Inference of Forcing Kernels in Generalized, Multi-Physics, Dynamic Systems
Abstract
The multi-physics dynamics of high-speed, shocked, droplet-laden flows are intricate. Its accurate theoretical and experimental prediction in technologies that are relevant to the Air Force such as rotating detonation engines, and the erosion of missile cones by condensed droplets has proven tremendously challenging. Central to modeling and observation challenges are (1) the extreme dynamical and thermodynamical scales, and (2) a marginal general understanding of the non-linear, stochastic and intermolecular, inter-phase forcing. Fast processes limit the observable data to snapshots of shadowgraphs of the droplet shape, while a reduced fidelity of process-scale models for droplet deformation and shock dynamics prevents a reliable engineering analysis. We propose a disruptive new method to infer forcing kernels of systems of computational, multi-physics, and particle clouds from limited, high-fidelity data with confidence intervals. We start from the paradigmchanging assumption that the forcing is unknown, random, and time-dependent on phase lag only. Building upon a stochastic, particle-cloud modeling framework that was developed under AFOSR funding, we infer kernels through inverse methods that optimize the confidence intervals of the evolution of finite time flow maps of the droplet and shock dynamics towards observed data. In the process, the algorithm presents an unprecedented opportunity to uncover the physics of shock and droplet interaction.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 06, 2024
- Source ID
- FA95502310405
Entities
People
- Gustaaf B Jacobs
Organizations
- Air Force Office of Scientific Research
- Salk Institute for Biological Studies
- United States Air Force