SEARCH · Engineering Papers
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.
The relative advantages and disadvantages of Direct Numerical Simulation (DNS) and Large Eddy Simulation (LES) are discussed. The topics are presented in viewgraph form.
The results from direct numerical simulations of a spatially evolving, supersonic, flat-plate turbulent boundary-layer flow, with free-stream Mach number of 2.25 are presented. The simulated flow field extends from a transition region, initiated by wall suction and blowing near the inflow boundary, into the fully turbulent regime. Distributions of mean and turbulent flow quantities are obtained and an analysis of these quantities is performed at a downstream station corresponding to Re(sub x)= 5.548 x10(exp 6) based on distance from the leading edge.
This viewgraph presentation reviews direct numerical simulation in the late stages of the transition process to turbulence.
Direct numerical simulations (DNS) of Mach 6 turbulent boundary layer with nominal freestream Mach number of 6 and Reynolds number of Re(sub T) approximately 460 are conducted at two wall temperatures (Tw/Tr = 0.25, 0.76) to investigate the generated pressure fluctuations and their dependence on wall temperature. Simulations indicate that the influence of wall temperature on pressure fluctuations is largely limited to the near-wall region, with the characteristics of wall-pressure fluctuations showing a strong temperature dependence. Wall temperature has little influence on the propagation speed of the freestream pressure signal. The freestream radiation intensity compares well between wall-temperature cases when normalized by the local wall shear; the propagation speed of the freestream pressure signal and the orientation of the radiation wave front show little dependence on the wall temperature.
The purpose of our project is to develop, analyze, and test novel numerical technologies central to the long term goal of direct simulations of subsonic jet noise. Our current focus is on two issues: accurate, near-field domain truncations and high-order, single-step discretizations of the governing equations. The Direct Numerical Simulation (DNS) of jet noise poses a number of extreme challenges to computational technique. In particular, the problem involves multiple temporal and spatial scales as well as flow instabilities and is posed on an unbounded spatial domain. Moreover, the basic phenomenon of interest, the radiation of acoustic waves to the far field, involves only a minuscule fraction of the total energy. The best current simulations of jet noise are at low Reynolds number. It is likely that an increase of one to two orders of magnitude will be necessary to reach a regime where the separation between the energy-containing and dissipation scales is sufficient to make the radiated noise essentially independent of the Reynolds number. Such an increase in resolution cannot be obtained in the near future solely through increases in computing power. Therefore, new numerical methodologies of maximal efficiency and accuracy are required.
Here we report flow statistics and visualizations from molecular-gas-dynamics simulations using the direct simulation Monte Carlo (DSMC) method for turbulent Couette flow in a minimal domain where the lower wall is replaced by an idealized permeable fibrous substrate representative of thermal-protection-system materials for which the Knudsen number is O(10 -1 ). Comparisons are made with smooth-wall DSMC simulations and smooth-wall direct numerical simulations (DNS) of the Navier-Stokes equations for the same conditions. Roughness, permeability, and noncontinuum effects are assessed. In the range of Reynolds numbers considered herein, the scalings of the skin friction on the permeable substrate and of the mean flow within the substrate suggest that they are dominated by viscous effects. While the regenerative cycle characteristic of smooth-wall turbulence remains intact for all cases considered, we observe that the near-wall velocity fluctuations are modulated by the permeable substrate with a wavelength equal to the pore spacing. Additionally, the flow within the substrate shows significant rarefaction effects, resulting in an apparent permeability that is 13% larger than the intrinsic permeability. In contrast, the smooth-wall DSMC and DNS simulations exhibit remarkably good agreement for the statistics examined, despite the Knudsen number based on the viscous length scale being as large as O(10 -1 ). This latter result is at variance with classical estimates for the breakdown of the continuum assumption and calls for further investigations into the interaction of noncontinuum effects and turbulence.
Embedded machine-learned models (EMLMs) have the promise to improve the predictive accuracy of engineering simulators in environments of national interest. EMLMs often comprise complex input-output maps (e.g., neural networks), which make them unamenable to rigorous analysis and generally difficult to interpret. In the face of decades of theory, this lack of interpretability is a significant barrier to building confidence in these models. This work outlines an approach to interpret EMLMs using sparse polynomial regression for comparison with theoretical understanding. To do so, we build on the concept of Locally Interpretable Model-agnostic Explanations (LIME) using physics-informed clustering, prototype selection, and library construction. While general, we demonstrate our method on tensor-basis neural networks used in Reynolds-Averaged Navier-Stokes simulations of hypersonic fluid flows. Results are presented for a simulated toy model and for direct numerical simulations (DNS) of turbulent flows over a flat plate.
