Alternatively of course you can model the soil itself (bricks, 2D vertical plates in a grill) and join the piles to the soil at nodes, what apart of precluding some relative movements for battered piles won't look nice.
Yet these models that include the soil directly modeled as an elastic continuum show that the calculated settlements (in theory of elasticity) are somewhat impervious to the nature of the foundation, especially if a stiff layer is not reached.
In any case, whichever the model used for the interaction, better use a range of soil stiffnesses to bracket the likely responses.
Settlement of groups of vertical piles are dealt with in
Bridge Substructure and Foudnation Design
Petros P. Xanthakos
Prentice Hall
p. 611...624