Abstract: Abstract: In order to optimize the JSW10 type flow-ejecting self-priming pump, 4 key structural parameters are selected as target factors, which are outlet width of the impeller (Factor A), number of blades (Factor B), throat diameter of ejector (Factor C), and outlet diameter of ejector (Factor D), and the orthogonal experiment is designed with 4 factors and 3 levels, a total of 9 tests. The choice of structural parameters is mainly dependent on related research and the feasibility of parameter controlling in the tests is considered. Due to the complex geometry of the flow channel, the unstructured mesh which has good adaptability is adopted. Based on the grid independence, the numerical results tend to be stable when the grid number is above 3 million, and in this paper the mesh number of tests is between 3.6 million and 4 million. The whole flow field is simulated by the RNG k-ε model provided by ANSYS CFX, which turns out that the numerical results match well with the experiment data. Therefore, the numerical simulation can be used to predict the performance of the pump. We can get efficiency curves in different tests by numerical calculation and obtain the optimal level combination of the 9 tests. Range analysis at designed flow point is employed to get the order of each factor's impact on the performance and to obtain the optimal level of each factor from the 3 levels. The most influential parameter is Factor C i.e. the inlet throat diameter of ejector, and the optimal parameter level combination is A1B1C3D1 (the figure represents the level). Therefore we design the single variable test with Factor C as the optimization target, and perform the numerical calculation at low flow point, rated flow point and high flow point respectively, and the quadratic equation of head and throat diameter at different flow point is obtained by data fitting with MATLAB. We gain the optimal level of the inlet throat diameter of the ejector by function extreme value operation considering the operating conditions. Besides we get the optimal value of the inlet throat diameter of the ejector is 16.6 mm by the weighted average calculation. Finally, the verification is carried out. From the performance curves of the original model, the optimization model of orthogonal test, and the optimization model of the single variable test, we can see that the single factor optimization model is obviously higher than other schemes. In the flow field analysis, we can find that the throat of ejector has larger momentum exchange and velocity gradient transform, and it has a greater influence on the performance. Also the static pressure in the cavity and the outlet of the single variable optimization model is higher than the other 2 models from the pressure contours. Therefore, the single variable test is the most optimal judged from the efficiency. This paper provides a reliable method for the structural optimization of flow-ejecting self-priming pump by means of experimental design and data processing, which can be applied to the optimization of the similar pumps.