Epoxy systems, cure rates, fiber saturation, material density, material orientation. As you can see there are so many variables at play I don't think our meger home setups allow for much consistancy. Our homemade boards are literally one of a kind items which is good I think
The real goal of developing a model for the flex etc is to be able to put a bit of rigour around what changing a particular parameter will do to the board 'all else being equal'. For example, if you add x mm of concave what could you expect to happen to the stiffness, if I reduce the core thinkness by 10% what is the impact, is having a step profile at the tips better than graded and why, if I use thinner s-glass in place of e-glass what is the impact flex and strenght.
I'm with you on the issue of there being a few variables at play but all the ones you've mentioned and more have been modelled pretty extensively in the analysis of beams and composites that is pretty plentiful on the web. I think that the thing that will make it make it v. difficult to predict the actual flex of the board in advance (rather that the 'all else being equal what happens if I change ....' ) is how variable the properties of the materials are from batch to batch. If I calibrate the existing simple model to a test point on my current boards and take into account that I squeezed all the concave out of it by clamping the b'jesus out of it to test then the agreement when placed under other loads is not too bad. Already its helped me get my head around the issue of reducing core thickness vs reducing the amount of glass.
The thing that I'm not sure about because its been a long time since I did any continuum mechanics (20 years+) is whether there are affects that I have left out of the model that will have a significant impact.