Search NASA⌕ Search

SEARCH · Search NASA

Results for “Direct numerical simulation”

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 991 records · Page 55

Co-optimized machine-learned manifold models for large eddy simulation of turbulent combustion

Many modeling approaches in large eddy simulation (LES) of turbulent combustion employ a projection of the thermochemical state onto a low-dimensional manifold within state space to reduce the number of transported variables and hence computational cost. Flamelet-generated manifolds (FGM) is an example of a well-established, physics-based approach, but increasingly, principal component analysis (PCA) is being used as a data-driven method for generating manifold models. For both approaches, the nonlinear relationship between the location on the predefined manifold and the outputs of interest, such as reaction rates, can be tabulated or encoded in a neural network. This work proposes a new approach for manifold modeling that extends these existing approaches. A modified neural network structure simultaneously encodes the definition of the manifold variables, the nonlinear mapping, and the subfilter closure for LES. This allows all three of these aspects of the model to be co-optimized, generating a model from any source of combustion thermochemical state data. The manifold parameterizing variables are constrained to be linear combinations of species, as in FGM and PCA-based models, to aid in interpretability and implementation. For LES, subfilter variances of the manifold variables are also included as inputs. Two types of a priori analysis are performed to evaluate the new approach. In the first, the model is trained on data from one-dimensional premixed flames. In this case, the approach recovers the behavior of flamelet-based manifold approaches, and in fact slightly improves performance by identifying an optimized progress variable. The approach is also applied to data from direct numerical simulations of spherical ignition kernels in isotropic turbulence. For any specified manifold dimensionality, the new approach provides substantially lower prediction errors than a PCA-based model developed from the same data set. Additionally, the LES formulation of the new approach can provide accurate predictions for filtered reaction rates across a variety of filter widths.

97 MATHEMATICS AND COMPUTING↗

Numerical and boundary condition effects on the prediction of detonation engine behavior using detailed numerical simulations

High-fidelity numerical simulations of an experimental rotating detonation engine with discrete fuel/air injection were conducted. A series of configurations with different feed-plenum pressures but with constant equivalence ratio were studied. Detailed chemical kinetics for the hydrogen/air system is used. A resolution study for the full rotating detonation engine (RDE) system simulation is also conducted. Two kinds of boundary conditions, a total pressure boundary and a constant mass flow rate boundary, are used to assess the effects of the inlet boundary. As mass flow rate is increased, the total pressure boundary causes more error in the axial pressure distribution while the constant mass flow rate gives a better solution for all cases ran. The simulations confirm experimental findings, and reproduce qualitative as well as some of the quantitative trends. These results demonstrate that a) fuel-air mixing is highly non-uniform within the detonation chamber, leading to variations in local equivalence ratio, b) the fuel and oxidizer injectors experience significant backflow as the detonation wave passes over, but recover at different rates which further augments the inefficiencies in mixing, and c) parasitic combustion in the mixing region makes the detonation wave weak by extending the reaction zone across the wave.

33 ADVANCED PROPULSION SYSTEMS↗

Mechanisms of flame stabilisation at low lifted height in a turbulent lifted slot-jet flame

A turbulent lifted slot-jet flame is studied using direct numerical simulation (DNS). A one-step chemistry model is employed with a mixture-fraction-dependent activation energy which can reproduce qualitatively the dependence of the laminar burning rate on the equivalence ratio that is typical of hydrocarbon fuels. The basic structure of the flame base is first examined and discussed in the context of earlier experimental studies of lifted flames. Several features previously observed in experiments are noted and clarified. Some other unobserved features are also noted. Comparison with previous DNS modelling of hydrogen flames reveals significant structural differences. The statistics of flow and relative edge-flame propagation velocity components conditioned on the leading edge locations are then examined. The results show that, on average, the streamwise flame propagation and streamwise flow balance, thus demonstrating that edge-flame propagation is the basic stabilisation mechanism. Fluctuations of the edge locations and net edge velocities are, however, significant. It is demonstrated that the edges tend to move in an essentially two-dimensional (2D) elliptical pattern (laterally outwards towards the oxidiser, then upstream, then inwards towards the fuel, then downstream again). It is proposed that this is due to the passage of large eddies, as outlined in Su et al. ( Combust. Flame , vol. 144 (3), 2006, pp. 494–512). However, the mechanism is not entirely 2D, and out-of-plane motion is needed to explain how flames escape the high-velocity inner region of the jet. Finally, the time-averaged structure is examined. A budget of terms in the transport equation for the product mass fraction is used to understand the stabilisation from a time-averaged perspective. The result of this analysis is found to be consistent with the instantaneous perspective. The budget reveals a fundamentally 2D structure, involving transport in both the streamwise and transverse directions, as opposed to possible mechanisms involving a dominance of either one direction of transport. Lastly, it features upstream transport balanced by entrainment into richer conditions, while on the rich side, upstream turbulent transport and entrainment from leaner conditions balance the streamwise convection.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A Study of Physics-Informed Deep Learning for System Fluid Dynamics Closures

