Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Closure model”

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

MLUQ (Uncertainty quantification for ML closure models) [SWR-24-36]

Data-based closure models are increasingly being used to replace physic-based closure models because of their flexibility and the growing availability of data. However, closure models are subject to uncertainty because of lack of data in parts of the input space (epistemic uncertainty) or because of noise in the data (aleatoric uncertainty). This software contains a toolbox for training bayesian neural nets that can estimate both uncertainties. A specific treatment is provided to ensure accuracy outside of the data distribution. A set of tools are provided to reduce the dimensionality of the uncertain parameter space, thereby enabling fast uncertainty propagation.

Hassanaly, Malik↗
NASA NTRS1985

Predictions of transonic separated flow with an eddy-viscosity/Reynolds-shear-stress closure model

In previous work, a new turbulence-closure model was specifically developed for two-dimensional turbulent boundary layers subjected to strong, adverse-pressure gradients and the attendant separation. This closure model was shown to perform well, but the inverse, boundary-layer calculations used in evaluating the model were not truly predictive since the mass-flux parameter distribution obtained from the experimental data had to be specified. The objective of this study was to evaluate this closure model for a series of strong, transonic, inviscid-viscous interactions with varying degrees of separation using a fully predictive calculation method based on the Reynolds-averaged, Navier-Stokes equations. Calculated results are compared with experimental results and are shown to be in excellent agreement, even for the interactions with massive separation. This closure model has the very favorable property of requiring little more computational effort than equilibrium, algebraic models.

Johnson, D. A.↗
NASA NTRS1984

Simulations of the stratocumulus-topped boundary layer with a third-order closure model

A third order closure model is proposed by Andre et al. (1982), in which the time rate of change terms, the relaxation and rapid effects for the pressure related terms, and the clipping approximation are included along with the quasi-normal closure, to study turbulence in a cloudy layer which is cooled radiatively from above. A spurious oscillation which is strongest near the inversion occurs. An analysis of the problem shows that the oscillation arises from the mean gradient and buoyancy terms of the triple moment equations; these terms are largest near the cloud top. The oscillation is physical, rather than computational. In nature the oscillation is effectively damped, by a mechanism which apparently is not included in our model. In the stably stratified layer just above the mixed layer top, turbulence can excite gravity waves, whose energy is radiated away. Because the closure assumption for the pressure terms does not take into account the transport of wave energy, the model generates spurious oscillations. Damping of the oscillations is possible by introducing diffusion terms into the triple moment equations. With a large enough choice for the diffusion coefficient, the oscillation is effectively eliminated. The results are quite sensitive to the ad hoc eddy coefficient.

Moeng, C. H.↗
NASA NTRS1986

Improvement of the second- and third-moment modeling of turbulence: A study of Reynolds-stress closure model

Four parts of the Reynolds-stress closure modeling are reported: (1) improvement of the k and epsilon equaitons; (2) development of the third-moment transport equation; (3) formulation of the diffusion coefficient of the momentum equation by using the algebraic-stress model of turbulence; and (4) the application of the Reynolds-stress model to a heat exchanger problem. It was demonstrated that the third-moment transport model improved the prediction of the triple-velocity products in the recirculating and reattaching flow regions in comparison with the existing algebraic models for the triple-velocity products. Optimum values for empirical coefficients are obtained for the prediction of the backward-facing step flows. A functional expression is derived for the coefficient of the momentum diffusion by employing the algebraic-stress model. The second-moment closure is applied to a heat transfer problem. The computations for the flow in a corrugated-wall channel show that the second-moment closure improves the prediction of the heat transfer rates by 30% over the k - epsilon model.

Amano, R. S.↗
DOE OSTI2020

Improvements to CTF Closure Models for Modelling of Two-Phase Flow

This report documents the efforts to improve the CTF prediction of void fraction and two-phase pressure drop by improving two-phase closure models. Previous validation activities have revealed that CTF tends to overpredict void fraction and two-phase pressure drop. In response to this, two approaches were taken to improve CTF’s predictive capabilities. Based on findings that the interfacial drag and subcooled boiling models significantly impact void prediction, alternative closure models for these physical effects were found and implemented into the code. An extensive assessment was performed by using the existing and newly added two-phase experimental data, which showed that void prediction, wall temperature, and two-phase pressure-drop results are improved by using the newly implemented models. A second approach for improving CTF modeling accuracy involved the use of a Bayesian calibration process that uses CTF validation data to optimize selected modeling coefficients and closure model multipliers to achieve a more accurate prediction of experimental results. A similar assessment was performed with calibrated models that improved the void and pressure-drop prediction for test cases that were both included and not included in the calibration dataset. Results of this study will be used to change the models used in CTF to achieve more accurate BWR analyses moving forward. This study also identified opportunities for improving additional closure models and other opportunities to use calibration techniques to improve CTF.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2022

