SEARCH · Engineering Papers

Results for “ERRORS”

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.

SourceResearch connectionFrom yearThrough year

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

At least 829 records · Page 46

All sources NASA NTRS DOE OSTI

DOE OSTI2025

Enhanced MPM framework with multipatch isogeometric analysis for geotechnical applications

Achieving stable stress solutions at large strains using the Material Point Method (MPM) is challenging due to the accumulation of errors associated with geometry discretization, cell-crossing noise, and volumetric locking. Several simplified attempts exist in the literature to mitigate these errors, including higher-order frameworks. However, the stability of the MPM solution in such frameworks has been limited to simple geometries and the single-phase formulation (i.e., neglecting pore fluid). Although never explored, multipatch isogeometric analysis offers desirable qualities to simulate complex geometries while mitigating errors in the MPM. The degree of required high-order spatial integration has also never been investigated to infer a minimum limit for the stability of the stress solution in MPM. This paper presents a general-purpose numerical framework for simulating stable stresses in porous media, capturing both near incompressibility and multiphase interactions. First, the numerical framework is presented considering Non-Uniform Rational B-splines (NURBS) to perform isogeometric analysis (IGA) in MPM. Additionally, a volumetric strain smoothing algorithm is used to alleviate errors associated with volumetric locking. Second, the manifestation of cell-crossing errors is assessed via a series of problems with orders ranging from linear to cubic interpolation functions. Third, the use of NURBS is investigated and verified for problems with circular geometries. Finally, multipatch analysis is deployed to simulate plane strain and 3D penetration in soils, considering nearly incompressible elastoplastic (total stress) analysis and fully-coupled hydro-mechanical (effective stress) analysis. The stability of the solution is also analyzed for different constitutive models. From the results, it can be concluded that the framework using cubic interpolation functions with strain smoothing is the most convenient, presenting stable stress solutions for a broad range of multiphase geotechnical applications.

58 GEOSCIENCES↗

DOE OSTI2026

Short-term electricity load forecasting: Application-driven evaluation of machine learning models across spatial and temporal scales

As we transition towards a decarbonized economy, the integration of variable renewable energy resources and new demands (e.g., electric vehicles, heat pumps) into the electricity grid places unprecedented pressure on grid operators to effectively anticipate and manage peak load. In this context, machine learning algorithms are proving to be indispensable for accurate short-term load forecasting, a crucial task to address these challenges. This study benchmarks 6 machine learning algorithms, including three neural networks and three tree-based algorithms, across various levels of spatial aggregation and time horizons (1, 4, 8, 24, and 48 h). The central contribution of this work is the comparison and analysis of load forecasting models not only based on statistical metrics, but also based on a novel error metric, which evaluates the cost implications of forecast errors for power system stakeholders. Results show that tree-based models outperform neural networks, based on statistical metrics, and yield less skewed error distributions for most spatial scales. However, through the lens of the novel error metric, neural networks are the more competitive choice, especially for forecast horizons that exceed 8 h. The study concludes with actionable recommendations to grid operators and highlights the need for the development of error metrics that link forecasting accuracy to operational costs. To promote transparency and open science, the datasets and Python code are open-sourced via a supplementary repository.

Houben, Nikolaus↗

DOE OSTI2022

Design and performance of a variable gap system for thermal conductivity measurements of high temperature, corrosive, and reactive fluids

We report high-temperature fluids such as molten salts, liquid metals, and gasses are being proposed for many advanced energy systems including thermal energy storage devices, concentrating solar plants, and advanced nuclear reactor designs. However, the chemical behavior and thermophysical properties of many of these fluids have not been well characterized, which hinders the design, modeling, safety analysis, and deployment of these systems. Thermal conductivity is a property that is especially limited by existing measurement capabilities, which are subject to errors caused by convection, material interaction, radiative heat transfer, and instrument degradation. Therefore, there is a lack of standard, systematic measurement techniques for high-temperature, reactive, and corrosive fluids. In this work, the development of a variable gap thermal conductivity measurement system is detailed. The system is designed to measure the thermal conductivity of highly corrosive and reactive fluids, and survive operation between 100 °C and 800 °C. The effects of convection are minimized by limiting the thickness of the specimen to thin sizes (<0.3 mm). Corrections for radiative heat transfer were included in the working equations to consider specimens with varying optical properties. The design, construction, instrumentation, operating principles, and data analysis techniques are discussed in detail. The system was tested up to 500 °C using helium gas and molten KNO 3 -NaNO 3 to verify the measurement technique and determine the sources of error. At 300 and 400 °C KNO 3 -NaNO 3 , results showed maximum relative error of 6% when compared to results in the literature. The helium results were within 13% of those in the literature at 300 and 400 °C. Higher errors were observed at 500 °C for both fluids, and the sources of these errors are discussed.

