CEAS undergraduate excels during internship
Thomas Savage, an undergraduate student in the Department of Chemical Engineering and Analytical Science (CEAS), has received high praise for his work in a collaborative research project between The University of Manchester and Imperial College London.
As part of this role, Tom was charged with creating a data-driven modelling framework to identify the optimal operating conditions of a novel liquefied natural gas (LNG) refrigeration cycle. The main challenge was optimising the process using the conventional rigorous model (HYSYS Flowsheet) - a very time-consuming task.
I developed a surrogate modelling strategy by integrating a range of unsupervised (dimensionality reduction) and supervised (artificial neural network, Gaussian process) machine learning technologies to resolve this challenge.
Through the use of a bespoke evolutionary algorithm, I was able to gain a similar optimal solution compared to the rigorous model, meanwhile reducing the computational time to ten minutes, enabling possible use for operational optimisation as well as reduced overall project times.
After the completion of his project, Tom was invited to present at the 36th Centre for Process Integration Annual Research Consortium, after which he received significant attention, including praise from leading companies such as Shell and BP.
The Principal Technical Expert from Shell was very interested in Tom's work and talked to him over dinner. He later sent Tom an email showing the advanced technique recently developed in Shell, and shared his experience regarding process modelling and control.
Professor Robin Smith, Fellow of the Royal Academy of Engineering, and Professor Megan Jobson of the Centre for Process Integration said they were "extremely impressed both with the quality and maturity of his work. This assessment was shared by the industrial members of the Research Consortium who participated in the meeting. Tom not only produced an excellent piece of work, but also presented it with great confidence and clarity".