The Vehicle Routing Problem with Time Windows (VRPTW) is a well-known and complex combinatorial problem. The static version of this problem has received special attention in last years. However, the most real world vehicle routing problems are dynamic, where information relevant to the routing can change after the initial routes have been constructed. The common approach in this scenario is restart the algorithm when novel information arrives, but his type of restart methods does not use the memory to improve the quality of the final result. In this paper, the hybrid CGH Column Generation Heuristic (Alvarenga et al. 2005) is modified in order to use the routes generated during the optimization process.