Computational Fluid Dynamics has a long history. This history relied on validation models. We never will know if predictions of any analytical model are correct. However, engineering analysis has always relied on validation of the models for known cases. For example, we know that flow over a cylinder has 0 lift and the numerical product should reflect that. From there, we can advance to wind tunnel data and then to flight test data to validate a numerical model.
As computational methodologies advance, we must be careful to not introduce errors from the computation and also that the computational analysis has built up its reputation on validation with example for cases where the answer is known. We can then progress into the unknown with a level of confidence in the prediction.
Currently, I am writing a book on the history of transonic flows and it's approximate 40 years of evolution. We must constantly keep in mind that faith in CFD solutions only comes about after years of painstaking and accurate comparison with real life accurate test data. This dance between analysis and test data has been in existence since we started modelling problems for their engineering solutions. This process cannot stop now that we are dealing with airborne viruses. We know what we know because we have developed faith in the answers from the analysis through years of comparison with real life test data. Airborne viruses, and their associated low speed flows are difficult to model and their is a stochastic nature to their CFD solution. However, there must still be validation before we just churn out a colourful particle tracker. More on this to come. Please come back to read more on the current state of CFD modelling and airborne viruses.