Multi-Objective Optimization of a Thermal Energy System using Surrogate Modeling and NSGA-II

📌 Project Overview

This project demonstrates an end-to-end framework for optimizing a thermal energy system (e.g., a heat exchanger). Since high-fidelity physical simulations (like CFD) are computationally expensive, this project uses Machine Learning to create a surrogate model, coupled with a Genetic Algorithm (NSGA-II) to find the optimal trade-off between competing objectives.

⚙️ Methodology & Workflow

1. Data Generation: Simulated 1000 data points for a thermal system with 3 design variables (Flow Rate, Temperature, Length).
2. Surrogate Modeling (AI): Trained RandomForestRegressor models to learn the underlying thermodynamics and predict Heat Transfer and Pressure Drop in fractions of a second.
3. Multi-Objective Optimization: Formulated the problem using the pymoo library to Maximize Heat Transfer while Minimizing Pressure Drop.
4. Decision Making: Applied the Epsilon-Constraint method to filter out solutions exceeding the pump's physical limits and extracted the ultimate Golden Design Point.

📊 Results: The Pareto Front

The optimization successfully converged, generating a distinct Pareto front that physically proves the trade-off between heat transfer and pressure drop.

🛠️ Technologies Used

• Python (NumPy, Pandas)
• Scikit-Learn (Random Forest for Surrogate Modeling)
• Pymoo (NSGA-II for Evolutionary Optimization)
• Matplotlib (Data Visualization)