•  
  •  
 

Document Type

Original Study

Keywords

Modelling, Optimization, Solution Polymerization, Artificial Intelligence, PYTHON

Abstract

This research project focuses on the modeling and optimization of the solution polymerization of ethylene, using artificial intelligence (AI) via machine learning (ML) tools specifically PYTHON programming language enabled workspaces (jupyter notebook and google colab). The objective of the research was to develop a robust empirical model capable of predicting the polymer yield and mol. wt. based on key process parameters, and to identify optimal conditions that maximize yield and mol. wt. while ensuring operational safety and efficiency.

A comprehensive dataset containing process parameters gotten from previous experimental works was used for model development while predicting missing parameters to ensure data completeness for efficient data training. The modeling was carried out using two ML models, linear regression and random forest for dependability and reliability of the model with the yield and molecular weight of polyethylene as the dependent variables. The empirical models demonstrated strong predictive performance with a coefficient of determination (R²) of 0.86 and 0.93 and low MSE value of 0.0056 and 0.25 for yield and Mol. Wt. prediction respectively, indicating a high correlation between predicted and actual yields. Further, the model was optimized using the SCIPY quasi-Newton optimization algorithm on python, to determine the optimal operating conditions for maximum yield and molecular weight which were 90% and 3.2×106 g/mol respectively. The model’s performance was further validated using a scatter plot (actual versus predicted yields) and residual distribution curve which showed a strong alignment along the line of perfect prediction. This supports the model’s reliability and consistency, confirming effective captured patterns in the data thus highlighting the model’s robustness within the operational range relevant to industrial applications.

The study concludes that empirical modeling, when combined with machine learning and optimization tools, serve as a powerful decision-making tool in polymer production showing the importance of integration between chemical engineering principles and AI techniques.

Additional Files
APPENDIX.docx (5857 kB)
Download

Share

COinS