This study investigates the fire behaviour of building-integrated photovoltaic (PV) claddings, focusing on the progression from glass failure to ignition under different cavity conditions. Experimental tests were conducted on two common PV cladding types: bifacial dual-glass (GG) and monofacial glass–plastic (GP) modules. Results revealed that GP modules exhibited faster burning and higher peak heat release rates (HRR), reaching up to 600 kW, while GG modules burned more slowly with peak HRR between 50 and 100 kW. Cavity conditions, including depth, ventilation, and operational energization, were found to be vital in determining glass breakage, occurring between 400 and 550 °C, and cavity ignition and subsequent flame spread. The relationship between cavity fire dynamics and glass breakage suggests the importance of system design, particularly regarding cavity ventilation and flame barriers, for mitigating upward fire propagation. These results establish a basis for advancing numerical fire models through integration of critical parameters such as material properties, glass breakage, cavity ignition, and cavity configuration. This approach supports comprehensive real-scale analysis to guide the development of effective design recommendations, ultimately improving fire safety in PV-integrated construction.

Numerical fire models of photovoltaic (PV) modules are crucial for fire safety in building integration, supporting fire safety engineering analyses in designing such PV modules. This study develops a numerical fire model for PV panels using the ‘SPyro’ pyrolysis sub-model within Fire Dynamics Simulator (FDS), which enhances pyrolysis modeling through direct incorporation of cone calorimeter heat release rate per unit area (HRRPUA) data. Eight cone calorimeter experiments, exposing both glass and plastic sides of PV panels to 40, 50, 70, and 80 kW/m² incident heat fluxes, were conducted for calibration. The ‘SPyro’ sub-model utilizes experimental time-dependent HRRPUA curves to predict heat release. Thermodynamic properties, including ignition temperature and emissivity, significantly influence the simulated heat release rate (HRR). Multiple linear regression (MLR) was used to optimize these parameters by minimizing the deviation between simulated and experimental HRRPUA. Optimized emissivity values were 0.6 for glass and 0.7 for plastic, with ignition temperatures of 350 °C and 310 °C, respectively. Other thermodynamic parameters, such as material densities, thermal conductivities and specific heats were obtained from the literature. Comparing the numerical and experimental HRRPUA after using the optimized thermodynamic properties shows good agreement, especially regarding ignition time. Furthermore, results show that careful selection and optimization of input variables is crucial in using the ‘SPyro’ sub-model. This enables the model to be used in fire scenarios where fire spread over PV panels is an important fire hazard and important fire behavior characteristic for risk assessment.

Experimental investigations on the behaviour of ignition on the combined effects of wind, radiant heating, and ember accumulation, simulating short-spotting ignition of common construction materials, including plywood, oriented strand board (OSB), and polyisocyanurate (PIR) were conducted. Experiments were limited to red-hot glowing embers that were deposited sequentially at an ember flux rate of ∼2.0 pcs/m²s. The wind velocity, amount of radiant heating, and initial radiant heat exposure time were varied in this study. The results indicated that wind enhances ember combustion by increasing oxygen availability and turbulent mixing, leading to a higher probability of ignition, particularly for plywood. Wind also reduces the number of embers required to achieve ignition. However, at lower heat fluxes, wind can hinder ignition by dispersing volatile gases before they accumulate sufficiently to sustain combustion. This effect was particularly pronounced for PIR samples, where ignition probability decreased under high wind conditions but increased at intermediate wind speeds.