The International Linear Collider (ILC) is a 200–500 GeV center-of-mass linear electron-positron collider, based on 1.3 GHz superconducting radio-frequency accelerating technology. This installation will require ~16,000 RF superconducting cavities operating within two linear accelerators at near absolute zero [[1]]. These SRF cavities are fabricated from pure Nb; to take full advantage of the Nb superconducting properties, the inner surface must be polished to a microscale roughness, and cleaned to be free of impurities that could degrade performance. Current methods use high viscosity electrolytes containing hydrofluoric acid, which is not conducive to low-cost, high volume manufacturing and is potentially harmful to workers. Faraday is developing an electropolishing process for niobium SRF cavities, based on a new and evolving paradigm of non-viscous dilute acid processing, enabled by a pulse-reverse electric field. Based on our understanding to date [[2]], we have speculated that the process works via oxide film formation controlled during a designed anodic pulse, followed by an off-time to remove heat and waste byproducts, followed by a cathodic pulse that removes the oxide film from the surface. This cycle is repeated many times per second, effectively removing niobium. The waveform design is such that the niobium is preferentially removed from the peaks on the surface, thus smoothing the surface. This talk will describe two recent efforts undertaken to improve understanding of various factors influencing the uniformity and speed of pulse-reverse electropolishing of niobium SRF cavities. The first is a flow study performed in a transparent plastic model of a single-cell (single-bell) cavity (Figure 1, left), to examine the flow dynamics in the absence and presence of an axisymmetric baffle fixed to the rod counter-electrode within the cavity bell. High-speed photography clearly shows the presence of a slow-moving eddy in the equatorial region of the bell (Figure 1, right), which is appreciably reduced in size when the baffle is present. Furthermore, rapid clearance of electrolysis gases and niobium oxide precipitates from the bell is expected to be strongly dependent on a proper configuration of flow throughout the bell. The second effort to be discussed comprises multiphysics modeling of the actual distribution of material removal in the EP process, as a function of position within the cavity. Modeling of EP of passive materials is complex, as numerous coupled phenomena must be accounted for, including: primary, secondary and tertiary current distributions; multi-phase effects, including fluid flow; and oxide formation/removal at the working surface. The strongest effects appear to be the primary and secondary current distributions, along with the surface oxide dynamics, inclusion of these physics (or semi-empirical approximations thereof) provides a significantly improved match between the model to the experimentally observed distribution of material removal, as compared to simulations incorporating only the primary current distribution. Figure 1 Caption (Left) Photograph of the transparent SRF cavity model, showing flow direction and early dye injection streaming pattern. (Right) High-speed photograph with annotations for bypass and eddy flows observed in a representative flow study test. References [[1]] http://www.linearcollider.org/ILC/What-is-the-ILC/The-project [[2]] M. Inman et al “Electropolishing of Passive Materials in HF-Free Low Viscosity Aqueous Electrolytes, J. Electrochem. Soc., 160 (9) E94-E98 (2013). Figure 1