Search NASA⌕ Search

SEARCH · Search NASA

Results for “arbitrary meshes”

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 433 records · Page 24

PRISMS-PF: A general framework for phase-field modeling with a matrix-free finite element method

Abstract A new phase-field modeling framework with an emphasis on performance, flexibility, and ease of use is presented. Foremost among the strategies employed to fulfill these objectives are the use of a matrix-free finite element method and a modular, application-centric code structure. This approach is implemented in the new open-source PRISMS-PF framework. Its performance is enabled by the combination of a matrix-free variant of the finite element method with adaptive mesh refinement, explicit time integration, and multilevel parallelism. Benchmark testing with a particle growth problem shows PRISMS-PF with adaptive mesh refinement and higher-order elements to be up to 12 times faster than a finite difference code employing a second-order-accurate spatial discretization and first-order-accurate explicit time integration. Furthermore, for a two-dimensional solidification benchmark problem, the performance of PRISMS-PF meets or exceeds that of phase-field frameworks that focus on implicit/semi-implicit time stepping, even though the benchmark problem’s small computational size reduces the scalability advantage of explicit time-integration schemes. PRISMS-PF supports an arbitrary number of coupled governing equations. The code structure simplifies the modification of these governing equations by separating their definition from the implementation of the numerical methods used to solve them. As part of its modular design, the framework includes functionality for nucleation and polycrystalline systems available in any application to further broaden the phenomena that can be used to study. The versatility of this approach is demonstrated with examples from several common types of phase-field simulations, including coarsening subsequent to spinodal decomposition, solidification, precipitation, grain growth, and corrosion.

36 MATERIALS SCIENCE↗

Free stream capturing in fluid conservation law for moving coordinates in three dimensions

The free-stream capturing technique for both the finite-volume (FV) and finite-difference (FD) framework is summarized. For an arbitrary motion of the grid, the FV analysis shows that volumes swept by all six surfaces of the cell have to be computed correctly. This means that the free-stream capturing time-metric terms should be calculated not only from a surface vector of a cell at a single time level, but also from a volume swept by the cell surface in space and time. The FV free-stream capturing formulation is applicable to the FD formulation by proper translation from an FV cell to an FD mesh.

Obayashi, Shigeru↗

High-order methods for wind turbine simulations and moving to next-generation platforms

The goal of the ExaWind project is to enable predictive simulations of wind farms comprised of many megawatt-scale turbines situated in complex terrain. Predictive simulations will require computational fluid dynamics (CFD) simulations for which the mesh resolves the geometry of the turbines and captures the rotation and large deflections of blades. Whereas such simulations for a single turbine are arguably petascale class, multi-turbine wind farm simulations will require exascale-class resources. The primary physics codes in the ExaWind project are Nalu-Wind, which is an unstructured-grid solver for the acoustically incompressible Navier-Stokes equations, and OpenFAST, which is a whole-turbine simulation code. The Nalu-Wind model consists of the mass-continuity Poisson-type equation for pressure and a momentum equation for the velocity. For such modeling approaches, simulation times are dominated by linear-system setup and solution for the continuity and momentum systems. For the ExaWind challenge problem, the moving meshes greatly affect overall solver costs as reinitialization of matrices and recomputation of preconditioners is required at every time step. This milestone represents an effort to increase the fidelity of Nalu-Wind at a fixed resolution through the implementation of a tensor-product based, matrix-free high order scheme. High order finite element methods have increased local work per datum communicated and have the potential to provide significantly more accurate solutions at a fixed number of degrees of freedom. Previous to this milestone, Nalu-Wind had an arbitrary order Control Volume Finite Element Method discretization as a solver option, but it required too much memory and was too slow to be of practical use. The work in this milestone addresses these issues by first implementing an implicit, high order solver that only partially assembles the global system. This reduces the memory footprint of the high-order scheme by orders of magnitude for higher polynomial orders. Second, a faster, tensor-product based method for evaluating the action of the left-hand side was implemented. This reduces the amount of computational work required by the scheme and dramatically enhanced the time-to-solution on example problems. Finally, this milestone is an evaluation of the value of high order methods in the wind application space. With the enhancements to memory and computational cost, accuracy vs. time-to-solution was evaluated for several resolutions on an under-resolved Taylor Green vortex test case. Results show that the high order scheme is cost-competitive with the production low-order schemes in Nalu-Wind, being moderately more expensive than the production edge-based vertex centered finite volume scheme. The evaluation of accuracy on the test case shows a potential benefit to high order at the highest resolution while not deteriorating accuracy on the lowest tested resolution. More work is needed to show value in the wind application, but positive strides have been made.

