We deal with the problem of Vehicle Routing Problem with time windows in a dynamic environment (DVRPTW). Indeed, we will serve a set of customers using homogeneous vehicles while respecting their maximum load. Each customer has a known demand and geographic location and must be served within a predefined time. Two objectives should be optimized: (1) the total transportation cost and (2) the total travel risk, under dynamic environments. Besides, multiple constraints are to be considered to avoid violating transportation rules. The three major contributions are as follows: first, to model mathematically the problem by considering the possible dynamics in the environment, and with a new risk function. Second, two solutions approaches are proposed: a bi-population genetic algorithm (GA) and a hybrid evolutionary algorithm combining GA and Variable Neighborhood Search (VNS). And finally, to develop a decision support system.