Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Conditional Normalizing Flow”

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 19 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Generative machine learning for detector response modeling with a conditional normalizing flow

In this paper, we explore the potential of generative machine learning models as an alternative to the computationally expensive Monte Carlo (MC) simulations commonly used by the Large Hadron Collider (LHC) experiments. Our objective is to develop a generative model capable of efficiently simulating detector responses for specific particle observables, focusing on the correlations between detector responses of different particles in the same event and accommodating asymmetric detector responses. Here, we present a conditional normalizing flow model ($\mathcal{CNF}$) based on a chain of Masked Autoregressive Flows, which effectively incorporates conditional variables and models high-dimensional density distributions. We assess the performance of the $\mathcal{CNF}$ model using a simulated sample of Higgs boson decaying to diphoton events at the LHC. We create reconstruction-level observables using a smearing technique. We show that conditional normalizing flows can accurately model complex detector responses and their correlation. This method can potentially reduce the computational burden associated with generating large numbers of simulated events while ensuring that the generated events meet the requirements for data analyses. We make our code available at https://github.com/allixu/normalizing_flow_for_detector_response

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗
DOE OSTI2024

Active learning for the design of polycrystalline textures using conditional normalizing flows

Generative modeling has opened new avenues for solving previously intractable materials design problems. However, these new opportunities are accompanied by a drastic increase in the required amount of training data. This is in stark juxtaposition to the high expense and difficulty in curating such large materials datasets. In this work, we propose a novel framework for integrating generative models within an active learning loop. Further, this enables the training of generative models with datasets significantly smaller than what has previously been demonstrated, providing a direct route for their application in data constrained environments. The functionality of this framework is then demonstrated by addressing the challenge of designing polycrystalline textures associated with target anisotropic mechanical properties. The developed protocol exhibited a cost reduction between 14 to 18 times over a randomly sampled experimental design.

36 MATERIALS SCIENCE↗
DOE OSTI2025

ForestFlow: predicting the Lyman-α forest clustering from linear to nonlinear scales

On large scales, the Lyman-α forest provides insights into the expansion history of the Universe, while on small scales, it imposes strict constraints on the growth history, the nature of dark matter, and the sum of neutrino masses. This work introduces ForestFlow, a novel framework that bridges the gap between large- and small-scale analyses, which have traditionally relied on distinct modeling approaches. Using conditional normalizing flows, ForestFlow predicts the two Lyman-α linear biases (b δ and b η ) and six parameters describing small-scale deviations of the three-dimensional flux power spectrum (P 3D ) from linear theory as a function of cosmology and intergalactic medium physics. These are then combined with a Boltzmann solver to make consistent predictions, from arbitrarily large scales down to the nonlinear regime, for P 3D and any other statistics derived from it. Trained on a suite of 30 fixed-and-paired cosmological hydrodynamical simulations spanning redshifts from z = 2 to 4.5, ForestFlow achieves 3 and 1.5% precision in describing P 3D and the one-dimensional flux power spectrum (P 1D ) from linear scales to k = 5 Mpc −1 and k ∥ = 4 Mpc −1 , respectively. Thanks to its conditional parameterization, ForestFlow shows similar performance for ionization histories and two ΛCDM model extensions – massive neutrinos and curvature – even though none of these are included in the training set. This framework will enable full-scale cosmological analyses of Lyman-α forest measurements from the DESI survey.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

inverse-cnf

Inverse machine learning model using Conditional Normalizing Flow (CNF)

DeBardeleben, Nathan Andrew [Los Alamos National L↗
DOE OSTI2025

Conditional Pseudo-Reversible Normalizing Flow for Surrogate Modeling in Quantifying Uncertainty Propagation

