Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “complex computing”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
DOE OSTI2020

Growing field of materials informatics: databases and artificial intelligence

The paradigm of molecular discovery in the chemical and pharmaceutical industry has followed a repetitive succession of screening and synthesis, involving the analysis of individual molecules that were both natural and produced. This ability to generate and screen libraries of compounds has found an echo in solid-state physics with the demand to explore and produce new materials for testing. In response to this demand, a golden age of materials discovery is being developed, with progress on important areas of both basic science and device applications. The confluence of theoretical and simulation methods, together with the availability of computation resources, has established the “materials genome” approach that is used by a growing number of research groups around the world with the goal of innovating on materials through systematic discovery. Here, an overview of this group of methodologies in tackling the ever-increasing complexity of computational materials science simulations is provided. Computational simulation is highlighted as a major component of rational design and synthesis of new materials with targeted properties, describing progress on databases and large data treatment. Tools for new materials discovery, including progress on the deployment of new data repositories, the implementation of high-throughput simulation approaches, and the development of artificial intelligence algorithms, are discussed.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Spin-free formulation of the multireference driven similarity renormalization group: A benchmark study of first-row diatomic molecules and spin-crossover energetics

We report a spin-free formulation of the multireference (MR) driven similarity renormalization group (DSRG) based on the ensemble normal ordering of Mukherjee and Kutzelnigg (1997). This ensemble averages over all microstates of a given total spin quantum number, and therefore, it is invariant with respect to SU(2) transformations. As such, all equations may be reformulated in terms of spin-free quantities and they closely resemble those of spin-adapted closed-shell coupled cluster (CC) theory. The current implementation is used to assess the accuracy of various truncated MR-DSRG methods (perturbation theory up to third order and iterative methods with single and double excitations) in computing the constants of 33 first-row diatomic molecules. The accuracy trends for these first-row diatomics are consistent with our previous benchmark on a small subset of closed-shell diatomic molecules. We then present the first MR-DSRG application on transition-metal complexes by computing the spin splittings of the [Fe(H 2 O) 6 ] 2+ and [Fe(NH 3 ) 6 ] 2+ molecules. A focal point analysis (FPA) shows that third-order perturbative corrections are essential to achieve reasonably converged energetics. The FPA based on the linearized MR-DSRG theory with one- and two-body operators and up to a quintuple- ζ basis set predicts the spin splittings of [Fe(H 2 O) 6 ] 2+ and [Fe(NH 3 ) 6 ] 2+ to be –35.7 and –17.1 kcal mol –1 , respectively, showing good agreement with the results of local CC theory with singles, doubles, and perturbative triples.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2025

Structure of the Majorana Clifford group

In quantum information science, Clifford operators and stabilizer codes play a central role for systems of qubits (or qudits). In this study, we study their analogs for systems composed of Majorana fermions. In this case, a crucial role is played by fermion parity symmetry, which is an unbreakable symmetry present in any system with fundamentally fermionic degrees of freedom. We prove that the subgroup of parity-preserving Majorana Cliffords can be represented by the orthogonal group over the binary field 𝔽 2 , and we show how it can be generated by braiding operators and used to construct any (even-parity) Majorana stabilizer code. We also analyze the frame potential for this so-called p-Clifford group when acting on a fixed-parity sector of the Hilbert space, proving that it is equivalent to the frame potential of the ordinary Clifford group acting on the same sector.

Computational complexity↗
DOE OSTI2022

Hybrid quantum-classical approach for coupled-cluster Green's function theory

The three key elements of a quantum simulation are state preparation, time evolution, and measurement. While the complexity scaling of time evolution and measurements are well known, many state preparation methods are strongly system-dependent and require prior knowledge of the system's eigenvalue spectrum. Here, we report on a quantum-classical implementation of the coupled-cluster Green's function (CCGF) method, which replaces explicit ground state preparation with the task of applying unitary operators to a simple product state. While our approach is broadly applicable to many models, we demonstrate it here for the Anderson impurity model (AIM). The method requires a number of T gates that grows as O ( N 5 ) per time step to calculate the impurity Green's function in the time domain, where N is the total number of energy levels in the AIM. Since the number of T gates is analogous to the computational time complexity of a classical simulation, we achieve an order of magnitude improvement over a classical CCGF calculation of the same order, which requires O ( N 6 ) computational resources per time step.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Using Numba for GPU acceleration of Neutron Beamline Digital Twins

