Model Reduction Methodology Research Articles

Data-driven nonlinear model reduction to spectral submanifolds in mechanical systems.

While data-driven model reduction techniques are well-established for linearizable mechanical systems, general approaches to reducing nonlinearizable systems with multiple coexisting steady states have been unavailable. In this paper, we review such a data-driven nonlinear model reduction methodology based on spectral submanifolds. As input, this approach takes observations of unforced nonlinear oscillations to construct normal forms of the dynamics reduced to very low-dimensional invariant manifolds. These normal forms capture amplitude-dependent properties and are accurate enough to provide predictions for nonlinearizable system response under the additions of external forcing. We illustrate these results on examples from structural vibrations, featuring both synthetic and experimental data.This article is part of the theme issue ‘Data-driven prediction in dynamical systems’.

Toward Model Reduction for Power System Transients With Physics-Informed PDE

This manuscript reports the first step towards building a robust and efficient model reduction methodology to capture transient dynamics in a transmission level electric power system. Such dynamics is normally modeled on seconds-to-tens-of-seconds time scales by the so-called swing equations, which are ordinary differential equations defined on a spatially discrete model of the power grid. Following Seymlyen (1974) and Thorpe, Seyler, and Phadke (1999), we suggest to map the swing equations onto a linear, inhomogeneous Partial Differential Equation (PDE) of parabolic type in two space and one time dimensions with time-independent coefficients and properly defined boundary conditions. We illustrate our method on the synchronous transmission grid of continental Europe. We show that, when properly coarse-grained, i.e., with the PDE coefficients and source terms extracted from a spatial convolution procedure of the respective discrete coefficients in the swing equations, the resulting PDE reproduces faithfully and efficiently the original swing dynamics. We finally discuss future extensions of this work, where the presented PDE-based modeling will initialize a physics-informed machine learning approach for real-time modeling, $n-1$ feasibility assessment and transient stability analysis of power systems.

Global sensitivity analysis informed model reduction and selection applied to a Valsalva maneuver model

In this study, we develop a methodology for model reduction and selection informed by global sensitivity analysis (GSA) methods. We apply these techniques to a control model that takes systolic blood pressure and thoracic tissue pressure data as inputs and predicts heart rate in response to the Valsalva maneuver (VM). The study compares four GSA methods based on Sobol’ indices (SIs) quantifying the parameter influence on the difference between the model output and the heart rate data. The GSA methods include standard scalar SIs determining the average parameter influence over the time interval studied and three time-varying methods analyzing how parameter influence changes over time. The time-varying methods include a new technique, termed limited-memory SIs, predicting parameter influence using a moving window approach. Using the limited-memory SIs, we perform model reduction and selection to analyze the necessity of modeling both the aortic and carotid baroreceptor regions in response to the VM. We compare the original model to systematically reduced models including (i) the aortic and carotid regions, (ii) the aortic region only, and (iii) the carotid region only. Model selection is done quantitatively using the Akaike and Bayesian Information Criteria and qualitatively by comparing the neurological predictions. Results show that it is necessary to incorporate both the aortic and carotid regions to model the VM.

Reduced-order models for coupled dynamical systems: Data-driven methods and the Koopman operator.

Providing efficient and accurate parameterizations for model reduction is a key goal in many areas of science and technology. Here, we present a strong link between data-driven and theoretical approaches to achieving this goal. Formal perturbation expansions of the Koopman operator allow us to derive general stochastic parameterizations of weakly coupled dynamical systems. Such parameterizations yield a set of stochastic integrodifferential equations with explicit noise and memory kernel formulas to describe the effects of unresolved variables. We show that the perturbation expansions involved need not be truncated when the coupling is additive. The unwieldy integrodifferential equations can be recast as a simpler multilevel Markovian model, and we establish an intuitive connection with a generalized Langevin equation. This connection helps setting up a parallelism between the top-down, equation-based methodology herein and the well-established empirical model reduction (EMR) methodology that has been shown to provide efficient dynamical closures to partially observed systems. Hence, our findings, on the one hand, support the physical basis and robustness of the EMR methodology and, on the other hand, illustrate the practical relevance of the perturbative expansion used for deriving the parameterizations.

