Robustness to Uncertainty: An Alternative Perspective in Realizing Uncertainty in Modeling Deflection of Reinforced Concrete Structures

This paper describes an alternative perspective in evaluating of computational models for predicting deflection of cracked reinforced concrete (RC) structures. The concept of robustness to uncertainty in modeling input parameters is introduced in addition to the classical perspective of modeling accuracy. A new method to quantify robustness to uncertainty is proposed and applied to the problem of modeling deflection of RC structures. Monte Carlo simulation is used for uncertainty propagation in serviceability calculations. The propagation of random uncertainties in deflection of a continuous one-way RC slab is examined. We introduce a method for the practical use of the proposed robustness metric by considering the reliability analysis. A cracked plane frame model to simulate deflection of the RC slab was adopted. We show that by considering a threshold of uncertainty in predicted deflection, the robustness to uncertainty in model inputs for different deflection prediction models can be evaluated. Our investigation proves that a tradeoff between model robustness and model accuracy exists in modeling deflection of RC structures.