Influence of modelling strategies on uncertainty propagation in the alternate path mechanism of reinforced concrete framed structures

• Uncertainty propagation in the alternate path mechanism of RC structures is studied using global variance-based sensitivity analysis.
• Compressive arching action significantly influences uncertainty propagation.
• Alternate path mechanism is found to be a strength-sensitive problem.
• Use of displacement-based element formulation is not reliable when plastification is extreme.
• Structural idealization significantly reduces computational costs in probabilistic analysis.

Reliable performance-based evaluation of structures subjected to extreme loading scenarios, which is a function of structural response uncertainty characterization is a vital question. This paper aims to study the major sources of uncertainty in the alternate path response quantities of reinforced concrete framed structures subjected to sudden column removal scenarios. Using global variance-based sensitivity analysis, influence of nonlinear modelling approaches on uncertainty propagation is studied. Sensitivity of structural response quantities to the input uncertainties using plastic analysis with lumped plastic hinges found to be quite distinct compared with those of obtained using fibre-based modelling approaches, where axial–flexural deformation interaction, and in turn, compressive arching action is taken into account. Moreover, uncertainty propagation at different load levels is investigated using nonlinear incremental dynamic analysis (NIDA). Results obtained from displacement-based element (DBE) formulation, and force-based element (FBE) formulation were in good agreement at low to moderate load factors, where use of DBE formulation found to be computationally more efficient. However, at high load factors, where the structure is prone to progressive collapse, FBE formulation provides more reliable solution as the DBE formulation underestimates local response quantities, and in turn, underestimates the probability of failure. Finally, as uncertainty evaluation using fibre-based modelling approaches is computationally expensive, a substructure technique is applied, and influence of structural idealization on uncertainty propagation in structural response quantities is investigated. Results show that structural idealization is an efficient technique for reducing computational costs in terms of probabilistic analysis, especially when hundreds or thousands of simulations are needed to be performed.