The sub-grid-scale (SGS) physics models, or so-called closure relations (CR), are essential in thermal-fluid modeling and simulation codes, ranging from large-eddy simulation to system-level thermo-fluid dynamic problems. While details vary, all thermo-fluid codes solve coupled mass-momentum-energy conservation partial differential equations (PDE). Traditionally, the analytical form of closure models has to be known, and then we calibrate the model using data from a set of applicable experiments. Characteristically, it takes long effort to gain insights and mechanistic understanding needed for developing closure models. Such a model is a compact form of data. The long time required for new model development limits the pace of model application while dealing with new geometries, new coolants or system conditions, particularly in newly designed systems. Alternative to mechanistic and semi-analytical models, it is possible to discover the pattern of closure relations by applying the modern machine learning methods to the body of relevant experimental data. Deep learning (DL) is a relatively new branch in the machine learning family, and a major breakthrough has been achieved since Hinton introduces a fast algorithm to efficiently train a feed-forward neural network (NN) with multilayer perceptrons. Generally, any NN with more than two layers is considered as deep learning, and it contains a large number of hyper-parameters to achieve pattern recognition or regression for complex datasets. Hornik shows that multilayer NN is compatible with universal approximation theorem. There is no constraint for a multilayer NN to capture the properties of any measurable information. This capability makes DL attractive for closure models in a system thermal-hydraulic code. As system geometry and/or operating conditions change, new physical experiments (increasingly sophisticated) are conducted to support evaluation and revision of closure models. Such a so-called data-driven modeling relaxes the structural form of system codes to directly assimilate data instead of hard-wiring the model to pre-calibrated correlations. The objective of this research is to enable cost-effective development of a DL-based closure that reflects the data. The previous work on applying machine learning techniques to fluid dynamics focused on statistical learning with basic NN from direct numerical simulations (DNS) dataset. In the present work, we show that DL-based closure models can cause the fluid calculation to diverge if multilayer NN is used. To address the issue, we investigate conditions by which the DL-based closure models work compatibly, stably, and effectively with PDE-constrained forward prediction problems. Optimally meeting the compatibility conditions constitutes a physics-informed deep learning (PIDL) strategy. We refer to this process as PDE-constrained machine learning or PDE-ML. To develop the technical basis for PDE-ML using DL-based fluid dynamics closure, numerical experiments are performed for manufactured problems of increasing complexity. (authors)

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Modification of the Local Cubic Law of fracture flow for weak inertia, tortuosity, and roughness

The classical Local Cubic Law (LCL) generally overestimates flow through real fractures. We thus developed and tested a modified LCL (MLCL) which takes into account local tortuosity and roughness, and works across a low range of local Reynolds Numbers. The MLCL is based on (1) modifying the aperture field by orienting it with the flow direction and (2) correcting for local roughness changes associated with local flow expansion/contraction. In order to test the MLCL, we compared it with direct numerical simulations with the Navier-Stokes equations using real and synthetic three-dimensional rough-walled fractures, previous corrected forms of the LCL, and experimental flow tests. The MLCL performed well and the effective errors (δ) in volumetric flow rate range from -6.7% to 13.4% with an arithmetic mean of |δ| (<|δ|>) equal to 3.7%. The MLCL is more accurate than previous modifications of the LCL. We also investigated the error associated with applying the Cubic Law (CL) while utilizing modified aperture field. The δ from the CL ranges from -14.2% to 11.2%, with a slightly higher <|δ|> = 6.1% than the MLCL. The CL with the modified aperture field considering local tortuosity and roughness may also be sufficient for predicting the hydraulic properties of rough fractures.

