In this article, we present a method to obtain a C 1 -surface, defined on a bounded polygonal domain , which interpolates a specific dataset and minimizes a certain "energy functional." The minimization space chosen is the one associated to the Powell-Sabin finite element, whose elements are C 1 -quadratic splines. We develop a general theoretical framework for that, and we consider two main applications of the theory. For both of them, we give convergence results, and we present some numerical and graphical examples.