17 WIND ENERGY↗

Verification of the 2D Finite Element SP3 Solutions

Method(FEM) is the free geometry handling. However many of the geometries of the nuclear engineering problems are not unstructured but rather structured, such as a hexahedral, or triangular prism. Thus, the need for the use of FEM has less portion in nuclear engineering. However it is also certain that it is very hard to exclude the possibility for use of an unstructured geometry such as a PBMR core problem, fusion blanket analysis of the source problem, another design issue with an unstructured geometry. In addition, research regarding the acceleration scheme such as Coarse Mesh Finite Difference (CMFD) showed up in nuclear engineering in recent years. By considering these facts, the FEM cannot be totally exclusive. To utilize the FEM for nuclear engineering problems, several types of assistance are needed. The first is parallelization because of its slow calculation speed. The second is an acceleration scheme. With these two types of assistance the FEM can be utilized freely in the nuclear engineering field. In this research, before conducting parallelization and acceleration, the final verification for the implementation of the up-scattering treatment with an arbitrary geometry conducted through a comparison with the McCARD solution through the well-known C5G7 2D benchmark problem. Note that the earlier work for verification of the FEM for the multi-group problem without up-scattering and with a structured geometry including a mathematical formulation has already been done. The main points that will be tested here are as follows. The first is a check of the right implementation for the multi-group problem with an up-scattering cross section from the perspectives of the multiplication factor and power. The second is the transport effect for this problem by comparing the solutions from P{sub 1} and Simplified P{sub 3}(SP{sub 3}). The third is the effectiveness of the higher order basis function for the same number of unknowns. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

VPIC v.1.2

Vector Particle-in-Cell (VPIC) is a general purpose particle-in-cell simulation code for modeling kinetic plasmas in one, two, or three spatial dimensions. It employs a second-order, explicit, leapfrog algorithm to update charged particle positions and velocities in order to solve the relativistic kinetic equation for each species in the plasma, along with a full Maxwell description for the electric and magnetic fields evolved via a second- order finite-difference-time-domain (FDTD) solve. The VPIC code has been optimized for modern computing architectures and uses Message Passing Interface (MPI) calls for multi-node application as well as data parallelism using threads. VPIC employs a variety of short-vector, single-instruction-multiple-data (SIMD) intrinsics for high performance and has been designed so that the data structures align with cache boundaries. The current feature set for VPIC includes a flexible input deck format capable of treating a wide variety of problems. These include: the ability to treat electromagnetic materials (scalar and tensor dielectric, conductivity, and diamagnetic material properties); multiple emission models, including user-configurable models; arbitrary, user-configurable boundary conditions for particles and fields; user- definable simulation units; a suite of "standard" diagnostics, as well as user-configurable diagnostics; a Monte-Carlo treatment of collisional processes capable of treating binary and unary collisions and secondary particle generation; and, flexible checkpoint-restart semantics enabling VPIC checkpoint files to be read as input for subsequent simulations. VPIC has a native I/O format that interfaces with the high-performance visualization software Ensight and Paraview. While the common use cases for VPIC employ low-order particles on rectilinear meshes, a framework exists to treat higher-order particles and curvilinear meshes, as well as more advanced field solvers.

Bowers, Kevin↗

Creep crack-growth: A new path-independent integral (T sub c), and computational studies

The development of valid creep fracture criteria is considered. Two path-independent integral parameters which show some degree of promise are the C* and (Delta T)sub c integrals. The mathematical aspects of these parameters are reviewed by deriving generalized vector forms of the parameters using conservation laws which are valid for arbitrary, three dimensional, cracked bodies with crack surface tractions (or applied displacements), body forces, inertial effects, and large deformations. Two principal conclusions are that (Delta T)sub c has an energy rate interpretation whereas C* does not. The development and application of fracture criteria often involves the solution of boundary/initial value problems associated with deformation and stresses. The finite element method is used for this purpose. An efficient, small displacement, infinitesimal strain, displacement based finite element model is specialized to two dimensional plane stress and plane strain and to power law creep constitutive relations. A mesh shifting/remeshing procedure is used for simulating crack growth. The model is implemented with the quartz-point node technique and also with specially developed, conforming, crack-tip singularity elements which provide for the r to the n-(1+n) power strain singularity associated with the HRR crack-tip field. Comparisons are made with a variety of analytical solutions and alternate numerical solutions for a number of problems.

