Dear Friends
I am simulating flow around a circular cylinder at Re=3900 ...but I got the value of Cd=1.4-1.6 which is very high compared to the experimental results. In available literature, the Cd=1.0
My problem details are:
Domain size=5m * 3m * 0.2m
Dia of cylinder=0.1m
Density=1.185kg/m^3
Viscousity=1.6e-5 kg/m-sec
Timestep=0.02 which is small enough
In Reference,
Characteristic lenght=0.1m
Projected area= 0.1 * 0.2=0.02
Velocity=0.5275m/s.
The velocity and pressure contour plots are in good pattern as expected, but the Cd and Strouhal number is not looking fine. My St=0.22 where experimental is around St=0.2..
Thanks in advance for help.
Dear Doctor,
It's better to use:
1- INFLATION around the cylinder about 5-7 with growth ratio 10.05-1.1
2- use k-omega turbulent style to solve the case
3- use COUPLE to get the solution of the pressure and the velocity.
GOOD LUCK
1) Check meshing around cylinder .. u did not give any information on it.
2) also what turbulence model u r using?
3) which mathematical scheme r using?
Just some of the factors u need to tune up in your model
In meshing, I used INFLATION layers technique around the cylinder with first node at value of 0.3mm in order to ensure the y+ value
Cd is sensitive to flow so will have different values for laminar and turbulent flow.
k-omega SST turbulent model is ok
try adjusting your y+ to be near from 1.
try Pressure-Velocity Coupling -Phase Coupled SIMPLE.
r u using First Order Upwind for Discretization?
In case of k-w SST, my y+ value < 1,
I am using 2nd order upwind. In above plot of K-w SST model SIMPLE scheme is used..
Actually I am not very much familiar with the uses of mathematical scheme, so as thumb rule, I use higher order Upwind scheme for discretization, with SIMPLE.
In reference vale, I set the Characteristic length=Dia of Cylinder,
Area= Diameter * Depth=0.1*0.2
and so on...
Still not sure with your meshing... as correct inflation meshing with suitable turbulence model should be able to capture flow field accurately provided u have 8-15 inflate layer within BL of your flow to resolve BL, separation and reattachment points.
Plus I can see an oscillatory behavior in all cases above... how are the residuals values and especially the continuity residual? are they also oscillatory like cl and cd?
Did u try this as steady state first? If u did as steady state than was this case also oscillatory? it occur because of incorrect meshing in zone of high gradients e.g. boundary layer. may be its because there is a geometry symmetry but not in the grid meshing.
try same meshed geometry with steady state if not tried with 1 order discretization and check all residual.
Try joining this forum u may find quick and lot of help also : www.cfd-online.com > Forums › FLUENT
Thank you so much Ms Shazia Farman Ali,
I will try the case as steady state and will then let you know.. (unfortunately, my workstation have some problem from last one week due to which I couldnot upload images of mesh).But I remember, the residual plots are in zig-zag form.
Actually, after looking the oscillatory plot of Cd and Cl, my observation was that the simulation is fully converged and now should have accurate result. But now I feel that only oscillatory plot of Cd and Cl are not enough to verify accuracy of results.
Thanks for the Forum suggestion. It is very good forum and very helpful
Now, I made the Meshing symmetry and prepare the O-grid using ICEM. After performing the Steady-state analysis the result is perfect and got the Cd=1.003 which is perfect( I used the kw model SST). But using the same meshed model, the transient result is still over-predicted (Cd=1.4).
congrats! Try following ... (1) start your transient simulations from your steady state results i.e. switch over with smaller time steps.
Remeber there might be difference in two sets of simulations (1) since u have 2d and the phenomena u r capturing is 3d the values of Cd in transient might not be near perfect value (2) also note Cd is strongly influenced by the location of boundary layer separation (3) My observation is that Cd can be overestimated by upto 30% in simulations from experimental results (4) What is the reference depth (in and out of viewing plane z direction) in new simulation, i think default is 1m and is good enough for your simulations and if working SI units ... did u change that value? Is your value less than 1m. Note this might affect your calcutaions also expect a change if u have change in length units
Dear Doctor,
It's better to use:
1- INFLATION around the cylinder about 5-7 with growth ratio 10.05-1.1
2- use k-omega turbulent style to solve the case
3- use COUPLE to get the solution of the pressure and the velocity.
GOOD LUCK
Dear Niaz,
Did you get the correct results for your 2D case? If yes, how did you do that?
Or did you switch to 3D?
Dear Paulina,
Yes I got the correct answer with 2D case... Later on, I switched to 3D study as well to capture the wake characteristics.
Can I ask how did you obtain it?
I am doing similar research and get correct cd value using Curvature Correction and Kato-Launder Limiter for k-omega SST simulation. But then, values of cp differ from the results reported in literature.
My case is very similar to this, but I'm getting a Strouhal number of 0.1, whereas I should be getting in the range of 0.2. What might be the problem? Can you please direct me on the right path to correct my problem?
- Badges
- Science topic