In this study, a numerical framework which jointly uses discrete element method, kinetic Monte Carlo method and lattice Boltzmann method is proposed to predict the electrochemical performance of solid oxide fuel cell cathode from raw powder. A total of 770 La0.6Sr0.4Co0.2Fe0.8O3-delta (LSCF) cathodes with different microstructures are numerically synthesized using various particle size, size distribution, pore former content and sintering time. Best performance is found at a porosity of approximately 0.40. The cathode performance degrades tremendously at about a critical porosity of 0.10-0.25 because the pores become disconnected. The critical porosity is found to have a positive linear dependency on the mean pore size. Cathodes that have smaller mean pore size may still operate functionally at a low porosity of 0.10. A response surface model (RSM) is proposed to predict the overpotential from microstructural parameters. The RSM is also found to be valid for real cathode microstructures and can be served as an off-line assessment tool for the cathode performance.