I analysed a wing using euler as well as N-S models. I used Spalart Allmaras model for viscous flow. The main problem I am facing is a very high value of coefficient of lift and a very low value of coefficient of drag. For instance I am getting a total value of Cd which is by far less than the theoretical Induced drag coefficient. For example, at angle of attack 2 deg, I am getting a Cl value of 0.3238 but a Cd value of 0.0068 which is less than induced drag predicted by the formula, Cl^2/(3.14xexA.R). I am not able to find out the reason behind this discrepancy. As I am involved in a time bound project, it would be really helpful if anyone shares their views as soon as possible.
Hi Balachander Gnanasekaran,
some years ago I did a project where I had to calibrate these Fluent models according to the wind tunnel tests results at Liege University (Belgium). The difference is that it was an enclosed space and I think yours it is not. First:
. If you are simulating an enclosed volume S-A is not a suitable model.
. If you are simulating an open volume (I suposse this is your case), S-A should fit properly, for compresible flows with high velocity (aeronautical conditions), also in 2D.
I have noticed that in the *.docx you sent, you specify a "Gauge Pressure" equal to atmospheric pressure. Perhaps this is your mistake, you must know that Fluent considers "Absolute Pressure" = "Operating Pressure" + "Gauge Pressure"
If you are setting "Gauge Pressure" equal to atmospheric pressure, you should set "Operating Pressure" to zero, check it and have a look to http://imechanica.org/files/fluent_13.0_workshop02-airfoil.pdf.
I would additionally suggest you to verify the domain size in order to let develop the flow correctly during simulation (avoid reversed flow, etc.), to provide the model with a simple boundary layer and to start with a 2D model of the airfoil until you check the validity of this one, before going into 3D models.
I hope this can help you. Be constant!, , I also passed hard moments in the past with airfoils and so. It will work.
My boundary conditions are as follows.
The domain is a rectangular domain with one surface as inlet, one as outlet, one as symmetry plane and the remaining three as wall.
The inlet, outlet and walls are pressure far fields with gauge pressure 101325Pa and Mach number 0.21. To define the flow direction which is towards the y axis I have used the cosine and sine of the angles of attack that I would be investigating.
The operating pressure is 0Pa. As far as the monitors are concerned.
For Cl,
X=0; Y= -sin(alpha) and z= cos(alpha)
For Cd
X=0, Y= cos(alpha) and Z= sin(alpha)
The aerofoil of the wing is on the Y-Z plane. The chord is 1m and span of wing is 4m. I did meshing with ICEM CFD.
I am using Fluent V11 code. The number of elements in my mesh is 4.8 million with prism layers and 2.6million without prism layers. The domain is rectangular with boundary of domain at 15times the chord length in the trailing and leading region and 8times in the direction perpendicular to chord and 25times the chord in the spanwise direction.
I am using a density based solver .
These are general specifications.
For Navier stokes simulation, I used the S-A model and while defining boundary conditions for pressure far field, I used turbulent viscosity ratio as 10. I selected ideal gas for density parameters and used sutherland for viscosity value.
For flow discretisation I used second order upwind and the same goes for modified turbulent viscosity discretisation also. Courant number 5 and flux type Roe-FDS.
I hope I am not wrong..
How do I resolve the components of force with respect to chord of the wing in fluent ,sir? Omar Qazi
Correction please. Lift perpendicular to the direction of flow and drag along the direction of flow. When you get CL and CD values directly from FLUENT, by default you get the coefficients along the x- and y-axes of the domain. Either use these components to manually resolve in the direction and normal to the flow or else specify the force vector in FLUENT to get the coefficient in that direction.
So, If I understand correctly, are you asking me to specify either one of the two parameters..either the definition of flow direction or changing the force vector in FLUENT and not both? Is that correct?
I am using pressure far field inlet, so it shows me a dialog box where i have to enter gauge pressure, mach number and direction of flow direction. I am giving the flow direction as cos and sine.
Also, when i am specifying the force vector I am specifying the direction of drag as parallel to the flow and the direction of lift as perpendicular. Both in FLUENT only.
OKay...i will try to explain my confusion again. I need to find Cl and Cd values for a particular angle of attack say alpha. I have designated my domain inlet as pressure far field in FLUENT. When I do so, i have the option of setting flow angle in cartesian form. I give the cosine of alpha in the y direction and sine of alpha in the z direction as my flow is in the y direction. Also during my monitor setting, for drag i give cos(alpha) as Y component and sin(alpha) as my z component. and for Cl i give -sin(alpha) as y component and cos(alpha) as z component. I am attaching screenshots of fluent screen. this is for alpha=5deg.
