Nonlinear global design resistance: Case studies of post-tensioned concrete bridges made of I-73 and KA-61 girders

Abstract

The paper portrays a comprehensive computational procedure for determining the global structural resistances of two existing bridges made of I-73 and KA-61 precast post-tensioned concrete girders using advanced statistical assessment methods in combination with nonlinear fracture mechanics-based finite element method analysis. Although this combination is a powerful tool for realistic modeling of structures, its practical application is still very time consuming. Therefore, a statistical sampling approach for the determination of the structural design resistance is compared to selected efficient semi-probabilistic methods based on the estimation of coefficient of variation-estimation of coefficient of variation (ECoV) method according to fib Model Code 2010 and improved approach called Eigen ECoV method. Load-bearing capacity is determined for the ultimate as well as several serviceability limit states. The sensitivity of the input parameters burdened with uncertainties on the response of the structure is quantified using a sensitivity analysis supported by a surrogate model based on polynomial chaos expansion. The paper shows that the applicability of nonlinear modeling with respect to uncertainties is possible when using these ECoV methods and a surrogate model and can be applied in a routine manner. The shortcomings and advantages of all the used safety design/assessment methods are discussed.The paper portrays a comprehensive computational procedure for determining the global structural resistances of two existing bridges made of I-73 and KA-61 precast post-tensioned concrete girders using advanced statistical assessment methods in combination with nonlinear fracture mechanics-based finite element method analysis. Although this combination is a powerful tool for realistic modeling of structures, its practical application is still very time consuming. Therefore, a statistical sampling approach for the determination of the structural design resistance is compared to selected efficient semi-probabilistic methods based on the estimation of coefficient of variation-estimation of coefficient of variation (ECoV) method according to fib Model Code 2010 and improved approach called Eigen ECoV method. Load-bearing capacity is determined for the ultimate as well as several serviceability limit states. The sensitivity of the input parameters burdened with uncertainties on the response of the structure is quantified using a sensitivity analysis supported by a surrogate model based on polynomial chaos expansion. The paper shows that the applicability of nonlinear modeling with respect to uncertainties is possible when using these ECoV methods and a surrogate model and can be applied in a routine manner. The shortcomings and advantages of all the used safety design/assessment methods are discussed.