Digital twins of neutron instruments using Monte Carlo ray tracing have proven to be useful in neutron data analysis and verifying instrument and sample designs. However, these simulations can become quite complex and computationally demanding with tens of billions of neutrons. In this paper, we present a GPU accelerated version of MCViNE using Python and Numba to balance user extensibility with performance. Numba is an open-source just-in-time (JIT) compiler for Python using LLVM to generate efficient machine code for CPUs and GPUs with NVIDIA CUDA. The JIT nature of Numba allowed complex instrument kernels to be generated easily. Initial simulations have shown a speedup between 200-1000x over the original CPU implementation. The performance gain with Numba enables more sophisticated data analysis and impacts neutron scattering science and instrument design.

Kendrick, Coleman↗
DOE OSTI2021

Revealing the nature of electron correlation in transition metal complexes with symmetry breaking and chemical intuition

Here, in this work, we provide a nuanced view of electron correlation in the context of transition metal complexes, reconciling computational characterization via spin and spatial symmetry breaking in single-reference methods with qualitative concepts from ligand-field and molecular orbital theories. These insights provide the tools to reliably diagnose the multi-reference character, and our analysis reveals that while strong (i.e., static) correlation can be found in linear molecules (e.g., diatomics) and weakly bound and antiferromagnetically coupled (monometal-noninnocent ligand or multi-metal) complexes, it is rarely found in the ground-states of mono-transition-metal complexes. This leads to a picture of static correlation that is no more complex for transition metals than it is, e.g., for organic biradicaloids. In contrast, the ability of organometallic species to form more complex interactions, involving both ligand-to-metal σ-donation and metal-to-ligand π-backdonation, places a larger burden on a theory's treatment of dynamic correlation. We hypothesize that chemical bonds in which inter-electron pair correlation is non-negligible cannot be adequately described by theories using MP2 correlation energies and indeed find large errors vs experiment for carbonyl-dissociation energies from double-hybrid density functionals. A theory's description of dynamic correlation (and to a less important extent, delocalization error), which affects relative spin-state energetics and thus spin symmetry breaking, is found to govern the efficacy of its use to diagnose static correlation.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

GAHLS: an optimized graph analytics based high level synthesis framework

The urgent need for low latency, high-compute and low power on-board intelligence in autonomous systems, cyber-physical systems, robotics, edge computing, evolvable computing, and complex data science calls for determining the optimal amount and type of specialized hardware together with reconfigurability capabilities. With these goals in mind, we propose a novel comprehensive graph analytics based high level synthesis (GAHLS) framework that efficiently analyzes complex high level programs through a combined compiler-based approach and graph theoretic optimization and synthesizes them into message passing domain-specific accelerators. This GAHLS framework first constructs a compiler-assisted dependency graph (CaDG) from low level virtual machine (LLVM) intermediate representation (IR) of high level programs and converts it into a hardware friendly description representation. Next, the GAHLS framework performs a memory design space exploration while account for the identified computational properties from the CaDG and optimizing the system performance for higher bandwidth. The GAHLS framework also performs a robust optimization to identify the CaDG subgraphs with similar computational structures and aggregate them into intelligent processing clusters in order to optimize the usage of underlying hardware resources. Finally, the GAHLS framework synthesizes this compressed specialized CaDG into processing elements while optimizing the system performance and area metrics. Evaluations of the GAHLS framework on several real-life applications (e.g., deep learning, brain machine interfaces) demonstrate that it provides 14.27× performance improvements compared to state-of-the-art approaches such as LegUp 6.2.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Generalized fractional approach to solving partial differential equations with arbitrary dispersion relations