Please advise me.
Can you give some informations about the physical model used, the x/c value and boundary condition used in the spanwise direction and mesh used for the simulation ? This may explain some discrepancies. The lift value seems not so bad with your positive camber wing.
The model is S-A and in the span wise direction the boundary condition is a pressure far field with ambient gauge pressure and temperature. Mesh is unstructured and the number of elements is 4.14 million. with a quality of 10e-2 with prism layers. I am meshing with ICEM CFD software. The chord length is 1m. and the flow is in the Y direction., ie the cross section of the wing is on y-z plane.
Also the wing is in an open environment in the center of your flow domain ? What is the size of the flow domain in chord units ? What is the wing span and aspect ratio (AR = b^2/S) ? Can you send a mesh zoom in a 2D cut around the wing ?
check with a number of standard airfoils from data books with your values on CD and CL first.
Then take another real aircraft wings existing currently and check its CL only. compare with your results.
Also at cruising speeds of existing air crafts find the CD indirectly from the T=D considerations. Note the difference between wing Cd and the aircraft total Cd. You will then get a practical feel for the design values.
Of course the aspect ratios and wing configuration play a role in these values.
Hence you must do your simulations till you strike at the correct values, as different software packages, assumptions, aerodynamic interference effects etc have a significant effect. One must build the confidence that he can design a pair of wing that will make an aircraft fly at cruising or level flight conditions. Further things will follow..
Hi Balachander Gnanasekaran,
some years ago I did a project where I had to calibrate these Fluent models according to the wind tunnel tests results at Liege University (Belgium). The difference is that it was an enclosed space and I think yours it is not. First:
. If you are simulating an enclosed volume S-A is not a suitable model.
. If you are simulating an open volume (I suposse this is your case), S-A should fit properly, for compresible flows with high velocity (aeronautical conditions), also in 2D.
I have noticed that in the *.docx you sent, you specify a "Gauge Pressure" equal to atmospheric pressure. Perhaps this is your mistake, you must know that Fluent considers "Absolute Pressure" = "Operating Pressure" + "Gauge Pressure"
If you are setting "Gauge Pressure" equal to atmospheric pressure, you should set "Operating Pressure" to zero, check it and have a look to http://imechanica.org/files/fluent_13.0_workshop02-airfoil.pdf.
I would additionally suggest you to verify the domain size in order to let develop the flow correctly during simulation (avoid reversed flow, etc.), to provide the model with a simple boundary layer and to start with a 2D model of the airfoil until you check the validity of this one, before going into 3D models.
I hope this can help you. Be constant!, , I also passed hard moments in the past with airfoils and so. It will work.
I think u should try to solve this problem by another numerical method then compare the error u got it by eulare form and as example rung kutta method or any other numerical method..
The best solution to tackle this would be as follow
1. Modify your mesh to obtain a y+ value to fall in the log law layer. Adding prism layers too close to the wall would calculate your co-effecients using a different method (viscous sub-layer) giving you a shift in the values as expected.
2. Set the turbulence parameters correctly based on the physics and geometry. The Fluent database will give you the method to evaluate the same and this would also help in better convergence.
Hope it works!
Hello...
Cd is very sensitive to flow regime.it means that, Cd in laminar and turbulent regime can have different values. if considerable part of flow over the wing is laminar, its better to use transient SST k-omega turbulent model, which also accounts for laminar effects.
Hello..
1) for Cl value, try adding more grid points along the surface of the wing, Did ur results converge for simulations done at 0 degree angle of attack?
2) for Cd, its usually, the boundary layer, cross check the calculations and you should be good
Cl and Cd are normalized by reference area. so make sure you have entered the proper reference area. Also make sure you have entered the proper sin and cos values so fluent compute the correct contribution to get the lift and drag force
Adjust your y+ to be near from 1 and use k-omega SST turbulence model.
Hello i have the same problem with a 2d simulation of DU 97 W 300 :
The chord of the model il 0.6m and i have only the reynolds number RE=3000000 of the experimental simulation ; i used a spalart-allmaras model in density based,with density ideal gas and viscosity sutherland;i exctract the velocity from the reynolds number with density e viscosity value for T=300K.
after that i used pressure far field for inlet and outlet specifying the cos and sin of angle of attack the temperature and the gauge pressure(0 Pa).
The operating pressure i took 101325 Pa and i set the turbulent viscosity ratio to 10
The Cd and Cl are not the same of the experimental.
someone can help me?
Please do the problem with two high aspect ratios and one with a lower aspect ratio and see whether the same percentage error exists
- Badges
- Science topic