Navier-Stokes Equations

Navier-Stokes Equations:

Greetings everyone, Today I would like to offer an interesting puzzle in the fields of mathematics, for discussion and attempted solving. A few days ago, I came across the website of the Clay Mathematics Institute. There, I found an interesting page listing seven so-called "Millenium Problems." These are seven extremely challenging puzzles, each with a prize fund of $1,000,000:

http://www.claymath.org/millennium/

So far, only the Poincare Conjecture has been solved. Remaining are the Berch and Swinnerton-Dyer Conjecture, the Hodge Conjecture, the Navier-Stokes Equations, P vs NP, the Riemann Hypothesis, and the Yang-Mills theory. Of these, I have chosen to work on the Navier-Stokes equations:

http://www.claymath.org/millennium/Navier-Stokes_Equations/

The Navier-Stokes equations, a group of hypotheses meant to explain both turbulence and breezes, have yet to be fully understood since their creation in the 19th century. The official problem description can be found here:

http://www.claymath.org/millennium/Navier-Stokes_Equations/navierstokes.pdf

In short, the problem can be done in four different ways. All of them require a proof which will be held up to analysis for two years before announced as winning. Below, I have posted the four possible proofs:

(A) Existence and smoothness of Navier–Stokes solutions on R3. Take ν > 0 and n = 3. Let u◦(x) be any smooth, divergence-free vector field satisfying (4). Take f(x,t) to be identically zero. Then there exist smooth functions p(x,t),ui(x,t) on R3 × [0, ∞) that satisfy (1), (2), (3), (6), (7).

(B) Existence and smoothness of Navier–Stokes solutions in R3/Z3. Take ν > 0 and n = 3. Let u◦(x) be any smooth, divergence-free vector field satisfying (8); we take f (x, t) to be identically zero. Then there exist smooth functions p(x, t), ui (x, t) on R3 × [0, ∞) that satisfy (1), (2), (3), (10), (11).

(C) Breakdown of Navier–Stokes solutions on R3. Take ν > 0 and n = 3. Then there exist a smooth, divergence-free vector field u◦(x) on R3 and a smooth f (x, t) on R3 × [0, ∞), satisfying (4), (5), for which there exist no solutions (p, u) of (1), (2), (3), (6), (7) on R3 × [0, ∞).

(D) Breakdown of Navier–Stokes Solutions on R3/Z3. Take ν > 0 and n = 3. Then there exist a smooth, divergence-free vector field u◦(x) on R3 and a smooth f (x, t) on R3 × [0, ∞), satisfying (8), (9), for which there exist no solutions (p, u) of (1), (2), (3), (10), (11) on R3 × [0, ∞).

That pretty much sums it up. For further information, I have posted a number of links below:

Video Lecture:

http://claymath.msri.org/navierstokes.mov

Simplified Explanations:

http://en.wikipedia.org/wiki/Navier–Stokes_equations

http://www.efunda.com/formulae/fluids/navier_stokes.cfm

PDF Lesson:

http://depts.washington.edu/chemcrs/bulkdisk/chem520A_aut05/notes_Week_05_Lecture_07.pdf

Wikipedia:

In physics, the Navier–Stokes equations, named after Claude-Louis Navier and George Gabriel Stokes, describe the motion of fluidsubstances. These equations arise from applying Newton's second law to fluid motion, together with the assumption that the fluid stress is the sum of a diffusing viscous term (proportional to the gradient of velocity), plus a pressure term.

The equations are useful because they describe the physics of many things of academic and economic interest. They may be used tomodel the weather, ocean currents, water flow in a pipe and air flow around a wing. The Navier–Stokes equations in their full and simplified forms help with the design of aircraft and cars, the study of blood flow, the design of power stations, the analysis of pollution, and many other things. Coupled with Maxwell's equations they can be used to model and study magnetohydrodynamics.

The Navier–Stokes equations are also of great interest in a purely mathematical sense. Somewhat surprisingly, given their wide range of practical uses, mathematicians have not yet proven that in three dimensions solutions always exist (existence), or that if they do exist, then they do not contain any singularity (smoothness). These are called the Navier–Stokes existence and smoothness problems. The Clay Mathematics Institute has called this one of the seven most important open problems in mathematics and has offered a US$1,000,000 prize for a solution or a counter-example.^[1]

The Navier–Stokes equations dictate not position but rather velocity. A solution of the Navier–Stokes equations is called a velocity field or flow field, which is a description of the velocity of the fluid at a given point in space and time. Once the velocity field is solved for, other quantities of interest (such as flow rate or drag force) may be found. This is different from what one normally sees in classical mechanics, where solutions are typically trajectories of position of a particle or deflection of a continuum. Studying velocity instead of position makes more sense for a fluid; however for visualization purposes one can compute various trajectories.

Nonlinearity

The Navier–Stokes equations are nonlinear partial differential equations in almost every real situation. In some cases, such as one-dimensional flow and Stokes flow (or creeping flow), the equations can be simplified to linear equations. The nonlinearity makes most problems difficult or impossible to solve and is the main contributor to the turbulence that the equations model.

The nonlinearity is due to convective acceleration, which is an acceleration associated with the change in velocity over position. Hence, any convective flow, whether turbulent or not, will involve nonlinearity. An example of convective but laminar (nonturbulent) flow would be the passage of a viscous fluid (for example, oil) through a small converging nozzle. Such flows, whether exactly solvable or not, can often be thoroughly studied and understood.

