Thermoelectric generators (TEGs) often use plate-fin heat sinks as cold side heat exchangers under forced convection. The available net electrical power obtained from these TEGs corresponds to that generated (Seebeck effect) minus that consumed (cooling fan). Generation and self-consumption have different trends as a function of the air flow speed, so a maximum of the net electrical power is expected when varying the cooling flow rate. Here, a semi-analytical model was developed to predict the maximum net electrical power of a single TEG module with a plate-fin heat sink with non-bypassed forced convection. The model was successfully validated with experimental data. It was applied to determine the heat sink design (fin thickness and fin-to-fin distance) that optimized the net electrical power for given values of hot source temperature, TEG properties, and duct cross-section. Numerical results indicated that the optimal dimensions of the plate-fin heat sink depended, among others, on the TEG effective properties. For a given TEG, the net output power was less sensitive to changes in fin thickness than in fin spacing. The optimal heat sink designs predicted by the model for the cases studied had fin thicknesses of 0.32 and 0.44 mm with fin-to-fin distances of 1 mm. This work was partially funded by the University of Girona under grant MPCUdG2016-4 Open Access funding provided thanks to the CRUE-CSIC agreement with Elsevier