Computational fluid dynamics (CFD) has matured to the point where it is becoming ubiquitous in the design and analysis of aerospace vehicles and their propulsion systems. The current state-of-the-art of CFD in hypersonic propulsion design and analysis as well as a historical perspective of its development over the last 50 years is presented. The current state-of-the-art for the design of finite rate chemically reacting turbulent hypersonic flow is shown to currently be Reynolds averaged simulations (RAS) where the Reynolds averaged Navier-Stokes (RAS) equations are solved to obtain steady state solutions. An overview of the physical modeling techniques employed by RAS for the simulation of hypersonic chemically reacting flow are presented. However, scale resolving simulation techniques, such as detached eddy simulation (DES), large eddy simulations (LES), hybrid RAS/LES and direct numerical simulations (DNS), are becoming becoming powerful tools for point design analysis. One example of such a scale resolving technique, hybrid RAS/LES, will be briefly described as well as how it can be used to help address model form uncertainties in RAS.
Abstract The traditional approach of using the Monin–Obukhov similarity theory (MOST) to model near-surface processes in large-eddy simulations (LESs) can lead to significant errors in natural convection. In this study, we propose an alternative approach based on feedforward neural networks (FNNs) trained on output from direct numerical simulation (DNS). To evaluate the performance, we conduct both a priori and a posteriori tests. In the a priori (offline) tests, we compare the statistics of the surface shear stress and heat flux, computed from filtered DNS input variables, to the stress and flux obtained from the filtered DNS. Additionally, we investigate the importance of various input features using the Shapley additive explanations value and the conditional average of the filter grid cells. In the a posteriori (online) tests, we implement the trained models in the System for Atmospheric Modeling (SAM) LES and compare the LES-generated surface shear stress and heat flux with those in the DNS. Our findings reveal that vertical velocity, a traditionally overlooked flow quantity, is one of the most important input features for determining the wall fluxes. Increasing the number of input features improves the a priori test results but does not always improve the model performance in the a posteriori tests because of the differences in input variables between the LES and DNS. Last, we show that physics-aware FNN models trained with logarithmic and scaled parameters can well extrapolate to more intense convection scenarios than in the training dataset, whereas those trained with primitive flow quantities cannot. Significance Statement The traditional near-surface turbulence model, based on a shear-dominated boundary layer flow, does not represent near-surface turbulence in natural convection. Using a feedforward neural network (FNN), we can construct a more accurate model that better represents the near-surface turbulence in various flows and reveals previously overlooked controlling factors and process interactions. Our study shows that the FNN-generated models outperform the traditional model and highlight the importance of the near-surface vertical velocity. Furthermore, the physics-aware FNN models exhibit the potential to extrapolate to convective flows of various intensities beyond the range of the training dataset, suggesting their broader applicability for more accurate modeling of near-surface turbulence.
The advection-diffusion equation is simulated via several quantum algorithms. Three formulations are considered: (1) Trotterization, (2) variational quantum time evolution (VarQTE), and (3) adaptive variational quantum dynamics simulation (AVQDS). These schemes were originally developed for the Hamiltonian simulation of many-body quantum systems. The finite-difference discretized operator of the transport equation is formulated as a Hamiltonian and solved without the need for ancillary qubits. Computations are conducted on a quantum simulator (IBM Qiskit Aer) and a superconducting quantum hardware (IBM Fez). The former emulates the latter without the noise. The actual hardware implementation experiences significant noise. The results of the quantum simulator are compared with data from direct numerical simulation (DNS) with infidelities of the order 10 −5 . In the quantum simulator, Trotterization is observed to have the lowest infidelity and is suitable for fault-tolerant computation. The AVQDS algorithm requires the lowest gate count and circuit depth. The VarQTE algorithm is the next best in terms of gate counts, but the number of its optimization variables is directly proportional to the number of qubits. Due to current hardware limitations, Trotterization cannot be implemented, as it has an overwhelmingly large number of operations. Meanwhile, AVQDS and VarQTE can be executed at the hardware level. These algorithms present a new paradigm for computational transport phenomena on quantum computers.
