6
$\begingroup$

I'm trying to understand the basics of fluid dynamics and the Navier-Stokes equations by following the short book A Mathematical Introduction to Fluid Dynamics by Chorin and Marsden (I'm an applied mathematician, not a physicist, so please be patient with me. Also I wasn't sure if this question was better suited for MSE or PSE. I chose here.)


As I understand it, to uniquely determine the three important quantities in a fluid mechanics problem which are $\mathrm u$, $P$ and $\rho$, one needs five equations - one for each component of $\mathrm u$ (which can be written as a single vector equation, reducing our number of equations to 3), and two more for the scalars $P,\rho$. To get these three equations, we need three physical conditions. They are

  • Conservation of mass
  • Conservation of momentum
  • Conservation of energy

I want to derive these three equations in the most general form possible. What I mean by that is our fluid may be

  • compressible ($\nabla\cdot\mathrm u\neq0$)
  • viscous ($\mu,\lambda\neq0$)
  • vortical ($\nabla\times\mathrm u\neq0$)
  • Subject to thermodynamic effects

The easiest part is conservation of mass. A short derivation will show that mass conservation may be described mathematically via the continuity equation,

$$\partial_t\rho+\nabla\cdot(\rho~\mathrm u)=0\tag{1}$$

Conservation of momentum is considerably more difficult, but after some work I was able to eventually arrive at

$$\rho\frac{D\mathrm u}{Dt}=-\nabla P+\nabla\cdot \boldsymbol{\tau}+\mathrm{F}\tag{2A}$$

Where,

  • $D/Dt=\partial_t +\mathrm u\cdot\nabla$ is the material derivative,
  • $\mathrm{F}$ denotes any external forces ($\mathrm{F}=0$ in an isolated system),
  • $\boldsymbol{\tau}$ is the deviatoric stress tensor, a $(1,1)$ tensor

defined as $$\boldsymbol{\tau}=\lambda\mathbf{I}~\nabla\cdot\mathrm u+\mu\big(\nabla\mathrm u+(\nabla\mathrm u)^{\mathrm T}\big)$$ Where $\mathbf I$ is the identity, $\mu$ is the dynamic viscosity, and $\lambda$ is the bulk viscosity. Using the identity $\nabla\cdot(\nabla\mathrm u)^{\mathrm T}=\nabla(\nabla\cdot\mathrm u)$ we can expand out this equation to reach

$$\rho(\partial_t\mathrm u+\mathrm u\cdot\nabla\mathrm u)=-\nabla P+(\lambda+\mu)\nabla(\nabla\cdot\mathrm u)+\mu\boldsymbol{\triangle}\mathrm u+\mathrm F\tag{2B}$$

This equation is given on page 33 of the linked text.

However, conservation of energy still eludes me. NASA appears to have a published formula, but it is only given in the special case of Cartesian coordinates, and I would much rather a coordinate-free representation. This page gives a coordinate-free description:

$$\partial_t\left[\rho~\left(e+\frac{|\mathrm u|^2}{2}\right)\right]+\nabla\cdot\left[\rho\mathrm u~\left(e+\frac{|\mathrm u|^2}{2}\right)\right]=\nabla\cdot(k\nabla T-P\mathrm u+\boldsymbol{\tau}\cdot\mathrm u)+\mathrm u\cdot\mathrm F+\mathcal{Q}\tag{3}$$

But they don't offer a derivation, nor do they even explain what the symbols $e,k,\mathcal Q$ mean. My guess(?) is that $k$ is some kind of thermal diffusivity, $e$ is a function of position and time and tells you something about the internal or potential energy, and $\mathcal{Q}$ is some external heat source. Am I right? And in most cases we would know $e(\mathrm r,t)$ and $T(\mathrm r,t)$ as given, and not have to solve for them, correct?

