Data‐driven reliability and cost‐based design optimization of steel fiber reinforced concrete suspended slabs

Abstract

AbstractThe governing failure mode for flat slabs is punching shear, which occurs when the punching forces exceed the slab resistances in the critical area around the column surface. Recently, considerable knowledge has been obtained on understanding steel fiber reinforced concrete (SFRC) suspended slabs as a sustainable and reliable technique to enhance post‐cracking behavior and resist loading effects, including punching shear. The properties of the materials, loading, and dimension vary inevitably and may be represented in the form of statistical distributions. In this study, all parameters affecting the punching shear of the SFRC suspended slabs were treated as random variables. Monte Carlo simulations (MCS) and punching shear model developed previously using Artificial Neural Network (ANN) were combined (ANN‐MCS) to conduct a reliability analysis, cost optimization, and model calibration. Reliability results have shown that the inclusion of fibers increased the safety level (20% in terms of the reliability index β) of the slabs against punching shear. Furthermore, a reliability‐based cost optimization was performed on the ANN‐MCS model. The optimization outcomes demonstrated steel ratio has a significant impact on SFRC slab cost, and thus it should be minimized/reduced when the steel fibers are included in the mix. Moreover, the ANN model was calibrated following a code calibration procedure to account for the model, material, and fabrication uncertainties. The calibration results have shown that a reduction factor of 0.6 should be applied to the ANN‐based model to achieve the target reliability level, β ≥ 2.5.