Stonesifer, R. B.↗

A Thermo‐Flow‐Mechanics‐Fracture Model Coupling a Phase‐Field Interface Approach and Thermo‐Fluid‐Structure Interaction

This work proposes a novel approach for coupling non-isothermal fluid dynamics with fracture mechanics to capture thermal effects within fluid-filled fractures accurately. This method addresses critical aspects of calculating fracture width in enhanced geothermal systems, where the temperature effects of fractures are crucial. The proposed algorithm features an iterative coupling between an interface-capturing phase-field fracture method and interface-tracking thermo-fluid-structure interaction using arbitrary Lagrangian–Eulerian coordinates. We use a phase-field approach to represent fractures and reconstruct the geometry to frame a thermo-fluid-structure interaction problem, resulting in pressure and temperature fields that drive fracture propagation. We developed a novel phase-field interface model accounting for thermal effects, enabling the coupling of quantities specific to the fluid-filled fracture with the phase-field model through the interface between the fracture and the intact solid domain. We provide several numerical examples to demonstrate the capabilities of the proposed algorithm. In particular, we analyze mesh convergence of our phase-field interface model, investigate the effects of temperature on crack width and volume in a static regime, and highlight the method's potential for modeling slowly propagating fractures.

fracture↗

Discrete crack growth analysis methodology for through cracks in pressurized fuselage structures

A methodology for simulating the growth of long through cracks in the skin of pressurized aircraft fuselage structures is described. Crack trajectories are allowed to be arbitrary and are computed as part of the simulation. The interaction between the mechanical loads acting on the superstructure and the local structural response near the crack tips is accounted for by employing a hierarchical modeling strategy. The structural response for each cracked configuration is obtained using a geometrically nonlinear shell finite element analysis procedure. Four stress intensity factors, two for membrane behavior and two for bending using Kirchhoff plate theory, are computed using an extension of the modified crack closure integral method. Crack trajectories are determined by applying the maximum tangential stress criterion. Crack growth results in localized mesh deletion, and the deletion regions are remeshed automatically using a newly developed all-quadrilateral meshing algorithm. The effectiveness of the methodology and its applicability to performing practical analyses of realistic structures is demonstrated by simulating curvilinear crack growth in a fuselage panel that is representative of a typical narrow-body aircraft. The predicted crack trajectory and fatigue life compare well with measurements of these same quantities from a full-scale pressurized panel test.

Potyondy, David O.↗

Active Vibration Control for Helicopter Interior Noise Reduction Using Power Minimization

This report describes work performed by United Technologies Research Center (UTRC) for NASA Langley Research Center (LaRC) under Contract NNL11AA06C. The objective of this program is to develop technology to reduce helicopter interior noise resulting from multiple gear meshing frequencies. A novel active vibration control approach called Minimum Actuation Power (MAP) is developed. MAP is an optimal control strategy that minimizes the total input power into a structure by monitoring and varying the input power of controlling sources. MAP control was implemented without explicit knowledge of the phasing and magnitude of the excitation sources by driving the real part of the input power from the controlling sources to zero. It is shown that this occurs when the total mechanical input power from the excitation and controlling sources is a minimum. MAP theory is developed for multiple excitation sources with arbitrary relative phasing for single or multiple discrete frequencies and controlled by a single or multiple controlling sources. Simulations and experimental results demonstrate the feasibility of MAP for structural vibration reduction of a realistic rotorcraft interior structure. MAP control resulted in significant average global vibration reduction of a single frequency and multiple frequency excitations with one controlling actuator. Simulations also demonstrate the potential effectiveness of the observed vibration reductions on interior radiated noise.

