Hi,

my colleages and I (research students) are thinking of solving the unsteady incompressible 3D NS eqns codes to model flapping wings. hence, the boundary'll be moving. we've been looking thru many schemes and we need some opinons...

1. projection or fractional or artificial compress, SIMPLER which is more suitable? we're abit confused.

2. 2nd order crank nicolson or backward euler for time discretization

3. how to move the boundary? what we know is to rotate the whole grids, suggested by liu & kawachi. is there a better solution?

4. we'll be using structured grids. is it better to write ourselves or get a ready program? Is it efficient to generate 2d grids and stacked them together?

Sorry for asking so many qns. Anyway, thanks in advance!

hi, i m a research student and i have written a 2D code for flapping wings. there can be two methods as 1. either use body fitted grids (Lui and Kawachi) 2. or use immersed boundary method.(my recomendation)

there are many disadvantages in using 1st one. structured grid movement is difficult and if u use multiple bodies grid generation is very tricky (go for block grid generator)

1. SIMPLER algorithm might be slow.

2. 4th order R-K shud be better for time discretisation. 4. Try to use some standard grid generators. Structured grid might be giving some problems near the boundaries

-H

Hi, thanks alot! btw, if SIMPLER is slow, is there a better scheme? what about fractional step/projection or PISO?

Also, will the RK4 be slow since i'm using finite vol?

Thanks again

with RK-$ memory becomes an issue rather than time. Try to look for memory preserving runga kutta schemes.I do not have any idea about projection methods but i would suggest you to read through websites. You can find articles and papers suggesting experience of people with differenct schemes for the specific problem that they had worked on .

-H