CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Philosophical CFD question (

Richard Howe May 29, 2001 08:57

Philosophical CFD question
I hope this is an appropriate message for this board -- I certainly cannot think of anywhere else where I might find an educated and objective response to a question that has been weighing heavily on me.

I am a third year NSF-funded Ph.D. student studying full DNS treatments of multiphase flow physics. My research is going very well, and the faculty are happy with my progress. Things are great in every respect save one -- I am beginning to wonder about a troubling question:

"Who cares?"

It seems as if, although a huge corpus of literature exists on these topics, there is a massive amount of "academic inbreeding;" in other words, paper after paper quotes itself and other papers that quote it. When I attend conferences and listen to presentations, the most repeated adjective is "interesting," not "applicable" or "useful." There appears to be a fascination with difficult solutions to NS equations as if the solutions themselves had inherent value (ars gratia artis?). Am I jousting windmills?

I am fully aware that, as a young and inexperienced graduate student, these impressions may well be ill-founded. That is why I post this message -- hopefully, there are those who have faced and overcome these doubts, and can offer sage advice on how to keep my research relevant to the real world.

Thanks, Richard

Barry May 29, 2001 09:57

Re: Philosophical CFD question
I guess you could ask this question about everything in life....and probably about life itself and the meaning of it all. (A philosophical answer)

I think the key thing is...enjoy what you are doing!

Cheers Barry

Jim Park May 29, 2001 15:26

Re: Philosophical CFD question

I don't think your observations are off base.

When I was in your shoes, I was reminded that my doctoral research was to demonstrate that I was capable of independent research. That was up to the faculty committee. They set the rules.

After you jump the barrier of having your dissertation accepted, you can ponder these philosophical questions while seeking a permanent job. If you wish more societal or scientific relevance, change your focus.

Good Luck!

kalyan May 29, 2001 17:34

Re: Philosophical CFD question

I am not sure what you mean by "applicable". Sure you can not use DNS for most high Re engineering problems. But that is not the point of DNS. It is an analysis tool and not a design/rapid prototyping tool. No one wants to see a huge database and a lot of unexplained plots in a DNS paper. Here is what I (and I am sure many others in turbulencing modeling community) look for.

i) numerical accuracy : When you present DNS results, not many people look for numerical details. They assume that you have checked that your simulation qualifies as a DNS in terms of the requirements on accuracy, resolution etc (i.e., low numerical dissipation/dispersion, wall layer grid spacing etc.). This may not be the case, if you use unstructured meshes, immersed boundary methods and some such fancy ideas.

ii) Objectives : You have to identify the reasons why you are doing DNS. If you do not have a set of questions that you are seeking answers for to begin with, conducting DNS is an wasteful exercise in high performance computing.

iii) Conclusions : How far were you able to answer the questions. If you have asnwered some long-standing questions, validated ot invalidated some assumptions, confirmed a hypothesis, what are the implications for the models in use and future models.

I am sure you being a DNS researcher realize these things. The importance of your work is gauged by the value the research community of the day attaches to knowing answers to the questions. Showing that a DNS of a complicated high Re problem is feasible with the new parallel computer on the block does not serve much purpose. Because "feasible" and "practical" are two different things.

The relevance of your DNS work may not immediately be apparent to many people. But if some one develops a model based on observations and analysis of your DNS results and it performs well, they would be interested. PDF and LES models are a couple of such models. Many industries now want to look into LES (though they have not given up K-e yet). For combustion, they want PDf models in place of algebraic closures. That's why many commercial codes claim to have LES and PDF capabilities. People, especially the industry folk, want the final model. They may not be inclined or they may not have the time to follow the model development from where you started. It may not be in their job description. Most of the industry wants immediate gratification. If you are doing DNS (by virtue of the fact that you simulations could run for months), you have to persist and settle for delayed gratification.

