A coupled computational fluid dynamics (CFD) and heat-transfer model for an ice-covered fiberglass-reinforced plastic (FRP) hot stick, elaborated in a previous study, could well explain why the flow of partial-discharge current could be sufficient to raise the temperature of an iced pollution layer just below freezing, where the cold-fog flashover mechanism prevails. However, the ice-covered hot stick was modeled as a solid “ice rod” having an equivalent cross section of ice, meaning that the exposed ice surface is smaller in the model compared to reality. In addition, the simulations were performed for a relatively low wind speed of 1 m/s, while average wind speeds of 6.1–14.4 m/s were reported for the two Manitoba flashovers. Both of these problems are addressed in this paper to deal with the site incident conditions. The ice cover is considered as a thin layer having a thickness of 1 mm on the FRP hot stick. The effects of wind speeds of 0.1–15 m/s and wind direction as parallel and perpendicular to the ice-covered FRP hot stick are studied. This paper also presents experimental investigations on the most reliable reproduction of four separate FRP hot stick flashover incidents in Canada achieved at CIGELE laboratories.