Given sufficient time and funds you could design an F3A model and know precisely how it would fly before you even cut your first piece of balsa. This is how modern airliners are built. However this cost millions of $$ to do which is why it is effectively restricted to the design of airliners and military aircraft.

We are involved in the design of aircraft costs at most a few thousand $$. This might seem like a lot - it is for most individuals, but compared to the budgets for designing a new fighter jet for instance it is peanuts. Therefore it is generally not worth going through the expensive process of computer simulation of new model aircraft.

The next step down from computer simulation is of course the testing of models in wind tunnels. Whilst cheaper than full simulations it is still quite expensive to do any meaningful tests (a half hour borrowed in a wind tunnel just would not be worth the effort). And as you would need to build one if not several models to go in the wind tunnel, why not just build a real model and try flying it and see what happens?

Now this does not mean that aerodynamic theory does not have any place in the design of F3a models. It's just perhaps not as rigorously or as obviously applied during the design stage. You might not even know you are using it, but every time you make a change that makes an improvement you are effectively doing by trial and error, or educated guess, or intuition, what you could equally have tried and tested on a computer or in a wind tunnel. It just would not have been cost effective or practical for models costing such a relatively small amount of money.

There are of course overlying aerodynamic principals governing how all pattern models fly - that is why most of them look so similar now. These govern things like the required layout and relative areas of the wing and tail etc. The model specific differences come in at a secondary level, and a thick trailing edge will be this kind of secondary effect that will be effective on some models, less so on others. Ultimately if it works ok on one model then use it on that model. It might be worth trying on a different model but don't necessarily expect it to have the same level of effect. The reasons for it either working or not working could be explained (the maths is always right, you just might not have asked the right question!!). It just probably isn't worth the effort or expense to figure it out!

James