Tubular pumps and turbines are the water-to-electricity conversion devices widely used in pumped hydro storage systems for low heads and high flow rates. An ecological concern is the high proportion of fish injuries caused by the rotating blades. In this study, a tubular pump blade is retrofitted to balance the energy loss and fish friendliness through multi-objective optimization, combining computational fluid dynamics with a mathematical blade strike model. Sensitivity analysis identifies nine geometric variables out of thirteen for optimizing blade design. An approximation model is developed to implement optimization algorithm and achieve the objective function. By artificially assigning weights to targeted performance metrics, three scenarios are obtained: the optimal efficiency scheme (A), the optimal fish friendliness scheme (C) and the balanced scheme (B). Scheme C achieves a fish survival ratio of 100 % at multiple flow rates (for fish lengths of Lf/D=1/12), but results in a 6.4 % decrease in efficiency compared to scheme A. Additionally, a negative blade attack angle can improve flow pattern. Considerable reduction in fish mortality can be obtained by equipping the rotor with a larger size but lower shaft speed, while also limiting the fish size to pass through the pump running at a reduced flow rate.