Complex Frequency Domain Model Reduction: A Comparison of Techniques

A mathematical model of the plant to be controlled must first be identified in order to design a controller to improve the process’ performance and stability. This mathematical model usually is in the form of complex frequency domain state-space realization or continuous/ discrete time domain regression algorithm. This paper works on the complex frequency domain state-space realization. Reducing the order of the state-space model of the physical plant is usually desirable since it reduces computational time and hardware implementation for the resulting controller designed based on the model. For a frequency domain state space realization of a linear time-invariant and non-causal plant, there are numerous methods available to reduce its order. These include Balancing Model Reduction, Frequency-Weighted Balancing, Low-Frequency Approximation and Balanced Truncation with Reduced Error Bound. Besides background information and a summary, the model-reduction methodology for each of the above-said techniques is also presented. Using a two-input two-output six-state model for a large turbo-generator and a large industrial boiler with 8 states, 2 inputs and 2 outputs model, this paper provides a comparison between three of the four methodologies mentioned above in terms of the spectral norms of the model reduction error.

Sampling-free model reduction of systems with low-rank parameterization

We consider the reduction of parametric families of linear dynamical systems having an affine parameter dependence that allow for low-rank variation in the state matrix. Usual approaches for parametric model reduction typically involve exploring the parameter space to identify representative parameter values and the associated models become the principal focus of model reduction methodology. These models are then combined in various ways in order to interpolate the response. The initial exploration of the parameter space can be a forbiddingly expensive task. A different approach is proposed here that requires neither parameter sampling nor parameter space exploration. Instead, we represent the system response function as a composition of four subsystem response functions that are non-parametric with a purely parameter-dependent function. One may apply any one of a number of standard (non-parametric) model reduction strategies to reduce the subsystems independently, and then conjoin these reduced models with the underlying parameterization to obtain the overall parameterized response. Our approach has elements in common with the parameter mapping approach of Baur et al. (PAMM 14(1), 19–22 2014) but offers greater flexibility and potentially greater control over accuracy. In particular, a data-driven variation of our approach is described that exercises this flexibility through the use of limited frequency-sampling of the underlying non-parametric models. The parametric structure of our system representation allows for a priori guarantees of system stability in the resulting reduced models across the full range of parameter values. Incorporation of system theoretic error bounds allows us to determine appropriate approximation orders for the non-parametric systems sufficient to yield uniformly high accuracy across the parameter range. We illustrate our approach on a class of structural damping optimization problems and on a benchmark model of thermal conduction in a semiconductor chip. The parametric structure of our reduced system representation lends itself very well to the development of optimization strategies making use of efficient cost function surrogates. We discuss this in some detail for damping parameter and location optimization for vibrating structures.

Koopman-Based Approach to Nonintrusive Reduced Order Modeling: Application to Aerodynamic Shape Optimization and Uncertainty Propagation

A methodology for nonintrusive projection-based nonlinear model reduction originally presented by Renganathan et al. (“Koopman-Based Approach to Nonintrusive Projection-Based Reduced-Order Modeling with Black-Box High-Fidelity Models,” AIAA Journal, Vol. 56, No. 10, 2018, pp. 4087–4111) is further extended toward parametric systems with a focus on application to aerospace design. Specifically, the method is extended to address static systems with parametric geometry (that deforms the mesh) in addition to parametric freestream boundary conditions. The main idea is to first perform a transformation on the governing equations such that it is lifted to a higher-dimensional but linear underdetermined system. This enables one to extract the system matrices easily as compared to that of the original nonlinear system. The underdetermined system is closed with a set of model-dependent nonlinear constraints upon which the model reduction is finally performed. The methodology is validated on the subsonic and transonic inviscid flows past the NACA0012 and the RAE2822 airfoils with parametrized shapes. The utility of the approach is further demonstrated by applying it to two common problems in aerospace design, namely, derivative-free global optimization and parametric uncertainty quantification with Monte Carlo sampling. Overall, the methodology is shown to achieve accuracy up to 5% and computational speedup of two to three orders of magnitude relative to the full order model. Comparison against another nonintrusive model reduction method revealed that the proposed approach is more robust, accurate, and retains the consistency between the state variables.

Deep multiscale model learning

The objective of this paper is to design novel multi-layer neural networks for multiscale simulations of flows taking into account the observed fine data and physical modeling concepts. Our approaches use deep learning techniques combined with local multiscale model reduction methodologies to predict flow dynamics. Using reduced-order model concepts is important for constructing robust deep learning architectures since the reduced-order models provide fewer degrees of freedom. We consider flow dynamics in porous media as multi-layer networks in this work. More precisely, the solution (e.g., pressures and saturation) at the time instant n+1 depends on the solution at the time instant n and input parameters, such as permeability fields, forcing terms, and initial conditions. One can regard the solution as a multi-layer network, where each layer, in general, is a nonlinear forward map and the number of layers relates to the internal time steps. We will rely on rigorous model reduction concepts to define unknowns and connections between layers. It is critical to use reduced-order models for this purpose, which will identify the regions of influence and the appropriate number of variables. Furthermore, due to the lack of available observed fine data, the reduced-order model can provide us sufficient inexpensive data as needed. The designed deep neural network will be trained using both coarse simulation data which is obtained from the reduced-order model and observed fine data. We will present the main ingredients of our approach and numerical examples. Numerical results show that using deep learning with data generated from multiscale models as well as available observed fine data, we can obtain an improved forward map which can better approximate the fine scale model.

