) which is based on the standard Q1 /Q1 -element and stability is achieved by augmenting the velocity space by local cubic bubble functions. The stabilized Q1 /Q1 -Stokes element has been implemented in several 2D and 3D Navier-Stokes codes (see, e.g., Harig [38], Becker [7], and Braack [17]). However, it was already reported in Harig [38] that the convergence properties of this element sensitively depend on the parameter and may deteriorate on strongly stretched meshes. We will come back to this point below. 3.3 The algebraic problems

The discrete Navier-Stokes problem (19), (20), possibly including pressure stabilization (25), has to be converted into an algebraic system which can be solved on a computer. To this end, we choose appropriate local nodal bases i {i h , i = 1, ..., Nv } of the velocity space Hh , and {h , i = 1, ..., Np } of the 25

pressure space Lh and expand the unknown solution {vh , ph } in the form
Nv Np

vh =

in vh

+
i=1

xi i h

ph =
j =1

xi j h.

We introduce the following matrices:

j A = ah (i h , h ) Nv , i,j =1 j B = bh (i h , h ) Np ,Nv , i,j =1 j C = ch (i h , h ) Nv , i,j =1 Np . j =1 Np , i,j =1

j j k i in i in v N (x) = nh (vh +N k =1 xk h , h , h ) + nh (h , vh , h ) j j in in in b = (f, j h ) a(vh , h ) nh (vh , vh , h ) Nv , j

c(x) = gh (vh ; j h)

Here, A is the stiness matrix, B the gradient matrix with the associated divergence matrix B T ; N () is the (nonlinear) transport matrix and b the load vector into which the nonhomogeneous inow-boundary data have been incorporated. Further, C is the matrix arising from pressure stabilization and c the (nonlinear) correction term on the right-hand side. Occasionally, we will use the abbreviation A() := A + N () . For later use, we also introduce the velocity and pressure mass matrices:
j Mv = (i h , h ) Nv , i,j =1 j Mp = (i h , h ) Np . i,j =1

With this notation the variational problem (19), (20) can equivalently be written in form of an algebraic system for the vectors x RNv and y RNp of expansion coecients: Ax + N (x)x + By = b , B T x + Cy = c(x) . (28) (29)

Notice that for this system has the structure of a saddle-point problem (for C = 0 ) and is generically nonsymmetric. This poses a series of problems in solving it by iterative methods. This point will be addressed in more detail in Section 5, below. 3.4 Anisotropic meshes

In many situations it is necessary to work with (locally) anisotropic meshes, i.e., in some areas of the computational domain the cells are stretched in order to better resolve local solution features. Such anisotropies generically occur when tensor-product meshes are used to resolve boundary layers. In this case the mesh Th is no longer quasi-regular and the discretization may strongly depend on the deterioration of the cells measured in terms of cell aspect ratios. On such meshes three dierent phenomena occur: - The constant cI in the interpolation estimates (6), (7) may blow up. - The constant in the infsup stability estimate (21) may become small. - The conditioning of the algebraic system may deteriorate. 26

Figure 15: A sequence of locally anisotropic tensor-product meshes with aspect ratios h = 10, 100, 1000, for computing the driven-cavity ow.

It is known that for most of the lower-order nite elements (including the elements considered here) the local interpolation estimates remain valid even on highly stretched elements (maximum angle condition versus minimum angle condition). Accordingly, the failure of the considered Stokes elements on stretched meshes is not so much a problem of consistency but rather one of stability. Hence, we will discuss the stability in more detail; the approximation aspects have been systematically analyzed in Apel/Dobrowolski [2], Apel [3], and the literature cited therein. The main technical diculty arises from the dete1 rioration of the inverse inequality for nite elements h K ch K h K on stretched cells. Further, the solution of the resulting algebraic systems, e.g., by multigrid methods, becomes increasingly dicult. For simplicity, we concentrate the following discussion on the special case of cartesian tensorproduct cells as shown in the gures above and below; here the cell aspect ratio is dened by K = hx /hy and the maximum mesh aspect ratio by h := maxK Th K . We consider aspect ratios of size h 1 104 . hy

K hx

As a model problem, we consider the stationary Stokes equations u + p = f, u = 0, in , (30)

with homogeneous Dirichlet boundary conditions u| = 0 , on a bounded polygonal domain R2 . Using the notation introduced above, the nite element formulation of this problem reads as follows: ah (uh , h ) + bh (ph , h ) = (f, v ) h Hh , bh (uh , h ) = 0 h Lh . 27 (31) (32)

1. First, we consider the nonconforming Q1 /P0 Stokes approximation which uses rotated bilinear shape functions for the velocities and piecewise constants for the pressure. Above, we have introduced its non-parametric version where local cell coordinates {K , K } are used for dening the local shape 2 2 functions on each cell as vh|K span{1, K , K , K K } . In this way one obtains a discretization which is robust with respect to deviations of the cell from parallelogram shape. It has been shown in [12] that this non-parametric element can be modied to be also robust with respect to increasing aspect ratio. This modication employs a scaling of the local coordinate system according to the cell aspect ratio K := hx /hy ,
2 2 2 vh|K span{1, K , K K , K K K } .

Figure 16: Pressure and velocity norm isolines for a driven-cavity computation with the standard isotropic Q1 /P0 element (left) compared to the anisotropically scaled version (right); from Becker [7].

Furthermore, the inf-sup stability estimate (21) is preserved on such meshes with a constant independent of the mesh aspect ratio h . To demonstrate that this scaling is actually necessary for the stability of this element, we show in Figure 16 the results of a driven cavity calculation on meshes with h = 32 using the standard isotropic approximation compared to the anisotropically scaled version. The instability caused by the large aspect ratio exhibits spurious pressure peaks and vortices along the boundary. 2. Next, we consider the stabilized Q1 /Q1 Stokes approximation which uses continuous (isoparametric) bilinear shape functions for both the velocity and 28

the pressure. As seen before, this discretization becomes inf-sup stable if the discrete model is augmented by a least-squares term of the form (uh , h ) + c(ph , h ) = correction terms. (33)

On quasi-uniform meshes, we obtain a stable and consistent approximation of the Stokes problem but this approximation sensitively depends on the choice of the form c(, ) and may deteriorate on strongly stretched meshes. Again, the approximation property of the Q1 /Q1 element is not the problem. The interpolation estimates (6) and (7) remain valid also on high-aspect-ratio meshes (as dened above) with constants independent of h . However, the proper design of the stabilization (33) is delicate. We consider the following three dierent choices for the stabilizing bilinear form: c1 (p, q ) = K Th |K |(p, q )K , c2 (p, q ) = K Th h2 c(p, q ) = K (p, q )K , 2 c3 (p, q ) = K Th h2 x (x p, x q )K + hy (y p, y q )K .

The form c1 (, ) is built in analogy to the MINIelement, since condensation of the bubble functions leads directly to the cell-wise scaling factor |K |. We see that c1 (, ) gets smaller with increasing h , an undesirable eect which is avoided by c2 (, ) . Finally, c3 (, ) distinguishes between the dierent coordinate directions which requires the use of a local coordinate system in the denition of the stabilization. By a local inverse estimate for bilinear functions on (arbitrary) rectangles we get the stability relation c3 (ph , ph ) ph 2 , which appears necessary for the stabilization to achieve uniformity with respect to the mesh aspect ratio. This may be seen by writing the discrete system (31), (33) in matrix notation A B B T Ci x y = b c ,

where Ci corresponds to the stabilizing bilinear form ci (, ) . The Schur complement of the main diagonal block A is = Ci B T A1 B . Then, the stability constant in (33) is given by (see [13]): 2 = min (M 1 ) , (34)

where M denotes the mass matrix of the pressure space Lh (piecewise constants in this case). This correspondence can be used in order to experimentally determine the dependence of the stability constant on the various parameters of the discretization, particularly the cell aspect ratio. We may detect by counting the number of cgiterations (preconditioned by the mass matrix M ) needed to invert the Schur complement. The convergence rate of the cgiteration applied to the preconditioned Schur complement M 1 is linked to the condition number = cond(M 1 ) by the following well-known formula 1 1/ . 2 1 + 1/ 29

These test calculations use a sequence of anisotropic grids obtained by onedirectional renements. The results are given in Table 1. Table 1: Number of cgiterations; from Becker [7]. c1 c2 c3 2 8 8 8 4 8 16 32 64 18 39 98 559 18 39 88 193 16 29 31 29 27 128 24

Figure 17: Pressure isolines for a jet ow in a channel calculation with the Q1 /Q1
element using isotropic stabilization (top and middle) compared to the anisotropic stabilization (bottom); from Becker [7].

The interpretation of these observations is as follows: The increase of the stability constant for c1 (, ) stems from the fact, that 2 0 with increasing aspect ratio, whereas the bad behavior of c2 (, ) can be explained by the growth of max () 2 due to fact that we only have c2 (p, q ) p q . We also see that the anisotropic stabilization by c3 (, ) leads to an aspectratio-independent behavior. Similar eects are also observed for the accuracy of the dierent stabilizations; see Becker [7] and [13]. Open Problem 3.1: Prove the approximation property (27) on general meshes with arbitrary aspect ratio h . The special case of tensor-product meshes has been treated in Becker [7] (see also Apel [3].

30

3.5

Treatment of dominant transport

In the case of higher Reynolds numbers (e.g., Re > 1000 for the 2-D driven cavity, and Re > 100 for the ow around an cylinder) the nite element models (19), (20) or (19), (25) may become unstable since they essentially use centraldierences-like discretization of the advective term. This instability most frequently occurs in form of a drastic slow-down or even break-down of the iteration processes for solving the algebraic problems; in the extreme case the possibly existing mathematical solution contains strongly oscillatory components without any physical meaning. In order to avoid these eects some additional numerical damping is required. The use of simple rst-order articial viscosity is not advisable since it introduces too much numerical damping. Below, we describe two approaches used in the context of nite element discretization: i) an adaptive upwinding, and ii) the streamline diusion method. Alternative techniques are the characteristics Galerkin method (for nonstationary ows) and the discontinuous nite element method which require major changes in the discretization and will therefore not be discussed here; for references see Pironneau [67], Morton [63] and Johnson [51]. 3.5.1 Upwinding

In the nite element context upwinding can be dened in a quite natural way; see, e.g., Tobiska/Schieweck [90] and Turek [97], and the literature cited therein. Here, the upwinding eect is accomplished in the evaluating of the advection term through shifting integration points into the upwind direction. This modication leads to system matrices which have certain Mmatrix properties and are therefore amenable to ecient and robust solution techniques. This is widely exploited in the nite element codes described in Schreiber/Turek [83], Turek [97] and Schieweck [82]. Following [97], we briey describe the upwind strategy for the nonconforming rotated bilinear Stokes element. Each quadrilateral K Th is divided into eight barycentric fragments Sij , and for each edge l and mid point ml on l the lumping region Rl is dened by Rl := kl Slk , where l = {k, ml and mk belong to the same element K }. The boundary of the lumping region Rl consists of the edges lk := Slk Skl , i.e., Rl = kl lk . In this way we obtain an edge oriented partition of the mesh domain h = l Rl . m s n @ S Snm @ nk Smn @ Skn @ smm mk s @ Skl S @ ml Slk Slm @ @ s ml
@ @ @ K K @ @ @ R l @ s @ m l @ @ @ @ @ @ s @ @

mk

31

A modication of the nonlinear form n(uh , vh , wh ) is now dened by n h (uh , vh , wh ) :=

K,l

(1 lk (uh )) (vh (mk ) vh (ml )) wh (ml )

uh nlk ds ,

where lk are parameters depending on the local ux direction. Setting x := possible choices are lk := lk := 1, for x 0 0, for x < 0 (simple upwinding), (Samarskij upwinding), 1 uh nlk ds ,

lk

(1/2 + x)/(1 + x), for x 0 1/(2 2x), for x < 0

It can be be shown (see Tobiska/Schieweck [90]) that this upwind scheme is of rst order accurate and, what is most important, the main diagonal blocks of () become M-matrices. This is the the corresponding system matrix A + N key property for its inversion by fast multigrid algorithms. The described upwind discretization can generically be extended to the three dimensional case. An analogous construction is possible for the conforming Q1 /Q1 element with pressure stabilization also in two as well as in three dimensions; see Harig [38]. 3.5.2 Streamline diffusion

The idea of streamline diusion is to introduce articial diusion acting only in the transport direction while maintaining the second-order consistency of the scheme. This can be achieved in various ways, by augmenting the test space by direction-oriented terms resulting in a Petrov-Galerkin method, or by adding certain least-squares terms to the discretization. For the (stationary) Navier-Stokes problem, we propose the following variant written in terms of in pairs {, } H L : Find vh vh + Hh and ph Lh , such that ah (vh , h ) + nh (vh , vh , h ) + bh (ph , h ) + sh ({vh , ph }, {h , h }) = (f, h ) + rh ({vh , ph }, {h , h }) (35) for all {h , h } H Lh , where, with some reference velocity v h , sh ({vh , ph }, {h, h }) =
K Th

K (ph + vh vh , h + v h h )K +(vh , h )K ,

rh ({vh , ph }, {h, h }) =

K Th

K (f + vh , h + v h h )K . 32

The stabilization parameters K are chosen according to K = min h2 K hK . , |v |K (36)

This discretization contains several features. The rst term in the sum K (ph , h )K + (vh vh , v h h )K + (vh , h )K

K Th

stabilizes the pressure-velocity coupling for the conforming Q1 /Q1 Stokes element, the second term stabilizes the transport operator, and the third term enhances mass conservation. The other terms introduced in the stabilization are correction terms which guarantee second-order accuracy for the stabilized scheme. Theoretical analysis shows that this kind of Galerkin stabilization actually leads to an improvement over the standard upwinding scheme, namely an error behavior like O(h3/2 ) for the nite elements described above; see Tobiska/Verf urth [91], and also Braack [17] where the same kind of stabilization has been applied for weakly compressible ows with chemical reactions. For linear convection-diusion problems the streamline diusion method is known to have even O(h2 ) convergence on fairly general meshes; see [107]. Open Problem 3.2: Derive a strategy for choosing the stabilization parameter K in the streamline diusion method on general meshes with arbitrarily large aspect ratio h .

Open Problem 3.3: The streamline diusion method (like the least-squares pressure stabilization) leads to a scheme which lacks local mass conservation. Recently, for convection-diusion problems an alternative approach has been proposed which uses a discontinuous Galerkin approximation on the transport term and combines (higher-order) upwinding features with local mass conservation. The extension of this method to the incompressible Navier-Stokes equations (and its practical realization) has yet to be done.

33

Time discretization and linearization

We now consider the nonstationary Navier-Stokes problem: Find v v in + H and p L , such that v (0) = v 0 and (t v, ) + a(v, ) + n(v, v, ) + b(p, ) = (f, ) , H , b(, v ) = 0 , L . (1) (2)

The choice of the function spaces H H1 ()d and L L2 () depends again on the specic boundary conditions chosen for the problem to be solved; see the discussion in Section 2. In the past, explicit time stepping schemes have been commonly used in nonstationary ow calculations, mainly for simulating the transition to steady state limits. Because of the severe stability problems inherent to this approach (for moderately sized Reynolds numbers) the very small time steps required prohibited the accurate solution of really time dependent ows. In implicit time stepping one distinguishes traditionally between two dierent approaches called the Method of Lines and the Rothe Method. 4.1 The Rothe Method

In the Rothe Method, at rst, the time variable is discretized by one of the common time dierencing schemes; for a general account of such schemes see, e.g., Thom ee [89]. For example, the backward Euler scheme leads to a sequence of Navier-Stokes-type problems of the form:
1 n (v v n1 , ) + a(v n , ) + n(v n , v n , ) + b(pn , ) = (f n , ), kn b(, v n ) = 0 .

(3) (4)

for all {, } H L , where kn = tn tn1 is the time step. Each of these problems is then solved by some spatial discretization method as described in the preceding section. This provides the exibility to vary the spatial discretization, i.e. the mesh or the type of trial functions in the nite element method, during the time stepping process. In the classical Rothe method the time discretization scheme is kept xed and only the size of the time step may change. The question of how to deal with varying spatial discretization within a time-stepping process while maintaining higher-order accuracy and conservation properties is currently subject of intensive research. It is essential to do the mesh-transfer by L2 projection which is costly, particularly in 3D if full remeshing is used in each time step, but is easily manageable if only meshes from a family of hierarchically ordered meshes are used. 4.2 The Method of Lines

The traditional approach to solving time-dependent problems is the Method of Lines. At rst, the spatial variable is discretized, e.g. by a nite element 34

method as described in the preceding section leading to a system of ordinary dierential equations of the form: Mx (t) + Ax(t) + N (x(t))x(t) + By (t) = b(t) , B T x(t) + Cy (t) = c(t), t > 0 , (5) (6)

with the initial value x(0) = x0 . The mass matrix M , the stiness matrix A and the gradient matrix B are as dened above in Section 3. The matrix C and the right-hand side c stem from the pressure stabilization when using the conforming Q1 /Q1 Stokes element. Further, the (nonlinear) matrix N () is thought to contain also all terms arising through the transport stabilization by upwinding or streamline diusion. For abbreviation, we will sometimes use the notation A() := A + N () . For solving this ODE system, one now applies a time dierencing scheme. The most frequently used schemes are the socalled One-Step- Schemes: One-step -scheme: Step tn1 tn (k = time step): [M + kAn ]xn + By n = [M (1 )kAn1 ]xn1 + kbn + (1 )kbn1 B T xn + Cy n = cn , where xn x(tn ) and An := A(xn ) . Special cases are the forward Euler scheme for = 0 (rst-order explicit), the backward Euler scheme for = 1 (rst-order implicit, strongly A-stable), and the most popular Crank-Nicolson scheme for = 1/2 (second-order implicit, A-stable). These properties can be seen by applying the method to the scalar model equation x = x . In this context it is related to a rational approximation of the exponential function of the form
1 1 ( 2 ) 1 R () = = e + O ( 2 )| | 2 + | | 3 , 1 +

| | 1 .

The most robust implicit Euler scheme ( = 1 ) is very dissipative and therefore not suitable for computing really nonstationary ow. In contrast, the Crank-Nicolson scheme has only very little dissipation but occasionally suers from unexpected instabilities caused by the possible occurrence of rough perturbations in the data which are not damped out due to the only weak stability properties of this scheme (not strongly A-stable). This defect can in principle be cured by an adaptive step size selection but this may enforce the use of an unreasonably small time step, thereby increasing the computational costs. For a detailed discussion of this issue see [71]. A good time-stepping scheme of the described type should possess the following properties: A-stability ( local convergence): |R()| 1

Global stability ( global convergence): limRe |R()| 1 O(k ) . 35

 Strong A-stability ( smoothing property): limRe |R()| 1 < 0 . Low dissipation ( energy preservation): |R()| = 1 O(|Im |), for Re 0 . Alternative schemes of higher order are based on the (diagonally) implicit Runge-Kutta formulas or the backward dierencing multi-step formulas, both being well known from the ODE literature. These schemes, however, have not yet found wide applications in ow computations, mainly because of their higher complexity and storage requirements compared with the Crank-Nicolson scheme. Also less theoretical analysis is available for these methods when applied to large sti systems. Some comparison of their stability and approximation properties is made in [72]; see also [65]. However, there is still another method which is an attractive alternative to the Crank-Nicolson method, the so-called Fractional-Step- Scheme originally proposed by Glowinski [30] and Bristeau et al. [22]. Fractional-Step--scheme: (three substeps: tn1 tn1+ tn tn ) (1) (2) (3) [M + kAn1+ ]xn1+ + kBy n1+ = [M kAn1 ]xn1 + kbn1 , B T xn1+ + Cy n1+ = cn1+ , [M + kAn ]xn + kBy n = [M kAn1+ ]xn1+ + kbn , B T xn + Cy n = cn , n [M + kAn ]xn + kBy n = [M kAn ]xn + kbh , T n n n B x + Cy = c .

