This paper presents an approach that combines Integer Linear Programming (ILP) and energy use in order to find near-optimal solutions in a reduced amount of time. Considering that the resulting design can be unfeasible because of the probable changes in the nodes' heads and therefore in the demand flows, and also because of the reconstruction of the looped network, there is a necessity for various iterations of the methodology that explore the head assignation space in an intelligent way. The methodology is tested for different scenarios showing the advantages of this approach.