Mr. Ziaul

Exactly, you have the bingoThis language is the french. The Victor Hugo's, the Balsac's, the Curie's, the Ampere's Language. Did you know this names?

Juan Carlos

Hi

Fortunately the 3 years of experimental testing had resulted in a Masters degree with the PhD covering the numerical work. I do however agree that if you start a PhD with no experimental work in hand you cannot seriously consider using a commercial code. And even if you write your own code it by no means ensures you of good results, there are numerical and modeling aspects that even good codes struggle with, turbulence is but one example.

I must also admit that my personel belief is that if you are serious about CFD then writing your own code is essential. At some stage you must aquire a fundamental understanding of how the whole thing is put together. In fact I have resolved myself to writing my own code in my spare time although the going is slow due to time constraints. It seems to be rather difficult to get postdoc positions that allows one to do this type of thing for a "living". I'll keep looking anyway!!!

It also seems that more and more academic institutions are moving away from the more fundamental CFD development to using commercial codes to solve or investigate problems for industry. This is especially true for institutions that are only now venturing into the field of CFD. Very few companies are willing to give money for code development that some time in the future might yield some rewards, it is easier to get money for a problem that can be "solved" now with a commercial code. Although I have respect for the circumstances that "enforce" this way of thinking I do not necessarily agree with it.

I think I have said quite enough already, I hope this was not to tedious for all involved!

Regards

Chris

.

Hi

I have a little vocabulary problem

comercial CFD package = comercial CFD code

Regards

Mr Salas, i think you have not gotten the answers to the questions you asked. if i'm not mistaken you'd like to know the algorithms,characteristics, and properties etc of the Fluent code you'll be using to model a problem you know experimentally.

to answer this question i think you should contact the company that makes Fluent. they are online at www.fluent.com (if i'm not mistaken). also you may have distributors and/or branches of the company in France so you can speak to someone over the phone or correspond with in some other way. they should be able to provide you with references written by company members and documentation describing some of the workings of the code. they should also be able to provide you with the assumptions, strengths, and weaknesses inherent in the code and tips to improve your accuracy and performance.

i think the use of a commercial codes like fluent would definitely be appropriate in this case because i think it is suited to your problem. but i think you should maintain contact with someone at the company who can assist you.

you must remember that some of the people on this group have a different background from yourself and so may not understand your problem or may have preconceived ideas. good luck with your work

Hi Gentlemans

Thank you for your advices. I know that some of the people on this group have a differnet backgroud..this is normal. At this moment I am very happy because I'am understandig (a little only) what is the technology CFD and the applications into engineering problems.

Regards from Lyon France

This time I am really lost! I don't know french.

I can't agree with you more.

Different people may have different ideas and opinions. Yet I want to point out that, interestingly enough, many state-of-the-art technologies in CFD these days are geared toward bringing CFD tools (not necessarily commercial for commercial CFD code averse folks) to the hands of designers and engineers who may have some fluid mechanics background but not necessarily experts in CFD, i.e., toward making CFD tools easy-to-use, efficient and affordable.

And considering that you're based in France, I must mention that Fluent France has a highly competent and knowledgeable group of CFD engineers who can help you get up to speed with using FLUENT software. And Fluent France holds every year their Fluent Users Group Meeting (we call UGM in short). I strongly recommend you attend the UGM next year unless you plan to graduate this year. Last September this year, they had about 150 client attendance, and lots of excellent user papers were presented by users like you, whose major expertises are not necessarily in "writing CFD code", but something else like design, trouble-shooting, experimentalists, etc. So take heart !

Among all the replies to your original query, I like the ones from Amadou Sowe and Clifford Bradford, which will help you develop a positive and objective attitude toward CFD.

