I am asked this question:

what is the tendency of CFD?

I do nnot know how to answer.

It is appreciated it you can address this.

Thanks in advance.

tendency to do what?

although i have not clearly understand in what context u are asking. are u a teacher and some student ask u this question but i share my feeling about it.

cfd help us in predicting many problems but still it have limitations. it have eased our life by preventing us to endulge in difficult and expensive experimentation.

i think i didnot add any in your knowledge and u must already know all that i tell you.

Thanks.

This is a question from my student who attended an interview for a R&D position and the committee asked him this question.

After the interview, he told me this question.

I feel difficult to answer him.

Thanks.

This is a question from my student who attended an interview for a R&D position and the committee asked him this question.

After the interview, he told me this question.

I feel difficult to answer him

Maybe it was one of those weird psychological questions. If your student had said "converge", then he is an optimist, but if he said "diverge" then he is a pesimist. Or maybe the person who asked it was just an idiot trying to be clever, whilst not really knowing what he was talking about. You'd be surprised how many CFD "experts" get by on seeming to know what they're talking about, without actually doing so.

CFD's now such a broad subject that I mistrust anyone who professes to be an expert, apart from those humble people who are and normally dont go on about it

In terms of 'tendency' of CFD an analogy to another tool might help. Automobiles are a tool, a tool to help you get from A to B quickly. CFD's a tool as well.

Automobile tools are somewhat more mature than CFD tools. If we look back to the origins of cars (the most common form of automobile) then we find that the tool user had to be quite expert to utilise the tool. One had to have a lot of mechanical engineering experience. This was in response to the immaturity of the tool technology. This is common for most emerging technologies.

Over the last 100 years automobile technology has changed in a number of ways, more efficient, more robust, more stable etc. Interesting that the core underlying technology hasn't changed (Otto cycle and the diesel cycle variant). The tool user has changed as well, now anyone can drive a car, even my wife Do they know how the car works? Not really, just which peddles to press and when. Does it breakdown? Sometimes but then there's garages to deal with that.

So, is CFD today analogous to the automobile 100 years ago? Yes (although at this stage please feel free to jump in with a number of analogy busting observations). The tools require expert usage, expert in terms of understanding of numerical methods, behaviour of fluids and thermal effects.

So, what of the future? Well even today the new breed of CFD tools are focussed towards less expert users (there's more $$s there just as the car industry has benefited from the masses owning cars). In terms of user type it's those CAD centric engineers, mechanical design engineers. The classical CFD community is sometimes up in arms about this. Lots of experts protecting their careers, both academic and industrial. Indsutry managers love the concept though, less reliance on over paid expert people who everyone has to rely on and never seem fast enough in providing answers/solutions to the rest of the design group; that's just not good business sense.

It's fascinating watching the big CFD players of today going schizo in trying on the one hand to placate their existing user base and on the other trying to tap into the new one ("the CFD expert should be the internal support guru for the CAD centric CFD users" etc. etc. ....bllx)

In summary of tendency:

FV will continue to be the underlying technology (the otto cycle of the CFD world)

FE will be a continued alternative (the diesel cycle alternative or maybe something else?)

Lots of people who have built expert castles on sand will be upset and looking for new rented accommodation

More effort will be put into automisation of CFD codes to preclude the need for expertise in usage

Lots more people will benefit from CFD which can only be a good thing

De-skilling.

Hi,

Very clever analysis.

I however think that the analogy with car mass owning only holds up to a point. Automotive is a final tool for everyday life. CFD is a tool for innovation.

That is a kind of law of physics: you must have experts of something somewhere, or you stop innovating and you simply have cheap Color Fluid Dynamics for basic engineering work.

PS: I am not an expert in CFD. I have no carreer to protect. That is simply what appears to me as a logical deduction.

"CFD is a tool for innovation"

Interesting, I hadn't considered it as such directly; in the same way I don't consider a calculator a tool for inovation. It does sums, an alternative to paper based calculation, much quicker (just like CFD is much quicker than wind tunnel/prototype testing).

Yes, innovation in the design of the tool itself critical. With such a competitive automtive indusry there is now really good value for money. Is the same true for the monopolised PC O/S market? Will the same be true for the Ansys dominated CFD market? I think innovation will be pushed to vertical application specific CFD tool design whilst Ansys dominate the general purpose 'fit for all' market.

I hate to be "that guy" but comments like "CFD is much quicker than wind tunnel/prototype testing" is one of those comments that always makes me cringe... It's a tool, and like any other tool you have to know when it is and isn't worth it.

I've worked on programs where we never bothered with CFD analysis (and this is recently... in the last couple of years... not 1960s... and we had planty of computer resources available). We started with simple hand calculations, built prototypes, tested, and finalized within a month or two. It's an exception to the rule, I admit. Typically prototypes have long lead times, tunnels have limited availability, and it makes sense to do any down selects you can before going to a tunnel. But again, it all depends on what you're trying to accomplish.

Also, there are some things that are just too complicated to be done in CFD in any reasonable time. So you make assumptions and simplifications, but without testing you may never pick up on "real" effects of the geometry. I've seen months spent designing a system in CFD... and from that a model was built and tested (in less time than was run on CFD) and nothing worked because of the original simplifications of the model. We essentially re-designed the thing in the tunnel and got it to work.

It's a tool. Treat it like a tool.

Like my old shop teacher used to say: misusing a tool can cause someone to lose an eye... I think he was talking about power tools, but I like the reference anyway.

Jason

hi dudes i have strongly believed this:

"the temdency of cfd is to simulate nature and thereby predict its vagaries"

now some of u may say cfd doesnt predict nature as a whole. then we have numerical analysis as a whole. but cfd is always tending to be an astrologer. isnt it?

I would venture that modern cfd is still in its infancy. not in terms of fancy interfaces with all the bells-&-whistles, but rather in terms of the understanding of the physics they are purported to be able to simulate. Most commercial solvers are essentially based on 1970's knowledge.

There appears to a huge void in workable numeric solutions for the region of flows from just above the onset of instability all the way up to say 90% (nominal), or more, of the speed-of-sound.

Numerous numeric kludges have been introduced to provide solution stability, but, precious little is really known about the physics of this region. Most folks really have no idea of why cfd solvers suddenly diverge for no good apparent physical reason. To say that once a 'magic' Peclet, or Reynold's number is reached, invokes the gods of instability, is a difficult one to swallow. Once this region becomes better-explored, then only will the numeric tools be able to more accurately predict reality.

I believe that it is time for the next academic, theoretical breakthrough in the fluids field. CFD can then follow at its leasure.

desA

I assume they are asking about the current direction of CFD software and/or research. My answer would be towards multi-physics analysis (although I hate the term multi-physics as the same physical laws apply to all the disciplines I have seen combined).

yes. This may be the original meaning they asked. The future direction in the community.