In general, I would say yes. testing against well known cases is a standard method of testing codes.However, that is to test whether the code correctly solves the relevant equations. You can use simple problems to test this against analytical solutions. This is a far more complicated case. As Abdul points out, there really are two things to be tested here: the code itself and the setup created by the user. In this case, I would have liked to see more detail about the way they did this. That is why some people nowadays add the source files of their code to the paper. This gives readers a chance to examine the setup in detail. While the test is certainly useful, I'm not sure whether you can really claim that it validates the work. Personally I would say that 'it behaves as expected', which is a good thing, but no more than that.

Yes Mohammad-Reza Moharreri it is a good way to validate your case provided your case is very similar to the test case for which you are validating your data. As Allard Jan Van Marle mentioned, you can not always rely only on the numerical results. So it is always a good practice to validate your numerical solution with the experimental solution which is more convincing.

The so-called well-known phenomenon should have a theoretical basis, so CFD merely verify the case with the specific flow/heat etc.. models and settings for simulations.  Unless the case study in point can validate all the models and settings similarly in theory, it is not possible to verify the case with past CFD study.  You may re-run the previous case to ensure it works theoretically before making the comparison and also using it for verification.  Good experimental data is also useful for verifications.

In my opinion, The validation of CFD results in the papre you mentioned is insufficient. They only obtain the well-known phenomenon, but not verified against with the experimental data. 

In fact, the validation of CFD should be divided into two parts. One is the global performance, such as entrainment ratio. The other one is local performance, such as the pressure pfofile inside the ejector. Most of the previous studies only valadated the first one. In a way, it is acceptable. But if you want to further investigate the mechanism, you should consider two factors mentioned above.

Good luck for you.

Most CFD simulations are specific in objectives, geometry, boundary conditions, etc.. If analytical or theoretical solutions or equations are applicable for the case, they can be used to verify and validate the simulations, otherwise experiments will be useful.

I agree with Weixiong Chen , we can say that this code is validated and can be used for the simulation of known phenomenon, but not for detailed and specific analysis when we want to verify for example the effect of one local parameters , In this case in general we have to adapt and adjust the code .

There is a comprehensive method for verification and validation of CFD results. Please see

Freitas, C. J. (2002). The issue of numerical uncertainty. Applied Mathematical Modelling, 26(2):237 –248.

Roache, P. J. (1997). Quantification of uncertainty in computational fluid dynamics. Annual Review of Fluid Mechanics, 29(1):123–160.

Roache, P. J. (1998). Verification of codes and calculations. AIAA Journal, 36(5):696–702.

There are two cases: 1) where you have written your own code for CFD study. Then validation of the code with a well defined published result is a must. 2) Where you are using a commericial code. In this case trust the software for solving accurately whatever you have asked it to solve. Normally, here we make a mistake in our choice of equations, boundary conditions etcs. So it is a good practice to first solve a well defined problem using the software so that you get a hang of all the choices offered. i also helps you in diving into the problem at hand.

So basically in both cases it is a very good practice that one should validate the results.

I agree with Kheng Ka Tan. Go to the compressible, 1 D equations and solve them. Compare the results with your simulation results. BTW/ I have gone through the paper you mentioned. The authors claimed that they solved the mathematical model!!. they are not. As Abdul Aziz said, the simulation solved the equation. The authors just presented the general governing equations of the flow. nothing about mathematical solution procedure. So, my advise; go to the basics and solve them by simple program in MATLAB, and validate you simulation results. I personally, dont agree to validate by comparing CFD simulation by commercial software with CFD simulation by commercial software. May be the visualization looks the same, but you must validate by values of pressure and velocity. 

Dear friends, it's all my pleasure to see your note to my question.

The matter is that I need to simulate flow in a steam ejector over and over again in order to optimize geometry. Unfortunately, it's too late and expensive (maybe impossible) to order an experimental setup. So, I'm searching for other methods of validation, as I found the latter paper that I shared.

In the case of the paper I shared, I'm agreed with Dr Chen and Belkacemi, the validation is insufficient or in my opinion, is weak. What concerns me is that whether I can use this method to validate my work or not?

Dr Chen

It's such a nice feeling to see your answer, it helped a lot. 

Here I should proudly note that one of your recent works was cited in my seminary article a couple of weeks ago :-) .

Thanks, Dr Moharreri.   And Good luck for your CFD works.

Thanks again for your affirmation of our work.

Yes, very much so. But not any CFD work. For each fluid dynamic problems there are pioneering CFD solutions which can be used to validate your Fluent /CFD reults. These solutions are called the 'Benmark Solutions'. For example, for a driven cavity flow, benchmark CFD solutions by Ghia and Ghia, for natural convection flows CFD solutions by Glen V Davis and analytical solutions by Chenoweth are used for validations.

Thanks Rajeev.