Valuable information from reliability analysis of pile foundation

• Autor(es): Teixeira, Ana ; Correia, A. Gomes ; Henriques, A.
• Data: 2011
• Identificador Persistente: http://hdl.handle.net/1822/15465
• Origem: RepositóriUM - Universidade do Minho
• Assunto(s): Pile; Axial load; Reliability; Uncertainties; Sensitivity analysis; SPT
• Tipo de Documento: Documento de conferência
• Idioma: Inglês

Descrição

The work presented in this paper was developed under the PhD thesis entitled “Reliability and Cost Models of Pile Foundations”. Our main goal is to present guidelines for geotechnical engineers to carry out reliability based designs (RBD). This analyses help evaluate the probability of a particular behavior in a time period, with the knowledge of the input parameters randomness (uncertainties). The biggest benefit is that it quantifies and gives information about the parameters that mostly influence the behavior under study. This capacity is important, not only because of the new regulation codes and social concerns, but also because these probabilistic formats support decision making under uncertainties, providing qualitative judgments and investments, very important in geotechnical area. Based on that, this work shows valuable information that a geotechnical engineers can obtain from a simple reliability analysis of a pile foundation, such as the most influential uncertainties in pile design or the minimum dimensions of the pile (Figure 3a) and maximum load (Figure 3b) that lead to a previously established probability of failure. Furthermore, a comparison between two widely known RBD methodologies was done, First Order Reliability Method (FORM) and Monte Carlo Simulations (MCS). FORM is the most traditional one, an approximate method (level II of reliability), while ordinary MCS has a higher level of accuracy (level III, pure probabilistic) and is a very straight forward method. Nevertheless, FORM has some limitations when complex performance functions are necessary and it is not possible to approximate normal distributions. The uncertainties considered in this work and the methodology scheme is presented in Figure 1 (Honjo et al., 2010 and Honjo et al., 2011).