Variational multiscale reinforcement learning for discovering reduced order closure models of nonlinear spatiotemporal transport systems

Abstract A central challenge in the computational modeling and simulation of a multitude of science applications is to achieve robust and accurate closures for their coarse-grained representations due to underlying highly nonlinear multiscale interactions. These closure models are common in many nonlinear spatiotemporal systems to account for losses due to reduced order representations, including many transport phenomena in fluids. Previous data-driven closure modeling efforts have mostly focused on supervised learning approaches using high fidelity simulation data. On the other hand, reinforcement learning (RL) is a powerful yet relatively uncharted method in spatiotemporally extended systems. In this study, we put forth a modular dynamic closure modeling and discovery framework to stabilize the Galerkin projection based reduced order models that may arise in many nonlinear spatiotemporal dynamical systems with quadratic nonlinearity. However, a key element in creating a robust RL agent is to introduce a feasible reward function, which can be constituted of any difference metrics between the RL model and high fidelity simulation data. First, we introduce a multi-modal RL to discover mode-dependant closure policies that utilize the high fidelity data in rewarding our RL agent. We then formulate a variational multiscale RL (VMRL) approach to discover closure models without requiring access to the high fidelity data in designing the reward function. Specifically, our chief innovation is to leverage variational multiscale formalism to quantify the difference between modal interactions in Galerkin systems. Our results in simulating the viscous Burgers equation indicate that the proposed VMRL method leads to robust and accurate closure parameterizations, and it may potentially be used to discover scale-aware closure models for complex dynamical systems.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1992

Comments on the present state of second-order closure models for incompressible flows

Second-order closure models account for history and nonlocal effects of the mean velocity gradients on the Reynolds stress tensor. Turbulent flows involving body forces or curvature, Reynolds stress relaxational effects, and counter-gradient transport are usually better described. The topics are presented in viewgraph form and include: (1) the Reynolds stress transport equation; (2) issues in second-order closure modeling; and (3) near wall models.

Speziale, Charles G.↗
DOE OSTI2022

Machine Learning Moment Closure Models for the Radiative Transfer Equation III: Enforcing Hyperbolicity and Physical Characteristic Speeds

This is the third paper in a series in which we develop machine learning (ML) moment closure models for the radiative transfer equation. In our previous work (Huang et al. in J Comput Phys 453:110941, 2022), we proposed an approach to learn the gradient of the unclosed high order moment, which performs much better than learning the moment itself and the conventional $P_N$ closure. However, while the ML moment closure has better accuracy, it is not able to guarantee hyperbolicity and has issues with long time stability. In our second paper (Huang et al., in: Machine learning moment closure models for the radiative transfer equation II: enforcing global hyperbolicity in gradient based closures, 2021. arXiv:2105.14410), we identified a symmetrizer which leads to conditions that enforce that the gradient based ML closure is symmetrizable hyperbolic and stable over long time. The limitation of this approach is that in practice the highest moment can only be related to four, or fewer, lower moments. In this paper, we propose a new method to enforce the hyperbolicity of the ML closure model. Motivated by the observation that the coefficient matrix of the closure system is a lower Hessenberg matrix, we relate its eigenvalues to the roots of an associated polynomial. Here, we design two new neural network architectures based on this relation. The ML closure model resulting from the first neural network is weakly hyperbolic and guarantees the physical characteristic speeds, i.e., the eigenvalues are bounded by the speed of light. The second model is strictly hyperbolic and does not guarantee the boundedness of the eigenvalues. Several benchmark tests including the Gaussian source problem and the two-material problem show the good accuracy, stability and generalizability of our hyperbolic ML closure model.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2004

The Liquid-Water Oscillation in Modeling Boundary-Layer Cumuli with Third-Order Turbulence Closure Models

