SEARCH · Search NASA
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.
Presentation describes research to analyze radiation transport using reduced order modeling as an alternate approach to Monte Carlo-based methods, with a goal of developing a variance reduction method that increases computing efficiency.
Natural convection in porous media is a highly nonlinear multiphysical problem relevant to many engineering applications (e.g., the process of CO 2 sequestration). Here, we extend and present a non-intrusive reduced order model of natural convection in porous media employing deep convolutional autoencoders for the compression and reconstruction and either radial basis function (RBF) interpolation or artificial neural networks (ANNs) for mapping parameters of partial differential equations (PDEs) on the corresponding nonlinear manifolds. To benchmark our approach, we also describe linear compression and reconstruction processes relying on proper orthogonal decomposition (POD) and ANNs. Further, we present comprehensive comparisons among different models through three benchmark problems. The reduced order models, linear and nonlinear approaches, are much faster than the finite element model, obtaining a maximum speed-up of 7 × 10 6 because our framework is not bound by the Courant–Friedrichs–Lewy condition; hence, it could deliver quantities of interest at any given time contrary to the finite element model. Our model’s accuracy still lies within a relative error of 7% in the worst-case scenario. We illustrate that, in specific settings, the nonlinear approach outperforms its linear counterpart and vice versa. We hypothesize that a visual comparison between principal component analysis (PCA) and t-Distributed Stochastic Neighbor Embedding (t-SNE) could indicate which method will perform better prior to employing any specific compression strategy.
For over a century, reduced order models (ROMs) have been a fundamental discipline of theoretical fluid mechanics. Early examples include Galerkin models inspired by the Orr–Sommerfeld stability equation and numerous vortex models, of which the von Karman vortex street is one of the most prominent. Subsequent ROMs typically relied on first principles, like mathematical Galerkin models, weakly nonlinear stability theory, and two- and three-dimensional vortex models pioneered the data-driven proper orthogonal decomposition (POD) modeling. In early POD modeling, available data were used to build an optimal basis, which was then utilized in a classical Galerkin procedure to construct the ROM, but data have made a profound impact on ROMs beyond the Galerkin expansion. In this paper, we take a modest step and illustrate the impact of data-driven modeling on one significant ROM area. Specifically, we focus on ROM closures, which are correction terms that are added to the classical ROMs in order to model the effect of the discarded ROM modes in under-resolved simulations. Through simple examples, we illustrate the main modeling principles used to construct the classical ROMs, motivate and introduce modern ROM closures, and show how data-driven modeling, artificial intelligence, and machine learning have changed the standard ROM methodology over the last two decades. Finally, we outline our vision on how the state-of-the-art data-driven modeling can continue to reshape the field of reduced order modeling.
A study in Kalman-filter theory has led to a method of processing relative measurements to estimate the current state of a physical system, using less computation than has previously been thought necessary. As used here, relative measurements signifies measurements that yield information on the relationship between a later and an earlier state of the system. An important example of relative measurements arises in computer vision: Information on relative motion is extracted by comparing images taken at two different times. Relative measurements do not directly fit into standard Kalman filter theory, in which measurements are restricted to those indicative of only the current state of the system. One approach heretofore followed in utilizing relative measurements in Kalman filtering, denoted state augmentation, involves augmenting the state of the system at the earlier of two time instants and then propagating the state to the later time instant.While state augmentation is conceptually simple, it can also be computationally prohibitive because it doubles the number of states in the Kalman filter. When processing a relative measurement, if one were to follow the state-augmentation approach as practiced heretofore, one would find it necessary to propagate the full augmented state Kalman filter from the earlier time to the later time and then select out the reduced-order components. The main result of the study reported here is proof of a property called reduced-order equivalence (ROE). The main consequence of ROE is that it is not necessary to augment with the full state, but, rather, only the portion of the state that is explicitly used in the partial relative measurement. In other words, it suffices to select the reduced-order components first and then propagate the partial augmented state Kalman filter from the earlier time to the later time; the amount of computation needed to do this can be substantially less than that needed for propagating the full augmented Kalman state filter.
