Biot-Savart law¶
References
Saffman [Saf95]
Fig. 2 Biot-Savart law¶
Let us consider the following conditions for a rotational flow: \begin{enumerate} \item The velocity field, \(\mathbf{v}\), satisfies \(\nabla \cdot \mathbf{v} = 0\). \item The fluid region is singly connected. \item The normal component, \(\mathbf{n} \cdot \mathbf{v}\), of the velocity is given at all bounding surfaces \(S\). \item The velocity \(\mathbf{v}\) vanishes at infinity when the fluid is unbounded. \item The normal component, \(\mathbf{n} \cdot \boldsymbol{\omega}\), of vorticity vanishes on \(S\). \item The vorticity field \(\boldsymbol{\omega}\) is compact when the fluid is unbounded. \end{enumerate} The velocity field under these conditions can be expressed as the sum of a solenoidal velocity potential component, \(\mathbf{v}_{v}\), for which \(\nabla \cdot \mathbf{v}_{v} = 0\), and an irrotational scalar component, \(\nabla \phi\), i.e.,
where the irrotational component is determined by the Poisson equation:
with the boundary condition
or
when the fluid is unbounded. Taking \(rot\) of Eq. (38) yields
since \(\nabla \times \nabla \phi = 0\). The solenoidal component is constructed by the vorticity distribution (\ref{Auton_Auton-BiotSavart})
where \(\mathbf{r}' = (x', y', z')\) and \(dV' = dx' dy' dz'\). This is analogous to the relation between a electric current density and a magnetic flux field. Eq. (43) is therefore called the Biot-Savart law for the fluid velocity induced by the vorticity distribution. Since
we can rewrite Eq. (43) in the following form
Let us see the Biot-Savart velocity field satisfies \(\nabla \cdot \mathbf{v}_{v} = 0\) and \(\nabla \times \mathbf{v}_{v} = \boldsymbol{\omega}\) in the following. For simplicity, we write \(\boldsymbol{\omega}' = \boldsymbol{\omega} (\mathbf{r}', t)\), \(\nabla' = \partial / \partial \mathbf{r}'\) and \(r = | \mathbf{r} - \mathbf{r}' |\). Taking \(div\) of Eq. (45) we have
We can take \(\boldsymbol{\omega}'\) out from the divergence since it is a function of \(\mathbf{r}'\), not \(\mathbf{r}\). Therefore,
\(- \nabla \cdot \left( \boldsymbol{\omega}' \times \nabla \frac{1}{r} \right) \rightarrow - \frac{\partial }{\partial x_{k}} \left( \epsilon_{kij} \omega'_{i} \frac{\partial }{\partial x_{j}} \frac{1}{r} \right) = - \omega'_{i} \epsilon_{kij} \frac{\partial }{\partial x_{k}} \left( \frac{\partial }{\partial x_{j}} \frac{1}{r} \right)\) \(= \omega'_{i} \epsilon_{ikj} \frac{\partial }{\partial x_{k}} \left( \frac{\partial }{\partial x_{j}} \frac{1}{r} \right) \rightarrow \boldsymbol{\omega}' \cdot \left( \nabla \times \nabla \frac{1}{r} \right)\)
However, because of the identity \(\nabla \times \nabla~\text{(any scalar)} = 0\),
The Biot-Savart velocity field is thus confirmed to be solenoidal. Then, we take \(rot\) of Eq. (45):\footnote{
}
\(\nabla^{2} (1/r)\) behaves like Dirac’s delta, i.e., \(\nabla^{2} (1/r) = - 4 \pi \delta (\mathbf{r} - \mathbf{r}')\) (see Appendix \nabla^{2} (1/r) behaves as Dirac’s delta). Therefore the integration of the first term gives \(\boldsymbol{\omega} (\mathbf{r})\):
For the second term, using \(\nabla (1/r) = - \nabla' (1/r)\) gives
where \(\nabla' \cdot \boldsymbol{\omega}' = 0\) was used. By the divergence theorem, we have
However, this integral vanishes because of the condition \(\mathbf{n} \cdot \boldsymbol{\omega} = 0\) on all bounding surfaces or the compactness of the vorticity field. Thus,