A hierarchy of third-order turbulence closure models are used to simulate boundary-layer cumuli in this study. An unrealistically strong liquid-water oscillation (LWO) is found in the fully prognostic model, which predicts all third moments. The LWO propagates from cloud base to cloud top with a speed of 1 m/s. The period of the oscillation is about 1000 s. Liquid-water buoyancy terms in the third-moment equations contribute to the LWO. The LWO mainly affects the vertical profiles of cloud fraction, mean liquid-water mixing ratio and the fluxes of liquid-water potential temperature and total water, but has less impact on the vertical profiles of other second-moments and third-moments. In order to minimize the LWO, a moderate large diffusion coefficient and a large turbulent dissipation at its originating level are needed. However, this approach distorts the vertical distributions of cloud fraction and liquid-water mixing ratio. A better approach is to parameterize liquid-water buoyancy more reasonably. A minimally prognostic model, which diagnoses all third moments except for vertical velocity, is shown to produce better results, compared to a fully prognostic model.

Cheng, A.↗
NASA NTRS1981

Prediction of fatigue-crack growth under variableamplitude and spectrum loading using a closure model

An existing analytical crack closure model was used to study crack growth under various load histories. The model was based on a concept like the Dugdale model, but modified to leave plastically deformed material in the wake of the advancing crack tip. The model was used to correlate crack growth rates under constant amplitude loading, and to predict crack growth under variable amplitude and aircraft spectrum loading on 2219-T851 aluminum alloy sheet material. The predicted crack growth lives agreed well with experimental data. For 80 crack growth tests subjected to various load histories, the ratio of predicted-to-experimental lives (N(P)/n(T)) ranged from 0.5 to 1.8. The mean value of N(P)/N(T) was 0.97 and the standard deviation was 0.27.

Newman, J. C., Jr.↗
NASA NTRS1999

Ocean Turbulence: Two-Point Closure Model Momentum, Heat and Salt Vertical Diffusivities in the Presence of Shear - Paper 3

In papers 1 and 2 we have presented the results of the most updated 1-point closure model for the turbulent vertical diffusivities of momentum, heat and salt, K(sub m,h,s). In this paper, we derive the analytic expressions for K(sub m,h,s) using a new 2-point closure model that has recently been developed and successfully tested against some approx. 80 turbulence statistics for different flows. The new model has no free parameters. The expressions for K(sub m, h. s) are analytical functions of two stability parameters: the Turner number R(sub rho) (salinity gradient/temperature gradient) and the Richardson number R(sub i) (temperature gradient/shear). The turbulent kinetic energy K and its rate of dissipation may be taken local or non-local (K-epsilon model). Contrary to all previous models that to describe turbulent mixing below the mixed layer (ML) have adopted three adjustable "background diffusivities" for momentum. heat and salt, we propose a model that avoids such adjustable diffusivities. We assume that below the ML, K(sub m,h,s) have the same functional dependence on R(sub i) and R(sub rho) derived from the turbulence model. However, in order to compute R(sub i) below the ML, we use data of vertical shear due to wave-breaking measured by Gargett et al. (1981). The procedure frees the model from adjustable background diffusivities and indeed we use the same model throughout the entire vertical extent of the ocean. Using the new K(sub m,h, s), we run an O-GCM and present a variety of results that we compare with Levitus and the KPP model. Since the traditional 1-point (used in papers 1 and 2) and the new 2-point closure models used here represent different modeling philosophies and procedures, testing them in an O-GCM is indispensable. The basic motivation is to show that the new 2-point closure model gives results that are overall superior to the 1-point closure in spite of the fact that the latter rely on several adjustable parameters while the new 2-point closure has none. After the extensive comparisons presented in papers 1 and 2, we conclude that the new model presented here is overall superior for it not only is parameter free but also 2 because is part of a more general turbulence model that has been previously successfully tested on a wide variety of other types of turbulent flows.