Comparative analysis of reduced-order spectral models and grouping strategies for non-equilibrium radiation

Radiative transfer calculations in complex three-dimensional domains are plagued by computational bottlenecks due to the combined cost of resolving spectral, angular, and spatial dependence. The current work presents a systematic study into the efficacy of three commonly used reduced-order wide-band models, Planck-averaging, and statistics-based k-distribution and theory of homogenization, for modeling non-equilibrium radiation. A reinterpretation of Planck-averaging based on the maximum entropy principle combined with the commonly used multi-band opacity binning allows a direct equivalence to be drawn with the two statistics-based approaches. Additionally, the three seemingly distinct methods are shown to be strictly derived only for local thermodynamic equilibrium conditions which limits their accuracy when simulating non-equilibrium radiation. This shortcoming is addressed through the development of a novel grouping strategy for defining larger groups of individual frequencies from detailed radiation databases (based on line-by-line or narrow-band models) while accounting for variation in all radiative properties under non-equilibrium. Radiative transfer calculations for different Earth and Jupiter entry problems are performed using the different spectral models. Conventional reduced-order wide-band approaches converge slowly to the solution obtained using detailed spectral models. However, the new non-equilibrium grouping strategy allows both total quantities-of-interest and their detailed spectral variation to be predicted accurately while employing fewer reduced-order groups. The current model-reduction methodology provides nearly two orders-of-magnitude decrease in required spectral evaluations along a given line-of-sight with respect to narrow-band methods (and three to four orders with respect to original LBL databases) and is ideal for coupled flow-radiation calculations.

Issues arising in the construction of QSSA mechanisms: the case of reduced n-heptane/air models for premixed flames

A model reduction methodology, based on the quasi steady-state approximation (QSSA), is employed for the construction of reduced mechanisms in the case of an n-heptane/air premixed flame. Several issues related to the construction of these reduced mechanisms are discussed; such as the influence of the size of the starting skeletal mechanism, the stiffness reduction, and the truncation/simplification of (i) the expressions of the global rates and (ii) the steady-state relations. The starting point for the reduction is two skeletal mechanisms that involve 177/768 and 66/326 species/reactions, respectively [J. Prager, H.N. Najm, M. Valorani, and D.A. Goussis, Skeletal mechanism generation with CSP and validation for premixed n-heptane flames, Proc. Combust. Inst. 32 (2009), pp. 509–517] and which were derived from the detailed mechanism of Curran et al. [H.J. Curran, P. Gaffuri, W.J. Pitz, and C.K. Westbrook, A comprehensive modeling study of iso-octane oxidation, Combust. Flame 129 (2002), pp. 253–280], which involves 561/2538 species/reactions. From these two skeletal mechanisms, a number of reduced mechanisms of various sizes are produced and analysed. The validity of the reduced mechanism with the minimum size is demonstrated by considering its accuracy regarding the mass fractions of major and minor species, the temperature, and the flame speed, over a wide range of equivalence ratios and pressures.

Model reduction methodology for computational simulations of endovascular repair

Endovascular aneurysm repair (EVAR) is a current alternative treatment for thoracic and abdominal aortic aneurysms, but is still sometimes compromised by possible complications such as device migration or endoleaks. In order to assist clinicians in preventing these complications, finite element analysis (FEA) is a promising tool. However, the strong material and geometrical nonlinearities added to the complex multiple contacts result in costly finite-element models. To reduce this computational cost, we establish here an alternative and systematic methodology to simplify the computational simulations of stent-grafts (SG) based on FEA. The model reduction methodology relies on equivalent shell models with appropriate geometrical and mechanical parameters. It simplifies significantly the contact interactions but still shows very good agreement with a complete reference finite-element model. Finally, the computational time for EVAR simulations is reduced of a factor 6–10. An application is shown for the deployment of a SG during thoracic endovascular repair, showing that the developed methodology is both effective and accurate to determine the final position of the deployed SG inside the aneurysm.

