Final Year Project Topics

FYP Proposals

A collection of project ideas in CFD, machine learning, turbulence, and aerospace applications. Each topic is meant to be exploratory, hands-on, and engaging.

How these projects are approached

These projects are designed to be both challenging and enjoyable. The CFD component will be personally supervised by me, with regular guidance on setup, interpretation, and next steps, so that students can focus on understanding the flow physics and building confidence as the work progresses.

D0302026 S1AerodynamicsAssigned to Saxena Raunak

Towards Intelligent Aerodynamic Design: CFD and Machine Learning for Airfoil Optimization

Objective

What happens when two airfoils work together instead of one? In this project, you will explore the aerodynamic behaviour of tandem airfoil configurations and study how their interaction influences lift, drag, and flow development. Computational fluid dynamics (CFD) will be used to generate detailed flow data, while machine learning will be introduced as a way to uncover patterns and predict aerodynamic performance across different tandem arrangements. The aim is to connect physical flow behaviour with data-driven modelling, and to see how intelligent tools can help accelerate aerodynamic design.

Scope

The project begins with a structured introduction to CFD and tandem airfoil configurations, where you will simulate flow over two interacting airfoils under different geometric and operating conditions. Parameters such as spacing, relative positioning, and angle of attack will be varied to observe how the flow field and aerodynamic forces respond. From these simulations, a dataset will be built using quantities such as lift, drag, and pressure distribution. Machine learning models will then be developed to learn relationships between configuration parameters and aerodynamic outcomes, allowing faster prediction of performance trends. Throughout the project, guidance will be provided in both the CFD and machine learning components, so that the focus remains on understanding how the two approaches work together.

D0292026 S1CFD + MLAssigned to Huang Nuo Yan

CFD and Machine Learning-Based Optimization of Active Stack Ventilation Systems in High-Rise Residential Buildings

Objective

How does airflow actually move through a high-rise apartment when assisted by a fan-driven system? In this project, you will explore the behaviour of active stack ventilation and how design choices influence indoor airflow. Using computational fluid dynamics (CFD), you will study how parameters such as stack size, placement, and inlet openings affect air movement and ventilation quality. Machine learning will then be used to uncover patterns in the results and help predict how different designs perform. The goal is to build a practical understanding of how ventilation systems work and how they can be improved.

Scope

The project begins with a structured introduction to airflow in built environments and the principles behind active stack ventilation. You will develop a simplified residential model and simulate airflow generated by a fan-driven vertical stack using CFD. Key parameters such as stack dimensions, location, inlet opening geometry, and fan speed will be varied to observe their impact on airflow distribution and ventilation effectiveness. From these simulations, datasets will be generated using metrics such as air velocity and flow uniformity. Machine learning models will then be trained to learn relationships between design parameters and ventilation outcomes, allowing you to explore improved configurations. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

D0282026 S1CFD + MLAssigned to Yeo Guo Ting

Multi-Objective Optimization of Passive Flow-Control Structures in a 3D Channel Using CFD and Machine Learning

Objective

How can simple structures inside a channel be used to shape and control the flow? In this project, you will explore how passive flow-control elements such as obstacles and baffles influence flow behaviour in a three-dimensional channel. Using computational fluid dynamics (CFD), you will study how different geometric parameters such as size, spacing, orientation, and placement affect flow distribution, mixing, and pressure losses. Machine learning will then be used to identify patterns in the data and help predict performance across different configurations. The aim is to understand how flow can be guided and improved through design, while introducing data-driven methods for optimisation.

Scope

The project begins with a structured introduction to CFD and internal flow behaviour in channels. You will develop a three-dimensional channel model with passive structures and perform simulations to observe how the flow responds to different configurations. Key quantities such as velocity distribution, pressure drop, and mixing characteristics will be extracted and analysed. A dataset will then be constructed to link geometric parameters with flow performance. Machine learning models will be trained to act as fast predictors, allowing you to explore improved designs without running full simulations each time. Guidance will be provided throughout to support both the CFD workflow and the interpretation of results.

D0272026 S1CFD + MLAssigned to Krishnan Raghavendra Aniruddh

Identification of Flow Regimes in Turbulent Separated Flows Using Autoencoders

Objective

What does turbulent flow actually look like when it separates and reattaches? In this project, you will explore complex flow behaviour in diffuser geometries, where separation, recirculation, and reattachment naturally occur. Using computational fluid dynamics (CFD), you will generate detailed flow fields, and then use autoencoder-based machine learning techniques to uncover hidden patterns within the data. The aim is to identify different flow regimes and understand how these structures evolve, providing a new way to interpret complex turbulent flows.