[edit]Turbulence

Turbulence is the time dependent chaotic behavior seen in many fluid flows. It is generally believed that it is due to the inertia of the fluid as a whole: the culmination of time dependent and convective acceleration; hence flows where inertial effects are small tend to be laminar (the Reynolds number quantifies how much the flow is affected by inertia). It is believed, though not known with certainty, that the Navier–Stokes equations describe turbulence properly.

The numerical solution of the Navier–Stokes equations for turbulent flow is extremely difficult, and due to the significantly different mixing-length scales that are involved in turbulent flow, the stable solution of this requires such a fine mesh resolution that the computational time becomes significantly infeasible for calculation (see Direct numerical simulation). Attempts to solve turbulent flow using a laminar solver typically result in a time-unsteady solution, which fails to converge appropriately. To counter this, time-averaged equations such as the Reynolds-averaged Navier–Stokes equations (RANS), supplemented with turbulence models, are used in practical computational fluid dynamics (CFD) applications when modeling turbulent flows. Some models include the Spalart-Allmaras, k-ω (k-omega), k-ε (k-epsilon), and SST models which add a variety of additional equations to bring closure to the RANS equations. Another technique for solving numerically the Navier–Stokes equation is the Large eddy simulation (LES). This approach is computationally more expensive than the RANS method (in time and computer memory), but produces better results since the larger turbulent scales are explicitly resolved.

[edit]Applicability

Together with supplemental equations (for example, conservation of mass) and well formulated boundary conditions, the Navier–Stokes equations seem to model fluid motion accurately; even turbulent flows seem (on average) to agree with real world observations.

The Navier–Stokes equations assume that the fluid being studied is a continuum (it is infinitely divisible and not composed of particles such as atoms or molecules), and is not moving at relativistic velocities. At very small scales or under extreme conditions, real fluids made out of discrete molecules will produce results different from the continuous fluids modeled by the Navier–Stokes equations. Depending on the Knudsen number of the problem, statistical mechanics or possibly even molecular dynamics may be a more appropriate approach.

Another limitation is simply the complicated nature of the equations. Time tested formulations exist for common fluid families, but the application of the Navier–Stokes equations to less common families tends to result in very complicated formulations which are an area of current research. For this reason, these equations are usually written for Newtonian fluids. Studying such fluids is "simple" because the viscosity model ends up being linear; truly general models for the flow of other kinds of fluids (such as blood) do not, as of 2011, exist.

The derivation of the Navier–Stokes equations begins with an application of Newton's second law: conservation of momentum (often alongside mass and energy conservation) being written for an arbitrary portion of the fluid. In an inertial frame of reference, the general form of the equations of fluid motion is:^[2]

where  is the flow velocity, ρ is the fluid density, p is the pressure,  is the (deviatoric) stress tensor, and  represents body forces (per unit volume) acting on the fluid and  is thedel operator. This is a statement of the conservation of momentum in a fluid and it is an application of Newton's second law to a continuum; in fact this equation is applicable to any non-relativistic continuum and is known as the Cauchy momentum equation.

This equation is often written using the material derivative Dv/Dt, making it more apparent that this is a statement of Newton's second law:

The left side of the equation describes acceleration, and may be composed of time dependent or convective effects (also the effects of non-inertial coordinates if present). The right side of the equation is in effect a summation of body forces (such as gravity) and divergence of stress (pressure and shear stress).

[edit]Convective acceleration

An example of convection. Though the flow may be steady (time independent), the fluid decelerates as it moves down the diverging duct (assuming incompressible or subsonic compressible flow), hence there is an acceleration happening over position.

A very significant feature of the Navier–Stokes equations is the presence of convective acceleration: the effect of time independent acceleration of a fluid with respect to space. While individual fluid particles are indeed experiencing time dependent acceleration, the convective acceleration of the flow field is a spatial effect, one example being fluid speeding up in a nozzle. Convective acceleration is represented by the nonlinear quantity:

which may be interpreted either as  or as  with  the tensor derivative of the velocity vector  Both interpretations give the same result, independent of the coordinate system — provided  is interpreted as the covariant derivative.^[3]

[edit]Interpretation as (v·∇)v

The convection term is often written as

where the advection operator  is used. Usually this representation is preferred as it is simpler than the one in terms of the tensor derivative ^[3]

[edit]Interpretation as v·(∇v)

Here  is the tensor derivative of the velocity vector, equal in Cartesian coordinates to the component by component gradient. The convection term may, by a vector calculus identity, be expressed without a tensor derivative:^[4][5]

The form has use in irrotational flow, where the curl of the velocity (called vorticity)  is equal to zero.

Regardless of what kind of fluid is being dealt with, convective acceleration is a nonlinear effect. Convective acceleration is present in most flows (exceptions include one-dimensional incompressible flow), but its dynamic effect is disregarded in creeping flow (also called Stokes flow) .

[edit]Stresses

The effect of stress in the fluid is represented by the  and  terms; these are gradients of surface forces, analogous to stresses in a solid.  is called the pressure gradient and arises from the isotropic part of the stress tensor. This part is given by normal stresses that turn up in almost all situations, dynamic or not. The anisotropic part of the stress tensor gives rise to , which conventionally describes viscous forces; for incompressible flow, this is only a shear effect. Thus,  is the deviatoric stress tensor, and the stress tensor is equal to:^[6]

where  is the 3×3 identity matrix. Interestingly, only the gradient of pressure matters, not the pressure itself. The effect of the pressure gradient is that fluid flows from high pressure to low pressure.

The stress terms p and  are yet unknown, so the general form of the equations of motion is not usable to solve problems. Besides the equations of motion—Newton's second law—a force model is needed relating the stresses to the fluid motion.^[7] For this reason, assumptions on the specific behavior of a fluid are made (based on natural observations) and applied in order to specify the stresses in terms of the other flow variables, such as velocity and density.

