Desenvolvimento de modelos preditivos para a resistência ao fogo de paredes LSF – uma abordagem experimental, numérica e de aprendizagem automática

The two main objectives of FireLSF are i) to deliver predictive models to enable the performance-based design of Light Steel Frame (LSF) walls without the need to perform experimental tests and ii) to increase the knowledge about the fire performance of LSF walls, in particular with external configurations which are much less studied in the literature. Different predictive models are planned: - Simplified design methods that shall incorporate future revisions of prEN1993-1-2 design standard to calculate the fire resistance (FR) regarding the insulation (I) and load-bearing (R) criteria. For the Criterion I, the new method will be based on the additive component method the method and it is targeted to offer an accuracy within the range of 50-100 °C, in comparison with the reference results. For criterion R, the new method will be based on the correlation between the load level and the critical temperature, with the target set at an average ratio between the predicted value and the reference result between 0.85 and 1.0; - Numerical models based on FEM that will offer high-accuracy and high-fidelity results; - AI/ML models that will be trained and developed offering a high-accuracy and low-cost predictive model encompassing both criteria I and R, and accounting with the effect of thermal and noise insulation performance. An R2 score between 0.95 and 1.0 is targeted here between predicted and reference results. In addition to these major objectives, FireLSF aims to establish a correlation between functional requirements – thermal and noise insulation – and the fire resistance of Light Steel Frame (LSF) walls. This holistic approach holds a distinct innovative edge, as there is a notable absence of similar studies in the literature. In particular, it helps evaluate the dilemma between keeping the fire resistance and improving the thermal comfort at room temperature behaviour, because the use of cavity insulation contributes to improve the thermal comfort but reduces the fire resistance due to the introduction of a larger temperature gradient between the exposed fire side and the unexposed side. The acoustic effect of these insulation materials will also be also investigated. Also, we aim to provide experimental results on the fire performance of external walls, which are currently unavailable, to generate a comprehensive set of results, including the effect of heating in the external side and on both sides. It is planned 30 physical tests, providing valuable insights into the performance characteristics of external walls under fire conditions. This research presents also an opportune moment to conduct simultaneously small- and full-scale tests, allowing for a thorough comparison and exploration of the advantages and disadvantages inherent in each testing method thus establishing the influence of the scale effect. Also, it is required to evaluate the effect of the openings in LSF walls. This effect cannot be neglected as may be suggested by EN 1365-1, because the fire effect is transmitted to the surrounding LSF structure of the opening, reducing its ability to sustain the load both influence criteria I and R. Finally, our project will reach a Technology Readiness Level (TRL) 4 resulting from the comprehensive experimental campaign on LSF walls.