The dynamics of physical systems are typically modeled by partial differential equations (PDEs). Because of the complexity in physical systems, PDE models rely on approximations or limiting cases. To model the full complexity of physical systems, it is necessary to use multiscale approaches where appropriate models are used at each level. Alternatively, complexity can be directly tackled by fractional integrodifferential equations that must be derived for each system. Here, we argue that complexity can be restored in PDEs by describing them from a fractional calculus perspective. Rather than deriving a fractional integrodifferential equation, we reinterpret the dispersion relation of the system by use of the Riesz definition, which contains the required information relating the energy and momentum space of the system and thus fully describes their dynamics. The approach is demonstrated by two examples: the Landau–Lifshitz equation in a 1D ferromagnetic chain and a modified KdV equation supporting surface gravity waves or Euler dispersion. The presented approach is applicable to fluids, soft matter, and solid-state matter and can be readily generalized to higher dimensions and more complex systems. While numerical calculations are needed to determine the fractional operator, the approach is analytical and can be utilized to determine analytical solutions and investigate nonlinear problems.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2022

Meso- to microscale modeling of atmospheric stability effects on wind turbine wake behavior in complex terrain

Abstract. Terrain-induced flow phenomena modulate wind turbine performance and wake behavior in ways that are not adequately accounted for in typical wind turbine wake and wind plant design models. In this work, we simulate flow over two parallel ridges with a wind turbine on one of the ridges, focusing on conditions observed during the Perdigão field campaign in 2017. Two case studies are selected to be representative of typical flow conditions at the site, including the effects of atmospheric stability: a stable case where a mountain wave occurs (as in ∼ 50 % of the nights observed) and a convective case where a recirculation zone forms in the lee of the ridge with the turbine (as occurred over 50 % of the time with upstream winds normal to the ridgeline). We use the Weather Research and Forecasting Model (WRF), dynamically downscaled from the mesoscale (6.75 km resolution) to microscale large-eddy simulation (LES) at 10 m resolution, where a generalized actuator disk (GAD) wind turbine parameterization is used to simulate turbine wakes. We compare the WRF–LES–GAD model results to data from meteorological towers, lidars, and a tethered lifting system, showing good qualitative and quantitative agreement for both case studies. Significantly, the wind turbine wake shows different amounts of vertical deflection from the terrain and persistence downstream in the two stability regimes. In the stable case, the wake follows the terrain along with the mountain wave and deflects downwards by nearly 100 m below hub height at four rotor diameters downstream. In the convective case, the wake deflects above the recirculation zone over 40 m above hub height at the same downstream distance. Overall, the WRF–LES–GAD model is able to capture the observed behavior of the wind turbine wakes, demonstrating the model's ability to represent wakes over complex terrain for two distinct and representative atmospheric stability classes, and, potentially, to improve wind turbine siting and operation in hilly landscapes.

17 WIND ENERGY↗
DOE OSTI2024

Developing And Scaling an OpenFOAM Model to Study Turbulent Flow in a HFIR Coolant Channel