In the ODE context this scheme reduces to a rational approximation of the exponential function of the form R () = (1 )(1 )2 = e + O(||3 ) , || 1 . (1 + )2 (1 + )

Here = 1 2/2 = 0.292893... , = 1 2 , (1/2, 1] , and = 1 , in order to ensure second-order accuracy, and strong A-stability, limRe |R ()| = < 1.

For the special choice = (12)/(1) = 0.585786... , there holds = which is useful in building the system matrices in the three substeps. This scheme was rst proposed in form of an operator splitting scheme separating the two complications nonlinearity and incompressibility within each cycle tn tn+ tn+1 . However, the Fractional-Step- scheme has also very 36

attractive features as a pure time-stepping method. It is strongly A-stable, for any choice of (1/2, 1] , and therefore possesses the full smoothing property in the case of rough initial data, in contrast to the CrankNicolson scheme (case = 1/2). Furthermore, its amplication factor has modulus |R()| 1 , for approaching the imaginary axis (e.g., |R(0.8i)| = 0.9998... ), which is desirable in computing oscillatory solutions without damping out the amplitude. Finally, it also possesses very good approximation properties, i.e., one cycle of length (2 + )k = k provides the same accuracy as three steps of the Crank-Nicolson scheme with total step length k/3 ; for more details on this comparison see [72] and [65]. We mention some theoretical results on the convergence of these schemes. For the Crank-Nicolson Scheme combined with spatial discretization as described in Section 3, an optimal-order convergence estimate
n vh v (, tn ) = O(h2 + k 2 )

(7)

has been given in [44]. This estimate requires some additional stabilization of the scheme but then holds under realistic assumptions on the data of the problem. A similar result has been shown by M uller [64] for the FractionalStep -Scheme. Due to its stronger damping properties (strong A-stability) this scheme does not require extra stabilization. 4.2.1 Computational tests

Below, we present some results of the computational comparison between the backward Euler scheme, the Crank-Nicolson scheme and the Fractional-Step- scheme. The ow conguration is shown in Figure 18: ow around an inclined plate in the cross-section of a channel at Re = 500. The spatial discretization is by the nonconforming rotated bilinear Stokes element described in Section 3 on a uniformly rened mesh with 13, 000 cells.

Figure 18: Conguration of plate-ow test, coarse mesh and streamline plot. The rst test concerns accuracy. Figure 19 shows that the backward Euler (BE) scheme is not suitable for computing time-periodic ows with acceptable time-step widths, while the Crank-Nicolson (CN) and the Fractional-Step- (FS) scheme show equally satisfactory results. This similar accuracy is further conrmed by comparing a more sensitive error quantity (mean pressure) in 37

Figure 20. Finally, we look at the stability of the schemes. Figure 21 demonstrates the lack of robustness of the Crank-Nicolson scheme combined with linear time-extrapolation in the nonlinearity for larger time steps.

Figure 19: Pressure isolines of the plate-ow test: BE scheme with 3k = 1 (top
left), BE scheme with 3k = 0.1 (top right), CN scheme with 3k = 1 (bottom left), FS scheme with k = 1 (bottom right); from [65].

10

20

30

40

50

60

10

20

30

40

50

60

Figure 20: Mean pressure plots for the plate test with fully implicit treatment of
the nonlinearity; left: CN scheme with 3k = 0.33; right: FS scheme with k = 0.33, both compared to a reference solution (dotted line); from[65].

10

20

30

40

50

60

10

20

30

40

50

60

Figure 21: Mean pressure plots for the plate test with linear time-extrapolation;
left: CN scheme with 3k = 0.11; right: FS scheme with k = 0.11, both compared to a reference solution (dotted line); [65].

38

4.3

Splitting and projection schemes

As already mentioned, the Fractional-Step- scheme was originally introduced as an operator splitting scheme in order to separate the two main diculties in solving problem (1) namely the nonlinearity causing nonsymmetry and the incompressibility constraint causing indeniteness. At that time, handling both complications simultaneously was not feasible. Therefore, the use of operator splitting seemed the only way to compute nonstationary ows. Using the notation from above, the splitting scheme reads as follows (suppressing here the terms stemming from pressure stabilization). Splitting-Fractional-Step- Scheme: (1) [M + kA]xn1+ + kBy n1+ = [M kA]xn1 + kbn1 kN n1 xn1 , B T xn1+ = 0 , [M + kAn ]xn = [M kAn1+ ]xn1+ kBy n + kbn , ..... [M + kA]xn + kBy n = [M kA]xn kN n xn + kbn , B T xn = 0 .

(2) (3)

The rst and last step solve linear Stokes problems treating the nonlinearity explicitly, while in the middle step a nonlinear Burgers-type problem (without incompressibility constraint) is solved. The symmetric form of this scheme follows the ideas from Strang [87], in order to achieve a second-order splitting approximation. The results of M uller [64] suggest that the optimal-order convergence estimate (7) remains true also for this splitting scheme. However, a complete proof under realistic assumptions is still missing. Open Problem 4.1: Prove that the Splitting-Fractional-Step- scheme is 1 actually second order accurate for all choices of the parameter ( 2 , 1] .

In these days, the ecient solution of the nonlinear incompressible NavierStokes equations is standard by the use of new multigrid techniques. Hence, the splitting of nonlinearity and incompressibility is no longer an important issue. One of these new approaches uses the Fractional-Step- scheme in combination with the idea of a projection method due to Chorin [24]; for a survey see Gresho/Sani [35]. Finally, Turek [95] (see also [97]) has designed the Discrete Projection Fractional-Step- scheme as component in his solver for the nonstationary Navier-Stokes problem. Next, we address the problem of how to deal with the incompressibility constraint v = 0 . The traditional approach is to decouple the continuity equation from the momentum equation through an iterative process (again operator splitting). There are various schemes of this kind in the literature referred to, e.g., as quasi-compressibility method, projection method, SIMPLE method, etc. All these methods are based on the same principle 39

idea. The continuity equation v = 0 is supplemented by certain stabilizing terms involving the pressure, e.g., v + p = 0, v p = 0, v + t p = 0, v t p = 0, (8) (9) (10) (11)

n p| = 0, p|t=0 = 0, n p| = 0, p|t=0 = 0,

where the small parameter is usually taken as h , or k , depending on the purpose of the procedure. For example, (8) corresponds to the classical penalty method, and (9) is the simplest form of the least squares pressure stabilization scheme (11) described above, with h2 in both cases. Further, (10) corresponds to the quasi-compressibility method. These approaches are closely related to the classical projection method of Chorin [24]. Since this method used to be particularly attractive for computing nonstationary incompressible ow, we will discuss it in a some detail. For simplicity consider the case of homogeneous Dirichlet boundary conditions, v| = 0 . The projection method reads as follows. For an admissible initial value v 0 , choose a time step k , and solve for n 1 : (i) v n H (implicit Burgers step): k 1 ( v n v n1) vn + v n v n = f n . (ii) v n = P v n J0 () (Projection step): v n = 0, nv|n = 0. (13) (12)

Here, the function space J0 () is obtained through the completion of the space { D (), 0} of solenoidal test functions with respect to the norm of L2 () . This time stepping scheme can be combined with any spatial discretization method, e.g. the nite element methods described in Section 3. The projection step (ii) can equivalently be expressed in the form (ii) vn = v n + k p n , (14)

with some pressure p n H 1 () , which is determined by the properties (ii) pn = k 1 v n , n p n | = 0. (15)

This amounts to a Poisson equation for p n with zero Neumann boundary conditions. It is this non-physical boundary condition, n p n | = 0 , which has caused a lot of controversial discussion about the value of the projection method. Nevertheless, the method has proven to work well for representing the velocity eld in many ow problems of physical interest (see, e.g. Gresho [32] and Gresho/Chan [33]). It is very economical as it requires in each time step 40

only the solution of a (nonlinear) advection-diusion system for v n (of Burgers equation type) and a scalar Neumann problem for p n . Still, it was argued that the pressure p n were a mere ctitious quantity without any physical relevance. It remained the question: How can such a method work at all? A challenging problem for mathematical analysis! The rst convergence results for the projection method was already given by Chorin, but concerned only cases with absent rigid boundaries (all-space or spatially periodic problems). Later on, qualitative convergence was shown even for the pressure, but in a measure theoretical sense, too weak for practical purposes. Only recently, stronger results on the error behavior of this method have been obtained (see, e.g., Shen [84, 85] as well as [73], and the literature cited therein). The best known error estimate is v n v (tn )

+ p n p(tn )

= O (k ),

(16)

where is a subdomain with positive distance to the boundary . This shows that the quantities p n are indeed reasonable approximations to the pressure p(tn ) , and nally conrms that Chorins original method is a rst-order time stepping scheme for the incompressible Navier-Stokes problem. The key to this result is the re-interpretation of the projection method in the context of the pressure stabilization methods. To this end, one inserts the quantity v n1 = v n1 k p n1 into the momentum equation, obtaining k 1 ( vn v n1 ) v n + ( v n ) v n + p n1 = f n , v n k pn = 0, n p n | = 0. v |n = 0, (17) (18)

This looks like an approximation of the Navier-Stokes equations involving a rst-order (in time) pressure stabilization term, i.e., the projection method can be viewed as a pressure stabilization method with a global stabilization parameter = k , and an explicit treatment of the pressure term. Moreover, it appears that the pressure error is actually conned to a small boundary strip of width k and decays exponentially into the interior of . In fact, it was conjectured that, setting d(x) = dist(x, ) , d(x) |p n (x) p(x, tn )| c exp k k + O (k ). (19)

This conjecture is supported by numerical experiments for the pressure stabilization method applied to the stationary Stokes problem and by some model situation analysis in E/Liu [25]. The analysis of this boundary layer phenomenon requires the study of the singularly perturbed Neumann problem
1 1 D N q = p, in ,

n q| = n p| ,

(20)

where D denotes the Laplacian operator corresponding to Dirichlet bound1 2 ary conditions. Clearly, D is a zero-order operator mapping L () 41

into L2 () . For this problem, one would like to know a decay estimate of the form q c exp p + c p , (21) for interior subdomains := {x , dist(x, ) > } , from which a pointwise result like (19) could be inferred. Such an estimate could be proven in 1 [73] only for the case that the global operator D is replaced by the local identity operator. In the general case the corresponding result is still an open problem. Open Problem 4.2: Prove an analogue of the a priori decay estimate (21) 1 for the non-local operator D .

The occurrence of the pressure boundary layer is demonstrated in Figure 22 for a simple model problem on the unit square with known polynomial solution. It is even possible to recover the optimal-order accuracy of the pressure, O(h2 ) , at the boundary by postprocessing, e.g. by linear or quadratic extrapolation of pressure values from the interior of the domain; see Figure 23 and Blum [15] for more details on this matter.

Figure 22: Sequence of pressure-error isolines obtained by the Chorin scheme with
k = 2.5102 , 6.25103 , 1.56103 (model problem with = 1 on the square); from Prohl [69].

Figure 23: Pressure error plots for a polynomial Stokes solution before (left) and
after (right) correction by extrapolation to the boundary; from Blum [15].

42

An important step towards the solution of the boundary layer problem has been made in Prohl [69, 70] by introducing the Chorin-Uzawa scheme, which reads as follows: (i) Implicit Burgers step: k 1 ( v n v n1 ) vn + v n v n + ( pn 1 pn 1 ) = f n , (ii) Pressure Poisson problem: pn = k 1 v n , (iii) Pressure and velocity update: vn = v n k p n , pn = pn1 v n , < 1. n p n | = 0. v |n = 0.

The reference to the name Uzawa is due to the fact that this scheme partially resembles the structure of the well-known Uzawa algorithm for solving stationary saddle-point problems; see Girault/Raviart [29]. It corresponds to a quasi-compressibility method using the regularization v n + 1 kt pn = 0. (22)

This splitting scheme does not introduce a singular perturbation in the pressure equation and is therefore supposed to be free of any spatial boundary layer. However, it suers from a boundary layer at time t = 0 in case of natural initial data not satisfying unrealistic global compatibility conditions; recall Section 2 for a discussion of such conditions. The conjectured suppression of the spatial pressure boundary layer by the Chorin-Uzawa scheme is conrmed by computational tests; see the example presented in Figure 24. A supporting analysis has been given in Prohl [70] for a modication of the Chorin-Uzawa method to a multi-component scheme which allows for the convergence estimate pn p(tn ) ck, tn 1. (23)

Figures 24 show pressure error plots obtained for a given polynomial solution on the unit-square with viscosity = 1 ; the time step is k = 1/100 and the spatial discretization uses the Q1 /Q1 Stokes element with pressure stabilization on a uniform mesh with mesh-size h = 1/64 .

Figure 24: Pressure error plots for a polynomial solution produced by the standard Chorin scheme (left) and the Chorin-Uzawa scheme (right); from Prohl [69, 70].

43

The projection approach can be extended to formally higher order projection methods. The most popular example is Van Kans Method [100]: For 1 admissible starting values v 0 and p0 compute, for n 1 and some 2 : (i) v n H (second order implicit Burgers step), satisfying
1 ( v n + v n1 ) + v n v n + pn1 = f n1/2 ; k 1 ( v n v n1 ) 2

(ii)

A careful examination of this scheme shows that it can also be interpreted as a certain pressure stabilization method using a stabilization of the form v k 2 t p = 0, in , n p| = 0, (24)

pn H 1 () :

vn = v n k (pn pn1 ) .

Open Problem 4.3: The ecient use of projection methods requires an automatic time-step-size control which should monitor deviation from the fully coupled solution. Design such a method for high-order schemes.

i.e., this method may be viewed as an (implicit) quasi-compressibility method of the form (11) with k 2 ; see [74] and Shen [86]. The projection method may be combined with any of the spatial discretizations described in Section 3. It should be remarked that the simple rst-order Chorin scheme is not suitable for computing stationary limits since it has not the from of a xed-point iteration. In contrast to that, the second-order scheme of Van Kan is designed as a defect-correction iteration and may therefore lead to convergence towards steady state solutions. However, in this case it requires extra pressure stabilization when used together with the conforming Q1 /Q1 Stokes element; in fact the stabilizing eect of the projection step disappears as k 2 t p 0 .

44

Solution of the algebraic systems

In this section, we describe solution algorithms for the nite-dimensional problems arising from the discretization presented in the previous sections. These problems form huge and highly nonlinear algebraic systems with a characteristic structure which is exploited by the algorithms. The solution procedure consists of several nested loops. Usually the outermost loop is an implicit time-iteration. In each time step, the arising nonlinear system is solved by a quasi-Newton or defect-correction iteration. The discretization by nite elements leads to a sparse structure in the system matrices which is exploited by the iterative solution method. Even in the case of the Laplace operator (which is always a part of the system), the inversion by a direct solver or a simple iterative scheme like the conjugate gradient (CG) method is prohibitive due to the bad conditioning of the matrix with decreasing mesh size. Therefore, the use of multigrid methods is mandatory, either directly as solvers or as preconditioners for a robust iterative schemes like the generalized minimal residual (GMRES) algorithm. Since the systems to be solved are in general non-symmetric and indenite, the construction of good multigrid algorithms requires special care. 5.1 Linearization

The time stepping schemes described above require in each time step the solution of nonlinear systems of the form [M + A + N (v )]v + Bp = g , B T v + Cp = c , (1) (2)

where = (k )1 and (on a quasi-uniform mesh) h2 . The operators involved correspond to dierential operators as follows: M id., A diag(D ), N (v ) v , B , B T , C N , where D and N denote the Laplacian operator combined with (homogeneous) Dirichlet or Neumann boundary conditions, respectively. The righthand sides g and c contain information from the preceding time level. Here and below, the same notation is used for the (discrete) velocity v and pressure p and the corresponding nodal vectors. The following iteration schemes are formulated on the continuous level without incorporating stabilization, i.e., we set = 0 and c = 0 . a) Newton method: Starting from some initial values v 0 , p0 H L (for example, taken from the preceding time level), one iterates: 1. Defect: dl = g M + A + N (v l ) v l Bpl . 45

2. Correction: 3. Update:

[M + A + N (v l )]w l + Bq l = dl , v l+1 = v l + l w l , pl+1 = pl + l q l

B T w l = 0. (l damping factor).

This iteration has been observed to converge very fast, provided that it converges at all. The problem is its lack of robustness particularly in the case of larger Reynolds numbers. This is due to the structure of the operator to be inverted in each iteration step: N (v )w = v w + w v. It contains a reaction term w v which eects the main diagonal of the system matrix in an uncontrolled manner and may cause divergence of the iteration. This problem may be avoided by simply dropping the reaction term in the Jacobian which results in the following xed-point defect correction iteration. b) Fixed-point defect correction: Starting from some initial values v 0 , p0 H L (taken again from the preceding time level), one iterates: 1. Defect: dl = g M + A + N (v l ) v l Bpl . [M + A + N (v l )]w l + Bq l = dl , v l+1 = v l + l w l , pl+1 = pl + l q l B T w l = 0. (l damping factor).

2. Correction: 3. Update:

(v l ) = v l only contains In this scheme the preconditioning operator A a transport term which can be stabilized by any of the methods described above: upwinding, streamline diusion, etc. Normally, within the time stepping scheme, only a few (usually 3-5) steps of the defect correction iteration are necessary for reducing the initial residual down to the level of the discretization error. This is our method of choice used in the codes mentioned in the Introduction. c) Nonlinear multigrid iteration: The multigrid method can be applied directly to the nonlinear system; see Hackbusch [36]. This may lead to faster convergence but its optimization is dicult and depends very much on the particular problem. Because of this lack of robustness, we do not advocate nonlinear multigrid for solving the Navier-Stokes equations. d) Nonlinear least-squares cg method: The (nonlinear) least squares cg-method for solving systems like (1) has been proposed by Glowinski/Periaux [31]. Starting from an initial guess x0 , a sequence of approximate solutions (xl )l0 is obtained by minimizing the least squares functional w 2 min! (3) 46

where w is determined by {v, p} through the equation [M + A]w Bq = defect of {v, p}, B T w = 0. (4) It can be seen that each nonlinear cg-step actually requires only the solution of three linear Stokes problems which can be eciently done by linear multigrid techniques. This method is very robust as it is based on the minimization of a positive functional, but the speed of convergence drastically slows down for larger Reynolds numbers. For example, the 3-D driven cavity problem can be solved by the stationary version of the least squares cg method up to about Re = 2000; for further details, see [38]. 5.2 Solution of the linearized problems

The problem to be solved has the form S B T B C g v , = c p A= S B , T B C (5)

where, with some initial guess v , S = M + A + N ( v ). The diculty with this system is that the matrix A is neither symmetric nor denite. It is usually too large for the application of direct solvers (like the LU decomposition by Gaussian elimination) and also the traditional iterative methods (like SOR iteration or Krylov space schemes) do not work suciently well. This suggests the use of multigrid methods which are particularly suited on very ne meshes. However, the construction of ecient multigrid algorithms for solving the indenite system (5) is not at all straightforward. Therefore, as a simpler alternative the Schur complement approach has become popular which will be described in the following subsection. 5.2.1 Schur-complement iteration

In the system matrix A the main block S is regular and usually robust to be inverted. Hence, the velocity unknowns may be eliminated from the system by inverting S which leads to: [B T S 1 B + C ]p = B T S 1 g + c , v = S 1 (g Bp). (6)

The Schur-complement matrix = B T S 1 B + C is regular. Neglecting the inuence of the nonlinear term N ( v ) , its condition number behaves like cond() = O(h2 ) for k h2 , cond() = O(1) for k h2 .

