Understanding and Predicting Flow Behaviour of Water and Air in Serpentine Pipes Using Machine Learning and CFD Modelling - Master by Research Thesis

Aug 3, 2026

Upload by Maeve

Two-phase flow of water and air inside serpentine pipes remains complex and insufficiently understood. This research combines regression-based machine learning with computational fluid dynamics (CFD) to improve prediction of key hydraulic characteristics under a fixed pipe diameter and varying water and gas velocities. Models are trained on experimental data from Cranfield’s process engineering laboratory to estimate pressure drop, void fraction, and liquid film thickness efficiently. CFD simulations with two geometries further support selecting the best representation for intermediate geometry behaviour. Results offer insight into two-phase flow physics and demonstrate a route to faster, more accurate serpentine pipe geometry optimization.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages196

Reading Time8 h 14 min

AccessFree to download

Abstract

Acknowledgements

Table of Contents

1 Introduction

1.1 Background and Motivation

CRANFIELD UNIVERSITY RAJESH SEKAR UNDERSTANDING AND PREDICTING FLOW BEHAVIOUR OF WATER AND AIR IN SERPENTINE PIPES USING MACHINE LEARNING AND CFD MODELLING THE SCHOOL OF WATER, ENERGY AND ENVIRONMENT Msc By Research In Energy And Power MSc by Research Academic Year: 2022-2023 Supervisor: Dr Patrick Verdin Associate Supervisor: Dr Liyun Lao Spetember 2023 CRANFIELD UNIVERSITY SCHOOL OF WATER, ENERGY AND ENVIRONMENT Msc By Research In Energy And Power MSc By Research Academic Year 2022-2023 RAJESH SEKAR UNDERSTANDING AND PREDICTING FLOW BEHAVIOUR OF WATER AND AIR IN SERPENTINE PIPES USING MACHINE LEARNING AND CFD MODELLING Supervisor: Dr Patrick Verdin Associate Supervisor: Dr Liyun Lao September 2023 This thesis is submitted in partial fulfilment of the requirements for the degree of Master by Research © Cranfield University 2023. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright owner. ACADEMIC INTEGRITY DECLARATION I declare that: 􀁸 The thesis submitted has been written by me alone. 􀁸 The thesis submitted has not been previously submitted to this university or any other. 􀁸 All content, including primary and/or secondary data, is true to the best of my knowledge. 􀁸 All quotations and references have been duly acknowledged according to the requirements of academic research. I understand that to knowingly submit work in violation of the above statement will be considered by examiners as academic misconduct. ABSTRACT The flow behaviour of two-phase fluid (water and air) in serpentine pipes is complex and poorly understood. This study aims to address this research problem by using machine learning concepts to gain a deeper understanding of the flow behaviour and optimize the geometry of serpentine pipes. Experimental data from the Cranfield process engineering laboratory was used in this work, fora fixed pipe diameter and varying water and gas velocities. Regression models were developed and trained. The study aimed to accurately predict the pressure drop, void fraction and liquid film thickness in serpentine pipes in a timely manner with high accuracy. In addition, Computational Fluid Dynamics (CFD) models were developed to predict the flow behaviour with two different geometries, and machine learning was applied to determine the best model for capturing the intermediate geometry flow behaviour. Results provide valuable insights into the behaviour of two-phase fluid in serpentine pipes. The use of machine learning in this research contributes to the field by offering a new approach for optimizing the geometry of serpentine pipes with improved accuracy and efficiency. The findings demonstrate the potential for machine learning to play a role in improving our understanding of two-phase fluid flow in serpentine pipes. This research is expected to have potential future applications in various sectors, including automotive, electronics cooling systems, and industrial and chemical processing systems. Keywords: process engineering, multiphase, void fraction, liquid film thickness, ACKNOWLEDGEMENTS I would like to extend my sincere gratitude to my esteemed supervisor, Dr. Patrick Verdin, and my associate supervisor, Dr. Liyun Lao, for their unwavering guidance and support throughout my research journey in the realm of computerbased studies. Your expertise, encouragement, and mentorship have been instrumental in shaping the trajectory of my project and helping me navigate through various challenges. Dr. Verdin, your profound knowledge and insightful guidance have provided me with a solid foundation in this intricate field. Your constructive feedback and thoughtful discussions have been invaluable in refining my research approach and expanding my horizons. Dr. Lao, your guidance and expertise have been a constant source of inspiration. Your insightful suggestions and academic insights have enriched my understandi

FAQs

What problem does the thesis address?

The thesis addresses the limited understanding of two-phase flow behaviour of water and air in serpentine pipes.

How is machine learning used in the study?

Regression models are developed and trained to predict pressure drop, void fraction, and liquid film thickness from experimental data.

How are CFD models and machine learning combined?

CFD models are built for two geometries, and machine learning is applied to identify which model best captures the intermediate geometry flow behaviour.

Aug 3, 2026

Upload by Maeve