1

Most fluid mechanics textbooks deal with mass, momentum and energy transport in fluid flows. Are there any books that deal with entropy transport in fluid flows?

AB2008
  • 11
  • 1
  • What do you need? Maybe an entire book on entropy transport in FM is too much, and an answer with some references can solve your doubts better than a book. Please provide some details: are you dealing with incompressible (even though it's hard to talk about thermodynamics in incompressible flows...)compressible flows? Shocks/no shock? One/many phases? and so on... – basics Dec 20 '22 at 08:17
  • I am mostly dealing with heat transfer in single phase compressible flows with no shock. – AB2008 Dec 20 '22 at 08:24
  • ok, nice. Can we treat the fluid as a Newtonian fluid for viscosity, namely stress tensor as $\mathbb{T} = -p \mathbb{I} + 2 \mu \mathbb{D} + \lambda (\nabla \cdot \mathbf{u}) \mathbb{I}$, with $\mathbb{D}$ the symmetric part of the velocity gradient tensor? – basics Dec 20 '22 at 08:32
  • Yes, the fluid is Newtonian. – AB2008 Dec 20 '22 at 08:39
  • Let's start from this answer. Then we'll go into details of your needs. – basics Dec 20 '22 at 09:09
  • Another question I didn't ask before. Are you interested in differential or integral balances? I went to the differential, because it's easier to manipulate the principles and derive from that balance equations for kinetic energy, internal energy and entropy – basics Dec 20 '22 at 14:28
  • I answered a few related questions on the matter of entropy transport: https://physics.stackexchange.com/q/717990/25301, https://physics.stackexchange.com/q/78921/25301. – Kyle Kanos Dec 20 '22 at 15:04
  • @KyleKanos am I missing something, or your links have nothing, or very little, to do with thi answer? – basics Dec 20 '22 at 19:16
  • @basics Second one was not right. I should have linked https://physics.stackexchange.com/q/116779/25301 instead (which is actually linked in first). Both involve fluid flows and entropy transport (though in a short sentence in q/717990). – Kyle Kanos Dec 20 '22 at 20:53
  • Transport Phenomena by Bird, Stewart, and Lightfoot, Chapter 24 – Chet Miller Dec 23 '22 at 13:04

2 Answers2

1

Differential equations

From the principle differential equations in form:
$D_t \rho = -\rho \nabla \cdot \mathbf{u} \qquad \qquad \qquad \qquad \qquad$ (mass)
$\rho D_t \mathbf{u} = \rho \mathbf{g} + \nabla \cdot \mathbb{T} \qquad \qquad \qquad \qquad \ $ (mometum)
$\rho D_t e^{t} = \rho \mathbf{g} \cdot \mathbf{u} + \nabla \cdot ( \mathbb{T} \cdot \mathbf{u}) -\nabla \cdot \mathbf{q}\quad \ \ $(total energy)

it's possible to write some balance equations for derived physical quantities, like kinetic energy, internal energy and the entropy:

  • kinetic energy, doing $\mathbf{u} \cdot$ the momentum equation, to get a balance for the kinetic energy $k = \frac{|\mathbf{u}|^2}{2}$:
    $\rho D_t k = \rho \mathbf{g} \cdot \mathbf{u} + \nabla \cdot \mathbb{T} \cdot \mathbf{u}$;

  • internal energy, $e = e^{t} - k$, subtracting the kinetice energy equation from the total energy:
    $\rho D_t e = \nabla \mathbf{u} : \mathbb{T} - \nabla \cdot \mathbf{q}$

  • entropy, putting together the internal energy and the reversible contribution of the internal energy, $\nabla \mathbf{u} \cdot \mathbb{T} = \nabla \mathbf{u} : \left( -p \mathbb{I} + 2\mu \mathbb{D} + \lambda (\nabla \cdot \mathbf{u} ) \mathbb{I} \right) = - p \nabla \cdot \mathbf{u} + 2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2$,

    so that

    $\rho D_t e + p \nabla \cdot \mathbf{u} = 2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2 - \nabla \cdot \mathbf{q}$
    $\rho T D_t s = 2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2 - \nabla \cdot \mathbf{q}$

    to get:

    $\rho D_t s = \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} - \dfrac{\nabla \cdot \mathbf{q}}{T} = \\ \qquad \ = \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} - \dfrac{\mathbf{q} \cdot \nabla T}{T^2} - \nabla \cdot \left(\dfrac{\mathbf{q}}{T}\right)$

    and if we add the Fourier assumption on the heat conduction flux $\mathbf{q} = -k \nabla T$, we can recast the last term as

    $\rho D_t s = \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} + k\dfrac{|\nabla T|^2}{T^2} + \nabla \cdot \left( \dfrac{ k \nabla T}{T} \right)$

    where we can recognize two contributions at the right-hand side, namely:

    • a volume entropy source,

      $\dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} - \dfrac{\mathbf{q} \cdot \nabla T}{T^2} = \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} + k\dfrac{|\nabla T|^2}{T^2}$

      always positive, because of the norms, of the positive values of absolute temperature and the constraints on the viscosity coefficients and the conductivity.

    • a flux contribution, that should remind you the contribution in Clausius formulation of the second principle of thermodyamics, $dS \ge \frac{\delta Q}{T}$,

      $ - \nabla \cdot \left(\dfrac{\mathbf{q}}{T}\right) = \nabla \cdot \left( \dfrac{ k \nabla T}{T} \right)$

Integral equations

Integrating over a material volume, we get the integral balance for a closed system,

$\dfrac{d}{dt} \displaystyle \int_{V_t} \rho s = \int_{V_t} \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} + k\dfrac{|\nabla T|^2}{T^2} - \oint_{\partial V_t } \dfrac{\mathbf{q}}{T} \cdot \mathbf{\hat{n}} $

As you can see, there is a non-negative volume source due to non-rigid motion in viscous fluids, and due to heat transfer between different regions of the domain: viscosity and heat transfer are causes of irreversibility; the other term is the entropy flux through the boundary of the domain, and reminds the expression of Clausius statement of the second principle of thermodynamics.

Exploiting Reynolds' transport theorem we get the balance for an arbitrary domain,

$\dfrac{d}{dt} \displaystyle \int_{v_t} \rho s + \oint_{\partial v_t} \rho s (\mathbf{u} - \mathbf{u}_s) \cdot \mathbf{\hat{n}} = \int_{v_t} \dfrac{2 \mu |\mathbb{D}|^2 + \lambda (\nabla \cdot \mathbf{u})^2}{T} + k\dfrac{|\nabla T|^2}{T^2} - \oint_{\partial v_t } \dfrac{\mathbf{q}}{T} \cdot \mathbf{\hat{n}} $.

where some flux of entropy is associated with the flux of mass through the boundaries of the domain.

basics
  • 8,365
0

For an in-depth resource on entropy and fluid dynamics, you may want to look at Evans' Entropy and PDEs (NB: PDF link to author's website), specifically Chapter 5 (starting on page 106). The text itself is written in a more mathematically formal manner than one might find in a physics textbook, though it is done in way that should be understood by anyone with vector calculus under their tool belt.

Kyle Kanos
  • 28,229
  • 41
  • 68
  • 131