This suggests the use of iterative methods for its inversion, e.g., Krylov space methods like the GMRES or the bi-cg-stab method. In the essentially nonstationary case, k h2 , only a few iteration steps suce. In nonstationary 47

computations where k h2 , preconditioning by an approximation of the Neumann-type operator B T M 1 B is necessary. In each iteration step the operator S 1 has to be evaluated which amounts to solving a linear transport 1 /P0 Stokes elements diusion problem. In the case of the nonconforming Q combined with upwind stabilization of advection S becomes an M-matrix. This facilitates the iterative inversion of , particularly by multigrid meth 1 q a preconditioning ods. The non-exact inversion of makes the step q step within the iteration for inverting . Hence, the number of inner iteration steps should be kept xed during the whole solution process. Another strategy for compensating for the error in the evaluation of S 1 is to embed the outer iteration (6) into a defect correction process; see Bank, et al., [6]. The convergence usually deteriorates for increasing Reynolds number, because of loss of symmetry, and for decreasing time step k , because of the bad conditioning of the operator B T B . For larger Reynolds number the convergence of the Schur complement iteration becomes slow and special preconditioning is necessary. The construction of eective preconditioners is not easy since the operator is not available as an explicit matrix. Another stability problem occurs on meshes containing cells with large aspect ratio. Because of this lack of robustness, the Schur complement method has less potential than the direct multigrid approach which will be described below. Open Problem 5.1: Derive a formula for the dependence of the conditioning of the Schur complement operator = B T S 1 B + C on the Reynolds number and on the mesh aspect ratio h . 5.3 Linear multigrid solution

The main idea underlying a multigrid method is the fast reduction of highfrequency error components by cheap relaxation (smoothing) on the ne meshes and the reduction of the remaining low-frequency error components by defect correction on coarser meshes (coarse-grid correction); see Hackbusch [36] and Wesseling [104], for an introduction to multigrid methods. 5.3.1 Multigrid as a preconditioner

Let A be the nite element system matrix of the linearized equation (5) or an appropriate approximation. While the theory of multigrid is well developed for scalar elliptic equations, the situation is less clear for complicated systems as considered in this paper. From mathematical analysis, we know that the use of the multigrid method as a preconditioner in an outer iteration (e.g., a Krylov space method such as GMRES) requires less restrictive assumptions than using the multigrid method directly as a solver. In the rst case, denoting by M the action of a multigrid step, it is sucient to have an upper bound for the condition of the product MA , whereas in the second case, the eigenvalues of the iteration matrix B = I MA have to be uniformly bounded away from one. Therefore, we choose the rst option to construct 48

a robust iteration scheme for the system (5). As basic solver, one may use the generalized minimal residual method (GMRES) for the preconditioned matrix MA . Here, the multigrid operator M can be interpreted as a certain approximate inverse M A1 . It is not necessary to calculate this matrix explicitly; it is sucient to evaluate the matrix-vector product M , i.e., to apply the multigrid iteration for a xed right-hand side. 5.3.2 Multigrid as a solver

The multigrid iteration makes use of the hierarchy of nite element spaces V0 V1 . . . VL , obtained, for example, in the course of a systematic mesh renement process; strategies for an automatic adaptive mesh renement will be discussed below in Section 7. The connection between these spaces is given by prolongation l 1 operators Pll1 : Vl1 Vl and restriction operators Rl : Vl Vl1 . In the nite element context, these operators are given naturally as Pll1 injection,
l 1 Rl L2 Projection.

The main ingredients of a multigrid scheme are the smoothing operators Sl on each grid level 0 l L (l = 0 corresponding to the coarse initial mesh and l = L to the nest mesh). The explicit form of these operators will be described below. The multigrid iteration M = M(l, z0 , ), (7)

on level l with initial guess z0 and with m1 pre- and m2 post-smoothing steps is recursively dened as follows: Multigrid Algorithm M(l, z0 , ) for l 0 : For l = 0, the multigrid algorithm is given by an exact solver M(l, z0 , ) := 1 A 0 . For l > 0 , the following recursive iteration is performed: 1. Pre-smoothing m1 times: 2. Residual on level l: z1 := Slm1 z0 . rl := Ak z1 . rl1 := Rl rl . q := M(l 1, q0 , rl1) .

3. Restriction to level l 1 : 5. Prolongation to level l : 6. Post-smoothing m2 times:

4. Coarse grid correction starting with q0 = 0: z2 := z1 + Pl1 q .

M(l, z0 , ) := Slm2 z2 .

49

If the multigrid recursion is applied -times on each mesh-level, one speaks of a V -cycle for = 1 and of a W -cycle for = 2 . If the multigrid iteration is used only as a preconditioner for a robust outer iteration scheme, usually the V -cycle suces. If multigrid is used as the primary solver, particularly in the case of nonsymmetric problems, the W -cycle is more robust and therefore to be preferred. In this case, the F -cycle as indicated in the gure below is a compromise between V - and W -cycle. The multigrid cycle with > 2 becomes too expensive and is not used.

v4 v3 v2 v1 v0
Figure 25: Scheme of the multigrid V-cycle (left) and the F-cycle (right) The design of a multigrid algorithm for solving the system (5) requires special care. In particular, the choice of the smoother is a delicate matter since the standard xed-point iterations do not work for the indenite matrix A . This problem can be tackled in various ways.

(1) Damped Jacobi smoother: In the case > 0 , the matrix A is weakly denite which makes it possible to apply even standard methods like the damped Jacobi iteration. However, the resulting algorithm is not very robust and parameter tuning requires care; see [38] for an application to 3-dimensional model problems. For larger Reynolds number the method slows down and multigrid convergence may get lost.

(2) Block-Gauss-Seidel smoother: A simple and successful smoother for the matrix A can be obtained by a cell-wise blocking of the physical variables within a global Gauss-Seidel iteration. This was originally proposed by Vanka [101] for a nite dierence discretization of the Navier-Stokes problem. We 1 /P0 Stokes briey discuss its analogue for the nonconforming rotated Q element. The velocity and pressure unknowns corresponding to a cell K or a patch of cells are grouped together. Indicating the corresponding element system matrices by index loc, these blocks of local velocity and pressure unknowns are simultaneously updated according to
t+1 +1 Sloc vloc + Bloc pt loc = known, T t+1 Bloc vloc = known,

where Sloc = Mloc +Aloc +Nloc ( vh ) . This iteration sweeps over all cell-blocks. The local Stokes problems have the dimension dloc = 9 (in 2D) or dloc = 19 (in 3D), respectively. The corresponding matrices (in 2D) are described in the following gure.

50

node for v : : node for p

Aloc

For cost reduction, the main diagonal blocks Sloc,i may be lumped, Sloc,i t+1 Dloc,i . Furthermore, for increasing robustness, the iteration is damped, vh = t +1 t t vh + ( vh vh ) with 0.9 . The good performance of this smoother for 1 /P0 -element has been demonstrated in Schreiber/Turek [83], Schieweck the Q [82], and Turek [97], and for the Q1 /Q1 -element in Becker [7]. We illustrate 1 /P0 the performance of the multigrid algorithm described above for the Q element by results obtained for solving the driven-cavity problem on grids as shown below.

Sloc,1 O Bloc,1 Sloc,2 Bloc,2 . = O T T Bloc,1 Bloc, 0 2

Figure 26: Driven cavity mesh (left) and computed results: pressure isolines (middle), velocity plot (right). Table 2: Multigrid convergence rates (2 pre- and 1 post-smoothing step by the
Vanka smoother) and number of outer xed-point iterations on uniformly rened meshes.

#cells Re = 1 Re = 100 Re = 1000 Re = 5000

1600 0.081 0.098 0.227 0.285

6400 25600 #iter 0.096 0.121 4 0.099 0.130 6 0.245 0.168 9 0.368 0.370 18

We note that a similar block-iteration can also be used in the context of a incomplete block-LU-decomposition for generating a multigrid smoother; for a detailed discussion of this approach see Braack [17]. From the common multigrid theory for elliptic equations we know that point iterations loose the smoothing property for large mesh-aspect ratios h . The remedy is the use of a smoother which becomes a direct solver in the limit h . Consequently, since our smoother acts like a pointGaussSeidel 51

iteration on the velocity unknowns, we expect problems in the case of strongly stretched grids. Our strategy to overcome this diculty is as follows: Since we expect the cell aspect ratio K to be large only in a small part of the computational domain, we should use an adaptive smoother. This means that we will combine the point smoother with a more robust version just where we need it, for instance on elements with large K . In this approach the nodes are grouped in the direction of the anisotropic mesh renement and iterated implicitly leading to a process which may be termed stringwise block-GaussSeidel method. Let us nally mention a critical problem especially in the use of non uniform grids. The use of iterative solvers makes it necessary to dene a stopping criterion. To this end, we need to measure the residual in the right norm. Clearly, the common weighting by the number of unknowns is not appropriate on nonuniform grids. For an approach towards a solution of this problem based on the Galerkin orthogonality inherent to the multigrid process, we refer to [11] and Becker [8]. (3) Discrete projection smoother: Finally, we present an approach to constructing multigrid solvers for the indenite system (5) which uses the idea of operator splitting as introduced above in Section 4 on time-discretization schemes; see Turek [95, 97]. This method is particularly ecient in the nonstationary case when = 1/k balances /h2 . In the following, we consider the linearized problem arising within a time-stepping scheme as described above in combination with spatial discretization by a Stokes element which does not need pressure stabilization. This problem has the form Sv n + Bpn = g n , B T v n = 0, (8)

with the (momentum) matrix S = M + A + N ( v n ) . The right-hand side g n and the approximation v n are given from the preceding time level. Elimination of the velocity unknown yields again the Schur complement formulation B T S 1 Bpn = B T S 1 g n , v n = S 1 (g n Bpn ). (9)

We have already mentioned that the solution of this problem by Krylov space methods with evaluation of S 1 by multigrid iteration becomes increasingly inecient for small time step k , larger Reynolds number, and on strongly anisotropic meshes. This problem can be overcome by using instead a simple Richardson iteration for the Schur complement equation (9) with a preconditioner of the form B T C 1 B . Popular choices for the preconditioning operator C are: C 1 = I (corresponds to the SIMPLE algorithm). 1 (lumped mass preconditioning). C 1 = M 1 + 1 B T B (Tureks preconditioner) C 1 = M 52

The resulting iteration is termed discrete projection method (see Turek [95]): pn,l+1 = pn,l (B T C 1 B )1 B T S 1 Bpn,l B T S 1 g n . (10)

After L iteration steps, on sets pn := pn,L and computes the corresponding velocity component by solving: Sv n = g n Bpn + 1 (I Sc1 )B (pn,L pn,L1), with some relaxation parameter (0, 1) . This construction of v n ensures that the resulting velocity is in the discrete sense divergence-free, B T v n = 0 , and suggests the name projection method for the whole scheme. The discrete projection method is then used as a smoother within an outer multigrid iteration. In the special case L = 1 , this scheme corresponds to a discrete version of the classical projection methods of Chorin (for the choice pn,0 := 0 ) and of Van Kan (for the choice pn,0 := pn1 , see Gresho [32]. This operator-splitting time-stepping scheme has the form: 1. S v n = g n kBpn1 1 Bq n = k 1 B T v 2. B T M n 1 Bq n 3. v n = v n k M 4. pn = pn1 + q n (Burgers step), (Pressure Poisson equation), (Velocity update), (Pressure update).

All these schemes are variants of the segregated solution approach containing the schemes of SIMPLE-type and other pressure correction schemes as special cases; for a survey see [97] and [98]. The multigrid method with smoothing by the discrete projection iteration (10) has proven to be a very ecient solution method for the fully coupled problem (8); it is robust for all relevant Reynolds number (laminar ows) and time steps. The whole solution process is based on ecient and robust inner multigrid solvers for the subproblems Burgers equation and pressure Poisson equation. The concrete implementation of this algorithm (as described in Turek [97]) requires about 1 KByte memory per mesh cell and shows almost meshsize-independent convergence behavior. As the result, 3D simulations with more than 107 unknowns requiring about 1 GByte of memory can be done on modern workstations. Open Problem 5.2: Derive a good preconditioner (smoother) for the Schur complement iteration (10) in the transport-dominant case.

53

A review of theoretical analysis

In this section, we give an account of the available theoretical analysis for the discretization described in the previous sections. We concentrate on the practical impact of these theoretical results; the main topics are: Problem of regularity at t = 0. Problem of global convergence up to t = . Problem of realistic error constants. We will identify some critical shortcomings of the available theory which lead to challenging questions for further analysis. We assume that the stationary or nonstationary Navier-Stokes equations are discretized by the nite element method as specied in Section 3 combined with one of the time-stepping schemes described in Section 4. (I) For the spatial discretization, we recall the following two representative examples of (quadrilateral) Stokes elements: a) the nonconforming rotated d-linear Q1 /P0 element; b) the conforming d-linear Q1 /Q1 element with pressure stabilization. a)
e e D D

ph e
e

Devh D D D

b)

u D D D D

Duvh (a), ph (a)

These discretizations are of second order expressed in terms of local approximation properties of the nite element functions used:
h Hh

inf

v h c h2 2 ,

H H2 ().

(II) For the time discretization, we think of the Crank-Nicolson scheme or the Fractional-Step- scheme which are both of second order in terms of local truncation error, e.g., for the Crank-Nicolson scheme applied to the homogeneous heat equation, there holds
1 k 1 (v n v n 1 ) + 2 (An v n + An1 v n1 ) c k 2 max [tn1 ,tn ] 2 t v 1

In view of these local approximation properties and the stability of the n schemes, we expect a global a priori estimate for the errors en v := u(, tn ) uh n and en p := p(, tn ) ph of the form
0<tn T

max

n en v + ep

C (, T, data){h2 + k 2 }, 54

(1)

with an error constant C (, T, data) depending on the viscosity parameter , the time-interval length T > 0 , and assumed bounds M for the data of the problem, e.g., M := 2 v 0 + sup f + t f
[0,T ]

< .

If additionally the domain is suciently regular (say, convex or with C 2 boundary), it is guaranteed that the solution {v, p} satises at least the a priori estimate sup 2 v + t v + p < . (2)
(0,T ]

Clearly, the size of the error constant C (, T, data) is of crucial importance for the practical value of the error estimate; we will come to this point in more detail, below. At rst, we have to consider the question of whether an error estimate of the form (1) can be expected to hold at all. In general, the answer is no, unless certain additional conditions are satised. This leads us to the following discussion of the smoothing property. 6.1 The problem of regularity at t = 0

The second-order convergence of the time-stepping scheme expressed in the 2 estimate (1) requires an a priori bound of the form sup(0,T ] t v < . We have seen in Section 2 that there is a principle problem with assuming this degree of regularity in general. Even for arbitrarily smooth data the solution of the Navier-Stokes problem may suer from lim
t0

3 v (t) + t v (t)

= ,

(3)

unless certain non-local (and non-veriable) compatibility conditions are satised for the initial data. We recall from Section 2 the natural regularity assumption for the (nonstationary) Navier-Stokes equations (without additional compatibility condition): v 0 J1 () H2 () sup
t(0,T ]

2 v (t) + t v (t)

< .

(4)

Accordingly, the best possible error estimate for the velocity which can be obtained under these realistic assumptions is sup
tn (0,T ] 2 en v = O{h + k }.

(5)

This estimate is only of rst order in time, in contrast to the postulated secondorder error estimate (1). As a result of the foregoing discussion we obtain the following:

55

Conclusion: For any discretization of the nonstationary Navier-Stokes equations which requires more than the natural regularity inherent to the problem, meaningful higher-order error estimates must be of smoothing type. We call an error estimate of type (1) a smoothing error estimate if it is of the form sup
tn (0,T ] 3/2 en tn en v + tn p 1

C (, T, data){h2 + k 2 }.

(6)

This estimate reects the well-known smoothing behavior of the exact solution {v, p} as t 0 in the (realistic) situation (4): sup
t(0,T ] r tr/21 r v (t) + tr1 t v (t)

2 v 0 + data .

(7)

Smoothing error estimates of the form (6) have been established earlier for standard parabolic problems like the heat equation in the case of rough initial data; see, e.g., Thom ee [89], as well as [62] and [71]. Corresponding results for the Navier-Stokes equation have been given in [43] for higher-order spatial semi-discretization and in [44] for the Crank-Nicolson time-stepping scheme. It turns out that due to the nonlinearity of the problem, the maximal achievable orders of smoothing error estimates under assumption (4) is O(h6 ) for the spatial discretization and accordingly O(k 3 ) for the time stepping (provided that the scheme is strongly A-stable). This particularly implies the result (6) stated above. The existence of a natural order-barrier for the smoothing property of nite element Galerkin schemes applied to nonlinear problems has been established by Johnson, et al. [53]. We adapt the following example from [53] for the situation of H 2 -regular initial data as relevant for the case of the nonlinear Navier-Stokes equations. Example: Example of limited smoothing property For x (, ) and t > 0 , we consider the system of equations
2 t u x u = 4 min{v 2 , 1}, u(x, 0) = u0 (x) := 0, 2 t v x v = 0, v (x, 0) = v 0 (x) := mr cos(mx),

with periodic boundary conditions. For any xed m N and r N {0} , the exact solution is u(x, t) = m2r2 1 e2m
2 2t

1 + e2m t cos(2mx) ,

v (x, t) = mr em t cos(mx) .

