0
$\begingroup$

I have seen in numerous literature sources (e.g. this chapter from MIT open courseware, pages 414-415) the following method of relating the pressure to particle velocity. Begin with the pressure field of an acoustic monopole (point source with uniform spherical waves): $$p(r) = \frac{A}{r}e^{j(\omega t - kr)} $$ where $r$ is radial distance from the source, $A$ is some arbitrary amplitude, $\omega$ is the angular temporal frequency of the signal, and $k$ is the angular spatial frequency. Newton's second law (as described in the literature) relates particle velocity $\textbf{u}$ to the pressure gradient: $$\rho \frac{\partial \textbf{u}}{\partial t} = -\nabla p$$ Radial symmetry in this case means that the pressure gradient is dependent only on $r$: $$\nabla p = \frac{\partial p}{\partial r} \textbf{e}_r = \left[ -\frac{A}{r^2}e^{j(\omega t - kr)} - \frac{A}{r}jke^{j(\omega t - kr)}\right] \textbf{e}_r$$ $$\nabla p = (-\frac{1}{r} - jk)p \textbf{e}_r$$ Newton's second law can therefore be written: $$\rho \frac{\partial u_r}{\partial t} = (\frac{1}{r} + jk)p $$ Assuming an oscillatory solution for $u_r$, $\dot u_r = j\omega u_r$, which gives $$ \rho j \omega u_r = (\frac{1}{r} + jk)p $$ Dividing through by $jk$, $$ \frac{\rho \omega u_r}{k} = (\frac{1}{jkr} + 1)p $$ $$ u_r = \frac{1}{\rho c} (1- j \frac{c}{2\pi fr})p$$

That's the result that I see wherever I look, but had originally taken an approach in which I considered a differential spherical element of volume $\textrm{d}V = r^2 \sin(\theta)\textrm{d}r \textrm{d}\theta \textrm{d}\phi$.

A force balance in the $\hat r$ direction on the element yields $$ \Sigma F_r = \left( p - \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left[ \left(r- \frac{\textrm{d}r}{2} \right)^2 \sin(\theta)\textrm{d}\theta \textrm{d}\phi \right] - \left( p + \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left[ \left(r + \frac{\textrm{d}r}{2} \right)^2 \sin(\theta)\textrm{d}\theta \textrm{d}\phi \right]$$ Dividing through by $\sin(\theta)\textrm{d}\theta\textrm{d}\phi$ and distributing terms, $$ \frac{\Sigma F_r}{\sin(\theta)\textrm{d}\theta\textrm{d}\phi} = \left( p - \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left( r^2 - r \textrm{d}r + \frac{\textrm{d}r^2}{4} \right)- \left( p + \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left( r^2 + r \textrm{d}r + \frac{\textrm{d}r^2}{4} \right) $$ $$ = pr^2 - pr\textrm{d}r + p \frac{\textrm{d}r^2}{4} - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r^2 + \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r\textrm{d}r - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} \frac{\textrm{d}r^2}{4} $$ $$ - pr^2 - pr\textrm{d}r - p \frac{\textrm{d}r^2}{4} - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r^2 - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r\textrm{d}r - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} \frac{\textrm{d}r^2}{4} $$ Neglecting higher order terms, this force balance reduces to $$ \frac{\Sigma F_r}{\sin(\theta)\textrm{d}\theta\textrm{d}\phi} = -2pr\textrm{d}r - \frac{\partial p}{\partial r} r^2 \textrm{d}r $$ Dividing through by $r^2\textrm{d}r$, the sum of the forces (in the $\hat r$ direction) has been divided by the differential volume $$ \frac{\Sigma F_r}{r^2\sin(\theta)\textrm{d}r\textrm{d}\theta\textrm{d}\phi} = -\frac{2p}{r} - \frac{\partial p}{\partial r} $$ The discrepancy between the result of this approach and that of the literature is clear at this point. Moving on, recalling that $$ \frac{\partial p}{\partial r} = (-\frac{1}{r} - jk)p $$ gives the equation of motion for the fluid due to the acoustic monopole $$ \rho \frac{\partial u_r}{\partial t} = -\frac{2p}{r} + \frac{p}{r} +jkp$$ $$ \rho j \omega u_r = \left(-\frac{1}{r} + jk\right)p $$ Rearranging, we arrive at my original result: $$ u_r = \frac{1}{\rho c} (1+ j \frac{c}{2\pi fr})p$$ And compare to the solution that I find elsewhere: $$ u_r = \frac{1}{\rho c} (1- j \frac{c}{2\pi fr})p$$

It's easy to look at this inconsistency and think of it as a sign error, but I don't think that's the case. I suspect that the problem begins with the formulation of the equation of motion. All external sources that I have seen have taken as given $$\rho \frac{\partial u}{\partial t} = -\nabla p$$ but is this really correct for a spherical wave, in which the "front" and "back" surfaces have different areas on which the pressure acts? This expression looks like a plane wave solution to me.

