Search NASA⌕ Search

SEARCH · Search NASA

Results for “data flow statistics”

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 91 records · Page 5

High-Fidelity Velocity and Concentration Measurements of Turbulent Buoyant Jets

Accurate models of turbulent buoyant flows are essential for the design of the cooling circuit of nuclear reactors and passive safety systems. However, available models fail to fully capture the physics of turbulent mixing when buoyancy becomes predominant with respect to momentum. Therefore, high-fidelity experiments of well-controlled fundamental flows are needed to develop and validate more accurate models. We analyze experiments of positive and negative turbulent buoyant jets, both in uniform and stratified environments, with the aim of understanding the thermal hydraulics of turbulent mixing with variable density and providing high-fidelity data for the development and validation of turbulence models. Non-intrusive, simultaneous particle image velocimetry and laser-induced fluorescence measurements were carried out to acquire instantaneous velocity and concentration fields on a vertical section parallel to the axis of a jet in the self-similar region. The refractive index matching method was applied to measure high-resolution buoyant jets with up to 8.6% density difference. These data are free of the typical errors that characterize optical measurements of buoyancy-driven flows (e.g. natural and mixed convection) where the refractive index of the fluid is inhomogeneous throughout the measurement domain. Turbulent statistics and entrainment of buoyant jets in uniform and stratified environments are presented. These data are compared with non-buoyant jets in a uniform environment, as a reference to investigate the effects of buoyancy and stratification on turbulent mixing. The results will be used for the assessment of current turbulence models and as a basis for the development of a new model that captures turbulent mixing.

laser induced fluorescence↗

Statistical study of reconnection exhausts in the solar wind

Magnetic reconnection is a fundamental process that changes magnetic field configuration and converts a magnetic energy to flow energy and plasma heating. This paper presents a survey of the plasma and magnetic field parameters inside 418 reconnection exhausts identified in the WIND data from 1995-2012. The statistical analysis is oriented on the re-distribution of the magnetic energy released due to reconnection between a plasma acceleration and its heating. The results show that both the portion of the energy deposited into heat as well as the energy spent on the acceleration of the exhaust plasma rise with the magnetic shear angle in accord with the increase of the magnetic flux available for reconnection. The decrease of the normalized exhaust speed with the increasing magnetic shear suggests a decreasing efficiency of the acceleration and/or the increasing efficiency of heating in high-shear events. However, we have found that the already suggested relation between the exhaust speed and temperature enhancement would be rather considered as an upper limit of the plasma heating during reconnection regardless of the shear angle.

79 ASTRONOMY AND ASTROPHYSICS↗

Large-deviation joint statistics of the finite-time Lyapunov spectrum in isotropic turbulence

One of the hallmarks of turbulent flows is the chaotic behavior of fluid particle paths with exponentially growing separation among them while their distance does not exceed the viscous range. The maximal (positive) Lyapunov exponent represents the average strength of the exponential growth rate, while fluctuations in the rate of growth are characterized by the finite-time Lyapunov exponents (FTLEs). In the last decade or so, the notion of Lagrangian coherent structures (which are often computed using FTLEs) has gained attention as a tool for visualizing coherent trajectory patterns in a flow and distinguishing regions of the flow with different mixing properties. A quantitative statistical characterization of FTLEs can be accomplished using the statistical theory of large deviations, based on the so-called Cramér function. To obtain the Cramér function from data, we use both the method based on measuring moments and measuring histograms and introduce a finite-size correction to the histogram-based method. We generalize the existing univariate formalism to the joint distributions of the two FTLEs needed to fully specify the Lyapunov spectrum in 3D flows. The joint Cramér function of turbulence is measured from two direct numerical simulation datasets of isotropic turbulence. Results are compared with joint statistics of FTLEs computed using only the symmetric part of the velocity gradient tensor, as well as with joint statistics of instantaneous strain-rate eigenvalues. When using only the strain contribution of the velocity gradient, the maximal FTLE nearly doubles in magnitude, highlighting the role of rotation in de-correlating the fluid deformations along particle paths. We also extend the large-deviation theory to study the statistics of the ratio of FTLEs. The most likely ratio of the FTLEs λ{sub 1} : λ{sub 2} : λ{sub 3} is shown to be about 4:1:−5, compared to about 8:3:−11 when using only the strain-rate tensor for calculating fluid volume deformations. The results serve to characterize the fundamental statistical and geometric structure of turbulence at small scales including cumulative, time integrated effects. These are important for deformable particles such as droplets and polymers advected by turbulence.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Data-driven Mori–Zwanzig modeling of Lagrangian particle dynamics in turbulent flows

