A practice-oriented approach to control the effects of epistemic uncertainties on the seismic fragility of reinforced concrete frames

Uncertainties related to material properties and structural details in existing buildings can strongly influence the hierarchy of strength at both member and beam-column joint levels, particularly in the case of reinforced concrete frames. In turn, this can affect the global plastic mechanism, the force/displacement capacity of the structure and its seismic fragility. A rigorous probabilistic approach is arguably the most appropriate method to quantify the effects of such uncertainties on the parameters of the desired fragility curve. However, such an approach is not (yet) feasible in the current engineering practice. In this study, a practice-oriented approach to achieve the above goal is presented. More specifically, the proposed method relies upon a sensitivity analysis based on the Simple Lateral Mechanism Analysis (SLaMA). The latter is an equilibrium-based analytical approach to compute the plastic mechanism and the non-linear force-displacement capacity curve for a given structure. The capacity spectrum method is then adopted to assess the expected performance at different intensity levels and derive fragility curves. Initially, tentative but realistic values of the material properties (e.g., concrete and steel strengths) and structural details are assumed to calculate the above-mentioned quantities. Then, specific variations of those parameters are appraised, and the analyses are repeated to propagate the considered epistemic uncertainties in a simplified way. The effects of such variations are tested in terms of local hierarchy of strength, global capacity curve and fragility curve parameters. Within the seismic performance assessment of existing structures, the results of such sensitivity analysis can be used to drive/define in-situ testing campaigns targeted to the structural members most sensitive to variations of the considered parameters. This can potentially help reducing the overall cost/invasiveness of the testing campaign. The proposed approach is demonstrated for two case study frames, respectively sensitive and non-sensitive to variations of the considered parameters.