mazibazi/seville-displacement_PlayGround

🏗️ Surface Displacement Predictor (Playground Edition)

This interactive Gradio app predicts tunnel-induced surface settlement in Seville using a trained ConvLSTM model based on FLAC3D simulations.

Developed by Maziyar Bahri, PhD (University of Seville). The tool allows users—especially architecture and civil engineering students—to visualize how excavation parameters and soil models affect surface displacement during twin-tunnel construction.

🎯 Features

• —Fixed Face Pressure = 250 kPa
• —Fixed Grout Pressure = 300 kPa
• —Adjustable soil type, ground treatment, and loads
• —Predicts surface displacement using a trained ConvLSTM network
• —Lightweight and interactive Gradio interface

📉 Figure 1. Twin Tunnel Geometry and Surface Settlement Pattern

Figure 1. Geometry of twin tunnels and the corresponding surface displacement trough. The first 50 excavation steps correspond to the advance of Tunnel A, followed by the remaining steps for Tunnel B. The maximum surface displacement occurs at X = –6 m, corresponding to the area above the midpoint between the two tunnels. The excavation extends along the Y direction from 0 to 40 m, representing the longitudinal tunnel length.

⚙️ Technical Details

• —Framework: TensorFlow + Gradio
• —Model file: best_convlstm_model.keras
• —Metadata file: metadata.json
• —Python version: 3.10
• —Gradio version: 5.0.0

📚 Citation

If you use this work or find it helpful in your research or teaching, please cite:

Bahri, M., Mascort-Albea, E.J., Romero-Hernández, R., & Soriano-Cuesta, C. (2025). Predicting Maximum Surface Displacement from Mechanized Twin Tunnel Excavation in Seville Using Machine Learning and FLAC3D Simulation. Geotechnical and Geological Engineering, Springer.