The dynamics of Lagrangian particles in turbulence play a crucial role in mixing, transport, and dispersion in complex flows. Their trajectories exhibit highly nontrivial statistical behavior, motivating the development of surrogate models that can reproduce these trajectories without incurring the high computational cost of direct numerical simulations of the full Eulerian field. This task is particularly challenging because reduced-order models typically lack access to the full set of interactions with the underlying turbulent field. Novel data-driven machine learning techniques can be powerful in capturing and reproducing complex statistics of the reduced-order/surrogate dynamics. In this work, we show how one can learn a surrogate dynamical system that is able to evolve a turbulent Lagrangian trajectory in a way that is point-wise accurate for short-time predictions (with respect to Kolmogorov time) and stable and statistically accurate at long times. This approach is based on the Mori–Zwanzig formalism, which prescribes a mathematical decomposition of the full dynamical system into resolved dynamics that depend on the current state and the past history of a reduced set of observables, and the unresolved orthogonal dynamics due to unresolved degrees of freedom of the initial state. We show how by training this reduced order model on a point-wise error metric on short time-prediction, we are able to correctly learn the dynamics of Lagrangian turbulence, such that also the long-time statistical behavior is stably recovered at test time. This opens up a range of applications, for example, for the control of active Lagrangian agents in turbulence.

97 MATHEMATICS AND COMPUTING↗

Physics-informed machine learning of the Lagrangian dynamics of velocity gradient tensor

Reduced models describing the Lagrangian dynamics of the velocity gradient tensor (VGT) in homogeneous isotropic turbulence (HIT) are developed under the physics-informed machine learning (PIML) framework. We consider the VGT at both Kolmogorov scale and coarse-grained scale within the inertial range of HIT. Building reduced models requires resolving the pressure Hessian and subfilter contributions, which is accomplished by constructing them using the integrity bases and invariants of the VGT. The developed models can be expressed using the extended tensor basis neural network (TBNN) introduced by Ling et al. [J. Fluid Mech. 807, 155 (2016)]. Physical constraints, such as Galilean invariance, rotational invariance, and incompressibility condition, are thus embedded in the models explicitly. Our PIML models are trained on the Lagrangian data from a high-Reynolds number direct numerical simulation (DNS). To validate the results, we perform a comprehensive out-of-sample test. We observe that the PIML model provides an improved representation for the magnitude and orientation of the small-scale pressure Hessian contributions. Statistics of the flow, as indicated by the joint PDF of second and third invariants of the VGT, show good agreement with the “ground-truth” DNS data. A number of other important features describing the structure of HIT are reproduced by the model successfully. We have also identified challenges in modeling inertial range dynamics, which indicates that a richer modeling strategy is required. This helps us identify important directions for future research, in particular towards including inertial range geometry into the TBNN.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Freight Analysis Framework Version 5 (FAF5) Base Year 2017 Data Development Technical Report

The Freight Analysis Framework (FAF) integrates data from a variety of sources to create a comprehensive national picture of freight movements among states and major metropolitan areas by all modes of transportation. The latest of this data series is FAF5, which is the fifth generation FAF and is benchmarked on Commodity Flow Survey (CFS) 2017. Except for FAF1 that provided estimates for truck, rail, and water tonnage for calendar year 1998, later generations of FAF (FAF2 through FAF5) were built based on their benchmark year CFS data, for 2002, 2007, 2012, and 2017 respectively. The FAF is produced under a partnership between Bureau of Transportation Statistics (BTS) and Federal Highway Administration (FHWA). As a major data product of the FAF program, the FAF regional database provides a national picture of freight flows to, from, and within the United States (among regions and states), by commodity and mode for the base year, as well as for forecasts up to 30 years into the future in a 5-year interval. Additional FAF data products also include FAF network flows database, where truck movements are routed onto the national highway network, estimates of annual projections, and synchronized historical data series. This report is a technical document prepared to describe the data sources and methodologies applied in the process of building the FAF5 base-year 2017 regional database, released as FAF5.0 in February 2021. This report offers a description of the diverse data sources and modeling methods used in constructing the base year FAF5 regional database. The FAF5 base-year database is used as the base for development of forecasts and for assignment of truck flows on highway network. Similarly, the FAF5 base-year database will be used as the base to generate FAF5 annual estimates. In addition to this report, users are encouraged to refer to the FAF5 User’s Guide, which provides basic information of the data, including definitions of the data attributes, information on how to access the data and tool, as well as detailed data dictionary and code tables.

