Fuzzy-based approach for determination of characteristic values of measured geotechnical parameters

Abstract

Determination of the nominal (characteristic) values of geotechnical properties plays a crucial role within the limits states design (LSD or LRFD) concepts. The interrelationship between the process of the selection of the nominal value and the safety level is not clearly addressed in most of the new limits states design codes of practice for geotechnical engineering. Estimation of the characteristic values (p% fractile or the mean value) using the stochastic models is often linked up with some assumptions regarding the probability distribution functions. Probability theory has been perceived as a unique methodology to handle uncertainty in these geotechnical parameters despite the fact that some of the uncertainties associated with these geotechnical properties may be nonstochastic in nature. In this paper, the uncertainty connected with measured geotechnical properties is modeled using the fuzzy-reliability techniques. The measured parameters are rendered into fuzzy variables and the nominal values are characterized by fuzziness. The procedure presented is proposed as an alternative or complementary method to the estimate of the nominal values of geomaterials. The approach is illustrated with computational algorithms and a numerical example.Key words: characteristic value, nominal value, fuzzy model, fuzzy variable, resistance factor, probability.