Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Turbulence simulation”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 379 records · Page 21
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Subcell limiting strategies for discontinuous Galerkin spectral element methods

Here, we present a general family of subcell limiting strategies to construct robust high-order accurate nodal discontinuous Galerkin (DG) schemes. The main strategy is to construct compatible low order finite volume (FV) type discretizations that allow for convex blending with the high-order variant with the goal of guaranteeing additional properties, such as bounds on physical quantities and/or guaranteed entropy dissipation. For an implementation of this main strategy, four main ingredients are identified that may be combined in a flexible manner: (i) a nodal high-order DG method on Legendre–Gauss–Lobatto nodes, (ii) a compatible robust subcell FV scheme, (iii) a convex combination strategy for the two schemes, which can be element-wise or subcell-wise, and (iv) a strategy to compute the convex blending factors, which can be either based on heuristic troubled-cell indicators, or using ideas from flux-corrected transport methods. By carefully designing the metric terms of the subcell FV method, the resulting methods can be used on unstructured curvilinear meshes, are locally conservative, can handle strong shocks efficiently while directly guaranteeing physical bounds on quantities such as density, pressure or entropy. We further show that it is possible to choose the four ingredients to recover existing methods such as a provably entropy dissipative subcell shock-capturing approach or a sparse invariant domain preserving approach. We test the versatility of the presented strategies and mix and match the four ingredients to solve challenging simulation setups, such as the KPP problem (a hyperbolic conservation law with non-convex flux function), turbulent and hypersonic Euler simulations, and MHD problems featuring shocks and turbulence.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Development of a CPU/GPU portable software library for Lagrangian–Eulerian simulations of liquid sprays

The Lagrangian–Eulerian method is widely used for simulations of fuel sprays in turbulent combustion because of the advantage of treating the spray droplets as discrete points. One challenge of the Lagrangian–Eulerian method is the intense computational requirement when tracking the large number of Lagrangian particles needed for high fidelity. We have developed a performance-portable library, Grit, to track the Lagrangian particles in parallel on central processing unit (CPU) and graphics processing unit (GPU) accelerated high performance computing (HPC) architectures. Grit is a C++ library which employs Message Passing Interface (MPI) for distributed memory parallelism and Kokkos programming model for on-node shared memory parallelism with performance portability across different architectures of GPUs and multi-core/manycore CPUs. The parallel algorithms, key parallel kernels, and their performances on the pre-exascale supercomputer, Summit, are presented. Grit is coupled with a direct numerical simulation (DNS) solver, S3D, for multiphase simulations. A conservative formulation has been developed and implemented in Grit for phase coupling with thermodynamic consistency. The formulation separates the conservation of mass, momentum and energy from the physical models to prevent accidental violation of conservation laws due to inconsistent models. The formulation also enforces consistent definitions of enthalpies of the fuel for both phases and the latent heat of evaporation. Finally, simulations of turbulent particle-laden flow and the evaporation of dilute turbulent spray jet are performed to verify the software implementation, and to demonstrate the scalability of Grit for large-scale multiphase simulations.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

A new mechanism for filament disconnection at the X-point: poloidal shear in radial E × B velocity

Plasma filaments generated by turbulence in the scrape-off layer are sometimes observed in experiments and simulations to disconnect from the target plate in the vicinity of the X-point, resulting in a ‘quiescent zone’ with reduced fluctuations, possibly affecting the fluctuating part of the divertor heat loads. This phenomena is usually explained by the flux tubes squeezing, induced by flux expansion, down to scales where collisions dominate. In this work, we consider an additional mechanism spontaneously arising at the X-point: the poloidal shear in radial E × B velocity. Through a simple model, tested against 3D global turbulence simulations, herein we quantify the effectiveness of this disconnection mechanism and identify ways to control it.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Pass-efficient methods for compression of high-dimensional turbulent flow data