Scope

The project begins with a structured introduction to turbulent separated flows and diffuser configurations. You will simulate flow in an axisymmetric diffuser using CFD and examine key features such as shear layers, recirculation zones, and reattachment regions. Flow data such as velocity and vorticity fields will be extracted and organised into datasets. These datasets will then be used to train an autoencoder, which learns a compact representation of the flow. By analysing this reduced representation, you will identify dominant flow regimes and recurring structures. Guidance will be provided throughout to support both the CFD simulations and the interpretation of the machine learning results.

D0262026 S1CFD + MLAvailable

Machine Learning Prediction of Reattachment Length in Axisymmetric Diffuser Flows

Objective

Where does the flow reattach after it separates, and why does it matter? In this project, you will explore turbulent separated flows in diffuser geometries and focus on predicting reattachment length, which plays a key role in flow performance. Using computational fluid dynamics (CFD), you will study how adverse pressure gradients lead to separation and downstream reattachment. Machine learning will then be used to learn relationships between geometry, flow conditions, and reattachment behaviour, allowing faster prediction of these complex flow features.

Scope

The project begins with a structured introduction to separated flows in axisymmetric diffusers. You will simulate flow behaviour using CFD and analyse key features such as velocity profiles, pressure distribution, wall shear stress, and reattachment length. The simulation results will be compared with available reference data to build confidence in the analysis. A dataset will then be constructed from the CFD results, linking geometric and flow parameters to reattachment behaviour. Machine learning models will be trained to predict reattachment length based on these inputs, providing a faster alternative to full simulations. Guidance will be provided throughout to support both the CFD workflow and the interpretation of results.

B0472026 S1CFDAvailable

Optimization of Inkjet Droplet Formation

Objective

How does a tiny droplet form and break off so precisely in an inkjet printer? In this project, you will explore the physics of droplet formation and breakup using computational fluid dynamics (CFD). You will study how liquid is ejected from a nozzle and how factors such as velocity, nozzle size, and pulse timing influence droplet shape and stability. The aim is to understand how clean, single droplets are formed while avoiding unwanted satellite droplets, and how these behaviours can be improved through design.

Scope

The project begins with a structured introduction to multiphase flow and droplet dynamics. You will simulate the transient ejection of liquid from an inkjet nozzle using a Volume of Fluid (VOF) approach to capture the interface between liquid and air. A baseline model will be developed to reproduce droplet formation, followed by parametric studies where key variables such as nozzle diameter, injection velocity, and pulse duration are varied. The resulting flow fields will be analysed to observe droplet size, breakup behaviour, and the formation of satellite droplets. Based on these results, operating conditions that promote stable droplet formation will be identified. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0462026 S1CFDAvailable

CFD Investigation of Vortex Shedding Suppression Behind a Square Cylinder Using Passive Control Devices

Objective

Why do structures in the wind sometimes vibrate or experience fluctuating forces? In this project, you will explore vortex shedding behind a square cylinder and how it affects aerodynamic forces. Using computational fluid dynamics (CFD), you will study how alternating vortices form in the wake and how this leads to unsteady behaviour. You will then investigate how simple passive control devices can be used to modify the flow and reduce these effects. The aim is to understand how flow-induced forces arise and how they can be controlled through design.

Scope

The project begins with a structured introduction to wake flows and vortex shedding behind bluff bodies. You will simulate flow around a square cylinder using CFD and analyse key features such as velocity fields, recirculation regions, and lift and drag fluctuations. A baseline case will first be established to capture the natural vortex shedding behaviour. Passive control strategies such as splitter plates, corner modifications, or small surface attachments will then be introduced and evaluated. Transient simulations will be performed to observe how these modifications influence the wake structure and vortex shedding patterns. The results will be analysed to identify configurations that reduce flow unsteadiness and aerodynamic force fluctuations. Guidance will be provided throughout to support both the CFD setup and interpretation of results.

B0452026 S1CFDAvailable

Assessment of Turbulence Models for Crossflow Through a Staggered Tube Bundle

Objective

When fluid flows across multiple cylinders, how well do our models actually capture what is happening? In this project, you will explore turbulent crossflow through a staggered tube bundle, a configuration commonly found in heat exchangers. Using computational fluid dynamics (CFD), you will compare different turbulence models and see how they predict velocity distribution, wake interactions, and overall flow behaviour. The aim is to understand how modelling choices influence simulation results and to identify which approaches best capture complex turbulent flows.

