The most reliable way to validate your models is using prototypes, you can use scale size prototypes, with a wind tunnel or similar. You can use other software and run the same models under the same conditions, only in order to compare results, buy it is not totally reliable. My recomendation is the use of scale size prototypes, with sensors a meters.
By validation, we actually mean to validate the method to be used for the present computations. ANSYS, in general, uses Finite Volume Method for computations. So, in that case, run the computations for a sample problem (available in literature and which has been performed either experimentally or numerically) using your methodology under the same set of conditions as mentioned in the literature. If the results obtained by your method match appreciably with the existing results in literature, your methodology is validated.
To validate your numerical model, you'd better review the literature for at least one basic common test-case as a reference result which you can simulate. Obviously, you should select the test-case such that they have identical dimensionless numbers. Then you can compare your results with those reference results using quantitative parameters that have been reported in literature and see how your numerical model is accurate and precise. Depending on your specific problem, you can choose experimental, analytical or numerical results. The more reference results you use, the better you have validated your results. I can suggest you to use at least 2 separate reference results for validation.
Enrique Isaza Thanks for your valuable response. I have noted your valuable suggestion. I understand that I would need to make a prototype of the model I am intending to optimize and check how it responds to various parameters. Unfortunately , in the place I reside I do not have an easy access to the wind tunnel. Hence, the only option left is to compare my results with the results obtained from other software.
Kumar Sourav According to you, I need to validate my methodology, by performing a sample analysis with the same boundary conditions of the existing experimental/numerical work that is a bit similar to my research work.
You can look into any of the papers. You will get your answer.
For example, I perform computations using an in-house code for a problem which is already available in literature (experimental/numerical). If my result matches with that already available, I say that my methodology is validated and hence, the results which will be produced by my methodology will be accepted.
Also, do not forget to show the mesh convergence test for your present problem.
You should validate your simulations at least for every class of flow configurations you will be working on, if not for every specific case. The best source of validation would be benchmark quality measurements from suitable experimental campaigns that provide sufficient detail to allow a thorough comparison toward a holistic validation. High quality numerical simulations that have been accepted within the community as a benchmark can also be used as an alternate (or supplementary) source of validation. NASA and the American Institute of Aeronautics and Astronautics (AIAA) has run a number of workshops in recent years that deal with CFD validation for specific flow configurations, e.g., NASA CFD Validation Workshop, AIAA Drag Prediction Workshop, BANC Workshops for unsteady CFD simulations for aeroacoustic prediction, High Lift Prediction Workshops, Sonic Boom Prediction Workshops, etc.