A hybrid fuzzy zoning approach for 3-dimensional exploration geotechnical modeling: a case study at Semilan dam, southern Iran

This paper proposes a hybrid fuzzy zoning approach to compute the spatial variation and design the layout of additional samples during geotechnical investigation of an engineering structure. The geotechnical and geological data of Semilan dam site in southern Iran are utilized to evaluate this methodology. The optimal zoning of the site is found using the Fuzzy c-means (FCM) and comparison of results of various fuzzy clustering validity indices. The borehole samples are assigned to the selected zones according to their fuzzy membership degrees. In each zone, the geostatistical procedures are performed to model the 3-dimensional spatial variability and the related kriging variances. The estimated outputs of four zones are merged to generate the unit and reliable model. After fuzzifying the variables, the uncertainty index is defined as the function of rock quality designation (RQD), Lugeon, lithology index, index of dam structures, and estimation error. To model uncertainty in this site, the Mamdani fuzzy inference system (FIS) is used to establish the rule base relationship between the input and output variables. These outputs of uncertainty can be implemented as a guideline to design the layout of additional exploratory boreholes. In the studied area, five additional boreholes are suggested between the primary network.