CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   CFD Technology Issues (http://www.cfd-online.com/Forums/main/26-cfd-technology-issues.html)

CFDsurvey July 3, 1998 11:46

CFD Technology Issues
 
Hello,

We are a group of CFD professionals who believe that CFD technology is currently under-utilized within industry and government sectors. We believe specific technology issues are at fault and your response to the questions below would assist us in deciding whether additional development energies should be focused in the area indicated below. Thank you for your inputs.

1) How frequently are you tasked with performing a CFD analysis?

(a) occasionally

(b) regularly

(c) full-time

2) What type of analysis codes do you use most?

(a) structured

(b) unstructured

(c) both

3) What are the major difficulties with performing your analyses?

(a) pre-processing (geometry & grid generation)

(b) solver setup (boundary conditions, inputs)

(c) solver execution time

(d) post-processing (visualization, data extraction)

Regarding your pre-processing methodology:

4) What preprocessing software do you currently use?

5) Are you satisfied with the current state of your pre-processing software?

6) Are their adequate tools to perform the desired modeling tasks?

7) Is the CAD interface seamless?

8) Do the tools as implemented assume that you are a mesh generation expert?

9) Are the available tools intuitive to use?

10) Is your pre-processing software platform independent?

11) Would you be interested in a pre-processor which embodies the following features:

state-of-the-art graphical user interface

multi-platform operation (Win95/NT, UNIX, LINUX)

interactive and independent geometry and grid generation modules

sophisticated and intuitive geometric and grid generation tool kits

software management of geometric and grid linkage

12) Would you like additional information?

Please reply to cfdsurvey@aol.com


Jonas Larsson July 3, 1998 12:12

Re: CFD Technology Issues
 
Perhaps you should inform us about who you are and who you are working for? What do you intend to use the survey results for? Will you post the results here?

John C. Chien July 7, 1998 00:56

Re: CFD Technology Issues
 
In real world applications, 3-D geometry modeling is one of the most difficult area for an average engineer to handle. Most engineers are not trained to handle 3-D complex geometry. Therefore, an easy-to-use intuitive modeling tool ( similar to 3-D Studio, 3-D Studio Max and the equivalent ) is definitely needed. The mesh generation although is based upon the geometry, the requirement is linked to the final solution, that is the problem you are trying to solve. My feeling is that, normally, you don't start the CFD process by first building the geometry. Instead, you analyze your problem first. Then, you try to identify possible flow field features, such as boundary layers at the walls, shear layers generated at the trailing edges, shock waves at the ramp, jet mixing, etc.. After that, you then try to make some preliminary mesh arrangement, mesh density calculation. At this point, you then go back to the geometry modeling, to figure out how to divide the model boundary and interior volume such that the geometry construction will give you the best mesh arranement and the most accurate flow field solutions. Remember that you are not dealing with linear solution of finite element structure problems. Instead, you are dealing with highly non-linear Navier-Stokes equations. What I am trying to say is, regardless of how you define the CFD processes, the computational geometry definition, the mesh will have to work with the solution several times or iteratively to produce the optimum result. Frequently, the mesh must be changed to obtain improved results. Sometimes, a poor mesh will create convergence problem, solution oscillation. Sometimes, you can't get a good mesh ( or any mesh at all when using automatic mesh generation methods ) because the surface model is not consistently patched for mesh generation. This often creates highly skewed cell or element. In this case, you are forced to modify the geometry model ( not the geometry itself but the way you create it ). Similar to the software for 3-D animation, all of the needed elements should be there, such as the tools to create geometry model, texture mapping, lighting, rendering, animation etc.. because these things got changed in the design process. By creating only the geometry modeling module and the mesh generation module, you have to create additional interfaces with the so-called solver and the post-processor. Interface support will be expensive. you will be forced to cover many different versions of codes. It also makes the real time interactive process very difficult. Going through (1) file (2) geometry (3) graphics (4) file (5) mesh (6) graphics (7) file (8) solver (9)file (10) graphics (11) file and then back to (1) again is feasible, but it's the technology of 1970. The conclusion: Why not go all the way to include everythings in one package. In this way, you can forget about the interface problem( that is the file format from both sides must be available for testing and interface development ). I am not trying to convince you that, because I have been developing CFD software in this way since early 1980 ( for my own use ). By the way, your approach may have unique power to control the industries and government. ( they will be forced to shut down when you stopped the pre-processor support . just a joke. ) Thank you for your time.


All times are GMT -4. The time now is 08:28.