Stokes solved this 1851. I have a question regarding the derivation. Following Batchelor the equations to be solved are \begin{align} \nabla \left( \frac{p - p_0}{\mu} \right) = \nabla^2 \vec{u} = -\nabla \times \vec{\omega} \\ \nabla \cdot \vec{u} = 0 \end{align} with boundary conditions $\vec{u} = \vec{U}$ at the surface of the sphere where U is the velocity of the sphere in z-direction. Also $\vec{u} \rightarrow 0, p-p_0 \rightarrow 0 \quad {\rm as} \quad r \rightarrow \infty$. The center of the sphere is instantaneous at the origin of the co-ordinate system. $\vec{\omega}=\nabla \times \vec{u}$ is the vorticity and $\mu$ the viscosity. Since $p-p_0$ solves the Laplace equation it can be written as a series in solid harmonics \begin{align} \frac{p-p_0}{\mu} = \sum_{l=0}^{\infty} \sum_{m=-l}^{l} c_{lm} I_{l}^m \quad {\rm .} \end{align} The m-sum is absent because of the azimutal symmetry. He argues that $p-p_0 \propto \vec{U} \cdot \vec{r} \propto \cos \theta$ and therefore the only term $\neq 0$ is $l=1$. I cannot see why $p-p_0 \propto \cos \theta$ should be fulfilled though.

  • $\begingroup$ Is it consistent with the differential equations and boundary conditions? $\endgroup$ – Chet Miller May 30 '18 at 11:15
  • $\begingroup$ Is what consistent? $\endgroup$ – Diger May 30 '18 at 11:27
  • $\begingroup$ The pressure dependence on the cosine of the angle. $\endgroup$ – Chet Miller May 30 '18 at 12:04

Stokes equation is linear in its dependent variables. Further its boundary condition is linearly dependent on translation velocity $\mathbf{U}$. This means that if $\{\mathbf{u},p\}$ is the solution of Stokes equation for translation velocity $\mathbf{U}$, then if the translation velocity is scaled to $\alpha\mathbf{U}$, the new solution solution corresponding to this new boundary condition is simply $\{\alpha\mathbf{u},\alpha p\}$. This suggests that both $\mathbf{u}$ and $p$ must be linear functions of $\mathbf{U}$.

In 3-dimensions, a solution of the Laplace equation is $1/r$, i.e. $\nabla^2(1/r)=0$, in which $r$ is the radial distance to a point in the flow field from the sphere center. Further solutions to the Laplace equation may be generated by taking successive gradients of $1/r$. Of course each new solution generated in this way is of higher tensorial rank than the previous one. The solutions are: $$\frac{1}{r},\nabla\frac{1}{r},\nabla\nabla\frac{1}{r},\ldots $$

Now the pressure field, which is a scalar field, must be linear in $\mathbf{U}$ while involving one (or more) of the solutions listed above. The only viable combination is $$p=\lambda \mathbf{U}\cdot\nabla\frac{1}{r}=-\lambda\frac{\mathbf{U}\cdot\mathbf{r}}{r^3}$$ in which $\lambda$ is a constant. This is how $p$ (or $p-p_0$, if your datum is not zero) is found to be proportional to $\cos\theta$, in which $\theta$ is the angle between the direction of motion and the position vector of a point in the flow field.

Reference: Guazzelli & Morris, A Physical Introduction to Suspension Dynamics, Cambridge University Press.

  • $\begingroup$ Why is it necessary that p depends on $\vec{U}$ via scalar product and not just in terms of the absolute value? It would still scale with $\alpha$. $\endgroup$ – Diger May 30 '18 at 17:39
  • $\begingroup$ @Diger Because the boundary condition involves the vector velocity and not its magnitude. If $p$ depended on the absolute magnitude $U$ then so would the flow velocity $\mathbf{u}$, via Stokes equation ($p~\alpha~U/r$ and so $\nabla p~\alpha~U\nabla (1/r)$). It would then not be possible to satisfy the boundary condition. $\endgroup$ – Deep May 31 '18 at 7:35
  • $\begingroup$ Another way to see it: Consider 2 distinct cases, in which sphere moves with velocity $\mathbf{U}$ and $\mathbf{U}'$; both have same magnitude $U$ but different directions. The corresponding solutions are $\{\mathbf{u},p\}$ and $\{\mathbf{u}',p'\}$. Now consider a third case in which the sphere moves with velocity $\tilde{\mathbf{U}}=\mathbf{U}-\mathbf{U}'$, whose solution is $\{\tilde{\mathbf{u}},\tilde{p}\}$. If pressure depended on magnitude of translation velocity then we must have $p~\alpha~U,p'~\alpha~U,\tilde{p}~\alpha~|\mathbf{U}-\mathbf{U}'|$. $\ldots$ $\endgroup$ – Deep May 31 '18 at 9:55
  • $\begingroup$ But since Stokes equations are linear we must also have $\tilde{\mathbf{u}}=\mathbf{u}-\mathbf{u}',\tilde{p}=p-p'$. But then $\mathbf{u}-\mathbf{u}'$ and $p-p'$ do not contain a term proportional to velocity magnitude. To avoid this contradiction we must suppose that the flow velocity and pressure do not depend on velocity magnitude. $\endgroup$ – Deep May 31 '18 at 9:55
  • $\begingroup$ @Diger Batchelor has given in Chapter 3 of his book the derivation that constitutes my answer. $\endgroup$ – Deep May 31 '18 at 10:01

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.