- Opens: Wednesday 9 June 2021
- Deadline: Thursday 30 September 2021
- Number of places: 1
- Duration: 36 months
- Funding: Home fee, Stipend
OverviewTake advantage of recent machine learning frameworks to accelerate the solution of fluid flow problems. Therefore, shortening the design-test-build-refine methodology cycle and allowing for the generation of tailored manufacturing equipment designs.
First or upper second, BEng, BSc degree or equivalent background in Computer Science, Mathematics, Chemical Engineering, or similar disciplines.
In recent years, the demands on pharmaceutical and biopharmaceutical process development have increased due to development schedules and molecule complexity. To address this, the industry is looking to new technologies and ways of development. As part of a wider project, we propose to build an adaptable platform technology based around Additive Manufacturing and Machine Learning, incorporating design optimisation, modelling, sensing and materials development that will set the basis for an agile, evolving, manufacturing platform. Using, at first, simplified physical models of the flow and process conditions, we will create a generative design platform that will seek the optimal geometry / flow paths for specific tasks. The flow-problem is relatively simple and thus computationally tractable but has complexities that make it challenging to fully optimise and lend itself to optimisation / machine learning approaches. Furthermore, common approaches for computational solution of the flow problem can be time consuming and therefore can become the rate limiting steps in the design-test-build-refine methodology cycle. Whilst brute forcing speeding up the solution through additional hardware is possible, recent advances in deep learning allow for speed up through more computationally elegant and scalable solutions.
Take advantage of recent machine learning frameworks to accelerate the solution of flow problems. Therefore, shortening the design-test-build-refine methodology cycle and allowing for the generation of more and more complex designs. As a result, enabling the platform technology development to extend to greater applications.
In this 3-year PhD programme at the University of Strathclyde, we will develop rapid solutions to flow problems using a combination of traditional computational fluid dynamic simulations and machine learning approaches. This programme will contribute to the wider EPSRC funded partnership with GlaxoSmithKline and the Universities of Strathclyde and Nottingham on the Accelerated Discovery & Development of New Medicines. Our vision is to enable the production of novel, transformative medicines at lower costs, with reduced waste production and shorter manufacturing times. The PhD student is expected to engage with the wider community by attending regular partnership meetings.
Applicants must have a first or upper second, BEng, BSc degree or equivalent background in Computer Science, Mathematics, Chemical Engineering or similar disciplines along with knowledge of computer programming (for example, python, MATLAB) and simulation. Experience or knowledge of machine learning and deep learning approaches would be advantageous.