42 ENGINEERING↗

DOE OSTI2020

An analysis of the spatio-temporal resolution of the immersed boundary method with direct forcing

The immersed boundary method (IBM) with direct forcing is very popular in the simulation of rigid particulate flows. In the IBM, an interaction force is introduced at the interface between fluid and particle in order to approximate the no-slip boundary condition. The interaction force is calculated through dividing the velocity difference (or error) between fluid and particle at the interface by the time step. Here, a dynamic equation for the velocity difference is derived. Additionally, analyses on the dynamic equation provide a few new findings: (i) The interaction force is the solution of a least-squares error problem, with the direct implication that the Lagrangian marker distribution has no effect on the large scale flow structure once the distribution of Lagrangian markers become saturated along the interface (i.e., each marker remains properly correlated with all its neighbors); (ii) The Lagrangian volume-weight is a relaxation factor to control how fast the velocity error decays to the ideal value of zero; (iii) The optimal choice of the Lagrangian volume-weight is the largest value permissible by a stability condition. A comprehensive convergence analysis with regard to the spatial and temporal resolution is presented for the velocity error and also for the shear-stress and surface pressure. In three simple canonical problems, it is analytically and numerically shown that the IBM results converge to the theoretical solutions obtained with precise imposition of no-slip and no-penetration boundary conditions. It is observed that it is not necessary to match the Lagrangian marker volume-weight to that of the local Eulerian cell volume and in fact this matching leads to lower than optimal computational efficiency. However, it is found that extremely high Eulerian grid resolution and small time step have to be used to obtain high precision simulation results. Especially, the time step should be inversely proportional to the particle Reynolds number for low Reynolds number flows. For high frequency oscillation problems, the grid size needs to be reduced by a factor of the square root of the frequency, and the time step to be reduced by a factor of the frequency. The theoretical findings here can be used to alleviate the technical difficulties in simulating non-spherical particles by not requiring the Lagrangian marker distribution to match the Eulerian grids and also in the implementation of IBM on non-uniform Eulerian grids. The present work also provides simple practical guidance on the choice of temporal and spatial resolution so as to control the simulation error a priori.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2020

NSFnets (Navier-Stokes flow nets): Physics-informed neural networks for the incompressible Navier-Stokes equations