Mendoza, J.↗

LDRD Abbreviated report: High-Order General-Discrete-Ordinates Method Enabling Efficient Deterministic Transport in Hydrodynamic Simulations

Deterministic transport simulations for national-security and energy applications often operate in high-dimensional phase-space, where accuracy and cost both become major challenges. A common numerical artifact in such problems is the “ray-effect,” which appears as unphysical streaks. Beyond misinterpretation, these artifacts can contaminate tightly coupled physics, such as fluid dynamics, radiation-hydrodynamics, and laser-plasma interactions, eroding the predictive capability of entire multiphysics workflows. Our objective was to make high-dimension studies practical on modern hardware while mitigating the ray-effect without relying on prohibitively expensive sampling approaches such as Monte Carlo methods. We developed the Generic Discretization Library (GenDiL), a Graphics Processing Unit (GPU)-first framework that uses high-order Discontinuous Galerkin (DG) methods and matrix-free algorithms to reduce memory usage and improve computational efficiency, critical for phase-space simulations. GenDiL supports phase-space adaptivity in both mesh size and polynomial order (hp-adaptivity) to place resolution only where it is needed. A central capability is Local Dimensional Refinement (LDR), which couples lower-dimension continuum models to higher-dimension kinetic models through stable and conservative interfaces, so that high-fidelity physics is applied only in regions where it is essential. Building on the GenDiL framework, we developed the General SN (GSN) family of algorithms as a true generalization of the polar SN approach (discrete ordinates, often denoted SN). Rather than tying discrete ordinates to a specific polar change of coordinates, GSN formulates transport on an arbitrary change of coordinates chosen to reduce ray-effect. We studied two complementary variants: an analytic variant, where the coordinate map is prescribed in advance by a closed-form function; and a data-driven variant, where a quantity of interest, such as the net flux, guides the coordinate system. GenDiL provides the library infrastructure for efficient GPU execution, but the GSN concept is algorithmic and independent of any one library. Across representative high-dimension tests, including non-symmetric solutions, both variants delivered strong ray-effect mitigation at practical cost, moving four- to six-dimensional analysis toward repeatable, routine studies.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

A fully implicit, asymptotic-preserving, semi-Lagrangian algorithm for the time dependent anisotropic heat transport equation

In this paper, we extend the operator-split asymptotic-preserving, semi-Lagrangian algorithm for time dependent anisotropic heat transport equation proposed in Chacón et al. (2014) [18] to use a fully implicit time integration with backward differentiation formulas. The proposed implicit method can deal with arbitrary heat-transport anisotropy ratios $\mathcal{X}$∥ /$ \mathcal{X}$⟂ $\ggg$ 1 (with $\mathcal{X}$∥, $ \mathcal{X}$⟂ the parallel and perpendicular heat diffusivities, respectively) in complicated magnetic field topologies in an accurate and efficient manner. Further, the implicit algorithm is second-order accurate temporally and demonstrates an accurate treatment at boundary layers (e.g., island separatrices), which was not ensured by the operator-split implementation. The condition number of the resulting algebraic system is independent of the anisotropy ratio, and is inverted with preconditioned GMRES. We propose a simple preconditioner that renders the finite-dimensional linear operator compact, resulting in mesh-independent convergence rates for topologically simple magnetic fields, and convergence rates scaling as ~ (NΔt) 1/4 (with N the total mesh size and Δt the timestep) in topologically complex magnetic-field configurations. We demonstrate the accuracy and performance of the approach with test problems of varying complexity, including an analytically tractable boundary-layer problem in a straight magnetic field, and a topologically complex magnetic field featuring magnetic islands with extreme anisotropy ratios $\mathcal{X}$∥ /$ \mathcal{X}$⟂ = 10 10 ) .

97 MATHEMATICS AND COMPUTING↗

The MOOSE fluid properties module

The fluid properties module in MOOSE~\cite{lindsay2022moose} serves a variety of fluid simulation applications based on MOOSE, including the MOOSE Navier Stokes module~\cite{moose_ns}, Pronghorn~\cite{pgh}, the MOOSE thermal hydraulics module, SAM~\cite{sam}, RELAP-7~\cite{relap7}, Sockeye~\cite{sockeye}, Pronghorn-subchannel~\cite{subchannel} and the MOOSE porous flow module~\cite{porous}. These applications are used to solve coarse mesh multi-dimensional thermal-hydraulics~\cite{pgh}, 1D systems analysis~\cite{sam,relap7} in nuclear reactor analysis, heat pipe modeling~\cite{sockeye} and porous flow simulations~\cite{porous} for underground gas storage and water seepage. The use of consistent fluid properties across fluid flow applications facilitates coupled flow simulations~\cite{anl_sam_pgh,osti_1889653}. Each application has historically driven the implementation of several fluid properties, which were later extended to be compatible with other applications. The unique diversity of applications of the module, due to its presence in MOOSE, has driven its expansion to new fluids, such as advanced nuclear reactor coolants and, more recently, arbitrary functions or tables-based property definitions, as detailed in section~\ref{content}, as well as numerous thermophysical properties and variable sets, as detailed in subsection~\ref{sec:prop}. The need for different discretizations of flow equations based on the compressibility of the fluid has motivated support for both primitive (pressure- and temperature-based) and conservative (internal energy- and specific volume-based) flow variables in the module. Thermodynamic relations are used to convert between these two formulations, as needed. The module serves a dual purpose of providing fluid properties for direct use in flow simulations and facilitating the implementation of user-specific fluid properties. Contributions of new properties for existing fluids or new fluids are strongly encouraged.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