If we assume that the fluid has no thermal or potential energy, the energy can be written as purely kinetic: $$E_{\text{tot}}=\frac{1}{2}\int_{\Omega(t)}\rho(\mathrm r,t)~|\mathrm u(\mathrm r,t)|^2~\mathrm d\mu(\mathrm r)$$ Here $\Omega(t)$ is the region that our fluid occupies at time $t$. Using the transport theorem, one can arrive at $$\frac{\mathrm d E_{\text{tot}}}{\mathrm dt}=\frac{\mathrm d}{\mathrm dt}\int_{\Omega(t)}\rho |\mathrm u|^2\mathrm d\mu=\int_{\Omega(t)}\rho\mathrm u\cdot\frac{D\mathrm u}{Dt}\mathrm d\mu$$ So it would appear that a sufficient constraint for conservation of energy in this case is

$$\rho\mathrm u\cdot\frac{D\mathrm u}{Dt}=0\implies \mathrm u\cdot\big(\partial_t\mathrm u+\mathrm u\cdot\nabla\mathrm u\big)=0\tag{4}$$

Which is a slightly modified form of the well known Burgers' equation.

But what if there is thermal and kinetic energy? What can I do then? Can anyone please help me derive and make sense of equation $(3)$, and reduce it to $(4)$ in a special case?

Thanks so much.

EDIT: Definitions for various uses of the symbol $\nabla$.

If $\mathbf{T}$ is an $(r,s)$ tensor, then I define the components of $\nabla \mathbf{T}$ in a coordinate system with Christoffel symbols $\Gamma^i_{jk}$ as (Using Einstein summation)

$$( \nabla \mathbf{T})^{i_{1} \dotsc i_{r}}{}_{j_{1} \dotsc j_{s} \ k} =\begin{matrix} \partial _{k} T^{i_{1} \dotsc i_{r}}{}_{j_{1} \dotsc j_{s}}\\ +\Gamma _{lk}^{i_{1}} T^{l\ i_{2} \dotsc i_{r}}{}_{j_{1} \dotsc j_{s}} +\cdots +\Gamma _{lk}^{i_{r}} T^{i_{1} \dotsc i_{r-1} \ l}{}_{j_{1} \dotsc j_{s}}\\ -\Gamma _{j_{1} k}^{l} T^{i_{1} \dotsc i_{r}}{}_{l\ j_{2} \dotsc j_{s}} -\cdots -\Gamma _{j_{s} k}^{l} T^{i_{1} \dotsc i_{r}}{}_{j_{1} \dotsc j_{s-1} \ l} \end{matrix}$$

I define the components the divergence of a $(1,s)$ tensor $\mathbf{T}$ as

$$(\nabla\cdot\mathbf{T})_{j_1\dots j_s}= (\nabla \mathbf{T})^i{}_{j_1\dots j_s~i}$$

And lastly I define the Laplacian of an $(r,s)$ tensor as

$$(\boldsymbol{\triangle}\mathbf{T})^{i_1\dots i_r}{}_{j_1\dots j_s}=g^{kl}(\nabla(\nabla\mathbf{T}))^{i_1\dots i_r}{}_{j_1\dots j_s~kl}$$

Where $g^{ij}$ is the $i,j$th component of the inverse metric.

$\endgroup$
4
  • 1
    $\begingroup$ Your guesses were mostly right. e is the internal energy per unit mass, k is the thermal conductivity and Q is an external source of heat. The function $e(\rho, T)$ is usually know, as well as the equation of state $T(\rho,P)$ $\endgroup$
    – Prallax
    Aug 8, 2021 at 22:39
  • 1
    $\begingroup$ The main idea for the NASA energy equation is basically application of the first law of thermodynamics, and its use is typically to calculate the temperature distribution. The e in the equation is the internal energy per unit mass, and is a function of temperature and pressure. k is the thermal conductivity of the fluid, F is the body force per unit volume, and q $\endgroup$ Aug 9, 2021 at 3:10
  • $\begingroup$ I am probably at a level of expertise now where I can answer this question myself. I will post something if I find the time. $\endgroup$
    – K.defaoite
    Feb 1, 2023 at 20:46
  • $\begingroup$ See also: "Derivation of continuum expression of the first law of thermodynamics" physics.stackexchange.com/q/265026/226902 and "Forms of the first law of thermodynamics" physics.stackexchange.com/q/67163/226902 $\endgroup$
    – Quillo
    Jun 25, 2023 at 12:52

1 Answer 1

4
$\begingroup$

The kinetic energy equation for stationary flow is just Bernoulli: $$ ({\bf v}\cdot\nabla) \left( \frac 12 |{\bf v}|^2+ h\right)=0 $$ where $h$ is the enthalpy $E+PV$ per unit mass.