In the last 50 years there has been a tremendous progress in solving numerically the Navier-Stokes equations using finite differences, finite elements, spectral, and even meshless methods. Yet, in many real cases, we still cannot incorporate seamlessly (multi-fidelity) data into existing algorithms, and for industrial-complexity applications the mesh generation is time consuming and still an art. Moreover, solving ill-posed problems (e.g., lacking boundary conditions) or inverse problems is often prohibitively expensive and requires different formulations and new computer codes. Here, we employ physics-informed neural networks (PINNs), encoding the governing equations directly into the deep neural network via automatic differentiation, to overcome some of the aforementioned limitations for simulating incompressible laminar and turbulent flows. We develop the Navier-Stokes flow nets (NSFnets) by considering two different mathematical formulations of the Navier-Stokes equations: the velocity-pressure (VP) formulation and the vorticity-velocity (VV) formulation. Since this is a new approach, we first select some standard benchmark problems to assess the accuracy, convergence rate, computational cost and flexibility of NSFnets; analytical solutions and direct numerical simulation (DNS) databases provide proper initial and boundary conditions for the NSFnet simulations. The spatial and temporal coordinates are the inputs of the NSFnets, while the instantaneous velocity and pressure fields are the outputs for the VP-NSFnet, and the instantaneous velocity and vorticity fields are the outputs for the VV-NSFnet. This is unsupervised learning and, hence, no labeled data are required beyond boundary and initial conditions and the fluid properties. The residuals of the VP or VV governing equations, together with the initial and boundary conditions, are embedded into the loss function of the NSFnets. No data is provided for the pressure to the VP-NSFnet, which is a hidden state and is obtained via the incompressibility constraint without extra computational cost. Unlike the traditional numerical methods, NSFnets inherit the properties of neural networks (NNs), hence the total error is composed of the approximation, the optimization, and the generalization errors. Here, we empirically attempt to quantify these errors by varying the sampling (“residual”) points, the iterative solvers, and the size of the NN architecture. For the laminar flow solutions, we show that both the VP and the VV formulations are comparable in accuracy but their best performance corresponds to different NN architectures. The initial convergence rate is fast but the error eventually saturates to a plateau due to the dominance of the optimization error. For the turbulent channel flow, we show that NSFnets can sustain turbulence at , but due to expensive training we only consider part of the channel domain and enforce velocity boundary conditions on the subdomain boundaries provided by the DNS data base. We also perform a systematic study on the weights used in the loss function for balancing the data and physics components, and investigate a new way of computing the weights dynamically to accelerate training and enhance accuracy. In the last part, we demonstrate how NSFnets should be used in practice, namely for ill-posed problems with incomplete or noisy boundary conditions as well as for inverse problems. We obtain reasonably accurate solutions for such cases as well without the need to change the NSFnets and at the same computational cost as in the forward well-posed problems. As a result, we also present a simple example of transfer learning that will aid in accelerating the training of NSFnets for different parameter settings.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2026

Feedforward-feedback ammonia control at a water resource recovery facility based on a digital twin with hybrid model

Ammonia-based aeration control (ABAC) at full-scale Water Resource Recovery Facilities (WRRFs) can be challenged by diurnal loading and transport delays. This work addressed these challenges using a hybrid feedforward–feedback controller built on Activated Sludge Model 1 (ASM1), marking the first full-scale deployment to pair a mechanistic feedforward core with data-driven corrections. The objectives were to improve ammonia setpoint tracking, assess performance of the mechanistic model when enhanced with data-driven corrections, and document full-scale operation. The hybrid model incorporates two data-driven components: (1) a Mechanistic Error Forecasting Engine (MEFE), consisting of a multivariate linear regressor and a long short-term memory (LSTM) ensemble. Defying expectations, low-parameter models outperformed more complex alternatives, reducing the mechanistic error by 71%. (2) A Residual Oscillation Forecasting Engine (ROFE), based on Fast Fourier Transform, reduced the remaining error by another 35%. Two proportional–integral (PI) feedback loops further (i) trim the feedforward output and (ii) eliminate residual controller error in the final aerobic zone. In full-scale operation, the controller reduced mean-squared error (MSE) by 94% over the baseline and produced more stable dissolved oxygen (DO) setpoints. Overall, it was proven that layering multi-timescale data-driven models on a mechanistic core can yield reliable ABAC performance at WRRFs.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2020

A non-intrusive optical (NIO) approach to characterize heliostats in utility-scale power tower plants: Sensitivity study

Power towers, a type of concentrating power tower technology, use a large number of heliostats to concentrate sunlight and produce renewable energy. Optical errors of heliostats can cause drastic losses in the efficiency of power-tower plants. Accurately measuring optical errors is crucial to assessing and improving plant performance. This analysis discusses an innovate non-intrusive optical (NIO) approach to measure slope, canting, and tracking errors by detecting distortions in the reflected tower structure in heliostat images. Further in this work a sensitivity study is carefully conducted to determine the uncertainty requirements to allow the method to calculate slope errors with an accuracy of 0.25 mrad. Measurement uncertainty sources include camera resolution and position uncertainty, tower position uncertainty, and number of collected images. Each uncertainty source is investigated to determine its impact on the accuracy of the slope-error calculation. A combination of theoretical results and experimental results obtained from data collected on a heliostat at Sandia National Laboratories is used to determine and validate uncertainty requirements. The analysis shows that a measurement uncertainty of 0.25 mrad can be realized by realistically controlling uncertainty sources when implementing the NIO method. It demonstrates the superior performance of NIO in performing in-situ optical characterization.