Much effort has been made to model hydrogen releases from leaks during potential failures of hydrogen storage systems. A reduced-order jet model can be used to quickly characterize these flows, with low computational cost. Notional nozzle models are often used to avoid modeling the complex shock structures produced by the underexpanded jets by determining an “effective” source to produce the observed downstream trends. In our work, the mean hydrogen concentration fields were measured in a series of subsonic and underexpanded jets using a planar laser Rayleigh scattering system. Furthermore, we compared the experimental data to a reduced order jet model for subsonic flows and a notional nozzle model coupled to the jet model for underexpanded jets. The values of some key model parameters were determined by comparisons with the experimental data. Finally, the coupled model was also validated against hydrogen concentrations measurements for 100 and 200 bar hydrogen jets with the predictions agreeing well with data in the literature.
Here, this paper integrates nonlinear-manifold reduced order models (NM-ROMs) with domain decomposition (DD). NM ROMs approximate the full order model (FOM) state in a nonlinear-manifold by training a shallow, sparse autoencoder using FOM snapshot data. These NM-ROMs can be advantageous over linear-subspace ROMs (LS-ROMs) for problems with slowly decaying Kolmogorov n-width. However, the number of NM-ROM parameters that need to be trained scales with the size of the FOM. Moreover, for “extreme-scale” problems, the storage of high-dimensional FOM snapshots alone can make ROM training expensive. To alleviate the training cost, this paper applies DD to the FOM, computes NM-ROMs on each subdomain, and couples them to obtain a global NM-ROM. This approach has several advantages: Subdomain NM-ROMs can be trained in parallel, involve fewer parameters to be trained than global NM-ROMs, require smaller subdomain FOM dimensional training data, and can be tailored to subdomain specific features of the FOM. The shallow, sparse architecture of the autoencoder used in each subdomain NM-ROM allows application of hyper-reduction (HR), reducing the complexity caused by nonlinearity and yielding computational speedup of the NM-ROM. This paper provides the first application of NM-ROM (with HR) to a DD problem. In particular, this paper details an algebraic DD reformulation of the FOM, training a NM-ROM with HR for each sub domain, and a sequential quadratic programming (SQP) solver to evaluate the coupled global NM-ROM. Theoretical convergence results for the SQP method and a priori and a posteriori error estimates for the DD NM-ROM with HR are provided. The proposed DD NM-ROM with HR approach is numerically compared to a DD LS-ROM with HR on the 2D steady-state Burgers’ equation, showing an order of magnitude improvement in accuracy of the proposed DD NM-ROM over the DD LS-ROM.
We present an efficient data-driven regression approach for constructing reduced-order models (ROMs) of reaction-diffusion systems exhibiting pattern formation. The ROMs are learned non-intrusively from available training data of physically accurate numerical simulations. The method can be applied to general nonlinear systems through the use of polynomial model form, while not requiring knowledge of the underlying physical model, governing equations, or numerical solvers. The process of learning ROMs is posed as a low-cost least-squares problem in a reduced-order subspace identified via Proper Orthogonal Decomposition (POD). Numerical experiments on classical pattern-forming systems–including the Schnakenberg and Mimura–Tsujikawa models–demonstrate that higher-order surrogate models significantly improve prediction accuracy while maintaining low computational cost. The proposed method provides a flexible, non-intrusive model reduction framework, well suited for the analysis of complex spatio-temporal pattern formation phenomena.
This work presents and applies a workflow for performing design optimization on gas-cooled pebble-bed reactors. Based on previous research, a representative equilibrium core of a pebble-bed reactor and a depressurized loss-of-forced-cooling model are created. These applications are built using the Multiphysics Object-Oriented Simulation Environment (MOOSE), specifically utilizing Griffin, Pronghorn, and Bison. After defining design-related parameters and quantities of interest regarding reactor safety and efficiency, this multiphysics model is sampled using the MOOSE stochastic tools module. The result is a comprehensive dataset of configurations, enabling sensitivity analysis and the generation of reduced-order models. Subsequently, the dataset and reduced-order models are employed in an optimization study aimed at maximizing fuel utilization while adhering to safety and operational constraints. The optimization process leads to an improvement of fuel utilization by approximately 10\%, compared to engineering-judgment-based nominal conditions.