This paper reports results from the numerical implementation and testing of the compressible large eddy-simulation (LES). Relevant quantities from 32-cubed coarse grid LES solutions are compared with results generated from direct numerical simulations (DNS) of three-dimensional compressible turbulence that have been run both with sufficient resolution, at 96 cubed. The 32 cubed LES results overall agree well with their 96 cubed DNS counterparts. Moreover, the new DNS results confirm several recent conclusions about compressible turbulence that have been based primarily on two-dimensional simulations.
A research program for direct numerical simulations of compressible reacting flows is described. Two main research subjects are proposed: the effect of pressure waves on turbulent combustion and the use of direct simulation methods to validate flamelet models for turbulent combustion. The interest of a compressible code to study turbulent combustion is emphasized through examples of reacting shear layer and combustion instabilities studies. The choice of experimental data to compare with direct simulation results is discussed. A tentative program is given and the computation cases to use are described as well as the code validation runs.
A direct numerical simulation method is described for predicting the deformation of laminar liquid jets. In the present nonlinear direct simulation, the convective term, which has been discarded in past linear analyses by Rayleigh and others, is included in the hydrodynamic equations. It is shown that only by maintaining full complexity of the nonlinear surface tension term accurate drop formation can be predicted. The continuity and momentum equations in the transient form are integrated on an adaptive grid, conforming the jet and surface wave shape. The equations, which are parabolic in time and elliptic in space, are solved by a TVD scheme with characteristic flux splitting. The results of the present work are discussed and compared with available measurements and other analyses. The comparison shows that among the predictions, the current 1-D direct simulation results agree best with the experimental data. Furthermore, the computer time requirements are much (an order of magnitude) smaller than those of previously reported multidimensional analyses.
Hypersonic boundary-layer flows over a circular cone at moderate angle of incidence can support strong crossflow instability in between the windward and leeward rays on the plane of symmetry. Due to the more efficient excitation of stationary crossflow vortices by surface roughness, a possible path to transition in such flows corresponds to rapid amplification of the high-frequency secondary instabilities of finite amplitude stationary crossflow vortices. In the present paper, the previous analyses of crossflow instability over a 7- degree half-angle, yawed circular cone in a Mach 6 free stream have been extended to the nonlinear evolution of azimuthally localized crossflow vortex packets and the amplification characteristics and nonlinear breakdown of high-frequency secondary instabilities associated with those packets. A comparison between plane marching PSE and direct Navier-Stokes simulations (DNS) reveals favorable agreement in regard to mode shapes, most amplified disturbance frequencies, and N-factor evolution. In contrast, the quasi-parallel predictions are found to result in severe underprediction of the N-factors. The direct numerical simulations also indicate that the breakdown of secondary instabilities in a 3D hypersonic boundary layer shares certain common features with the previous computations of crossflow transition over subsonic swept wings.
Fourier pseudo-spectral methods for nonlinear partial differential equations are of wide interest in many areas of advanced computational science, including direct numerical simulation of three-dimensional (3-D) turbulence governed by the Navier-Stokes equations in fluid dynamics. This paper presents a new capability for simulating turbulence at a new record resolution up to 35 trillion grid points, on the world's first exascale computer, Frontier, comprising AMD MI250x GPUs with HPE's Slingshot interconnect and operated by the US Department of Energy's Oak Ridge Leadership Computing Facility (OLCF). Key programming strategies designed to take maximum advantage of the machine architecture involve performing almost all computations on the GPU which has the same memory capacity as the CPU, performing all-to-all communication among sets of parallel processes directly on the GPU, and targeting GPUs efficiently using OpenMP offloading for intensive number-crunching including 1-D Fast Fourier Transforms (FFT) performed using AMD ROCm library calls. With 99% of computing power on Frontier being on the GPU, leaving the CPU idle leads to a net performance gain via avoiding the overhead of data movement between host and device except when needed for some I/O purposes. Memory footprint including the size of communication buffers for MPI_ALLTOALL is managed carefully to maximize the largest problem size possible for a given node count. Detailed performance data including separate contributions from different categories of operations to the elapsed wall time per step are reported for five grid resolutions, from 2048 3 on a single node to 32768 3 on 4096 or 8192 nodes out of 9408 on the system. Both 1D and 2D domain decompositions which divide a 3D periodic domain into slabs and pencils respectively are implemented. The present code suite (labeled by the acronym GESTS, GPUs for Extreme Scale Turbulence Simulations) achieves a figure of merit (in grid points per second) exceeding goals set in the Center for Accelerated Application Readiness (CAAR) program for Frontier. The performance attained is highly favorable in both weak scaling and strong scaling, with notable departures only for 2048 3 where communication is entirely intra-node, and for 32768 3 , where a challenge due to small message sizes does arise. Communication performance is addressed further using a lightweight test code that performs all-to-all communication in a manner matching the full turbulence simulation code. Performance at large problem sizes is affected by both small message size due to high node counts as well as dragonfly network topology features on the machine, but is consistent with official expectations of sustained performance on Frontier. Overall, although not perfect, the scalability achieved at the extreme problem size of 32768 3 (and up to 8192 nodes — which corresponds to hardware rated at just under 1 exaflop/sec of theoretical peak computational performance) is arguably better than the scalability observed using prior state-of-the-art algorithms on Frontier's predecessor machine (Summit) at OLCF. New science results for the study of intermittency in turbulence enabled by this code and its extensions are to be reported separately in the near future.
