Space Information Flow (SIF) is a new research direction that studies network coding in a geometric space, for instance, Euclidean space. This paper proposes a heuristic 3-D SIF algorithm that can solve the problem of min-cost multicast network coding in 3-D Euclidean space. The algorithm combines Linear Programming and non-uniform partitioning so that it can adapt to any density distribution among terminal nodes in 3-D Euclidean space. By simulations, we obtain a class of 3-D pentagram networks and show the feasibility of the algorithm.