For spatial semi-discretization of this problem, let the Galerkin method be used with the trial spaces Sm := span 1, cos(x), sin(x), ..., cos((m 1)x), sin((m 1)x) , 56

if we set h := m1 . This demonstrates that, for v 0 H 2(, ) , i.e., for r = 2 , the best achievable order of approximation for t > 0 is indeed O(h6 ) .

and let Pm denote the L2 projection onto Sm . Since Pm v 0 = 0 , taking as usual Pm u0 and Pm v 0 as initial values for the Galerkin approximation results in the Galerkin solutions vm (t) = 0 and um (t) = 0 . Consequently, for xed t > 0 , there holds (um u)(t) = u(t) 2m2r2 = 2 v 0 r h2r+2 ,

There is another remarkable aspect of the estimate (6) which concerns the Crank-Nicolson scheme. This scheme, due to its absent damping properties (not strongly A-stable), possesses only a reduced smoothing property. In consequence, even in the case of the linear heat equation, for initial data v 0 L2 () only qualitative convergence en v 0 (h, k 0) can be guaranteed at xed tn = t > 0 . For even stronger initial irregularity (e.g., v 0 = x a Dirac measure) divergence en (h, k 0) occurs. However, the optimal smoothing behavior is recovered if one keeps the relation k h2 . This undesirable step-size restriction can be avoided simply by starting the computation with a few (two or three) backward Euler steps; for examples and an analysis, see [61], [71], and the literature cited therein. Surprisingly, such a modication is not necessary for more regular initial data (half way up to the maximum 1 regularity), v 0 H0 () H 2 () . In this case the Crank-Nicolson scheme admits an optimal-order smoothing error estimate of the form For the heat equation this is easily seen by a standard spectral argument; see Chen/Thom ee [23] and [71]. The extension of the smoothing error estimate (8) to the nonlinear (nonsymmetric and nonautonomous) Navier-Stokes equations is one of the main results in [44]. Finally, we mention some further results on smoothing error estimates relevant for the Navier-Stokes equations together with some open problems: Second-order projection schemes, particularly the Van Kan scheme, have been analyzed by Prohl [69, 70] and optimal-order smoothing error estimates have been established. The second-order smoothing property of the Fractional-Step- scheme has been proven by M uller [64]. Open Problem 6.1: The construction of compatible initial data from experimental data has already been formulated as a problem. If this is not possible, it would be interesting to estimate the time length over which the error pollution eect of incompatible initial data persists. Open Problem 6.2: Establish the optimal smoothing property of any higherorder (q 3) time discretization schemes for the Navier-Stokes equations. 57
2 0 en v C , T, v , data 1 2 h2 + t , n k

tn (0, T ].

(8)

6.2

The problem of convergence up to t =

The error constants in the a priori error estimates (1) usually grow exponentially in time, C (, T, data) KeT ,

unless the data of the problem is very small, actually of size 2 , such that nonlinear perturbation terms can be absorbed into the linear main part. This exponential growth is unavoidable in general, due to the use of Gronwalls inequality in the proof. In fact, the solution to be computed may be exponentially unstable, so that a better error behavior cannot be expected. To improve on this situation, one has to make additional assumptions on the stability of the solution. Instead of requiring the data of the problem to be unrealistically small, the solution {v, p} itself is supposed to be stable. This assumption may not be veriable theoretically; nevertheless it may be justied in many situations in view of experimental evidence. A discussion of various types of stability concepts for nonstationary solutions of the Navier-Stokes equations in view of numerical approximation can be found in [42, 45]. Exponential Stability: A solenoidal solution v is called (conditionally) exponentially stable, if for each suciently small initial perturbation w 0 J1 () , w 0 < , at any time t0 0 , the solution v (t) of the perturbed problem t v v+v v + q = 0 , starting from v (t0 ) = v (t0 ) + w 0 , satises (v v )(t) Ae(tt0 ) w 0 , t t0 , (9) t t0 ,

with certain constants A > 0 and > 0 . In this assumption it is essential that the decay of the perturbation is proportional to the size of the initial perturbation w 0 . For global strong solutions this concept of exponential L2 -stability is equivalent to corresponding stability concepts expressed in terms of stronger norms, e.g. the H 1 norm; see [45]. It has been proved in a series of papers [42, 43, 44] that exponentially stable solutions can be approximated uniformly in time, i.e.,
tn 1 n sup vh v (, tn ) C {h2 + k 2 }.

(10)

In this estimate the error constant C = C (A, ) depends on the stability parameters of the solution. The proof uses a continuation argument. We sketch its essential steps for semi-discretization in time by the backward Euler scheme.

58

Proof of the global error estimate (10): n (i) We recall the local bound for the error en = v (tn ) vk , en E (tn )k, tn 0, (11)

involving the exponentially growing error constant E (t) := Ket . By continuity, it can be assumed that this error estimate holds with the same constant for all solutions neighboring the true solution v . The proof of this statement is technical and uses the particular properties of the discretization scheme considered. Further, let T = Nk be a xed time length such that with the stability parameters of the solution v , there holds
1 AeT 2 .

(12)

Figure 27: Scheme of induction proof (following [42]) (ii) Suppose now that the desired error estimate is already known to hold on some time interval (0, tm ] , tm T , with error constant K := 2E (T ) . Let v (t) be the solution of the perturbed problem t v v+v v + p = 0, t tm ,

m starting at tm with initial value v (tm ) = vk . In virtue of the assumed exponential stability of the solution, there holds

(v v )(t) Ae(ttm ) em , for suciently small k guaranteeing time length T , we obtain

t > tm ,

em < . Then, stepping forward by

m+n em+n (v v )(tm + T ) + v (tm + T ) vk .

59

Here, the rst term is bounded by K k , in view of (11) and the induction assumption (12), while the second one can be controlled by E (T )k , using the local error estimate (11) for v vk . Hence, it follows that em+n K k. The assertion then follows by induction with respect to multiples of T . The argument presented for the global error estimate (10) appears simple and general; however, in concrete situations involving simultaneous discretization in space and time there are several technical diculties. The initial value for the perturbed solution v (tn ) in the induction step may not be admissible (i.e., not exactly divergence-free or even nonconforming). Further, the use of the local error bound (11) for the perturbed error v vh requires control on higher-order regularity of the corresponding initial value v (tn ) . These and some other complications can be overcome as shown in [42, 43, 44], for dierent types of spatial as well as time discretization. 6.3 The problem of realistic error constants

In the preceding sections, we have discussed the derivation of qualitative a priori error estimates, local as well as global in time. Now, we turn to the more quantitative aspect of the size of error constants relating to the question of practical relevance of the a priori results. To this end, let us briey summarize the results of a priori error analysis presented so far: a) In the stationary case, we can guarantee convergence behavior like ev Chp provided that the solution v is suciently smooth and locally unique (i.e., the linearization of the nonlinear Navier-Stokes operator at v is regular). Then, the error constant C (, v ) depends on bounds on the regularity of v as well as its stability, on the viscosity , and of course on the characteristics of the discretization. b) In the nonstationary case, we can guarantee convergence behavior like
p q en v C {h + k }

0 tn T,

provided again that the solution v is suciently smooth. The error constant C (, T, v ) depends on bounds on the regularity of v , on the viscosity , and additionally on the length of the time interval T . A question naturally arises: How large is C ? In normal situations as, for example, for the Poisson problem or the heat equation, the error constant may be shown to be of moderate size C 1 104 , depending on the situation and the care spent in the estimation. The qualitative conclusion from the estimates 60

may then be that the the error bound is reduced by a factor of 2 min{p,q} if the mesh size is halved in space and time. The (not unrealistic) hope is that this carries over to the true discretization error. Unfortunately, the Navier-Stokes equations do not at all a normal problem; it is of mixed elliptic-hyperbolic or parabolic-hyperbolic type with degenerating ellipticity. This has decisive consequences for the size of the error constants C . Normalizing the ow conguration as usual to characteristic length L = 1 and velocity U = 1 , the Reynolds number for common cases is Re = 1 1 105 , which relates to laminar ow, and the characteristic time length is T 1/ . This means that it takes the time T 1 for the ow to reach a characteristic limit behavior, e.g. stationary or time periodic. The question can now be made more precise: How do the error constants C depend on Re ? This dependence has several sources: the explicit occurrence of in the dierential operator, the dependence of the solutions regularity on (boundary layers), the dependence on the length of the time interval T 1/ , the dependence of the solutions stability on . Let us discuss the mechanisms of these dependencies separately. (i) Structure of the dierential operator: The standard procedure in the stationary case is to absorb the lower-order terms into the linear main part v which leads to the dependence C 1 . In the nonstationary case the lower-order terms are absorbed into the term t v by the use of Gronwalls inequality resulting in C eKT / , K sup(0,T ] v 1/2 .

This dependence on can be formally removed by using streamline-diusion damping for the transport term, but it leaves the T -dependence. (ii) Regularity of solution: For small boundary layers of width occur. This implies that sup |v | 1/2 , C p v (d,p) . This problem can be solved by proper mesh renement in the boundary layer. (iii) Length of the time interval: It was demonstrated above that the local worst case error constant C eKT / 61

becomes independent of the time interval-length T , C eKT / , if the solution can be assumed to be exponentially stable. Here, T is suciently large but xed. However, tracing constants in the proof, we see that T 1 rendering this formally global error bound practically meaningless for small . (iv) Stability of the solution: The argument for proving error estimates for nonlinear problems rely on assumptions on the stability of certain linearized tangent operators. The resulting error constants can in general not be assumed to behave better than C eKT / . This exponential dependence seems unavoidable unless something dierent is shown in particular situations. The observability of laminar ows even for higher Reynolds numbers indicates that these ows may possess better stability properties than expressed by the worst case scenario addressed above. An analogous conclusion may be true even for turbulent ows with respect to certain averaged quantities. Conclusion: In general, one has to admit that the error constants depend exponentially on 1 , unless something dierent is proven. Realizing that even in the range of laminar ows, 20 Re 104 , e20 5108 , e100 1043 , e1000 , the practical meaning of available a priori error estimates seems rather questionable! The above observation seems to indicate that there is a conceptual crisis in the theoretical support of CFD as far as it concerns the computational solution of the Navier-Stokes equations. This is contrasted by the abundant body of research papers reporting successful computations of viscous ows and the good agreement of the obtained results with experimental data. Hence, we reformulate the question: Is there any theoretical support that certain ows (i.e., solutions of the Navier-Stokes equations) can actually be computed numerically. If the answer were no, everybody should be worried. We again emphasize that the presence of an asymptotic error estimate of the form v vh C {hp + k q } cannot be taken as justication for the meaningful performance of a numerical scheme, unless the error constant C is shown to be of moderate size at least for certain model situations of practical interest. Reliable ow simulation requires computable error bounds in terms of the approximate solution; the elements of such an a posteriori error analysis will be described in Section 7 below. 62

In proving useful error estimates, we have to deal with the question of proper concepts for describing the stability of solutions relevant for numerical approximation. Qualitatively, all stability concepts may be equivalent but this strongly depends on the viscosity parameter . The choice of the wrong norm may lead to unfavorable dependence on , like O( 2 ) rather than the generic behavior O( 1 ) . Actually, the fundamental question whether there are practically interesting situations in which the solution of the Navier-Stokes equations are stable, with stability constant cS 1 seems open. Results in this direction appear necessary for a rigorous error analysis of discretization schemes. However, until now, practically meaningful a priori error bounds are not even available for such basic situations as Couette ow (constant sheer ow) and Poiseuille ow (constant pipe ow); we will address this question in more detail in the following section. 6.4 Towards a quantitative a priori error analysis

The following discussion is of conceptual nature. In order to abstract from the nonessential technicalities of nite element discretization, we consider the idealized situation of an exactly divergence-free approximation, using subspaces Vh V := J1 () . Accordingly, the discretization delivers only approximations vh Vh to the velocity v V . The associated pressures ph are then to be determined by post-processing. Further, we restrict us to the very basic case of homogeneous Dirichlet boundary conditions v| = 0 . 6.4.1 The stationary case

We begin with the stationary Navier-Stokes problem v + v v + p = f, Using again the notation a(v, ) = (v, ), n(v, v, ) = (v v, ), v = 0, in , v| = 0. (13)

the pressure-free variational formulation seeks v V , such that A(v ; ) := a(v, ) + n(v, v, ) = (f, ) V. The corresponding nite element discretization seeks vh Vh , such that A(vh ; h ) = (f, h ) h Vh . (15) (14)

All error analysis of this discretization is based upon the (nonlinear) Galerkin orthogonality: A(v ; h ) A(vh ; h ) = 0, h Vh . (16) In the following, we denote the error by e := v vh . 63

a) The small data case, v : The Fr echet derivative taken at v of the semi-linear Form A(; ) is given by L(v ; , ) = a(, ) + n(v, , ) + n(, v, ). Under the small data assumption, this bilinear form is coercive on V with stability constant cS = cS ( ) 1 : cS L(v ; , ). Linearization and Galerkin orthogonality then leads to the relation e cS L(v, e, e) = cS n(e, e, e) + A(v ; v h ) A(vh ; v h ) , with an arbitrary approximation h Vh to v . From this we infer that, for suciently small h , e cS Ch, (17) with an error constant C = C ( 2 v , data) . b) The general case of an isolated solution: Now, the solution v is assumed to be stable in the sense that cS sup L(v, , ) , V (18)

with some stability constant cS = cS (, v ) . Again, by linearization and Galerkin orthogonality, it follows that e cS Ch. Further, assuming stability of the Fr echet derivative in the form cS L(v, , ) , 2 VH2 () sup e cS Ch2 . (20) (19)

one may apply the usual duality argument to obtain the L2 -error bound (21)

These results rely on the assumption of stability of the problem expressed in terms of the stability constant cS ; see [56]. In order to use the resulting estimates, on has to determine these constants either analytically, which may rarely be possible, or computationally by solving dual problems. Numerical experiments for the driven-cavity problem reported by Boman [16] show a dependence like cS ( ) 1 in the (laminar) range 1 Re 103 . This investigation should be extended to other elementary ows in order to see whether linear growth cS Re is generic for laminar ow. The answer is not clear yet, as indicated by the following simple example. 64

An example of bad stability (from Tobiska/Verf urth [91]): For the worst-case scenario, we quote the one-dimensional Burgers equation vxx + vvx = 0, x (1, 1), v (1) = 1, v (1) = 1.

The exact solution is v (x) = 2 tanh( x) , where is the unique positive solution of 2 tanh( ) = 1 . Linearization at this solution results in the boundary value problem zxx + vzx + zvx = f x (1, 1), which has the solution
x t

z (1) = z (1) = 0,

z (x) = 1 eU (x)/

eU (t)/
1 0

f (s) ds + c

dt,

where U is a primitive of v and the constant c is determined by imposing the boundary condition z (1) = 0 . We ask for the best possible bound in the stability estimate z H 1 C ( ) f H 1 . For the particular choice f (x) = cosh( x) , there holds cosh3 ( ) cosh3 ( x) . z (x) = 2 cosh2 ( x) 3 Since z

2 z 1,

2 2 f
,

it follows that, for 1 , z

ce1/ f

and consequently, C ( ) e1/ . This seems to indicate that Burgers equation is not numerically solvable which, however, contradicts practical evidence. The explanation may be that for the performance of discretization stability is essential in other more local measures then those considered above. Open Problem 6.3: Explain the success of discretization methods in computing solutions to the Burgers equation despite its bad conditioning with respect to the energy norm. 6.4.2 The nonstationary case

We now turn to the nonstationary Navier-Stokes problem posed on a time interval I = [0, T ] , t v v + v v + p = f in I, 65 v| = 0, v|t=0 = v 0 . (22)

The corresponding (pressure-free) space-time variational formulation uses the function space V(I ) := H 1 (I ; V) and the space-time forms (, )I =
I

(, ) dt,

aI (, ) =
I

a(, ) dt,

nI (v, , ) =
I

n(v, , ) dt.

Then, a solution v V(I ) is sought satisfying A(v ; ) = F () V(I ), where F () := (f, )I + (v 0 , (0)) and A(v ; ) := (, )I + aI (, ) + nI (v, , ) + (v 0 , (0)). Here, the initial condition v (0) = v 0 is incorporated weakly. In the following conceptual discussion, we restrict ourselves to the semidiscretization in time leaving the spatial variable continuous. The discretization is by the discontinuous Galerkin method of degree r = 0 (dG(0) method) which is a variant of the backward Euler scheme. The time interval I = [0, T ] is decomposed like 0 = t0 < t1 < ... < tM +1 = T , and we set Im = [tm1 , tm ), k (t) k|Im = tm tm1 .
m m [v ]m = v+ v .

(23)

For piecewise continuous functions on this decomposition, we write

m v = lim v (tm s), s0+

Accordingly, we dene the discrete semi-linear form

M

Ak (v ; ) :=
m=0 M

(t v, )m + am (v, ) + nm (v, v, ) +
0 0 ([v ]m , m + ) + (v+ , + ), m=1

and the time-discrete Spaces Vk (I ) = {vk L2 (I ; V), v|Im P0 (Im ), m = 1, ..., M }. The time-discrete approximation then seeks a vk Vk (I ), such that Ak (vk ; k ) = F (k ) k Vk (I ). (24)

We note that this formulation contains the initial condition v (0) = v 0 in the weak sense. The essential feature of the dG(r) schemes are their Galerkin 66

orthogonality property. Using the fact that the continuous solution v also satises the discrete equation (24), we have Ak (v ; k ) Ak (vk ; k ) = 0, k V (I ). (25)

In order to estimate the error e := v vk , we may employ a parabolic duality argument. The adjoint of the Fr echet derivative of the governing semi-linear form taken at the solution v is given by L (v ; , z ) = (, t z )I + aI (, z ) + nI (v, , z ) + nI (, v, z ) ((T ), z (T )). Then, we introduce the dual solution z V(I ) as solution of the space-time dual problem L (v ; , z ) = (e(T ), (T )) V(I ). Taking now = e in (26), we obtain the error representation e(T ) = L (v ; e, z ) = (e, t z )I + aI (e, z ) + nI (v, e, z ) + nI (e, v, z ). Using the Galerkin orthogonality (25) and interpolation estimates on each subinterval Im , we conclude the estimate
M

(26)

e(T ) =
m=0

m m (f, z zk )m ([v ]m , z+ zk, +) I I

cS max [v ]m + max kf
T k M 0

with the stability constant cS = cS (, T, v ) given by | log(kM )|1/2 z

I

+ | log(kM )|1/2

t z dt cS z (T ) .

(27)

The result is the nal time a priori error estimate e(T ) cLk cS max kt v ,
I

(28)

where Lk := maxI 1 + log(k )

1/2

Conclusion: The foregoing discussion shows that proving a priori error estimates for numerical approximations is closely connected with the study of stability properties of linearizations of the Navier-Stokes equations, i.e. with hydrodynamic stability; for more details see [55]. The goal is to derive a priori estimates of the type (18), (20) and (27) with quantied constants cS = cS (, T, data) . The dependence of these constants on the Reynolds number may be linear in good cases or may deteriorate to exponential in bad cases. Such exponentially unstable ows are not computable over relevant intervals of time. The same question will also be crucial for the derivation of a posteriori error estimates discussed in Section 7, below. 67

6.5

The problem of stability constants (A critical review of hydrodynamic stability theory)

Most of the traditional stability theory for uid ow is of qualitative nature being based on eigenvalue criteria through a linearized stability argument. The theoretical results are in good agreement with experiments concerning the critical Reynolds number at which the rst bifurcation occurs; well-studied examples are the B enard problem and the Taylor-Couette problems. But they do not t with experiments for the other fundamental case of parallel ow. There are several paradoxes observed: Poiseuille ow (between two parallel plates) is predicted to turn turbulent at Re 5772 through 2d Tollmien-Schlichting waves, while experiments show instability with essential 3d features somewhere in the range Re 1000 10000 depending on the experimental setup. Couette ow (parallel shear ow) is supposed to be stable for all Re > 0 , but experiments show instability for Re 300 1500 . This failure of theory was blamed on the deciency of linearized stability theory being valid only for small perturbations. However, linearized stability theory is okay, but was only wrongly interpreted. In dynamic systems governed by nonnormal matrices, one has to look at the size of the total amplication factors for the initial perturbation and not only at the sign of the eigenvalues real parts. It is observed that, for example in Poiseuille ow there occurs amplication by a factor of 104 for Re 549 . Some of the relevant references on this subject are Landahl [59] and Trefethen et al. [92], to mention only a few. More references can be found in [54, 55] where this new concept in hydrodynamic stability theory is discussed in view of numerical approximation. In fact, the question of quantitative hydrodynamic instability and that of numerical computability of laminar ows are closely related. In transition to turbulence one seeks to establish lower bounds on the growth of perturbations in order to understand how a laminar ow may develop into a turbulent ow. In error control in CFD for laminar ows one seeks upper bounds of the growth of perturbations related to discretizations of the Navier-Stokes equations. We want to illustrate the phenomenon of error amplication by two simple examples taken from [55]. a) An ODE model: At rst, we consider the simple ODE system w 1 + w1 + w2 = 0, w 2 + w2 = 0. (29) (30)

Here, wi stand for the diusion terms and w2 in the rst equation for the coupling in the transport term of the linearized perturbation equation of the 68

Navier-Stokes equations. The corresponding coecient matrix A= 1 0

is non-normal; the only eigenvalue = has algebraic multiplicity two. This is just the situation described above. For this linear system the solution corresponding to the initial values w (0) = w 0 is given by
0 0 w1 (t) = et w1 tet w2 , 0 w2 (t) = et w2 .