The future of high-performance computing, specifically on future Exascale computers, will presumably see memory capacity and bandwidth fail to keep pace with data generated, for instance, from massively parallel partial differential equation (PDE) systems. Current strategies proposed to address this bottleneck entail the omission of large fractions of data, as well as the incorporation of in situ compression algorithms to avoid overuse of memory. To ensure that post-processing operations are successful, this must be done in a way that a sufficiently accurate representation of the solution is stored. Moreover, in situations where the input/output system becomes a bottleneck in analysis, visualization, etc., or the execution of the PDE solver is expensive, the number of passes made over the data must be minimized. In the interest of addressing this problem, this work focuses on the utility of pass-efficient, parallelizable, low-rank, matrix decomposition methods in compressing high-dimensional simulation data from turbulent flows. Additionally, a particular emphasis is placed on using coarse representation of the data – compatible with the PDE discretization grid – to accelerate the construction of the low-rank factorization. This includes the presentation of a novel single-pass matrix decomposition algorithm for computing the so-called interpolative decomposition. The methods are described extensively and numerical experiments on two turbulent channel flow data are performed. In the first (unladen) channel flow case, compression factors exceeding 400 are achieved while maintaining accuracy with respect to first- and second-order flow statistics. In the particle-laden case, compression factors of 100 are achieved and the compressed data is used to recover particle velocities. These results show that these compression methods can enable efficient computation of various quantities of interest in both the carrier and disperse phases.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2021

Experimental decoy-state Bennett-Brassard 1984 quantum key distribution through a turbulent channel

In free-space quantum key distribution (QKD) in turbulent conditions, scattering and beam wandering cause intensity fluctuations which decrease the detected signal-to-noise ratio. This effect can be mitigated by rejecting received bits when the channel's transmittance is below a threshold. Thus, the overall error rate is reduced and the secure key rate increases despite the deletion of bits. In this work, we implement recently proposed selection methods focusing on the prefixed-threshold real-time selection (P-RTS) where a cutoff can be chosen prior to data collection and independently of the transmittance distribution. We perform finite-size decoy-state Bennett-Brassard 1984 QKD in a laboratory setting where we simulate the atmospheric turbulence using an acousto-optical modulator. We show that P-RTS can yield considerably higher secure key rates for a wide range of the atmospheric channel parameters. In addition, we evaluate the performance of the P-RTS method for a realistically finite sample size. We demonstrate that a near-optimal selection threshold can be predetermined even with imperfect knowledge of the channel transmittance distribution parameters.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2022

Modelling of electrostatic ion-scale turbulence in divertor tokamaks with the gyrokinetic code $\mathrm{COGENT}$

Here, continuum gyrokinetic simulations of electrostatic ion scale turbulence are presented for the case of a diverted (single-null) tokamak geometry. The simulation model, implemented in the finite-volume code COGENT, solves the long-wavelength limit of the full-F gyrokinetic equation for ion species coupled to a vorticity equation for electrostatic potential variations, where a fluid model is used for an electron response. The model describes the ion scale ion temperature gradient (ITG) and resistive drift modes as well as neoclassical ion physics effects. Different turbulence regimes are observed depending on the plasma profiles, and the roles of a self-consistent background electric field and an X-point geometry are explored. In particular, increasing the pedestal density gradient and the corresponding radial electric field is demonstrated to suppress the ITG turbulence, whereas the same edge plasma background can still be destabilized by the resistive modes when the plasma resistivity is increased. The effects of X-point geometry are assessed by comparing cross-separatrix simulations with counterpart calculations performed for a toroidal annulus geometry. For the simulation parameters considered, similar global behaviour is observed in both cases, whereas strong local suppression of turbulence fluctuations is demonstrated near the X-point for the case of a single-null geometry.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Extreme Heating of Minor Ions in Imbalanced Solar-wind Turbulence

