Hi netters,

I'm tring to study Magnus Force on infinite (2D) cylinder using Fluent, but results I obtain doesn't match sperimental datas: in particoular I can't see Negative Magnus force that I should have during the transition from laminar-tourbulent to fully developed toubulent motus, the results Fluent gives to me for both Cd and Cl coefficient (note that I inserted in Fluent correct reference values for my problem) are too far from experimental results, even with no spin for the cylinder (Cl goes rightly to 0 but Cd doesn't give me good results (it should change with Reynolds number during the transition written before, but it's very far from the range it should belong: sperimentally from ~1.3 to 0~0.35, with Fluent always near 0.15-0.25)

I setted Fluent with uncompressible fluids (air max speed is near 20m/s), RSM model with standard coefficients, standard wall function for near wall treatment, double precision solver.

Mesh is very well done with Gambit, near cylinder I used Map scheme and first lenght is 1*10^-5, I can see very well velocity profiles changeing and stall point, I'm quite sure it's not a mesh problem.

I read in some old posts that to simulate more accurately magnus force you should change a coefficent in k-e model, but i don't know which and the correct value, maybe it's the same for RSM model (it solves k-e equations too....).

So at the end of this: can someone help me? did someone already studied Magnus force with Fluent (I can use both 5.4.8 and 6.0.12) and had good results and could help me?

If you need more details please ask.

Thank you very much Michele Spinolo

Fluent can not predict transition - if you turn on a turbulence model you will most likely get fully turbulent bouyndary-layers from the start. If you are trying to simulate a regime where boundary layer transition is important you will most likely run into trouble.

Btw, have you made sure that you have the first grid cell at y+ between something like 30 and 200 (a must if you use wall-functions)?

Y+ is "wall Yplus" I can find in display/contours/turboulence? If it's this I have a value under 5 near the 2D cylinder, the dominium I created is 60x20 and the cylinder has diameter 1.4.

I tried to simulate transition from laminar to tourbulent becouse I have this kind of experimental datas (from Re=~0 to Re=5*10^5), and i can't validate other simulation.

Could someone give me more Cd (and if someone have Cl too) for a spinning or at least non spinning cylinder in a fully developed tourbolent flow? (giving Re number too?)

Yep, that is the right Yplus - if it is below 5 in the cells next to the cylinder surface your wall-functions will not work very well. You need to have y+ above 20 (and below something like 200 - varies depending on the Re number) for wall functions. You can either refine your mesh so that y+ is below 1 everywhere and use the Low-Re Wolfstein two-layer model or you can coarsen your grid next to the walls to get y+ bigger than 20 so that you can use wall-functions. Fluent 6 has a new wall-treatment which they claim can swith seamlessly between these models to work on any grid. I haven't checked how it performs in the difficult case when you have y+=5 though (anyone?).

If you are looking for other good test-cases you can take a look at the links in the reference section.

I have Y+ under 1 near the cylinder, but it's higher than 1 (from 1 to 3000) near wall domains, now i'm tring to use 2 layer zonal model as wall treatment, do you think I should have Y+ under 1 also in the zone near domain's wall (it's very far from cylinder)?

Thanks

If it is a wall you should, but it will probably not affect your results close to the cylinder. Perhaps it is better to set it to a symmetry boundary - then you don't have to worry about y+.

Jonas, thank you very much! great idea to set as a symmetry boundary condition!

near the cylinder I have Y+=0, i did some run with 2 layer zonal model (more than 4000 iterations) but i get Cd=0.16 instead of Cd=0.4 (experimentally), the cylinder is not spinning. I don't really know why...or better maybe I know why: with 2 layer zonal model i don't have the negative pressure in the rear of the cylinder (opposite part of flow inlet) that i have with standard wall treatment...

Another question (you are very patient) how can i obtain a higerh value of Y+ near the cylinder (with a more larger mesh?) so I can use standard wall treatment?

Thankx Michele