The Navier–Stokes equations result from the following assumptions on the deviatoric stress tensor :^[8]

• the deviatoric stress vanishes for a fluid at rest, and – by Galilean invariance – also does not depend directly on the flow velocity itself, but only on spatial derivatives of the flow velocity
• in the Navier–Stokes equations, the deviatoric stress is expressed as the product of the tensor gradient  of the flow velocity with a viscosity tensor , i.e. : 
• the fluid is assumed to be isotropic, as valid for gases and simple liquids, and consequently  is an isotropic tensor; furthermore, since the deviatoric stress tensor is symmetric, it turns out that it can be expressed in terms of two scalar dynamic viscosities μ and μ”:  where  is the rate-of-strain tensor and  is the rate of expansion of the flow
• the deviatoric stress tensor has zero trace, so for a three-dimensional flow 2μ + 3μ” = 0

As a result, in the Navier–Stokes equations the deviatoric stress tensor has the following form:^[8]

with the quantity between brackets the non-isotropic part of the rate-of-strain tensor  The dynamic viscosity μ does not need to be constant – in general it depends on conditions like temperature and pressure, and in turbulence modelling the concept of eddy viscosity is used to approximate the average deviatoric stress.

The pressure p is modelled by use of an equation of state.^[9] For the special case of an incompressible flow, the pressure constrains the flow in such a way that the volume of fluid elements is constant: isochoric flow resulting in a solenoidal velocity field with ^[10]

[edit]Other forces

The vector field  represents body forces. Typically these consist of only gravity forces, but may include other types(such as electromagnetic forces). In a non-inertial coordinate system, other "forces" such as that associated with rotating coordinates may be inserted.

Often, these forces are so-called conservative forces and may be represented as the gradient of some scalar quantity , with  Gravity in the z direction, for example, is the gradient of − ρgz. Since pressure shows up only as a gradient, this implies that solving a problem without any such body force can be mended to include the body force by using a modified pressure p_m = p − F. The pressure and force terms on the right hand side of the Navier–Stokes equation become

[edit]Other equations

The Navier–Stokes equations are strictly a statement of the conservation of momentum. In order to fully describe fluid flow, more information is needed (how much depends on the assumptions made). This additional information may include boundary data (no-slip, capillary surface, etc.), the conservation of mass, the conservation of energy, and/or an equation of state.

Regardless of the flow assumptions, a statement of the conservation of mass is generally necessary. This is achieved through the mass continuity equation, given in its most general form as:

or, using the substantive derivative:

A simplification of the resulting flow equations is obtained when considering an incompressible flow of a Newtonian fluid. The assumption of incompressibility rules out the possibility ofsound or shock waves to occur; so this simplification is invalid if these phenomena are important. The incompressible flow assumption typically holds well even when dealing with a "compressible" fluid — such as air at room temperature — at low Mach numbers (even when flowing up to about Mach 0.3). Taking the incompressible flow assumption into account and assuming constant viscosity, the Navier–Stokes equations will read, in vector form:^[11]

Here f represents "other" body forces (forces per unit volume), such as gravity or centrifugal force. The shear stress term  becomes the useful quantity  ( is the vector Laplacian) when the fluid is assumed incompressible, homogeneous and Newtonian, where  is the (constant) dynamic viscosity.^[12]

It's well worth observing the meaning of each term (compare to the Cauchy momentum equation):

Note that only the convective terms are nonlinear for incompressible Newtonian flow. The convective acceleration is an acceleration caused by a (possibly steady) change in velocity over position, for example the speeding up of fluid entering a converging nozzle. Though individual fluid particles are being accelerated and thus are under unsteady motion, the flow field (a velocity distribution) will not necessarily be time dependent.

Another important observation is that the viscosity is represented by the vector Laplacian of the velocity field (interpreted here as the difference between the velocity at a point and the mean velocity in a small volume around). This implies that Newtonian viscosity is diffusion of momentum, this works in much the same way as the diffusion of heat seen in the heat equation (which also involves the Laplacian).

If temperature effects are also neglected, the only "other" equation (apart from initial/boundary conditions) needed is the mass continuity equation. Under the incompressible assumption, density is a constant and it follows that the equation will simplify to:

This is more specifically a statement of the conservation of volume (see divergence).

These equations are commonly used in 3 coordinates systems: Cartesian, cylindrical, and spherical. While the Cartesian equations seem to follow directly from the vector equation above, the vector form of the Navier–Stokes equation involves some tensor calculus which means that writing it in other coordinate systems is not as simple as doing so for scalar equations (such as the heat equation).

[edit]Cartesian coordinates

Writing the vector equation explicitly,

Note that gravity has been accounted for as a body force, and the values of g_x,g_y,g_z will depend on the orientation of gravity with respect to the chosen set of coordinates.

The continuity equation reads:

When the flow is at steady-state, ρ does not change with respect to time. The continuity equation is reduced to:

When the flow is incompressible, ρ is constant and does not change with respect to space. The continuity equation is reduced to:

The velocity components (the dependent variables to be solved for) are typically named u, v, w. This system of four equations comprises the most commonly used and studied form. Though comparatively more compact than other representations, this is still a nonlinear system of partial differential equations for which solutions are difficult to obtain.