Minor ions in the solar corona are heated to extreme temperatures, far in excess of those of the electrons and protons that comprise the bulk of the plasma. These highly nonthermal distributions make minor ions sensitive probes of the collisionless processes that heat the corona and power the solar wind. The recent discovery of the "helicity barrier" offers a mechanism in which imbalanced Alfvénic turbulence in low-β plasmas preferentially heats protons over electrons, generating high-frequency, proton-cyclotron-resonant fluctuations. We use the hybrid-kinetic particle-in-cell code Pegasus++ to drive imbalanced Alfvénic turbulence in a 3D low-β plasma with additional passive ion species, He 2+ and O 5+ . A helicity barrier naturally develops, followed by clear phase-space signatures of oblique proton-cyclotron-wave heating and Landau-resonant heating from the imbalanced Alfvénic fluctuations. The former results in characteristically arced ion velocity distribution functions, whose non-bi-Maxwellian features are shown by linear ALPS calculations to be critical to the heating process. Additional features include a steep transition-range electromagnetic spectrum, proton-cyclotron waves propagating in the direction of the imbalance, significantly enhanced proton-to-electron heating ratios, ion temperatures that are considerably more perpendicular with respect to magnetic field, and extreme heating of heavier species in a manner consistent with mass scalings inferred from spacecraft measurements. None of these features are realized in an otherwise equivalent simulation of balanced turbulence. If seen simultaneously in the fast solar wind, these signatures of the helicity barrier would testify to the necessity of incorporating turbulence imbalance in a complete theory for the evolution of the solar wind.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Development and preliminary results of 270 GHz microwave forward scattering diagnostic system on the experimental advanced superconducting tokamak (EAST)

To measure localized (intermediate, high) poloidal wavenumber electron density fluctuations, a tangential millimeter-wave collective scattering system has been designed and successfully developed on the experimental advanced superconducting tokamak (EAST). This innovative system employs a 270 GHz mm-wave probe beam, emitted from the mid-plane of Port K and directed towards Port P after passing through a combination of two in-vessel mirrors. Here these two ports are located on the EAST device’s midplane and are 110° apart. The downward scattered signals pass through two in-vessel mirrors, and exit through the P window. The received multi-channel separated scattering signals are arranged along the poloidal direction. This optical arrangement makes the monitored turbulent wave number is poloidal dominant and up to 40 cm −1 . The tangential microwave scattering scheme enables this diagnostic to monitor local density fluctuations, with typically a radial spatial resolution △ R = 5 cm, and a wavenumber resolution of △ k = 0.4 cm −1 . The scattering system underwent comprehensive laboratory testing in 2023, with installation finalized in 2024. The system features steerable launch and receiver optics, allowing remote control to position the scattering volume from normalized radius ρ –0 to the pedestal region on a shot-by-shot basis. Effective experimental data, characterized by distinct off-center spectral peaks, have been successfully obtained in neutral beam-heated plasmas on EAST. The beam tracing forward modeling has been employed for experimental data interpretation. The advanced millimeter-wave scattering system provides fluctuation measurement capability from ion temperature gradient to electron temperature gradient scale on low field side. Preliminary experimental density fluctuation data have been successfully obtained in neutral beam-heated plasmas on EAST. The localized measurement will be used for kinetic turbulence transport numerical simulation validation.

high-k collective scattering↗
DOE OSTI2021

Sensitivity of Grass Fires Burning in Marginal Conditions to Atmospheric Turbulence

Abstract Atmospheric forcing and interactions between the fire and atmosphere are primary drivers of wildland fire behavior. The atmosphere is known to be a chaotic system that, although deterministic, is very sensitive to small perturbations to initial conditions. We assume that as a result of the tight coupling between fire and atmosphere; wildland fire behavior, in turn, should also be sensitive to perturbations in atmospheric initial conditions. Observations suggest that low intensity prescribed fire, in particular, is susceptible to small perturbations in the wind field, which can significantly alter fire spread. Here, we employ a computational fluid dynamics model of coupled fire‐atmosphere interactions to answer the question: How sensitive is fire behavior to small variations in atmospheric turbulence? We perform ensemble simulations of fires in homogenous grass fuels. The only difference between ensemble members is the state of the turbulent atmosphere provided to the model throughout the simulation. The atmospheric state is a function of the initial conditions applied at the start of the simulation and boundary conditions applied throughout the simulation. We find a wide range of outcomes, with area burned ranging from 2,212 to 11,236 m 2 (>400% change), driven primarily by sensitivity to initial conditions, with nonnegligible contributions from boundary condition variability during the initial 30 s of simulation. Our results highlight the need for ensemble simulations, especially when considering fire behavior in marginal burning conditions.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2024

Simulating Mixed‐Phase Open Cellular Clouds Observed During COMBLE: Evaluation of Parameterized Turbulence Closure