Improving the understanding of how computational fluid dynamics (CFD) direct numerical simulations (DNS) of flows in the High Flux Isotope Reactor (HFIR) perform when run in parallel using the high performance computing (HPC) platform Summit at the Oak Ridge Leadership Computing Facility (OLCF) is of particular importance to boost the computational tools used to support HFIR conversion to low enriched fuel (LEU). Evaluation of scaling performance was driven by the increasing importance of graphics processing unit (GPU) usage in HPC, which is becoming the standard for modern supercomputers such as Summit. The desired results are to obtain a strong positive correlation between the computational resources dedicated to a problem and the relative speed-up of the simulation in comparison to a benchmark. This capability will allow substantially improvement in HFIR flow analytical capabilities, specifically when predicting turbulence properties at high Reynolds numbers. The study leverages previous simulation results performed with code PHASTA (finite element) on HPC platforms Cori (NERSC) and Theta (ALCF) [1] with computing options provided in the computing platform OpenFOAM (finite volume) at OLCF. Transitioning from PHASTA to OpenFOAM will (1) eliminate dependence on third-party software for mesh generation and manipulation, (2) reduce resource needs by employing modern architectures, and (3) build expertise for future modeling of HFIR-specific problems like heat transfer in involute geometry, entrance effects, flow structure in channel corners, and so on—all important issues when defining the available thermal margins in the transition to LEU. CPUs and GPUs differ significantly in their architecture and utilization, as discussed in the literature [2]. The most important differences are in the approach to computations and their memory. A single GPU contains a large quantity of cores, enabling it to perform with a much higher throughput than a CPU, but execution requires a different approach. GPU codes execute instructions using the Single-Instruction Multiple-Thread (SIMT) approach in which a single instruction is used for groups of threads called warps. A warp typically consists of 32 threads which must execute the same set of instructions, although on separate threads. Alternately, a CPU has far fewer cores that are much more flexible in their operation, excelling at quickly performing more complex serial computations. This is why GPUs have greater throughput when properly utilized. The second important difference is seen when comparing their memory spaces. Limited memory allocations and CPU–GPU communications cause a significant bottleneck in GPU-accelerated programs. Further study was required to properly take advantage of GPU resources. A comprehensive analysis of code performance and the model-specific features of turbulence constitutes the core of this work. In this study, a DNS simulation of HFIR channel turbulence was performed with the finite volume CFD code OpenFOAM v2112 and CUDA v11.0 on Red Hat Enterprise Linux v8.2. The OpenFOAM installation had AMGx integrated to enable GPU acceleration and utilizes the PETSc4FOAM library. The computational resources and the problem size were scaled on CPU and CPU + GPU architectures to gain a better understanding of the performance of a DNS problem on modern computing hardware. The study aimed to analyze the scaling of the code exclusively on CPUs and then to examine the scaling of the codes with GPU acceleration enabled. Scaling studies included CPU and GPU acceleration on a mesh of varying resolution to analyze the impact of problem size relative to computational resources. In the course of preparing the GPU configuration on Summit, mainly using the AMGX solvers, difficulties were encountered stemming from constant changes resulting from extensive ongoing development activities and the changing environment. This resulted in the inability to complete the GPU portion of the work. The code was compiled and tested, but production runs to assess acceleration were not performed because the used discretional compute time allocation expired as year-end approached. The Summit HPC platform is scheduled for decommissioning in 2024, making it unattractive for future use with Nvidia-based GPUs. Therefore, the work will be moved onto NERSC machines in FY24. An application was prepared and submitted, and sufficient node-hours were awarded to continue the research in the next calendar year. This report summarizes work performed thus far, which mostly focused on CPU OpenFOAM computing.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2022

Monitoring and prediction of porosity in laser powder bed fusion using physics-informed meltpool signatures and machine learning

In this work we accomplished the monitoring and prediction of porosity in laser powder bed fusion (LPBF) additive manufacturing process. This objective was realized by extracting physics-informed meltpool signatures from an in-situ dual-wavelength imaging pyrometer, and subsequently, analyzing these signatures via computationally tractable machine learning approaches. Porosity in LPBF occurs despite extensive optimization of processing conditions due to stochastic causes. Hence, it is essential to continually monitor the process with in-situ sensors for detecting and mitigating incipient pore formation. In this work a tall cuboid-shaped part (10 mm × 10 mm × 137 mm, material ATI 718Plus) was built with controlled porosity by varying laser power and scanning speed. This test caused various types of porosity, such as lack-of-fusion and keyhole formation, with varying degrees of severity in the part. The meltpool was continuously monitored using a dual-wavelength imaging pyrometer installed in the machine. Physically intuitive process signatures, such as meltpool length, temperature distribution, and ejecta (spatter) characteristics, were extracted from the meltpool images. Subsequently, relatively simple machine learning models, e.g., K-Nearest Neighbors, were trained to predict both the severity and type of porosity as a function of these physics-informed meltpool signatures. These models resulted in a prediction accuracy exceeding 95% (statistical F1-score). The same analysis was carried out with a complex, black-box deep learning convolutional neural network which directly used the meltpool images instead of physics-informed features. The convolutional neural network produced a comparable F1-score in the range of 89–97%. Finally, these results demonstrate that using pragmatic, physics-informed meltpool signatures within a simple machine learning model is as effective for flaw prediction in LPBF as using a complex and computationally demanding black-box deep learning model.