Abstract We present an innovative approach called boosting Barlow Twins reduced order modeling (BBT‐ROM) to enhance the reliability of machine learning surrogate models for multiphase flow problems. BBT‐ROM builds upon Barlow Twins reduced order modeling that leverages self‐supervised learning to effectively handle linear and nonlinear manifolds by constructing well‐structured latent spaces of input parameters and output quantities. To address the challenge of high contrast data in multiphase flow problems due to injection wells and faults, we employ a boosting algorithm within BBT‐ROM. This algorithm sequentially trains a set of weak models (i.e., inaccurate models), improving prediction accuracy through ensemble learning. To evaluate the performance of BBT‐ROM, we conduct three three‐dimensional multiphase flow problems, including waterflooding and geologic carbon storage (GCS), with varying numbers of input parameter cases and model domain features. The results demonstrate that BBT‐ROM excels at predicting non‐wetting phase saturation (e.g., oil or saturation) and fluid pressure, with average relative errors ranging from 0.5% to 3%. Importantly, BBT‐ROM showcases robustness when faced with limited input parameter space during GCS testing.
Review of recent work in recasting the energy type of approximation to aircraft flight in terms of singular perturbation theory and in extending this theory to three-dimensional maneuvers. Singular perturbations for differential equations arising in optimal control are first examined for a system of fairly general form but low order. The attitude dynamics for optimal flight of a rocket in vacuum is then studied as an introductory, and fairly transparent, example. The question of the choice of variables is then discussed. Finally, optimal aircraft flight in various reduced-order approximations is investigated. In particular, the problem of three-dimensional aircraft flight is formulated for singular perturbation treatment, and possibilities for decoupling into several lower-order problems are illustrated. The use of reduced-order approximation facilitates numerical computations by reducing the number of multiplier initial values that must be determined simultaneously and by improving the conditioning of the differential equations.
The most general H2 control problem is considered. The authors derive necessary and sufficient conditions when the infimum is attained by state feedback. They do the same for the measurement feedback case where necessary and sufficient conditions are derived when the infimum is attained by proper dynamic compensator. Reduced-order compensators are investigated if some states are observable without noise. For all of these cases the freedom that the non-uniqueness of optimal compensators gives in assigning the closed-loop eigenvalues is discussed. The case when the infimum cannot be attained is investigated. A constructive algorithm is presented to find a minimizing sequence of stabilizing controllers and the freedom in the asymptotic locations of the closed-loop eigenvalues is discussed. This procedure is repeated for three different cases: static state feedback, full-order measurement feedback, and reduced-order measurement feedback.
The present work proposes a self-consistent reduced-order NLTE kinetic model for radiating plasmas found in the outer layers of stellar atmospheres. A detailed collisional-radiative kinetic mechanism is constructed by leveraging the most up-to-date set of ab initio and experimental data available in the literature. This constitutes the starting point for the derivation of a reduced-order model, obtained by lumping the bound energy states into groups. In order to determine the needed thermo-physical group properties, uniform and Maxwell–Boltzmann energy distributions are used to reconstruct the energy population of each group. Finally, the reduced set of governing equations for the material gas and the radiation field is obtained based on the moment method. Applications consider the steady flow across a shock wave in partially ionized hydrogen. The results clearly demonstrate that adopting a Maxwell–Boltzmann grouping allows, on the one hand, for a substantial reduction of the number of unknowns and, on the other, to maintain accuracy for both gas and radiation quantities. Also, it is observed that, when neglecting line radiation, the use of two groups already leads to a very accurate resolution of the photo-ionization precursor, internal relaxation, and radiative cooling regions. The inclusion of line radiation requires adopting just one additional group to account for optically thin losses in the α , β , and γ lines of the Balmer and Paschen series. This trend has been observed for a wide range of shock wave velocities.
This software combines nonlinear-manifold reduced order models (NM-ROMs) with domain decomposition (DD) techniques. NM-ROMs, which utilize a shallow, sparse autoencoder trained with full order model (FOM) snapshot data, approximate the FOM state on a nonlinear manifold. These models offer advantages over linear-subspace ROMs (LS-ROMs) particularly in scenarios with slowly decaying Kolmogorov n-width. However, the training of NM-ROMs involves a number of parameters that scale with the size of the FOM, and storing high-dimensional FOM snapshots can significantly increase the cost of ROM training for extreme-scale problems. To mitigate these costs, the software employs DD to partition the FOM into smaller subdomains, computes NM-ROMs for each, and then integrates these to form a global NM-ROM. This strategy offers multiple benefits: it enables parallel training of subdomain NM-ROMs, reduces the number of parameters needed, decreases the dimensional requirements of subdomain FOM training data, and allows for customization to the unique characteristics of each FOM subdomain. The use of a shallow, sparse autoencoder architecture in each subdomain NM-ROM facilitates the application of hyper-reduction (HR), simplifying the nonlinear complexities and enhancing computational speed. This software marks the inaugural application of NM-ROM combined with HR to a DD problem. It features an algebraic DD reformulation of the FOM, training of NM-ROMs with HR for each subdomain, and employs a sequential quadratic programming (SQP) solver for the evaluation of the coupled global NMROM. The effectiveness of the DD NM-ROM with HR is numerically demonstrated on the 2D steady-state Burgers' equation, showing an order of magnitude improvement in accuracy over the DD LS-ROM with HR.
