CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   Airfoil 2D help (http://www.cfd-online.com/Forums/fluent/98854-airfoil-2d-help.html)

 pranavom8 March 20, 2012 16:30

Airfoil 2D help

1 Attachment(s)
Hello ppl,
I'm trying to do a 2D airfoil analysis using ansys fluent, I used the 3 part tutorial video to mesh a CH type mesh in ICEM CFD, when I import into fluent the mesh imports fine, I did scaling to 0.254 to x and y to convert the size to inches(my chord length is 10"), after that when checked for mesh it shows errors as show in the attached picture. When neglected and continued using steps from
https://confluence.cornell.edu/displ...irfoil-+Step+4
while running the solver it converges till 20 iterations and after that it starts throwing fatal error Reversed flow.
What am I doing wrong?
Is there anything wrong with the mesh,

 Zigainer March 22, 2012 10:11

You could lower the under-relaxation factors. And I always start my problems with less mass flow (lets say 1/3) and then increase mass flow slowly.
What are your boundary condition? Pressure inlet? Pressure outlet? Operating pressure?

 pranavom8 March 23, 2012 06:43

Quote:
 Originally Posted by Zigainer (Post 350897) You could lower the under-relaxation factors. And I always start my problems with less mass flow (lets say 1/3) and then increase mass flow slowly. What are your boundary condition? Pressure inlet? Pressure outlet? Operating pressure?
I have defined the walls are pressure fields and in the fluent I defined them as velocity inlet and pressure outlet, with a velocity of 20ms^-1, and the pressure outlet set to guage, 0Pa. I tried reducing the under relaxation factors from 0.3 to 0.2 and 0.1 but that didnt help. The chord length of my model is 1m in the ICEM CFD(leading edge 0,0,0 and trailing edge 1,0,0) I used the scaling in fluent to scale down to 10" by using 0.254 in the places of x and y in the scale.
Thanks in advance for any help

 robboflea March 23, 2012 10:24

Try to initialize your solution with FMG initialization: you find it from command line menu in Fluent under 'solve/initialize/FMG initialization'. This generally solved all my problems concerning backflow. Let me know if it doesn't work.

Cheers

Rob

 pranavom8 March 23, 2012 10:54

Quote:
 Originally Posted by robboflea (Post 351111) Try to initialize your solution with FMG initialization: you find it from command line menu in Fluent under 'solve/initialize/FMG initialization'. This generally solved all my problems concerning backflow. Let me know if it doesn't work. Cheers Rob
Sure Rob thank you!!!

 pranavom8 March 25, 2012 03:07

I still haven't solved, but for anyone who has this same problem, you could use this information in the site
https://www.sharcnet.ca/Software/Flu.../tg/node53.htm

 LuckyTran March 25, 2012 04:02

Quote:
 Originally Posted by pranavom8 (Post 350503) Hello ppl, I'm trying to do a 2D airfoil analysis using ansys fluent, I used the 3 part tutorial video to mesh a CH type mesh in ICEM CFD, when I import into fluent the mesh imports fine, I did scaling to 0.254 to x and y to convert the size to inches(my chord length is 10"), after that when checked for mesh it shows errors as show in the attached picture. When neglected and continued using steps from https://confluence.cornell.edu/displ...irfoil-+Step+4 while running the solver it converges till 20 iterations and after that it starts throwing fatal error Reversed flow. I dunno where I'm going wrong please help, my angle of attack tested is 8degrees What am I doing wrong? Is there anything wrong with the mesh,
Quote:
 Originally Posted by pranavom8 (Post 351081) I have defined the walls are pressure fields and in the fluent I defined them as velocity inlet and pressure outlet, with a velocity of 20ms^-1, and the pressure outlet set to guage, 0Pa. I tried reducing the under relaxation factors from 0.3 to 0.2 and 0.1 but that didnt help. The chord length of my model is 1m in the ICEM CFD(leading edge 0,0,0 and trailing edge 1,0,0) I used the scaling in fluent to scale down to 10" by using 0.254 in the places of x and y in the scale. Thanks in advance for any help

I am not familiar with ICEM but,

Fluent imports cell coordinates in the default units of meters [m]. 1 inch contains 25.4 mm or 0.0254 m. If you did a scaling of 0.254, then your scaling is off by a factor of 10.

Check the actual physical span of your mesh. Go to the scaling window, and check if the min and max x and y coordinates correspond to the actual extent of the domain.

 robboflea March 25, 2012 05:25

Sometimes I had problems when using 0 Pa as a reference condition. Try to see if you get something when using a non-zero operating condition.
Could ypu please specify what are your boundary conditons on the top and bottom walls of the domain?