36 MATERIALS SCIENCE↗
DOE OSTI2022

Scalable multiscale modeling of platelets with 100 million particles

Here, we developed the core components of the AI-aided multiple time stepping algorithm for multiscale modeling of cell dynamics. This algorithm was implemented and analyzed on two supercomputer architectures with an application of simulating the aggregation of 250 platelets, or 102 million particles. To scale on these computers with complex memory and network architectures with GPUs, we devised a biomechanics-informed task mapping scheme to optimize load imbalance, communications, and memory utilization. Our simulations, scaling well up to 192 nodes on a Summit-like supercomputer with a peak speed of 11 petaflops, achieved a rate of 423 μs/day which is 500 times faster than the conventional algorithm using static time step and this has enabled studies of record size blood clots at record spatial–temporal resolutions. Additionally, we discovered the sensitive dependence of the scalability and execution time on the methods of decomposition, CPU–GPU coupling, and task mapping.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2021

Lowering tomography costs in quantum simulation with a symmetry projected operator basis

Measurement in quantum simulations provides a means for extracting meaningful information from a complex quantum state, and for quantum computing, reducing the complexity of measurement will be vital for near-term applications. For most quantum simulations, the targeted state will obey several symmetries inherent to the system Hamiltonian. Here, we obtain an alternative symmetry projected basis of measurement that reduces the number of measurements needed by a constant factor. Our scheme can be implemented at no additional cost on a quantum computer, can be implemented under various measurement or tomography schemes, and is reasonably resilient under noise.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Dynamical Complexity of Non-Gaussian Many-Body Systems with Dissipation

We characterize the dynamical state of many-body bosonic and fermionic many-body models with intersite Gaussian couplings, on-site non-Gaussian interactions, and local dissipation comprising incoherent particle loss, particle gain, and dephasing. We first establish that, for fermionic systems, if the dephasing noise is larger than the non-Gaussian interactions, irrespective of the Gaussian coupling strength, the system state is a convex combination of Gaussian states at all times. Furthermore, for bosonic systems, we show that if the particle loss and particle gain rates are larger than the Gaussian intersite couplings, the system remains in a separable state at all times. Building on this characterization, we establish that at noise rates above a threshold, there exists a classical algorithm that can efficiently sample from the system state of both the fermionic and bosonic models. Finally, we show that, unlike fermionic systems, bosonic systems can evolve into states that are not convex Gaussian even when the dissipation is much higher than the on-site non-Gaussianity. Similarly, unlike bosonic systems, fermionic systems can generate entanglement even with noise rates much larger than the intersite couplings.

Computational complexity↗
DOE OSTI2021

A Simple Criterion for Feasibility of Heat Integration between Distillation Streams Based on Relative Volatilities

In a multicomponent distillation configuration, there are numerous sources and sinks of heat, and a potential way to reduce the heat duty requirement is to perform heat integration. Unfortunately, an algorithmic search of the optimal heat integration opportunities is intractable when the required temperatures of intermediate streams are computed via complex models. Instead, in this work, we introduce pressure-scaled pseudo relative volatility, a new metric to compare stream temperatures. Here, we justify the use of pseudo relative volatility by proving that this variable is a monotonically increasing function of the liquid fraction in a saturated mixture stream. Using this metric, we derive a shortcut criterion to check the feasibility of various heat integration opportunities, such as thermal coupling via heat transfer (TCH). The advantage of this approach is that it circumvents the need for explicit temperatures and instead relies on composition, component relative volatilities, and pressure—quantities that are readily available in shortcut models for optimization of distillation configurations. Leveraging this fact, we propose a new optimization framework that identifies feasible TCHs that we consider within the formulation while minimizing the total heat duty of a distillation configuration. We demonstrate, on a few examples, that our formulation can identify heat duty efficient configurations, some of which are multieffect configurations. Using this methodology, we discover configurations that are not only simpler than the fully thermally coupled (FTC) configuration but also have a much lower heat duty.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2026

Covalency of M–N Bonds in Isomorphous Lanthanide and Actinide 5-(2-Pyridyl)-1H-tetrazolate Complexes

