Hi again, I have a version of channelflow running on a mac (still working on getting it to run on our linux computer), and I'm starting to work on modifying the code to simulate boundary layer flow. I'm not sure if I should put this under the development section, if so I can move it.
Before I get too into changing the code, I wanted to run my idea by you just to see if it sounds reasonable and like the best way to approach things.
The simplest way is to put the top wall very far away, but that's not feasible. So I need to create boundary conditions at the upper wall that mimic what the flow would do if the wall wasn't there. My idea is to extend the way the baseflow is implemented so that it can vary in x, y, and z, rather than only in y. To do this, I thought that rather than defining variable “U” as type ChebyCoeff, I would define it as type FlowField. I'd then create methods that could take U as a FlowField. Another option would be to make a baseflow class, and it would deal with both cases (U as a ChebyCoeff or U as a FlowField), but that could be overly complicated.
I'm still learning the code, so I'm not entirely sure how much assumes U is ChebyCoeff, or if it will be easily changed to handle U as a new data-type. Do you think this approach would work, or have any other comments?
Brandt Belson, 2009-03-16 12:50