58 GEOSCIENCES↗

Coherent motions and time scales that control heat and mass transfer at wind-swept water surfaces

Forecast of the heat and chemical budgets of lakes, rivers, and oceans requires improved predictive understanding of air-water interfacial transfer coefficients. Here in this paper we present laboratory observations of the coherent motions that occupy the air-water interface at wind speeds (U 10 ) 1.1–8.9 m/s. Spatiotemporal near-surface velocity data and interfacial renewal data are made available by a novel flow tracer method. The relative activity, velocity scales, and time scales of the various coherent interfacial motions are measured, namely for Langmuir circulations, streamwise streaks, nonbreaking wind waves, parasitic capillary waves, nonturbulent breaking wind waves, and turbulence-generating breaking wind waves. Breaking waves exhibit a sudden jump in streamwise interfacial velocity wherein the velocity jumps up to exceed the wave celerity and destroys nearby parasitic capillary waves. Four distinct hydrodynamic regimes are found to exist between U 10 =0 and 8.9 m/s, each with a unique population balance of the various coherent motions. The velocity scales, time scales, and population balance of the different coherent motions are input to a first-principles gas transfer model to explain the waterside transfer coefficient (k w ) as well as experimental patterns of temperature and gas concentration. The model mixes concepts from surface renewal and divergence theories and requires surface divergence strength (β), the Lagrangian residence time inside the upwelling zone (t Lu ), and the total lifetime of new interface before it is downwelled (t LT ). The model’s output agrees with time-averaged measurements k w , patterns of temperature in infrared photographs, and spatial patterns of gas concentration and k w from direct numerical simulations. Finally, several nondimensional parameters, e.g. βt Lu and τ s t LT where ss is the interfacial shear rate, determine the effectiveness of a particular type of coherent motion for affecting k w .

16 TIDAL AND WAVE POWER↗

Model of non-stationary, inhomogeneous turbulence

Here, we compare results from a spectral model for non-stationary, inhomogeneous turbulence (Besnard et al. in Theor Comp Fluid Dyn 8:1–35, 1996) with direct numerical simulation (DNS) data of a shear-free mixing layer (SFML) (Tordella et al. in Phys Rev E 77:016309, 2008). The SFML is used as a test case in which the efficacy of the model closure for the physical-space transport of the fluid velocity field can be tested in a flow with inhomogeneity, without the additional complexity of mean-flow coupling. The model is able to capture certain features of the SFML quite well for intermediate to long times, including the evolution of the mixing-layer width and turbulent kinetic energy. At short-times, and for more sensitive statistics such as the generation of the velocity field anisotropy, the model is less accurate. We propose two possible causes for the discrepancies. The first is the local approximation to the pressure-transport and the second is the a priori spherical averaging used to reduce the dimensionality of the solution space of the model, from wavevector to wavenumber space. DNS data are then used to gauge the relative importance of both possible deficiencies in the model.

42 ENGINEERING↗

Vorticity dynamics after the shock–turbulence interaction

In this article, the interaction of a shock wave with quasi-vortical isotropic turbulence (IT) represents a basic problem for studying some of the phenomena associated with high speed flows, such as hypersonic flight, supersonic combustion and Inertial Confinement Fusion (ICF). In general, in practical applications, the shock width is much smaller than the turbulence scales and the upstream turbulent Mach number is modest. In this case, recent high resolution shock-resolved Direct Numerical Simulations (DNS) (Ryu and Livescu, J Fluid Mech 756, R1, 2014) show that the interaction can be described by the Linear Interaction Approximation (LIA). Using LIA to alleviate the need to resolve the shock, DNS post-shock data can be generated at much higher Reynolds numbers than previously possible. Here, such results with Taylor Reynolds number approximately 180 are used to investigate the changes in the vortical structure as a function of the shock Mach number, M s , up to M s = 10. It is shown that, as M s increases, the shock interaction induces a tendency towards a local axisymmetric state perpendicular to the shock front, which has a profound influence on the vortex-stretching mechanism and divergence of the Lamb vector and, ultimately, on the flow evolution away from the shock.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Sensible-Enthalpy-Based Conditional Moment Closure Model for Homogeneous Charge Compression Ignition with Temperature Inhomogeneity

