Experiments at JET showed locally enhanced, asymmetric beryllium (Be) erosion at outer wall limiters when magnetically connected ICRH antennas were in operation. A first modeling effort using the 3D erosion and scrape-off layer impurity transport modeling code ERO reproduced qualitatively the experimental outcome. However, local plasma parameters-in particular when 3D distributions are of interest-can be difficult to determine from available diagnostics and so erosion / impurity transport modeling input relies on output from other codes and simplified models, increasing uncertainties in the outcome. In the present contribution, we introduce and evaluate the impact of improved models and parameters with largest uncertainties of processes that impact impurity production and transport across the scrape-off layer, when simulated in ERO: (i) the magnetic geometry has been revised, for affecting the separatrix position (located 50-60 mm away from limiter surface) and thus the background plasma profiles; (ii) connection lengths between components, which lead to shadowing of ion fluxes, are also affected by the magnetic configuration; (iii) anomalous transport of ionized impurities, defined by the perpendicular diffusion coefficient, has been revisited; (iv) erosion yields that account for energy and angular distributions of background plasma ions under the present enhanced sheath potential and oblique magnetic field, have been introduced; (v) the effect of additional erosion sources, such as charge-exchange neutral fluxes, which are dominant in recessed areas like antennas, has been evaluated; (vi) chemically assisted release of Be in molecular form has been included. Sensitivity analysis highlights a qualitative effect (i.e. change in emission patterns) of magnetic shadowing, anomalous diffusion, and inclusion of neutral fluxes and molecular release of Be. The separatrix location, and energy and angular distribution of background plasma fluxes impact erosion quantitatively. ERO simulations that include all features described above match experimentally measured Be I (457.3 nm) and Be II (467.4 nm) signals, and erosion increases with varying ICRH antenna's RF power. However, this increase in erosion is only partially captured by ERO's emission measurements, as most contributions from plasma wetted surfaces fall outside the volume observed by sightlines.