An inverse reliability method and its applications in engineering design

The concepts and methods of reliability theory have been developed to take into account intervening engineering uncertainties in modern engineering design. To achieve a certain performance an engineer must find design parameters which meet corresponding pre-set target reliability levels for the different limit states considered. Traditionally, this inverse reliability problem is implemented by means of "trial and error," using a forward reliability procedure and interpolating the design parameters at the desired reliability. This approach is inefficient and involves difficulties resulting from repetitive forward reliability analysis. Thus, it is desired to develop an efficient and more direct approach to determine the design parameters for specified target reliabilities. A general inverse reliability methodology is proposed in the present work, which allows the direct determination of the design parameters when the corresponding target reliabilities are given. The design parameters can also be associated with random variables, such as their mean and standard deviation. As an extension of the inverse reliability method, an inverse reliability-based optimization procedure is also developed. In contrast to traditional reliability-based optimization, this method permits the separation of the ordinary optimization method and the inverse reliability analysis. For cases requiring an intensive computational effort or having a non-smooth limit state function, a response surface method for inverse analysis is proposed to implement and speed up the calculations. The efficiency and applicability of the proposed method are verified with several practical engineering applications, ranging from offshore and earthquake engineering to manufacturing of carbon fiber reinforced composite components.Applied Science, Faculty ofCivil Engineering, Department ofGraduat