Hello. I am solving the flow past n cylinders placed axis parallel. The flow is a compressible, viscous flow. Can you suggest good schemes for solving this. I will be starting with a single cylinder first. I have limited computational power.. A pentium computer with 32Mb ram. So please take that into account too.

I am currently solving the exact problem as part of my dissertation. Compressible viscous flow over single cylinder, both fixed and rotating. Mach number 0.5, Reynolds number 350,000 based on a diameter of 0.04m at low ambient pressure 700hPa (1013hPa is sea level).

I am using Harten's UTL1C TVD algorithm on a curvilinear grid for the inviscid part and central differences for the viscous portion of the computation. I have a Pentium processor (200Mhz) with 128MB of RAM. For a 60 by 60 mesh, the computation takes between 1 to 2 days, because of the boundary layer resolution ( delta/dia ~ Re^-0.5 ).

I would give you the source code, but I am planning to turn this into a commercial software package next year. What Mach number and Reynolds number are you looking for? I may be able to run one test for you.

I will send you a GIF image of one of my rotating cylinders.

Your biggest task will be the grid generation for multiple cylinders. Do you have a program for that?

Take care,

Axel Rohde

if you have limited computational resources an explicit scheme like jameson's 4 step runge kutta scheme is appropriate. you should note that he made improvements to the scheme (the artificial dissipation) since his 1981 paper that most people reference