Unexpected events often perturb railway traffic. The impact of these events may be very remarkable in terms of delay propagation. We analyze a mixed integer linear programming (MILP) formulation which aims at minimizing the delay propagation when traffic is perturbed. It does so by modifying train routing and scheduling at junctions. This formulation is able to solve to optimality many realistic instances in a computation time which is in line with real-time purposes. However, for the most difficult instances, finding the optimal solution is too time consuming. In this paper, we assess the performance of the MILP formulation when a short time limit is imposed. Moreover, we propose different methods for boosting this performance. We tackle instances representing traffic in the Lille-Flandres station (France), and we show that the boosted MILP formulation achieves very positive results, finding the optimal solution in more than 75% of the experimental runs.