Resistance of steel cross-sections with local buckling at elevated temperatures

• Resistance of slender I-shaped cross-sections prone to local buckling is investigated at elevated temperatures.
• Numerical parametric study considering several cross-sections is performed.
• Comparison of the results with the existing formulae available in Part 1.2 of the Eurocode 3 shows that they need to be improved.
• New methodology to account for the local buckling in steel I-sections at elevated temperatures is proposed.
• It is demonstrated that the new methodology leads to good results when compared to the numerical results obtained and other reference (experimental) results.

In this work, the resistance of slender I-shaped cross-sections, where local buckling has a predominant role in the ultimate capacity, is investigated at elevated temperatures. A numerical study considering several cross-sections submitted to compression or bending about the major-axis is performed using a finite element analysis software. The results are compared with the existing formulae available in Part 1.2 of Eurocode 3 showing that they need to be improved. For Class 3 cross-sections, it is observed that the existing rules lead to unsafe results because local buckling occurs at elevated temperatures prior to the development of the elastic bending resistance or the gross cross-section compression resistance. For Class 4 cross-sections, the results show that these rules are not adequate because it is recommended for the design yield strength of steel the use of the 0.2% proof strength even if the cross-section has plates not prone to local buckling. A new methodology to account for the local buckling in steel I-sections at elevated temperatures is presented based on the expressions previously developed by the authors to calculate the effective width of thin plates at elevated temperatures. According to this new methodology, an effective cross-section is calculated for Class 3 and Class 4 cross-sections and the yield strength at 2% total strain is used for Class 4 cross-sections as recommended by Eurocode 3 for the other section classes. Finally, it is demonstrated that this methodology leads to good results when compared against numerical and experimental results.