RANS problem. Cant get any Cd and Cl values.

Quek · February 3, 2016, 04:21

i have a problem of getting the Cd and Cl values and i think is due to the meshing method i use. Can anyone advise in this?

This existing results using Re=20, which calculated at velocity,v, of 0.00029m/s, with the default setting of density and viscosity. My dimension modal is 1m X 1m X 10m, which L=1m, D=10m. The boundary area is about 4D on both sides and height is also 4D from the top of the body.

Meshing method is in default setting. With this mesh i managed the iteration while calculating doesnt have any reverse flow occurs. However my Cd and Cl values are too small which considered to be zero.

I have tried many method example from the given link:
3D Multiblock Structured Hexahedral Mesh
This method given a problem of having reverse flow instead.
Can anyone please advise how to get the meshing correct in order to get my results and compared with the existing results?

Anyway im doing a project to illustrate the air flow around tall buildings.

FMDenaro · February 3, 2016, 04:45

At Re=20 you should work with the laminar model!
Are you sure about the congervenge to a steady state?
Could you post the plots of the solution?

Quek · February 3, 2016, 04:57

Quote:

Originally Posted by FMDenaro

At Re=20 you should work with the laminar model!
Are you sure about the congervenge to a steady state?
Could you post the plots of the solution?

Im using this journal article as my benchmark case.
http://www.math.u-psud.fr/~maury/pap...hmarkTurek.pdf

If is laminar, i tried and Cd and Cl still remain as zero.

At the end of simulation, I will be input velocity 5m/s as my boundary condition using RANS model.

All i can think of is the method of meshing i approached is wrong. Please guide me through if you could. Really appreciated.

adrin · February 3, 2016, 19:13

Filippo is correct; this is clearly laminar flow. You should turn off any turbulence stuff. My first impression is that your choice of using 4D for the far-field is insufficient, even at such low Re number. The downstream free stream requires time/space to reach its "free" state. But, I would have expected your domain to give even higher coefficients!

I'm curious, if you are benchmarking with the reference you provided, why are you not using the geometric definitions provided in the reference. I strongly recommend you start from there; also take a look at the number of grids/meshes used by various contributors. That should give you a hint if, at least, your grid resolution is somewhat acceptable. Other than that, you have not provided much else in terms of information to receive feedback on.

adrin

Quek · February 4, 2016, 10:23

Quote:

Originally Posted by adrin

Filippo is correct; this is clearly laminar flow. You should turn off any turbulence stuff. My first impression is that your choice of using 4D for the far-field is insufficient, even at such low Re number. The downstream free stream requires time/space to reach its "free" state. But, I would have expected your domain to give even higher coefficients!

I'm curious, if you are benchmarking with the reference you provided, why are you not using the geometric definitions provided in the reference. I strongly recommend you start from there; also take a look at the number of grids/meshes used by various contributors. That should give you a hint if, at least, your grid resolution is somewhat acceptable. Other than that, you have not provided much else in terms of information to receive feedback on.

adrin

Thanks for reply. What do you mean to turn off all turbulent stuffs? Can give me some examples? Sorry im still new in this software.
According to the meshing method that provide from other contributors, i tried to use the method given in this link:
http://www.cfd-online.com/Forums/ans...dral-mesh.html

And also tried many methods as well according from different reviews from other journals. As I define my element size smaller, more define, the solution give a reverse flow. IF i use as default method (No edge sizing or face meshing is involved), the solution can be converged but the Cd and Cl is relatively zero.

If let says if i managed to get the desire results from laminar flow, at the end I will be using velocity 5m/s as my basic condition, which calculated Re = 300,000+. With this Re i will need to use turbulent flow model such as RANS model, k-epsilon.

I would like to advice on the meshing method, how will I construct the meshing with my model dimension 1m X 1m X 10m (L X L X H)? How big is my boundary box away from my model? 4H at least?