[edit]Cylindrical coordinates

A change of variables on the Cartesian equations will yield^[11] the following momentum equations for r, ϕ, and z:

The gravity components will generally not be constants, however for most applications either the coordinates are chosen so that the gravity components are constant or else it is assumed that gravity is counteracted by a pressure field (for example, flow in horizontal pipe is treated normally without gravity and without a vertical pressure gradient). The continuity equation is:

This cylindrical representation of the incompressible Navier–Stokes equations is the second most commonly seen (the first being Cartesian above). Cylindrical coordinates are chosen to take advantage of symmetry, so that a velocity component can disappear. A very common case is axisymmetric flow with the assumption of no tangential velocity (u_ϕ = 0), and the remaining quantities are independent of ϕ:

[edit]Spherical coordinates

In spherical coordinates, the r, ϕ, and θ momentum equations are^[11] (note the convention used: θ is polar angle, or colatitude,^[13] 0 ≤ θ ≤ π):

Mass continuity will read:

These equations could be (slightly) compacted by, for example, factoring 1 / r² from the viscous terms. However, doing so would undesirably alter the structure of the Laplacian and other quantities.

[edit]Stream function formulation

Taking the curl of the Navier–Stokes equation results in the elimination of pressure. This is especially easy to see if 2D Cartesian flow is assumed (w = 0 and no dependence of anything on z), where the equations reduce to:

Differentiating the first with respect to y, the second with respect to x and subtracting the resulting equations will eliminate pressure and any conservative force. Defining the stream function ψ through

results in mass continuity being unconditionally satisfied (given the stream function is continuous), and then incompressible Newtonian 2D momentum and mass conservation degrade into one equation:

where  is the (2D) biharmonic operator and ν is the kinematic viscosity, . We can also express this compactly using the Jacobian determinant:

This single equation together with appropriate boundary conditions describes 2D fluid flow, taking only kinematic viscosity as a parameter. Note that the equation for creeping flow results when the left side is assumed zero.

In axisymmetric flow another stream function formulation, called the Stokes stream function, can be used to describe the velocity components of an incompressible flow with one scalarfunction.

[edit]Pressure-free velocity formulation

The incompressible Navier–Stokes equation is a differential algebraic equation, having the inconvenient feature that there is no explicit mechanism for advancing the pressure in time. Consequently, much effort has been expended to eliminate the pressure from all or part of the computational process. The stream function formulation above eliminates the pressure (in 2D) at the expense of introducing higher derivatives and elimination of the velocity, which is the primary variable of interest.

The incompressible Navier–Stokes equation is composite, the sum of two orthogonal equations,

,

,

where Π^S and Π^I are solenoidal and irrotational projection operators satisfying Π^S + Π^I = 1 and  and  are the nonconservative and conservative parts of the body force. This result follows from the Helmholtz Theorem (also known as the fundamental theorem of vector calculus). The first equation is a pressureless governing equation for the velocity, while the second equation for the pressure is a functional of the velocity and is related to the pressure Poisson equation.

The explicit functional form of the projection operator in 3D is found from the Helmholtz Theorem

.

with a similar structure in 2D. Thus the governing equation is an integro-differential equation and not convenient for numerical computation.

An equivalent weak or variational form of the equation, proved to produce the same velocity solution as the Navier–Stokes equation,^[14] is given by,

for divergence-free test functions  satisfying appropriate boundary conditions. Here, the projections are accomplished by the orthogonality of the solenoidal and irrotational function spaces. The discrete form of this is imminently suited to finite element computation of divergence-free flow, as we shall see in the next section. There we will be able to address the question, "How does one specify pressure-driven (Poiseuille) problems with a pressureless governing equation?"

The absence of pressure forces from the governing velocity equation demonstrates that the equation is not a dynamic one, but rather a kinematic equation where the divergence-free condition serves the role of a conservation law. This all would seem to refute the frequent statements that the incompressible pressure enforces the divergence-free condition.

[edit]Discrete velocity

With partitioning of the problem domain and defining basis functions on the partitioned domain, the discrete form of the governing equation is,

.

It is desirable to choose basis functions which reflect the essential feature of incompressible flow – the elements must be divergence-free. While the velocity is the variable of interest, the existence of the stream function or vector potential is necessary by the Helmholtz Theorem. Further, to determine fluid flow in the absence of a pressure gradient, one can specify the difference of stream function values across a 2D channel, or the line integral of the tangential component of the vector potential around the channel in 3D, the flow being given byStokes' Theorem. Discussion will be restricted to 2D in the following.

We further restrict discussion to continuous Hermite finite elements which have at least first-derivative degrees-of-freedom. With this, one can draw a large number of candidate triangular and rectangular elements from the plate-bending literature. These elements have derivatives as components of the gradient. In 2D, the gradient and curl of a scalar are clearly orthogonal, given by the expressions,

Adopting continuous plate-bending elements, interchanging the derivative degrees-of-freedom and changing the sign of the appropriate one gives many families of stream function elements.

Taking the curl of the scalar stream function elements gives divergence-free velocity elements.^[15][16] The requirement that the stream function elements be continuous assures that the normal component of the velocity is continuous across element interfaces, all that is necessary for vanishing divergence on these interfaces.

Boundary conditions are simple to apply. The stream function is constant on no-flow surfaces, with no-slip velocity conditions on surfaces. Stream function differences across open channels determine the flow. No boundary conditions are necessary on open boundaries, though consistent values may be used with some problems. These are all Dirichlet conditions.

The algebraic equations to be solved are simple to set up, but of course are non-linear, requiring iteration of the linearized equations. .

Similar considerations apply to three-dimensions, but extension from 2D is not immediate because of the vector nature of the potential, and there exists no simple relation between the gradient and the curl as was the case in 2D.

