Bayesian emulation of geotechnical deterioration curves using quadratic and B-spline hierarchical models

Abstract The stability of geotechnical infrastructure assets, such as cuttings and embankments, is crucial to the safe and efficient delivery of transport services. Factor of safety is a common metric used to quantify the stability of geotechnical infrastructure assets and computer experiments are an extremely useful method to model factor of safety over time. However, computer experiments are time-consuming to run. Therefore, we trained a fully Bayesian Gaussian process emulator using an ensemble of 75 computer experiments to predict factor of safety. We construct two different hierarchical models, one approximating the factor of safety temporal evolution with a quadratic model and one approximating the temporal evolution with a B-spline model; and we emulate their parameters. The Gaussian process emulator takes a slope’s initial conditions as inputs and outputs model parameters which provide a time-series of factor of safety. This work builds on Svalova et al. (2021) who modelled time to slope failure using a slope’s initial conditions. The successful emulation of factor of safety over time for slopes has the potential to inform slope design, maintenance, and remediation by introducing the time dependency of deterioration into geotechnical asset management.