where to go for the computational modelling?

This post is to share where to go for the computational modelling. Should computation modelling focus on advance the solver convergence or focus on experiment verification?

The experiment verification always expensive and high cost involved. For solver convergence, it is always take time but less cost involved, however even with advance computational power and smart equation, we always face problem on verfiy our result through experiement stage.  

 Computational modelling, as its name indicated, we use the computer program to model the real world application and hope can optimize our design and improve the efficiency of the application. The application can refer to the opimization of pump performance, the fluid structure interaction of artery wall and blood, the wind energy efficieny and others different sectors. However, recently, the mesh independent test is not longer ensure the accuracy the computational modelling. Most of reviewers require us to conduct the experiment to prove our simulation results is correct. For example, if it is in the pump performance, they will require to test the pump in the reality by considering different factors like pressure. In terms of fluid structure interaction of artery wall, they will require you to test it in the phantom. However, we always has limitation in doing the experiment, as one experiment invovle man power and cost, and it is understood it take time for an hardward calibration, which is expensive and request extra funding. 

so my point is in the future, in computational modelling, an an engineering background researchers,is the experiment verification our focus? if yes, it means that we not longer contended where the good convergence and mesh independent test  achievement?