14 SOLAR ENERGY↗

DOE OSTI2026

ARC physics basis–magnetohydrodynamics

ARC is designed to produce 400 ⁢MW of net electricity and prove the commercial feasibility of a fusion power plant. In order to achieve this goal ARC has to operate with optimal core performance in a stationary scenario that minimises wear on the first wall and divertor. This requires avoiding or mitigating magnetohydrodynamic (MHD) instabilities which have the potential to not only degrade the plasma core but also lead to deleterious transient heat loads on plasma facing components. Therefore, this work aims at characterising the MHD stability of the high performance ARC scenario and inform the design of error field correction coils. Firstly, simulations of vertical displacement events show that an in-vessel coil is not needed and instead the poloidal shaping coils can be used to control vertical stability. These simulations also inform the demands on the corresponding coil power supplies. Stability analysis of the ideal kink mode with or without a conducting wall and kinetic effects suggests that the ARC baseline scenario operates deeply in the stable region. Using RDCON, tearing modes at the 𝑚/𝑛 =2/1 and 3/2 surfaces (with poloidal mode number 𝑚, and toroidal mode number 𝑛) are shown to be linearly stable, and including thermal transport effects in the rational surfaces lead to further stabilisation. However, other transient plasma instabilities can seed neoclassical tearing modes (NTMs). The marginally stable width of NTMs in ARC strongly depends on the internal inductance and can fall below 0.1% of the normalised poloidal flux. Furthermore, an empirical cross-machine model of the 𝑛 =1 error field leading to a disruption predicts a critical error field larger than SPARC but smaller than ITER. Three-dimensional coils can be designed with the Generalised Purturbed Equilbium Code based on a simple model that calculates the maximum correctable error field that is limited by the neoclassical toroidal viscosity torque. Broad scans of different coil geometries identify a set of 2 rows of off-midplane coils to be a suitable solution. It is also determined that such a set of three-dimensional coils is capable of correcting 𝑛 =2 error fields to some degree and creating strong enough 𝑛 =2 or 𝑛 =3 edge resonant perturbation fields for the suppression of edge-localised modes at reasonable coil currents. The final design of the first ARC will be further informed by results from SPARC.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2020

Benchmark calculations of radiative forcing by greenhouse gases

Changes in concentrations of greenhouse gases lead to changes in radiative fluxes throughout the atmosphere. The value of this change, the instantaneous radiative forcing, varies across climate models, due partly to differences in the distribution of clouds, humidity, and temperature across models, and partly due to errors introduced by approximate treatments of radiative transfer. This paper describes an experiment within the Radiative Forcing Model Intercomparision Project that uses benchmark calculations made with line–by–line models to identify parameterization error in the representation of absorption and emission by greenhouse gases. Clear–sky instantaneous forcing by greenhouse gases is computed using a set of 100 profiles, selected from a re–analysis of present–day conditions, that represent the global annual mean forcing from pre–industrial times to the present–day with sampling errors of less than 0.01 W m –2 . Six contributing line–by–line models agree in their estimate of this forcing to within 0.025 W m –2 while even recently–developed parameterizations have typical errors four or more times larger, suggesting both that the samples reveal true differences among line–by–line models and that parameterization error will be readily identifiable. Agreement among line–by–line models is better in the longwave than in the shortwave where differing treatments of the water vapor vapor continuum affect estimates of forcing by carbon dioxide and methane. The impacts of clouds on instantaneous radiative forcing are estimated from climate model simulations. Here, the adjustment due to stratospheric temperature change by assuming fixed dynamical heating. Adjustments are large only for ozone and for carbon dioxide, for which stratospheric cooling introduces modest non–linearity.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2025

How Representative Are Uncrewed Aircraft System Measurements of the Convective Boundary Layer?