Modeling and Evaluating Zeolite and Amorphous Based Catalysts in Vacuum Gas Oil Hydrocracking Process

Abstract Hydrocracking is a significant process in a refinery which is commonly used for converting heavy fractions such as vacuum gas oil (VGO) to the valuable products such as naphtha and diesel. In this research, VGO hydrocracking process was studied in a pilot scale plant in the presence of a zeolite and two amorphous based commercial catalysts called RK-NiY, RK-MNi and KF-101, respectively. In order to study the effect of support on the yield of the process, a discrete 4-lump kinetic model, including feed (vacuum gas oil and unconverted materials), distillate (diesel and kerosene), naphtha and gas was proposed for each catalyst. At first, each network had six reaction paths and twelve kinetic coefficients, and then by using the model reduction methodology, only four main routes for RK-MNi and RK-NiY, and three ones for KF-101 were designated. Results showed that the absolute average deviation (AAD%) of reduced models decreased from 5.11 %, 10.1 % and 21.8 % to 4.54 %, 8.9 % and 19.67 % for RK-MNi, KF-101 and RK-NiY, respectively. Moreover, it was confirmed that amorphous and zeolite catalysts could be selected for producing middle distillate and naphtha products, respectively.

A Method for the Hankel-Norm Approximation of Fractional-Order Systems

A model-reduction methodology for fractional-order systems based on the Hankel-norm is presented. The methodology involves the truncation of a Laurent series associated with the fractional-order system in a transformed domain. The truncated Laurent series coefficients are used to construct a finite-order transfer function to approximate the original system. Standard model-reduction techniques are then applied to obtain a final low-order approximation. The Hankel norm of the approximation error can be specified a priori. The approximation method is applied to several fractional-order and other infinite-order systems. It is shown to be more generally applicable than standard finite-order modeling techniques.

Dimensional hyper-reduction of nonlinear finite element models via empirical cubature

We present a general framework for the dimensional reduction, in terms of number of degrees of freedom as well as number of integration points (“hyper-reduction”), of nonlinear parameterized finite element (FE) models. The reduction process is divided into two sequential stages. The first stage consists in a common Galerkin projection onto a reduced-order space, as well as in the condensation of boundary conditions and external forces. For the second stage (reduction in number of integration points), we present a novel cubature scheme that efficiently determines optimal points and associated positive weights so that the error in integrating reduced internal forces is minimized. The distinguishing features of the proposed method are: (1) The minimization problem is posed in terms of orthogonal basis vector (obtained via a partitioned Singular Value Decomposition) rather that in terms of snapshots of the integrand. (2) The volume of the domain is exactly integrated. (3) The selection algorithm need not solve in all iterations a nonnegative least-squares problem to force the positiveness of the weights. Furthermore, we show that the proposed method converges to the absolute minimum (zero integration error) when the number of selected points is equal to the number of internal force modes included in the objective function. We illustrate this model reduction methodology by two nonlinear, structural examples (quasi-static bending and resonant vibration of elastoplastic composite plates). In both examples, the number of integration points is reduced three order of magnitudes (with respect to FE analyses) without significantly sacrificing accuracy.

Parametric model reduction for aeroelastic systems: Invariant aeroelastic modes

A novel model reduction methodology for coupled aeroelastic systems undergoing parameter variations is presented based on a frequency-domain formulation and use of the Proper Orthogonal Decomposition. Typically, an aeroelastic system is a function of multiple parameters such as air density, speed, as well as structural parameters, hence its aeroelastic characteristics vary from one condition to another. It is shown that using the Modally Equivalent Perturbed System it is possible to interpret and analyze the parameter variations in the context of ordinary differential equation with forcing terms. The new procedure is applied to Goland wing modeled by a finite element and unsteady vortex to produce a new class of aeroelastic modes that are invariant under the parameter variations. When used for model reduction these aeroelastic modes are shown to produce accurate parametric reduced-order models for a wide range of the parameters.

