Engineered slopes are critical challenges in engineering practices. Failures of the engineered slopes can cause catastrophic losses of human lives and properties, which have been reported many times in history. Within the life cycle of a slope, its performance may vary with time due to deterioration and unexpected loadings or events. Regular monitoring is important since it can reflect the performance transition of the slope, and provide information for real-time evaluation of the slope reliability so that prompt effective risk mitigation measures can be undertaken when necessary. Besides, monitoring information also helps reduce the uncertainties associated with any calculation models and their input parameters. The primary objectives of this doctoral research are to develop a methodol...[ Read more ]

Engineered slopes are critical challenges in engineering practices. Failures of the engineered slopes can cause catastrophic losses of human lives and properties, which have been reported many times in history. Within the life cycle of a slope, its performance may vary with time due to deterioration and unexpected loadings or events. Regular monitoring is important since it can reflect the performance transition of the slope, and provide information for real-time evaluation of the slope reliability so that prompt effective risk mitigation measures can be undertaken when necessary. Besides, monitoring information also helps reduce the uncertainties associated with any calculation models and their input parameters. The primary objectives of this doctoral research are to develop a methodology to systematically characterize the variability of geotechnical systems; to evaluate the slope reliability based on the integration of multi-source and time-series monitoring information; and to analyse the performance transition of the engineered slope within its life cycle for effective slope management.

Analysis and design of an engineered slope face a significant amount of uncertainty. Inherent spatial variability associated with geologic conditions and parameters is a major source of uncertainty. An important task in site investigation is to characterize the spatial variability with limited field data. A method is proposed in this doctoral research to characterize the spatial variability using conditioned random fields. The limited measurement information within the site is used to constrain the random field. Due to spatial correlation, the site variability can be greatly reduced after incorporating the field data. Thus, a more accurate description of the geologic conditions and parameters can be obtained, which is desired for slope design and construction.

The spatial variability of soil and rock properties gives rise to scale-dependency in slope safety evaluation. Ignoring the spatial variability sometimes can lead to unconservative estimation of the failure probability of slopes. Random finite element method is a powerful tool to account for the spatial variability, but the computational efforts associated with this tool are large sometimes. In this doctoral research, a simplified reliability method is presented for slopes considering spatial soil or rock variability. In this method, equivalent homogeneous random parameters are used in a single random variable method (e.g., the first-order reliability method) to consider the spatial variability. The proposed method can produce a comparable failure probability as that calculated using a more rigorous random finite element method with the original spatially variable parameters, but the computational efficiency is largely improved.

As an additional safety measure, field monitoring is routinely conducted for engineered slopes. Observational information of different types and sources is collected. It remains a challenge to make use of the monitoring information to reveal failure mechanisms and assess the slope stability. A method is proposed in this study to assess the slope stability by integrating monitoring parameters with physical analysis. The observed information is first used to back analyse the strength and loading parameters, and then the updated basic parameters are used to calculate the factor of safety or failure probability of the slope. The dominant basic parameters whose uncertainties influence the observed results the most are identified from the probabilistic back analysis. Alert levels are defined in the monitoring parameter space based on a factor of safety or failure probability criterion.

Predicted performance of an engineered slope with complex geologic conditions and disturbance is subject to errors due to the presence of uncertainties. Monitoring data can also be used to complement the prediction of the future performance of the engineered slope by reducing the uncertainties in the prediction model and the input parameters simultaneously. A multi-step updating method is developed to enhance the prediction of future performance of the slope using incremental time-series monitoring data. The proposed method considers inherent uncertainty of the system, model uncertainty, and measurement uncertainty. The prediction is updated and improved gradually with new monitoring information using Bayes’ theorem. The proposed method is applied to a multi-stage excavation of a 530 m high rock slope.

Many high engineered slopes are stabilized using anchors. The performance of an anchored slope may deteriorate over time due to corrosion of anchors or other causes. Proper maintenance is essential to upkeep slope functions. However, there are no unified criteria for quantitatively evaluating the timing and engineering scheme for slope maintenance. In this research, a resilience model is presented to analyse the performance degradation of an engineered slope due to anchor corrosion, and evaluate the recovery of slope performance after maintenance. Failure probability is used as an indicator to characterize the evolution of slope performance within its service life. The timing for maintenance is determined when the failure probability reaches a designated intolerable value. Information obtained from regular inspections is used to reduce the uncertainties and improve the accuracy of the determined maintenance time. The resilience index attained after maintenance is used to indicate the effectiveness of a repair measure. A benefit index, which incorporates both the effectiveness and cost of a repair measure, is defined and used to conduct a quantitative evaluation of different maintenance schemes. The proposed resilience model may be used to establish an effective slope management program.