Precision Learning in Thermoacoustics
Full funding, fees and maintenance, is available only to UK students.
Thermoacoustic oscillations are a perpetual problem in aircraft and rocket engines. Unacceptably large oscillations often appear during full-scale engine tests, despite being absent during part-scale tests, leading to costly re-designs. The aim of this project is to combine adjoint-based sensitivity analysis with automated experiments and machine learning so that a computer can build up an accurate model of a system and use it to design out thermoacoustic oscillations, despite not knowing all the model parameters a priori.
This PhD project will combine experiments with theory, and be predominantly experimental. The proposed student will take over a bench-scale thermoacoustic test rig developed by an existing post doc and PhD student. The proposed student will set up the rig to perform automated experiments in which the rig's behaviour is monitored as several of the most influential parameters (as determined by the adjoint-based model) are altered. The proposed student and the existing student will then automate a process by which these results update the model, which will then be used in an optimization routine operating on-the-fly. This method will be applied to a system that we can characterize well in advance, and then to systems that cannot be characterized well in advance.
The expected outcome is a robust method for the passive control of thermoacoustic oscillations, combining adjoint-based optimization and machine learning from experiments. When this method has been proven, it will be applied to the mid-scale annular rig at Cambridge and will then be ready to be an industrially-focused project.
Applicants should have an excellent academic track record in a scientific, mathematical, or engineering discipline, with at least a high 2i degree, or equivalent.
To apply, please complete form CHRIS/6 (cover sheet for C.V.s) available at http://www.admin.cam.ac.uk/offices/hr/forms/ and send it with your C.V. and a covering letter to Matthew Juniper (email@example.com) to arrive no later than 28 Feb 2017.