A methodology needs to be developed to conduct safety analysis of optimized core design with mixed cores.
Non-uniform heat flux flow boiling
3D printing for heat transfer
Algae4Africa: Treatment of Tanzanian livestock water through algae to produce clean water and algae byproducts
Freshwater production and thermal control of greenhouses
EV battery cooling with 2-phase thermosiphons
Malaria Prevention: Mosquitos and airflow
The evolution of computational fluid dynamics (CFD) has witnessed significant advancements in numerical simulation techniques. The Lattice Boltzmann Method (LBM) has emerged as a promising alternative to traditional CFD solvers, leveraging its inherent parallelizability and efficiency in high-performance computing environments, particularly with GPU acceleration. Extensive literature documents the successful application of LBM in simulating fluid flow and multiphase dynamics. However, challenges remain in extending its application to complex engineering systems characterized by high Reynolds numbers, multiphase interactions, and intricate boundary conditions, highlighting the need for innovative approaches. This project entails incorporating a true multiphase model or a two-phase VOF model in the Lattice Boltzmann Method that can handle high-Reynolds number flow with high-density ratios between the fluids and the capturing reverse flow at the outlet boundaries. The method will be assessed through a case study on air-core formation in a hydrocyclone, using existing experimental data for validation.
Student funding: The student will have to self fund their studies or secure funding themselves, such as through the UP funding page, DSI-NRF Masters/PhD program, DSI-CSIR program or similar.
Deep learning methods have increasingly garnered attention within the CFD community. Techniques such as Physics-Informed Neural Networks (PINNs), Graph Neural Networks (GNNs), and Generative Adversarial Networks (GANs) are being explored for their potential to accelerate simulations and enhance predictive accuracy. Numerous studies have demonstrated that these methods can substantially reduce computational cost and improve the resolution of simulation outputs. Nevertheless, the literature also underscores inherent limitations, such as the dependence on extensive training datasets generated from conventional numerical methods and ongoing debates regarding whether these techniques provide true predictive capabilities or merely serve as computational accelerators. This study will develop a novel hybrid modeling approach integrating the Lattice Boltzmann Method (LBM) with deep learning techniques such as Physics-Informed Neural Networks (PINNs), Graph Neural Networks (GNNs) and Generative Adversarial Networks (GANs) to enhance the accuracy and computational efficiency of multiphase flow simulations. The effectiveness of these techniques will be tested against traditional numerical solvers and existing experimental data from literature.
Student funding: The student will have to self fund their studies or secure funding themselves, such as through the UP funding page, DSI-NRF Masters/PhD program, DSI-CSIR program or similar.
This research will apply and refine LBM for simulating multiphase flows in gas-solid cyclone separators, commonly used in mineral processing. The study will assess the method's strengths and limitations in handling turbulence, phase interactions, and particle separation efficiency. Simulations will be validated using existing experimental data from literature. with the goal of improving predictive accuracy and computational feasibility.
Student funding: The student will have to self fund their studies or secure funding themselves, such as through the UP funding page, DSI-NRF Masters/PhD program, DSI-CSIR program or similar.
A CFD simulation of a hydrocyclone seperator
This study will explore the application of machine learning techniques to enhance the accuracy and efficiency of multiphase flow modeling in computational fluid dynamics (CFD). The research will focus on integrating deep learning methods, such as Physics-Informed Neural Networks (PINNs) and Graph Neural Networks (GNNs), with traditional numerical approaches to improve phase interaction predictions. The methodology will be applied to industrial processes, including cyclone separators and fluidized beds, to evaluate its performance against conventional models.
Student funding: The student will have to self fund their studies or secure funding themselves, such as through the UP funding page, DSI-NRF Masters/PhD program, DSI-CSIR program or similar.
This project aims to investigate the role of deep learning-based surrogate models in accelerating CFD simulations, particularly for high-Reynolds number flows. The study will analyze the trade-offs between accuracy and computational cost, focusing on the applicability of the above approaches for rapid solution approximation. The effectiveness of these techniques will be tested against traditional numerical solvers and experimental datasets.
Student funding: The student will have to self fund their studies or secure funding themselves, such as through the UP funding page, DSI-NRF Masters/PhD program, DSI-CSIR program or similar.
Some of the advanced geometries possible with metal 3D printing
This project entails
Malaria prevention is an ongoing challenge in the tropical eastern portions of South Africa, particularly for impoverished rural communities, with the South African government employing a collection of strategies to tackle this problem, such as chemical sprays, mosquito repellents and mosquito nets.