Where in my derivation have I gone astray? Many thanks to anyone who attempts to clarify this concept for me!

$\endgroup$
0
$\begingroup$

Here it is:

The force balance on a differential fluid element around an acoustic monopole must include all forces in the $\textbf{e}_r$ direction. The symmetry of the spherical wavefront allows us to ignore any viscous effects. Forces on the element due to the pressure field include the following: $$ \left( p - \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left[ \left(r- \frac{\textrm{d}r}{2} \right)^2 \sin(\theta)\textrm{d}\theta \textrm{d}\phi \right] \textbf{e}_r \hspace{5mm} \textrm{on the $-\textbf{e}_r$ surface} $$

$$ - \left( p + \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left[ \left(r + \frac{\textrm{d}r}{2} \right)^2 \sin(\theta)\textrm{d}\theta \textrm{d}\phi \right] \textbf{e}_r \hspace{5mm} \textrm{on the $\textbf{e}_r$ surface} $$

$$ p \left[ r \sin{\theta} \textrm{d}r \textrm{d}\phi \right] \frac{\sin(\textrm{d}\theta)}{2} \textbf{e}_r \hspace{5mm} \textrm{on the $\textbf{e}_{\theta}$ and $-\textbf{e}_{\theta}$ surfaces} $$

$$ p \left[ r \textrm{d}r \textrm{d}\theta \right] \frac{\sin(\theta) \sin(\textrm{d}\phi)}{2} \textbf{e}_r \hspace{5mm} \textrm{on the $\textbf{e}_{\phi}$ and $-\textbf{e}_{\phi}$ surfaces} $$

Where $p$ is the pressure at the center of the element. The small angle approximations $\sin(\textrm{d}\phi) \approx \textrm{d}\phi$ and $\sin(\textrm{d}\theta) \approx \textrm{d}\theta$ simplify the force balance to

$$ \frac{\Sigma F_r}{\sin(\theta) \textrm{d}\theta \textrm{d}\phi} = \left( p - \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left(r- \frac{\textrm{d}r}{2} \right)^2 - \left( p + \frac{\partial p}{\partial r} \frac{\textrm{d}r}{2} \right) \left(r + \frac{\textrm{d}r}{2} \right)^2 + 2pr\textrm{d}r $$

$$ = pr^2 - pr\textrm{d}r + p \frac{\textrm{d}r^2}{4} - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r^2 + \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r\textrm{d}r - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} \frac{\textrm{d}r^2}{4} $$ $$ - pr^2 - pr\textrm{d}r - p \frac{\textrm{d}r^2}{4} - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r^2 - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} r\textrm{d}r - \frac{\partial p}{\partial r}\frac{\textrm{d}r}{2} \frac{\textrm{d}r^2}{4} + 2pr\textrm{d}r $$ Neglecting higher order terms, this force balance reduces to $$ \frac{\Sigma F_r}{\sin(\theta)\textrm{d}\theta\textrm{d}\phi} = - \frac{\partial p}{\partial r} r^2 \textrm{d}r $$

$$ \frac{\Sigma F_r}{r^2 \sin(\theta) \textrm{d}r \textrm{d}\theta\textrm{d}\phi} = - \frac{\partial p}{\partial r} $$ Knowing that the pressure field is $$p(r) = \frac{A}{r}e^{j(\omega t - kr)} $$ gives the equation of motion for the fluid element: $$\rho \frac{\partial u_r}{\partial t} = (\frac{1}{r} + jk)p $$ Assuming a temporally oscillatory solution for $u_r$, $\dot u_r = j\omega u_r$, which gives $$ \rho j \omega u_r = (\frac{1}{r} + jk)p $$ Dividing through by $jk$, $$ \frac{\rho \omega u_r}{k} = (\frac{1}{jkr} + 1)p $$ $$ u_r = \frac{1}{\rho c} (1- j \frac{c}{2\pi fr})p$$

In sum, my original approach failed to account for the radial force contributions of pressure on the $\textbf{e}_{\theta}$ and $\textbf{e}_{\phi}$ sides of the differential element. Careless! It should have been clear right away from the equation of motion that I derived which, for review, was $$ \rho \frac{\partial u_r}{\partial t} = -\frac{2p}{r} - \frac{\partial p}{\partial r} $$ that something was wrong. How can the particle acceleration be dependent on anything other than a pressure gradient? If this were true, then absent a pressure gradient the fluid would accelerate toward the origin under any finite pressure, which is absurd. The bottom line is that $$\rho \frac{\partial u_r}{\partial t} = -\nabla p $$ is the appropriate application of Newton's second law in this case. Hope this helps someone other than me.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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