MIT CHEFSI’s focus on scalable, uncertainty-quantified prediction of coupled multiphysics phenomena under extreme conditions directly aligns with NNSA’s mission to advance the frontiers of predictive science, high-performance computing, and national security applications involving complex materials and environments.
The overarching goal of CHEFSI is to simulate a model — but complete — heat shield system during atmospheric reentry along a prescribed representative trajectory. Failures observed in the Orion capsule’s heat shield during the Artemis I Lunar return, including the fracture and loss of large TPS fragments, have exposed critical gaps in the state-of-the-art of predictive simulation science and capability in this area important to national security.
We will adopt a representative heat shield geometry, and consider SiC and graphite as two separate TPS model materials, chosen for their relevance and availability of thermochemical and mechanical properties. The final full- system predictions (FSP) will contemplate a full trajectory starting at Mach number Ma=25, and initial altitude of 50km.These conditions are relevant to a significant range of orbital and deep space missions.
The complexity of the multiphysics models and solvers required to capture the full aero-thermo-chemo-mechanical response—including material failure—along with the broad range of spatial and temporal scales, the need for physics- informed surrogate models, and the demands of rigorous uncertainty quantification, place the computational require- ments of our overarching application squarely in the exascale regime.
CHEFSI will deliver a full-system, limited-physics predictive simulation by the end of Year 1. This will be accomplished by an initially weak coupling of our existing simulation capabilities in hypersonic flow (Exasim) and thermo-chemo-mechanical material response (ΣMIT). In subsequent years, legacy phenomenological models for rarefied flow, pyrolysis, oxidation, nitridation, and material failure will be progressively replaced by higher-fidelity, first- principles-informed models. These enhancements will incorporate dimensionality reduction and physics-constrained ML surrogates, with prioritization guided by our integrated UQ strategy combining experiments, forward UQ, statistical inference, and optimal experimental design. The computational demands resulting from increasing model fidelity will be addressed through implementations informed by accompanying advances in compiler infrastructure, domain- specific languages, and performance engineering. By Year 5, we will demonstrate a fully predictive, exascale-capable simulation encompassing all key physics.
Another major objective is to provide an educational platform focused on the development of a talented workforce of computational scientists at the nation’s service. As an integral part of its research and educational activities, CHEFSI will pursue extensive collaborations with colleagues at DOE/NNSA laboratories. These will include direct engagement of collaborators on MIT graduate courses, research sub-teams, joint publications, structured and unstructured visits and student/postdoc/researcher internships at NNSA labs, data and code sharing and transitioning.
Scientific Investigations
CHEFSI research will be organized into several subject areas.
Predictive Science
Advancing multiscale modeling to enable predictive simulation
Full system predictions (FSPs) will be conducted using our coupled full-physics aero-thermo-chemo-mechanics simulation capability, which will comprise: our hypersonic aero-thermo-chemistry flow solver Exasim; the rarefied, non-equilibrium direct simulation Monte Carlo (DSMC) solver SPARTA developed by Sandia National Laboratories; and our thermo-chemo-mechanics solver ΣMIT. The individual components and coupled facility will be subject to a comprehensive Verification & Validation/Uncertainty Quantification plan.
Verification & Validation/Uncertainty Quantification
Quantifying simulation accuracy and uncertainty with confidence
CHEFSI’s mathematical models of gas-phase chemical kinetics, thermal and chemical gas-surface interactions, and non- equilibrium response of solid materials encompass a vast array of uncertainties. Even when some model parameters or constitutive relations are developed from ab initio calculations, extrapolation away from idealized conditions, stochasticity in materials microstructure, and our broader program of bridging scales require a quantitative assessment of these uncertainties and their impact. Our rigorous and comprehensive UQ program aims to quantify, prioritize, and reduce these uncertainties in a goal-oriented fashion—i.e., with regard to their impact on the final predictive QoIs. Our UQ program will include aspects of forward UQ (e.g., sensitivity analyses, dimension reduction, accelerated with surrogate models and multi-fidelity methods); Bayesian inference and model calibration (including adaptive choices of prior and stochastic representations of model error); in-house and external experiments for model validation, augmenting available benchmark data; and methods of optimal experimental design (aimed at creating a closed-loop interaction between simulations and experiments). In each of these areas, we will not only harness leading UQ methodologies, but advance the methodological state of the art. In this vein, we will focus a portion of our UQ effort on quantifying uncertainty in our ML surrogate model predictions (ENNs, MLS) that we will employ throughout our multiscale simulations.
Exascale Computing
Harnessing exascale platforms for multiphysics simulation
CHEFSI will develop an integrated suite of software systems and programming technologies that enable our simulation software to scale efficiently on exascale computing systems. We will aim to improve the portability and performance of existing codes with a mix of existing tools and novel DSL/compiler technologies.
Artificial Intelligence and Machine Learning Integration
Accelerating discovery through AI-enhanced modeling
CHEFSI will extend and apply symmetry-equivariant neural networks (ENNs) to develop accurate, efficient, and physically grounded surrogate models across atomistic and continuum scales.
Software
Developing modular, scalable tools for scientific computing
CHEFSI will design the software framework to combine various components, including Exasim, ΣMIT, and new software components, in a flexible and modular fashion. The coupling between Exasim and ΣMIT will be the core simulation. To develop MD potentials and MD data (particularly for the QK gas-surface chemistry model and the constitutive model for continuum simulation), we will utilize LAMMPS for MD, E3NN for ENNs, and Quantum-ESPRESSO for DFT. We will also use DSMC via SPARTA to validate Exasim’s hypersonic flow. Outside of these other codes though, our main code will first utilize software such as PETSc, Trillinos, preCICE, and bespoke Julia packages to build a partitioned multi-physics simulation from Exasim and ΣMIT while utilizing other CS tools within one of the packages, the coupling, or UQ. This design strategy aims to take advantage of the capabilities and performance of Exasim and ΣMIT while using new software in the context of these existing codes and their coupling.
Integration
Unifying physics across scales through deep system integration
One of CHEFSI’s major integration goals is the coupling of Exasim and ΣMIT, which will serve to catalyze our reasoning about shared software development. Within this task, we can explore integration of CS/Exascale, UQ, predictive science, and ML technologies.
Collaboration and Education
In tandem with its scientific investigations, CHEFSI will devote significant effort to enacting collaboration and educational plans.
Collaboration
Working with colleagues in the US DOE science and engineering ecosystem
As an integral part of its research and educational activities, the Center will pursue extensive collaborations with colleagues at DOE/NNSA laboratories. These will include direct engagement of collaborators on MIT graduate courses, research sub-teams, joint publications, structured and unstructured visits and student/postdoc/researcher internships at NNSA labs, data and code sharing and transitioning. These will build upon our already numerous connections and vibrant collaborations with NNSA researchers.
Education
Educating future S&T leaders and strengething tomorrow’s workforce
Our educational activities will focus on (1) strengthening the pipelines that feed the graduate programs involved in this research, and (2) infusing our curricular offerings with modern, real-world material, providing a unique opportunity for the NNSA research agenda to shape CSE education at MIT for the foreseeable future.