Analysis of the behaviour of retaining structures through a novel data interpretation approach

Abstract

Retaining structures are generally designed to satisfy serviceability limits on allowed deflections while also providing sufficient strength to carry the bending moment caused by the soil pressure acting on the wall in tandem with any applied loads. Since wall displacement and mobilised soil pressure are the most important design parameters, direct measurement would be ideal, however instrumentation limitations leave the measurement of wall curvature using strain gauges and rotation using inclinometers as the industry standard for retaining wall monitoring. Pressure and deflection are normally inferred from these discrete, sparse and potentially inaccurate measurements through beam theory by fitting the discrete data to obtain differentiable continuous curves. Conventional fitting methods cannot simultaneously fit data from different types of instrumentation like inclinometers and strain gauges and there is generally little flexibility in addressing issues when the fits do not make geotechnical sense. A novel fitting approach is developed to address these shortcomings, with the ability to pull together imperfect data from a multitude of instruments (strain gauges, inclinometers, displacement transducers etc.) into a cohesive prediction. The method is validated based on field and centrifuge data in which it outperforms existing fitting solutions. This novel approach significantly simplifies data analysis by enabling practising engineers and researchers to simply input all instrument data from a retaining structure and retrieve sensible and accurate predictions of the parameters of soil-structure interaction which are compatible with all data measured and insensitive to measurement errors to a high degree.