Experimental and computational analyses of [M(pdtz) 3 (H 2 O) 3 ]·3.5H 2 O (M 3+ = Pu 3+ −Cm 3+ , La 3+ −Nd 3+ , and Sm 3+ −Ho 3+ , pdtz− = 5-(2-pyridyl)-1H-tetrazolate) were conducted to understand potential differences in bonding between lanthanide and actinide complexes with a N-donor ligand. Structural analyses show that the An−N bond distances in the Pu 3+ , Am 3+ , and Cm 3+ complexes are within error of one another. Whereas in the lanthanide series, there is a nearly linear decrease in the Ln−N bond lengths from La 3+ to Ho 3+ (excluding Pm 3+ ). The An−N bond lengths are ∼0.015 Å shorter than their similarly-sized lanthanide analogs, in agreement with computational results that suggest greater covalent character in these bonds versus those with lanthanides. QTAIM analysis indicates that the An−N orbital mixing remains essentially unchanged from Pu 3+ to Cm 3+ , consistent with the nearly identical An−N bond lengths. However, upon deconvolution of the NLMOs into orbital compositions, the metal orbital contributions to An−N bonding decreases slightly overall wherein the 6d involvement remains constant, 7s involvement slightly increases, and 5f participation decreases. The molecular orbital energy diagram indicates that energy degeneracy between the 5f metal and 2p ligand orbitals increases from Pu 3+ to Cm 3+ and counteracts the contraction of the 5f orbtials. Together with prior reports of decreasing energy degeneracy between 5f and 3p orbitals from Np 3+ to Cf 3+ , these observations provide guidance on understanding how chemical bonding evolves in the actinide series.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

DownScaleBench for developing and applying a deep learning based urban climate downscaling- first results for high-resolution urban precipitation climatology over Austin, Texas

Abstract Cities need climate information to develop resilient infrastructure and for adaptation decisions. The information desired is at the order of magnitudes finer scales relative to what is typically available from climate analysis and future projections. Urban downscaling refers to developing such climate information at the city (order of 1 – 10 km) and neighborhood (order of 0.1 – 1 km) resolutions from coarser climate products. Developing these higher resolution (finer grid spacing) data needed for assessments typically covering multiyear climatology of past data and future projections is complex and computationally expensive for traditional physics-based dynamical models. In this study, we develop and adopt a novel approach for urban downscaling by generating a general-purpose operator using deep learning. This ‘DownScaleBench’ tool can aid the process of downscaling to any location. The DownScaleBench has been generalized for both in situ (ground- based) and satellite or reanalysis gridded data. The algorithm employs an iterative super-resolution convolutional neural network (Iterative SRCNN) over the city. We apply this for the development of a high-resolution gridded precipitation product (300 m) from a relatively coarse (10 km) satellite-based product (JAXA GsMAP). The high-resolution gridded precipitation datasets is compared against insitu observations for past heavy rain events over Austin, Texas, and shows marked improvement relative to the coarser datasets relative to cubic interpolation as a baseline. The creation of this Downscaling Bench has implications for generating high-resolution gridded urban meteorological datasets and aiding the planning process for climate-ready cities.

Singh, Manmeet (ORCID:0000000233747149)↗
DOE OSTI2025

Harnessing distributed GPU computing for generalizable graph convolutional networks in power grid reliability assessments

Although machine learning (ML) has emerged as a powerful tool for rapidly assessing grid contingencies, prior studies have largely considered a static grid topology in their analyses. This limits their application, since they need to be re-trained for every new topology. Here, this paper explores the development of generalizable graph convolutional network (GCN) models by pre-training them across a range of grid topologies and contingency types. We found that a GCN model with auto-regressive moving average (ARMA) layers with a line graph representation of the grid offered the best predictive performance in predicting voltage magnitudes (VM) and voltage angles (VA). We introduced the concept of phantom nodes to consider disparate grid topologies with a varying number of nodes and lines. For pre-training the GCN ARMA model across a variety of topologies, distributed graphics processing unit (GPU) computing afforded us significant training scalability. The predictive performance of this model on grid topologies that were part of the training data is substantially better than the direct current (DC) approximation. Although direct application of the pre-trained model to topologies that are not part of the grid is not particularly satisfactory, fine-tuning with small amounts of data from a specific topology of interest significantly improves predictive performance. In general, this paper highlights the feasibility of training large-scale GNN models to assess the reliability of power grids by considering a wide variety of grid topologies and contingency types. With the advent of foundational models in ML and the exponential increase in GPU computing clusters, generalizable ML models will significantly enhance how utilities manage power systems and make decisions in real-time or near-real-time.

24 - POWER TRANSMISSION AND DISTRIBUTION↗