Experiments and their numerical twins in Rayleigh-Taylor instability

Abstract

Rayleigh–Taylor instability occurs in a wide range of natural and engineering contexts, from its role in ocean and atmospheric dynamics at the heart of climate models, to fuel mixing in inertial confinement fusion; a potential energy source of the future. It has therefore been a topic of particular interest since the 1950s, initially focussed on experimental and analytical work, but in the last forty years increasingly dominated by numerical simulation. However, this has be accompanied by a worrying trend of divergence of simulation from experiment, a trend that this thesis seeks to address.

A set of hybrid experimental-numerical studies is performed, using experimental data to inform construction and initialisation of the numerical method. The optimisation of the simulations to give good agreement with experiments emphasises the importance of initial conditions, which determine the structure of subsequent evolution of the instability. The need to diagnose and control experimental initial conditions motivates the development of the CAMPI rig, a novel variable acceleration apparatus that facilitates Rayleigh–Taylor growth under complex acceleration profiles. The rig design promotes experimental method that best suits parallel numerical study, and will form the basis for such work over the next decade.

Conventional numerical simulations, even those optimised to better match experiments, predict behaviour that inevitably diverges from that of experiments. An alternative approach is presented, in which a flexible numerical method is iteratively corrected to match experimental scalar data. This digital twin of the experiment provides all of the insights of a conventional simulation but with an experimental ground truth.

Date of Award	21 Jun 2022
Original language	English
Awarding Institution	University of Bristol
Supervisor	Andrew G W Lawrie (Supervisor), Robin Williams (Supervisor) & Harry Coules (Supervisor)