Scope

The project begins with a structured introduction to turbulent crossflow and tube bundle configurations. You will simulate flow through a staggered arrangement of cylinders using CFD and analyse key features such as velocity fields, wake development, and turbulence characteristics. Simulations will be performed using different turbulence models, such as k–ε and k–ω SST, to observe how predictions vary. The results will be compared against available benchmark or experimental data to assess model accuracy. Through this process, you will develop insight into how turbulence models behave and how to select appropriate models for engineering applications involving similar flow systems. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0442026 S1CFDAssigned to Liang Renxing

CFD Investigation of Fire-Extinguishing Agent Distribution in an Aircraft Engine Nacelle

Objective

How does a fire-extinguishing system ensure that every part of an engine nacelle is properly covered? In this project, you will explore how suppressant agents are dispersed within an aircraft engine nacelle and how airflow influences their distribution. Using computational fluid dynamics (CFD), you will study how factors such as nozzle placement, spray characteristics, and internal airflow affect the transport and coverage of the extinguishing agent. The aim is to understand how effectively the agent reaches different regions and how the system can be improved for better fire suppression.

Scope

The project begins with a structured introduction to airflow within an engine nacelle and the behaviour of spray-based systems. You will develop a simplified nacelle model and simulate the injection of a fire-extinguishing agent into the airflow. Droplet motion and dispersion will be analysed using appropriate modelling approaches to track how the agent spreads throughout the domain. Key parameters such as nozzle location, injection velocity, droplet size, and airflow conditions will be varied to observe their influence on coverage. Flow visualisation and concentration fields will be used to identify regions where the agent is effective or insufficient. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0432026 S1CFDAvailable

Effect of Piston Bowl Geometry on In-Cylinder Flow and NOx Formation in Diesel Engines

Objective

How does the shape of a piston influence how fuel burns inside an engine? In this project, you will explore how piston bowl geometry affects airflow, fuel-air mixing, and NOx formation in a diesel engine. Using computational fluid dynamics (CFD), you will study how different piston shapes influence swirl motion, temperature distribution, and combustion behaviour. The aim is to understand how design choices at the piston level can impact engine performance and emissions.

Scope

The project begins with a structured introduction to in-cylinder flow and combustion processes in diesel engines. You will develop a simplified combustion chamber model and simulate the flow using CFD with a moving mesh approach. Different piston bowl geometries will be tested to observe how they influence airflow patterns, fuel-air mixing, and temperature fields during the combustion cycle. Key quantities such as swirl structures, temperature distribution, and regions associated with NOx formation will be analysed. Results will be compared across different designs to identify how geometry affects combustion behaviour and emissions. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0422026 S1CFDAvailable

Transient Dispersion of Vehicle Exhaust Near Roadside Pedestrian Seating: A CFD Study

Objective

What do pedestrians actually breathe when sitting near a busy road? In this project, you will explore how vehicle exhaust disperses around roadside seating areas and how different conditions influence pollutant exposure. Using computational fluid dynamics (CFD), you will study how moving vehicles, ambient wind, and seating location affect the transport of exhaust gases. The aim is to understand how short-term exposure occurs at breathing height and how design choices can influence air quality in urban environments.

Scope

The project begins with a structured introduction to pollutant dispersion and airflow in urban settings. You will develop a simplified roadside model that includes a moving vehicle, pedestrian seating, and surrounding airflow. Simulations will be performed to capture the transient behaviour of exhaust as a vehicle passes by. Key factors such as vehicle speed, wind conditions, and distance between the road and seating will be varied to observe their effect on pollutant concentration. Transport of exhaust species will be modelled to evaluate exposure levels at pedestrian height. The results will be analysed to identify conditions that reduce pollutant exposure. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0412026 S1CFDAvailable

Turbulent Wake Dynamics of the Merlion: A Large Eddy Simulation Study

Objective

What kind of airflow forms around a complex landmark like the Merlion? In this project, you will explore turbulent wake behaviour behind a Merlion-inspired geometry and study how its shape influences flow patterns. Using Large Eddy Simulation (LES), you will capture unsteady flow features such as vortex formation and wake development that are not easily seen with simpler models. The aim is to understand how complex shapes interact with airflow and how advanced simulation techniques can reveal these hidden flow structures.