Abstract This paper presents a sensible-enthalpy-based conditional moment closure (CMC) model for homogeneous charge compression ignition (HCCI) with temperature inhomogeneity. The focus of study is placed on the effects of a conditioning variable on the prediction of heat release rates and developing the mixing models for sensible enthalpy, a reacting scalar. The sensible-enthalpy-based CMC model is validated a priori with 2-D direct numerical simulation (DNS) data for HCCI, and compared with the total-enthalpy-based formulation. Sensible enthalpy is shown to be strongly correlated with temperature, which makes sensible enthalpy well suited for a conditioning variable in CMC for HCCI. Results show that sensible-enthalpy-based CMC performs well for the prediction of heat release rates for all temperature stratification levels investigated in this study, while total-enthalpy-based CMC leads to the overprediction of heat release rates when the temperature stratification level is high. While conditional fluctuations of species mass fractions are substantial, sensible-enthalpy-based CMC is found to accurately estimate heat release rates during ignition, as well as reaction rates of key species, due to substantially suppressed conditional temperature fluctuations. The mapping closure models for the probability density function (PDF) and conditional dissipation rates of sensible enthalpy are proposed. The proposed models are found to capture the characteristics of reactive scalar mixing.

42 ENGINEERING↗

A multiscale study of damage in elastomeric syntactic foams

Damage mechanisms in elastomeric syntactic foams filled with glass microballoons (GMB) and resulting effects on the macroscale elastic constants have been investigated. Direct numerical simulations of the material microstructure, composite theory analyses, and uniaxial compression tests across a range of filler volume fractions were conducted. The room temperature and elastic behavior of composites with undamaged, fully debonded, and fully crushed GMBs were investigated for syntactic foams with a polydimethylsiloxane matrix. Good agreement was obtained between numerical studies, composite theory, and experiments. Debonding was studied via finite element models due to the difficulty of isolating this damage mechanism experimentally. The predictions indicate that the bulk modulus is insensitive to the state of debonding at low-GMB-volume fractions but is dramatically reduced if GMBs are crushed. The shear behavior is affected by both debonding and crush damage mechanisms. The acute sensitivity of the bulk modulus to crushed GMBs is further studied in simulations in which only a fraction of GMBs are crushed. We find that the composite bulk modulus drops severely even when just a small fraction of GMBs are crushed. Various material parameters such as GMB wall thickness, volume fraction, and minimum balloon spacing are also investigated, and they show that the results presented here are general and apply to a wide range of microstructure and GMB filler properties.

36 MATERIALS SCIENCE↗

Statistics of Point Vortex Turbulence in Non-neutral Flows and in Flows with Translational and Rotational Symmetries

A theory (Esler and Ashbee in J Fluid Mech 779:275–308, 2015) describing the statistics of N freely-evolving point vortices in a bounded two-dimensional domain is extended. First, the case of a non-neutral vortex gas is addressed, and it is shown that the density of states function can be identified with the probability density function of an infinite sum of independent non-central chi-squared random variables, the details of which depend only on the shape of the domain. Equations for the equilibrium energy spectrum and other statistical quantities follow, the validity of which are verified against direct numerical simulations of the equations of motion. Second, domains with additional conserved quantities associated with a symmetry (e.g., circle, periodic channel) are investigated, and it is shown that the treatment of the non-neutral case can be modified to account for the additional constraint.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Flow and axial dispersion in a sinusoidal-walled tube: Effects of inertial and unsteady flows

Dispersion in porous media flows has been the subject of much experimental, theoretical and numerical study. Here we consider a wavy-walled tube (a three-dimensional tube with sinusoidally-varying diameter) as a simplified conceptualization of flow in porous media, where constrictions represent pore throats and expansions pore bodies. A theoretical model for effective (macroscopic) longitudinal dispersion in this system has been developed by volume averaging the microscale velocity field. Direct numerical simulation using computational fluid dynamics (CFD) methods was used to compute velocity fields by solving the Navier-Stokes equations, and also to numerically solve the volume averaging closure problem, for a range of Reynolds numbers (Re) spanning the low-Re to inertial flow regimes, including one simulation at Re = 449 for which unsteady flow was observed. Dispersion values were computed using both the volume averaging solution and a random walk particle tracking method, and results of the two methods were shown to be consistent. Our results are compared to experimental measurements of dispersion in porous media and to previous theoretical results for the low-Re, Stokes flow regime. In the steady inertial regime we observe an power-law increase in effective longitudinal dispersion (DL) with Re, consistent with previous results. This rapid rate of increase is caused by trapping of solute in expansions due to flow separation (eddies). For the unsteady case (Re = 449), the rate of increase of DL with Re was smaller than that observed at lower Re. Velocity fluctuations in this regime lead to increased rates of solute mass transfer between the core flow and separated flow regions, thus diminishing the amount of tailing caused by solute trapping in eddies and thereby reducing longitudinal dispersion.

