This is a very interesting question.

On one hand side, commercial (and academic) CFD packages are to a certain extend representing the current state-of-the-art in modeling approaches in computational fluid dynamics.

On the other hand side I agree, that the result from a commercial CFD software package can to a certain extent only be as good as the fluid dynamics knowledge of the person who is using it. This is in the responsibility of each individual CFD software user, e.g. having a good idea about the CFD error hierarchy and CFD best practice guidelines for industrial CFD, and in the responsibility of all university staff to make students and young scientist sensible with respect to possible software and user errors in the application of such CFD packages. This aspect cannot be over-emphasized in university education, from my perspective.

Regards,

Dr. Th. Frank.

The C.F.D of the future will be profoundly affected by ongoing advances in deep machine learning. A typical C.F.D simulation of the future will, in addition to the usual static pictures and video animations, be used to generate a digital twin/reduced order model of the system under investigation. The digital twin will be an accurate replica of the system, but, orders of magnitude faster to evaluate. These digital twins can then be “plugged” into CAD/PLM software tools or even into spreadsheets. We assume that suitable software interfaces will be developed and standardized for this purpose. These can then be used for design without requiring the services of a C.F.D expert. A clear separation between the tasks of generating a digital twin and their use in design will expedite the future growth of C.F.D and render it substantially easier to use. We expect that this will result in the application of C.F.D to a wide range of problems and have an impact on virtually every aspect of human activity. For instance, expanded use of C.F.D and digital twins in bio-medical applications will substantially improve the quality and effectiveness of healthcare and keep it affordable too. The social benefits are obvious.

I think the discussion can be addressed in two different topics, the technological advances that are then used in CFD and the real advance in CFD topics.

Unfortunately, I suppose that very few academic people are really aware of what is the latter. The present time is characterized by people engaged only in technological advance in softwares and not in the CFD contents.

And the explosion of ML and AI is just a way to follow the trend.

Dear Abraha Kahsay

Yes, unfortunately, the academic guys' tendency toward CFD has deviated in two ways; first, is the voracious rush for publishing articles, while the project on the grounds in the industry may seem hilarious because the academic guys may fall behind the application of science due to indulging in theories. Secondly, the point of using CFD is missed. In fact, are we using CFD to replicate experiments? or are we using CFD to get a hand on the blind spot that is not touchable due to the limitations? Certainly, without profound knowledge of foundation, CFD results are nothing but flamboyant embellished pictures whether in uni or industry.

Regards

I appreciate you all for your deep and awesome responses!

Hamed Majidian You wrote: "Secondly, the point of using CFD is missed. In fact, are we using CFD to replicate experiments? or are we using CFD to get a hand on the blind spot that is not touchable due to the limitations?"

Here I would like to disagree. It is necessary to do both and replicating experiments with CFD has its own justification. It is usually called verification and validation (V&V). It is absolutely required to build trust and confidence in existing and newly developed physical models by applying to a certain experimental setup involving the modeled physics those CFD models and to quantify the remaining modeling error in this particular model combination. By that, of course, the CFD simulation has to be carried out very carefully and all numerical, software and user errors needs to be eliminated before one can judge about a remaining quantitative error of a particular combination of physical models for a simulation purpose.

Once this model validation has been done, it is of course (here I agree with you) the purpose of CFD to apply it to some investigation matter which is new, which has never done before, which helps to improve a certain engineering design or application and which is not accessible by experimental methods. That is the real purpose of CFD. But the V&V task comes beforehand and it cannot be passed by.

Regards,

Dr. Th. Frank.

CFD predictions can not replicate experimental measurements but in some cases when the problem is complex experimentally CFD can be used to predict the near results or the trend at least (save time and money).However, developing CFD techniques are necessary to face the industrial applications.Also verification and validation should be carried out in parallel with using CFD.

Dear Thomas Frank

Thanks for your conribution. I am on the same page with you. Perhaps, I started my saying from the middle. To put it another way, I meant the purpose of using CFD is not only replicating experiments but also is to investigate what we cannot access in reality. However, I believe the most important aspect is engineering application that means without engineering application the point of replicating experimental data remains fruitless.

Regards,