Adopting a Fuzzy C-Means Clustering Method to Evaluate Soil Liquefaction from SPT Measurements

Soil liquefaction occurs during major earthquakes in saturated, relatively shallow soils and is the primary cause of building foundation failures. Hence, assessing the likelihood of soil liquefaction has become one of the most important areas of study in geotechnical and earthquake engineering. However, the complex nonlinear behavior of liquefiable soils and the nondeterministic nature of earthquakes make liquefaction analysis difficult and uncertain. In the present study, we address this uncertainty by applying fuzzy set theory and adapting the SPT-based cyclic stress approach to the Fuzzy C-means Clustering Method (FCM). In this context, several liquefaction records are compiled from the cited literature and used to assess the possibility of liquefaction occurrence. The effectiveness of this proposed method in handling the vagueness of liquefaction occurrence is compared with the conventional deterministic method and a probabilistic method of liquefaction assessment. The results show the new method’s clear advantages over both the deterministic and probabilistic methods, particularly as the fine content of the liquefiable soil increases.