Liquefaction & Settlement of South Tehran Railway Using Energy-Al Method

The southern districts of Tehran are underlain by thick, saturated alluvial deposits highly prone to liquefaction. This study presents a novel hybrid framework combining the latest energy-based liquefaction evaluation method (۲۰۲۳-۲۰۲۵ updates) with an optimized artificial neural network (ANN) to assess triggering and long-term settlement of the Tehran-Qom high-speed railway embankments. Using ۴۲ detailed SPT/CPTu boreholes and seven scenario ground motions scaled to Mw ۷.۰ on the Rey Fault (PGA≈۰.۳۵ g), the energy-based approach predicts liquefaction (FS < ۱.۰) in ۶۹% of profiles between ۶-۱۸ m depth - ۶۸% more extensive than conventional stress-based methods. A feed-forward ANN (۱۲-۱۸-۱۴-۲ architecture) was trained and validated on ۴۸۷ high-quality case histories, including the ۲۰۲۳ Kahramanmaraş and ۲۰۲۴ Noto Peninsula events. The model achieved R۲ = ۰.۹۳۷ and RMSE = ۰.۴۱ % for post-liquefaction volumetric strain, outperforming current empirical and ML benchmarks by ۵۵-۶۴ % in accuracy. Predicted embankment settlements range from ۲۱ to ۴۸ cm, with differential settlements up to ۳۱ cm/۵۰ m far exceeding high-speed rail tolerances. Cost-effective mitigation via deep soil mixing, stone columns, and geosynthetic reinforcement is recommended. This hybrid energy-ANN methodology offers a rapid, accurate, and practical tool for liquefaction risk management of critical linear infrastructure in Iran's seismic urban corridors.

کلیدواژه ها:

Liquefaction ، Energy-based method ، Artificial Neural Networks ، Railway embankment ، Long-term settlement ، Southern Tehran ، Seismic hazard