CFD↗

Temporal breakdown and Borel resummation in the complex Langevin method

We reexamine the Parisi-Klauder conjecture for complex e{sup i{theta}/2}{phi}{sup 4} measures with a Wick rotation angle 0{<=}{theta}/2{<=}{pi}/2 interpolating between Euclidean signature and Lorentzian signature. Our main result is that the asymptotics for short stochastic times t encapsulates information also about the equilibrium aspects. The moments evaluated with the complex measure and with the real measure defined by the stochastic Langevin equation have the same t{yields}0 asymptotic expansion which is shown to be Borel summable. The Borel transform correctly reproduces the time dependent moments of the complex measure for all t, including their t{yields}{infinity} equilibrium values. On the other hand the results of a direct numerical simulation of the Langevin moments are found to disagree from the 'correct' result for t larger than a finite t{sub c}. The breakdown time t{sub c} increases powerlike for decreasing strength of the noise's imaginary part but cannot be excluded to be finite for purely real noise. To ascertain the discrepancy we also compute the real equilibrium distribution for complex noise explicitly and verify that its moments differ from those obtained with the complex measure. - Highlights: Black-Right-Pointing-Pointer The Parisi-Klauder conjecture is reexamined for complex e{sup i{theta}/2}{phi}{sup 4} measures. Black-Right-Pointing-Pointer The time dependent moments are evaluated by temporal Borel resummation. Black-Right-Pointing-Pointer The results disagree with the Langevin simulations beyond a critical time t{sub c}. Black-Right-Pointing-Pointer t{sub c} increases with decreasing strength of the noise's imaginary part. Black-Right-Pointing-Pointer The technical reason for the breakdown is identified.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Nonlinear localized modes in PT-symmetric optical media with competing gain and loss

The existence and stability of the nonlinear spatial localized modes are investigated in parity-time symmetric optical media characterized by a generic complex hyperbolic refractive index distribution with competing gain and loss profile. The exact analytical expression of the localized modes are found for all values of the competing parameter and in the presence of both the self-focusing and self-defocusing Kerr nonlinearity. The effects of competing gain/loss profile on the stability structure of these localized modes are discussed with the help of linear stability analysis followed by the direct numerical simulation of the governing equation. The spatial localized modes in two-dimensional geometry as well as the transverse power-flow density associated with these localized modes are also examined. -- Highlights: • Existence of localized modes is investigated in PT-symmetric complex potentials. • Exact analytical expression of the localized modes is obtained. • Effect of gain/loss profile on the stability of these localized modes is discussed. • Localized modes in 2D and associated transverse power-flow density are also examined.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Gaussian-type light bullet solutions of the (3+1)-dimensional Schrödinger equation with cubic and power-law nonlinearities in PT-symmetric potentials

Two kinds of Gaussian-type light bullet (LB) solutions of the (3+1)-dimensional Schrödinger equation with cubic and power-law nonlinearities in PT-symmetric potentials are analytically obtained. The phase switches, powers and transverse power-flow densities of these solutions in homogeneous media are studied. The linear stability analysis of these LB solutions and the direct numerical simulation indicate that LB solutions are stable below some thresholds for the imaginary part of PT-symmetric potentials in the defocusing cubic and focusing power-law nonlinear medium, while they are always unstable for all parameters in other media. Moreover, the broadened and compressed behaviors of LBs in the exponential periodic amplification system and diffraction decreasing system are discussed. Results indicate that LB is more stable for the sign-changing nonlinearity in the exponential periodic amplification system than for the non-sign-changing nonlinearity in the diffraction decreasing system at the same propagation distances.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Beyond the effectiveness factor: Multi-step reactions with intraparticle diffusion limitations

