Green's function for Stokes equation

  • #1
13
0
TL;DR Summary
How do I reach the expression for pressure?
So I've just started learning about Greens functions and I think there is some confusion. We start with the Stokes equations in Cartesian coords for a point force.

$$-\nabla \textbf{P} + \nu \nabla^2 \textbf{u} + \textbf{F}\delta(\textbf{x})=0$$
$$\nabla \cdot \textbf{u}=0$$

We can apply the second relation to the first to get

$$- \nabla^2 \textbf{P} + \textbf{F} \cdot \nabla \delta(\textbf{x})=0 $$
$$\nabla^2 \textbf{P} = \textbf{F} \cdot \nabla \delta(\textbf{x}) $$

The Greens function for the 3D laplacian, according to Wikipedia, is $$-1 \over{4 \pi r}$$ where $$r=(x^2+y^2+z^2)^{1 \over 2}$$ Generally, with Greens functions

$$u(x)= \int G(x,s) f(s) ds$$

So would we get something like

$$\textbf{P}= {-1\over {4 \pi}} \int {1 \over r} \textbf{F} \cdot \nabla \delta(\textbf{x}) ds$$

I believe you might have to convert $$\delta(\textbf{x})={1\over{4 \pi r^2}}\delta(r)$$ I'm a little confused as to where to go next and how to deal with an integral that might have delta functions in it.

Thank you
 

Answers and Replies

  • #2
Pressure is a scalar quantity; it doesn't need a bold symbol.

The Green's function is used to decompose a non-homogenous linear problem [itex]L(u) = f[/itex] into a superposition of problems [itex]L(G_{x_0}) = \delta(x - x_0)[/itex] from which it follows that [tex]u(x) = \int f(x_0) G_{x_0}(x)\,dx_0.[/tex] If we want the solution for a point source of strength [itex]k[/itex] at [itex]x_1[/itex], then we get [tex]
u(x) = \int k\delta(x_0 - x_1)G_{x_0}(x)\,dx_0 = kG_{x_1}(x)[/tex] and we are looking for a constant multiple of the Green's function [itex]G_{x_1}[/itex].

To solve [itex]L(G_{x_0}) = \delta(x - x_0)[/itex], we treat it as a homogenous problem [itex]L(G_{x_0}) = 0[/itex] subject to appropriate conditions on [itex]G_{x_0}[/itex] and its derivatives.

In this case, we want [itex]P[/itex] and [itex]\mathbf{u}[/itex] to be linear in [itex]\mathbf{F}[/itex] and we want [itex]-\nabla P + \mu\nabla^2 \mathbf{u}[/itex] to be proportional to [itex]\nabla^2(1/r)[/itex] with [tex]
\int_{r \leq a} \nabla P - \mu \nabla^2 \mathbf{u}\,dV = \mathbf{F}.[/tex] That leads us to [tex]\begin{split}
P &= \frac{1}{4\pi} \frac{\mathbf{F} \cdot \mathbf{r}}{r^3} = -\frac{1}{4\pi}\mathbf{F} \cdot \nabla\left(\frac1r\right),\\
\mathbf{u} &= \frac{1}{8\pi\mu}\left(\frac{\mathbf{F}}{r} + \frac{(\mathbf{F} \cdot \mathbf{r})\mathbf{r}}{r^3}\right) = \frac{1}{8\pi \mu}\left(\frac{\mathbf{F}}{r} - (\mathbf{F} \cdot \mathbf{r}) \nabla\left(\frac1r\right)\right).\end{split}[/tex]
 

Suggested for: Green's function for Stokes equation

Replies
3
Views
888
Replies
1
Views
661
Replies
5
Views
220
Replies
4
Views
2K
Replies
1
Views
512
Replies
9
Views
3K
Back
Top