Hi,

Does anybody have any burning desires, thoughts, dreams, applications or areas that they would like to see researched by CFD methods.

I have picked up a couple of general areas lately on this forum, such as Sports and FSI and I am also interested in other ideas.

Lets get the debate going, cheers.

Jas

Areas I think will be interesting are: a) design: i) cfd & general design methology ii) automated design and optimisation with cfd iii) integration with "design" databases

b) technical: i) real-time cfd for online process control

A lot of work is already being done in these areas - in the design area I think they'll be heaps of work to explore how best to come up with tightly integrated design methologies and which are reasonably automatic. To some extent, I think for engineering companies this will need to be an extension of their IT strategy etc.

Using cfd in online process control in chemical, oil and gas and other process industries (hasn't happened yet - or has it?) will also be possible in the future - here I think there will need to be work to allow the cfd to be part of a larger online model which can use and somehow assemble the deterministic cfd predictions, with current monitoring info - temperature or flowrate etc, and some type of statistical representation of the system - fuzzy logic, neural networks etc? - this area could be exciting - but really its just an extension of the simple system models - lumped models - currently used.

Greg

Hi,

Very interesting areas you have mentioned. I know that FEA is already well established in it's optimisation techniques and CFD would be a natural extension into this area. CFD in online process control, sounds exciting but for the complexity and scale possibly limiting it. Maybe CFD can be developed in modules (similar to symbols, blocks or library parts in CAD) that can be thrown together like process components enabling companies to build up their own process systems. I would love to see it in action.

Regards

Jas

The area of cfd research that interests me is in the rotodynamic machinery devices such as the pump.Toward that aim I have attempted to viewpoint world wide various studies and present them as best I can in my personal website at

http://www.fortunecity.com /meltingpot/broadway/1211/

Much difficulty arises because of market place issues on just how much actual cfd content reaches the final as sold product.I would appreciate any information particularly on pumps but Turbines and Compressors too.

I would like to think that CFD can continue to benefit the overall design and development processes.

The discussion thus far centres upon a rationlisation of resources for problem categories.

How far has CFD come from the 1960`s and what product benefits made.

The question on the future CFD Research is a good one.

Hi Ken,

Yes, an interesting topic, I have also had some experience in this area, but in positive displacement pumps, there are many areas for research in the application of these as most of the fluids pumped can have remarkable properties (foodstuffs and biopharm fluids). I would like to read your paper but cannot log onto your site.

A good area for research could be CFD on a device where one of the boundaries were rubber (eg diaphragm valve, disc pump). I would imagine this fell under the catogory of transient, deforming mesh, turbulent, fluid/structure interaction with rubber properties being the thorn in your side. Quite a challenge.

Regards

Jas

Sorry that you could not access but try again at http://www.fortunecity.com/meltingpot/broadway/1211/

The food industry does throw up real flow prediction challenges from heating and cooling extruded product such as reconstituted meat,potatoes and fish to the spray cooling of icecream and yougarts formulation with in house confidentiality the order of the day.

Yeah I think the areas he picks out are pretty good. As more (smaller) engineering companies start to use CFD I think they'll need to integrate it with their tools and design processes better. so design aspects of CFD are very important and many of the sommercial vendors don't pay enough attention to this. the process control area is also very big. In many traditional fluid applications control of the fluid flow is very important and not enough codes can be coupled to a controls modelling software like Matlab for either online or offline modelling. In general I'd like to see more easy coupling between CFD codes (commercial or otherwise) and other analysis and modeling software to cut engineering cyxle time and allow CFD to be a more integral part of the design cycle. right know I don't think solving CFD models is taking more time than people would like it's the setup and data extraction time. In some cases yo can write macros to automate a lot of it (especially data extraction) but that requires a lot of expertise.

