To investigate the influence of nozzle design on the homogeneity of mixing and smelting efficiency within a double-sided blowing furnace during the smelting process, a computational fluid dynamics simulation was carried out using the volume of fluid model, and the realizable turbulence k-ε model to simulate the liquid flow within the furnace. Analysis of furnace velocity distribution across operating conditions revealed that penetration depth variations correlated with bubble generation, bubble growth, bubble rise, and bubble breakage. To optimize molten pool mixing efficiency, four nozzle configurations were proposed: (1) cone straight shape; (2) excessive rounded corners; (3) conical; (4) anti-conical. Among these improved nozzle types, the conical nozzle demonstrated the best improvement effect. Compared to the ordinary nozzle, the dead zone proportion decreased by 16.57% and the penetration depth increased by 57.28%. Furthermore, this study also proposes combining nozzle types with distribution patterns. The relative conical mode stands out, showing an average velocity increase in 117.37%, a 37.17% reduction in dead zone proportion, and a 45.06% decrease in splash mass, compared to the ordinary nozzle.