Hazardous Facility Relocation Optimization: Gabès Phosphate Case

This project provides a robust decision-support framework for the relocation of the Groupe Chimique Tunisien (GCT) phosphate processing facility in Gabès, Tunisia. Utilizing Multi-Objective Goal Programming (MOGP) and Monte Carlo Simulation, the study evaluates potential sites based on population safety, economic costs, water security, and environmental integrity under conditions of uncertainty.

🎓 Academic Context

• Institution: Tunis Business School (TBS), University of Tunis
• Course: Business Optimization
• Authors: Khouloud Ben Younes & Montaha Ghabri
• Evaluated by: Pr. Dr. H. Essid
• Academic Year: 2025-2026

📝 Project Overview

The Gabès industrial complex is a pillar of the Tunisian economy but faces a severe socio-environmental crisis. This project addresses the "wicked problem" of relocation by:

1. Screening: Filtering 24 candidate sites down to 6 feasible locations using stochastic screening.
2. Modeling: Implementing a Lexicographic Goal Programming model that prioritizes human health (97% reduction in population exposure) over fiscal costs.
3. Simulation: Running 1,000 Monte Carlo iterations per site to account for parameter uncertainty (costs, hydrogeology, and demographics).
4. Optimal Solution: Identifying Boughrara (Medenine) as the robust optimal site.

📁 Directory Structure

.
├── Data and Plot Generation/      # Core computational scripts and datasets
│   ├── Monte Carlo Simulation.py  # Script for stochastic parameter sampling
│   ├── Plots_Generation.py        # Generates radar charts and deviation plots
│   ├── Rain_Prediction.py         # Supporting meteorological analysis
│   ├── System Parameters.xlsx     # Input data for the optimization model
│   ├── monte_carlo_results.csv    # Exported simulation data
│   └── tun_pop_CN_..._Image.png   # Population density visualization
├── GCT Docs/                      # External source material and audits
│   ├── 20100276_eia_fr.pdf        # Environmental Impact Assessment
│   └── Rapport_Audit_E_S-GCT.pdf  # Technical audit of the Gabès complex
├── Latex Report Files/            # Source files for the final document
│   ├── images/                    # Figure assets (raw deviations, radar charts)
│   ├── *.tex             # Modular LaTeX chapters (Introduction, Methodology, etc.)
│   └── main.tex                   # Main LaTeX compiler file
└── Project Report.pdf             # The final comprehensive research paper

🛠️ Requirements & Installation

The analysis is implemented in Python 3.12. To reproduce the results, you will need the following libraries:

pip install numpy scipy pandas pulp matplotlib seaborn

• PuLP: Used for solving the Mixed-Integer Linear Program (MILP).
• SciPy/NumPy: Used for the Cholesky decomposition and stochastic sampling.
• Pandas: For data manipulation and results aggregation.

🚀 Usage

1. Run Simulation: Execute Monte Carlo Simulation.py to generate the parameter distributions and perform feasibility screening.
2. Optimize: The optimization logic is embedded within the simulation scripts to find the optimal site (Boughrara) based on median values.
3. Visualize: Run Plots_Generation.py to produce the Radar Performance Comparison and Raw Deviation charts found in the report.

📊 Key Findings

• Optimal Site: Boughrara (Medenine).
• Safety Impact: 97% reduction in population exposure (from 8,500 to 280 persons).
• Economic Trade-off: Requires a 506 Million TND "safety premium" over the status quo.
• Robustness: Boughrara remains the optimal choice in 5 out of 6 weight sensitivity scenarios, proving it is not highly dependent on specific parameter variations (±20%).

📜 License & Contact

This project was prepared for academic purposes at Tunis Business School. For inquiries regarding the data or methodology, please reach out to moontahaghabry@gmail.com or khouloudbenyounes06@gmail.com