Routing optimization in telecommunication networks is based on the solution of multicommodity flow problems. We propose a new way to model multicommodity flow problems. The flows of commodities sharing a same source node are routed on a set of trees. For any tree rooted at a given source, the same proportion of all demands, originating at this source, is routed on the tree towards the destinations. To assess the efficiency of our approach, we apply it to the Maximum Concurrent Flow problem, which is known to be among the most difficult continuous multicommodity flow problems. We show that this tree-based model is exact, in other words, restricting flows to be routed on trees does not induce any reduction in the maximum concurrent flow value. Compared to the classical path-based approach, the new approach involves much less variables, while the pricing problem remains essentially the same, namely computing shortest-path trees. This leads to a drastic reduction of computing times as shown by our computational experiments. Finally, we provide a strongly polynomial-time algorithm for the special case where all commodities share the same source.