[edit]Pressure recovery

Recovering pressure from the velocity field is easy. The discrete weak equation for the pressure gradient is,

,

where the test/weight functions are irrotational. Any conforming scalar finite element may be used. However, the pressure gradient field may also be of interest. In this case one can use scalar Hermite elements for the pressure. For the test/weight functions  one would choose the irrotational vector elements obtainied from the gradient of the pressure element.

There are some phenomena that are closely linked with fluid compressibility. One of the obvious examples is sound. Description of such phenomena requires more general presentation of the Navier–Stokes equation that takes into account fluid compressibility. If viscosity is assumed a constant, one additional term appears, as shown here:^[17][18]

where μ^v is the volume viscosity coefficient, also known as bulk viscosity. This additional term disappears for an incompressible fluid, when the divergence of the flow equals zero.

The Navier–Stokes equations, even when written explicitly for specific fluids, are rather generic in nature and their proper application to specific problems can be very diverse. This is partly because there is an enormous variety of problems that may be modeled, ranging from as simple as the distribution of static pressure to as complicated as multiphase flow driven by surface tension.

Generally, application to specific problems begins with some flow assumptions and initial/boundary condition formulation, this may be followed by scale analysis to further simplify the problem. For example, after assuming steady, parallel, one dimensional, nonconvective pressure driven flow between parallel plates, the resulting scaled (dimensionless) boundary value problem is:

Visualization of a) parallel flow and b) radial flow.

The boundary condition is the no slip condition. This problem is easily solved for the flow field:

From this point onward more quantities of interest can be easily obtained, such as viscous drag force or net flow rate.

Difficulties may arise when the problem becomes slightly more complicated. A seemingly modest twist on the parallel flow above would be the radial flow between parallel plates; this involves convection and thus nonlinearity. The velocity field may be represented by a function f(z) that must satisfy:

This ordinary differential equation is what is obtained when the Navier–Stokes equations are written and the flow assumptions applied (additionally, the pressure gradient is solved for). The nonlinear term makes this a very difficult problem to solve analytically (a lengthy implicit solution may be found which involves elliptic integrals and roots of cubic polynomials). Issues with the actual existence of solutions arise for R > 1.41 (approximately; this is not the square root of 2), the parameter R being theReynolds number with appropriately chosen scales. This is an example of flow assumptions losing their applicability, and an example of the difficulty in "high" Reynolds number flows.

Some exact solutions to the Navier–Stokes equations exist. Examples of degenerate cases — with the non-linear terms in the Navier–Stokes equations equal to zero — are Poiseuille flow, Couette flow and the oscillatory Stokes boundary layer. But also more interesting examples, solutions to the full non-linear equations, exist; for example the Taylor–Green vortex.^[19][20][21] Note that the existence of these exact solutions does not imply they are stable: turbulence may develop at higher Reynolds numbers.

A three dimensional steady-state vortex solution

Some of the flow lines along a Hopf fibration.

A nice steady-state example with no singularities comes from considering the flow along the lines of a Hopf fibration. Let r be a constant radius to the inner coil. One set of solutions is given by:^[23]

g = 0

μ = 0

for arbitrary constants A and B. This is a solution in a non-viscous gas (compressible fluid) whose density, velocities and pressure goes to zero far from the origin. (Note this is not a solution to the Clay Millennium problem because that refers to incompressible fluids where ρis a constant.) It is also worth pointing out that the components of the velocity vector are exactly those from the Pythagorean quadrupleparametrization. Other choices of density and pressure are possible with the same velocity field:

Wyld diagrams are bookkeeping graphs that correspond to the Navier–Stokes equations via a perturbation expansion of the fundamental continuum mechanics. Similar to the Feynman diagrams in quantum field theory, these diagrams are an extension of Keldysh's technique for nonequilibrium processes in fluid dynamics. In other words, these diagrams assign graphsto the (often) turbulent phenomena in turbulent fluids by allowing correlated and interacting fluid particles to obey stochastic processes associated to pseudo-random functions inprobability distributions.^[24]

The Navier–Stokes equations are used extensively in video games in order to model a wide variety of natural phenomena. These include simulations of effects such as water, fire, smoke etc. Many of the implementations used are based on the seminal paper "Real-Time Fluid Dynamics for Games"^[25] by J. Stam. More recent implementations based upon this work run on the GPU as opposed to the CPU and achieve a much higher degree of performance.^[26] [27]

• Boltzmann equation
• Churchill–Bernstein equation
• Coandă effect
• Computational fluid dynamics
• Euler equations
• Fokker–Planck equation
• Hagen–Poiseuille flow from the Navier–Stokes equations
• Large eddy simulation
• Mach number
• Multiphase flow
• Navier–Stokes existence and smoothness — one of theMillennium Prize Problems as stated by the Clay Mathematics Institute
• Reynolds transport theorem
• Reynolds-averaged Navier–Stokes equations
• Adhémar Jean Claude Barré de Saint-Venant
• Vlasov equation

