An improved robust formulation for constituent distribution in metallic nuclear fuels is developed and implemented into the advanced fuel performance framework BISON. The coupled thermal diffusion equations are solved simultaneously to reanalyze the constituent redistribution in post irradiation data from fuel tests performed in Experimental Breeder Reactor II (EBR-II). Deficiencies observed in previously published formulation and numerical implementations are also improved. The present model corrects an inconsistency between the enthalpies of solution and the solubility limit curves of the phase diagram while also adding an artificial diffusion term when in the 2-phase regime that stabilizes the standard Galerkin Finite Element (FE) method used by BISON. An additional improvement is in the formulation of zirconium flux as it relates to the Soret term. With these new modifications, phase dependent diffusion coefficients are revaluated and compared with the previously recommended values.The model validation included testing against experimental data from fuel pins T179, DP16 and T459, irradiated in EBR-II. A series of viable material properties for U-Pu-Zr based materials was determined through a sensitivity study, which resulted in three cases with differing parameters that showed strong agreement with one set of experimental data, rod T179. Subsequently a full-scale simulation of T179 was performed to reduce uncertainties, particularly relating to the temperature boundary condition for the fuel. In addition a new thermal conductivity model combining all available data covering 0 to 100% zirconium concentration and a zirconium concentration dependent linear heat rate solution derived from Monte Carlo N-Particle (MCNP) simulations were developed. An iterative calibration process was applied to obtain optimized diffusion coefficients for U-Pu-Zr metallic fuels. Optimized diffusion coefficients suggest relative improvements in comparison to previous reported values. The most influential or uncertain phase is found to be the gamma phase, followed by alpha phase, and thirdly the beta phase; indicating separate effect testing should concentrate on these phases.