42 ENGINEERING↗

Isotherms for Water Adsorption on Molecular Sieve 3A: Influence of Cation Composition

This study is part of our continuing efforts to address engineering issues related to the removal of tritiated water from off-gases produced in used nuclear fuel reprocessing facilities. In the current study, adsorption equilibrium of water on molecular sieve 3A beads was investigated. Adsorption isotherms for water on the UOP molecular sieve 3A were measured by a continuous-flow adsorption system at 298, 313, 333, and 353 K. Experimental data collected were analyzed by the Generalized Statistical Thermodynamic Adsorption (GSTA) isotherm model. The K + /Na + molar ratio of this particular type of molecular sieve 3A was ~4:6. Our results showed that the GSTA isotherm model worked very well to describe the equilibrium behavior of water adsorption on molecular sieve 3A. The optimum number of parameters for the current experimental data was determined to be a set of four equilibrium parameters. This result suggests that the adsorbent crystals contain four energetically distinct adsorption sites. In addition, it was found that water adsorption on molecular sieve 3A follows a three-stage adsorption process. This three-stage adsorption process confirmed different water adsorption sites in molecular sieve crystals. In addition, the second adsorption stage is significantly affected by the K + /Na + molar ratio. In this stage, the equilibrium adsorption capacity at a given water vapor pressure increases as the K + /Na + molar ratio increases.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Heat-release equation of state and thermal conductivity of warm dense carbon by proton differential heating

Warm dense carbon is generated at 0.3–2.0 g/cc and 1–7 eV by proton heating. The release equation of state (EOS) after heating and thermal conductivity of warm dense carbon are studied experimentally in this regime using a Au/C dual-layer target to initiate a temperature gradient and two picosecond time-resolved diagnostics to probe the surface expansion and heat flow. Comparison between the data and simulations using various EOSs and thermal conductivity models is quantified with a statistical χ2 analysis. In conclusion, out of seven EOS tables and five thermal conductivity models, only L9061 with the Lee-More model provides a probability above 50% to match all data.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Atypical Behavior Identification in Large Scale Network Traffic

Cyber analysts are faced with the daunting challenge of identifying exploits and threats within potentially billions of daily records of network traffic. Enterprise-wide cyber traffic involves hundreds of millions of distinct IP addresses and results in data sets ranging from terabytes to petabytes of raw data. Creating behavioral models and identifying trends based on those models requires data intensive architectures and techniques that can scale as data volume increases. Analysts need scalable visualization methods that foster interactive exploration of data and enable identification of behavioral anomalies. Developers must carefully consider application design, storage, processing, and display to provide usability and interactivity with large-scale data. We present an application that highlights atypical behavior in enterprise network flow records. This is accomplished by utilizing data intensive architectures to store the data, aggregation techniques to optimize data access, statistical techniques to characterize behavior, and a visual analytic environment to render the behavioral trends, highlight atypical activity, and allow for exploration.

99 GENERAL AND MISCELLANEOUS↗

Discrete Mathematical Approaches to Graph-Based Traffic Analysis

Modern cyber defense and anlaytics requires general, formal models of cyber systems. Multi-scale network models are prime candidates for such formalisms, using discrete mathematical methods based in hierarchically-structured directed multigraphs which also include rich sets of labels. An exemplar of an application of such an approach is traffic analysis, that is, observing and analyzing connections between clients, servers, hosts, and actors within IP networks, over time, to identify characteristic or suspicious patterns. Towards that end, NetFlow (or more generically, IPFLOW) data are available from routers and servers which summarize coherent groups of IP packets flowing through the network. In this paper, we consider traffic analysis of Netflow using both basic graph statistics and two new mathematical measures involving labeled degree distributions and time interval overlap measures. We do all of this over the VAST test data set of 96M synthetic Netflow graph edges, against which we can identify characteristic patterns of simulated ground-truth network attacks.

