Probabilistic strength retention factors for steel and concrete and effect on structural reliability of columns in fire

Evaluating reliability of structures requires consideration of the uncertainties in demand and capacity. While material strengths exhibit a significant scatter at high temperature, no probabilistic model is available to quantify these uncertainties. To fill this gap, this work has compiled a database of test data on strength retention factors for steel and concrete, formulated a set of temperature-dependent probabilistic models based on these data, and applied the models in FE analyses of columns in fire. The proposed material models yield an average response similar to well-established deterministic models (Eurocode), but allow an explicit evaluation of the variability in structural fire response due to experimentally observed variability in material strength.