Geostatistical investigation of geotechnical and constructional properties in Kadikoy–Kartal subway, Turkey

Abstract

The Kadikoy–Kartal subway is a unique underground transit line in the Asian part of Istanbul, and has been in operation since August 2012. The 22km underground subway line includes 16 stations and two main tubes. This study focuses on the last 13km, where the excavation was performed by excavators and loaders, to obtain the distribution of geotechnical and construction parameters from geostatistical methods. Rock quality designation (RQD), geological strength index (GSI), uniaxial compressive strength (UCS), elasticity modulus of intact rock (Em), and rock mass elastic modulus (Er) were determined from laboratory and site studies as geotechnical parameters. The daily advance rate (AR) for the excavation of underground tubes was collected as a construction parameter. Geostatistical methods were applied to estimate the magnitude of the unsampled points from sampled points. The validity of the proposed methodologies was confirmed by the previous researches, working with site specific parameters, determining the regional dependence functions, and applicability of kriging matrices. After regionalization of the parameters, semivariograms (SV) and crossvariograms (CSV) were determined to run kriging and cokriging techniques that were used to estimate the magnitude of regionalized variables (ReV). Cokriging is used to estimate the magnitude of Er, which has scarce data, from UCS, which is highly correlated to Er. The effect of areas that are changing between 49m and 212m for geotechnical and construction parameters were determined by SVs. This is important to show the applicability of the geostatistical methodologies prior to the construction stage for tunnel projects especially to prepare a well-organized drill hole program. Distribution maps for each parameter were then determined. The distribution of Er was obtained from cokriging technique that was not possible to determine from classical statistical and geostatistical techniques because of the insufficient number of data. Consequently, this research demonstrates that the application of kriging, cokriging techniques, and SV functions as geostatistical methodologies are usable for tunnel projects, therefore the techniques and interpretations based on the distribution maps can be used to assist the similar high budget subway projects.