Unlocking the potential of digital 3D geological subsurface models for geotechnical engineers

Within any construction project the most significant factor in controlling the cost and feasibility is the subsurface ground conditions. This is particularly the case in underground construction. Geological modelling in three-dimensions (3D) can provide a detailed definition of sub-surface conditions. Such modelling requires the extension of traditional GIS methods to handle the volumetric representations. Over the past two decades, a series of sophisticated 3D modelling technologies have been developed to address this need. However, the adoption of these techniques in the geotechnical industry has lagged behind technological advances. Two contrasting approaches to 3D geological modelling are presented: a) the Thames Gateway Development Zone (TGDZ) in London, UK and b) subsurface characterisation studies in Boston, USA. The TGDZ studies used ‘GSI3D’ software, while the Boston studies involved geostatistical evaluations of the field data and the Environmental Visualization System (EVS) for model creation and visualisation. Both studies have created 3D geological models attributed with physical and mechanical property data, but this has been achieved in two different ways. The TGDZ study provides a single uniform property attribution to individual geological units, whereas the Boston studies used geostatistical methods (kriging) to interpolate borehole sample data onto a 3D structured mesh. This 'discretisation' allowed the development of volumetric models that quantified the variability of the data used to build the property model. These different modelling methods provide solutions to two very different problems. In the TGDZ, the requirement was for regional scale information for ground investigation design, for assessing water management strategies, and as a tool for communicating information to non-geo-specialists. In this situation, the best approach was a system for model building that did not require a specialist modeller, the use of bulk attribution, and the ability for modelling to be carried out quickly using a desktop computer. However, in Boston, a more specialist solution was required to provide a detailed understanding of the natural variability of the complex geology, thus discretisation and spatial interpolation of sample data values was necessary. The 2001 EuroConference in Spa, Belgium, that addressed characterisation of the shallow subsurface, identified four major constraints on the use of 3D digital geological data: This paper shows that these constraints are being overcome with the use of new modelling software and techniques and, more importantly, with an understanding of the needs of the client.