1. ^ Millennium Prize Problems, Clay Mathematics Institute, retrieved 2009-04-11
2. ^ Batchelor (1967) pp. 137 & 142.
3. ^ ^a b Emanuel, G. (2001), Analytical fluid dynamics(second ed.), CRC Press, ISBN 0849391148 pp. 6–7.
4. ^ See Batchelor (1967), §3.5, p. 160.
5. ^ Eric W. Weisstein, Convective Derivative, MathWorld, retrieved 2008-05-20
6. ^ Batchelor (1967) p. 142.
7. ^ Feynman, Richard P.; Leighton, Robert B.; Sands, Matthew (1963), The Feynman Lectures on Physics, Reading, Mass.: Addison-Wesley, ISBN 0-201-02116-1, Vol. 1, §9–4 and §12–1.
8. ^ ^a b Batchelor (1967) pp. 142–148.
9. ^ Batchelor (1967) p. 165.
10. ^ Batchelor (1967) p. 75.
11. ^ ^a b c See Acheson (1990).
12. ^ Batchelor (1967) pp. 21 & 147.
13. ^ Eric W. Weisstein (2005-10-26), Spherical Coordinates, MathWorld, retrieved 2008-01-22
14. ^ Temam, Roger (2001), Navier–Stokes Equations, Theory and Numerical Analysis, AMS Chelsea, pp. 107–112
15. ^ Holdeman, J.T. (2010), "A Hermite finite element method for incompressible fluid flow", Int. J. Numer. Meth. FLuids 64 (4): 376–408, Bibcode2010IJNMF..64..376H, doi:10.1002/fld.2154
16. ^ Holdeman, J.T.; Kim, J.W. (2010), "Computation of incompressible thermal flows using Hermite finite elements", Comput. Methods Appl. Mech. Engrg. 199(49–52): 3297–3304, Bibcode2010CMAME.199.3297H,doi:10.1016/j.cma.2010.06.036
17. ^ Landau & Lifshitz (1987) pp. 44–45.
18. ^ Batchelor (1967) pp. 147 & 154.
19. ^ Wang, C.Y. (1991), "Exact solutions of the steady-state Navier–Stokes equations", Annual Review of Fluid Mechanics 23: 159–177, Bibcode1991AnRFM..23..159W,doi:10.1146/annurev.fl.23.010191.001111
20. ^ Landau & Lifshitz (1987) pp. 75–88.
21. ^ Ethier, C.R.; Steinman, D.A. (1994), "Exact fully 3D Navier–Stokes solutions for benchmarking", International Journal for Numerical Methods in Fluids 19 (5): 369–375,Bibcode 1994IJNMF..19..369E,doi:10.1002/fld.1650190502
22. ^ Ladyzhenskaya, O.A. (1969), The Mathematical Theory of viscous Incompressible Flow (2nd ed.), p. preface, xi
23. ^ Kamchatno, A. M (1982), Topological solitons in magnetohydrodynamics
24. ^ McComb, W.D. (2008), Renormalization methods: A guide for beginners, Oxford University Press,ISBN 0199236526 pp. 121–128.
25. ^ Stam, J. (2003), Real-Time Fluid Dynamics for Games
26. ^ Harris, Mark J. (2004), "38", GPUGems - Fast Fluid Dynamics Simulation on the GPU
27. ^ Sander, P.; Tatarchuck, N.; Mitchell, J.L. (2007), "9.6",ShaderX5 - Explicit Early-Z Culling for Efficient Fluid Flow Simulation, pp. 553–564

• Acheson, D. J. (1990), Elementary Fluid Dynamics, Oxford Applied Mathematics and Computing Science Series, Oxford University Press, ISBN 0198596790
• Batchelor, G. K. (1967), An Introduction to Fluid Dynamics, Cambridge University Press, ISBN 0521663962
• Landau, L. D.; Lifshitz, E. M. (1987), Fluid mechanics, Course of Theoretical Physics, 6 (2nd revised ed.), Pergamon Press, ISBN 0 08 033932 8, OCLC 15017127
• Rhyming, Inge L. (1991), Dynamique des fluides, Presses Polytechniques et Universitaires Romandes, Lausanne
• Polyanin, A. D.; Kutepov, A. M.; Vyazmin, A. V.; Kazenin, D. A. (2002), Hydrodynamics, Mass and Heat Transfer in Chemical Engineering, Taylor & Francis, London, ISBN 0-415-27237-8
• Currie, I. G. (1974), Fundamental Mechanics of Fluids, McGraw-Hill, ISBN 0070150001

• Simplified derivation of the Navier–Stokes equations
• Millennium Prize problem description.
• CFD online software list A compilation of codes, including Navier–Stokes solvers.

EXISTENCE AND SMOOTHNESS OF THE NAVIER–STOKES EQUATION

CHARLES L. FEFFERMAN

The Euler and Navier–Stokes equations describe the motion of a fluid in Rn (n = 2 or 3). These equations are to be solved for an unknown velocity vector u(x, t) = (ui(x, t))1≤i≤n ∈ Rn and pressure p(x, t) ∈ R, defined for position x ∈ Rn and time t ≥ 0. We restrict attention here to incompressible fluids filling all of Rn. The Navier–Stokes equations are then given by

(6) and

p,u∈C∞(Rn ×[0,∞)) |u(x, t)|2dx < Cfor all t ≥ 0(bounded energy).

∂􏰄n∂ui ∂p n

(1)∂tui+uj∂x =ν∆ui−∂x +fi(x,t) j=1j i

(2)divu=􏰄n ∂ui =0 i=1 ∂xi

with initial conditions

(3)u(x, 0) = u◦(x)(x ∈ Rn).

(x∈R ,t≥0), (x∈Rn,t≥0)

Here, u◦(x) is a given, C∞ divergence-free vector field on Rn,fi(x,t) are the com- ponents of a given, externally applied force (e.g. gravity), ν is a positive coefficient

􏰄n ∂2 (the viscosity), and ∆ =∂x2 is the Laplacian in the space variables. The Euler

i=1i equations are equations (1), (2), (3) with ν set equal to zero.

