The heat transfer performance of a new fan-shaped pin-fin in a rectangular cooling channel is evaluated using three-dimensional Reynolds-averaged Navier-Stokes equations in comparison with a circular pin-fin. A multi-objective optimization of the fan-shaped pin-fin was performed using a hybrid evolutionary algorithm with surrogate modeling to maximize the heat transfer and to minimize the friction loss simultaneously. The fluid flow and heat transfer were analyzed using the shear stress transport turbulence model in the Reynolds number range of 5000-100,000. The results of our analysis indicate that the averaged Nusselt number with the fan-shaped pin-fin was improved considerably in comparison with that of the circular pin-fin over the whole range of Reynolds numbers. For the optimization of the fan-shaped pin-fin, the ratio of the radius of front part of the pin-fin to that of the rear part, and the lateral reduction angle of the pin-fin, were selected as the design variables. Fifteen design points were generated using Latin hypercube sampling, and the values of the objective function were evaluated at these points. The surrogate models for the objective functions were constructed using the radial basis neural network method. The Pareto-optimal solutions of the fan-shaped pin-fin show that the designs were improved with respect to heat transfer and pressure drop in comparison with the circular and reference pin-fins. (C) 2014 Elsevier Ltd. All rights reserved.