FMDenaro · February 4, 2016, 11:00

Quote:

Originally Posted by Quek

Thanks for reply. What do you mean to turn off all turbulent stuffs? Can give me some examples? Sorry im still new in this software.
According to the meshing method that provide from other contributors, i tried to use the method given in this link:
http://www.cfd-online.com/Forums/ans...dral-mesh.html

And also tried many methods as well according from different reviews from other journals. As I define my element size smaller, more define, the solution give a reverse flow. IF i use as default method (No edge sizing or face meshing is involved), the solution can be converged but the Cd and Cl is relatively zero.

If let says if i managed to get the desire results from laminar flow, at the end I will be using velocity 5m/s as my basic condition, which calculated Re = 300,000+. With this Re i will need to use turbulent flow model such as RANS model, k-epsilon.

I would like to advice on the meshing method, how will I construct the meshing with my model dimension 1m X 1m X 10m (L X L X H)? How big is my boundary box away from my model? 4H at least?

sorry but your reply is not clear....
Maybe you want first to assess your method for a laminar solution? that's ok but demonstrates only that your discretization is consistent and has some degree of accuracy.

When you use the same method for solving the RANS equations at high Re number, you cannot be sure that things work. You are solving equations having additional (continuous) terms that should take into account for non-resolved flow contribution. That is simply a model and can or can not work independently from your discretization or mesh issues.

Quek · February 5, 2016, 01:14

mesh.jpg

Attached is the mesh method i did. No reverse flow occur, but cd and cl values are zero

Quek · February 5, 2016, 01:21

mesh2.jpg

Attached is another method of mesh. I sliced the box and make the element size more fine near the wall-solid. However this gives me reverse flow. Before that, during updating the mesh, error shows that the all cells zones in fluent may be automatically set to fluid. What does this means?

February 3, 2016, 04:21	RANS problem. Cant get any Cd and Cl values.	#1
Quek New Member Quek Join Date: Feb 2016 Posts: 9 Rep Power: 10	i have a problem of getting the Cd and Cl values and i think is due to the meshing method i use. Can anyone advise in this? This existing results using Re=20, which calculated at velocity,v, of 0.00029m/s, with the default setting of density and viscosity. My dimension modal is 1m X 1m X 10m, which L=1m, D=10m. The boundary area is about 4D on both sides and height is also 4D from the top of the body. Meshing method is in default setting. With this mesh i managed the iteration while calculating doesnt have any reverse flow occurs. However my Cd and Cl values are too small which considered to be zero. I have tried many method example from the given link: 3D Multiblock Structured Hexahedral Mesh This method given a problem of having reverse flow instead. Can anyone please advise how to get the meshing correct in order to get my results and compared with the existing results? Anyway im doing a project to illustrate the air flow around tall buildings.

February 3, 2016, 04:45		#2
FMDenaro Senior Member Filippo Maria Denaro Join Date: Jul 2010 Posts: 6,776 Rep Power: 71	At Re=20 you should work with the laminar model! Are you sure about the congervenge to a steady state? Could you post the plots of the solution?

February 3, 2016, 19:13		#4
adrin Senior Member adrin Join Date: Mar 2009 Posts: 115 Rep Power: 17	Filippo is correct; this is clearly laminar flow. You should turn off any turbulence stuff. My first impression is that your choice of using 4D for the far-field is insufficient, even at such low Re number. The downstream free stream requires time/space to reach its "free" state. But, I would have expected your domain to give even higher coefficients! I'm curious, if you are benchmarking with the reference you provided, why are you not using the geometric definitions provided in the reference. I strongly recommend you start from there; also take a look at the number of grids/meshes used by various contributors. That should give you a hint if, at least, your grid resolution is somewhat acceptable. Other than that, you have not provided much else in terms of information to receive feedback on. adrin

February 5, 2016, 01:14		#7
Quek New Member Quek Join Date: Feb 2016 Posts: 9 Rep Power: 10	mesh.jpg Attached is the mesh method i did. No reverse flow occur, but cd and cl values are zero

February 5, 2016, 01:21		#8
Quek New Member Quek Join Date: Feb 2016 Posts: 9 Rep Power: 10	mesh2.jpg Attached is another method of mesh. I sliced the box and make the element size more fine near the wall-solid. However this gives me reverse flow. Before that, during updating the mesh, error shows that the all cells zones in fluent may be automatically set to fluid. What does this means?