Data-Driven Homogenization of Solid Propellants
Abstract
Performing simulations of a full solid propellant rocket motor require, among other things, the availability of thermomechanical constitutive relations for the propellant at the continuum scale. This is particularly challenging as solid propellants are generally complex conglomerates of energetic particles embedded in a soft matrix, resulting in highly nonlinear constitutive properties that depend on the composition of the particles, their size and shape distribution, their volume fraction, and the properties of the binder. Furthermore, in scenarios where the propellant is under heavy loads (such as high chamber pressures during firing), this can cause microstructural debonding, causing cracks that increase the burning surface area which may lead to dangerous instabilities. It is desired to have some capability to predict the material response of a propellant to external loads, to prevent motor failure during testing or once deployed in a real-world environment. Towards this goal, modern methodologies rely on a combination of homogenization schemes and micromechanical models thus minimizing empiricism and leading to a more fundamental understanding of the relationship between process, microstructure, and performance in the material development cycle. A common computational approach to this problem is via concurrent multiscale modeling. These schemes generally can handle arbitrary microstructures in the nonlinear and history-dependent regimes as seen in solid rocket propellants, but usually at a prohibitive computational cost. This research will build on the recent development of Smart Constitutive Laws, which combine advanced modeling with cutting edge machine learning techniques to obtain surrogate models of the microstructure, by adding features that allow them to work on the large deformation regime typical of solid rocket propellants. A cornerstone of this approach will be the development of new neurons that can automatically satisfy frame indifference and material symmetries of the systems under consideration. We expect that these novel architectures will enable the simulation of these complex materials at a fraction of the computational cost of current approaches with minimal sacrifice in accuracy or stability of simulations.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 05, 2025
- Source ID
- FA95502410014
Entities
People
- Julian J Rimoli
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of California, Irvine