[quarkz]

Hi,

I need to simulate a flow past a vertical infinitely thin plate. One way to do it is to have a Cartesian grid, use a single column of cells to represent the wall.

However, I am thinking of using a line to represent it instead. This is because it's supposed to be a very thin wall and I don't want to add thickness to the problem.

I am using the fractional step which has the momentum and poisson equations. I have no problem specifying the BC for momentum but I am not sure about the poisson equation.

If I use dp/dn=0 at the walls, my answer will just explode. How can I solve this problem? I've heard of using jump conditions but I can't find a paper which is suitable for my case.

Can someone help?

Thanks!

[Jonas Holdeman]

I assume you are talking about 2D horizontal flow past a vertical plate. I have done something like this using separate blocks on the left and right. This allows BCs on the left and right of the plate to be independent like the column of cells, but the plate has zero thickness.

However, I use a pressure-free momentum equation with Hermite finite elements, so your remark about dp/dn is not an issue. Come to think of it, the line option would work just as well. I have used something like the line option to model flow in a room from a (2D) ceiling fan where the fan was a flow-through line element instead of a barrier (flow controlled by the stream function difference along the fan). In this case there is a pressure jump somewhere.

[quarkz]

Thanks Jonas for the reply. I will look into it.