• Research Reports
• Geographic Roadway Inventory Tool (GRIT)
• Event Proceedings
• Staff Presentations
• Other Resources (Surface Selection Tool, Grain Industry Data, etc.)

Research Reports
Report Details

Abstract

This report presents several advancements in the empirical modeling of liquefaction-induced lateral spread. It starts with a newly collected dataset of 5,560 historical lateral spread displacement vectors, a sample size over 10 times larger than the existing databases and subsurface data comprising over 633 standard penetration test boreholes. This work presents a comprehensive comparison of state-of-the-art empirical models for lateral spreads through Monte Carlo simulations and sensitivity analyses and proposes new evaluation metrics to measure performance. It also quantifies the uncertainty of model weights of the Multiple Linear Regression (MLR) model using Bayesian Statistics. A new functional form is proposed for the MLR model using the least absolute shrinkage and selection operator method. Importantly, the conventional probabilistic framework for predicting lateral spread is expanded to account for the probability of lateral spread triggering given the triggering of liquefaction. This expansion allows us to model zero-displacement lateral spreads despite having liquefaction susceptibility. A convolutional neural network classifier is developed to model the probability of lateral spread triggering with an out-of-fold model accuracy of 90.5%. A new mathematical representation of soil types is presented and trained in the context of liquefaction and lateral spread and boosted model performance.

How to Cite

Bartlett, Steve, and Massoud Hosseinali. Development of Next Generation Liquefaction (NGL) Database for Liquefaction-induced Lateral Spread, MPC-22-477. North Dakota State University - Upper Great Plains Transportation Institute, Fargo: Mountain-Plains Consortium, 2022.