(1). It is really very hard to understand your message. (2). The need to use CFD comes from the need to solve fluid dynamics problems. These can be problems to reduce the auto emission level to reduce the health harzard, problems to design turbine blade to reduce fuel consumption and improve engine life, problems to design fighter aircraft to improve performance, problems to reduce the pollution from a combustor or a furnace to clean up the air, problems to design a clean room for IC manufacturing plant for productivity, ....etc... (3). These fluid dynamics problems occur at the design stage, the production stage, and the service stage. So, it is linked to the product from the begining to the end. (4). It becomes clear that the ability to provide accurate and reliable answers to these fluid dynamics problems is the most important issue. (5). So, as an engineer working in this field, the most important issue is to know whether the solutions obtained are accurate and reliable. (6). To achieve this goal, researchers have been working on various numerical methods to provide more accurate results (assuming that the converged solution is achievable), different turbulence models to provide more realistic representation of the turbulent flows, etc... (7). So, there are numerous methods and models around, which were developed under different conditions and for different problems in mind. When some of those were implemented in a package ( a collection of libraries, modules, subroutines and programs), and sold on the market under "as is" condition, it is highly questionable that a user (without proper pre-caution and a systematic approach) can easily obtain the desirable accurate and reliable results in a so-called user-friendly program environment. (8). All of the questions and answers presented here were trying to make sure that the solution obtained is indeed the desirable accurate and reliable solution. (9). So, if the vendor of the commercial code is willing to say that the solution obtained by a user is always accurate and reliable, then, both the vendor of the code and the user of the code would be very happy. But, so far this approach is not practical, and no vendor is taking this approach. (10). So, Ask yourself, would you want to buy a car under "as is" condition? Would you want to buy a car which was designed by a commercial code under "as is" condition? And, next time, when you travel by air, would you want to fly in an airplane which was designed by a commercial code under "as is" condition? (11). If your fluid dynamics problems are not in these areas, then perhaps, the use of CFD or commercial CFD codes is not essential at all. I mean, if your problem is happy with a 1-D code, then most questions and answers related to the 3-D problems presented here are just like Saturday night music.

Hi John,

I agree with Amadou's assessment above, and I'll give you my take. With your car analogy, you're asking if someone wants to buy a car "as is", with no "warranty" or guarantee of accurate results. Well, if I need transportation, and my only two options are buying a car "as is" or building it myself, I'll take the former. Likewise, most people don't have the time to write their own codes, and probably are not expert enough programmers to do a better job than the team of professionals who work as commercial code developers.

However, as Amadou points out, one should have a strong enough background in both fluid dynamics and numerical methods to be able to understand exactly what is being done inside the code. Even an expert mechanic would be more efficient to buy the "as is" car than build it himself. He has the advantage of being able to recognize the potential problems. And it is important to note that this expertise does not automatically come from code-writing. I have written my own codes, mostly to gain an appreciation of what it takes. And it usually comes down to wrestling with exactly how to implement a certain scheme (that I didn't develop) and models (that I also didn't develop) and get them to work properly. Implementing a k-e model in your own code by no means makes you aware of exactly what problems are going to show up in what flow situations.

Finally, one more analogy. If I was an experimentalist, and had spent a decade of my life building a complicated measurement apparatus, I'd be pretty proud of it. And then if my colleague next door went out and bought the same thing, I'd probably try to convince people that really, if you don't build it yourself, then you probably aren't going to get useful results. It's baloney, of course, but I'd have an interest to protect. My colleague's ability to get useful results would have to do with his knowledge and experience, not on whether or not he completely pieced the thing together himself.

Should one be competent and well-versed in fluid dynamics and numerical methods to use CFD properly? Absolutely. Should one have to write his own code? Of course not.

Regards, Keith

I am very sorry that you found it hard to understand my message. You are very scornful of the CFD codes because they sell their software 'as is'. If I am not mistaken every software company usually has this disclaimer in their license agreements. I do not think that this is unique to the CFD software companies alone.

I will try to clarify my point by answering your questions about buying products 'as is'. It is very funny that you ask whether I will buy a car as is; infact I just bought a 1988 Buick as is. Inorder to buy a product 'as is ' you have to do one of two things:

(1) Make sure that you understand the product well enough so you can judge whether the product will do the task you want to buy it for.

(2)On the other hand if you cannot do (1) for yourself hire some one qualified to do it for you.

I am a mathematician by training who does not know much about cars so I had a good mechanic check the car for me. The only problem he found was a torn CV joint cover. I bought the car and had a new boot put on.

The point I was trying to make was that some of us do not have the required background to make an objective condemnation of the CFD software industry. The required background has to include the key components of CFD( numerical analysis and fluid dynamics). Some of our community members seem to be strong in one or the other but not both.

Another point you made was whether I will fly on a plane whose designers use commercial code under 'as is ' condition. I am sure the designers of the commercial planes use software that has a disclaimer somewhere in their license agreement. This disclaimer precisely declares the 'as is' condition. By the way I hate flying because I am not in control of the plane. Let me get your feed back.

Well put, Keith.

