Modeling spatial geotechnical parameter uncertainty and quantitative tunneling risks

Abstract

In underground construction and tunneling (UCT) works, geological and geotechnical uncertainty is often the most significant source of risk associated with a project. Unforeseen adverse geological/geotechnical conditions can lead to significant construction issues causing reduced tunnel advance rates and schedule delays, cost increases, damage to existing infrastructure, and/or damage to construction equipment. However, current practice falls short of reliably assessing spatial uncertainty in geological conditions and geotechnical parameters. This thesis addresses the spatial uncertainty encountered in UCT works, utilizing the site investigation data from several tunneling projects. The analysis of typical geotechnical site investigation data for tunneling projects demonstrates that there is often sufficient data to assess the spatial correlation structure, and that this structure can vary by engineering soil unit type. Therefore, an approach to modeling the spatial variability and uncertainty of geotechnical parameters, while jointly considering the geological variability and uncertainty, was developed. In addition, the levels of uncertainty in engineering soil unit classification and a representative geotechnical parameter, N1(60), encountered in three distinct UCT projects is assessed. These projects enable a broad assessment of different project scales, degrees of site investigation effort and geological conditions, and the corresponding uncertainties. The results of this analysis reveal that spatial uncertainty can differ significantly both within a project and across multiple projects. Furthermore, while a higher degree of site investigation effort leads to lower average uncertainty in the engineering soil unit estimation, the same is not necessarily the case for N1(60). The analysis also conveys that quantified uncertainty in geotechnical parameters is influenced by other factors in addition to site investigation sample spacing including uncertainty in geological variability and measurement error. This thesis also presents a novel application of geostatistical simulation for risk assessment and mitigation planning for two frequent risks encountered during mechanized tunneling: clogging and cutter tool wear. The quantification of spatial uncertainty in geotechnical parameters enhances the risk assessment as the uncertainty in risk estimates can also be quantified spatially. The results of the proposed mitigation plans (water injection and cutter tool replacements) from the geostatistical model approach were found to be in strong agreement with the project data, validating both the methodology and the value of performing such detailed analysis for tunneling risks.