This paper deals with the choice of the optimal configuration of the district heating network to be built in an urban area. The users to be connected with the network are determined so that an economic objective function is optimized. In this approach, the average unit cost of heat is considered as the function to be minimized. An alternative heating system is considered for the users not connected with the network.In this work, three different iterative procedures are presented. All these procedures start with an initial superstructure connecting the possible users. The initial structure is progressively simplified by disconnecting one user at each iteration. The three procedures differ in the algorithm for the network simplification: the first procedure is deterministic, while the others use probabilistic approaches derived from the simulated annealing technique.The procedures are applied to a small portion of the urban tissue of Turin and their effectiveness is compared.