Farhana Taiyebah - Dissertation Defense
"Toward Data-Driven Subgrid-Scale Modeling of the Zel'dovich Deflagration-To-Detonation Mechanism in Dense Stellar Plasmas II. Multi-Dimensional Subgrid-Scale Model"
Abstract:
The deflagration-to-detonation transition (DDT) remains a central challenge in modeling Type Ia supernovae, as the local processes responsible for detonation initiation can remain unresolved in large-scale calculations. This dissertation develops a multidimensional, data-driven subgrid-scale model of spontaneous detonation initiation through the Zel'dovich reactivity-gradient mechanism. Building on an earlier one-dimensional framework, the model incorporates the lateral extent and curvature of reactive hotspots to examine the influence of geometry on their evolution. A database of two-dimensional direct numerical simulations, informed by three-dimensional turbulence data, is constructed to investigate the conditions under which these perturbations develop into detonations. The database supports both analysis of the combustion outcomes and the training of neural-network classifiers for prediction at coarser resolution. The trained model is integrated into FLASH/Proteus and evaluated on the database, including coarsened and rotated profiles and a synthetic turbulent background. This work provides a framework for connecting direct numerical simulations (DNS) with stellar-scale astrophysical simulations, strengthening the basis for modeling detonation initiation in thermonuclear supernovae and for future studies of multiscale reactive flows.