Partnering with the University of Pretoria’s Institute for Sustainable Malaria Control, this project will investigate another possible tool that could be used – airflow. Airflow is often used to prevent mosquito bites (typically through fans) by dispersing the carbon dioxide and other volatiles that mosquitos use to locate humans, as well as directly interfere with their flight, but much uncertainty still remains over this approach.
This project aims to better understand the interaction between mosquito behaviour and the fluid dyanmics of air flow.
Mosquito biting behaviour can be disrupted by certain airflows
(© Department of Foreign Affairs and Trade, CC BY 2.0)
Flow boiling is an important heat transfer mechanism. In thermal solar energy systems, such as direct steam generation plants or solar driven desalination plants, the working fluid is heated in collector tubes exposed to focused solar irradiation. Several types of collector tube and solar reflections systems exist, but they all result in circumferentially non-uniform heat flux conditions on the outer surface of the collector tube. Because most flow boiling literature is for fully uniform heat flux conditions, relatively little is known about what impact the heat flux distribution has on the internal heat transfer performance (heat transfer coefficient). In this investigation the influence of the heat flux distribution is to be investigated experimentally. For this purpose one or more horizontal test sections are to be constructed with specially designed heating elements with which different solar heat flux distribution conditions can be mimicked in a laboratory environment. Test are to be conducted at different mass flow rates, heat flux distributions and heat flux levels. Wall temperature heat flux measurements are to be made and processed into heat transfer coefficients. Relevant correlations are to be developed to describe the impact of the investigated parameters.
Different flow regimes of two phase flow boiling in tubes, typical of direct steam generation systems
The use of PCMs is a viable method of storing thermal energy collected from solar sources to be utilized at night. Liquid-solid PCM’s support high energy concentrations and do not suffer as much from a high volumetric contraction and expansion as is the case with vapour-liquid PCM’s. The phase change temperature is important and should match the requirements of the application. For solar power thermal storage this limits the list of suitable materials. These include for instance inorganic salts and metal alloys. Inorganic molten salts are already used in some solar power plant types as the heat transfer fluid (only in its liquid phase), but has not yet been fully considered as a phase change material in, for instance, possibly simpler type direct steam generation plants, where water is used as the heat transfer fluid directly. A draw-back of inorganic salts are that they have relatively low thermal conductivities which result in a significant thermal barrier during the charging (solidifying) and discharging (melting) modes of thermal storage modules.
In this numerical optimization topic, a commercial numerical software package is to be used to model a thermal storage module where heat transfer rates between (to and from) the heat transfer fluid and (a) selected phase change material(s) is to be maximized during the charging as well as discharging modes. The model is to be validated against experimental data obtained from literature before optimization can commence. Optimization design variables include the thickness of the phase change material plate layers, the length of the plate layers and the number of phase change plate layers
The field of study combines the 3-dimensional reactor physics analysis of a large number of random groupings of fuel elements with a wide variety of operational histories, that are placed in the fixed geometry of the spent fuel pool at the Koeberg Nuclear Power Station [4]. The study will identify the probability of an accidental super-critical geometry being created and the resultant heat production and removal by natural convection heat transfer mechanisms in the surrounding water of the spent fuel pool.
In general the project requires existing knowledge of reactor physics coupled to heat transfer phenomena [2]. The novelty of the project is to determine the extent of the random groupings of the packings coupled to the burn-up history of the fuel elements and to determine the risk, in terms of overheating and fission product release that such an accidental criticality event will pose.
The proposed cooperative research project will investigate the risk of super criticality and boiling in the SFP. This proposed framework will utilize risk informed approaches to identify parameters necessary to ensure that risks of super criticality and boiling in the SFP are minimized. According to the definition risk is a probability multiplied by consequences. The proposed assessment will utilize probabilistic risk assessment (PRA) methods combined with deterministic studies in the areas of thermal hydraulics, and reactivity (criticality) to evaluate consequences [1]. This framework will form a technical foundation to be used to devise mitigation strategies and provide input to developing regulatory changes by NNR.
The tools to be used in this project consist of MCNP6 [4], SCALE-6.2 [5], COBRA-SFS [2] and MCNP6/CTF [3-6]. MCNP will be utilized to carry out analysis of criticality safety while SCALE-6.2 will be used to confirm independently the MCNP criticality calculations, perform depletion calculations when needed, and conduct uncertainty analysis and propagation. COBRA-SFS, a thermal-hydraulic code developed for steady-state and transient analysis of multi-assembly spent-fuel storage will be used to model important physical behavior governing the thermal performance of SFPs, with internal and external natural convection flow patterns, and heat transfer by convection, conduction, and thermal radiation. Of particular significance is the capability for detailed thermal radiation modeling within the fuel rod array. The multi-physics code MCNP6/CTF, developed at NCSU, will help investigate criticality (reactivity) and boiling in SFPS taking into account complete modeling of all feedback effects involved. The proposed project will develop models for the Koeberg nuclear power plant spent fuel pool for the computation tools involved in the project: MCNP6, SCALE-6.2, COBRA-SFS and MCNP6/CTF.
