Evaluating the reflection of solar radiation by Building Integrated Photovoltaics (BIPV) with structured front-glass is challenging for two reasons. First, the resulting irregular scattering of light cannot be accounted for by simple reflection models. Second, the locations where the evaluation needs to take place cannot be predicted by simple geometric considerations, as in the case of mirror-like reflection. The detailed modelling of a BIPV module featuring a periodical, linear structure based on the scanning of its surface is described. The application of the model is demonstrated in the assessment of solar reflection by a concave array of the modules. The stochastic sampling of the BIPV modules from each location in a dense sensor grid, that is required to ensure that no local concentration of reflected solar radiation is missed, is supported by re-orienting the sensors. Rather than relying on a data-driven Bidirectional Reflectance Distribution Function (BRDF) model of limited directional resolution, the geometric structure of the front-glass is modelled by modulating the normal of a mirror-like surface. The method allows an accurate and dense simulation with moderate computational demand.