Light Steel Frame (LSF) walls are widely used in building structures, used as partition walls and load bearing
walls. The LSF is usually protected by layers of homogeneous plates or composite plates, with or without
insulation materials in the cavity. This investigation presents the simulation results of composite LSF walls in
reduced scale and full scale, based on variable load levels (20 to 80%). The numerical model uses a hybrid model approximation, based on the experimental tests to accurately determine the temperature field. The numerical model is validated with experimental results, at reduced and full scale, both at room temperature and under fire conditions. This modelling technique can follow the thermal and mechanical degradation of the protection layers of the LSF wall and determine the fire rating for load bearing (R) and insulation (I). The fire resistance (R) decreases with the increase of the load level. The insulation ability is also predicted for different protection materials. Relevant conclusions are presented to increase the insulation ability of LSF walls, especially when using non combustible double protection layers. The insulation ability to sustain fire usually increases with the number of studs and with the application of insulation material in the cavity region. The ability to sustain the load under fire also increases with the number of studs, especially for higher load levels. The ability to sustain the load also increases with the number of protection layers, changing from 16% at 20% load level, to 42% at 50% load level. A new proposal is presented for the critical temperature of the LSF, based on the maximum temperature of the LSF during the fire, allowing the calculation of the critical temperature, based on the load level of the specimen. This relation can predict the fire resistance time, based on the preliminary thermal analysis of the specimen. The reduced scale specimens present higher critical temperatures when compared to the full scale specimens, due to the typical failure mode (local modes for the reduced scale and global modes for the large scale).