The proposed work will require use of high performance computing facilities. The Virtual Computing Laboratory at NCSU (https://vcl.ncsu.edu/) will be utilized. The Office of Information Technology (OIT) High Performance Computing (HPC) services provide NCSU students, faculty high performance computing resources, and consulting support for research and instruction. Campus Linux Cluster, henry2 has 1192 dual socket servers with Intel Xeon Processors (mix of single-, dual-, quad-, six-, and eight-core), 2-4GB per core distributed memory, dual gigabit or 10Gb Ethernet interconnects. Also integrated into henry2 are a number of nodes with 16 cores and up to 128GB of memory. These nodes are intended to support shared memory (OpenMP) jobs or other jobs with large memory requirements. The HPC services are available allowing for running jobs up to 128 processor cores up to 48 hours. The number of nodes can also be expanded on demand to accommodate higher computational requirements. In addition, the Reactor Dynamics and Fuel Modeling Group (RDFMG) at NCSU, led by Dr. Avramova, has the fowling computational resources
The proposed cooperative research project will investigate the risk of super criticality and boiling in the SFP. This proposed framework will utilize risk informed approaches to identify parameters necessary to ensure that risks of super criticality and boiling in the SFP are minimized. According to the definition risk is a probability multiplied by consequences. The proposed assessment will utilize probabilistic risk assessment (PRA) methods combined with deterministic studies in the areas of thermal hydraulics, and reactivity (criticality) to evaluate consequences [1]. This framework will form a technical foundation to be used to devise mitigation strategies and provide input to developing regulatory changes by NNR. METHODOLOGY The tools to be used in this project consist of MCNP6 [4], SCALE-6.2 [5], COBRA-SFS [2] and MCNP6/CTF [3-6]. MCNP will be utilized to carry out analysis of criticality safety while SCALE-6.2 will be used to confirm independently the MCNP criticality calculations, perform depletion calculations when needed, and conduct uncertainty analysis and propagation. COBRA-SFS, a thermal-hydraulic code developed for steady-state and transient analysis of multi-assembly spent-fuel storage will be used to model important physical behavior governing the thermal performance of SFPs, with internal and external natural convection flow patterns, and heat transfer by convection, conduction, and thermal radiation. Of particular significance is the capability for detailed thermal radiation modeling within the fuel rod array. The multi-physics code MCNP6/CTF, developed at NCSU, will help investigate criticality (reactivity) and boiling in SFPS taking into account complete modeling of all feedback effects involved. The proposed project will develop models for the Koeberg nuclear power plant spent fuel pool for the computation tools involved in the project: MCNP6, SCALE-6.2, COBRA-SFS and MCNP6/CTF. The proposed work will require use of high performance computing facilities. The Virtual Computing Laboratory at NCSU (https://vcl.ncsu.edu/) will be utilized. The Office of Information Technology (OIT) High Performance Computing (HPC) services provide NCSU students, faculty high performance computing resources, and consulting support for research and instruction. Campus Linux Cluster, henry2 has 1192 dual socket servers with Intel Xeon Processors (mix of single-, dual-, quad-, six-, and eight-core), 2-4GB per core distributed memory, dual gigabit or 10Gb Ethernet interconnects. Also integrated into henry2 are a number of nodes with 16 cores and up to 128GB of memory. These nodes are intended to support shared memory (OpenMP) jobs or other jobs with large memory requirements. The HPC services are available allowing for running jobs up to 128 processor cores up to 48 hours. The number of nodes can also be expanded on demand to accommodate higher computational requirements. In addition, the Reactor Dynamics and Fuel Modeling Group (RDFMG) at NCSU, led by Dr. Avramova, has the fowling computational resources
The diurnal swing in ambient temperature is a known fact. Taking advantage of this, phase-change materials (PCMs) can utilise the lower ambient temperature during night time to solidify, storing 'cold' to absorb heat during daytime. This can either be used to assist the HVAC cooling system during the day, or to act as a temperature source in a low temperature difference Stirling engine to harvest power. The guaranteed complete solidification of a PCM, the development and construction of a low-temperature self-starting Stirling engine, and heat exchangers to provide a constant temperature to the Stirling engine, are some of the topics that require investigation.
Copyright © University of Pretoria 2025. All rights reserved.
Virtual Campus
Get Social With Us
Download the UP Mobile App