Based on the assumption that every section of pipeline is composed of two diameters, a mathematical model for optimizing the pumping tree pipe network was presented with the objective function of annual cost, using pump head, pipe length and standard diameter as decision variable. The infeasible degree function of solution in the species group was defined, and the constraint condition was handled by adopting annealing algorithm based on the infeasible degree of solution. The infeasible degree of solution was locally optimized through simulated annealing, to make it gradually approach feasible solution. Combined with the genetic algorithm global optimization and good local search capabilities of simulated annealing, a simulated annealing-genetic algorithm for the optimized design of the pumping tree pipe network was established. Simulation results showed that the model and algorithms have good performance for optimizing the pumping tree pipe network.