Marine cold-air outbreaks, or CAOs, are airmass transformations whereby relatively cold boundary layer (BL) air is transported over relatively warm water. To more deeply understand BL and mixed-phase cloud properties during CAO conditions, the Cold-Air Outbreaks in the Marine Boundary Layer Experiment (COMBLE) took place from late 2019 into early 2020. During COMBLE, the U.S. Department of Energy's first Atmospheric Radiation Measurement Mobile Facility (AMF1) was deployed to Andenes, Norway, far downstream (~1,000 km) from the Arctic pack ice. This study examines the two most intense CAOs sampled at the AMF1 site. The observed BL structures are open cellular with high (~3–5 km) and cold (–30 to –50 °C) cloud tops, and they often have pockets of high liquid water paths (LWPs; up to ~1,000 g m –2 ) associated with strong updrafts and enhanced turbulence. We use a high-resolution mesoscale model to explore how well four turbulence closure methods represent open cellular clouds. After applying a radar simulator to model outputs for direct evaluation, cloud top properties agree well with AMF1 observations (within ~10%), but radar reflectivity and LWP agreement is more variable. Results suggest that the turbulent Prandtl number may play an important role for the simulated BL and cloud properties. All simulations produce enhanced precipitation rates that are well-correlated with a cloud transition. Finally, the eddy-diffusivity/mass-flux approach produces the deepest cloud layer and therefore the largest and most coherent cellular structures. Furthermore, we recommend the use of a non-local turbulence closure approach to better capture turbulent processes in intense CAOs.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2022

The role of breakup and coalescence in fine-scale bubble-induced turbulence. I. Dynamics

We study the effect of bubble breakup and coalescence on fine-scale dynamics of bubbly turbulent flows using direct numerical simulations. We perform two different simulations of dilute bubbly flows of void fraction 0.5%: one with bubbles breaking up and coalescing and the other without these physical processes. The volume of the fluid method is used for simulating bubbles undergoing breakup and coalescence while the bubbles are treated as rigid spheres in the immersed boundary method simulation. The energy spectrum in both types of simulation, consistent with previous studies, exhibits a -3 slope. We follow a single infinitesimal fluid element as it evolves to understand velocity gradient dynamics using conditional mean trajectories. We note finite-time divergence when the fluid element evolves under the action of inertial and pressure dynamics. The inertial, pressure, and viscous velocity gradient dynamics, when considered individually, produce the same results for bubble-induced turbulence (BIT) as with the classical homogeneous isotropic turbulence (HIT). Yet when the overall velocity gradient dynamics is considered, BIT results in non-cyclic trajectories moving toward stable node and unstable saddle while classical HIT shows cyclic behavior in their trajectories that move toward the origin. Interestingly, both the volume of fluid and immersed boundary simulations produce similar results. Therefore, there are two main takeaways from this research. First, new velocity gradient models are needed for BIT as their velocity gradients behavior is entirely different from the HIT. Second, we can neglect the bubble topology, breakup, and coalescence while studying or modeling the fine-scale dynamics of BIT.

42 ENGINEERING↗
DOE OSTI2024

Conservative discontinuous Galerkin interpolation: Sheared boundary conditions

Local studies of accretion disks and laboratory magnetized plasmas employ analytical coordinate mappings that introduce sheared boundary conditions (BCs). Here, we present a discontinuous Galerkin (DG) algorithm to apply such BCs based on projections and quadrature-free integration. The procedure is high-order accurate, preserves moments exactly and works in multiple dimensions. Tests of increasing complexity are provided, beginning with translations of one and two dimensional fields, followed by 3D and 5D simulations with sheared (twist-shift) BCs. Results show that the algorithm is (p + 1)-order accurate in the DG representation and (p + 2) -order accurate in the cell averages, with p being the order of the polynomial basis. Quantification of the algorithm's hyperdiffusion and discussion of aliasing errors are given. This technique enables conservative local simulations of plasma turbulence with DG, not possible until now.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Planck 2018 results XII. Galactic astrophysics using polarized dust emission

