Pitch-based foam carbon, a novel lightweight material, boasts excellent mechanical and thermoelectric properties, and tantalum film deposition on its surface can further enhance its performance. However, this deposition process often suffers from non-uniform deposition and suboptimal coating quality. To address these issues, this study systematically optimized the furnace structure by tuning pipe diameter, tilt angle, and porous media height. Numerical simulations of 216 models were conducted to evaluate the effects of these parameters on axial velocity, turbulence intensity (quantified by the vortex criterion Q > 1), and reactant concentration uniformity. The results showed that pipe diameters below 70 mm increased the mean axial velocity by 8-fold compared to larger diameters, whereas tilt angles of 15° and porous media heights of 60–80 mm yielded limited velocity enhancements of only 2%. Pipe diameter was identified as the dominant factor governing flow stability, inducing up to a 300% variation in the volume fraction of Q > 1, with minimal turbulence observed at the maximum diameter. In contrast, adjustments to tilt angle and porous media height had weaker effects, altering the Q > 1 volume fraction by 26% and 5%, respectively. Smaller pipe diameter (70–80 mm) also optimized TaCl₅ concentration uniformity; tilt angles between 0° and 30° showed negligible influence, while porous media height exhibited no definitive trend. Guided by the practical priorities of process evaluation, a multi-objective optimization was performed. The globally optimal structural parameters were determined to be a pipe diameter of 70 mm, a tilt angle of 15°, and a porous media height of 60 mm, which comprehensively balance deposition uniformity, process stability, and deposition efficiency. These findings establish pipe diameter as the pivotal factor for deposition homogeneity and provide a reference scheme for the structural design of industrial tantalum deposition furnaces and lay a foundation for subsequent multi-physics coupling studies and experimental validation.