Cheers,

Rob

 Tom Lucius March 28, 2012 21:37

Hi, i am running 2D flow over NACA 0012 simulation based on this tutorial https://confluence.cornell.edu/displ...il+-+All+Pages.

The default chord length i get is 1 m.
Means if i am analyze based on Reynolds number = 360,000
Re= (Density X Chord length X Velocity)/(dynamic viscosity)
Density = 1.225
Chord Length = 1m
Re= 360,000
Viscosity = 1.7894 x 10 ^-5

The velocity i get is 5.2586 m/s .

With this velocity input, i cant get the accurate drag coefficient.

Anyone help me to remove my doubts?

 LuckyTran March 28, 2012 21:47

Quote:
 Originally Posted by Tom Lucius (Post 352052) Hi, i am running 2D flow over NACA 0012 simulation based on this tutorial https://confluence.cornell.edu/displ...il+-+All+Pages. The default chord length i get is 1 m. Means if i am analyze based on Reynolds number = 360,000 Re= (Density X Chord length X Velocity)/(dynamic viscosity) Density = 1.225 Chord Length = 1m Re= 360,000 Viscosity = 1.7894 x 10 ^-5 The velocity i get is 5.2586 m/s . With this velocity input, i cant get the accurate drag coefficient. Anyone help me to remove my doubts?
Are the drag coefficients calculated by Fluent? If so, did you remember to specify the reference values?

 Tom Lucius March 28, 2012 22:35

Quote:
 Originally Posted by LuckyTran (Post 352053) Are the drag coefficients calculated by Fluent? If so, did you remember to specify the reference values?
Re = 360,000
Angle of attack = 4 degree

Type of Solver = Density Based

Viscous = Inviscid

Boundary Conditions:
Inlet = Velocity inlet
Outlet = Pressure outlet

X velocity = 5.2586 cos 4 = 5.2458
Y velocity = 5.2586 sin 4 = 0.3668

Reference Value:
Compute from inlet

Residual continuity = 1e-6

Solution Steering:
Flow type = Incompressible

My result:
Lift = 0.411
Drag = 0.00377

Published paper result:
Lift = 0.44
Drag = 0.0098

 LuckyTran March 28, 2012 22:40

Quote:
 Originally Posted by Tom Lucius (Post 352061) Reference Value: Compute from inlet
That didn't really help...

Where is this magic lift and drag coefficient coming from? Is that the Fluent output or did you calculate it? It seems like you are just taking whatever Fluent is spitting at you, is this the case?

Have the solver report the raw Forces on the surfaces and calculate (by hand + calculator) independently and see if you can the same lift and drag coefficients as the solver, even if they do not match published literature. This way you are at least sure the coefficients are being calculated properly (that is Lift = 0.411 and Drag = 0.00377).

And what are your reference values? You need to define all the correct reference values to get the right coefficients.

You need to have the reference area, density, and velocity specified consistent with your problem. I do not think the reference length matters (they are not used in the calculation of the lift & drag coefficient), but you can set that to be your chord length too if it helps.

 Tom Lucius March 28, 2012 23:07

1 Attachment(s)
Quote:
 Originally Posted by LuckyTran (Post 352063) You are not really helping... Where is this magic lift and drag coefficient coming from? Is that the Fluent output or did you calculate it? It seems like you are just taking whatever Fluent is spitting at you, is this the case? Have the solver report the raw Forces on the surfaces and calculate (by hand + calculator) independently and see if you can the same lift and drag coefficients as the solver, even if they do not match published literature. This way you are at least sure the coefficients are being calculated properly (that is Lift = 0.411 and Drag = 0.00377). And what are your reference values? You need to define all the correct reference values to get the right coefficients. You need to have the reference area, density, and velocity specified consistent with your problem. I do not think the reference length matters (they are not used in the calculation of the lift & drag coefficient), but you can set that to be your chord length too if it helps.

I thought the lift and drag coefficient is obtained from Fluent. Maybe I am misunderstood the tutorial.

As you mention, after I getting Fluent result (lift & drag coefficient), I must calculate by myself to compare with the Fluent result?