NASTRAN hydroelastic modal studies. Volume 1: Introduction, theory, and results

The implementation of a general three-dimensional hydroelastic capability in the NASTRAN computer program is described. A method for analyzing the combined mode shapes of arbitrary fluid and structure finite element models was developed. The fluid was modeled with three-dimensional solid elements with options for tetrahedron, wedge, and hexahedron shapes. The elements were then connected to fluid grid points which defined the pressure in the fluid at the specified location. Matrices were formulated to provide efficient solutions for large-order problems. The structure matrices were processed separately and could be reduced using matrix condensation procedures or through a modal formulation using the normal modes of the empty structure as solution coordinates. The fluid/structure interface and the free surface were defined by the user with special NASTRAN boundary elements. A special purpose mesh generator program was provided to generate the actual NASTRAN data cards for the fluid, the structure, and the boundary elements for typical tank-type models.

Source record↗

Efficient computation of volume in flow predictions

An efficient method for calculating cell volumes for time-dependent three-dimensional flow predictions by finite volume calculations is presented. Eight arbitrary corner points are considered and the shape face is divided into two planar triangles. The volume is then dependent on the orientation of the partitioning. In the case of a hexahedron, it is noted that any open surface with a boundary that is a closed curve possesses a surface vector independent of the surface shape. Expressions are defined for the surface vector, which is independent of the partitioning surface diagonal used to quantify the volume. Using a decomposition of the cell volume involving two corners, with each the vertex of three diagonals and six corners which are vertices of one diagonal, gives portions which are tetrahedra. The resultant mesh is can be used for time-dependent finite volume calculations one requires less computer time than previous methods.

Vinokur, M.↗

A CFD Simulation Process for Fast Reactor Fuel Assemblies

A CFD modeling and simulation process for large-scale problems using an arbitrary fast reactor fuel assembly design was evaluated. Three-dimensional flow distributions of sodium for several fast reactor fuel assembly pin spacing configurations were simulated on high performance computers using commercial CFD software. This research focused on 19-pin fuel assembly “benchmark” geometry, similar in design to the Advanced Burner Test Reactor, where each pin is separated by helical wire-wrap spacers. Several two-equation turbulence models including the k–e and SST (Menter) k–? were evaluated. Considerable effort was taken to resolve the momentum boundary layer, so as to eliminate the need for wall functions and reduce computational uncertainty. High performance computers were required to generate the hybrid meshes needed to predict secondary flows created by the wire-wrap spacers; computational meshes ranging from 65 to 85 million elements were common. A general validation methodology was followed, including mesh refinement and comparison of numerical results with empirical correlations. Predictions for velocity, temperature, and pressure distribution are shown. The uncertainty of numerical models, importance of high fidelity experimental data, and the challenges associated with simulating and validating large production-type problems are presented.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗

Three-dimensional boundary layer calculation by a characteristic method

A numerical method for solving the three-dimensional boundary layer equations for bodies of arbitrary shape is presented. In laminar flows, the application domain extends from incompressible to hypersonic flows with the assumption of chemical equilibrium. For turbulent boundary layers, the application domain is limited by the validity of the mixing length model used. In order to respect the hyperbolic nature of the equations reduced to first order partial derivative terms, the momentum equations are discretized along the local streamlines using of the osculator tangent plane at each node of the body fitted coordinate system. With this original approach, it is possible to overcome the use of the generalized coordinates, and therefore, it is not necessary to impose an extra hypothesis about the regularity of the mesh in which the boundary conditions are given. By doing so, it is possible to limit, and sometimes to suppress, the pre-treatment of the data coming from an inviscid calculation. Although the proposed scheme is only semi-implicit, the method remains numerically very efficient.

