Dear Friends
I am doing flow around cylinder at Re=1e7 and want to study the Cd, St and wake area using Fluent/CFD.
Problem description or actual dimensions:
Cylinder dia=1m
cylinder height=3m
Depth of water=12m
so I made a domain size 70D X 30D X 12m(depth of water) (also shown in figure, while height of cylinder=3m)
this is very big domain. I want to reduce it to minimize the computational cost.
Kindly suggest plz...
1.One option in my mind is to make domain depth up-to cylinder height only (3m instead of water depth of 12m) and assign the SYMMETRY boundary condition to the bottom surface.
2. Can I use similarity theory like froude number similarity model or Reynold number similarity model? which will be better
the case study is water flow around cylinder.
Thanks
Hi,
may by you can model just 1/4 of the domain and apply symmetry constrains (if the solution is steady). Then try to reduce the mesh size on the external boundaries.
To understand if the size of the box affect the solution you should perform a parametric study... start from a small box and increment its size until the results does not change any more.
Franco
Thanks Franco for your answer..
The problem is transient and we have to study the kamran vortex, drag coefficient and pressure differences...
Due to this, we can not reduce the model to 1/4th symmetric part... One option may be the scaling the domain using froude similarity or reynold similarity... but I am not very much sure about that....
Regards
Niaz
You would need to perform an analysis whether reducing a dimension would lead to different results than those you achieve with this domain. The width of the domain is 30 x cylinder diameter, I think you would be just fine with a narrower domain. The same goes with the domain length, maybe I would rather make the downstream space longer. Also you may decrease the height of the domain, I do not expect that the wall effect of the bottom boundary would extend as far as the cylinder bottom in your current domain. It should be said that it also depends on the flow regime since the turbulent and laminar flow velocity profiles differ. I guess that your case will be somewhere in the transition area if you are going to study the Karman vortexes.
I do not see scaling the domain as a solution. If I have a large domain and a small domain of the same shape, still I would apply the same number of cells - for example, I would have, let's say, 50 cells around the cylinder circumference, no matter what the real size is.
To conclude, I would run a simulation with your current domain and then reduce the dimension of the cylinder enclosure one by one to see their effect on the results.
Thanks Zbynek for your answer...
you are right . I also think that I should only reduce the depth of the domain to cylinder height (i.e 3m only instead of 12m) since the bottom wall boundary condition will have very little effect on the flow around cylinder ....
@ Niaz Bahadur Khan
Hi Dr. Niaz, I will be very grateful if you can help me to calculate average and local Nusselt number over a cylinder, based on degree (Tetta), as usually FLUENT gives the graph based on length.
Dear Muhammad Reza
If you are able to find any calculations over cylinder surface based on length, then its very easy to convert the same data using simple trigonometry, and plot graph of nusset number versus thetha...
For example,i have cylinder of diameter=2,
, so from fluent, after getting the data of pressure(or any output) versus length(here length is diameter). Copy both x-y data to excel sheet.
Using trigonometry on x-data column(which is of length data)
Thetha= cos(inverse)(base/hyp) where base=x-data and
hypotenuse=radius of cylinder(which will be 1 in this example)
Using this method, you will get thetha for whole x-data column. Then sort the data "smallest to largest".
You will get thetha from 0deg to 180 deg.
- Badges
- Science topic