In my view the real challenge will come from the Bio-fluid dynamics. Also, extension of CFD software for aerodynamic noise prediction. Turbulent combustion with finite rate kinetics is yet another challenging area. Magneto-Hydro-Dynamics(MHD) and Magneto-Aero-Dynamics(MAD) will also add the complexity to the existing codes. In future, Mesh-Less CFD will become a common practice and must be considered an area of intensive research for the moment. The integration of different numerical technique (e.g. FDM, FVM, FEM and BEM) in a single code may also be considered, if one needs to utilize these technique in a most economic way. More and more CFD codes will be written for network computing using PVM/MPI and performance optimization can be considered as an active area of research.

GS

I think it is really a good idea to concentrate on the data extraction parts of CFD codes, and directly coupling it to process control options as that offered by MATLAB.

(1). Future CFD Research is going to be based on the Current CFD Research, if there is Current CFD Research at all. So, what is the Current CFD Research? (2). I think, currently, CFD is still in the research phase. Every individual problem one is trying to solve requires extensive Journal Research to understand the background of the problem. Then, the geometry and the mesh generation will have to be created, step-by-step. (3). The geometry and the mesh generation can be organized and simplified, only for very simple geometry and configuration. And even for the simplest case of the driven cavity flow, there is no standard formula to generate the mesh, which will result in the optimum solution accuracy. (4). Assuming that the geometry generation is the job of a CAD designer, then, the question is :" what is the formula to generate the optimum mesh for a cavity flow problem?" If you don't have a definite answer to this question, then you need to do some research on your own. As a result, even for this seemingly simple cavity problem, the mesh generation itself is a research topic. If you don't think so, then you can just pick some conditions and try to generate a mesh. Then study the solution and find out whether the mesh is the right one to give you the right solution. (5). In the mesh generation area, mesh independent solution will demand the research work to be carried out, so that the solution will be the right one. To use CFD in design, to obtain the cavity flow solution and use the result in design, you must already know how to generate the optimum mesh for the cavity in the first place. If you don't have the optimum mesh formula for the cavity flow, you will have to do some research first. (6). In general, the real problem will be many times more complicated than the square cavity flow, thus, the first step in CFD is mesh generation research. (7). If I move on to the turbulence modeling and the solution algorithm areas of CFD, the question will be even more difficult to answer. Do you know which turbulence model will give you what solution in the high Reynolds number cavity flow? Do you know which solution algorithm will give you converged solution, when using with a particular turbulence model? And as you know that the cavity flow requires the solution of the Navier-Stokes equations ( or Reynolds averaged equations), and the main diffficulty is to capture the corner flow separation bubbles accurately, the question is then: which low Reynolds number model will be the choice for such simple flow problem? I should say, for such geometrical simple flow problem, instead. (8). Therefore, I think, each CFD problem is a research problem today, even for the cavity flow case. (9). We can not say that "a converged CFD solution " is "the right solution" of the problem, unless you have carried out the mesh independent research and the validation process. (10). I would say that in the near future, CFD will remain as the research approach in the flow simulation. The CFD process can be coupled with the long term product development, where there is enough time for the cfd research and validation. (11). As a short term design tools, the only result which can be produced by a CFD code is the wrong result. This may seem to be un-realistic statement, but in reality this is true, because most product design is the result of the engineer's experience, rather than the result of a CFD run. (12). To use only the CFD result and CFD optimization result in the design, you are likely to get the wrong answer to the design, based on my recent experience with in-house modified codes and the commercial CFD codes. (13). So, for future CFD research, each problem has to be studied for the mesh independent solution, and validated for the particular flow situation. When this is done and coupled with the design development activities, there is a chance for the CFD result to have impact on the product design. (14). So, the experience of the engineer is still and will be in the future, the only factor to determine the usefulness of the CFD solutions. This will apparently come from the engineer's knowledge about his trade and the particular problem being solved. (15). So, in the near future, the CFD solution will remain to be a research solution on a problem-to-problem basis. That is, each problem is a research problem.

[vidhatedigambar]

hi... we are doing our BE project in cfd analysis of Vertical Axis Wind Turbine....and looking for better blade shape.....
so, what should i use????? cfx or fluent for pre & post processing... and why???