A reduced low-temperature electro-thermal coupled model for lithium-ion batteries

A low-temperature electro-thermal coupled model, which is based on the electrochemical mechanism, is developed to accurately capture both electrical and thermal behaviors of batteries. Activation energies reveal that temperature dependence of resistances is greater than that of capacitances. The influence of frequency on polarization voltage and irreversible heat is discussed, and frequency dependence of polarization resistance and capacitance is obtained. Based on the frequency-dependent equation, a reduced low-temperature electro-thermal coupled model is proposed and experimentally validated under different temperature, frequency and amplitude conditions. Simulation results exhibit good agreement with experimental data, where the maximum relative voltage error and temperature error are below 2.65% and 1.79°C, respectively. The reduced model is demonstrated to have almost the same accuracy as the original model and require a lower computational effort. The effectiveness and adaptability of the proposed methodology for model reduction is verified using batteries with three different cathode materials from different manufacturers. The reduced model, thanks to its high accuracy and simplicity, provides a promising candidate for development of rapid internal heating and optimal charging strategies at low temperature, and for evaluation of the state of battery health in on-board battery management system.

Surrogate model reduction for linear dynamic systems based on a frequency domain modal analysis

A novel model reduction methodology for linear dynamic systems with parameter variations is presented based on a frequency domain formulation and use of the proper orthogonal decomposition. For an efficient treatment of parameter variations, the system matrices are divided into a nominal and an incremental part. It is shown that the perturbed part is modally equivalent to a new system where the incremental matrices are isolated into the forcing term. To account for the continuous changes in the parameters, the single-composite-input is invoked with a finite number of predetermined incremental matrices. The frequency-domain Karhunen---Loeve procedure is used to calculate a rich set of basis modes accounting for the variations. For demonstration, the new procedure is applied to a finite element model of the Goland wing undergoing oscillations and shown to produce extremely accurate reduced-order surrogate model for a wide range of parameter variations.

Zoning Evaluation for Improved Coordinated Automatic Voltage Control

Hierarchically structured automatic voltage control (AVC) architecture has attracted increased interest as networks operate closer to their capacity limits. Hierarchical AVC enables wide-area coordinated voltage regulation (CVR). Due to the inherent complexity of the task, it is based on reduced control models, i.e., simplified models of the system suitable for voltage control. It is a fact however that a single reduced control model (static RCM) cannot be optimal for all network configurations and operating conditions. In pursuit of an improved CVR, this paper investigates the applicability of zoning methodologies in adaptively determined RCM. It further argues that the selection of a zoning methodology affects not only the CVR operation, but also its robustness to erroneous data and proposes a comprehensive generic framework for evaluating its performance. Lastly, it extends and evaluates several zoning-based control model reduction methodologies: namely, hierarchical clustering employing two different proximity metrics, spectral ${\rm k}$ -way and fuzzy ${\rm C}$ -means, on both static and adaptive schemes.

Reduced order modeling of the Newton formulation of MODFLOW to solve unconfined groundwater flow

Projection-based model reduction techniques have been shown to be very effective for reducing the computational burden of high-dimensional groundwater simulations, but only applied to confined groundwater flow. A new methodology is proposed that reduces the dimension of a discretized, transient, unconfined groundwater-flow model. This unconfined model reduction technique is based on Galerkin projection and the Newton formulation of MODFLOW. The method is implemented following the standard package design and code structure that MODFLOW employs for all its features. When the package is invoked within MODFLOW it can collect snapshots, produce a basis, construct the reduced model and propagate the reduced model forward in time. The new formulation accurately represents the water-table surface under a variety of nonlinear settings, such as intraborehole flow from a Multi-Node Well. The unconfined model reduction is applied to four test cases to illustrate its flexibility in handling nonlinear features. Several test cases are discussed to demonstrate the unconfined model reduction applicability. The final test case applies the new model reduction methodology to a scoping MODFLOW model of Santa Barbara, CA composed of 113,578 cells, which requires solving 113,578 equations per time step, and reduces it to 127 equations.

Predicting Critical Transitions in ENSO models. Part II: Spatially Dependent Models

Abstract The present paper is the second part of a two-part study on empirical modeling and prediction of climate variability. This paper deals with spatially distributed data, as opposed to the univariate data of Part I. The choice of a basis for effective data compression becomes of the essence. In many applications, it is the set of spatial empirical orthogonal functions that provides the uncorrelated time series of principal components (PCs) used in the learning set. In this paper, the basis of the learning set is obtained instead by applying multichannel singular-spectrum analysis to climatic time series and using the leading spatiotemporal PCs to construct a reduced stochastic model. The effectiveness of this approach is illustrated by predicting the behavior of the Jin–Neelin–Ghil (JNG) hybrid seasonally forced coupled ocean–atmosphere model of El Niño–Southern Oscillation. The JNG model produces spatially distributed and weakly nonstationary time series to which the model reduction and prediction methodology is applied. Critical transitions in the hybrid periodically forced coupled model are successfully predicted on time scales that are substantially longer than the duration of the learning sample.