Search NASA⌕ Search

SEARCH · Search NASA

Results for “normalizing flows”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
NASA NTRS1996

Stability of Wavy Films in Gas-Liquid Two-Phase Flows at Normal and Microgravity Conditions

For flow rates of technological interest, most gas-liquid flows in pipes are in the annular flow regime, in which, the liquid moves along the pipe wall in a thin, wavy film and the gas flows in the core region. The waves appearing on the liquid film have a profound influence on the transfer rates, and hence on the design of these systems. We have recently proposed and analyzed two boundary layer models that describe the characteristics of laminar wavy films at high Reynolds numbers (300-1200). Comparison of model predictions to 1-g experimental data showed good agreement. The goal of our present work is to understand through a combined program of experimental and modeling studies the characteristics of wavy films in annular two-phase gas-liquid flows under normal as well as microgravity conditions in the developed and entry regions.

Balakotaiah, V.↗
NASA NTRS1999

Detailed Studies on the Structure and Dynamics of Reacting Dusty Flows at Normal and Microgravity

Dusty reacting flows are of particular interest for a wide range of applications. Inert particles can alter the flammability and extinction limits of a combustible mixture. Reacting particles can release substantial amount of heat and can be used either for power generation or propulsion. Accumulation of combustible particles in air can result in explosions which, for example, can occur in grain elevators, during lumber milling and in mine galleries. Furthermore, inert particles are used as flow velocity markers in reacting flows, and their velocity is measured by non-intrusive laser diagnostic techniques. Despite their importance, dusty reacting flows have been less studied and understood compared to gas phase as well as sprays. The addition of solid particles in a flowing gas stream can lead to strong couplings between the two phases, which can be of dynamic, thermal, and chemical nature. The dynamic coupling between the two phases is caused by the inertia that causes the phases to move with different velocities. Furthermore, gravitational, thermophoretic, photophoretic, electrophoretic, diffusiophoretic, centrifugal, and magnetic forces can be exerted on the particles. In general, magnetic, electrophoretic, centrifugal, photophoretic, and diffusiophoretic can be neglected. On the other hand, thermophoretic forces, caused by steep temperature gradients, can be important. The gravitational forces are almost always present and can affect the dynamic response of large particles. Understanding and quantifying the chemical coupling between two phases is a challenging task. However, all reacting particles begin this process as inert particles, and they must be heated before they participate in the combustion process. Thus, one must first understand the interactions of inert particles in a combustion environment. The in-detail understanding of the dynamics and structure of dusty flows can be only advanced by considering simple flow geometries such as the opposed-jet, stagnation-type. In such configurations the imposed strain rate is well characterized, and the in-depth understanding of the details of the physico-chemical processes can be systematically obtained. A number of computational and experimental studies on spray and particle flows have been conducted in stagnation-type configurations. Numerically, the need for a hybrid Eulerian-Lagrangian approach has been identified by Continillo and Sirignano, and the use of such approach has allowed for the prediction of the phenomenon of droplet flow reversal. Gomez and Rosner have conducted a detailed study on the particle response in the opposed-jet configuration, and the particle thermophoretic diffusivities were determined experimentally. Sung, Law and co-workers have conducted numerical studies on the effect of strain rate and temperature gradients on the dynamics of inert particles, as a way of understanding potential errors in experimental LDV data that may arise from thermophoretic forces. This investigation is a combined experimental and numerical study on the details of reacting dusty flows. The specific tasks are: (1) Experimental determination of laminar flame speeds, and extinction strain rates of dusty flows at normal- and micro-gravity as functions of the particle type, particle initial diameter, particle initial number density, and gas phase chemical composition; (2) Detailed numerical simulation of the experiments. Results are compared with experiments and the adequacy of theoretical models is assessed; and (3) Provision of enhanced insight into the thermo-chemical coupling between the two phases.

Andac, M. Gurhan↗
DOE OSTI2025

Solving high-dimensional inverse problems using amortized likelihood-free inference with noisy and incomplete data

Here, we present a likelihood-free probabilistic inversion method based on normalizing flows for high-dimensional inverse problems. The proposed method is composed of two complementary networks: a summary network for data compression and an inference network for parameter estimation. The summary network encodes raw observations into a fixed-size vector of summary features, while the inference network generates samples of the approximate posterior distribution of the model parameters based on these summary features. The posterior samples are produced in a deep generative fashion by sampling from a latent Gaussian distribution and passing these samples through an invertible transformation. We construct this invertible transformation by sequentially alternating conditional invertible neural network and conditional neural spline flow layers. The summary and inference networks are trained simultaneously. We apply the proposed method to an inversion problem in groundwater hydrology to estimate the posterior distribution of the log-conductivity field conditioned on spatially sparse time-series observations of the system’s hydraulic head responses. The conductivity field is represented with 706 degrees of freedom in the considered problem. Comparison with the likelihood-based iterative ensemble smoother PEST-IES method demonstrates that the proposed method accurately estimates the parameter posterior distribution and the observations’ predictive posterior distribution at a fraction of the inference time of PEST-IES.

