tlemmin,
Pretty much all of our attempts to use divalents with complex organics were troublesome. Inclusion of polarization helped a bit. We played with Drude polarizable Mg++, Zn++ and Ca++ and NMA, MeOH and H2O and they seems to be more or less o`k. The problem with divalents is that they are good example where classical force-fields start to fail for a variety of reason. To give an example, consider Zn++ and Mg++. They would have pretty similar pair distribution functions (ion-oxygen in water) and yet their experimental solvation free energies display significant difference. Partial charge transfer, diffuse electronic shell in d-elements and many many other reasons, you name it.
I never try to use it with C-36 lipids, but our attempts to simulate ++ in organic solvents were not a glaring success. So, the bottom line is that if you want to use ++, then go ahead but be prepared to be very critical in accepting results of simulations.
Another problem, as Rick pointed out lack of reliable experimental information. It seems that gold age of solution chemistry happened to be in 60-70ties and then it start to die out.
