A hybrid framework for developing empirical model for seismic deformations of anchored sheetpile bulkheads. Issue 116 (January 2019)

• Record Type: Journal Article
• Authors: Gong, Wenping; Martin, James R.; Juang, C. Hsein; Dickenson, Stephen E.; McCullough, Nason J.
• Is Part Of: Soil dynamics and earthquake engineering. Issue 116(2019)
• Page Start: 192
• Page End: 204
• Publication Date: 2019-01
• Subjects: Anchored sheetpile wall -- Bayesian inference method -- Data-driven empirical model -- Seismic deformation; Soil dynamics -- Periodicals; Earthquake engineering -- Periodicals; Sols -- Dynamique -- Périodiques; Génie parasismique -- Périodiques; 624.176205
• DOI: 10.1016/j.soildyn.2018.09.032
• Languages: English
• ISSNs: 0267-7261

Abstract

Abstract: Damages to port facilities supported by anchored sheetpile walls are commonly observed at moderate to high seismic load levels. Predicting the extent of seismic deformations of anchored sheetpile walls is a necessary step in the performance-based design; however, the lack of straightforward and practical models poses great challenges for this prediction. In this study, a hybrid framework is proposed for developing practical empirical models for predicting seismic deformations of anchored sheetpile walls. First, a 2-D numerical model is calibrated with field case history data. Next, a series of numerical simulations are undertaken to characterize the relationship between the responses (i.e., seismic deformations of anchored sheetpile walls) and the inputs (i.e., variables that affect the responses). The model bias factor and coefficients (i.e., model parameters) in the derived coarse solution model are then characterized probabilistically, and further updated and refined with the collected database of field cases and/or experimental data. Finally, this hybrid framework is demonstrated through the development of data-driven empirical models for predicting seismic deformations of anchored sheetpile walls. The results show that the proposed framework is effective in developing a data-driven empirical model, and the obtained model can be easily updated with additional case history data, as needed, to improve the accuracy. Highlights: A hybrid framework is proposed for developing data-driven empirical models. The framework takes the advantage of cost-efficiency of numerical simulations and high fidelity of field cases and/or experimental data. The framework is adopted to develop data-driven empirical models for seismic deformations of anchored sheetpile walls. The proposed framework and the developed data-driven empirical models are shown effective.