A framework is introduced that leverages known physics to reduce overfitting in machine learning for scientific applications. The partial differential equation (PDE) that expresses the physics is augmented with a neural network that uses available data to learn a description of the corresponding unknown or unrepresented physics. Training within this combined system corrects for missing, unknown, or erroneously represented physics, including discretization errors associated with the PDE's numerical solution. For optimization of the network within the PDE, an adjoint PDE is solved to provide high-dimensional gradients, and a stochastic adjoint method (SAM) further accelerates training. Additionally, the approach is demonstrated for large-eddy simulation (LES) of turbulence. High-fidelity direct numerical simulations (DNS) of decaying isotropic turbulence provide the training data used to learn sub-filter-scale closures for the filtered Navier–Stokes equations. Out-of-sample comparisons show that the deep learning PDE method outperforms widely-used models, even for filter sizes so large that they become qualitatively incorrect. It also significantly outperforms the same neural network when a priori trained based on simple data mismatch, not accounting for the full PDE. Measures of discretization errors, which are well-known to be consequential in LES, point to the importance of the unified training formulation's design, which without modification corrects for them. For comparable accuracy, simulation runtime is significantly reduced. A relaxation of the typical discrete enforcement of the divergence-free constraint in the solver is also successful, instead allowing the DPM to approximately enforce incompressibility physics. Since the training loss function is not restricted to correspond directly to the closure to be learned, training can incorporate diverse data, including experimental data.
Different implications of employment of the temporal or spatial model in direct numerical simulations of transition based on the complete Navier-Stokes equations are outlined, and it is noted that in situations where approximations are not justified, the temporal model cannot be employed. The discussion is then concentrated on developing numerical methods for solving the Navier-Stokes equations based on the spatial model that are applicable for realistic situations for which the temporal model would not be valid. It is concluded that in the future numerical simulations based on the spatial approach will replace, to a large extent, the temporal simulations.
Simulation of two-phase flows is relevant for reactor design and safety at normal operation or during accident scenarios. Often, the two-phase flow is in a regime in which slugs are formed or where the flow stratifies. Modeling such situations using standard single-phase Reynolds-averaged Navier-Stokes (RANS) turbulence models fails due to an overestimation of the eddy viscosity at the resolved two-phase interface. To solve this, an ad hoc turbulence damping term has been proposed in the literature that reduces the turbulence production locally at a two-phase interface, analogously to turbulence wall functions. However, this approach must be tailored to the specific setting and does not consider physical contributions such as surface tension or flow topology. Therefore, the problem of two-phase interfacial turbulence must be studied more in-depth. In this work, we consider co-current turbulent Taylor bubble flow using high-fidelity numerical simulation. The Basilisk code is used to simulate a Taylor bubble rising in a vertical pipe. By simulating the bubble in a moving frame of reference, we may study the turbulent kinetic energy (TKE) budgets ahead of the bubble, in its wake, and across the interface. The implementation of the TKE budget computation and the underlying averaging techniques are validated for the single-phase region ahead of the Taylor bubble using reference direct numerical simulation data. The analysis of the TKE budgets in the setting of Taylor bubble flow allows for the study of how turbulence behaves due to the presence of a two-phase interface and, in turn, supports the improvement of two-phase RANS models.