Though common in many industrial chemical processes, multi-step reactions occurring within porous catalyst particles provide significant challenges to previously established modeling frameworks. In this work, a system of reaction-diffusion equations is solved analytically via eigenanalysis. The intraparticle concentration profiles are integrated over the particle volume to obtain effectiveness factors for each reacting species, referred to herein as a multi-step effectiveness vector (MEV). The MEV solution is tested for the case of uniform flow past a stationary catalyst particle with a state-of-the-art, multi-step reaction scheme for fluid catalytic cracking (FCC). Outputs from particle-resolved direct numerical simulation (DNS) are compared to outputs from ordinary differentials with effective reaction closures coming from the MEV solution, the classic single-step effectiveness factor (SEF) (R. Aris, 'The mathematical theory of diffusion and reaction in permeable catalysts: The theory of the steady state,' Oxford University Press, 1975), and utilizing the intrinsic rates as-is. When compared to DNS, the MEV solution shows excellent agreement for the formation of all species and quantitatively captures the intraparticle concentration profiles. By contrast, the intrinsic treatment shows an over-prediction for product formation and the SEF treatment shows mixed levels of accuracy that depend upon the reaction coupling. Specifically, the SEF treatment accurately characterizes product formation if the kinetic scheme is weakly coupled but shows significant departure if stronger coupling is present. The computational cost associated with the MEV solution is associated with eigendecomposition of the Thiele matrix, an N x N matrix where N is the number of reacting components. Overall, the method is shown to be relatively easy to implement and may significantly enhance the fidelity of chemical reactor simulations that employ multi-step reaction schemes.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

A discontinuous Galerkin spectral element method for compressible reacting flows

High-order methods have recently been shown to be an effective tool for high-fidelity flow computations like direct numerical simulations and large-eddy simulations because of their strong balance between accuracy and computational cost. In this work, a high-order discontinuous Galerkin spectral element method (DGSEM) is developed to solve the chemically reacting Navier-Stokes equations. To handle the disparate length and time scales associated with these equations, we develop a novel method which combines the spectral accuracy of the SEM with the flexibility of the DG approach. The framework, implemented in the spectral element code Nek5000, is well suited to capture turbulence in smooth regions of the flow, while maintaining numerical stability in the presence of shocks. An entropy-residual based artificial viscosity is added to smooth shocked regions of flow, and a positivity-preserving limiter is implemented to suppress non-physical oscillations. These enhancements support the numerical stability of the hydrodynamic sub-step, which is decoupled from the chemistry integration through a second-order operator splitting method. Here, a series of smooth and discontinuous validation cases are presented in increasing physical and computational complexity for both inviscid and viscous flows. In particular, simulations of canonical one-dimensional and two-dimensional detonations are performed, and the high-order numerical results are validated against available literature data. Additional validation studies are carried out for classical three-dimensional numerical simulations of incompressible and compressible turbulent flows.

Compressible reacting flows↗

An artificial neural network framework for reduced order modeling of transient flows

Here, this paper proposes a supervised machine learning framework for the non-intrusive model order reduction of unsteady fluid flows to provide accurate predictions of non-stationary state variables when the control parameter values vary. Our approach utilizes a training process from full-order scale direct numerical simulation data projected on proper orthogonal decomposition (POD) modes to achieve an artificial neural network (ANN) model with reduced memory requirements. This data-driven ANN framework allows for a nonlinear time evolution of the modal coefficients without performing a Galerkin projection. Our POD-ANN framework can thus be considered an equation-free approach for latent space dynamics evolution of nonlinear transient systems and can be applied to a wide range of physical and engineering applications. Within this framework we introduce two architectures, namely sequential network (SN) and residual network (RN), to train the trajectory of modal coefficients. We perform a systematic analysis of the performance of the proposed reduced order modeling approaches on prediction of a nonlinear wave-propagation problem governed by the viscous Burgers equation, a simplified prototype setting for transient flows. We find that the POD-ANN-RN yields stable and accurate results for test problems assessed both within inside and outside of the database range and performs significantly better than the standard intrusive Galerkin projection model. Finally, our results show that the proposed framework provides a non-intrusive alternative to the evolution of transient physics in a POD basis spanned space, and can be used as a robust predictive model order reduction tool for nonlinear dynamical systems.

42 ENGINEERING↗