I also study this problem. I tried to use very small shells with very small bending thickness, additional to my real shells. The results seams allright but I don't really trust these kind of tricks, so if anybody knows a "legal" way of modelling pin connected shells, please write me.