Inverse response surface method for prestressed concrete bridge design

The paper describes application of the inverse response surface method when performing reliability-based design of selected bridge parameters. Employed adaptive procedure combines response surface method with an artificial neural network-based inverse reliability method. The first method serves for approximation of computationally demanding limit state function based on finite element method model. The second method provides design parameters for prescribed reliability indicators which correspond to analyzed limit states. Since the initial set of design parameters is not known in advance both methods must be efficiently combined to obtain sufficiently accurate results. The proposed method was employed for determination of selected uncertain material and prestress parameters of aging concrete bridge made of post-tensioned precast segments. Presented analysis is part of lifetime and reliability assessment of the aging bridge subjected to gradual deterioration which brings increasing level of uncertainty to its parameters.