Abstract Uncrewed aircraft systems (UAS) demonstrate significant potential for filling data gaps in the atmospheric boundary layer. However, the extent to which UAS observations—typically vertical profiles taken over 15 min—are representative of the boundary layer as a whole remains poorly characterized. Using large eddy simulations (LES) of the daytime convective boundary layer (CBL), we quantify random errors in UAS measurements that occur due to insufficient statistical convergence of the time average to the true ensemble mean. Random errors in first‐order moments increase as the CBL becomes increasingly unstable, and are largest near the surface for most quantities. Errors are on the order of 2–6 m for wind speed, 15–60 for wind direction, 0.2–3 K for potential temperature, and 0.1–1 g for specific humidity, with errors in turbulent fluxes on the order of 50%–100%. Sampling strategies that mitigate random errors are discussed in light of our results.

Greene, Brian R. [Now at Verisk Extreme Event Solu↗

DOE OSTI2023

Logical quantum processor based on reconfigurable atom arrays

Suppressing errors is the central challenge for useful quantum computing, requiring quantum error correction (QEC) for large-scale processing. However, the overhead in the realization of error-corrected ‘logical’ qubits, in which information is encoded across many physical qubits for redundancy, poses substantial challenges to large-scale logical quantum computing. Here we report the realization of a programmable quantum processor based on encoded logical qubits operating with up to 280 physical qubits. Using logical-level control and a zoned architecture in reconfigurable neutral-atom arrays, our system combines high two-qubit gate fidelities, arbitrary connectivity, as well as fully programmable single-qubit rotations and mid-circuit readout. Operating this logical processor with various types of encoding, we demonstrate improvement of a two-qubit logic gate by scaling surface-code distance from d = 3 to d = 7, preparation of colour-code qubits with break-even fidelities, fault-tolerant creation of logical Greenberger–Horne–Zeilinger (GHZ) states and feedforward entanglement teleportation, as well as operation of 40 colour-code qubits. Finally, using 3D [[8,3,2]] code blocks, we realize computationally complex sampling circuits with up to 48 logical qubits entangled with hypercube connectivity with 228 logical two-qubit gates and 48 logical CCZ gates. We find that this logical encoding substantially improves algorithmic performance with error detection, outperforming physical-qubit fidelities at both cross-entropy benchmarking and quantum simulations of fast scrambling. These results herald the advent of early error-corrected quantum computation and chart a path towards large-scale logical processors.

74 ATOMIC AND MOLECULAR PHYSICS↗

DOE OSTI2020

Hybridizing pseudo-Hamiltonians and non-local pseudopotentials in diffusion Monte Carlo

An accurate treatment of effective core potentials (ECPs) requires care in continuum quantum Monte Carlo (QMC) methods. While most QMC studies have settled on the use of familiar non-local (NL) pseudopotentials with additional localization approximations, these approaches have been shown to result in moderate residual errors for some classes of molecular and solid state applications. Here, we revisit an idea proposed early in the history of QMC ECPs that does not require localization approximations, namely, a differential class of potentials referred to as pseudo-Hamiltonians. We propose to hybridize NL potentials and pseudo-Hamiltonians to reduce residual non-locality of existing potentials. We derive an approach to recast pseudopotentials for 3d elements as hybrid pseudo-Hamiltonians with optimally reduced NL energy. We demonstrate the fidelity of the hybrid potentials by studying atomic ionization potentials of Ti and Fe and the binding properties of TiO and FeO molecules with diffusion Monte Carlo (DMC). We show that localization errors have been reduced relative to potentials with the same NL channels for Sc–Zn by considering the DMC energy change with respect to the choice of approximate localization. While localization error decreases proportionate to the reduced NL energy without a Jastrow, with a Jastrow, the degree of reduction decreases at higher filling of the d-shell. Our results suggest that a subset of existing ECPs may be recast in this hybrid form to reduce the DMC localization error. They also point to the prospect of further reducing this error by generating ECPs within this hybrid form from the start.

