The kinetic CO2-over-N2 sieving capabilities in narrow pore zeolites are dependent on the free-energy barriers of diffusion between the zeolite pores, which can be fine-tuned by altering the framework composition. An ab initio level of theory is necessary to accurately compute the energy barriers, whereas it is desirable to predict the macroscopic scale diffusion for industrial applications. Using ab initio molecular dynamics on the picosecond time scale, the free-energy barriers of diffusion can be predicted for different local pore properties in order to identify those that are rate-determining for the pore-to-pore diffusion. Specifically, we investigate the effects of the Na(+)-to-K(+) exchange at the different cation sites and the CO2 loading in Zeolite NaKA. These computed energy barriers are then used as input for the Kinetic Monte Carlo method, coarse graining the dynamic simulation steps to the pore-to-pore diffusion. With this approach, we simulate how the identified rate-determining properties as well as the application of skin-layer surface defects affect the diffusion driven uptake in a realistic Zeolite NaKA powder particle model on a macroscopic time scale. Lastly, we suggest a model by combining these effects, which provides an excellent agreement with the experimental CO2 and N2 uptake behaviors presented by Liu et al. (Chem. Commun. 2010, 46, 4502-4504).