I am also genuinely interested. I tried to add two vector field sampling modules in the same emitter, with the corresponding apply vector field modules. In pure theory it should work, as each of the nodes are adding to the transient forces to be finally used by the solver at the end of the stack. But in reality it does not work, it only applies the first of them. I was thinking about developing a customised module performing vectors sum in the sampled points. Any suggestion?
I really miss somebody from Epic giving us some feedback on this subject. I consider global fields a super powerful tool for Niagara, but as of today I found no solid reference.
David