Frequent changes between inflow and outflow operations pose significant challenges in the design of bidirectional flow passage components with high efficiency and stability. In this study, hydraulic optimization of the inlet/outlet diffusion segment of a pumped storage, power station was performed. First, a 3D-optimization platform was established for the inlet/outlet diffusion segment, and consists of parametric modelling, automatic mesh generation, CFD numerical calculation, and an optimization strategy. Three objective functions; the head loss, velocity uneven distribution, and discharge uneven distribution; were adopted to evaluate the overall performance of the inlet/outlet diffusion segment. Both dual- and triple-objective optimizations were adopted to optimize the shape of the inlet/outlet diffusion segment, the response surface methodology (RSM) was used to generate approximate functions relating to the objectives and design parameters, and the non-dominated sorting genetic algorithm (NSGA-II) was selected to conduct the optimizations. The objective of the present study was to use a numerical optimization method to determine the optimal inlet/outlet structure configurations yielding better hydraulic performance with bidirectional flow conditions. The results show that with triple-objective optimization, the head loss decreased by 2.71%, velocity uneven distribution decreased by 21.05%, and discharge uneven distribution decreased by 2.24%. (C) 2017 Elsevier Ltd. All rights reserved.