A sectorisation-based method for geostatistical modeling of pressuremeter test data: Application to the Grand Paris Express project (France)

In the context of underground construction projects such as large-scale linear projects, knowledge of subsoil geotechnical properties is of major importance. An effective way to analyze and represent the variability of these properties is to use geostatistical methods. Despite the good performance of such methods, their use is not yet widespread in the field of underground projects. This paper aims to show the usefulness of geostatistics in the field of tunneling by proposing an original 3D sectoral modeling approach that allows dividing the geological formations in homogeneous geotechnical areas to better take into account the vertical and lateral variations of geotechnical properties. The approach is tested by developing a 3D geostatistical model of the pressuremeter tests on a section of a massive tunneling project in Paris (France): the Grand Paris Express. A strong focus is also made on the model input data: a methodology is proposed to overcome the limitations of the pressuremeter test in order to better study the spatial variability of the net limit pressure. A twofold analysis is conducted by comparing the modeling results obtained with and without pressuremeter data processing and those obtained with and without sectoral modeling. The results of the comparison show that: a) data processing is useful to better discern areas of extremes values of net limit pressure; b) dividing the geological units into homogeneous geotechnical sectors provides an advantage for geological units with geotechnical heterogeneities, as it leads to improved reliability of results. •An original 3D geostatistical modeling method for tunneling project is proposed.•Net limit pressure re-evaluation overcomes the pressuremeter test limitations.•Statistical methods identify homogeneous geotechnical areas within heterogeneous geological formations.•Sectorisation modeling outperforms global modeling for lateral geotechnical heterogeneities.