Joslyn, Cliff A.↗

Multiscale Flow for robust and optimal cosmological analysis

We propose Multiscale Flow, a generative Normalizing Flow that creates samples and models the field-level likelihood of two-dimensional cosmological data such as weak lensing. Multiscale Flow uses hierarchical decomposition of cosmological fields via a wavelet basis and then models different wavelet components separately as Normalizing Flows. The log-likelihood of the original cosmological field can be recovered by summing over the log-likelihood of each wavelet term. This decomposition allows us to separate the information from different scales and identify distribution shifts in the data such as unknown scale-dependent systematics. The resulting likelihood analysis can not only identify these types of systematics, but can also be made optimal, in the sense that the Multiscale Flow can learn the full likelihood at the field without any dimensionality reduction. We apply Multiscale Flow to weak lensing mock datasets for cosmological inference and show that it significantly outperforms traditional summary statistics such as power spectrum and peak counts, as well as machine learning–based summary statistics such as scattering transform and convolutional neural networks. We further show that Multiscale Flow is able to identify distribution shifts not in the training data such as baryonic effects. Finally, we demonstrate that Multiscale Flow can be used to generate realistic samples of weak lensing data.

79 ASTRONOMY AND ASTROPHYSICS↗

Particle Loading Dynamics of SRNL Electrostatic Particle Collectors - 19398

The Department of Energy is interested in decontamination methods and tools for the removal of airborne radioactive particulate from facilities prior to decommissioning. Electrostatic particle collectors (EPCs) have historically been used industrially to scrub particles from air streams for large scale operations, and Savannah River National Laboratory (SRNL) researchers have expanded on the fundamental collection capabilities of EPCs to develop unique samplers focused on particulate effluent collection. One distinct advantage of an electrostatic particle collection over filtration or inertial impaction-based methods is the low pressure drop associated with prolonged use of EPCs. Electrostatic particle collectors previously developed by SRNL have proven effective in the collection of atmospheric aerosols in multiple field trials, and their applicability as a decontamination method for airborne particulates. Initial experiments using aerosolized NaCl were performed to determine EPC collection efficiencies, the times required to achieve a desired particle loading, and the size distribution of collected particulates. The amount of NaCl collected on Si substrates was found to increase linearly with time (∼ 2 mg per hour) and showed little sign of change after 24 hours of collection. Experiments showed that NaCl, or a similar aerosol, would likely continue to collect unperturbed until physical limitations within the EPC are reached. Collection efficiencies (i.e., the number of particles collected vs. the number of particles entering the EPC) were calculated independently using an assessment scheme based on Particle Depletion Dynamics (PDD) and by Computer Controlled Scanning Electron Microscope (CCSEM) analysis of collection substrates for a range of particle sizes. For the ACE 2.0 with a corona voltage of 15 kV, collection efficiencies were determined using PDD to be 54, 29, and 13% for the flow rates 100, 300, and 490 L/min respectively. While CCSEM data suggests a very high collection efficiency for the MiniACE, the static sampling rate of 28.1 L/min (1 ft{sup 3}/min) was found to be too low volume for PDD to provide statistically meaningful data. Number density maps of NaCl particles detected via CCSEM revealed that particles are preferentially collected near the EPC inlet (i.e., opposite the direction of airflow) when using slower flow rates. Particles collected using higher flow rates (e.g., 300 and 490 L/min) were collected more evenly across the substrate. When segregated by size, smaller (e.g., 200-500 nm) particles are collected evenly across the collection substrates. As particle sizes increase they are collected preferentially near the EPC inlet, but not in sufficient quantity to obstruct airflow and hinder long-term operations. Subsequent deployment of the SRNL developed system into plutonium-238 contaminated hot cells will be performed at Savannah River Site Building 235-F. Presented in this work are the testing and considerations utilized by SRNL to show this technique as a viable means of rapidly decontaminating airborne contaminated areas. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Generating Emissions Inventory for Carbon Capture and Storage Analysis for Carbon-Intensive Industrial Sectors

