Data-driven uncertainty reduction in geotechnical engineering: Optimal preloading of a road embankment

Abstract

A typical geotechnical problem is the design of prefabricated vertical drains (PVD) and pre-loading with surcharge for an embankment on clayey soil. If not designed accordingly, excessive residual settlements might occur after the embankment has been taken into service, which result in damages to the embankment and its superstructure. Engineers face the task of finding an optimal surcharge design, which guarantees the integrity of the embankment while minimizing the amount of surcharge material used. This task is hindered by the considerable uncertainties inherent to the consolidation process of soft soils. To address this challenge, in 2021 one of the authors proposed a probabilistic geotechnical model that describes the long-term primary compression settlement of an embankment with initial pre-loading and PVDs.

Using this model, we proposed in previous work a risk-based approach for finding the optimal design of a surcharge for an embankment. We extended the geotechnical model, such that sequential preloading strategies based on settlement measurements can be considered. The optimal preloading strategies minimize the expected construction costs and penalties for construction delay or insufficient soil consolidation. On this basis, we investigated different heuristics and the resulting optimal preloading strategies for the case of a single measurement and one possible additional loading.

In practice, settlement measurements can be available at regular time intervals, for the case of continually monitored settlement. In this contribution, we extend the methodology of the previous decision optimization process to include predictions, which are based on continually monitored settlement. This extension also introduces the possibility of addition of surcharge at multiple time steps. We demonstrate the effectiveness in reducing uncertainties by considering more information and increasing the number of possible mitigation measures in the decision optimization process for the embankment preloading problem introduced in our previous work.