Observations of the submillimetre emission from Galactic dust, in both total intensity I and polarization, have received tremendous interest thanks to the Planck full-sky maps. In this paper we make use of such full-sky maps of dust polarized emission produced from the third public release of Planck data. As the basis for expanding on astrophysical studies of the polarized thermal emission from Galactic dust, we present full-sky maps of the dust polarization fraction p, polarization angle ψ, and dispersion function of polarization angles S . The joint distribution (one-point statistics) of p and NH confirms that the mean and maximum polarization fractions decrease with increasing NH. The uncertainty on the maximum observed polarization fraction, pmax = 22.0 -1.4 +3.5% at 353 GHz and 80' resolution, is dominated by the uncertainty on the Galactic emission zero level in total intensity, in particular towards diffuse lines of sight at high Galactic latitudes. Furthermore, the inverse behaviour between p and S found earlier is seen to be present at high latitudes. This follows the S ∝ p -1 relationship expected from models of the polarized sky (including numerical simulations of magnetohydrodynamical turbulence) that include effects from only the topology of the turbulent magnetic field, but otherwise have uniform alignment and dust properties. Thus, the statistical properties of p, ψ, and S for the most part reflect the structure of the Galactic magnetic field. Nevertheless, we search for potential signatures of varying grain alignment and dust properties. First, we analyse the product map S × p, looking for residual trends. While the polarization fraction p decreases by a factor of 3-4 between NH = 10 20 cm -2 and NH = 2 x 10 22 cm -2 , out of the Galactic plane, this product S × p only decreases by about 25%. Because S is independent of the grain alignment efficiency, this demonstrates that the systematic decrease in p with N H is determined mostly by the magnetic-field structure and not by a drop in grain alignment. This systematic trend is observed both in the diffuse interstellar medium (ISM) and in molecular clouds of the Gould Belt. Second, we look for a dependence of polarization properties on the dust temperature, as we would expect from the radiative alignment torque (RAT) theory. We find no systematic trend of S × p with the dust temperature T d , whether in the diffuse ISM or in the molecular clouds of the Gould Belt. In the diffuse ISM, lines of sight with high polarization fraction p and low polarization angle dispersion S tend, on the contrary, to have colder dust than lines of sight with low p and high S . We also compare the Planck thermal dust polarization with starlight polarization data in the visible at high Galactic latitudes. The agreement in polarization angles is remarkable, and is consistent with what we expect from the noise and the observed dispersion of polarization angles in the visible on the scale of the Planck beam. The two polarization emission-to-extinction ratios, R P/p and R S/V , which primarily characterize dust optical properties, have only a weak dependence on the column density, and converge towards the values previously determined for translucent lines of sight. We also determine an upper limit for the polarization fraction in extinction, p V /E(B - V), of 13% at high Galactic latitude, compatible with the polarization fraction p ≈ 20% observed at 353 GHz. Taken together, these results provide strong constraints for models of Galactic dust in diffuse gas.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

Computational Thermal Hydraulics of a High-Performance Low-Enriched-Uranium Annular Target for HFIR Irradiation

Molybdenum-99 has historically been generated via isolation from fissioned highly enriched uranium (HEU) targets. Here, this isotope is in high demand due to its daily use across the world in radiopharmaceutical medical procedures. The primary objective of this work was to design and analyze an experimental target assembly containing one low-enriched-uranium (LEU) annular target for irradiation at the High Flux Isotope Reactor (HFIR). Efforts included incorporating spatially dependent energy sources from neutron and gamma interactions, quantifying thermal contact conductance at material interfaces, performing grid-independent studies, comparing turbulence models, and simulating various steady-state and transient scenarios relevant for irradiation qualification and eventual insertion. These models provide velocity, pressure, and temperature distributions in both space and time. Such results enable the selection of an appropriate irradiation location, fission rate density, and flow-limiting orifice size and demonstrate compliance with HFIR safety requirements such that insertion into the reactor can be approved. This analysis shows that across all scenarios, wetted surface temperatures remain below the coolant saturation temperature with no net vapor formation in the coolant. In every scenario, all components stay below 30% of the aluminum 6061 melting temperature. Computational fluid dynamics and system-level models predict peak target temperatures that agree within 4%, though the predicted axial location of the peak differs by about 10% of the heated length due to differences in flow development length. These results de-risk the irradiation of LEU (annular targets) and strengthen a domestic, HEU-independent 99 Mo supply by providing important fuel performance data to form the foundation for a robust licensing basis.