Realize that most people in industry research also would have done some kind of basic research during their Ph.D. Just becasue they rely on "what works" (even if it empirical and makes little scientific sense) does not mean they do not value fundamental research and basic understanding during your graduate school. If some one is looking for a Ph.D to run commercial codes, you are better off not being in such a job.

Also researchers rarely get to work in the same area as their Ph.D thesis. Ph.D is not an initiation into a specific area of research, it is (as some one pointed out) a process where you learn to propose, work and communicate. e.g. You might be doing LES after your Ph.D. It is not DNS but it is close.

Many graduate students take their research very seriously in the beginning of their Ph.D. You have a feeling that what you are working on is "the" thing. You are fairly self-absorbed. Then you go to a conference and realize that there are other areas that people are working in that they think are as important. You have to realize that conferences are not always about science. Academics take their research too personally and do not like some one disproving their work. The lack of objectivity is sometimes quite disturbing. There are lot of clashes of egos. Industry people are there to look for some thing they can use immediately or something they can get done quickly by hiring a student about to finish. The commercial code developers try to tell you that their codes can give you faster and/or better solutions than any other. In all this circus, people sometimes can be unwilling to accept or appreciate a very significant result. If you result is too subtle, you can forget it. Lot of times, you may need more than 20-30 minuted to get your point across. So, try to have a broader outlook when you go to conferences. You seldom gain research knowledge from them but they give you good exposure to what your life might be like in the future if you remain in research.

Finally, if you continue in research, be prepared to jump into new fields, learn them quickly and publish fast in order to be successful. New fields or buzz words are where the money is. It was LES a decade ago and now it seems to be MEMS. I am not sure if you expected such a long reply but having graduate only a couple of years ago I remember very well what it wasl like going through a phase you seem to be going through now.

John C. Chien May 29, 2001 19:43

Re: Philosophical CFD question
(1). You are right. (2). I think, a degree is just a degree. It is a good idea to get the degree as soon as possible. (3). But since the school is an excellant environment for learning, it is important to learn "subjects which will be important when you are outside the school". (4). You can always try to get another MBA after PhD, or even at the same time. And in some fields, you don't even need an advanced degree. (a computer programmer with a PhD ?, No, that's too much.)

Kevin May 30, 2001 01:56

Re: Philosophical CFD question
What made you decide to work on that topic for PhD? The NSF fund? Why did NSF decide to give you money for that topic? What did you write in the proposal? What do you mean by "My research is going very well"? By that, do you mean "the faculty are happy with your progress"? "Things are great in every respect save one", great in what sense?

As a 3rd-year PhD student, you should be able to answer most of the questions above.

I. Dotsikas May 30, 2001 07:08

Re: Philosophical CFD question
Hi Richard,

As it is already said: PhD is only to proove that you can you can work by your own; that you can ceep yourself obeyd. It has few to do with something like helping the living world and solve the big problems of today or even contribute to their solution. Donīt forget it. If something usefull comes out of it; itīs good. If you can make a living out of it, it is good for you. On the other way, Academics are much like small children. They play with their toys and think that what they do is really great. Everyone believes that his topic is the most important and every PhD Work beginns with the question why nobody else has not investigated this very importand question!!! Itīs good if you have a plan. A big idea that you follow. That may be to finish your PhD with as few work as possible and get a good paid job; or try to make a good PhD and try be become a Prof. Why not? Everything is possible. But what ever you do: you must like what are you doing. You must get up every morning look yourself at the mirror and say I am a very lucky person, because I go to work and I do something I love; and they pay me for doing this. If you say it you are a lucky man, and not only you ...

Best regards


George Bergantz May 31, 2001 00:01

sorry, but hogwash
To much of the the previous post, I politely say, "hogwash." Many of these themes arise around the misconceptions about the many roles that academic/intellectual life plays in things of immeadiate societal relavance and the way that a society creates opportunity for the continued growth of its technical and social well being.