We see the exponential decay of the second component and the linear growth 0 over the interval [0, 1 ] to size w1 ( 1 ) = 1 e1 w2 of the rst component before the exponential decay sets in. The component w2 acts like a catalyst in the rst equation. Although exponentially decaying it rst causes w1 to grow; 0 the later exponential decay is irrelevant when by the growth to size 1 w1 the linearization is no longer valid. b) A simple ow model: Next, we consider a very simple conguration: the ow in an innite pipe = R extending in the x1 -axis with cross section in the (x2 , x3 )-plane. The ow is driven by a volume force f = (f1 (x2 , x3 , t), 0, 0)T (gravitation) in x1 -direction. The solution is supposed to have the form (like x1 -independent Poiseuille ow): v = (v1 (x2 , x3 , t), 0, 0)T , p = p(x, t).

Then the Navier-Stokes equations take the form t v1 v1 + 1 p = f1 in , v1| = 0 .

The corresponding linearized perturbation equation is t w1 w1 + v1 1 w1 + 2 v1 w2 + 3 v1 w3 + 1 q = 0 , t w2 w2 + v1 1 w2 + 2 q = 0 , t w3 w3 + v1 1 w3 + 3 q = 0 , with the incompressibility condition 2 w2 + 3 w3 = 0, and the initial and boundary conditions w|t=0 = w 0 and w| = 0 . Even this simple problem is still too complex for an explicit solution. Therefore, we simplify it further by assuming that the perturbed solution {w, q } is also independent of x1 . This corresponds to looking at a uid in a long vertical tube under gravity or in a long rotating tube with varying speed of rotation. Under this assumption the perturbation equation reduces to t w1 w1 + 2 v1 w2 + 3 v1 w3 = 0, t w2 w2 + +2 q = 0 , t w3 w3 + +3 q = 0 . 69

In this system the equations for the components w := {w2 , w3} together with the constraint 2 w2 + 3 w3 = 0 form a two-dimensional Stokes problem which can be solved independently of the rst equation. Hence, we are in a similar situation as in the above ODE example. For the Stokes subsystem we have the standard a priori estimate w (t) et w 0 , t 0,

with = diam( ) . The rst equation does not contain the pressure. Using the result for w we obtain for the rst component w1 the bound
0 w1 (t) cet t w1 + w 0

t 0.

(31)

Hence, we see that for this model problem, one can show that the error constant in the a priori error estimate (1) grows at most linearly with the Reynolds number: C (, T, data) max{T, Re}. It is an open question whether this linear dependence on Re is generic for a larger class of ow problems. Numerical experiments for the lid-driven cavity ow show such a dependence. Open Problem 6.4: Prove a posteriori stability estimates like (31) for more practical problems (e.g. Poiseuille ow) possibly with respect to dierent norms. Is there any indication that linear growth in time of perturbations may be generic to the Navier-Stokes equations?

70

Error control and mesh adaptation

This section is devoted to concepts of error estimation and mesh optimization. The goal is to develop techniques for reliable estimation of the discretization error in quantities of physical interest as well as economical mesh adaptation. The use of a nite element Galerkin discretization provides the appropriate framework for a mathematically rigorous error analysis. On the basis of computable a posteriori error bounds the mesh is locally rened within a feed-back process yielding economical mesh-size distributions for prescribed error tolerance or maximum number of cells. On the resulting sequence of rened meshes the discrete problems are solved by multi-level techniques. The general concept of residual-based error control for nite element methods is described in the survey article by Eriksson/Estep/Hansbo/Johnson [26]; this technique has then been further developed for various situations in [12, 14]. The application to incompressible ows is extensively discussed in Becker [7, 9]. Extensions to compressible ow including chemical reactions are given in Braack [17]; see also [18]. A survey of applications of this approach to a variety of other problems can be found in [75]. 7.1 Principles of error estimation

The discretization error in a cell K splits into two components, the locally produced error (truncation error) and the transported error (pollution error)
loc trans etot . K = eK + eK

(1)

The eect of the cell residual K on the local error eK , at another cell K , is governed by the Green function of the continuous problem. This is the general philosophy underlying our approach to error control.

Figure 28: Scheme of error propagation (I) A priori error analysis: The classical a priori error estimation aims at estimating the error to be expected in a computation which is still to be done. These bounds are expressed in terms of powers of a mesh size h and involve constants which depend on the (unknown) exact solution. In this way, only asymptotic (as h 0) information about the error behavior is provided but 71

no quantitatively useful error bound. In particular, no criterion for local mesh adaptation is obtained. (II) A posteriori error analysis: The a posteriori error analysis generates error estimates in the course of the computation. Accordingly, these bounds are in terms of computable local residuals of the approximate solution and do not require information about the exact solution. However, a posteriori error analysis usually does not provide a priori information about the convergence of the discretization process as h 0. We illustrate the basic principles underlying error estimation by considering Rnn , b, perturbations of linear algebraic systems. Let A, A b Rn be given and solve x Ax = b , A = b (perturbed problem). (2) For estimating the error e := x x , there are several approaches. The a priori (x x method uses the truncation error := Ax b=A ), 1 e=A e c S , (3) 1 . The a posteriori method with the discrete stability constant c S := A uses the residual := b Ax = A(x x ), e = A1 e cS , (4)

with the continuous stability constant cS := A1 . Alternatively, we may use the solution z of the dual problem A z = e 1 e , to obtain e = (e, A z ) = (b Ax , z ) = (, z ) z c S , (5)

1 with the dual stability constant c . Of course, this approach does S := A not yield a new result in estimating the error in the l2 -norm. But it shows the way to bound other error quantities as for example single components |ei |. An analogous argument can also be applied in the case of nonlinear equa : Rn Rn be (dierentiable) vector functions and solve tions. Let F, F

F (x) = b ,

( F x) = b (perturbed problem).

(6)

Then, the residual := b F ( x) satises

1

= F (x) F ( x) =

F ( x + se) ds e =: L(x, x )e,

0

(7)

with the Jacobian F . The term in parentheses denes a linear operator L(x, x ) : Rn Rn which depends on the (unknown) solution x. It follows that e cS , with the (nonlinear) stability constant cS := L(x, x )1 . Below, we will use this duality technique for generating a posteriori error estimates in Galerkin nite element methods for dierential equations. 72

7.1.1

A diffusion model problem

For illustrating our concept, we start with the (scalar) model diusion problem u = f in , u = 0 on , (8)

posed on a polygonal domain R2 . In its variational formulation one seeks 1 u V := H0 () satisfying (u, ) = (f, ) V. (9)

We consider a nite element approximation using piecewise (isoparametric) bilinear shape functions (see Section 3). The corresponding nite element into quadrilaterals spaces Vh V are dened on decompositions Th of (cells) K of width hK := diam(K ). We write again h := maxK T hK for the maximal global mesh width. Simultaneously, the notation h = h(x) is used for the continuously distributed mesh-size function dened by h|K = hK . For ease of mesh renement and coarsening we allow hanging nodes, but at most one per edge. The shape of the corresponding modied basis function is shown in Figure 29.

l 1
+1 l 2

+1 l 1

+1 1 l+1 l = l 1 1 4 2

4 x

Figure 29: Q1 nodal basis function on a patch of cells with a hanging node The discrete problem determines uh Vh by (uh , h ) = (f, h ) h Vh . We recall the Galerkin orthogonality of the error e := u uh , (e, h ) = 0, 73 h Vh . (11) (10)

We seek to derive a posteriori error estimates. Let J () be an arbitrary error functional dened on V and z V the solution of the corresponding dual problem (, z ) = J () V. Setting = e in (12) results in the error representation J (e) = (e, z ) = (e, (z Ih z )) =
K T

(12)

(u + uh , z Ih z )K (n uh , z Ih z )K
1 (f + uh , z Ih z )K 2 (n[uh ], z Ih z )K ,

(13)

=
K T

where [uh ] is the jump of uh across the interelement boundary. In the second equation, we have used galerkin orthogonality. This gives us the a posteriori error estimate |J (e)| (uh ) := with the cell residuals
1 K (uh ) := h K f + uh K,

h4 K K (uh )K (z ) + K (uh )K (z ) ,
K Th

(14)

K (uh ) := hK

3/2

n[uh ]

K ,

and the weights

3 K (z ) := h K z Ih z K, 1 K (z ) := 2 hK 5/2

z Ih z

These quantities are normalized, such that they can be expected to approach certain meshindependent limits as h 0 . The interpretation of the relation (14) is that the weights K (z ) describe the dependence of J (e) on variations of the cell residuals K (uh ) , J (e) 4 2 h4 K K (z ) hK maxK | z |. K We remark that in a nite dierence discretization of the model problem (8) the corresponding inuence factors behave like K (z ) h2 K maxK |z |. In practice the weights K (z ) have to be determined computationally. Let zh Vh be the nite element approximation of z , (h , zh ) = J (h ) h Vh . We can estimate
1 2 K (z ) cI h K z K

(15)

cI max |2 h zh |,
K

(16)

74

2 where 2 h zh is a suitable dierence quotient approximating z . The interpolation constant is usually in the range cI 0.1 1 and can be determined by calibration. Alternatively, we may construct from zh Vh a patchwise (2) biquadratic interpolation Ih zh and replace z Ih z in the weight K (z ) by (2) Ih zh zh . This gives an approximation to K (z ) which is free of interpolation constants. One may try to further improve the quality of the error estimate by solving local defect equations, either Dirichlet problems (` a la Babuska/Miller) or Neumann problems (` a la Bank/Weiser); see Backes [4]. References for these approaches are Verf urth [102] and Ainsworth/Oden [1]. Comparison with simpler mesh adaptation techniques, e.g. renement criteria based on dierence quotients of the computed solution, local gradient recovery ZZ technique (` a la Zienkiewicz/Zhu [106]), or other local ad hoc criteria have been reported in Braack [17] and in [75]. By the same type of argument, one can also derive the traditional global error estimates in the energy and the L2 norm.

(i) Energy-norm error bound: Using the functional J () := e

1

(e, )

in the dual problem, we obtain the estimate e h4 K K (uh ) K (z ) cI h2 K K (uh ) z

, K

K T

K T

is the union of all cells neighboring K . In view of the a priori bound where K z cS = 1 , this implies the a posteriori error estimate e E (uh ) := cI
2 h4 K K (uh ) K T 1/2

(17)

(ii) L2 -norm error bounds: Using the functional J () := e

1

(e, )

in the dual problem, we obtain the estimate e h4 K K (uh ) K (z ) cI

2 h3 K K (uh ) z K.

K T

K T

In view of the a priori bound 2 z cS (cS = 1 if is convex), this implies the a posteriori error bound e L2 (uh ) := cI cS 75
2 h6 K K (uh ) K T 1/2

(18)

7.1.2

A transport model problem

As a simple model, we consider the scalar transport equation u = f , (19)

on a domain R2 with inow boundary condition u = g along the inow boundary = {x , n < 0} . Accordingly, + = \ is the outow boundary. The transport vector is assumed as constant for simplicity; therefore, the natural solution space is V := {v L2 (), v L2 ()}. This problem is discretized using the Galerkin nite element method with streamline diusion stabilization as described above. On quadrilateral meshes Th , we dene again subspaces Vh = {v H 1 (), v|K Q1 (K ), K Th } , where Q1 is the space of isoparametric bilinear functions on cell K . The discrete solution uh Vh is dened by ( uh f, + ) + (n (g uh ), ) = 0 Vh , (20) where the stabilization parameter is determined locally by K = hK . In this formulation the inow boundary condition is imposed in the weak sense. This facilitates the use of a duality argument in generating a posteriori error estimates. Let J () be a given functional with respect to which the error e = u uh is to be controlled. Following our general approach, we consider the corresponding dual problem ( , z + z ) (n, z ) = J () V, (21)

which is a transport problem with transport in the negative -direction. We note that the stabilized bilinear form Ah (, ) is used in the duality argument, in order to achieve an optimal treatment of the stabilization terms; for a detailed discussion of this point see [75] and [47]. The error representation reads J (e) = ( e, z zh + (z zh )) (ne, z zh ) , for arbitrary zh Vh . This results in the a posteriori error estimate |J (e)| (uh ) := cI with the cell residuals
1 K (uh ) := h K f uh K,

h4 K K (uh )K (z ) + K (uh )K (z ) ,
K T

(22)

K (uh ) := hK

3/2

n (uh g )
5/2

K ,

and cell weights (setting := z zh )

3 K (z ) := h K

+ K

K (z ) := hK

K .

76

We note that this a posteriori error bound explicitly contains the mesh size hK and the stabilization parameter K as well. This gives us the possibility to simultaneously adapt both parameters, which may be particularly advantageous in capturing sharp layers in the solution. We want to illustrate the features of the error estimate (22) by a simple thought experiment. Let = (0, 1)(0, 1) and f = 0 . We take the functional J (u) := (1, nu) + . The corresponding dual solution is z 1 , so that J (e) = 0. This implies (1, nuh ) + = (1, nu) + = (1, ng ) . recovering the well-known global conservation property of the scheme. 7.1.3 Evaluation of the error estimates

To evaluate the error estimates (14) or (22), one may solve the corresponding perturbed dual problem numerically by the same method as used in computing uh , yielding an approximation zh Vh to the exact dual solution z . However, the use of the same meshes for computing primal and dual solution is by no means obligatory. In fact, in the case of dominant transport it may be advisable to compute the dual solution on a dierent mesh; see [47] for examples. Then, the weights K can be determined numerically in dierent ways: 1. We may take zh = Ih z Vh as the nodal interpolation of z and use the local interpolation properties of nite elements to obtain
3 K = h K z Ih z K 1 cI hK 2 z K,

with an interpolation constant cI 0.1 1 . Here, 2 z is the tensor of second derivatives of z . Then, approximation by second-order dierence quotients of the computed discrete dual solution zh Vh yields K cI |2 h zh (xK )| , xK being the center point of K . 2. Computation of a discrete dual solution zh Vh in a richer space Vh Vh (e.g., on a ner mesh or by higher-order elements) and setting
3 K h K zh Ih zh K

(23)

(24)

where Ih zh Vh denotes the generic nodal interpolation. 3. Interpolation of the discrete dual solution zh Vh by higher order poly(2) nomials on certain cell-patches, e.g., biquadratic interpolation Ih zh :
3 K h K Ih zh zh (2) K

(25)

77

Analogous approximations can be used for the weights K . Option (2) is quite expensive and rarely used. Since we normally do not want to spend more time in evaluating the error estimate than for solving the primal problem, we recommend option (1) or (3). Notice that option (3) does not involve an interpolation constant which needs to be specied. The computational results reported in [14] indicate that the use of biquadratic interpolation on patches of four quadrilaterals is more accurate than using the nite dierence approximation (23). 7.2 Strategies for mesh adaptation

We use the notation introduced above: u is the solution of the variational problem posed on a 2-dimensional domain , uh is its piecewise linear (or bilinear) nite element approximation. Further, e = uuh is the discretization error and J () a linear error functional for measuring e. We suppose that there is an a posteriori error estimate of the form |J (e)| := h4 K K (uh ) K (z ),
K Th

(26)

with the cell residuals K (uh ) and weights K (z ) . Accordingly, we dene the local error indicators K := h4 K K (uh ) K (z ). The mesh design strategies are oriented towards a prescribed tolerance T OL for the error quantity J (e) and the number of mesh cells N which measures the complexity of the computational model. Usually the admissible complexity is constrained by some maximum value Nmax . There are various strategies for organizing a mesh adaptation process on the basis of the a posteriori error estimate (26). Error balancing strategy: Cycle through the mesh and equilibrate the local error indicators, K T OL N T OL. (27)

This process requires iteration with respect to the number of cells N . Fixed fraction strategy: Order cells according to the size of K and rene a certain percentage (say 30%) of cells with largest K (or those which make up 30% of the estimate value ) and coarsen those cells with smallest K . By this strategy, we may achieve a prescribed rate of increase of N (or keep it constant as may be desirable in nonstationary computations).

78

 Mesh optimization strategy: Use the representation :=

K Th

h4 K K (uh ) K (z )

h(x)2 A(x) dx

(28)

for generating a formula for an optimal mesh-size distribution hopt (x). We want to discuss the strategy for deriving an optimal mesh-size distribution in more detail. As a side-product, we will also obtain the justication of the error equilibration strategy. Let Nmax and T OL be prescribed. We assume that for T OL 0, the cell residuals and the weights approach certain limits, K (uh ) hK
3/2 5/2

K (z ) hK

n[uh ]

K K

z Ih z

(xK ) |D 2 u(xK )|,

(xK ) |D 2 z (xK )|.

These properties can be proven on uniformly rened meshes by exploiting super-convergence eects, but still need theoretical justication on locally rened meshes. This suggests to assume that := h(x)2 A(x) dx,

N=
K Th

2 h2 K hK

h(x)2 dx,

(29)

with the weighting function A(x) = (x) (x). Now, let us consider the mesh optimization problem min!, N Nmax . Applying the usual Lagrange approach yields the necessary optimality conditions d dt (h + t)2 A dx + ( + t) (h + t)2 dx Nmax = 0,
t=0

for any variations and . From this, we infer that 2h(x)A(x) 2h(x)3 = 0, Hence, we obtain h(x) = 1 A(x) and 1/2
1/4

h(x)2 dx Nmax = 0.

= h4 A 1 , W :=

A(x)1/2 dx = Nmax ,

A(x)1/2 dx.

This gives us a formula for the optimal mesh-size distribution: = W Nmax

2

hopt (x) = 79

W Nmax

1/2

A(x)1/4 .

(30)

In an analogous way, we can also treat the adjoint optimization problem N min!, T OL . We note that even for a rather singular error functional J () the quantity W is bounded, e.g., J (e) = e(0) A(x) |x|3 W =

|x|3/2 dx < .

