Physics Informed Machine Learning for Gas Solids Multi-Phase Flow Characterisation

Project Reference Number

Please refer to project number SCEBE-21SF-PIML -McGlinchey.

Background

Gas Solids multi-phase flow is common in nature, essential in many process and life science industries for the transport of grains and powders, and produced sand is a major issue in the oil and gas industry. Understanding and characterising the flow behaviour of gas solids flow is a significant challenge for the science and engineering community, however, developments in machine learning have shown promising results for gas liquid flow which may be transferable. 

This research project intends to use machine learning techniques, e.g. physics informed neural networks, to characterise the flow behaviour of gas solids flow in pipelines. Glasgow Caledonian University has a successful track record in the study of gas solids flow and related measurement techniques, and has two industrial scale test rigs to provide data on flow behaviour. These data and underlying system physics would be used in an investigation applying machine learning techniques.

Aims

- Conduct experiments and collect data on gas solids flow behaviour, potentially developing novel instrumentation and measurement techniques on GCU’s test facility.

- Investigate machine learning techniques, such as, PIGPs or PINNs, in the characterisation of gas solids flow behaviour.

- Integrate the measurement techniques and data analysis to form a gas solids flow characterisation instrumentation system.

Glasgow Caledonian University’s research is framed around the United Nations Sustainable Development Goals. This project addresses the goal of Industry, Innovation and Infrastructure and is part of the research activity of the Engineering Simulation and Advanced Manufacturing Research Group https://www.gcu.ac.uk/cebe/research/researchgroups/engineeringsimulationandadvancedmanufacturing/

Candidates for this PhD position will normally have a Masters or an honours degree in a science or engineering subject and be able to demonstrate an interest in data analysis and machine learning. Candidates are requested to submit an outline research proposal (maximum of 1000 words) as part of their application. 

How to Apply

This project is available as a 3 years full-time or 6 years part-time PhD study programme. 

Candidates are encouraged to contact the research supervisors for the project before applying. 

Please note that emails to the supervisory team or enquires submitted via this project advert do not constitute formal applications; applicants should apply using our Application Process page, choosing Mechanical Engineering and their preferred intake date.  

Please send any other enquires regarding your application to: [Email Address Removed]