OPTIMIZING CARBON-CARBON COMPOSITES MANUFACTURING BY IDENTIFYING AND REDUCING KEY UNCERTAINTIES FOR HYPERSONIC APPLICATIONS
Abstract
The overarching goal of this project is to develop a multi-fidelity physics-informed Bayesian deep learning framework to identify and reduce the uncertainties involved in the manufacturing process of carbon-carbon (C-C) composites with respect to their performance properties. A system of coupled, non-linear partial differential equations (PDEs) governing the mass, momentum and energy transfer involved in the manufacturing process and materials responses to thermal and mechanical loads will be assembled with a set of constitutive relations. Constitutive relation parameters (e.g., reaction rate constants, phase-dependent mechanical and thermal properties) will be modeled as uncertain parameters in the Bayesian framework. The initial framework will be trained and validated against existing data (e.g., process history datasets and corresponding material performance) from Honeywell, an industry leader for C-C composites manufacturing. Based on this trained model, uncertainties will be quantified using both forward and inverse uncertainty quantification (UQ), and parameters and-or constitutive relations with the largest uncertainties will be identified. This will direct active data collection in new manufacturing processes with sets of process parameters designed to reduce the identified uncertainties in the model.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Mar 07, 2023
- Source ID
- FA95502210065
Entities
People
- Tengfei Luo
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Notre Dame