Open Problem 7.1: Make the mesh optimization strategy rigorous, i.e., prove the proposed convergence of cell weights and residuals under (local) mesh renement. This could be accomplished by proving that for piecewise linear or d-linear approximation, there holds lim h0 {max |2 h uh |} c(u),

where 2 h uh is a suitable second-order dierence quotient. 7.2.1 Computational tests

(I) The diusion model problem: We begin with the model diusion problem (8) posed on the rectangular domain = (1, 1) (1, 3) with slit at (0, 0). In the presence of a reentrant corner, here a slit, with angle = 2 , the solution involves a corner singularity. It can be written in the form u = r 1/2 + u , with r being the distance to the corner point and u H 2 (). We want to illustrate how the singularity introduced by the weights interacts with the pollution eect caused by the slit singularity. Let the goal be the accurate computation of the a derivative value J (u) = 1 u(P ) at the point P = (0.75, 2.25). In this case the dual solution z behaves like |2 z (x)| d(x)3 + r (x)3/2 , where d(x) and r (x) are the distance functions with respect to the points P and (0, 0) , respectively. Notice that in this case, the dual solution does not 1 exist in the sense of H0 () , such that for practical use, we have to regularize the functional J (u) = 1 u(P ) appropriately. It follows that |1 e(P )| cI
3 h4 K K (uh ) dK + rK K Th 3/2

(31)

Equilibrating the local error indicators yields K and, consequently, N=

K Th 2 h2 K hK =

T OL h4 K 3 dK N N T OL

h2 K dK

3/2

T OL N

1/2

1/2 K Th

h2 K dK

3/2

N T OL

1/2

80

This implies that Nopt T OL1 which is better than what could be achieved on a uniformly rened mesh. In fact, the global energy-error estimate leads to a mesh eciency like J (e) N 1/2 , i.e., Nopt T OL2 . This predicted asymptotic behavior is well conrmed by the results of our computational test shown in Figures 30 and 31 (for more details, we refer to [14]).

Figure 30: Rened meshes with about 5, 000 cells for computing 1 u(P ) using the
weighted error estimate weight (middle) and the energy error estimate E (right); from Backes [4].

"ERROR" "ETA_weight"

0.1

"ETA_weight" "ETA_E"

0.01 e_x(-2.,.5) e_x(-2.,.5) 0.001 0.0001 1000 10000 Number of elements N

0.01

0.001

0.0001

1e-05 1000 10000 Number of elements N 100000

Figure 31: Test results on the slit domain obtained by the weighted error estimator weight: comparison with the true error (left) and comparison of eciency with that of the energy error estimator E (right); from Backes [4].

(II) The transport model problem: Next, we consider the model problem (19) on the unit square = (0, 1) (0, 1) R2 with the right-hand side f 0, the (constant) transport coecient = (1, 0.5)T , and the inow data g (x, 0) = 0, 81 g (0, y ) = 1 .

The quantity to be computed is part of the outow as indicated in Figure 32: J (u) :=

nu ds .

The mesh renement is organized according to the xed fraction strategy described above. In Table 3, we show results for this test computation. The corresponding meshes and the primal as well as the dual solution are presented in Figure 32. Notice that there is no mesh renement enforced along the upper line of discontinuity of the dual solution since here the residual of the primal solution is almost zero.

Figure 32: Conguration and grids of the test computation for the model transport problem (19): primal solution (left) and dual solution (right) on an adaptively rened mesh.

Table 3: Convergence results of the test computation for the model transport problem (19).

Level 0 1 2 3 4 5 6 7

N 256 310 634 964 1315 1540 2050 2128

J (e) 2.01e-2 1.82e-2 1.09e-2 7.02e-3 6.25e-3 5.37e-3 4.21e-3 4.11e-3

/J (e) 2.38e-2 1.18 1.96e-2 1.08 1.21e-2 1.11 8.23e-3 1.17 7.88e-3 1.26 6.94e-3 1.29 5.37e-3 1.27 5.21e-3 1.27

7.3

A general paradigm for a posteriori error estimation

The approach to residualbased error estimation described above for the linear model problem can be extended to general nonlinear systems. We outline the underlying concept in an abstract setting following the general paradigm introduced by Johnson, et al. [26]. 82

Let V be a Hilbert space with inner product (, ) and corresponding norm , A(; ) a continuous semi-linear form and F () a linear form dened on V . We seek a solution u V to the abstract variational problem A(u; ) = F () V. (32) This problem is approximated by a Galerkin method using a sequence of nite dimensional subspaces Vh V parameterized by a discretization parameter h. The discrete problems seek uh Vh satisfying A(uh ; h ) = F (h ) h Vh . (33) The key feature of this approximation is the Galerkin orthogonality which in this nonlinear case reads as A(u; h ) A(uh ; h ) = 0, By elementary calculus, there holds
1

h Vh .

(34)

A(u; ) A(uh ; ) =

A (su + (1 s)uh ; e, ) ds,

with A (v ; , ) denoting the tangent form of A(; ) at some v V . This leads us to introduce the bilinear form
1

L(u, uh; , ) :=
0

A (su + (1 s)uh ; , ) ds ,

which depends on the solutions u as well as uh . Then, denoting the error by e = u uh , there holds
1

L(u, uh ; e, h ) =
0

A (su + (1 s)uh ; e, h ) ds h Vh .

= A(u; h ) A(uh ; h ) = 0 ,

Suppose that the quantity J (u) has to be computed, where J () is a linear functional dened on V . For representing the error J (e) , we use the solution z V of the dual problem L(u, uh ; , z ) = J () V. (35) Assuming that this problem is solvable and using the Galerkin orthogonality (34), we obtain the error representation J (e) = L(u, uh; e, z zh ) = F (z zh ) A(uh ; z zh ), (36)

with any approximation zh Vh . Since the bilinear form L(u, uh ; , ) contains the unknown solution u in its coecient, the evaluation of (36) requires approximation. The simplest way is to replace u by uh , yielding a perturbed dual problem L(uh , uh ; , z ) = J () V. (37) 83

We remark that the bilinear form L(uh , uh ; , ) used in (37) is identical to the tangent form A (uh ; , ) . Controlling the eect of this perturbation on the accuracy of the resulting error estimate may be a delicate task and depends strongly on the particular problem under consideration. Our own experience with several dierent types of problems (including the Navier-Stokes equations) indicates that this problem is less critical as long as the continuous solution is stable. The crucial problem is the numerical computation of the perturbed dual solution z by solving a discretized dual problem L(uh , uh ; , z h ) = J () Vh . (38)

This then results in a practically useful error estimate J (e) (uh ).

Open Problem 7.2: Analyze the eect of the error introduced by the linearization steps (37) and (38) on the quality of the a posteriori error bound and design a reliable strategy for controlling this error. 7.3.1 The nested solution approach

For solving the nonlinear problem (32) by the adaptive Galerkin nite element method (33), we employ the following iterative scheme. Starting from a coarse initial mesh T0 , a hierarchy of rened meshes Ti , i 1 , and corresponding nite element spaces Vi is generated by a nested solution process. (0) Initialization i = 0 : Start on coarse mesh T0 with v0 V0 . (1) Defect correction iteration: For i 1, start with vi
(0) (0)

= vi1 Vi .
(j )

(2) Iteration step: Evaluate the defect (di , ) := F () A(vi ; ),

(j )

Vi .

(v (j ) ; , ) to the derivative A (v (j ) ; , ) Choose a suitable approximation A i i (with good stability and solubility properties) and solve the correction equation vi Vi :
(j )

(v (j ) ; v (j ) , ) = (d(j ) , ) Vi . A i i i

For this, Krylov-space or multigrid iterations are employed using the hierarchy (j +1) (j ) (j ) of already constructed meshes {Ti , ..., T0 }. Then, update vi = vi + i vi ( i (0, 1] a relaxation parameter), set j = j + 1 and go back to (2). This process is repeated until a limit vi Vi , is reached with a certain required accuracy. (3) Error estimation: Solve the (linearized) discrete dual problem zi Vi : A (vi ; , zi) = J () Vi , 84

and evaluate the a posteriori error estimate |J (ei )| (vi ). For controlling the reliability of this bound, i.e. the accuracy in the determination of the dual solution z , one may check whether zi zi1 is suciently small; if this is not the case, additional global mesh renement is advisable. If (vi ) T OL or Ni Nmax , then stop. Otherwise cell-wise mesh adaptation yields the new mesh Ti+1 . Then, set i = i + 1 and go back to (1). This nested solution process is employed in the application presented below. Notice that the derivation of the a posteriori error estimate (3) involves only the solution of linearized problems. Hence, the whole error estimation may amount only to a relatively small fraction of the total cost for the solution process. 7.4 Application to the Navier-Stokes equations

The results in this section are collected from Becker [7, 9]; see also [12]. We consider the stationary Navier-Stokes equations v + v v + p = 0, v = 0 , (39)

in a bounded domain R2 , with boundary conditions as described in Section 2, v |rigid = 0, v |in = v in , n v pn|out = 0. As an example, we consider the ow around the cross section of a cylinder in a channel shown in Figure 33. This is part of a set of benchmark problems discussed in Sch afer/Turek [81].

Figure 33: Conguration of the benchmark ow around a cylinder

85

Quantities of physical interest are, for example, pressure drop: drag coecient: lift coecient: Jp (v, p) = p(af ront ) p(aback ), 2 n (v, p)ex ds, Jdrag (v, p) = 2 U D S 2 Jlif t (v, p) = 2 n (v, p)ey ds, U D S

where S is the surface of the cylinder, D its diameter, ex and ey the cartesian the reference velocity, and (v, p) = 1 (v + v T ) + pI the unit vectors, U 2 stress force acting on S . In our example, the Reynolds number is Re = 2 D/ = 20 , such that the ow is stationary. For evaluating the drag and U lift coecients, one may use another representation obtained by the Stokes formula, e.g., for the drag: 2 Jdrag := 2 U D 2 n (v, p)ex ds = 2 U D S { (v, p)e x + (v, p)e x } dx,

where e x is an extension of ex to the interior of with support along S ; see Giles, et al. [28] and Becker [9]. This representation in terms of a domain integral is more robust and accurate than the original one involving a contour integral. The discretization is by the nite element Galerkin method using the conforming Q1 /Q1 Stokes element described in Section 3 with least-squares pressure stabilization and streamline diusion stabilization for the transport. In order to incorporate this scheme in the abstract framework described above, we rewrite it in a more compact form. To this end, we introduce the Hilbert-spaces V := HL of pairs u := {v, p} and their discrete analogues Vh := HhLh of pairs uh := {vh , ph } . Accordingly the Navier-Stokes equations can be written in vector form as follows: Lu := v + v v + p f = . v 0

Further, for u := {v, p} and = {, } , we dene the semi-linear form A(u; ) := (v, ) + (v v, ) (p, ) + (v, ), and the linear functional F () := (f, ) . Then, the stationary version of the variational formulation (8) is written in the following compact form: Find in , 0)T , such that u V + (vh A(u; ) = F () V. Using the weighted L2 -bilinear form (v, w )h :=
K Th

(40)

K (v, w )K ,

86

the stabilized nite element approximation reads as follows: Find uh Vh + (v in , 0)T such that A(uh ; h ) + (Luh , Sh )h = (F, h ) + (F, Sh )h where the stabilization operator S is dened by S := + v + 0 , h Vh , (41)

with the parameter specied by (36). This formulation contains the pressure and transport stabilization as described in Section 3. The question is now how to construct a mesh as economical as possible on which the quantities Jp (v, p) , Jdrag (v, p) and Jlif t (v, p) can be computed to the required accuracy, say, of 1% . The a priori design of such a mesh is a dicult task as will be demonstrated by the results of numerical tests below; Table 34 shows a collection of possible a priori meshes.

Figure 34: Examples of meshes designed for the benchmark problem ow around
a cylinder; the rst three meshes on the left, Grid 1, Grid 2, and Grid 3, are coarse initial meshes which are to be uniformly rened.

Now, we will discuss the use of a posteriori techniques for constructing economical meshes. We denote the discretization error for the pressure by ep := p ph and that for the velocity by ev := v vh . By standard arguments relying on the coerciveness properties of the Fr echet derivative of the operator L , one derives the following energy-norm a posteriori error estimate ev + ep cI cS (h2 K + K ) R(uh ))
K Th 2 K

+ + ...
1/2

(42) ,

+ vh

2 K

+ hK n[vh ] 87

2 K

with the residual R(uh ) := vh vh vh ph . In this estimate the ... stand for additional terms representing the errors in approximating the inow data and the curved boundary S ; they can be expected to be small compared to the other residual terms and are usually neglected. In this estimate the interpolation constant cI can be determined and is of moderate size cI 0.2 . The most critical point is the stability constant cS which is completely unknown. It is related to the constant in the coerciveness estimate of the tangent form A (v ; , ) of A(; ) taken at the solution v , w + q cS sup
V

A (v ; z, ) +

where z = {w, q } and = {, } . In order to use this error bound for mesh-size control, we have set it to cS = 1 . The error estimate (42) is not appropriate for controlling the error in local quantities like drag and lift since it measures the residual uniformly over the whole computational domain. One way of introducing more a priori information into the mesh renement process based on (42) is to start from an initial mesh which is already rened towards the contour S . Alternatively, one may also use (on heuristic grounds) additional weighting factors which enforce stronger mesh renement in the neighborhood of S . The resulting global error indicator reads as follows: ev + ep cI cS K (h2 K + K ) R(uh ))
K Th 2 K 1/2

(43) + ... ,

+ vh

2 K

+ hK n[vh ]

2 K

where the weights K are chosen large along S . Correctly weighted a posteriori error estimates can be obtained following the general line of argument described above. The approximate dual problem seeks z := {w, q } V satisfying A (uh ; , z ) + (L (uh ) , Sz ) = J () V , (44) where A (uh ; , ) and L (uh ) are the tangent form and adjoint tangent operator of A(; ) and L() , respectively. The resulting weighted a posteriori estimate for the error e := u uh becomes |J (e)|
div K K + K K + div K K + ... , K Th

(45)

with the local residual terms and weights dened by K = R(uh ) K , K = w wh K + K vh (w wh ) + (q qh ) 1 K = 2 n[vh ] K , K = w wh K , div K = vh
K, div K = q qh K. K,

88

The dots ... stand again for additional terms measuring the errors in approximating the inow and the curved cylinder boundary. For more details on this aspect, we refer to [14] and Becker [9]. The bounds for the dual solution z = {w, q } are obtained computationally by replacing the unknown solution u in the convection term by its approximation uh and solving the resulting linearized problem on the same mesh. From this approximate dual solution z h , patchwise biquadratic interpolations are taken to approximate z in eval(i) (2) uating the weights K , Ih z h zh z zh . This avoids the occurrence of interpolation constants. Table 4 shows the corresponding results for the pressure drop computed on four dierent types of meshes: (i) Hierarchically rened meshes starting from coarse meshes of type Grid 1 and Grid 2 as shown in Figure 34. (ii) Adapted meshes using the global energy-norm error estimate (42) with enforced renement along the contour S ; see Figure 43. (iii) Adapted meshes using the weighted error estimate (45) for the pressure drop; see Figure 42. These results demonstrate clearly the superiority of the weighted error estimate (45) in computing local quantities. It produces an error of less than 1% already after 6 renement cycles on a mesh with less than 1400 unknowns while the other algorithms use more than 21000 unknowns to achieve the same accuracy (the corresponding values are printed in boldface). Corresponding sequences of meshes generated by the weighted energy error estimate (42) and the energyerror estimate (43) are seen in Figures 35 and 36. Table 5 contains some results of the computation of drag and lift coecients using the corresponding weighted error estimates. The eectivity index is dened by Ief f := (uh )/|J (e)| . Finally, Figure 37 shows plots of the dual solutions occurring in the computation of pressure drop, drag and lift.

89

Table 4: Results of the pressure drop computation (ref. value p = 0.11752016... );

a) upper row: on uniformly rened meshes of type Grid1 and Grid2, b) lower row: on adaptively rened meshes starting from a coarse mesh Grid1; from Becker [7]. Uniform Renement, Grid1 N p 2268 0.109389 8664 0.110513 33840 0.113617 133728 0.115488 531648 0.116486 Uniform Renement, Grid2 L N p 1 1296 0.106318 2 4896 0.112428 3 19008 0.115484 4 74880 0.116651 5 297216 0.117098 Weighted Adaptive Renement L N p 4 650 0.115967 6 1358 0.116732 9 2858 0.117441 11 5510 0.117514 12 8810 0.117527

L 1 2 3 4 5

Adaptive Renement, Grid1 L N p 2 1362 0.105990 4 5334 0.113978 6 21546 0.116915 8 86259 0.117379 10 330930 0.117530

Figure 35: A sequence of rened meshes generated by the (heuristically) weighted

global energy estimate; from Becker [7].

Figure 36: A sequence of rened meshes generated by the weighted error estimate
for the pressure drop; from Becker [7].

90

Table 5: Results of the cylinder ow computations of drag and lift (ref. values cdrag = 5.579535... and clif t = 0.0106189... ) on adaptively rened meshes starting from a coarse mesh of type Grid1; from Becker [9].
L 3 4 5 6 7 8 9 Computation N cdrag 251 5.780186 587 5.637737 1331 5.568844 3953 5.576580 8852 5.578224 16880 5.578451 34472 5.578883 of drag drag 2.0e 1 5.8e 2 1.0e 2 2.5e 3 8.7e 4 6.5e 4 2.1e 4 Ief f 0.5 0.6 1.6 2.0 2.5 1.6 2.0 L 3 4 5 6 7 8 9 Computation N clif t 296 0.007680 764 0.009249 1622 0.009916 4466 0.010144 8624 0.010267 18093 0.010457 34010 0.010524 of lift lif t 2.9e 3 1.4e 3 7.3e 4 5.0e 4 3.8e 5 1.9e 5 1.2e 4 Ief f 5.0 5.0 5.0 2.5 2.0 2.0 1.6

Figure 37: Velocity plots of the dual solution for pressure drop (top), drag (middle),
and lift (bottom); from Becker [9].

7.5

The nonstationary case

The extension of the approach to mesh adaptivity described above to the nonstationary Navier-Stokes equations is presently under development. (I) The traditional method for a posteriori time-step selection is based on the concept of controlling the local truncation error but neglecting the global error accumulation. In its simplest form this strategy uses the condition
n 1 c k 2 (Uk/ 2 3 S n n Uk ) T OL,

(46)

n n where Uk and Uk/ 2 are the solutions computed from the preceding approxn1 imation U at tn1 by a second-order scheme (e.g. the Crank-Nicolson scheme) with time-step sizes k and k/2 , respectively. For a more detailed description of techniques of this type, we refer to Turek [97].

