In computational grid, resource backup is the most important method for improving the availability of applications. It can be implemented by making advance reservation of redundant resources. But in order to make the grid applications more cost-effective, it is absolutely necessary to devise multiple policies for resource backup in order to provide flexible resource backup services. In this paper, we give the detailed discussion of several resource backup policies, which were briefly presented in our former work, and evaluate them with a grid resource backup simulator, GReBSim, which is designed as one of the simulating tools in our grid project. The simulating results show that combination of multiple policies is quite important for trading off in resource utilization and time to recovery for failed applications.