We introduce a conditional pseudo-reversible normalizing flow (PR-NF) that directly learns conditional probability distributions from noisy physical models to efficiently quantify both forward and inverse uncertainty propagation. Traditional surrogate modeling approaches approximate only the deterministic component of physical models, requiring separate noise characterization and computationally expensive sampling methods for inverse problems. Here, in this work, we develop the conditional PR-NF model to directly learn and efficiently generate samples from the conditional probability density functions (PDFs). The training process utilizes dataset consisting of input-output pairs without requiring prior knowledge about the noise and the function. Once trained, our model efficiently generates samples from conditional PDFs for any input within the training domain. Moreover, the pseudo-reversibility feature allows for the use of fully connected neural network architectures, which simplifies the implementation and enables theoretical analysis. We provide a rigorous convergence analysis of the conditional PR-NF model, showing its ability to converge to the target conditional PDF using the Kullback−Leibler divergence. To demonstrate the effectiveness of our method, we apply it to several benchmark tests and a real-world geologic carbon storage problem.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

The influence of temperature history and flow mixing on the vapor-phase speciation of uranium oxide nanoparticles

The transport of radioactive material following a nuclear event depends heavily on fallout characteristics (e.g., composition, shape, and size), which in turn are controlled by the conditions under which the fallout forms. To this end, we use a novel plasma flow reactor (PFR) to explore how conditions including temperature history and gas mixing influence the chemical speciation and physical characteristics of uranium oxide nanoparticles. The PFR consists of a glass tube attached to an inductively coupled plasma (ICP) torch via an adaptor piece (ring flow injector) located downstream of the plasma-generating RF coil. We increase the gas flow rates from normal ICP operating conditions to higher values both upstream at the RF coil and downstream at the ring flow injector to create four distinct temperature profiles and two distinct gas flow mixing conditions (normal and higher Ar atomic ratios). Ex situ transmission electron microscopy (TEM) analysis is used to determine the crystal structure, morphology, and size distribution of the synthesized particles as a function of temperature and mixing. The particles display a striking dependence on these conditions in terms of the resulting speciation of uranium oxide: lower temperature (higher upstream) and slower cooling (normal downstream with normal Ar atomic ratio mixing) favor the formation of α-UO 3 particles, while faster cooling (higher downstream with higher Ar atomic ratio mixing) and higher temperature (normal upstream) favor the formation of UO 2 particles. The extent of gas mixing dictates particle aggregation: better mixing encourages the particles to aggregate more extensively. Here these results demonstrate the sensitivity of uranium to initial local conditions and can be used to improve our understanding of the kinetics that drive uranium speciation and ultimately inform fallout transport models.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2019

Climate variability and fill rate impacts on downstream flows from the Grand Ethiopian Renaissance Dam [Sheets]

Conclusions of this report include: Flow conditions (wet, normal, dry) are strong determinant of success in filling the dam; Some improvement for longer fill periods; Downstream Flow Compact only met under wetter flow conditions, rare otherwise; Results generated using other inflow datasets (not shown) indicate selection of data important for filling dam and highlight difficulty in determining which one to trust.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2026

Monte Carlo Event Generation with Continuous Normalizing Flows

We apply continuous normalizing flows trained with the flow matching method to the problem of phase-space sampling in Monte Carlo event generation for high-energy collider physics. Focusing on lepton-pair and top-quark pair production with multiple jets, the two computationally most expensive processes at the Large Hadron Collider, we train helicity-conditioned continuous normalizing flows to remap the random numbers used in matrix element evaluation. Compared to standard methods, we achieve unweighting efficiency improvements by factors of up to 184 and 25 for the two processes at their respective highest jet number, at the cost of an increased evaluation time. When combining the advantages of continuous normalizing flows with the fast evaluation times of coupling-layer-based flows, using the RegFlow approach, we find parton-level unweighted event generation walltime gains of about a factor of 10 at the highest jet numbers. These substantial gains highlight the promise of samplers based on machine learning for next-generation collider experiments.

