For the planning of telecommunication transmission networks, a cost- optimal network structure must be calculated under reliability constraints. The situation can be modelled by a multicommodity network with (fixed charge) concave cost functions, where (some) traffic loads must be split up and routed over more than one independent (edge disjoint) path. For the resulting concave cost minimization problem a heuristic procedure is proposed, to find a sequence of local minima with decreasing value of the objective function; the procedure is based on an improved marginal pricing technique and a multirouting extension of Yaged's method [see \textit{B. Yaged}, Networks 1, 139-172 (1971; Zbl 0228.90011)] to find minimum cost routing for static network models.