 LuckyTran March 28, 2012 23:19

Quote:
 Originally Posted by Tom Lucius (Post 352066) I thought the lift and drag coefficient is obtained from Fluent. Maybe I am misunderstood the tutorial. As you mention, after I getting Fluent result (lift & drag coefficient), I must calculate by myself to compare with the Fluent result?
As I mentioned, you need to specify the correct reference values in order to get any meaningful results from lift and drag coefficients. At the minimum, for your problem, reference area, density, and velocity should be specified consistent with your definition of lift and drag coefficient. Use the same definition as your "published paper" so that you can directly compare the two.

I doubt your reference area is 1 so it appears to be the first thing that is wrong. For reference area, I am not sure which area is being used in your "published paper" definition. There are many ways to define these coefficients, some based on planform area and some based on projected frontal area. You will have to look it up and then specify the correct reference area.

Make sure that the reference density and velocity are correct! Don't be lazy. Also do not assume that the reference density and velocity are the same as the material property or boundary condition. Double, triple check them!

You can always double check by calculating it by hand as I mentioned earlier to make sure everything is making sense.

 robboflea March 29, 2012 02:17

Quote:
 Originally Posted by Tom Lucius (Post 352061) Viscous = Inviscid
Sorry but I'm missing something. Are you trying to calculate the drag by using a Euler solver? Does Fluent use a kind of magic to do that?

Rob

 Tom Lucius March 29, 2012 04:34

Quote:
 Originally Posted by LuckyTran (Post 352069) As I mentioned, you need to specify the correct reference values in order to get any meaningful results from lift and drag coefficients. At the minimum, for your problem, reference area, density, and velocity should be specified consistent with your definition of lift and drag coefficient. Use the same definition as your "published paper" so that you can directly compare the two. I doubt your reference area is 1 so it appears to be the first thing that is wrong. For reference area, I am not sure which area is being used in your "published paper" definition. There are many ways to define these coefficients, some based on planform area and some based on projected frontal area. You will have to look it up and then specify the correct reference area. Make sure that the reference density and velocity are correct! Don't be lazy. Also do not assume that the reference density and velocity are the same as the material property or boundary condition. Double, triple check them! You can always double check by calculating it by hand as I mentioned earlier to make sure everything is making sense.

The chord length which I used in Fluent is by default (1m).
The published paper's chord length is 15.24 cm and span is 0.91 m.
I not good in using Fluent, so I do not know how to make the chord length from 1 m to 15.24 cm, and also the span.
Is there any idea or advice for me?
My NACA 0012 shape is formed by the list of coordinates which provided by the tutorial.

 robboflea March 29, 2012 05:08

Assuming that your geometry is correct you don't need to change it! you only need to change the reference values in Fluent. Those values are used by Fluent to calculate meaningful non-dimensional parameters like Re or Lift and Drag coefficients.

For info on houw to change them please refer to Fluent manual.

Cheers,

Rob

 LuckyTran March 29, 2012 12:34

Quote:
 Originally Posted by Tom Lucius (Post 352110) The chord length which I used in Fluent is by default (1m). The published paper's chord length is 15.24 cm and span is 0.91 m. I not good in using Fluent, so I do not know how to make the chord length from 1 m to 15.24 cm, and also the span. Is there any idea or advice for me? My NACA 0012 shape is formed by the list of coordinates which provided by the tutorial.
Sorry, what I meant was that you should use the same "type" or reference area/length as the published paper. However, define the reference values according to your model.

i.e. if they use the chord length as the reference length, then you also use the chord length. if they use planform area, then you also use planform area, etc. But when it comes to the actual values, use your chord length (not the paper's chord length).

You don't need to change your chord length, you just need to make sure that the right reference values are defined so that Fluent calculates your coefficients properly. Hope that helps.

 pranavom8 April 10, 2012 02:59

Mesh file

1 Attachment(s)
Hello people I have included the mesh file in the zip file attached, I still get the flow reversal problem :confused: And I have no clue about the reference are and lengths modification. The boundary conditions can be verified from the file attached within.
Hope anyone could help.

 robboflea April 10, 2012 05:19

http://hpce.iitm.ac.in/website/Manua...g/node1208.htm

from the 2nd result from google's when searching for "fluent reference values". What's not clear in that? Use the reference values of your profile but using the same definition of drag and lift coefficients used in the reference paper.

For what concerns your mesh for me it has some quality problems starting from the too big density step on the leading edge, the high aspect ratio of the cells far from the airfoil, the cells to big on the outlet boundary.

By the way this couldn't be the only problem to your simulation. What are your boundary conditions? What solver are you using? I read a few posts above that you were using a Euler solver? Was that a mistake or is that true?

All times are GMT -4. The time now is 12:15.