Houdeville, R.↗

A CFD M&S PROCESS FOR FAST REACTOR FUEL ASSEMBLIES

A CFD modeling and simulation process for large-scale problems using an arbitrary fast reactor fuel assembly design was evaluated. Three dimensional flow distributions of sodium for several fast reactor fuel assembly pin spacing configurations were simulated on high performance computers using commercial CFD software. This research focused on 19-pin fuel assembly “benchmark” geometry, similar in design to the Advanced Burner Test Reactor, where each pin is separated by helical wire-wrap spacers. Several two-equation turbulence models including the k-e and SST (Menter) k-? were evaluated. Considerable effort was taken to resolve the momentum boundary layer, so as to eliminate the need for wall functions and reduce computational uncertainty. High performance computers were required to generate the hybrid meshes needed to predict secondary flows created by the wire-wrap spacers; computational meshes ranging from 65 to 85 million elements were common. A general validation methodology was followed, including mesh refinement and comparison of numerical results with empirical correlations. Predictions for velocity, temperature, and pressure distribution are shown. The uncertainty of numerical models, importance of high fidelity experimental data, and the challenges associated with simulating and validating large production-type problems are presented.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗

Differentiable Multiphysics Codes: A Breakthrough Technology for Simulation and Computing

This document summarizes the findings of a strategic planning exercise commissioned by the Weapons Simulation and Computing, Computational Physics (WSC/CP) program at the Lawrence Livermore National Laboratory (LLNL) in FY24. During the year, the committee met with multiple stakeholder communities to gather input, opinions, suggestions and concerns which have been incorporated throughout this document. The key findings from this exercise are summarized: • The development of multiphysics modelling and simulation (mod/sim) codes and software technologies, their deployment on exascale compute platforms, and their broad adoption across the NNSA is a major success of the Advanced Simulation and Computing (ASC) program and the Exascale Computing Project (ECP). Sustained investment in these core technologies is essential. • Today’s state of the art involves running ensembles of O(100K) simulations to perform uncertainty quantification (UQ) and design studies using multiple statistical methods such as Bayesian optimization to understand sensitivities of our models and explore parameterized design spaces. Even with exascale computing, we are practically limited to O(10) parameters in these studies since the number of simulations required to sample the space scales exponentially with the number of design parameters. • The data from these simulation ensembles is increasingly being used to train machine learned (ML) surrogates (or reduced order models, ROMs) which can then be used for optimization or real time design exploration. However, the trained surrogates are still limited in the number of parameters they can represent due to the sampling limitations previously noted. • Augmenting our suite of integrated multiphysics simulation codes, both current and emerging, with the ability to compute gradients (solution derivatives) of arbitrary simulation outputs with respect to (some or all) simulation inputs would be a breakthrough technology, opening the door to a new era of efficient and automated inverse design based on verified and validated mod/sim capabilities. • This capability, which we refer to as differentiable multiphysics codes (DMCs), would revolutionize both UQ and optimization studies by breaking the curse of dimensionality that presently limits our “gradient-free” ensemble based computing approach. A similar breakthrough occurred in the AI/ML community once the ability to compute gradients of arbitrary loss functions using back-propagation became commonplace. Gradient information from the multiphysics codes can also be used to dramatically improve the efficiency and scale of training of ML/ROM surrogates for rapid assessments. • Achieving this in our suite of codes will be a grand challenge, similar to the amount of effort that was required to transition from CPU to GPU computing. It will require buy-in from the entire WSC/CP program and beyond, including all integrated codes, physics and engineering models, third-party library dependencies and performance portability abstractions. It will also require investment in research and development of numerical methods for computing adjoints of coupled physics across multiple adaptively refined moving meshes and of stochastic (Monte Carlo) and mesh free (SPH) methods. • New software and numerical techniques, largely pioneered by the AI/ML community, make this feasible. Chief among these is automatic differentiation (AD), the ability to employ AD at point-wise locations in a physics calculation (instead of traditional black-box approaches) and the ability to perform “back-propagation in time” (or reverse mode AD) for non-linear partial differential equations (PDEs). Fundamentally, the conclusion of this strategic planning exercise is that the time is right to undertake a large scale effort in WSC, centered on the existing integrated codes, to continue the natural evolution of mod/sim in the age of AI/ML. Instead of attempting to replace mod/sim with purely data driven AI/ML models, we believe the key to success is to integrate AI/ML by building on top of the decades of hard-won knowledge and the verified/validated multiphysics modelling capability that is the hallmark of the ASC program.

97 MATHEMATICS AND COMPUTING↗