Consider this simple example: someone out there wants to uinderstand rocks, perhaps because they are interesting. But it turns out that many people share this interest. In the process and after two huundred years, a whole body of understanding emerges. And then one day you have a terrible toothache and have to go to the dentist- luckily for you the dentist has special alloys and materials that depended critically at some point on that knowledge of the geologist. So your tooth gets fixed and you can go on with your life, largely ignorant of the incredible way that things are deeply interconnected just so that you can put on your pants in the morning.

Now consider this same situation for the point of view often expressed on this web page: really smart cfd guys see no need for geologists 200 hundreds year ago- just academics fooling around. So no university, no funding. Now the toothache day comes- ouch- but sorry to say, no help. You are miserable and disfigured, and can't wait for 200 years for someone to do all the work to solve your problem. Tough luck.

More to the point in my own reserach- volcanic eruptions- if no one tries all kinds of things no understanding emerges. Sometimes CFD is useful, sometimes not. And maybe we will find that some part of multiphase DNS will play a role- we just don't know yet. But maybe I find your paper and it is a real help. It is the job of the student to do sufficient scholarship to understand why and how their work may be useful, not just to the building of some gizmo, but to facilitate an entire field.

Multiphase DNS is about as hard a problem as there is (if you add reactions). If you master all the physics and computational/numerical skills to do this, I commend you. There are very few in the US today who can take on this challange. It is a pity that you do not appreciate how unique you are, how special the opportunity for unique contribution. Perhaps look to other fields for applications- there are many.

The point of most academic reserach is *specifically* not to come up with a very particular result to some local problem. The mandate is much larger. It is to develop a wide base on which many points of view, experience, can co-emerge, and new, previously unanticipated knowledge can arise. This is serious and thrilling business.

If you think that learning is simply a matter of answering questions on game shows, stick to daytime TV and away from higher education.

I. Dotsikas May 31, 2001 08:13

Re: sorry, but hogwash
We used to talk about the use of science many years ago. We were children. Probably you read my hogwash quite fast, so that you didnīt understand what I was talking about. There are two completely different things. The one is the real life after the work and the other thing is the science itself. I didnīt talk at all about science. I apreciate any scientific work, because I know that it is not only hard work but a great portion of competence and intelligence. Surelly science does not produce only gold but sometimes , too. Anyway, as you probably didnīt realise I was talking about life.

For watching daytime TV I have unfortunatelly not the time. I donīt even posess a TV! Besides my day work I develop a new honeycomb reactor (with CFD-support). What made me write my last contribution? Dissapointment about the worth of pure academic work and the way this work is beeing seen by other people.

Johan Larsson June 1, 2001 06:24

Re: Philosophical CFD question
I agree with basically all messages posted above. I totally think that we need to do fundamental research in order to apply it to real life problems later on.

I also think that there is an interesting point in what you say. All people in academics, including myself, need to constantly question why we're doing what we're doing. In many cases the outcome of research has minor if any practical impact, and it is our responsibility to avoid that as much as possible.

Due to this, I am a big fan of research in collaboration with industry, since this usually means that there is a demand for the results. In Sweden it is fairly common with Ph.D. students employed by industry. For applied research, this situation will hopefully add to the usefulness of the research.

I totally think that fundamental research is of vital importance and that it should primarily be funded by non-industrial organsations.

I think that your questioning of your work is sound, and I think that we all should do that on a regular basis. Mind you, let's not forget how fun it is! When you get your Ph.D. you can go work in industry doing very applied CFD, and this will definitely add value to the design process of that company.

clifford bradford June 1, 2001 18:22

Re: Philosophical CFD question
I think the answer to your question is obvious: people who're interested in multiphase flows are (should be) interested in your work. Now they may not be interested in all areas of your work; hey may not really care about the DNS part of it but it truth the DNS is just the instrument that gets you to the better understanding of multiphase flow.