For the thermal energy I'm going to ignore viscosity and heat flow, but the general idea should be obtained from the following:

For a non-relativistic fluid we can write the equations of fluid flow as $$ \left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right)\rho= -\rho\,{\rm div\,} {\bf v}\nonumber\\ \left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right) {\bf v}= -\frac{1}{\rho} \nabla p\nonumber\\ \left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right)E=-\frac{p}{\rho} {\rm div\,}{\bf v} .\nonumber $$ In deriving the last (energy) equation we have assumed that there is no heat flow so only thing changing $E$, the energy per unit mass, is the work $-pdV=-pd(1/\rho)$. It is therefore not surprising that we can combine the first and last of these equations to get $$ \frac 1 T \left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right) E+ \frac{p}{T}\left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right)\frac{1}{\rho}=0. $$ Since $$ TdS= dE+pd\left(\frac{1}{\rho}\right), $$ this last is $$ \left(\frac{\partial}{\partial t}+{\bf v}\cdot \nabla\right) S=0. $$ Combining this convective constancy of the entropy with mass conservation, we have $$ \frac{\partial \rho S}{\partial t}+ {\rm div\,} (\rho S {\bf v})=0. $$ Equivalently, this is $$ \frac{\partial s}{\partial t}+ {\rm div\,}(s {\bf v})=0, $$ where $s= \rho S=S/V$.

With $\varepsilon= \rho E= E/V$, the internal energy (non) conservation equation can be written as $$ \frac{\partial \varepsilon }{\partial t}+ {\rm div\,} \varepsilon {\bf v}+ P\,{\rm div\,}{\bf v}=0. $$ Thus $$ T\left(\frac{\partial s}{\partial t}+ {\rm div\,}s {\bf v}\right) =T\left(\frac{\partial s}{\partial t}+ {\rm div\,}s {\bf v}\right)+\mu \left(\frac{\partial \rho}{\partial t}+ {\rm div\,} \rho {\bf v}\right) -\left( \frac{\partial \varepsilon}{\partial t}+ {\rm div\,} \varepsilon {\bf v} +P\,{\rm div\,}{\bf v}\right)\nonumber \\ =\left(T \frac{\partial s}{\partial t}+\mu \frac{\partial \rho}{\partial t}-\frac{\partial \varepsilon}{\partial t}\right)+ {\bf v}\cdot \left(T \nabla s +\mu \nabla \rho-\nabla \varepsilon \right) \nonumber\\ \qquad +\left(Ts+\mu \rho -\varepsilon-P\right){\rm div\,} {\bf v}.\nonumber $$ The last RHS is zero because the first two terms compose the convective derivative of $ Tds+\mu d\rho -d\varepsilon$, which is zero, and also $TS+\mu N-E-PV$ is identically zero.

Thermo relations: We have that the Gibbs free energy is $E-TS+pV = \mu N$, so, with the definitions $$ \epsilon = E/V,\quad n=N/V,\quad s=S/V, $$ we can write the first law $$ dE=TdS-pdV +\mu dN $$ as $$ d\epsilon = \frac 1 V (TdS-pdV+\mu dN) - E \frac{dV}{V^2}\nonumber\\ = \frac 1 V (TdS+\mu dN) - \frac {dV}{V^2}(TS+\mu N)\nonumber\\ = Tds +\mu dn\nonumber $$

$\endgroup$
3
  • 1
    $\begingroup$ I'm still a bit confused... What is the difference between $p,P$? And what are $V,N$? How do I link these equations with (1),(2) in my question to be able to arrive at a unique solution for $\mathrm u,\rho,P$ given suitable initial and boundary conditions? Thanks for the answer. $\endgroup$
    – K.defaoite
    Aug 8, 2021 at 23:19
  • $\begingroup$ @K.defaoite See en.wikipedia.org/wiki/Gibbs_free_energy for a definition of the notation. $\endgroup$
    – alephzero
    Aug 8, 2021 at 23:39
  • 1
    $\begingroup$ Sorry! Copying from various notes caused a mess. There is no difference in the $p$'s $p=P$, but $N$ is total number of particles, $V$ is volume of a unit mass so mass density os $\rho= 1/V$. The rest is standard thermodynamics. My first two equations are your 1 and 2 and the my third is the energy equation for the internal energy that I think you are asking for. As to solving them, that is a vast subject. $\endgroup$
    – mike stone
    Aug 8, 2021 at 23:39

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.