Decarbonizing the industrial sector is critical to achieve carbon dioxide (CO2) emissions reductions goals of the Biden Administration. Currently available decarbonization options include electrification, fuel switching to zero carbon fuels like green hydrogen (H2) and carbon capture and storage (CCS). Application of post-combustion carbon capture (PCCC) technology in the power sector, as well as research at the U.S. Department of Energy's Fossil Energy and Carbon Management (FECM) Office has shown that its application in the industrial sector could have co-benefits in the form of emissions reductions of non-CO2 regulated pollutants. For example, solvent based PCCC systems require pre-conditioning of flue gas to remove sulfur and particulate matter (PM) upstream of the CO2 absorber. However, there is a lack of understanding about the type of non-CO2 pollutants which can be captured and the amount of reduction possible. PCCC application differs across industrial sectors as it depends on the availability of decarbonization options, characteristics of industrial processes and the amount and composition of pollutant flows. Certain facilities can also have multiple effluent flows with or without a CO2 stream. As such, understanding industrial processes and their effluent flows in detail is required to quantify the co-benefits opportunities presented by PCCC. Considering this requirement, the goal of this analysis is to develop a high-resolution inventory of effluent flows from facilities of 8 industrial sectors in the U.S. These industrial sectors - ethanol, ammonia, cement, steel, natural gas processing, hydrogen, petroleum refining and wood and pulp products - have carbon-intensive effluent flows, and thus are prime candidates for PCCC applications. In this study, we map the composition of pollutant flow from flue stacks across the identified facilities. Using data available in three Environmental Protection Agency (EPA) databases - the Green House Gas Reporting Program (GHGRP), the National Emissions Inventory (NEI) and the Toxic Release Inventory (TRI), we create a combined inventory which lists the type, amount, and concentration of pollutant flows. Using total weight of the pollutant flow back calculated from observed data for CO2 concentrations in flue gas for individual sectors, we calculate the concentration of each pollutant in the flue gas stream. Thus, the resultant emissions inventory includes the following details for each facility in the sector: facility-level and if possible, process-level pollutant flows, concentrations of pollutants in the flue gas, and geographical coordinates of the facilities. A detailed statistical analysis and summary allows us to search for erroneous data and remove them from the final inventory. The generation of the inventory is achieved using a python-based framework which can recreate this inventory for other industrial sectors as well as using newer releases of emission inventories from EPA. The statistical analysis performed on the inventory is also calibrated and automated to identify outliers efficiently.

air pollutants↗

Data-driven Feature Analysis in Control Design for Series-Compensated Transmission Systems

One challenge in power-system control designs is the gap between numerical model-based analysis and complex real-world power systems. With increased data and measurements being collected from power systems, data-driven analysis (e.g., machine learning) may provide an alternative approach to reveal hidden information through learning from the real system data, and provide insights for better control scheme design during the utility planning process. Data-driven feature analysis was performed to evaluate the relationships between series compensation, power generation, and path flows in a real transmission system, as well as temporal patterns. The main data-driven analysis methods, including statistical cross-correlation, multinomial logistical regression, and classification and regression trees, were integrated for feature selection and developing predictive models of series compensation. Analysis results demonstrated the effectiveness of the proposed methodology in feature analysis and the potential to help improve power-system control scheme design.

42 ENGINEERING↗

Large-Eddy Simulation of Flow Over Boeing Gaussian Bump Using Multiagent Reinforcement Learning Wall Model: Preprint

We develop a wall model for large-eddy simulation (LES) that takes into account various pressure-gradient effects using multi-agent reinforcement learning. The model is trained using low-Reynolds-number flow over periodic hills with agents distributed on the wall at various computational grid points. It utilizes a wall eddy-viscosity formulation as the boundary condition to apply the modeled wall shear stress. Each agent receives states based on local instantaneous flow quantities at an off-wall location, computes a reward based on the estimated wall-shear stress, and provides an action to update the wall eddy viscosity at each time step. The trained wall model is validated in wall-modeled LES of flow over periodic hills at higher Reynolds numbers, and the results show the effectiveness of the model on flow with pressure gradients. The analysis of the trained model indicates that the model is capable of distinguishing between the various pressure gradient regimes present in the flow. To further assess the robustness of the developed wall model, simulations of flow over the Boeing Gaussian bump are conducted at a Reynolds number of 2 x 10^6, based on the free-stream velocity and the bump width. The results of mean skin friction and pressure on the bump surface, as well as the velocity statistics of the flow field, are compared to those obtained from equilibrium wall model (EQWM) simulations and published experimental data sets. The developed wall model is found to successfully capture the acceleration and deceleration of the turbulent boundary layer on the bump surface, providing better predictions of skin friction near the bump peak and exhibiting comparable performance to the EQWM with respect to the wall pressure and velocity field. We also conclude that the subgrid-scale model is crucial to the accurate prediction of the flow field, in particular the prediction of separation.

