The vibration and noise of pipe system is always a key problem which prevent the development of air condition. For finite element simulation of pipe system, the calculation precision contradicts the consumed time and cost. So, the approximation model method, which is based on the statistical theory, is applied to damping optimal design of pipe system in this paper. A hybrid experimental design approach including the orthogonal and uniform experimental design is introduced to produce the training data. The approximate model between the position, width of damping and frequency, damping ratio response are established by response surface method, Kriging and neural network. The Pareto curve between damping ratio and mass is got by NSGA-Ⅱ algorithm. At last, a test is carried out to validate the optimal result. In the damping optimization of pipe system, the approximation model methods enable to shorten the design process and reduce the computational cost, which give a feasible method of air condition simulation.