Bothmann, Enrico [CERN; Gottingen U.] (ORCID:00000↗
DOE OSTI2014

Bypass flow computations on the LOFA transient in a VHTR

Bypass flow in the prismatic gas-cooled very high temperature reactor (VHTR) is not intentionally designed to occur, but is present in the gaps between graphite blocks. Previous studies of the bypass flow in the core indicated that the cooling provided by flow in the bypass gaps had a significant effect on temperature and flow distributions for normal operating conditions. However, the flow and heat transports in the core are changed significantly after a Loss of Flow Accident (LOFA). This study aims to study the effect and role of the bypass flow after a LOFA in terms of the temperature and flow distributions and for the heat transport out of the core by natural convection of the coolant for a 1/12 symmetric section of the active core which is composed of images and mirror images of two sub-region models. The two sub-region models, 9 x 1/12 and 15 x 1/12 symmetric sectors of the active core, are employed as the CFD flow models using computational grid systems of 70.2 million and 117 million nodes, respectively. It is concluded that the effect of bypass flow is significant for the initial conditions and the beginning of LOFA, but the bypass flow has little effect after a long period of time in the transient computation of natural circulation.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2012

Vortex Diode Analysis and Testing for Fluoride Salt-Cooled High-Temperature Reactors

Fluidic diodes are presently being considered for use in several fluoride salt-cooled high-temperature reactor designs. A fluidic diode is a passive device that acts as a leaky check valve. These devices are installed in emergency heat removal systems that are designed to passively remove reactor decay heat using natural circulation. The direct reactor auxiliary cooling system (DRACS) uses DRACS salt-to-salt heat exchangers (DHXs) that operate in a path parallel to the core flow. Because of this geometry, under normal operating conditions some flow bypasses the core and flows through the DHX. A flow diode, operating in reverse direction, is-used to minimize this flow when the primary coolant pumps are in operation, while allowing forward flow through the DHX under natural circulation conditions. The DRACSs reject the core decay heat to the environment under loss-of-flow accident conditions and as such are a reactor safety feature. Fluidic diodes have not previously been used in an operating reactor system, and therefore their characteristics must be quantified to ensure successful operation. This report parametrically examines multiple design parameters of a vortex-type fluidic diode to determine the size of diode needed to reject a particular amount of decay heat. Additional calculations were performed to size a scaled diode that could be tested in the Oak Ridge National Laboratory Liquid Salt Flow Loop. These parametric studies have shown that a 152.4 mm diode could be used as a test article in that facility. A design for this diode is developed, and changes to the loop that will be necessary to test the diode are discussed. Initial testing of a scaled flow diode has been carried out in a water loop. The 150 mm diode design discussed above was modified to improve performance, and the final design tested was a 171.45 mm diameter vortex diode. The results of this testing indicate that diodicities of about 20 can be obtained for diodes of this size. Experimental results show similar trends as the computational fluid dynamics (CFD) results presented in this report; however, some differences exist that will need to be assessed in future studies. The results of this testing will be used to improve the diode design to be tested in the liquid salt loop system.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2021

Hanford Waste Treatment Plant LAB Facility Stack Effluent Monitoring: Sampling Probe Location Qualification Evaluation