Canuto, V. M.↗
DOE OSTI2024

a priori uncertainty quantification of reacting turbulence closure models using Bayesian neural networks

While many physics-based closure model forms have been posited for the sub-filter scale (SFS) in large eddy simulation (LES), vast amounts of data available from direct numerical simulations (DNS) create opportunities to leverage data-driven modeling techniques. Albeit flexible, data-driven models still depend on the dataset and the functional form of the model chosen. Increased adoption of such models requires reliable uncertainty estimates both in the data-informed and out-of-distribution regimes. Here, in this work, we employ Bayesian neural networks (BNNs) to capture both epistemic and aleatoric uncertainties in a reacting flow model. In particular, we model the filtered progress variable scalar dissipation rate which plays a key role in the dynamics of turbulent premixed flames. We demonstrate that BNN models can provide unique insights about the structure of uncertainty of the data-driven closure models. We also propose a method for the incorporation of out-of-distribution information in a BNN, which can be used for out-of-distribution query detection. The efficacy of the model is demonstrated by a priori evaluation on a dataset consisting of a variety of flame conditions and fuels.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Achievements and Ongoing Challenges for advanced CFD Boiling Closure Model Development using Physical Insights from Validation-Oriented High-Fidelity Boiling Experiment

The fundamental understanding of the boiling-associated heat transfer, including sliding bubble effect, is still challenging despite the intensive research for decades in both experimental and computational boiling research communities. One of the main difficulties of boiling experiment come from the fact that direct observation of the underlying principle through experiment is very difficult due to the high complexity of the boiling phenomenon. Also, since the heat transfer characteristics associated with boiling is the result of non-linear interaction of various physical parameters coupled to each other, it is challenging to find the relationship between the parameters for modeling. This paper discusses the authors’ past and ongoing boiling research under CASL (Consortium for Advanced Simulation of Light Water Reactors) program, including the efforts to overcome the difficulties of boiling measurement and model development. In particular, we focus on describing the major achievements, lessons learnt, and ongoing challenges for the CFD boiling closure model development based on the high-fidelity subcooled flow boiling experiment at Texas A&M University (TAMU). The major research achievements discussed through this paper include (i) novel boiling measurement strategy and high-fidelity validation data production, (ii) dedicated effort to identify the critical measurement issues in boiling experiment, (iii) new physical insight into sliding bubble heat transfer mechanisms, and (iv) new boiling closure model development. Discussion is also made on the ongoing challenges for the boiling closure model development and the future research plan based on the lessons learnt.

42 ENGINEERING↗
DOE OSTI2021

Initial use of Nek5000/Cardinal to improve closure models in Pronghorn

Heat transfer coefficient closure models for pebble bed reactors are built using a data-driven approach by leveraging the capabilities of an Evolutionary Algorithm entitled Particle Swarm Optimization (PSO). In the present work, the Computational Fluid Dynamics code nekRS was used in order to collect the high-fidelity flow data for a core with 1,568 pebbles. To characterize the heat transfer, multiple concentric regions were considered to extract the physical quantities of interest, e.g./ the Reynolds number. The PSO algorithm is employed as part of an inverse problem targeting determine what are the coefficients for a Nusselt number correlation to match the collected data. Such correlation should follow any given format that is defined a priori. Finally, two correlations are proposed, one with an implicit dependence on the pebbles’ wall temperatures and another expressed as a fully explicit correlation depending on the flow conditions and the position within the core. Anyway, given the generic nature of the proposed approach, correlations following different formats could be tested. Preliminary results for the high-fidelity simulation of a fast MSR core are presented. The target Reynolds number is currently 20K, with the expectation that this will increase, pending the availability of further computational resources. These simulations will be used to inform lower fidelity models, including a coarse CFD turbulence model in Pronghorn. Additionally, they will serve as a reference for the RANS models in Nek5000/NekRS.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗
NASA NTRS1974

Application of an invariant second-order closure model to compressible turbulent shear layers

A second-order closure model for two-dimensional, compressible shear flows is investigated using the invariant modeling technique developed by Donaldson. The invariant model parameters were originally selected by comparison of model predictions with critical experimental data on basic incompressible flows. Additional modeling for compressible flows has been introduced. A number of different shear flows such as the free shear layer, flat plate boundary layer and a simulation of the mixing region of a chemical laser are computed. The model is consistent with first-order closure turbulence models for equilibrium flows and is further capable of predicting nonequilibrium flows that cannot be correctly solved by eddy viscosity models. The results are in generally good agreement with experimental measurements but suggest the need for inclusion of a Mach number dependent model of the pressure diffusion terms in order to adequately represent high speed compressible flows.

Varma, A. K.↗