I'm reading a book "A Mathematical Introduction to Fluid Mechanics" by Alexandre J. Chorin, and I came across the derivation of Euler's equations for isentropic flow. Page 15, the author goes from

$$\frac{d}{dt}\int_{W_t} (\frac{1}{2} \rho ||\vec{u}^2|| + \rho \epsilon ) dV = -\int_{\partial W_t} p \vec{u}\cdot \vec{n} dA + \int_{W_t}\rho \vec{u}\cdot \vec{b}dV $$


$$\frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u} = -\nabla \omega + \vec{b}$$

Now, this is supposed to be a compressible flow, so $\nabla \cdot \vec{u}$ is not necessarily equal to 0, and change in internal energy $\epsilon$ is not necessarily zero either.

The author writes

This follows from the balance of momentum using our earlier expressions for $(d/dt)E_{kinetic}$, the transport theorem, and $p = \rho^2 \frac{\partial \epsilon}{\partial \rho}$`

This is what I believe to be the earlier expressions for the $(d/dt)E_{kinetic}$

$$d/dt E_{kinetic} = \frac{d}{dt}\int_{W_t} (\frac{1}{2} \rho ||\vec{u}^2||)dV = \int _{W_t} \rho ( \vec{u}\cdot(\frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u}))dV$$

When I tried to reach the result myself, I get stuck at:

$$ \frac{d}{dt}\int_{W_t} (\frac{1}{2} \rho ||\vec{u}^2|| + \rho \epsilon ) dV = -\int_{\partial W_t} p \vec{u}\cdot \vec{n} dA + \int_{W_t}\rho \vec{u}\cdot \vec{b}dV\\ \int_{W_t} (\rho(\vec{u}\cdot \frac{\partial \vec{u}}{\partial t} + \vec{u} \cdot ((\vec{u} \cdot \nabla)\vec{u})) + \rho \frac{D}{Dt}\epsilon ) dV = \int_{W_t} (- \nabla \cdot (p \vec{u}) + \rho\vec{u}\cdot \vec{b}) dV\\ \rho(\vec{u}\cdot \frac{\partial \vec{u}}{\partial t} + \vec{u} \cdot ((\vec{u} \cdot \nabla)\vec{u})) + \rho \frac{D}{Dt}\epsilon = - (\vec{u}\cdot(\nabla p) + p\nabla\cdot \vec{u}) + \rho\vec{u}\cdot \vec{b} \\ \rho\vec{u}\cdot(\frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u}) + \rho \frac{\partial \epsilon}{\partial t} + \rho \nabla \cdot (\epsilon \vec{u})= - \vec{u}\cdot(\rho \nabla \omega) - p\nabla\cdot \vec{u} + \rho\vec{u}\cdot \vec{b} \\ $$

which doesn't seem to be reducible any further. UNLESS I presume it's incompressible, that is; $(D/Dt) \epsilon = 0$ and $\nabla \cdot \vec{u} = 0$. When I do, I can then do: $$\rho\vec{u}\cdot(\frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u}) + \rho \frac{\partial \epsilon}{\partial t} + \rho \nabla \cdot (\epsilon \vec{u})= - \vec{u}\cdot(\rho \nabla \omega) - p\nabla\cdot \vec{u} + \rho\vec{u}\cdot \vec{b} \\ \rho\vec{u}\cdot(\frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u})= - \vec{u}\cdot(\rho \nabla \omega) + \rho\vec{u}\cdot \vec{b} \\ \frac{\partial \vec{u}}{\partial t} + (\vec{u} \cdot \nabla)\vec{u} = -\nabla \omega + \vec{b}$$ which is exactly the answer the book claims. But this equation is supposed to describe (together with equation of conservation of mass and boundary condition for trapped volume $\vec{u}\cdot \vec{n} = 0$) compressible isentropic flow. How do I get there?


1 Answer 1


Let's start from

$$ \frac{\mathrm{d}}{\mathrm{d}t} \int_{W_t}\left(\tfrac{1}{2}\rho\rvert\rvert\mathbf{u}\rvert\rvert^2+\rho\epsilon\right)\mathrm{d}V = -\int_{W_t}p\mathbf{u}\cdot\mathbf{n}\mathrm{d}A + \int_{W_t}\rho\mathbf{u}\cdot\mathbf{b}\mathrm{d}V, $$

and use the transport theorem and divergence theorem to obtan

$$ \int_{W_t}\left(\rho\mathbf{u}\cdot\frac{D\mathbf{u}}{Dt} + \rho\frac{D\epsilon}{Dt}\right)\mathrm{d}V = \int_{W_t}\left(-\nabla\cdot(p\mathbf{u}) + \rho\mathbf{u}\cdot\mathbf{b}\right)\mathrm{d}V $$

$$ \Longrightarrow\;\,\rho\mathbf{u}\cdot\frac{D\mathbf{u}}{Dt} + \rho\frac{D\epsilon}{Dt} = -\nabla\cdot(p\mathbf{u}) + \rho\mathbf{u}\cdot\mathbf{b}. $$

Now, we divide through by $\rho$ and use $\nabla w = \nabla p/\rho$ to obtain

$$ \mathbf{u}\cdot\frac{D\mathbf{u}}{Dt} + \frac{\partial\epsilon}{\partial\rho}\frac{D\rho}{Dt} = -\mathbf{u}\cdot\nabla w - \frac{p}{\rho}\nabla\cdot\mathbf{u} + \mathbf{u}\cdot\mathbf{b}. $$

Finally, using $D\rho/Dt=-\rho\nabla\cdot\mathbf{u}$ and $p/\rho = \rho\partial\epsilon/\partial\rho$, we find

$$ \mathbf{u}\cdot\frac{D\mathbf{u}}{Dt} - \frac{p}{\rho}\nabla\cdot\mathbf{u} = -\mathbf{u}\cdot\nabla w - \frac{p}{\rho}\nabla\cdot\mathbf{u} + \mathbf{u}\cdot\mathbf{b} $$

$$ \Longrightarrow\;\, \frac{D\mathbf{u}}{Dt} = -\nabla w + \mathbf{b}. $$

P.S. Note that $D\epsilon/Dt = \partial\epsilon/\partial t + \mathbf{u}\cdot\nabla\epsilon$, which is different from $\partial\epsilon/\partial t + \nabla(\epsilon\mathbf{u})$.

  • $\begingroup$ Thank you very much. $D\rho / Dt = (-) \rho \nabla \cdot \vec{u}$. I tried to edit your post, but it won't let me do edits that are less than 6 characters. $\endgroup$ Feb 18, 2018 at 9:47
  • $\begingroup$ No problem! And you're right, I've edited it. $\endgroup$
    – Elbers
    Feb 18, 2018 at 13:34

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.