The WTP LAB stack monitor locations were qualified using scale model stacks to mitigate the risk of identifying that sampling locations do not meet the qualification criteria on the full-scale stack. As required by the ANSI/HPS N13.1-1999 standard, the scale model and its sampling location were geometrically similar to the actual stack and the Reynolds numbers for both the actual and model stacks were >10,000. An additional criterion is that the DV of the full-scale stack must be between 1/6 DV and 6 DV of the scale model stack tests. Verification tests of the LAB stacks were performed at normal operating conditions. The maximum 6 DV value from the scale model testing determines the maximum conditions for which the full-scale stack may be operated and remain in compliance with the stack verification criterion. The remaining criteria for the stack verification to be considered valid involve the flow angle and velocity uniformity results. First, the flow angle at the full-scale stack must be = 20°. Second, the velocity uniformity at the full-scale stack must be = 20% COV. Finally, the velocity uniformity results for the actual and scale model stacks must agree within 5% COV. In general, these criteria were met through the full-scale stack tests at the LAB facility. Some specific items for each scale model and full-scale stack comparison should be noted in assessing the validity of the verification tests. The LB-C2 scale model stacks were performed with single fan operations at then-minimum flow conditions. These conditions resulted in DV values that were too low to meet the DV range criterion, and therefore are unable to be used in qualifying single fan operations on the full-scale stack. However, the single fan operations demonstrated COV values for velocity uniformity as well as flow angles that were comparable to the dual-fan operations at high flows. Overall, the test conditions and test results were within the range of acceptable values based on current design flow rates. Single fan operations are expected to only occur infrequently for maintenance needs and is therefore not a planned operating condition at this time. The LB-S1 scale model stacks were performed at three fan combinations, and in each combination, at least one test was performed at the then-minimum flow conditions as well as at then-maximum flow conditions. While the then-minimum flow conditions result in a DV range that is lower than the verification test DV value, and therefore does not meet the criterion, these tests are un-necessary for the verification test acceptance. The normal and maximum flow conditions from the scale model stack tests meet the DV range criterion, and the velocity uniformity test results compare favorably with the full-scale stack results. The LB-S2 scale model stacks were performed at nominally the same maximum flow condition; however, the 6 DV value for Fan A operation was slightly lower than the full-scale stack test DV. The COV values from the scale model stacks were comparable between the Fan A and Fan B results, and the overall test conditions are within the range of acceptable values. The stack verification is therefore considered acceptable for both Fan A and Fan B operations. The verification tests were performed at flows that were appreciably higher than the design conditions, and further elevated flow rates would be beyond the range of acceptable DV. Based on these stack verification test results, the three LAB filtered exhaust stacks meet the qualification criteria provided in the ANSI/HPS N13.1-1999 standard. Further changes to the system configuration or operating conditions that are outside the bounds described in this and the scale model test reports (Glissmeyer, Flaherty, and Piepel (2001), Glissmeyer and Geeting (2013)) may require additional tests and additional analyses to determine compliance with the standard.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2015

CFD Analysis of Upper Plenum Flow for a Sodium-Cooled Small Modular Reactor

Upper plenum flow behavior is important for many operational and safety issues in sodium fast reactors. The Prototype Gen-IV Sodium Fast Reactor (PGSFR), a pool-type, 150 MWe output power design, was used as a reference case for a detailed characterization of upper plenum flow for normal operating conditions. Computational Fluid Dynamics (CFD) simulation was utilized with detailed geometric modeling of major structures. Core outlet conditions based on prior system-level calculations were mapped to approximate the outlet temperatures and flow rates for each core assembly. Core outlet flow was found to largely bypass the Upper Internal Structures (UIS). Flow curves over the shield and circulates within the pool before exiting the plenum. Cross-flows and temperatures were evaluated near the core outlet, leading to a proposed height for the core outlet thermocouples to ensure accurate assembly-specific temperature readings. A passive scalar was used to evaluate fluid residence time from core outlet to IHX inlet, which can be used to assess the applicability of various methods for monitoring fuel failure. Additionally, the gas entrainment likelihood was assessed based on the CFD simulation results. Based on the evaluation of velocity gradients and turbulent kinetic energies and the available gas entrainment criteria in the literature, it was concluded that significant gas entrainment is unlikely for the current PGSFR design.

computational fluid dynamics↗
DOE OSTI2017

Fines migration during CO 2 injection: Experimental results interpreted using surface forces

