Bayesian updating of subsurface spatial correlation through monitoring of infrastructure and building Developments

, 2016. As land space becomes increasingly sparse in urban areas, development of underground infrastructure becomes more important in major cities around the world. In these cases, uncertainties in subsurface ground conditions account for a significant proportion of construction problems and infrastructure underpcrformance. Recent advances in new sensing techniques open up opportunities to establish a holistic framework that duly considers spatial uncertainties of the subsurface environment, meanwhile processes sensing data to continuously update the predictions on system performance during construction and service life. Such potentials in promoting flexible design, 'observational method∗ in construction, and life cycle assessment of infrastructure have not yet been fully realised. This study introduces a preliminary framework for such purposes, by incorporating the Bayesian approach into recent findings of spatial correlations in geological profiles and geotechnical properties. The variables representing the subsurface spatial structure can be updated using the monitoring data obtained at various construction stages, through inverse analyses coupled with conditional random field simulations. This turns the subsurface environment into a self-Aware system fed by multi-source sensing information, where the risks and uncertainties are quantified and gradually reduced as the construction progresses. The updating process can continue as long as the sensors remain active, throughout the life cycle of the infrastructure where disturbances may occur due to subsequent construction activities nearby. The proposed framework is illustrated with a case study of residual building development in London, and it can be extended to different infrastructure development pertaining to different urban settings.