Molybdenum-99↗
DOE OSTI2024

Reconstruction and interpretation of ionization asymmetry in magnetic confinement via synthetic diagnostics

Abstract Strong poloidal refueling asymmetry in the DIII-D tokamak is inferred from line radiation measurements. Synthetic diagnostics in neutral transport modeling coupled to gyrokinetic simulations illuminate implications for the plasma flow profile in the scrape-off layer of single-null beam-driven discharges. Recycling occurs primarily either on the inner or outer divertor legs, depending on the toroidal magnetic field direction. By reversing the toroidal magnetic field, the observed line radiation asymmetry is nearly eliminated or reversed. It is determined that, while relatively simple physics can describe the observed ionization asymmetry, predicting the overall brightness of the hydrogenic Lyman- α signal requires detailed simulation of the plasma and resulting turbulence. To this end, kinetic plasma simulations fully coupled to comprehensive neutral transport calculations—a novel capability—provide first-principles reproduction of Lyman- α observations on DIII-D.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2021

Wind Energy Forecasting with the Weather Research and Forecasting Model

This was a collaborative effort between Lawrence Livermore National Security, LLC as manager and operator of Lawrence Livermore National Laboratory (LLNL) and Siemens Energy, Inc. (Siemens) to develop a wind resource forecasting tool. LLNL was to develop an independent high-resolution mesoscale modeling capability forecasting tool that could be implemented in conjunction with existing wind farm control and monitoring software to provide forecasting of wind resources using local observations of winds and temperature. Research with LLNL’s state-of-the-art large-eddy simulation meteorological prediction model, based on the community WRF model and innovative turbulence parameterizations, would improve that model’s applicability to large wind farms offshore and in complex terrain. The modeling capability would include uncertainty quantification. Finally, the application of the modeling tool and existing global climate change predictions would enable the delineation of the likely effects of climate change on wind resources. Siemens was to provide high time resolution hub-height wind speed and other meteorological data streams, including temperature profiles from wind farms, for LLNL to incorporate into the modeling system, to validate and tune this forecasting model for their locations of interest. These data streams would also be used for longer-term studies of correlations of wind resources to climate oscillations to indicate how long-term climate change trends may affect the available wind resource. Siemens would also provide information and observations of turbine wakes for incorporation into the modeling tool. By implementing state-of-the-art turbulence parameterizations into a simulation model and/or ensembles of simulation models, and by integrating real-time hub height wind speed and other meteorological datastreams from wind farms into that model or ensemble of models, LLNL would develop a forecasting tool that could be implemented by Siemens as an add-on to existing wind farm control and monitoring software to provide owners with useful resource forecasting. The desired outcome was that the accuracy level of the output would be sufficient to substantiate power output commitments. The final deliverable for this work would consist of a document outlining the algorithms and software tools that could be integrated into Siemens Wind Park Supervisor.

17 WIND ENERGY↗
DOE OSTI2024

Symbolic diagnostics to interpret and analyze neural network models

Embedded machine-learned models (EMLMs) have the promise to improve the predictive accuracy of engineering simulators in environments of national interest. EMLMs often comprise complex input-output maps (e.g., neural networks), which make them unamenable to rigorous analysis and generally difficult to interpret. In the face of decades of theory, this lack of interpretability is a significant barrier to building confidence in these models. This work outlines an approach to interpret EMLMs using sparse polynomial regression for comparison with theoretical understanding. To do so, we build on the concept of Locally Interpretable Model-agnostic Explanations (LIME) using physics-informed clustering, prototype selection, and library construction. While general, we demonstrate our method on tensor-basis neural networks used in Reynolds-Averaged Navier-Stokes simulations of hypersonic fluid flows. Results are presented for a simulated toy model and for direct numerical simulations (DNS) of turbulent flows over a flat plate.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Physics-Reinforced Machine Learning Algorithms for Multiscale Closure Model Discovery

The central objective of this project was to address the challenge of modeling and simulating complex multiscale turbulence phenomena by leveraging physics-guided machine learning (PGML) and hybrid modeling approaches. By integrating physics-based methods with data-driven models, the research focused on achieving robust and scalable solutions for geophysical turbulence, enhancing numerical weather prediction and climate research tools. The project resulted in significant advancements in computational modeling paradigms, predictive tools for reduced-order modeling, and innovative algorithms for fluid dynamics.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