You did touch on an important issue of academic inbreeding (it's not quite incest but it can be pretty close). I think this is due to some small extent to improper use of references in articles and perhaps a lack of stringent refereeing in some journals but it isn't the main issue. I think as a person undertaking advanced engineering research you have to realize that you need make what you've learned available to the engineering community that needs to know it. If you have a lot of great stuff in your thesis it does the guy who design sewage pump no good because he can't get your thesis (he's never even heard of you). Even if you've presented it in journals and at conferences you're still missing your audience because that sewage pump engineer who needs more understanding of multiphase flows doesn't get Journal of Computational Physics in the mail.

Then also you have take into account that there are diffent audiences for your work. Some people will be interested in the "DNS" and some in the multiphase flow physics at different levels.

The reason why it is good to get the information out to the engineering community beyond just the academic world is that if you can do this you will often feel gratified for it particularly if you take what could otherwise seem like something particularly esoteric and break it down for someone is not as research oriented as yourself.

In addition it may give you an additional appreciation of your own work to get out a little and find out what's going on in the world of multiphase flow. The guy who design the "shit pumps" might give you something to think about!

When I am discouraged about working on my CFD (particularly when the job won't converge or hen it crashes etc) I always think on the beautiful (at least I think they are) machines that come out of my and my colleagues work, and the beneficial uses for them.

I think that is essentially what you're asking: "what's the useful purpose of my work?". Perhaps you're feeling uninspired. My advice in a nutshell is think of the practical knowledge that other engineers can gain from your work (remembering that you're mohammed and they're the mountain and mohammed must go to the mountain) and the useful things that they can make better with that knowledge. Remember people don't realise it but they need good multiphase pumps so that their crap won't back up on them!

Inspire yourself and you'll answer that question.

clifford bradford June 16, 2001 05:49

Re: Philosophical CFD question
Just another thing: when I was in school I tried as much as possible to expose myself to things outside of my research field so i could keep fresh and have some breadth (have you ever met the PhD student, or even professor, who is an expert in his research field but his knowledge of areas outside is at the undergraduate level)

I find structural dynamics to be a trifling diversion from the day to day grind of the Navier Stokes equations. It's all linear!

John C. Chien June 16, 2001 12:35

Re: Philosophical CFD question
(1). When Wright brothers invented the airplane, they did not use FEM structure dynamics. So, the hard part is aerodynamics.

TOT KTO 3HAET June 16, 2001 16:46

Re: Philosophical CFD question
If you keep doing this on a regular basis, you'll realize quite soon that most of the real-world problems in SD are non-linear too...

Just my 2 cents.

John C. Chien June 17, 2001 14:41

Re: Philosophical CFD question
(1). Only when he is convinced to write his own codes in structure dynamics or CFD. (2). The problem in turbomachinery business is: people "think" that CFD is important to have. So, they hire someone and bring in some codes. they "think" that they have done the right thing. So, they decided to take over other companies. then they realized that they can't just buy another company like buying CFD codes. after that they don't know what to do next. (3). It is sad to hear that the wires in new commercial aircrafts were cut ,even after the company has acquired many other aerospace and aircraft companies. The same is true for the turbomachinery companies. (4). Acquiring other companies (to eliminate competition I guess), buying more CFD codes and hiring more engineers will not create anything new (new airplanes, light bulbs, or more reliable aircraft or engines, etc.). (5). There is only one thing for sure, that is: a company without real CFD capability will disappear from the surface of the earth very soon. (6). The structure dynamics world and CFD world are non-linear. Unfortunately, many people like to invent linear world out of it, trying to make it as easy as buying a license or a can of Coke. (7). Well, a can of cold Coke is better than nothing, when a company is dying. (8). In the industrialied world (after the steam engine was invented), there are two ways to survive: (a). maximize the "entropy production", in this way you can out run other people or companies, (b). minimize the "entropy production", in this way you can improve your way of thinking and thus higher probability of survival. The bigger will disappear faster, because it must take more resources. (Chinese communists sort of learned it the hard way. To feed more people, will only create more people. And there is no end to it.)

All times are GMT -4. The time now is 01:58.