Equation (1) is just Newton’s law f = ma for a fluid element subject to the ex- ternal force f = (fi (x, t))1≤i≤n and to the forces arising from pressure and friction. Equation (2) just says that the fluid is incompressible. For physically reasonable solutions, we want to make sure u(x, t) does not grow large as |x| → ∞. Hence, we will restrict attention to forces f and initial conditions u◦ that satisfy

(4)|∂xαu◦(x)| ≤ CαK (1 + |x|)−Kon Rn, for any α and K and (5)|∂xα∂tmf(x,t)|≤CαmK(1+|x|+t)−KonRn×[0,∞), foranyα,m,K. We accept a solution of (1), (2), (3) as physically reasonable only if it satisfies

􏰅

(7) Alternatively, to rule out problems at infinity, we may look for spatially periodic

Rn solutions of (1), (2), (3). Thus, we assume that u◦(x),f(x,t) satisfy

(8)u◦(x+ej)=u◦(x), f(x+ej,t)=f(x,t) for1≤j≤n 12CHARLES L. FEFFERMAN (ej = jth unit vector in Rn).

In place of (4) and (5), we assume that u◦ is smooth and that (9)|∂xα∂tmf(x,t)|≤CαmK(1+|t|)−KonR3×[0,∞), foranyα,m,K. We then accept a solution of (1), (2), (3) as physically reasonable if it satisfies (10)u(x,t)=u(x+ej,t) onR3×[0,∞)for 1≤j≤n and (11)p, u ∈ C∞(Rn × [0, ∞)).

A fundamental problem in analysis is to decide whether such smooth, physically reasonable solutions exist for the Navier–Stokes equations. To give reasonable lee- way to solvers while retaining the heart of the problem, we ask for a proof of one of the following four statements.

(A) Existence and smoothness of Navier–Stokes solutions on R3. Take ν > 0 and n = 3. Let u◦(x) be any smooth, divergence-free vector field satisfying (4). Take f(x,t) to be identically zero. Then there exist smooth functions p(x,t),ui(x,t) on R3 × [0, ∞) that satisfy (1), (2), (3), (6), (7).

(B) Existence and smoothness of Navier–Stokes solutions in R3/Z3. Take ν > 0 and n = 3. Let u◦(x) be any smooth, divergence-free vector field satisfying (8); we take f (x, t) to be identically zero. Then there exist smooth functions p(x, t), ui (x, t) on R3 × [0, ∞) that satisfy (1), (2), (3), (10), (11).

(C) Breakdown of Navier–Stokes solutions on R3. Take ν > 0 and n = 3. Then there exist a smooth, divergence-free vector field u◦(x) on R3 and a smooth f (x, t) on R3 × [0, ∞), satisfying (4), (5), for which there exist no solutions (p, u) of (1), (2), (3), (6), (7) on R3 × [0, ∞).

(D) Breakdown of Navier–Stokes Solutions on R3/Z3. Take ν > 0 and n = 3. Then there exist a smooth, divergence-free vector field u◦(x) on R3 and a smooth f (x, t) on R3 × [0, ∞), satisfying (8), (9), for which there exist no solutions (p, u) of (1), (2), (3), (10), (11) on R3 × [0, ∞).

These problems are also open and very important for the Euler equations (ν = 0), although the Euler equation is not on the Clay Institute’s list of prize problems.

Let me sketch the main partial results known regarding the Euler and Navier– Stokes equations, and conclude with a few remarks on the importance of the ques- tion.

In two dimensions, the analogues of assertions (A) and (B) have been known for a long time (Ladyzhenskaya [4]), also for the more difficult case of the Euler equations. This gives no hint about the three-dimensional case, since the main difficulties are absent in two dimensions. In three dimensions, it is known that (A) and (B) hold provided the initial velocity u◦ satisfies a smallness condition. For initial data u◦(x) not assumed to be small, it is known that (A) and (B) hold (also for ν = 0) if the time interval [0,∞) is replaced by a small time interval [0,T), with T depending on the initial data. For a given initial u◦(x), the maximum allowable T is called the “blowup time.” Either (A) and (B) hold, or else there is a smooth, divergence-free u◦(x) for which (1), (2), (3) have a solution with a finite blowup time. For the Navier–Stokes equations (ν > 0), if there is a solution with

EXISTENCE AND SMOOTHNESS OF THE NAVIER–STOKES EQUATION3

a finite blowup time T, then the velocity (ui(x,t))1≤i≤3 becomes unbounded near the blowup time.

Other unpleasant things are known to happen at the blowup time T , if T < ∞. For the Euler equations (ν = 0), if there is a solution (with f ≡ 0, say) with finite blowup time T , then the vorticity ω(x, t) = curlx u(x, t) satisfies

􏰅T􏰀 􏰁 sup |ω(x, t)|dt = ∞(Beale–Kato–Majda),

0x∈R3

so that the vorticity blows up rapidly. Many numerical computations appear to exhibit blowup for solutions of the

Euler equations, but the extreme numerical instability of the equations makes it very hard to draw reliable conclusions.

The above results are covered very well in the book of Bertozzi and Majda [1].

