Degradation analyses of empirical inferential predictors for the development of improved dynamic mechanistic/empirical equations

Mammen, Ashlen

Degradation analyses of empirical inferential predictors for the development of improved dynamic mechanistic/empirical equations

Thesis / Dissertation

2025

Publisher

University of Cape Town

Department

Department of Chemical Engineering

Faculty

Faculty of Engineering and the Built Environment

Abstract

The paper presents an in-depth exploration of a debutaniser distillation column, a critical component in a typical separation train. The primary function of this unit is to separate the upstream distillation column product flow into LPG and a heavier stream of catalytic naphtha. The operation of the Debutaniser is crucial for maintaining the total C5 vol% and RVP within specified limits, ensuring optimal operation of downstream units. Given the high costs associated with real-time analysers, the study explores the development of various modelling techniques, including principal component analyses, decision trees, random forests, gradient boosting, neural networks and partial least squares, to optimize the prediction accuracy and process control. By leveraging these models, the study aims to enhance the automation and optimization of process units within chemical process plants, ultimately contributing to the overall efficiency of the chemical process plant.

Keywords

Degradation analyses

Reference:

Collections

Masters

Full item page