We use a rolling two-stage procedure for solving a multistage stochastic program to assess the effects of uncertain fuel costs on optimal energy flows in the U.S. The optimal solution to the largest deterministic equivalent is obtained via Benders decomposition. We apply methods including temporal aggregation and scenario reduction to find approximate solutions which require less computational effort. These methods exploit both the network structure of the model and the multistage nature of forecast revision and uncertainty resolution. We evaluate the approximations based on similarity of the effects of uncertainty on the optimal flows compared to the exact solution.