91

(II) The extension of the residual-based error control described above to nonstationary problems is based on a time discretization which has also the features of a Galerkin method. These are for example the socalled continuous or discontinuous Galerkin methods of polynomial degree r 0 (cG(r) or dG(r) methods). The lowest-order examples are the dG(0) method which (in the autonomous case) is equivalent to the backward Euler scheme, the dG(1) method which is similar to an implicit Runge-Kutta scheme of third order, and the cG(1) method which can be interpreted as a variant of the Crank-Nicolson scheme. In particular, the dG(1) method is attractive for solving the nonstationary Navier-Stokes problem because of its superior accuracy (compared to the dG(0) method). The result of error estimation using a duality argument is an a posteriori error estimate of the form (see [55] and Hartmann [39])
n n Uk

u(, tn )

3 m km m dt Uk + ... m=1

(47)

m1 m 1 m where dt Uk = km (Uk Uk ) are the time-dierence quotients of the computed solution and m are weighting factors obtained by solving a backward in time space-time dual problem. The dots ... refer to residual terms of the spatial discretization. The main problem with this approach is its huge m n computational work; in a nonlinear problem the forward solution {Uk }m=0 enters the linearized dual problem as coecient and needs therefore to be stored over the whole time interval. Moreover, in this way error control can be achieved only at single times tn or for the time-averaged error. Controlling the error uniformly in time requires (theoretically) to solve a dual problem at each discrete time level resulting in prohibitively high cost. The economical realization of this concept for computing nonstationary ows involving global error control is still an open problem.

Open Problem 7.3: Devise a strategy for adapting the stabilization parameters K simultaneously with the mesh size hK on the basis of the a posteriori error estimate (45). Open Problem 7.4: Derive an a posteriori error estimate of the form (45) for the full space-time discretization of the Navier-Stokes equations and device a strategy for simultaneous adaptation of mesh sizes hK and time steps kn .

92

Extension to weakly compressible ows

In this last section, we discuss the extension of the computational methodology described above to certain compressible ows. The ows of interest are those in which density changes are induced by temperature gradients resulting for example from heat release by chemical reactions. Such weakly compressible ows are characterized by low-Mach-number speed and hydrodynamically incompressible behavior. Here, the dominant problem is that of sti velocitypressure coupling while shocks or large pressure gradients do not develop. We recall the system of conservation equations for mass, momentum and energy, in the case of a stationary ow: [v ] = 0 , 1 vI ] + ptot = f , v v [v + 3 cp v T [T ] = h . (1) (2) (3)

Here, again v is the velocity, the density, ptot the (total) pressure and T the temperature of the uid occupying a two- or three-dimensional region . The dynamic viscosity > 0 , the heat capacity cp > 0 , the heat conductivity , the external volume force f and the heat source h are given. Since we only consider low-speed ows, the inuence of stress and hydrodynamic pressure in the energy equation can be neglected. In general, f as well as h implicitly depend on the temperature T and on further quantities describing the release of heat for example through chemical reactions. Here, we will simply consider the heat source h as given. The coupling between pressure and density is assumed as that of a perfect gas, ptot = RT, (4)

where R is the gas constant. As mentioned above, we consider hydrodynamically incompressible ows. Accordingly, the pressure is split into two parts, ptot (x, t) = p (t) + phyd (x, t), namely the spatial mean value p := ||1 ptot (x, t) dx,

and the hydrodynamic pressure phyd (x, t) . In a weakly compressible ow, the pressure variation due to hydrodynamic mechanisms is assumed to be small compared to the mean value of the total pressure, |phyd | |p |, which is determined by thermodynamic eects. Accordingly, we call Pth (t) = p (t) the thermodynamic pressure. In the low-Mach-number approximation the hydrodynamic pressure occurs in the momentum equation v v [v + 1 vI ] + phyd = f, 3 93 (5)

while the pressure-density coupling in the equation of state (4) is expressed in terms of the thermodynamic pressure Pth . (6) RT In many applications, this set of equations has to be supplemented by further conservation equations for species concentrations and complicated nonlinear source terms representing the chemical reactions. Here, we restrict ourselves to the simple case of low-Mach-number ow, where temperature variations are induced by outer source terms. The thermodynamic pressure Pth (t) is supposed to be determined by a priori considerations; for more details, see Braack [17] and also [18]. = Open Problem 8.1: Estimate the error caused by neglecting stress and hydrodynamic pressure in the energy equation. Prove corresponding error bounds for the low-Mach-number approximation in terms of the Mach number. Since in the above approximation the density occurs as a secondary variable determined by the temperature through the equation of state, it appears natural to use the pressure p := phyd together with the velocity v and the temperature T as primal variables in the computational model. We use the equation of state to rewrite the continuity equation as an equation for velocity and temperature: v T 1 v T = 0 . (7) Furthermore, introducing the modied pressure p := phyd 1 v , the mo3 mentum equation can be written as v v [v ] + p = f , while the energy equation keeps the form cp v T [T ] = h . (9)

(8)

The temperature-dependent functions = (T ) and cp = cp (T ) are usually given in terms of polynomial ts from data bases. The density is expressed by the algebraic relation (6) in terms of the temperature. The system is closed by imposing appropriate boundary conditions, v|rigid = 0, v|in = v in , n v + pn|out = 0, , T| = T (10)

where again rigid , in , out are the rigid part, the inow part and the outow part of the boundary , respectively. For questions of well-posedness of this type of problem, we refer to the relevant literature, e.g., Feistauer [27] and Lions [60]. The starting point for a nite element discretization of problem (7), (8), (9), and (6) is again its variational formulation. To formulate this, we introduce the natural function spaces as already used above, L L2 (), H H 1 ()d , 94 R H 1 ().

for the pressure p L , the velocity v H , and the temperature T R . For a compact notation, we set V := L H R . Prescribed Dirichlet data v and can be included by seeking the weak solutions in appropriate sub-manifolds, T p L, vv + H, +R. T T (11) (12) (13)

Then, the triple u := {p, v, T } is determined by the variational equations (v, ) (T 1 v T, ) = 0, L, (v v, ) + (v, ) (p, ) = (f, ) H, (cp v T, ) + (T, ) = (h, ) R.

In the following analysis, we consider for simplicity only the case of pure Dirichlet boundary conditions. In this case the pressure is determined only modulo constants and the corresponding solution space is L = L2 0 () . Now, the nite element discretization replaces the (innite dimensional) function spaces L , H , and R by nite dimensional discrete spaces denoted by Lh , Hh , and Rh . Here, we think of nite element spaces based for example on conforming Q1 approximation for all physical quantities. The corresponding h + Rh are determined discrete solutions ph Lh , vh v h + Hh , and Th T through the system
1 (vh , h ) (Th vh Th , h ) = 0, h Lh , (14) (vh vh , h ) + (vh , h ) (ph , h ) = (f, h ) h Hh , (15) (cp vh Th , h ) + (Th , h ) = (h, h ) h Rh , (16)

with coecients = (Th ) and cp = cp (Th ) . The compact formulation of the system (14)-(16) makes use of the semi-linear form A(u; ) := (v, ) (T 1 v T, ) + (v v, ) + (v, ) (p, ) + (cp v T, ) + (T, ), and the linear form F () = (f, ) + (h, ), dened for triples u = {p, v, T }, = {, , } V . With this notation, the problem reads as follows: Find u u + V , such that A(u; ) = F () V. (17) where u represents Dirichlet boundary data for all components. The corresponding discrete problem reads: Find uh u h + Vh , such that A(uh ; h ) = F (h ) h Vh . (18) In general this system is unstable and needs stabilization with respect to the sti velocity-pressure coupling as well as the transport terms. 95

8.1

Least-squares stabilization

The stabilization is introduced into the system (18) by using pressure stabilization and streamline diusion as discussed above in the context of the incompressible Navier-Stokes equations. The corresponding stabilization terms are listed below: Pressure stabilization: sp h (uh , h ) =
K Th p rh (uh , h )

K ( vh vh [vh ] + ph , h )K , K (f, h )K .

=
K Th

Streamline diusion for the velocities: sv h (uh , h ) =

K Th v rh (uh , h )

K (v h vh [vh ] + ph , v h h )K , K (f, v h h )K .

=
K Th

Streamline diusion for the temperature: sT h (uh , h , h ) =

K Th T rh (uh , h , h ) = K Th

K (cp v h Th [Th ], cp v h h )K , K (h, cp v h h )K .

Here, v h is a suitable approximation to the current velocity eld vh , taken for example from a preceding iteration step. We denote the sum over these h-dependent stabilization terms by sh (, ) and rh () , respectively,
v T sh (uh , ) := sp h (uh , ) + sh (uh , ) + sh (uh , ), p v T (u h , ). (u h , ) + r h rh (uh , ) := rh (uh , ) + rh

Then, with Ah (; ) := A(; ) + sh (, ) and Fh () := F () + rh () , the discrete equations can be written in compact form Ah (uh ; ) = Fh (h ) Vh . (19)

In order to ensure symmetry for the resulting stabilized system, K should be taken equal to K . The stability and consistence of this formulation can be analyzed by similar techniques as used in the case of the incompressible Navier-Stokes equations; see Braack [17]. One obtains the following condition for the parameters K : |v h | K = 2 + hK hK
1

|cp v h | K = 2 + hK hK 96

(20)

Open Problem 8.2: Derive a formula for the stabilization parameters K and K which leads to a robust scheme on general meshes with large aspect ratio h . 8.2 Computational approach

For solving the model for weakly compressible ow introduced above, we want to use the methodology discussed for incompressible ows. (i) Explicit defect correction coupling: The simplest way to use an incompressible solver for computing weakly com1 l 1 l 1 pressible ows is by a defect correction iteration. The step {pl h , vh , Th } l l { pl h , vh , Th } of this scheme proceeds as follows:

1. The nonlinear coecients (density, transport vectors, etc.) are frozen at 1 l 1 l 1 l l l { pl h , vh , Th } . Corresponding corrections {ph , vh , Th } Vh are determined by solving the linearized system:
l 1 (vh , h ) = (dl p , h ),

l 1 l (vh vh , h )

l + (vh , h ) (pl h , h ) l 1 l l (cp vh Th , h ) + (Th , h )

= =

1 (dl v , h ) 1 (dl t , h )

h Hh , h Rh ,

h Lh ,

l 1 1 l 1 l 1 1 l 1 where dl are the defects of the iteration {pl p , dv , and dT h , vh , Th } . For the sake of robustness, the pressure and transport stabilization described above has also to be applied to this problem.

2. The new solution vector is obtained by

i1 pl + l pl h = ph h, l i1 l vh = vh + l vh , l i1 l Th = Th + l Th ,

with some relaxation parameter l (0, 1] , and the density is updated acl cording to l h = Pth /(RTh ) . 3. The iteration is continued until some stopping criterion is satised. In each step of this iteration a linearized Navier-Stokes problem supplemented by a heat transfer equation is to be solved. This may be accomplished by using the methods described above for the incompressible Navier-Stokes equations. Hence, the incompressible solver is used for preconditioning the defect correction iteration for solving the full system (18). However, this simple defect correction process may converge very slowly in the case of large temperature gradients (e.g., caused by strong heat release in chemical reactions). This lack of robustness can be cured by making the iteration more implicit. (ii) Semi-implicit defect correction coupling: In order to achieve better control on the variation of temperature, one may use the following more implicit iteration: 97

1. The nonlinear coecients (density, transport vectors, etc.) are frozen at 1 l 1 l 1 l l l { pl h , vh , Th } . Corresponding corrections {ph , vh , Th } Vh are then determined by solving the linearized system:
l l 1 1 l 1 l 1 (vh , h ) + ((Th ) vh Th , ) = (dl p , h ) h Lh ,

l 1 l l l 1 (vh vh , h ) + (vh , h ) (pl h , h ) = (dv , h ) h Hh , l 1 l l 1 (cp vh Th , h ) + (Th , h ) = (dl t , h ) h Rh , l 1 1 l 1 l 1 1 l 1 where dl are the defects of the iteration {pl p , dv , and dT h , vh , Th } . Again, the pressure and transport stabilization described above has to be applied.

2. The new solution vector is obtained by

i1 pl + l pl h = ph h, l i1 l vh = vh + l vh , l i1 l Th = Th + l Th ,

with some relaxation parameter l (0, 1] , and the density is updated acl cording to l h = Pth /(RTh ) . 3. The iteration is continued until some stopping criterion is satised. This solution method has been used in Braack [17] for the simulation of lowMach-number combustion processes; see also [10] and [18]. 8.3 The algebraic system

In each substep of the defect correction iterations described above, we have to (p) (v) (T ) solve linear problems for the coecients xj = {xj , xj , xj } including the components for pressure, velocity and temperature in the basis representations
N N (p) xj j j =1 N (v) xj j j =1

ph =

vh =

Th =
j =1

xj j .

(T )

The system sub-matrices corresponding to the dierent components are obtained from the coupled system by taking rst test functions of the form h = {h , 0, 0}: 1 v Ah (uh ; {h , 0, 0}) = (vh , h ) (T h h Th , h ) +
K Th

K (v h vh (vh ) + ph , h )K .

Analogously, taking the test functions h = {0, h , 0}, we obtain the equation for the velocity components, Ah (uh ; {0, h , 0}) = (v h vh , h ) + (vh , h ) (ph , h ) +
K Th

K (v h vh (vh ) + ph , v h h )K , 98

and by taking the test functions h = (0, 0, h ) the equation for the temperature component, Ah (uh ; {0, 0, h}) = (cp v h Th , h ) + (Th , h ) +
K Th

K (cp v h Th [Th ], cp v h h ) .

Ordering the unknowns in a physically block-wise sense, i.e., marching through the set of nodal points and attaching to each node the corresponding submatrix containing the unknowns of all physical quantities, we obtain nodal matrices Aij of the form Bpp Bpv BpT Aij = Bvp Bvv BvT , BT p BT v BT T where the indices p, v, T indicate the corresponding contributions. Looking at the equations, we see that almost all physical components are coupled with each other; only the pressure does not appear in the temperature equation, i.e., BT p = 0 . Several of the other couplings are of minor importance and may be neglected in building an approximating nodal matrix Aij to Aij . One could think of a complete decoupling of the ow variables {p, v } from the temperature T (or other state variables describing for example chemical reactions) resulting in an approximation of the form Bpp Bpv 0 Aij = Bvp Bvv 0 . 0 0 BT T However, such a simplication is not appropriate in computing processes in which the temperature has a signicant inuence on the ow eld and vice versa. For example, in combustion problems, density variations are mainly caused by changes of the temperature. A detailed discussion of this issue can be found in Braack [17] and in [18].

99

8.4

An example of chemically reactive ow

We close this section by presenting some results from Braack [17] on computations for low-Mach-number ows with chemical reactions. The conguration considered is the model of a methane burner with a complicated geometry and using a sophisticated reaction mechanism. A stoichiometric mixture of methane CH4 and air O2 /N2 ows from the bottom of the burner through a sample of slots of uniform width 2 mm and three dierent heights (varying from 14 mm to 11 mm). The columns have a uniform width of 1.5 mm . The inow velocity is uniformly 0.2 m/s . The Reynolds number in this model is about Re = 90. The geometry is shown in Figure 38.

burnt gas

flamefront

14 mm 13 mm

1,5

2,0 mm

unburnt gas
11 mm

uniform inflow of CH4 / O2 / N2

Figure 38: Geometry of a methane burner; from Braack [17]. Due to the heating of the slots, the ow accelerates up to approximately 1 m/s. Since this is higher than the ame velocity of a stoichiometric methane ame the ame front is located above the slots. For lower inow velocities, the ame moves downstream into the slots and extinguishes as a result of the heat loss by the cold walls. If the solution is assumed to be spatially periodic, it is sucient to restrict the computational domain to only three slots, as shown in Figure 39. The boundaries at the left and right hand of are symmetry boundary conditions. The walls of the slots are described by Dirichlet conditions for the temperature and Neumann conditions for the species. The calculation on the coarsest mesh (with 1344 cells) uses a time-stepping procedure to provide a physically correct starting value. Then, on the ner meshes the stationary xed-point defect correction iteration converges. In order to obtain ignition, the temperature for the initial solution is set to 2000 K at the points above the slots. The reaction mechanism is that of Smooke [80] with 15 species and 84 elementary reactions (42 bidirectional), supplemented by two further species, N and NO , and 4 additional reactions to describe their formation. 100

The solution is obtained on an adaptively rened mesh with renement criterion based on the linear functional J (u) = ||1 T dx ,

in order to capture the temperature distribution accurately. The nest mesh is shown in Figure 39; we see local mesh renement at the ame front and below the slots where the velocity eld changes. The mesh is automatically adapted and no hand-tting on the basis of a priori knowledge of the solution is necessary to nd the appropriate balance of the mesh-size distribution. The CPU time required for such a simulation with about 5,000 cells ( 100,000 unknowns) is approximately 6 hours on a Pentium II (233 Mhz) when the initial guess on the coarse grid with approximately 1300 cells is given. The computed pressure and the main velocity component are shown in Figure 39. Due to the strong heat release the ow accelerates by a factor of 10 at the outow of the slots. At the walls of the slots, Dirichlet conditions for the temperature are imposed, varying linearly from 298 K at the bottom up to 393 K , 453 K and 513 K for the three dierent walls. This leads to a higher outow velocity at the longer slot compared to the shorter ones. Therefore, the lift-o of the ame is substantially higher at the longer slot, leading to the common Bunsen cone formed by two neighboring longer slots.

Figure 39: Results of the methane burner simulation: velocity and temperature
proles (left and middle), nest mesh with 5, 000 cells (right); from Braack [17].

101

References
[1] M. Ainsworth and J. T. Oden (1997), A posteriori error estimation in nite element analysis, Comput. Meth. Appl. Mech. Engrg., 142, pp. 1-88. [2] T. Apel and M. Dobrowolski (1992), Anisotropic interpolation with applications to the nite element method, Computing, 47, pp. 277293. [3] T. Apel (1999), Anisotropic Finite Elements: Local Estimates and Applications, Habilitation Thesis, Preprint 99-03, SFB 393, University of Magdeburg. [4] E. Backes (1997), Gewichtete a posteriori Fehleranalyse bei der adaptiven Finite-Elemente-Methode: Ein Vergleich zwischen Residuen- und Bank-WeiserSch atzer, Diploma Thesis, Institute of Applied Mathematics, University of Heidelberg. [5] W. Bangerth and G. Kanschat (1999), deal.II Homepage, Technical Reference, Release 1.0, SFB 359, University of Heidelberg, http://gaia.iwr.uniheidelberg.de/deal/. [6] R. E. Bank, B. Weiser, and H. Yserentant (1990), A class of iterative methods for solving saddle point problems, Numer. Math., 56, 645666. [7] R. Becker (1995), An Adaptive Finite Element Method for the Incompressible Navier-Stokes Equations on Time-Dependent Domains, Doctor Thesis, Preprint 95-44, SFB 359, Nov. 1995, University of Heidelberg. [8] R. Becker (1998), An adaptive nite element method for the Stokes equations including control of the iteration error, ENUMATH95, Paris, Sept. 18-22, 1995, in: Proc. ENUMATH97 (H. G. Bock, et al., eds.), pp. 609-620, World Scientic Publisher, Singapore. [9] R. Becker (1998): Weighted error estimators for nite element approximations of the incompressible Navier-Stokes equations, Preprint 98-20, SFB 359, University of Heidelberg, submitted for publication. [10] R. Becker, M. Braack, R. Rannacher, and C. Waguet (1998), Fast and reliable solution of the Navier-Stokes equations including chemistry, Proc. Conf. Applied Mathematics for Industrial Flow Problems (AMIF), San Feliu de Guixols (Spain), Oct. 1-3. 1998, Preprint 99-03 (SFB 359), University of Heidelberg, January 1999, to appear in Computing and Visualization in Sciences. [11] R. Becker, C. Johnson, and R. Rannacher (1995), Adaptive error control for multigrid nite element methods, Computing, 55, pp. 271-288. [12] R. Becker and R. Rannacher (1995), Weighted a posteriori error control in FE methods, ENUMATH95, Paris, Sept. 18-22, 1995, Proc. ENUMATH97 (H. G. Bock, et al., eds.), pp. 621637, World Scientic Publishers, Singapore.

