I'm reading a book "A Mathematical Introduction to Fluid Mechanics" by Alexandre J. Chorin that says that the flow is isentropic if

$$ \nabla w = \frac{1}{\rho}\nabla p $$

where $w$ is an enthalpy, $\rho$ is density and $p$ is a pressure.

Then Chorin recalls the following two facts from thermodynamics

$$ dw = Tds + \frac{1}{\rho}dp $$ and $$ de = Tds + \frac{p}{\rho^2}d\rho $$ where $e$ is the internal energy, $s$ entropy and $T$ temperature.

But I do not understand the next claim of the book:

if $p$ is a function of $\rho$ only, then the flow is clearly isentropic.

Why is this true?

  • 1
    $\begingroup$ In the equation for de, shouldn't that be a $d\rho$? $\endgroup$ Nov 8, 2018 at 2:12
  • $\begingroup$ @ChesterMiller Yes You are right, sorry for the typo $\endgroup$
    – tgtt
    Nov 8, 2018 at 2:21
  • 1
    $\begingroup$ What the author says about p being a function of $\rho$ only makes no sense to me either, because, even for an ideal gas, the temperature must also be varying in tandem for an isentropic deformation. $\endgroup$ Nov 8, 2018 at 2:36
  • $\begingroup$ Maybe it means that, since dw is an exact differential, it follows that at constant S, the term $dp/\rho$ must be an exact differential. $\endgroup$ Nov 8, 2018 at 3:55
  • $\begingroup$ Perhaps he meant the remark to apply to a constant density fluid. $\endgroup$
    – Deep
    Nov 8, 2018 at 10:03

1 Answer 1


The underlying assumption in local equilibrium theories is that any thermodynamic variable locally depends only on two parameters, here they are $s$ and $p$. So the enthalpy is taken to be $w=w(s,p)$ and also $dw=\frac{\partial w}{\partial s}ds + \frac{\partial w}{\partial p} dp=Tds+\frac {1}{\rho}dp$ at every point; here $T=T(s,p)$ and $\rho = \rho (s,p)$.

If one also assumes that $\rho = \rho (p)$ independent of entropy then one must also have both $\frac{\partial w}{\partial s} = T$ and $\frac{\partial w}{\partial p} =\frac {1}{\rho}$. So integrating the latter we get $w(s,p)=f(p)+g(s)$ and then $T(s,p)=g'(s)$ and thus the enthalpy change is really the sum of two independent exact differentials, one in the $s$ and the other in $p$ (or $\rho$) so the two are completely unrelated $dw=g'(s)ds + f'(p)dp = dg(s) + df(p)$ and one can change without affecting the other.

It may be a misnomer to say that it is isentropic, i.e., $ds=0$, as a consequence of $\rho$ being dependent on $p$ alone; the important point is that entropy change itself has no effect on the fluid flow and specifically on its mechanical state.

  • $\begingroup$ This seems like a nice answer; however, the statement in the question was 'if $p=p(\rho)$'. Is that implied, if $\rho=\rho(p)$? $\endgroup$
    – Time4Tea
    Nov 12, 2018 at 18:15
  • 1
    $\begingroup$ @Time4Tea thermodynamically $p=p(\rho)$ is the same as $\rho = \rho (p)$, that is neither variable depends on $s$; obviously they are the same assumptions whenever either is invertible, but even if it is not invertible globally it should be equivalent over finite monotonic segments of the variables. $\endgroup$
    – hyportnex
    Nov 12, 2018 at 18:28
  • $\begingroup$ Ok, I see. So, it's the fact that $\rho=\rho (p)$ that implies that the enthalpy can be split into separate functions, $f(p)$ and $g(s)$? So, could another way of looking at it be that $T=T(s)$ (since $T(s,p)=g'(s)$), which means that entropy could change if heat were added or removed, but as you say, not due to the mechanical state of the flow? $\endgroup$
    – Time4Tea
    Nov 13, 2018 at 14:25
  • 1
    $\begingroup$ Yes, that is my interpretation, too. I would also make the distinction that $\rho = \rho (p)$ being equivalent to $p=p(\rho)$ is a mathematical "detail" regarding continuous functions without apparent physical significance but that we can write the enthalpy as $g(s)+f(p)$ as equal consequence of either assuming $\rho = \rho (p)$ or $T=T(s)$ is physically interesting in that both imply the uncoupled, ie., separate thermal and mechanical nature of the flow. $\endgroup$
    – hyportnex
    Nov 13, 2018 at 14:35

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.