Scope

The project begins with a structured introduction to turbulent wake flows and the basics of Large Eddy Simulation. You will develop a simplified three-dimensional model of a Merlion-like geometry and simulate external airflow around it. LES will be used to resolve unsteady flow features such as vortices, wake structures, and pressure variations. Flow visualisation techniques will be applied to identify and interpret coherent structures in the wake. The results will provide insight into how geometry influences turbulent behaviour and demonstrate the capability of LES in analysing complex flows. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0402026 S1CFDAvailable

Turbulence over Waves: A CFD Investigation of Flow over Sinusoidal Surfaces

Objective

What happens to airflow when it moves over a wavy surface? In this project, you will explore how turbulence behaves over sinusoidal geometries and how surface shape influences the flow. Using computational fluid dynamics (CFD), you will study how the flow accelerates and decelerates as it follows the wave, and how this affects quantities such as velocity, wall shear stress, and turbulence intensity. The aim is to understand how surface curvature modifies the structure of turbulent boundary layers and influences overall flow behaviour.

Scope

The project begins with a structured introduction to turbulent boundary layers and flow over curved surfaces. You will develop a channel flow model with a sinusoidal wall and perform CFD simulations to observe how the flow responds to the wavy geometry. Key quantities such as velocity profiles, wall shear stress, and turbulence characteristics will be extracted at different positions along the wave. Parametric variations in wave amplitude and wavelength may be explored to study their influence on the flow. The results will be analysed to understand how turbulence adapts to surface curvature and how flow structures evolve along the wave. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0392026 S1CFDAvailable

When Turbulence Turns: An ω-Based Model for Curvature-Driven Secondary Flows

Objective

What happens when turbulent flow is forced to turn? In this project, you will explore how curvature affects turbulence in duct flows and why standard models sometimes struggle to capture these effects. Using computational fluid dynamics (CFD), you will study flow through a curved duct and observe how secondary flows develop due to curvature and rotation. You will then investigate how ω-based turbulence models can be adapted to better account for these effects. The aim is to understand the limitations of existing models and explore how they can be improved for more complex flow situations.

Scope

The project begins with a structured introduction to curved duct flows and turbulence modelling. You will develop a computational domain consisting of a straight inlet, a 180° bend, and an outlet section, and perform CFD simulations to analyse flow behaviour. Key quantities such as velocity profiles, pressure loss, and secondary flow intensity will be extracted and compared with reference data. A baseline ω-based turbulence model will first be used, followed by the introduction of curvature-aware modifications. The modified model will be implemented and tested to evaluate its ability to better capture the flow features. The results will be analysed to understand how curvature influences turbulence and how modelling approaches can be improved. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0382026 S1CFDAvailable

Low-Dimensional Modelling of Secondary Flows in a 180° Duct Bend using Proper Orthogonal Decomposition

Objective

Can a complex flow be broken down into a few simple patterns? In this project, you will explore secondary flows in a curved duct and study how they form due to curvature effects. Using computational fluid dynamics (CFD), you will generate flow data for a 180° duct bend and then apply Proper Orthogonal Decomposition (POD) to identify the dominant flow structures. The aim is to understand how complex flow behaviour can be represented using a reduced number of modes, providing insight into the key features that drive the flow.

Scope

The project begins with a structured introduction to curved duct flows and the concept of secondary motion. You will simulate turbulent flow through a duct with a 180° bend using CFD and extract velocity data at multiple locations. This data will then be analysed using Proper Orthogonal Decomposition to identify dominant coherent structures and quantify their contribution to the overall flow. The reduced set of modes will be used to reconstruct the flow and assess how well simplified representations capture the key behaviour. The results will be interpreted in relation to secondary vortex structures within the duct. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.

B0372026 S1CFDAvailable

The Birth of a Horseshoe Vortex: A CFD Study of Boundary Layer–Obstacle Interactions

Objective

What happens when a smooth flow suddenly meets an obstacle? In this project, you will explore how a horseshoe vortex forms when a boundary layer interacts with a surface-mounted structure. Using computational fluid dynamics (CFD), you will study how the flow wraps around the obstacle and generates complex three-dimensional vortex structures near the wall. The aim is to understand how these vortices develop and how well commonly used turbulence models capture their behaviour.

Scope

The project begins with a structured introduction to boundary layers and flow interaction with surface-mounted obstacles. You will simulate a developing boundary layer encountering an obstacle using CFD and analyse how the flow separates, accelerates, and forms vortex structures. Key features such as streamline patterns, vorticity distribution, and wall shear stress will be examined to identify the formation and evolution of the horseshoe vortex system. Different turbulence models may be compared to assess their ability to capture these flow features. The results will be interpreted in relation to known experimental observations. Guidance will be provided throughout to support both the CFD setup and the interpretation of results.