The paper presents a procedure to design water network. First of all, water reuse system, water regeneration reuse system (including regeneration recycle) and wastewater treatment system are designed separately. But the interaction between different parts demands that each part is designed iteratively to optimize the whole water network. Therefore, on the basis of the separated design a water network superstructure including reuse, regeneration and wastewater treatment is established from the system engineering point of view. And a multiobjective adaptive simulated annealing genetic algorithm is adopted to simultaneously integrate the overall water network to balance the economic and environmental effects. The algorithm overcomes the defect of local optimum of simulated annealing (SA), avoids the pre-maturation of genetic algorithm (GA) and finds a set of solutions (pareto front) in acceptable computer time. From the pareto front, a point with minimum fresh water consumption will be extended to zero discharge as our ultimate goal.