36 MATERIALS SCIENCE↗

DOE OSTI2024

Hyperfine interactions for small systems including transition-metal elements using self-interaction corrected density-functional theory

The interactions between the electronic magnetic moment and the nuclear spin moment, i.e., magnetic hyperfine (HF) interactions, play an important role in understanding electronic properties of magnetic systems and in realizing platforms for quantum information science applications. We investigate the HF interactions for atomic systems and small molecules, including Ti or Mn, by using Fermi–Löwdin orbital (FLO) based self-interaction corrected (SIC) density-functional theory. We calculate the Fermi contact (FC) and spin-dipole terms for the systems within the local density approximation (LDA) in the FLO-SIC method and compare them with the corresponding values without SIC within the LDA and generalized-gradient approximation (GGA), as well as experimental data. For the moderately heavy atomic systems (atomic number Z ≤ 25), we find that the mean absolute error of the FLO-SIC FC term is about 27 MHz (percentage error is 6.4%), while that of the LDA and GGA results is almost double that. Therefore, in this case, the FLO-SIC results are in better agreement with the experimental data. For the non-transition-metal molecules, the FLO-SIC FC term has the mean absolute error of 68 MHz, which is comparable to both the LDA and GGA results without SIC. For the seven transition-metal-based molecules, the FLO-SIC mean absolute error is 59 MHz, whereas the corresponding LDA and GGA errors are 101 and 82 MHz, respectively. Therefore, for the transition-metal-based molecules, the FLO-SIC FC term agrees better with experiment than the LDA and GGA results. We observe that the FC term from the FLO-SIC calculation is not necessarily larger than that from the LDA or GGA for all the considered systems due to the core spin polarization, in contrast to the expectation that SIC would increase the spin density near atomic nuclei, leading to larger FC terms.

Chemistry↗

DOE OSTI2025

Resolving discrepancies in bang-time predictions for indirect-drive ICF experiments on the NIF: Insights from the Build-A-Hohlraum campaign

This study investigated discrepancies between measured and simulated x-ray drive in Indirect-Drive Inertial Confinement Fusion (ID-ICF) hohlraums at the National Ignition Facility. Despite advances in radiation-hydrodynamic simulations, a consistent “drive deficit” remains. Experimentally measured ID-ICF capsule bang-times are systematically 400–700 ps later than simulations predict. The Build-A-Hohlraum (BAH) campaign explored potential causes for this discrepancy by systematically varying hohlraum features, including laser entrance hole (LEH) windows, capsules, and gas fills. Overall, the agreement between simulated and experimental x-ray drive was found to be largely unaffected by these changes. The data allow us to exclude some hypotheses put forward to potentially explain the discrepancy. Errors in the local thermodynamic equilibrium (LTE) atomic modeling, errors in the modeling of LEH closure, and errors due to a lack of plasma species mix physics in simulations are shown to be inconsistent with our measurements. Instead, the data support the hypothesis that errors in NLTE emission modeling are a significant contributor to the discrepancy. X-ray emission in the 2–4 keV range is found to be approximately 30% lower than in simulations. This is accompanied by higher than predicted electron temperatures in the gold bubble region, pointing to errors in non-LTE modeling. Introducing an opacity multiplier of 0.87 on energy groups above 1.8 keV improves agreement with experimental data, reducing the bang-time discrepancy from 300 to 100 ps. These results underscore the need for refined NLTE opacity models to enhance the predictive power of hohlraum simulations.

Band emission↗

DOE OSTI2026

Deployment of Traditional and Hybrid Machine Learning for Critical Heat Flux Prediction in the CTF Thermal-Hydraulics Code