102

[13] R. Becker and R. Rannacher (1994), Finite element solution of the incompressible Navier-Stokes equations on anisotropically rened meshes, Proc. Workshop Fast Solvers for Flow Problems, Kiel, Jan. 14-16, 1994 (W. Hackbusch and G. Wittum, eds.), pp. 5261, NNFM, Vol. 49, Vieweg, Braunschweig. [14] R. Becker and R. Rannacher (1996), A feed-back approach to error control in nite element methods: Basic analysis and examples, East-West J. Numer. Math., 4, pp. 237-264. [15] H. Blum (1990), Asymptotic Error Expansion and Defect Correction in the Finite Element Method, Habilitation Thesis, University of Heidelberg. [16] M. Boman (1995), A Model Study of Hydrodynamic Stability, Masters Thesis, Chalmers University of Technology, Gothenburg, Sweden. [17] M. Braack (1998), An Adaptive Finite Element Method for Reactive Flow Problems, Doctor Thesis, Institute of Applied Mathematics, University of Heidelberg. [18] M. Braack and R. Rannacher (1999), Adaptive nite element methods for lowMach-number ows with chemical reactions, Lecture Series 1999-03, 30th Computational Fluid Dynamics, (H. Deconinck, ed.), von Karman Institute for Fluid Dynamics, Belgium. [19] S. C. Brenner and R. L. Scott (1994), The Mathematical Theory of Finite Element Methods, Springer, Berlin-Heidelberg-New York. [20] F. Brezzi and M. Fortin (1991), Mixed and Hybrid Finite Element Methods, Springer, Berlin-Heidelberg-New York. [21] F. Brezzi and J. Pitk aranta (1984), On the stabilization of nite element approximations of the Stokes equations, Proc. Workshop Ecient Solution of Elliptic Systems (W. Hackbusch, ed.), Vieweg, Braunschweig. [22] M. O. Bristeau, R. Glowinski, and J. Periaux (1987), Numerical methods for the Navier-Stokes equations: Applications to the simulation of compressible and incompressible viscous ows, Comput. Phys. Reports, 6, pp. 73187. [23] C. M. Chen and V. Thom ee (1985), The lumped mass nite element method for a parabolic problem, J. Austral. Math. Soc., Ser. B, 26, pp. 329-354. [24] A. J. Chorin (1968), Numerical solution of the Navier-Stokes equations, Math. Comp., 22, pp. 745-762. [25] W. E and J. P. Liu (1998), Projection method I: Convergence and numerical boundary layers, SIAM J. Num. Anal., to appear. [26] K. Eriksson, D. Estep, P. Hansbo, and C. Johnson (1995), Introduction to adaptive methods for dierential equations, Acta Numerica 1995 (A. Iserles, ed.), pp. 105-158, Cambridge University Press.

103

[27] M. Feistauer (1993), Mathematical Methods in Fluid Dynamics, Longman Scientic&Technical, England. [28] M. L. M. Giles, M. Larson, and E. S uli (1998), Adaptive error control for nite element approximations of the lift and drag coecients in viscous ow, SIAM J. Numer. Anal, to appear. [29] V. Girault and P.-A. Raviart (1986), Finite Element Methods for the NavierStokes Equations, Springer, Heidelberg. [30] R. Glowinski (1985), Viscous ow simulations by nite element methods and related numerical techniques, in Progress in Supercomputing in Computational Fluid Dynamics (E.M. Murman and S.S. Abarbanel, eds.), pp. 173-210, Birkh auser, Boston. [31] R. Glowinski and J. Periaux (1987), Numerical methods for nonlinear problems in uid dynamics, Proc. Int. Seminar on Scientic Supercomputers, Paris, North-Holland, Amsterdam. [32] P. M. Gresho (1990), On the theory of semi-implicit projection methods for viscous incompressible ow and its implementation via a nite element method that also introduces a nearly consistent mass matrix, Part 1: Theory, Int. J. Numer. Meth. Fluids, 11, pp. 587620. [33] P. M. Gresho and S. T. Chan (1990), On the theory of semi-implicit projection methods for viscous incompressible ow and its implementation via a nite element method that also introduces a nearly consistent mass matrix, Part 2: Implementation, Int. J. Numer. Meth. Fluids, 11, pp. 621660. [34] P. M. Gresho (1991), Incompressible uid dynamics: Some fundamental formulation issues, Ann. Rev. Fluid Mech., 23, pp. 413-453. [35] P. M. Gresho and R. L. Sani (1998), Incompressible Flow and the Finite Element Method, John Wiley, Chichester. [36] W. Hackbusch (1985), Multi-Grid Methods and Applications, Springer, Heidelberg-Berlin. [37] P. Hanbo and C. Johnson (1995), Streamline diusion nite element methods for uid ow, in Finite Element Methods for Compressible and Incompressible Flow, Selected Topics from Previous VKI Lecture Series, (H. Deconinck, ed.), von Karman Institute for Fluid Dynamics, Belgium. [38] J. Harig (1991), Eine robuste und eziente Finite-Elemente Methode zur L osung der inkompressiblen 3D-Navier-Stokes Gleichungen auf Vektorrechnern, Doctor Thesis, Institute of Applied Mathematics, University of Heidelberg. [39] R. Hartmann (1998), A posteriori Fehlersch atzung und adaptive Schrittweiten- und Ortsgittersteuerung bei Galerkin-Verfahren f ur die W armeleitungsgleichung, Diploma Thesis, Institute of Applied Mathematics, University of Heidelberg.

104

[40] J. G. Heywood (1980), The Navier-Stokes equations: On the existence, regularity and decay of solutions, Indiana Univ. Math. J., 29, pp. 639-681. [41] J. G. Heywood and R. Rannacher (1982), Finite element approximation of the nonstationary Navier-Stokes Problem. I. Regularity of solutions and second order error estimates for spatial discretization, SIAM J. Numer. Anal., 19, pp. 275-311. [42] J. G. Heywood and R. Rannacher (1986), Finite element approximation of the nonstationary Navier-Stokes Problem. II. Stability of solutions and error estimates uniform in time, SIAM J. Numer. Anal., 23, pp. 750-777. [43] J. G. Heywood and R. Rannacher (1988), Finite element approximation of the nonstationary Navier-Stokes Problem. III. Smoothing property and higher order error estimates for spatial discretization, SIAM J. Numer. Anal., 25, pp. 489-512. [44] J. G. Heywood and R. Rannacher (1990), Finite element approximation of the nonstationary Navier-Stokes Problem. IV. Error analysis for second-order time discretization, SIAM J. Numer. Anal., 27, pp. 353-384. [45] J. G. Heywood and R. Rannacher (1986), An analysis of stability concepts for the Navier-Stokes equations, J. Reine Angew. Math., 372, pp. 133. [46] J. G. Heywood, R. Rannacher, and S. Turek (1992), Articial boundaries and ux and pressure conditions for the incompressible Navier-Stokes equations, Int. J. Numer. Math. Fluids, 22, pp. 325352. [47] P. Houston, R. Rannacher, and E. S uli (1999), A posteriori error analysis for stabilised nite element approximation of transport problems, Report No. 99/04, Oxford University Computing Laboratory, Oxford, England OX1 3QD, submitted for publication. [48] T. J. R. Hughes and A. N. Brooks (1982), Streamline upwind/PetrovGalerkin formulations for convection dominated ows with particular emphasis on the incompressible Navier-Stokes equation, Comput. Meth. Appl. Mech. Engrg., 32, pp. 199-259. [49] T. J. R. Hughes, L. P. Franca, and M. Balestra (1986), A new nite element formulation for computational uid mechanics: V. Circumventing the BabuskaBrezzi condition: A stable Petrov-Galerkin formulation of the Stokes problem accommodating equal order interpolation, Comput. Meth. Appl. Mech. Engrg., 59, pp. 85-99. [50] C. Johnson (1987), Numerical Solution of Partial Dierential Equations by the Finite Element Method, Cambridge University Press, Cambridge-Lund. [51] C. Johnson (1989), The streamline diusion nite element method for compressible and incompressible uid ow, Proc. Finite Element Method in Fluids VII, Huntsville.

105

[52] C. Johnson (1993), A new paradigm for adaptive nite element methods, Proc. MAFELAP93 Conf., Brunel Univ., Uxbridge, UK, John Wiley, Chichester. [53] C. Johnson, S. Larsson, V. Thom ee, and L. B. Wahlbin (1987), Error estimates for spatially discrete approximations of semilinear parabolic equations with nonsmooth initial data, Math. Comp., 49, pp. 331-357. [54] C. Johnson, R. Rannacher, and M. Boman (1995), Numerics and hydrodynamic stability: Towards error control in CFD, SIAM J. Numer. Anal., 32, pp. 10581079. [55] C. Johnson, R. Rannacher, and M. Boman (1995), On transition to turbulence and error control in CFD, Preprint 95-06, SFB 359, University of Heidelberg. [56] C. Johnson and R. Rannacher (1994), On error control in CFD, Proc. Int. Workshop Numerical Methods for the Navier-Stokes Equations, Heidelberg, Oct. 25-28, 1993 (F.-K. Hebeker, R. Rannacher, and G. Wittum, eds.), pp. 25-28, NNFM, Vol. 47, Vieweg, Braunschweig. [57] S. Kracmar and J. Neustupa (1994), Modelling of ows of a viscous incompressible uid through a channel by means of variational inequalities, Z. Angew. Math. Mech., 74, pp. T637 - T639. [58] O. Ladyshenskaya (1969), The Mathematical Theory of Viscous Incompressible Flow, Gordon and Breach, London. [59] M. Landahl (1980), A note on an algebraic instability of viscous parallel shear ows, J. Fluid Mech., 98, p. 243. [60] P.-L. Lions (1998), Mathematical Topics in Fluid Mechanics, Vol. 2, Compressible Models, Clarendon Press, Oxford. [61] M. Luskin and R. Rannacher (1982), On the smoothing property of the CrankNicolson scheme, Appl. Anal., 14, pp. 117-135. [62] M. Luskin and R. Rannacher (1981), On the smoothing property of the Galerkin method for parabolic equations, SIAM J. Numer. Anal., 19, pp. 93-113. [63] K. W. Morton (1996), Numerical Solution of Convection-Diusion Problems, Applied Mathematics and Mathematical Computation, Vol. 12, Chapman&Hall. [64] S. M uller-Urbaniak (1993), Eine Analyse des Zwischenschritt- -Verfahrens zur L osung der instation aren Navier-Stokes-Gleichungen, Doctor Thesis, Preprint 9401, SFB 359, University of Heidelberg. [65] S. M uller, A. Prohl, R. Rannacher, and S. Turek (1994), Implicit timediscretization of the nonstationary incompressible Navier-Stokes equations, Proc. Workshop Fast Solvers for Flow Problems, Kiel, Jan. 14-16, 1994 (W. Hackbusch and G. Wittum, eds.), pp. 175191, NNFM, Vol. 49, Vieweg, Braunschweig.

106

[66] H. Oswald (1999), L osung der instation aren Navier-Stokes-Gleichungen auf Parallelrechnern, Doctor Thesis, Institute of Applied Mathematics, University of Heidelberg. [67] O. Pironneau (1982), On the transport-diusion algorithm and its applications to the Navier-Stokes equations, Numer. Math., 38, pp. 309332. [68] O. Pironneau (1983), Finite Element Methods for Fluids, John Wiley, Chichester. [69] A. Prohl (1995), Projektions- und Quasi-Kompressibilit atsmethoden zur L osung der inkompressiblen Navier-Stokes-Gleichungen, Doctor Thesis, Institute of Applied Mathematics, University of Heidelberg. [70] A. Prohl (1998), On Quasi-Compressibility Methods and Projection Methods for Solving the Incompressible Navier-Stokes Equations, Teubner, Stuttgart. [71] R. Rannacher (1984), Finite element solution of diusion problems with irregular data, Numer. Math., 43, pp. 309-327. [72] R. Rannacher (1998), Numerical analysis of nonstationary uid ow (a survey), in Applications of Mathematics in Industry and Technology (V.C.Bo and H.Neunzert, eds.), pp. 34-53, B.G. Teubner, Stuttgart. [73] R. Rannacher (1992), On Chorins projection method for the incompressible Navier-Stokes equations, Proc. Oberwolfach Conf. Navier-Stokes Equations: Theory and Numerical Methods, September 1991 (J.G. Heywood, et al., eds.), pp. 167183, LNM, Vol. 1530, Springer, Heidelberg. [74] R. Rannacher (1993), On the numerical solution of the incompressible NavierStokes equations, Survey lecture at the annual GAMM-Conference 1992, Leipzig, March 24-27, Z. Angew. Math. Mech., 73, pp. 203-216. [75] R. Rannacher (1998), A posteriori error estimation in least-squares stabilized nite element schemes, Comput. Meth. Appl. Mech. Engrg., 166, pp. 99114. [76] R. Rannacher (1999), Error control in nite element computations, Proc. NATO-Summer School Error Control and Adaptivity in Scientic Computing Antalya (Turkey), Aug. 1998 (H. Bulgak and C. Zenger, eds.), pp. 247278, NATO Science Series, Series C, Vol. 536, Kluwer, Dordrecht. [77] R. Rannacher and S. Turek (1992), A simple nonconforming quadrilateral Stokes element, Numer. Meth. Part. Di. Equ., 8, pp. 97-111. [78] R. Rautmann (1983), On optimal regularity of Navier-Stokes solutions at time t = 0 , Math. Z., 184, pp. 141-149. [79] H. Reichert and G. Wittum (1993), On the construction of robust smoothers for incompressible ow problems, Proc. Workshop Numerical Methods for the Navier-Stokes Equations, Heidelberg, Oct. 25-28, 1993 (F.K. Hebeker, R. Rannacher, G. Wittum, eds.), pp. 207216, Vieweg, Braunschweig.

107

[80] M. D. Smooke (1991), Numerical Modeling of Laminar Diusion Flames, Progress in Astronautics and Aeronautics, 135. [81] M. Sch afer and S. Turek (1996), The benchmark problem ow around a cylinde, Flow Simulation with High-Performance Computers (E.H. Hirschel, ed.), pp. 547566, NNFM, Vol. 52, Vieweg, Braunschweig. [82] F. Schieweck (1992), A parallel multigrid algorithm for solving the NavierStokes equations on a transputer system, Impact Comput. Sci. Engrg., 5, pp. 345-378. [83] P. Schreiber and S. Turek (1993), An ecient nite element solver for the nonstationary incompressible Navier-Stokes equations in two and three dimensions, Proc. Workshop Numerical Methods for the Navier-Stokes Equations, Heidelberg, Oct. 25-28, 1993 (F.K. Hebeker, R. Rannacher, G. Wittum, eds.), pp. 133-144, Vieweg, Braunschweig. [84] J. Shen (1992), On error estimates of projection methods for the Navier-Stokes Equations: First order schemes, SIAM J. Numer. Anal., 29, pp. 57-77. [85] J. Shen (1994), Remarks on the pressure error estimates for the projection methods, Numer. Math., 67, pp. 513-520. [86] J. Shen (1996), On error estimates of the projection methods for the NavierStokes Equations: 2nd order schemes, Math. Comp., 65, pp. 10391065. [87] G. Strang (1968), On the construction and comparison of dierence schemes, SIAM J. Numer. Anal., 5, 506-517. [88] R. Temam (1987), Navier-Stokes Equations. Theory and numerical analysis, 2nd ed., North Holland, Amsterdam. [89] V. Thom ee (1997), Galerkin Finite Element Methods for Parabolic Problems, Springer, Berlin-Heidelberg-New York. [90] L. Tobiska and F. Schieweck (1989), A nonconforming nite element method of up-stream type applied to the stationary Navier-Stokes equation, M2 AN, 23, pp. 627-647. [91] L. Tobiska and R. Verf urth (1996), Analysis of a streamline diusion nite element method for the Stokes and Navier-Stokes equations, SIAM J. Numer. Anal., 33, pp. 107127. [92] L. N. Trefethen, A. E. Trefethen, S. C. Reddy, and T. A. Driscoll (1992), A new direction in hydrodynamical stability: Beyond eigenvalues, Tech. Report CTC92TR115 12/92, Cornell Theory Center, Cornell University. [93] S. Turek (1994), Tools for simulating nonstationary incompressible ow via discretely divergence-free nite element models, Int. J. Numer. Meth. Fluids, 18, pp. 71-105.

108

[94] S. Turek (1996), A comparative study of some time-stepping techniques for the incompressible Navier-Stokes equations, Int. J. Numer. Meth. Fluids, 22, pp. 9871011. [95] S. Turek (1997), On discrete projection methods for the incompressible NavierStokes equations: An algorithmic approach, Comput. Methods Appl. Mech. Engrg., 143, pp. 271288. [96] S. Turek (1998), FEATFLOW: Finite Element Software for the Incompressible Navier-Stokes Equations, User Manual, Release 1.3, SFB 359, University of Heidelberg, URL: http://gaia.iwr.uni-heidelberg.de/featow/. [97] S. Turek (1999), Ecient Solvers for Incompressible Flow Problems, NNCSE, Vol. 6, Springer, Berlin-Heidelberg-New York. [98] S. Turek, et al. (1999), Virtual Album of Fluid Motion, Preprint, Institute of Applied Mathematics, University of Heidelberg, in preparation. [99] M. Van Dyke (1982), An Album of Fluid Motion, The Parabolic Press, Stanford. [100] J. Van Kan (1986), A second-order accurate pressure-correction scheme for viscous incompressible ow, J. Sci. Stat. Comp., 7, pp. 870-891. [101] S. P. Vanka (1986), Blockimplicit multigrid solution of NavierStokes equations in primitive variables, J. Comp. Phys., 65, pp. 138158. [102] R. Verf urth (1996), A Review of A Posteriori Error Estimation and Adaptive Mesh-Renement Techniques, Wiley/Teubner, New York-Stuttgart, 1996. [103] C. Waguet (1999) Adaptive Finite Element Computation of Chemical Flow Reactors, Doctor Thesis, SFB 359, University of Heidelberg, in preparation. [104] P. Wesseling (1992), An Introduction to Multigrid Methods, J. Wiley, Chichester. [105] G. Wittum (1990), The use of fast solvers in computational uid dynamics, in Proc. Eighth GAMM-Conference on Numerical Methods in Fluid Mechanics (P. Wesseling, ed.), pp. 574581. LNFM, Vol. 29, Vieweg, Braunschweig. [106] O. C. Zienkiewicz and J. Z. Zhu (1987), A simple error estimator and adaptive procedure for practical engineering analysis, Int. J. Numer. Meth. Engrg., 24, pp. 337-357. [107] Guohui Zhou (1997), How accurate is the streamline diusion nite element method?, Math. Comp., 66, pp. 3144.

109