(1). You are absolutely right. "One should not have to write his own code." The same is true: " One should not have to build his own car." (2). I agree with you that "when there is no choice, one would probably use whatever available to him". (3). The fact is : many leading companies in industries are not spending research money on their CFD activities, they are trying to save engineering salary by puting commercial codes on their computer system. (4). Back to your comments : "one should have a strong enough background in both fluid dynamics and numerical methods to be able to understand exactly what is being done inside the code." This can be done only if " the source code listing is available". (5). Very few companies are willing to give you the source code listings. And this is exactly why the commercial codes are labeled as "black box". (6). The need to know how to write his own code is to have knowledge to protect himself when using a commercial code. The same is true that you can buy a car "as is", but you are likely to get it checked out by a car mechanics first. The car mechanics is just like a person with knowledge to write his own CFD code. He can check out a car without actually take the whole car apart. (7). Are you sure that the team members works on the commercial code development are "real professional"? (8). As a matter of fact, in most cases, you don't know their names, their background training, their experience, their knowledge in solving your problems, not to mention the code itself. (it is fairly common that an error identified by the user is not fixed for a long time, because the person who had worked on that portion of the program is no longer with the company. And if that part of the program was written by a subcontractor company, there is not much you can do or he can do to fix it. )

(1). I think we have the same view point. (2). We both agree that a car "as is" will always have some problem. (3). So, it is the approach taken to eliminate the problem. (4). When you take the car to the professional mechanics, you are taking the professional approach to solve the problem. This is because the mechanics works for you. It makes no difference whether you have the knowledge about the car or not. As long as you two can find the problem and fix it. It is the approach which is important. (5). So, I agree with you that there is no need for you to take the auto mechanics class at the nearby community college in the weekend, in order to buy a used car. (6). By the way, I used to have a Buik convertable, a Buik Skylark, a Buik Century. But recently, I am driving a small Honda. The difference is I used to visit the garage often to get my Buiks fixed, even when they are brand new. Now, with Honda, I can start the car with only one turn of the ignition switch, and I don't have to visit the garage to fix the car any more. (7). I must say that, the purpose of the discussions is to bring out the nature (of the state of the art ) of the CFD technology (commercial codes represent part of the technology). Through the awareness of it, some professional approaches on the users side could be developed to better utilize the CFD technology. (there is nothing personal in the discussion , it is always aimed at the rest of the readers.)

Hi

I want to make two points:

1) When you get into a plane (even a car) you know that it had to pass some set of standards set by some government agency. It does not mean that nothing will go wrong, but it means that great care has been taken to make the plane safe. To some that may be a comfort, to others not. I don't know of any agency that checks up on commercial CFD-codes. The "checking" is left to the user and the user has so little time......

2) A year or two ago I was injured participating in a organized game. While rushing me to hospital the person in charge was thoughful enough to remind me that I signed a form that proclaimed that the game was played at "own risk". At that stage I had other things to worry about. It was only much later that a lawyer friend of mine told me that even if I had signed their disclaimer it did not mean the company hosting the game was not liable. The very fact that they provided protective clothing for the game made them liable since they obviously knew that without proper protection injury was very likely. The important phrase is "proper protection". They provided protection, but was it enough? Are the companies selling CFD-codes doing enough to protect the unsuspecting users?

Finally, and this is not a third point, I use a commercial CFD-code for research purposes. It has taken me some years to feel "comfortable" with the numerical results. A lot of my time was taken up by testing various aspects of the code by comparing the results to experimental data. I could have saved a lot of time if there had been some organization of repute that had done the validating of the code and had given it a seal of approval. Maybe this is an impossible situation to have such an organization, but it would be very nice wouldn't it?

Regards

Chris

(1). I would say that you have just identified a great business opportunity, that is, a CFD testing laboratory to test and validate the commercial CFD codes. (2). In this way, at least, a commercial CFD code will have to meet some standard requirements before it can be placed in market for sale.

Is this what the WUA-CFD (World User Association in Applied Fluid Dynamics http://www.wua-cfd.com/wua-cfd/) are aiming for?

I don't know much about them but I stumbled accross their web pages the other day.

Regards

Althea

comment to point (1)........

You do not know of any agency that checks up on CFD may be true. Infact I do not know of any agency that checks up on any software including our operating systems. May be I am wrong but I will be glad to get some examples from some one out there. You are right when you said that the user is the one doing the checking. This is the main reason why I kept saying that to use CFD correctly you have to know both numerical analysis and fluid mechanics very well.

comment (2).........

To me, I do not think there should be any 'unsuspecting users' of CFD. It is too complicated a field to not suspect your every move. It is a lot more comforting to be 'suspecting' in a familiar environment than in one you are less familiar with.

(1). Based on reports, system software companies such as Microsoft, and computer games companies do test their software in-house and by outside professional programmers. (2). They also have a large number of in-house software testers. (3). They can't afford not to have their software thoroughly tested before release.