|March 30, 2009, 10:55||
Fluent parameter sensitivity study
I'm a mechanical engineering student, and working on my final project now.
Appreciate if you can give comment and advise to my subject.
My final project tentative tittle is "Sensitivity study of fluent parameter on 2 dimensional airfoil aerodynamics analysis".
The idea is to identify fluent characteristic in dealing with aerodynamic design. Well one of fluent function is as a tool to analyze aerodynamics properties of a design, therefore it is mandatory to well recognize the characteristic of the tool (fluent) so we can insert the required properties properly and resulting close to actual data that will validate the design optimally.
Temporarily I'm limiting the object on 2 dimensional model, infinite airfoil, and on subsonic wind.
But this is still a rough figure, really need to be define properly.
Frankly speaking though I like aerodynamics, I'm still new with CFD and I'm still confuse about "what are the (main) parameters that need to be defined" and later on studied it's sensitivity.
My adviser said it includes meshing properties, turbulence model (I think this one related with the reynold properties), etc.
Really appreciate if you can give me your comment/feedback so I can have a better figure on this subject.
|March 30, 2009, 14:09||
Join Date: Mar 2009
Blog Entries: 17Rep Power: 20
The main parameters affecting the problem on a physics side are:
Angle of Attack (AoA)
Reynolds Number (Re)
Because you are studying an incompressible flow, the Mach Number is not an issue. Anyway, the other two parameters can strongly affect the final solution if the numeric parameters are not set in the proper way. In the following i will assume that the simulations are steady and RANS based.
So, first of all, i'd pick three AoA values (low, medium and near to stall) and two Re values (low and high). So, for a given profile, you will have to perform 6 different groups of simulations; for each of these groups there will be several numerical parameters to change. However, if AoA and Re are assigned there will be only one group of tests.
On the numerical side there are several parameters to change:
Turbulence Model (Spalart-Allmaras, k-eps, k-omega, ecc.)
Pressure Coupling (Simplec, PISO)
Convective Scheme (Second Order Upwind, Third Order MUSCL, ecc.)
Gradient Reconstruction Scheme (green-gauss node based, green-gauss cell based, least squares)
Pressure Interpolation Scheme (Standard, Second Order, Presto!)
Inlet Turbulent Boundary Conditions
y+ value - wall models
Actually, you would need thousands of different simulations to analyze the full set of different conditions, so some choice is needed on which subset of this set of different parameters you are going to analyze.
The grid choise is obviously far from trivial. Because your study is directed toward the aerodynamic project, i would probably choose to use a triangular unstructured grid ensuring, of course, that all the solutions are properly converged on the given grid. In general, i'd concentrate on a single topological type of grid, the one which is best suited for the purpose of your computations.
Hope this helps
|April 3, 2009, 11:15||
After discussed in detail, the objective is to have optimum parameter to get sensitive/detail enough data but also fast enough to simulate/analyze with common computer spec (I'm working on core2duo T6400 2.0 Ghz). This optimum parameter will be use later on as reference to conduct all testing which has the same limitation (2d and on sub-sonic speed)
The testing is on Coefficient of Lift (CL) & Coefficient of Drag (CD) versus Angle of Attack compare to wind tunnel result on infinite wing (2D).
The parametera that will be analyzed are :
1. Mesh density, domain effect, and Yplus . After having the optimum mesh density on structured mesh, then the data will be study to nonstructural mesh, then study the domain effect. Right now I'm searching for mesh manual/theory. and also refreshing my mind bout boundary layer to understand the Yplus and later on domain effect. (please correct me if I'm wrong).
2. turbulence modeling with two models, Sparat-Allmaras and K-Omega. I'm still have no idea bout this turbulence modeling, what is the Sparat-Allmaras and K-Omega, what is their parameter, and what makes them different.
The study will be conducted with several Mach number and Reynold number based on the data which already obtained from wind tunnel testing.
Do you have any link/file regarding the explenation/elaboration of meshing, boundary layer, and turbulence modeling? Including the theory of it. Since this is the main parameter that I will study and I guess this is the basic knowledge that I have to well-understand.
Thanks a lot,
|Thread||Thread Starter||Forum||Replies||Last Post|
|Stopping a Fluent batch job AND saving the data! Possible?||Volker Pawlik||FLUENT||7||May 22, 2014 22:30|
|Comparing Fluent Time dependent result with Patran||venigall||FLUENT||0||January 13, 2007 20:30|
|FLUENT to ANSYS Temperature Mapping Procedure||schreiberc1||FLUENT||0||June 29, 2006 13:50|
|Integration of a Custom C++ Model into FLUENT||Syed Haider||FLUENT||1||February 20, 2006 05:49|
|Advanced Turbulence Modeling in Fluent, Realizable k-epsilon Model||Jonas Larsson||FLUENT||5||March 13, 2000 04:27|