hello, i am new to cfd and fluent. i am simulating unsteady flow past a stationary cylinder for differnt reynolds numbers. but i am having difficulty getting an accurate value of Cd and Cl. i am presently using fluent 6.3. i started by using the laminar model and my fluid is water. what role does the residual plot have in the convergence of simulation? any suggestions as to why my Cd and Cl values are 100% off the ones in the papers? when does one use the grid reorder caption option in fluent?

thank you.

have you set the reference values correctly?

hello, thank u for responding to my message i think i have set the reference values correctly. i set the area= diameter, depth=1, length= diameter. or is this not correct? because i still have Cl and Cd values different from that expected. can u please suggest anything i should try.i need a reply urgently.

thank u

Hi Nkiru!

I think referent values are inportant in here, not only area, depth and length, but fluid properties also.

Jack

Hi, Can I get some more information about the bc you are using? As well as the turbulence model + Re value ? One more important thing you are using "laminar model" and expecting the software to give you Cl Cd values? I think your missing something !

thank you for your reply.

i am currently simulating an unsteady flow past a stationary cylinder in which the fluid is water and the reynolds number is 200(although i am to simulate for different reynolds numbers). therefore i am using laminar model. the fluid (water) characteristics i specified are the density=1000, viscosity=0.001, and velocity=0.002(all in SI units). i have noticed that my answer varies with the time step choosen, such that the higher the time step, the smaller the Cd and Cl values are becoming. i have even tried for Reynolds numbers 500,100 and 185 (all using the same specifications except velocity which is dependant on the reynolds number selected) and it is still the same story.

someone told me that the variations in values are due to the fact that my solution is not converged and that i should try to run more simulations with decreasing values of time step until there is no/very little difference between the values obtained. and this value obtained for Cl and Cd would be the right one because the solution has finally converged. is this true? should i consider doing this? this would mean that each reynolds number will have its own time step size. is that correct?

please i require assistance urgently as i have a deadline.

any suggestions would be appreciated.

thank you.

On this issue how can you get a plot of Cd anf Cl over time for an unsteady flow? I have run the simulation writing this data as wel las writing the time data but cannot find where to generate a plot.

Also, check your solution for mesh convergance, adapt the mesh for areas of high velocity or pressure gradient by adapt-gradient. Choose a refine threshold that refines a sensible amount of cells.

thank you dale for your comments. i will try them and see what happens. fingers crossed.

cheers

Setting residual values to 1e-4 or 1e-6 by experience is the one way of checking for the convergence.

But for some case monitoring some of the flow parameters with respect to the number of iterations ( Time step for Transient) is the best way of judging the convergence.

For your case you may monitor the wall forces / shear forces on the cylinder wall with reference to the time step/Iterations. Once the oscillations are periodic or constant , then the solution is converged.

At the inlet you have to impose the proper values for the Turbulence parameters.Calculate the Turbulent kinetic for the given inlet boundary condition and Impose the same. And dissipation rate should be imposed as zero at the inlet.

This gives a better CFD prediction for Drag.This has given drag values close to theoritical values for ship hull hydrodynamics.