In this paper, we present a quantitative analysis of the robustness of a reduced-order pole-assignment state-space self-tuning controller for a multivariable adaptive control system whose order of the real process is higher than that of the model used in the controller design. The result of stability analysis shows that, under a specific bounded modelling error, the adaptively controlled closed-loop real system via the reduced-order state-space self-tuner is BIBO stable in the presence of unmodelled dynamics.
Uncertainty quantification of computational fluid dynamics (CFD) simulations is a complicated procedure which still relies in many cases on engineering judgment and factors of safety. This is in part because the computational cost of measuring the simulation's sensitivity to all meaningful parameters (e.g., body surface roughness) and hyperparameters (e.g., subiteration convergence criterion) is intractable for even a single simulation. Reduced-order modeling dramatically lowers this computational cost of simulating fluid flows, but usually only where similar data is already available. In this work, fluid reduced-order models are utilized to quantify a flow's sensitivity to certain physical parameters for the purposes of improved uncertainty quantification. Characteristic observability and sensitivity are both explored. The resulting sensitivity quantifications enable more informed CFD frameworks and more rigorous uncertainty bounds on the resulting data.
Uncertainty quantification of computational fluid dynamics (CFD) simulations is a complicated procedure which still relies in many cases on engineering judgment and factors of safety. This is in part because the computational cost of measuring the simulation's sensitivity to all meaningful parameters (e.g., body surface roughness) and hyperparameters (e.g., subiteration convergence criterion) is intractable for even a single simulation. Reduced-order modeling dramatically lowers this computational cost of simulating fluid flows, but usually only where similar data is already available. In this work, fluid reduced-order models are utilized to quantify a flow's sensitivity to certain physical parameters for the purposes of improved uncertainty quantification. Characteristic observability and sensitivity are both explored. The resulting sensitivity quantifications enable more informed CFD frameworks and more rigorous uncertainty bounds on the resulting data.
The Rayleigh-Taylor instability is a classical hydrodynamic instability of great interest in various disciplines of science and engineering, including astrophysics, atmospheric sciences and climate, geophysics, and fusion energy. Analytical methods cannot be applied to explain the long-time behavior of the Rayleigh-Taylor instability, and therefore, numerical simulation of the full problem is required. However, in order to capture the growth of amplitude of perturbations accurately, both the spatial and temporal discretizations need to be extremely fine for traditional numerical methods, and long-time simulation may become prohibitively expensive. In this paper, we propose efficient reduced order model techniques to accelerate the simulation of the Rayleigh-Taylor instability in compressible gas dynamics. Here, we introduce a general framework for decomposing the solution manifold to construct the temporal domain partition and temporally-local reduced order model construction with varying Atwood number. We propose two practical approaches in this framework, namely decomposition by physical time and by penetration distance. Numerical results are presented to examine the performance of the proposed approaches.
While the accuracy of assembly calculations has greatly improved due to the increase in computer power enabling more refined description of the phase space and use of more sophisticated numerical algorithms, the computational cost continues to increase which limits the full utilization of their effectiveness for routine engineering analysis. Reduced order modeling is a mathematical vehicle that scales down the dimensionality of large-scale numerical problems to enable their repeated executions on small computing environment, often available to end users. This is done by capturing the most dominant underlying relationships between the model's inputs and outputs. Previous works demonstrated the use of the reduced order modeling for a single physics code, such as a radiation transport calculation. This paper extends those works to coupled code systems as currently employed in assembly calculations. Finally, numerical tests are conducted using realistic SCALE assembly models with resonance self-shielding, neutron transport, and nuclides transmutation/depletion models representing the components of the coupled code system.