hello

I'm doing a project in which I do the simulation of a flow around a pipeline in 2D. I tried several calculations with Re=10e5 and the SST model but the drag and lift coefficient are lower and the strouhal number is higher. I tried a refinement of the grid and a smaller timesteps but no change. I did it with Re=100 and the results are good in this case.I'm now wondering about the turbulence model and his options. I chose the intensity and length scale option with a fractional intensity of 0.05 and a eddy length scale of 0.1m. I don't know how to choose the better option and his parameter.I hope someone could maybe help me on this point.

Thanks a lot for your help

Are you comparing your results number with other numerical results or with experimental data? In the latter case, be aware that there is no 2D flow over a cylinder at such high Reynolds number. You are missing a complex three-dimensionality, and shouldn't expect to get arbitrarily close to experimental data, even with grid and temporal refinement. You are doing well with Re=100 because that's laminar 2D flow. Anything above about Re=180 will be turbulent and develop three-dimensional shedding modes. How far off is your Strouhal number in the high-Re case ?

I'm comparing the results with experimental data and my Strouhal number is about 0,7 in Re=10e5 instad of about 1.

How are you computing your Strouhal number? I thought the Strouhal number for a cylinder was on the order of 0.2?

sorry I made a mistake I was speaking about the drag coefficient. My Strouhal number is 0,26.

Most of the data I have indicates that for turbulent flow a drag coefficient of 0.7 is not unreasonable, depending on the roughness of the cylinder. What data are you comparing to? Your Strouhal number also falls within the data spread for a turbulent flow. The information I am going by can be found in "Fluid Mechanics" by F. M. White, pp.279-285. Are you using 2nd-order time integration? That can have a large impact on the Strouhal number over first-order time differencing.

You are probably doing this right... but I'll ask anyway: How do you evaluate the drag coefficient? Is it the time-average or R.M.S. over one oscillation period of an unsteady computation, or does your 0.7 come from a steady-state computation?

The fact that you overpredict (not underpredict) the Strouhal number is consistent with my suspicion that your 2D analysis is missing essential flow features. In the low Reynolds number range, with 2D laminar or partially turbulent flow, the Strouhal number increases with Re. At higher Reynolds numbers, when the flow becomes increasingly 3-dimensional and turbulent, Strouhal number levels off at roughly 0.2. If you miss any of the essential physics (either turbulence or 3-dimensionality) you are likely to see your Strouhal number increase with Reynolds number beyond that limit.

I'd do three things: a) Look at the time-average point of separation on the cylinder and compare with experimental data (this is more informative than integrated global quantities), b) get results for various Reynolds numbers between 100 and 1e5 to plot a curve for Strouhal number and drag coefficient versus Reynolds number (this will likely tell you what you are missing, e.g. see if you're missing or overpredicting the drag crisis), b) try to find published articles on similar computations (although most of them will be 3D RANS or LES, which would be the right thing to do at moderate to high Re).

when the curbe becomes stable I evaluate the average value of the drag force on the curve and after the calculation of the drag coefficient is made with Cd=Fd/(0.5*ro*u^2*d*thickness). I ran a transient computation and compared my results with experimental data.

Hai stanc, i can suggest something! u said, u got the correct results for low Re(around 100) but not for high Re(1e5). if u use the same code for both Re, u won't get the results properly, bcoz, at low Re, flow is incompressible and at high Re flow is compressible. And, i think ur code was incompressible one and thus u got good results for low Re, and if u use the same code for high Re without changing any thing u will obviously get the wrong results. so what i can suggest is u plz check ur code once and take care of density and like things while running for low and high Re. hope this might help u a little! by amarnath

Actually the Reynolds number has no bearing on compressibility. The Mach number is the proper nondimensional number to consider when determining compressibility effects.

Ya i agree with u ag! so at high mach numbers compressibility effects comes into picture. since the mach number is high, velocity is also high and thus our Re. so if mach number is increasing, Re also increasing for the same fluid(constant viscosity) and the body(cylinder). so my doubt is y can't we justify the flow based on Re.

And a small help! could u plz tell where i can get the info to generate the C-grid over a circular cylinder for 2D case. or any other sources, if available, wher i can get the grid data itself. any suggestions plz

Because I can have a ship simulation at a Reynolds number of 1e+7 in water and a fighter jet simulation at the same Re. One flow is incompressible and the other is compressible, and transonic. The Reynolds number is NOT the parameter to consider when deciding whether compressibility effects are important. Compressibility effects become important when the flow kinetic energy is roughly the same order of magnitude as the flow thermal energy. This ratio is (essentially) the Mach number, not the Reynolds number.

ok! but i think in both the cases the fluid under consideration is water which is incompressible even at high mach numbers. where as air is incompressible at low mach numbers and compressible at high mach numbers. In the case of ur flight simulation, if the fluid is air at the mentioned Re(1e7), obviously, it is compressible flow. compressibility and incompressibility of a problem at hand depends on the fluid under consideration also! then what is ur openion!

The Reynolds number for a dirigible could be well into the tens of millions due to the length scale. Still flow in air, but essentially incompressible. The point is, the Reynolds number says nothing about compressibility effects.

What Le Stanc is doing is to compare the solution of his incompressible code with experimental data from water flow over a circular cylinder. Compressibility is not an issue, here.

Knowing the basic principles of fluid dynamics you should be aware that Mach number and Reynolds number are independent parameters in the non-dimensional NS equations. To say that high Reynolds number implies high Mach number is simply nonsense.