conditional invertible neural network↗
NASA NTRS1998

Bubble Formation and Detachment in Liquid Flow Under Normal and Reduced Gravity

Two-phase flows are present in a wide variety of applications for spacecraft thermal control systems design. Bubble formation and detachment is an integral part of the two-phase flow science. The authors objective is to experimentally investigate the effects of liquid cross velocity, gas velocity, and nozzle diameter on the bubble diameter at detachment under reduced and normal gravity and under relatively low gas flow rates. Results from ground (1 g) and reduced gravity experiments will be presented in this paper. For the 1 g experiment, a flow loop was designed and built to accommodate a range of liquid and gas flow rates. The reduced gravity experiment was conducted on the NASA DC-9 reduced gravity platform using the two-phase flow loop qualified for operation on the low-gravity platform. Flow visualization is accomplished using a high speed 500 frames/s camera. The results suggest that the existence of buoyancy force contributes to the faster detachment of bubbles. Buoyancy helps the detachment process which results in smaller bubbles being formed. In reduced gravity, although drag force is present, the virtual non-existence of buoyant force results in larger bubbles and longer times for detachment. Theoretical predictions are also presented in this paper and seem to agree with the experimental results.

Nahra, Henry K.↗
NASA NTRS1997

Detailed Studies on the Structure and Dynamics of Reacting Dusty Flows at Normal and Microgravity

Two-phase reacting flows are substantially less understood compared to gas phase flows. While extensive work has been done on sprays, less attention has been given to the details of dusty reacting flows. Dusty flows are of particular interest for a wide range of applications. Particles can be present in a gas intentionally or unintentionally, and they can be inert or reacting. Inert particles can be also present in an otherwise reacting gas flow, and that can lead to flame cooling and modification of the extinction limits of a combustible mixture. Reacting solid particles can release substantial amounts of heat upon oxidation, and can be used either for propulsion (e.g. Al, B, Mg) or power generation (coal). Furthermore, accidents can occur when a reacting dust accumulates in air and which, in the presence of an ignition source, can cause explosion. Such explosions can occur during lumber milling, in grain elevators, and in mine galleries.

Egolfopoulos, Fokion N.↗
DOE OSTI2026

Neural posterior unfolding

Differential cross section measurements are the currency of scientific exchange in particle and nuclear physics. A key challenge for these analyses is the correction for detector distortions, known as deconvolution or unfolding. Binned unfolding of cross section measurements traditionally rely on the regularized inversion of the response matrix that represents the detector response, mapping pre-detector (`particle level') observables to post-detector (`detector level') observables. In this paper we introduce Neural Posterior Unfolding, a modern, Bayesian approach that leverages normalizing flows for unfolding. By using normalizing flows for neural posterior estimation, NPU offers several key advantages including implicit regularization through the neural network architecture, fast amortized inference that eliminates the need for repeated retraining, and direct access to the full uncertainty in the unfolded result. In addition to introducing NPU, we implement a classical Bayesian unfolding method called Fully Bayesian Unfolding (FBU) in modern Python so it can also be studied. These tools are validated on simple Gaussian examples and then tested on simulated jet substructure examples from the Large Hadron Collider (LHC). We find that the Bayesian methods are effective and worth additional development to be analysis ready for cross section measurements at the LHC and beyond.

Analysis and statistical methods↗
NASA NTRS1992

Magnetospheric plasma flows associated with boundary waves and flux transfer events

We describe plasma flow perturbations in the outer magnetosphere during the passage of magnetopause boundary waves and cylindrically shaped flux transfer events (FTEs). Spacecraft which remain within the magnetosphere observe bipolar flows normal to the nominal magnetopause accompanied by flows nearly opposite to the direction of boundary wave and/or FTE motion. Flows are generally discontinuous across FTE and magnetopause boundaries, with reversals in the component of flow tangential to the nominal magnetopause expected during all magnetopause crossings and some entries into FTEs. No bipolar flows normal to the nominal magnetopause occur within FTEs unless they rotate and none occur in the magnetosheath unless the boundary waves propagate relative to the magnetosheath flow. IRM satellite observations on October 28, 1984 may be interpreted either in terms of FTEs or boundary waves.