The South West Hub project is one of the Australian Flagship Carbon Capture and Storage projects located in the south-west of Western Australia. To evaluate the injectivity potential during the forthcoming full-scale CO 2 injection, we conducted three core-flooding experiments using reservoir core plugs from the well Harvey-1. We aimed to investigate in this paper whether the injection of CO 2 leads to fines migration and permeability reduction due to the relatively high kaolinite content (up to 13%) in the injection interval of the target formation (i.e. the Wonnerup Member of the Lesueur Formation). We imaged the core samples before flooding to verify the presence of kaolinite at the pore-scale using scanning electron microscopy (SEM). We also examined the pore network of the core plugs before and after the core-flooding experiments using Nuclear Magnetic Resonance (NMR). Moreover, to gain a better understanding of any kaolinite fines migration, we delineated surface force using two models based on Derjaguin-Landau-Verwey-Overbeek (denoted by DLVO) theory coupled hydrodynamic force: (1) sphere/flat model representing interaction between kaolinite/quartz, and (2) flat/flat model representing interaction between kaolinite/kaolinite. Our core-flooding experimental results showed that CO 2 /brine injection triggered moderate to significant reduction in the permeability of the core samples with a negligible porosity change. NMR measurements supported the core-flooding results, suggesting that the relatively large pores disappeared in favour of a higher proportion of the medium to small pores after flooding. The DLVO calculations showed that some kaolinite particles probably lifted off and detached from neighbouring kaolinite particles rather than quartz grains. Moreover, the modelling results showed that the kaolinite fines migration would not occur under normal reservoir multiphase flow conditions. This is not because of the low hydrodynamic force. It is rather because the geometries of the particles dominate their interplay. Finally and overall, both of the experimental and analytical modelling results point to the fines migration to be the most likely cause of the permeability impairment observed during core-flooding experiments.

58 GEOSCIENCES↗
DOE OSTI2017

Scaling of thermal stratification in outlet plena of SFRs with gallium as a surrogate fluid

Due to its high thermal diffusivity, liquid sodium can develop thermally stratified regions in the hot and cold pools of SFRs under certain off normal operating conditions. The temperature gradients in the fluid give rise to mixed convection flow in which flow velocities result both from pressure gradients and from volume forces induced by density gradients. High thermal diffusivity and low momentum diffusivity leads to reduced thermal mixing and enhanced stratification. Thermal stratification, if not looked carefully in the design, can lead to severe consequences in normal operation with forced coolant circulation and during off-normal natural circulation. The resulting hydraulic and thermal loads can cause the failure modes in structures and components critical for reactor safety. Thermal stratification in the SFR pools can lead to thermal fatigue, which can lead to structural and material failure, or pose serious concerns for the reactor safety. In addition to thermo-structural issues, the reactivity changes are significantly impacted due to density or temperature driven movement of liquid sodium. Temperature changes or gradients in the Upper Instrumentation Structures (UIS) of SFRs can lead to thermal contraction, a case that can lead to positive reactivity insertion after core shut-down. The motivation of this work is to design a scaled down experimental facility which can generate high fidelity data for thermal stratification physics to validate 3-dimensional (3D) Computational Fluid Dynamics (CFD) models. These validated CFD codes can be used to improve the 1-dimensional (1D) system level codes such as SAS4A/SASSYS-1 which are currently not capable of simulating thermal stratification accurately. These improvements in 1D system level SAS4A/SASSYS-1 codes can significantly improve the safety analyses and design of SFRs. The GaTE facility outlet plenum model is designed based on physics based similarities. Thermal stratification physics is governed primarily by the Richardson numbers (Ri). The expected low flow rates during off-normal conditions can lead to a high degree of thermal stratification in the outlet plenum with a corresponding Ri = 530. Therefore, the inlet flow velocities and inlet hydraulic diameter for the gallium coolant entering the test vessel are required to achieve same Ri. The design of test vessel in GaTE facility is finalized to a 1/20. scale outlet plenum height of the ABTR. This design of the test facility will allow the decoupling of the thermal stratification physics at low flow rates with the transient natural circulation. In other words, it will enable effective control to study thermal stratification at very low flow rates. Numerical studies are conducted with a 3D CFD code to examine the thermal behavior of the selected test design. Simulation results show that thermal gradients are present only in the vertical or axial direction which are similar to the behavior reported in previous studies on full scale SFR models. Future work will involve the experimental studies using the GaTE facility with the high fidelity instrumentation.