Starting with Leray [5], important progress has been made in understanding weak solutions of the Navier–Stokes equations. To arrive at the idea of a weak solution of a PDE, one integrates the equation against a test function, and then integrates by parts (formally) to make the derivatives fall on the test function. For instance, if (1) and (2) hold, then, for any smooth vector field θ(x,t) = (θi(x,t))1≤i≤n compactly supported in R3 × (0, ∞), a formal integration by parts yields

􏰅􏰅 ∂θ 􏰄􏰅􏰅 ∂θi (12)u· ∂tdxdt−uiuj∂x dxdt

R3 ×RR3 ×R 􏰅􏰅 􏰅􏰅 􏰅􏰅

= νu · ∆θdxdt +f · θdxdt −p · (div θ)dxdt. R3×RR3×RR3×R

Note that (12) makes sense for u ∈ L2, f ∈ L1, p ∈ L1, whereas (1) makes sense only if u(x, t) is twice differentiable in x. Similarly, if φ(x, t) is a smooth function, compactly supported in R3 × (0, ∞), then a formal integration by parts and (2) imply

(13)

􏰅􏰅

R3 ×R

u · ▽xφdxdt = 0.

ij j

A solution of (12), (13) is called a weak solution of the Navier–Stokes equations.

A long-established idea in analysis is to prove existence and regularity of solutions of a PDE by first constructing a weak solution, then showing that any weak solution is smooth. This program has been tried for Navier–Stokes with partial success. Leray in [5] showed that the Navier–Stokes equations (1), (2), (3) in three space dimensions always have a weak solution (p,u) with suitable growth properties. Uniqueness of weak solutions of the Navier–Stokes equation is not known. For the Euler equation, uniqueness of weak solutions is strikingly false. Scheffer [8], and, later, Schnirelman [9] exhibited weak solutions of the Euler equations on R2 × R with compact support in spacetime. This corresponds to a fluid that starts from rest at time t = 0, begins to move at time t = 1 with no outside stimulus, and returns to rest at time t = 2, with its motion always confined to a ball B ⊂ R3.

Scheffer [7] applied ideas from geometric measure theory to prove a partial regularity theorem for suitable weak solutions of the Navier–Stokes equations.

4CHARLES L. FEFFERMAN

Caffarelli–Kohn–Nirenberg [2] improved Scheffer’s results, and F.-H. Lin [6] sim- plified the proofs of the results in Caffarelli–Kohn–Nirenberg [2]. The partial regu- larity theorem of [2], [6] concerns a parabolic analogue of the Hausdorff dimension of the singular set of a suitable weak solution of Navier–Stokes. Here, the singu- lar set of a weak solution u consists of all points (x◦,t◦) ∈ R3 ×R such that u is unbounded in every neighborhood of (x◦,t◦). (If the force f is smooth, and if (x◦,t◦) doesn’t belong to the singular set, then it’s not hard to show that u can be corrected on a set of measure zero to become smooth in a neighborhood of (x◦, t◦).)

To define the parabolic analogue of Hausdorff dimension, we use parabolic cylin-

dersQr =Br×Ir ⊂R3×R,whereBr ⊂R3 isaballofradiusr,andIr ⊂Risan

interval of length r2. Given E ⊂ R3 × R and δ > 0, we set 􏰂∞􏰃

PK,δ(E) = inf􏰄riK : Qr1,Qr2,··· cover E, and each ri < δ i=1

and then define The main results of [2], [6] may be stated roughly as follows.

Theorem. (A) Let u be a weak solution of the Navier–Stokes equations, satisfying suitable growth conditions. Let E be the singular set of u. Then P1(E) = 0.

(B) Given a divergence-free vector field u◦(x) and a force f(x,t) satisfying (4) and (5), there exists a weak solution of Navier–Stokes (1), (2), (3) satisfying the growth conditions in (A).

In particular, the singular set of u cannot contain a spacetime curve of the form {(x, t) ∈ R3 × R : x = φ(t)}. This is the best partial regularity theorem known so far for the Navier–Stokes equation. It appears to be very hard to go further.

Let me end with a few words about the significance of the problems posed here. Fluids are important and hard to understand. There are many fascinating prob- lems and conjectures about the behavior of solutions of the Euler and Navier–Stokes equations. (See, for instance, Bertozzi–Majda [1] or Constantin [3].) Since we don’t even know whether these solutions exist, our understanding is at a very primitive level. Standard methods from PDE appear inadequate to settle the problem. In- stead, we probably need some deep, new ideas.

References

[1] A. Bertozzi and A. Majda, Vorticity and Incompressible Flows, Cambridge U. Press, Cam- bridge, 2002.

[2] L. Caffarelli, R. Kohn, and L. Nirenberg, Partial regularity of suitable weak solutions of the Navier–Stokes equations, Comm. Pure & Appl. Math. 35 (1982), 771–831.

[3] P. Constantin, Some open problems and research directions in the mathematical study of fluid dynamics, in Mathematics Unlimited–2001 and Beyond, Springer Verlag, Berlin, 2001, 353–360.

[4] O. Ladyzhenskaya, The Mathematical Theory of Viscous Incompressible Flows (2nd edition), Gordon and Breach, New York, 1969.

[5] J. Leray, Sur le mouvement d’un liquide visquex emplissent l’espace, Acta Math. J. 63 (1934), 193–248.

[6] F.-H. Lin, A new proof of the Caffarelli–Kohn–Nirenberg theorem, Comm. Pure. & Appl. Math. 51 (1998), 241–257.

[7] V. Scheffer, Turbulence and Hausdorff dimension, in Turbulence and the Navier–Stokes Equa- tions, Lecture Notes in Math. 565, Springer Verlag, Berlin, 1976, 94–112.

PK(E)= lim PK,δ(E). δ→0+

EXISTENCE AND SMOOTHNESS OF THE NAVIER–STOKES EQUATION5

[8] V. Scheffer, An inviscid flow with compact support in spacetime, J. Geom. Analysis 3 (1993), 343–401.

[9] A. Shnirelman, On the nonuniqueness of weak solutions of the Euler equation, Comm. Pure & Appl. Math. 50 (1997), 1260–1286.