Sibeck, D. G.↗
NASA NTRS1981

Jet noise results from static, wind tunnel, and flight tests of conical and mechanical suppressor nozzles

Results of jet noise suppression tests conducted on a Rolls-Royce Viper 601 turbojet engine are reported. Seven exhaust nozzle configurations are tested, including two conical nozzles, two suppressor mixers, and three treated ejector configurations with different ejector inlets. Tests are conducted at the NASA Ames outdoor static test facility and the 40- by 80-ft wind tunnel facility at minimum tunnel flow velocity and normal flow velocities of 230 and 290 ft/sec. Near-field multiple sideline noise levels are projected to the far fields to compare far-field fixed microphone outdoor static noise levels, and wind tunnel near-field noise data are projected to the far field and flight distances to compare with noise levels recorded from an Hs-125 aircraft. Near-field outdoor noise data duplicate the far-field data recorded from fixed microphones within 2 PNdB, and the Douglas mechanical jet noise suppressor/treated ejector exhaust system achieves a noise reduction of 12 EPNdB relative to a conic reference nozzle at equal thrust in flight.

Mckinnon, R. A.↗
NASA NTRS2013

Effects of Bedrock Lithology and Subglacial Till on the Motion of Ruth Glacier, Alaska, Deduced from Five Pulses from 1973-2012

A pulse is a type of unstable glacier flow intermediate between normal flow and surging. Using Landsat MSS, TM, and ETM+ imagery and feature tracking software, a time-series of mostly annual velocity maps from 1973 to 2012 was produced that reveals five pulses of Ruth Glacier, Alaska. Peaks in ice velocity were found in the 1981, 1989, 1997, 2003, and 2010; approximately every 7 years. During these peak years the ice velocity increased 300%, from approximately 40 m/yr to 160 m/yr, and occurred in an area of the glacier underlain by sedimentary bedrock. Based on the spatio-temporal behavior of Ruth Glacier during the pulse cycles, we suggest the pulses are due to enhanced basal motion via deformation of a subglacial till. The cyclical nature of the pulses is theorized to be due to a thin till, with low permeability, that causes incomplete drainage of the till between the pulses, followed by eventual recharge and dilation of the till. These findings suggest care is needed when attempting to correlate changes in regional climate with decadal-scale changes in velocity, because in some instances basal conditions may have a greater influence on ice dynamics than climate.

Glacier↗
DOE OSTI2024

Sequential Kalman tuning of the t -preconditioned Crank-Nicolson algorithm: efficient, adaptive and gradient-free inference for Bayesian inverse problems

Ensemble Kalman Inversion (EKI) has been proposed as an efficient method for the approximate solution of Bayesian inverse problems with expensive forward models. However, when applied to the Bayesian inverse problem EKI is only exact in the regime of Gaussian target measures and linear forward models. Here, in this work we propose embedding EKI and Flow Annealed Kalman Inversion, its normalizing flow (NF) preconditioned variant, within a Bayesian annealing scheme as part of an adaptive implementation of the t-preconditioned Crank-Nicolson (tpCN) sampler. The tpCN sampler differs from standard pCN in that its proposal is reversible with respect to the multivariate t-distribution. The more flexible tail behaviour allows for better adaptation to sampling from non-Gaussian targets. Within our Sequential Kalman Tuning (SKT) adaptation scheme, EKI is used to initialize and precondition the tpCN sampler for each annealed target. The subsequent tpCN iterations ensure particles are correctly distributed according to each annealed target, avoiding the accumulation of errors that would otherwise impact EKI. We demonstrate the performance of SKT for tpCN on three challenging numerical benchmarks, showing significant improvements in the rate of convergence compared to adaptation within standard SMC with importance weighted resampling at each temperature level, and compared to similar adaptive implementations of standard pCN. The SKT scheme applied to tpCN offers an efficient, practical solution for solving the Bayesian inverse problem when gradients of the forward model are not available. Code implementing the SKT schemes for tpCN is available at https://github.com/RichardGrumitt/KalmanMC.

97 MATHEMATICS AND COMPUTING↗
NASA NTRSDate not supplied

Aerodynamic Assessment of Surface-Normal Active Flow Control for Lift Enhancement on the High-Lift Common Research Model

