How exactly does a wishbone work? I have observed it being used 1) at jacket to pile connections to restrict vertical loads and allow lateral loads at coincident joints and   2) for connecting a beam placed directly on top of another ( Bottom of steel on Top of steel). 

In the second case, is it appropriately used? will a rigid link made by offsetting one beam above another provide accurate results? Is there any other way to properly model a secondary beam resting directly on top of a primary beam like a UDL in such a way that loads are transferred from top to bottom while both act monolithically?