In this paper, a flat tube of an engine radiator is modeled numerically for improving the cooling process or heat recovery of the engine using nanofluids. Two hydrogen based fluids (water (H2O) and ethylene glycol or EG ((CH2OH)2) and four nanoparticles (CuO, TiO2, Al2O3 and Fe3O4) in different shapes (Brick, Cylindrical, Platelet and Spherical) are considered for modeling the nanofluids in four different Reynolds numbers (500, 1000, 1500 and 2000). Hamilton correlation is used to calculate the thermal conductivity of nanofluids in different shapes of nanoparticles. Furthermore, the effect of nanoparticles volume fraction on the Nusselt number for all nanoparticle shapes is discussed in this study. Results show that EG-TiO2 with platelet shape and larger volume fraction of nanoparticles has the best cooling performance for the engine among other modeled nanofluids.