This study explores aerodynamic feasibility of small surface-normal pneumatic jets (micro-jets) for lift enhancement on the high-lift Common Research Model. To date, lift enhancement studies using microjets have been conducted at wind tunnel Reynolds numbers less than three and half million. This paper computationally investigates microjet effects at a flight Reynolds number of24.6million. Prior to microjet activation, it is shown that flow features such as trailing edge (TE) flow separation on both inboard and outboard flaps have noticeable sensitivity to Reynolds number. At the flight Reynolds number, the onset of TE flow separation shifts downstream resulting in less flow separation compared to at the wind tunnel Reynolds numberof3.26million. The effects are insignificant for lift enhancement due to microjet on the inboard flap. However, the effects are significant for lift enhancement on the outboard flap. Microjets are shown to be more effective in circulation control on the outboard flap at the higher Reynolds number. For a microjet with velocity ratio of one, lift enhancement is increased by17%at the flight Reynolds number. This highlights the sensitivity of microjet performance to the flow features of the configuration it is implemented on. Microjet lift enhancement trends at both Reynolds numbers are found to be consistent in their ability to move the lift curve upwards in the linear regime, with the lift enhancement related to microjet momentum coefficient throughΔ𝐶𝐿=𝐾√︁𝐶𝜇. Analysis of the impact microjets have on drag shows that microjets improve the configuration’s Oswald efficiency by5.4%and2.5%at the wind tunnel and flight Reynolds numbers respectively, while providing lift enhancement. Microjet effects on the trim drag are found to be less than one percent of the total drag. The results of this study highlight the potential of small surface-normal pneumatic jets for improved performance at high-lift conditions.

ARMD↗
NASA NTRS1976

Performance of 1.15-pressure-ratio fan stage at several rotor blade setting angles with reverse flow

A 51 cm diameter low pressure ratio fan stage was tested in reverse flow. Survey flow data were taken over the range of rotative speed from 50 percent to 100 percent design speed at several rotor blade setting angles through both flat and feather pitch. Normal flow design values of pressure ratio and weight flow were 1.15 and 29.9 kg/sec with a rotor tip speed of 243.8 m/sec. The maximum thrust in reverse flow was 52.5 percent of design thrust in normal flow.

Kovich, G.↗
NASA NTRSDate not supplied

Surface-Normal Active Flow Control for Lift Enhancement and Separation Mitigation for High-Lift Common Research Model

This study explores the feasibility of using small surface-normal jets near the flap trailing edge as active aerodynamic load control on the high-lift Common Research Model wing/-body configuration. Chimera Grid Tools are utilized to generate structured curvilinear overset grids, and the Reynolds-averaged Navier-Stokes solver OVERFLOW is employed to solve for the flow-field around the geometry. The so-called microjet is initially employed across both the inboard and the outboard flaps on the pressure-side near the trailing edge. It is shown that implementing the microjet on the inboard flap is more effective compared to implementing it on the outboard flap. This is because, prior to microjet implementation, the flow on the outboard flap exhibits extensive separation, while the flow on the inboard flap exhibits moderate separation. For microjet implementation across the inboard flap only, the relationship between momentum coefficient of the microjet and lift-enhancement is found to be∆CL'1.66√Cμfor the rangeCμ= 0.00−0.012. We show that implementing a microjet with a jet velocity ratio of one, which corresponds toCμ= 0.003, can shift the linear region of the lift curve by ∆CL= 0.08. The linear shift in the lift curve is significant for enhancing airplane performance such as increasing its payload. Microjet implementation effects on the drag coefficient are also investigated through a drag decomposition analysis. Further, we employ an induced drag analysis based on the spanwise load distribution and show that the microjet-related increase in pressure drag coefficient is dominated by the increase in the induced drag while microjet implementation reduces the form drag. These preliminary results show that favorable changes in aerodynamic performance can be achieved by using the surface-normal jets presented in this study.

ARMD↗
DOE OSTI2026

Efficient many-jet event generation with flow matching

We apply for the first time, to the best of our knowledge, the flow matching method to the problem of phase-space sampling for event generation in high-energy collider physics. By training the model to remap the random numbers used to generate the momenta and helicities of the scattering matrix elements as implemented in the portable partonic event generator pepper, we find substantial efficiency improvements in the studied processes. We focus our study on the highest final-state multiplicities in Drell-Yan and top-antitop pair production used in simulated samples for the Large Hadron Collider, which computationally are the most relevant ones. We find that the unweighting efficiencies improve by factors of 184 and 25, respectively, when compared to the standard approach of using a vegas-based optimization. We also compare continuous normalizing flows trained with flow matching against the previously studied normalizing flows based on coupling layers and find that the former leads to better results, faster training and a better scaling behavior across the studied multiplicity range, while the latter evaluate faster. When combining the advantages of both methods using the regflow approach, we find parton-level unweighted event generation walltime gains of about a factor of 10 at the highest final-state multiplicities.

Bothmann, E. [CERN; Gottingen U.] (ORCID:000000016↗
NASA NTRS1994

Low-Pressure-Drop Shutoff Valve

Flapper valve remains open under normal flow conditions but closes upon sudden increases to high rate of flow and remains closed until reset. Valve is fluid/mechanical analog of electrical fuse or circuit breaker. Low-pressure-drop shutoff valve contains flapper machined from cylindrical surface. During normal flow conditions, flapper presents small cross section to flow. (Useful in stopping loss of fluid through leaks in cooling systems.)

Thornborrow, John↗