boundary layer↗

Constraining the local variance of H {sub 0} from directional analyses

We evaluate the local variance of the Hubble Constant H {sub 0} with low-z Type Ia Supernovae (SNe). Our analyses are performed using a hemispherical comparison method in order to test whether taking the bulk flow motion into account can reconcile the measurement of the Hubble Constant H {sub 0} from standard candles ( H {sub 0} = 73.8±2.4 km s{sup -1} Mpc {sup -1}) with that of the Planck's Cosmic Microwave Background data ( H {sub 0} = 67.8 ± 0.9km s{sup -1} Mpc{sup -1}). We obtain that H {sub 0} ranges from 68.9±0.5 km s{sup -1} Mpc{sup -1} to 71.2±0.7 km s{sup -1} Mpc{sup -1} through the celestial sphere (1 σ uncertainty), implying a Hubble Constant maximal variance of δ H {sub 0} = (2.30±0.86) km s{sup -1} Mpc{sup -1} towards the ( l,b ) = (315°,27°) direction. Interestingly, this result agrees with the bulk flow direction estimates found in the literature, as well as previous evaluations of the H {sub 0} variance due to the presence of nearby inhomogeneities. We assess the statistical significance of this result with different prescriptions of Monte Carlo simulations, obtaining moderate statistical significance, i.e., 68.7% confidence level (CL) for such variance. Furthermore, we test the hypothesis of a higher H {sub 0} value in the presence of a bulk flow velocity dipole, finding some evidence for this result which, however, cannot be claimed to be significant due to the current large uncertainty in the SNe distance modulus. Then, we conclude that the tension between different H {sub 0} determinations can plausibly be caused to the bulk flow motion of the local Universe, even though the current incompleteness of the SNe data set, both in terms of celestial coverage and distance uncertainties, does not allow a high statistical significance for these results or a definitive conclusion about this issue.

79 ASTRONOMY AND ASTROPHYSICS↗

A hybrid point-particle force model that combines physical and data-driven approaches

Here this study improves upon the physics-based pairwise interaction extended point-particle (PIEP) model. The PIEP model leverages our physical understanding to predict fluid mediated interactions between solid particles. By considering the relative location of neighboring particles, the PIEP model is able to provide better predictions than existing drag models. While the current physical PIEP model is a powerful tool, its assumptions lead to increased error in flows with higher particle volume fractions. To reduce this error, a regression algorithm makes direct use of the results of direct numerical simulations (DNS) of an array of monodisperse solid particles subjected to uniform ambient flow at varying Reynolds numbers. The resulting statistical model and the physical PIEP model are superimposed to construct a hybrid, physics-based data-driven PIEP model. It must be noted that the performance of a pure data-driven approach without the model-form provided by the physical PIEP model is substantially inferior. The hybrid model's predictive capabilities are analyzed using additional DNS data that was not part of training the data-driven model. In every case tested, the hybrid models resulting from the regression were capable of (1) improving upon the physical PIEP model's prediction and (2) recovering underlying relevant physics from the DNS data. As the particle volume fraction increases, the physical PIEP model loses the ability to approximate the forces experienced by the particles, but the statistical model continues to produce accurate approximations.

97 MATHEMATICS AND COMPUTING↗

Data-driven learning of Mori–Zwanzig operators for isotropic turbulence

Developing reduced-order models for turbulent flows, which contain dynamics over a wide range of scales, is an extremely challenging problem. In statistical mechanics, the Mori–Zwanzig (MZ) formalism provides a mathematically exact procedure for constructing reduced-order representations of high-dimensional dynamical systems, where the effects due to the unresolved dynamics are captured in the memory kernel and orthogonal dynamics. Turbulence models based on MZ formalism have been scarce due to the limited knowledge of the MZ operators, which originates from the difficulty in deriving MZ kernels for complex nonlinear dynamical systems. In this work, we apply a recently developed data-driven learning algorithm, which is based on Koopman's description of dynamical systems and Mori's linear projection operator, on a set of fully resolved isotropic turbulence datasets to extract the Mori–Zwanzig operators. With data augmentation using known turbulence symmetries, the extracted Markov term, memory kernel, and orthogonal dynamics are statistically converged and the generalized fluctuation–dissipation relation can be verified. The properties of the memory kernel and orthogonal dynamics, and their dependence on the choices of observables are investigated to address the modeling assumptions that are commonly used in MZ-based models. A series of numerical experiments are then constructed using the extracted kernels to evaluate the memory effects on prediction. The results show that the prediction errors are strongly affected by the choice of observables and can be further reduced by including the past history of the observables in the memory kernel.

97 MATHEMATICS AND COMPUTING↗