Critical heat flux (CHF) marks the transition from nucleate to film boiling, where heat transfer to the working fluid can rapidly deteriorate. Accurate CHF prediction is essential for efficiency, safety, and preventing equipment damage, particularly in nuclear reactors. Although widely used, empirical correlations frequently exhibit discrepancies when compared to experimental data, limiting their reliability in diverse operational conditions. Traditional machine learning (ML) approaches have demonstrated potential for CHF prediction but often suffer from limited interpretability, data scarcity, and insufficient knowledge of physical principles. Hybrid model approaches, which combine data-driven ML with base models, mitigate these concerns by incorporating prior knowledge of the domain. This study integrates an externally trained purely data-driven ML model and two hybrid models (using the Biasi and Bowring CHF correlations) within the CTF subchannel code via a custom Fortran framework. Performance was evaluated using two validation cases: a subset of the Nuclear Regulatory Commission (NRC) CHF database and the Bennett dryout experiments. In both cases, the hybrid models demonstrated significantly lower error metrics compared to conventional empirical correlations, with the best models often reducing relative error by about 5 percentage points. The pure ML model achieved comparable accuracy, outperforming the hybrid Biasi model in the NRC test case (3.3% versus 5.5% relative error) but exhibiting slightly higher error against the hybrid Bowring model in the Bennett test case (7.7% versus 6.1%). Trend analysis of error parity indicated that ML-based models reduced the tendency for CHF overprediction, improving overall accuracy. These results demonstrate that ML-based CHF models can be effectively integrated into subchannel codes and could potentially increase performance compared to conventional methods.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2023

Deep Learning for Multigroup Cross-Section Representation in Two-Step Core Calculations

Here we investigate using deep learning, a type of machine-learning algorithm employing multiple layers of artificial neurons, for the mathematical representation of multigroup cross sections for use in the Griffin reactor multiphysics code for two-step deterministic neutronics calculations. A three-dimensional fuel element typical of a high-temperature gas reactor as well as a two-dimensional sodium-cooled fast reactor lattice are modeled using the Serpent Monte Carlo code, and multigroup macroscopic cross sections are generated for various state parameters to produce a training data set and a separate validation data set. A fully connected, feedforward neural network is trained using the open-source PyTorch machine-learning framework, and its accuracy is compared against the standard piecewise linear interpolation model. Additionally, we provide in this work a generic technique for propagating the cross-section model errors up to the k eff using sensitivity coefficients with the first-order uncertainty propagation rule. Quantifying the eigenvalue error due to the cross-section regression errors is especially practical for appropriately selecting the mathematical representation of the cross sections. We demonstrate that the artificial neural network model produces lower errors and therefore enables better accuracy relative to the piecewise linear model when the cross sections exhibit nonlinear dependencies; especially when a coarse grid is employed, where the errors can be halved by the artificial neural network. However, for linearly dependent multigroup cross sections as found for the sodium-cooled fast reactor case, a simpler linear regression outperforms deeper networks.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

DOE OSTI2021

Parametric dependencies of locked mode thresholds in KSTAR L-mode plasmas

Resonant error fields with toroidal harmonics n = 1 and n = 2 in otherwise axisymmetric tokamak plasmas can drive disruptive locked modes when their strength exceeds a threshold value. This paper presents the sensitivity of error field threshold in global parameters such as line-averaged density and toroidal magnetic field as well as toroidal harmonics dependence in KSTAR L-mode plasmas. The error field threshold for mode-locking increases along with density up to a roll-over density, which coincides well with the linear ohmic confinement (LOC) to saturated ohmic confinement (SOC) density for the studied KSTAR plasmas. As a result, this non-monotonic density correlation of the error field threshold in LOC–SOC can be explained by a modification of existing theoretical scaling in SOC, while plasma rotation can also play an important role in setting the resonant error field threshold.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2025

Parameter uncertainties for imperfect surrogate models in the low-noise regime

Abstract Bayesian regression determines model parameters by minimizing the expected loss, an upper bound to the true generalization error. However, this loss ignores model form error, or misspecification, meaning parameter uncertainties are significantly underestimated and vanish in the large data limit. As misspecification is the main source of uncertainty for surrogate models of low-noise calculations, such as those arising in atomistic simulation, predictive uncertainties are systematically underestimated. We analyze the true generalization error of misspecified, near-deterministic surrogate models, a regime of broad relevance in science and engineering. We show that posterior parameter distributions must cover every training point to avoid a divergence in the generalization error and design a compatible ansatz which incurs minimal overhead for linear models. The approach is demonstrated on model problems before application to thousand-dimensional datasets in atomistic machine learning. Our efficient misspecification-aware scheme gives accurate prediction and bounding of test errors in terms of parameter uncertainties, allowing this important source of uncertainty to be incorporated in multi-scale computational workflows.

Swinburne, Thomas D. (ORCID:0000000232554257)↗