21 SPECIFIC NUCLEAR REACTORS AND ASSOCIATED PLANTS↗
DOE OSTI2016

Preliminary Development of a Drift-Flux Model Based Pin-Resolved Core Thermal-Hydraulic Code for Dedicated Use in Efficient High-Fidelity Neutronics Coupled Calculations

As the high-fidelity multiphysics coupled core calculation becomes more practical with the advent of high-performance computing technology, efficient coupling of a neutronics code with a thermal-hydraulic (T/H) code at the pin level poses a new challenge. The COBRA-TF (CTF) code which is one of the well-known subchannel analysis codes based on the two-fluid, three-field model was coupled with many neutron transport codes such as MPACT, TORTTD, and nTRACER due to the easiness of coupling by its well-established coupling interface. Nevertheless, the calculation efficiency of such coupled codes is not satisfactory because of the heavy computational burden of CTF. The MATRA subchannel analysis code of KAERI which is based on the homogeneous equilibrium model (HEM) was coupled with neutronics codes as well. The choice of the HEM and the marching scheme in MATRA reduces the computational burden significantly. However, the applicability of the HEM for the two-phase flow conditions is limited to normal operating conditions of pressurized water reactors (PWRs) and therefore a better model which would allow wider applicability than HEM is necessary in order to deal with the core states involving nontrivial voiding. The drift-flux model (DFM) which uses 5 equations for two-fluid analyses by adopting the mixture velocity instead of two phasic velocities can be a good compromise between the two-fluid model and the HEM. With the DFM, the computational cost can be reduced and moreover the modeling of the interfacial area constitutive relations can be avoided. The phase separation for the mixture velocity can be incorporated by using the drift-flux parameters, C0 and Vgj, which can be obtained by using empirical formulas. Several subchannel codes such as WOSUB and ASSERT were developed based on the DFM. It was also adopted in some system analysis codes such as RETRAN3D. In the work here, the development of a DFM based pin-resolved T/H analysis code, ESCOT (Efficient Simulator of Core Thermal-hydraulics), is introduced for the use in efficient coupled calculations with a high-fidelity neutronics code. The field equations and the numerical methods are briefly described. Then the efficiency tests of various linear system solvers including the Krylov subspace methods (KSMs) are presented. A test problem is solved by ESCOT, MATRA and CTF for performance comparison and validation. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2026

A Full-Induction Magnetohydrodynamics Solver for Liquid Metal Fusion Blankets in Vertex-CFD

Multiphysics modeling of liquid metal fusion blankets, which produce tritium and convert energy of neutrons created via fusion reactions into heat, is crucial for predicting performance, ensuring structural integrity, and optimizing energy production. While traditional blanket modeling of liquid metal flows during normal steady operating conditions commonly employs the inductionless approximation of the magnetohydrodynamics (MHD) equations, transient scenarios, when the plasma-confining magnetic field varies on millisecond time scales, require a full-induction MHD approach that dynamically evolves the magnetic field via the time-dependent induction equation. This paper presents the formulation, implementation, and initial verification of a full-induction MHD solver integrated within the open-source Vertex-CFD framework, which aims to achieve tight multiphysics coupling, a flexible software design enabling easy extension and addition of physics models, and performance portability across computing platforms. The solver utilizes finite element spatial discretization, implicit Runge–Kutta time integration, and an inexact Newton method to solve the resulting discrete nonlinear system, leveraging Trilinos packages for efficient computation. Verification against selected benchmark problems demonstrates accuracy and robustness of the solver. Furthermore, when the solver is applied to an idealized blanket model in 2.5D and full 3D, results obtained with Vertex-CFD are in good agreement with recently published quasi-2D simulations. These findings establish a computational foundation for future simulations of transient MHD phenomena in liquid metal blankets with Vertex-CFD, and open avenues for future extensions and performance optimizations.

Endeve, Eirik [ORNL] (ORCID:0000000312519507)↗