Reduced Order Model Technique Research Articles

Feasible model-order reduction approach for analysis of composite rotor blades based on geometrically exact beam formulation

This paper presents a novel and practical reduced-order model (ROM) approach designed specifically for analyzing composite rotor blades, leveraging the geometrically exact beam formulation. The slender nature of rotor blades necessitates aero-structure coupled analysis to precisely predict their response, which involved numerous iterative computations. By capturing large displacements and rotations, this approach enhances the accuracy of blade behavior predictions under operational conditions. To address the computational demands of large-scale analyses, a ROM technique that effectively reduces system dimensionality while preserving critical structural features is introduced. The proposed method is then applied to the aero-structure interaction analysis of a hingeless hovering rotor trim, incorporating geometrical nonlinearity and centrifugal effects. The analysis results demonstrate that the ROM accurately captures the dynamics of these complex aeroelastic systems while significantly reducing computational costs. This approach exhibits significant potential for various aerospace applications, particularly in the design and analysis of structures undergoing geometrical nonlinear behavior and rotation-induced loads.

Nonlinear Reduced Order Modelling for Korteweg-de Vries Equation

Efficient computational techniques that maintain the accuracy and invariant preservation property of the Korteweg-de Vries (KdV) equations have been studied by a wide range of researchers. In this paper, we introduce a reduced order model technique utilizing kernel principle component analysis (KPCA), a nonlinear version of the classical principle component analysis, in a non-intrusive way. The KPCA is applied to the data matrix, which is formed by the discrete solution vectors of KdV equation. In order to obtain the discrete solutions, the finite differences are used for spatial discretization, and linearly implicit Kahan's method for the temporal one. The back-mapping from the reduced dimensional space, is handled by a non-iterative formula based on the idea of multidimensional scaling (MDS) method. Through KPCA, we illustrate that the reduced order approximations conserve the invariants, i.e., Hamiltonian, momentum and mass structure of the KdV equation. The accuracy of reduced solutions, conservation of invariants, and computational speed enhancements facilitated by classical (linear) PCA and KPCA are exemplified through one-dimensional KdV equation.

Robustness evaluation of large-scale machine learning-based reduced order models for reproducing flow fields

The robustness of an artificial neural network that performs model order reduction for flow field data is studied. The network is trained with a large-scale distributed learning approach using up to 6259 nodes of the supercomputer Fugaku. Flow around two square cylinders with a varying distance between their centers is investigated. The network is trained and tested with data from numerical simulations. First, the capability to reproduce flow fields with 2, 12, and 24 modes is investigated by comparing the reconstructed flow data to simulated data. It is shown, that reconstructions based on 2 modes cannot capture both, low- and high-frequency flow structures correctly, whereas predictions based on 12 and 24 modes yield improved flow fields, especially in the case of high-frequency waves in the vicinity of the square cylinders. Reconstructions with 24 modes provide smooth velocity fields that reproduce all relevant low- and high-frequency waves for all variations of the distance between the two square cylinders. Second, the performance of the machine learning-based reconstructions are compared to proper orthogonal decomposition, which is a commonly used reduced order model technique. The comparison only includes flow fields based on 24 modes. For all geometric variations, the mean squared errors of the reconstructions by the conventional method are higher than those of the machine learning model. This underlines the advantage of artificial neural networks over linear methods like proper orthogonal decomposition for tasks like reconstructing flow fields that are characterized by non-linear governing equations.

A domain decomposing reduced-order method for solving the multi-domain transient heat conduction problems

In this article, a domain decomposition algorithm combined with a reduced-order model (ROM) is proposed to deal with the transient heat conduction problems in multi-domain structures. Using the domain decomposition algorithm and the finite element method (FEM), the original problem is decomposed into multiple non-overlapping subdomains, which can be solved independently under appropriate boundary conditions. However, additional inter-regional connected calculation will be brought through the simple geometry decomposition. Thus, the proper orthogonal decomposition (POD), a projection-based ROM technique, is utilized to further reduce the number of unknowns and the calculation time for each subdomain. To be more specific, the local POD bases of each subdomain are obtained by decomposing a snapshot matrix that gathered according to time by the application of the virtual boundary conditions on interfaces of the adjacent subdomains and other boundaries. Furthermore, the problems under different boundary conditions can be linked by the variable condensation technique on the interfaces, which enhances the efficiency of the numerical solver dramatically. Finally, the accuracy and efficiency of the proposed method are demonstrated by comparing the results calculated using both the proposed algorithm and the direct finite element simulation.

Automatic Proper Orthogonal Block Decomposition method for network dynamical systems with multiple timescales

In this work, we introduce a novel reduced order model technique, based on the Proper Orthogonal Decomposition method, for dynamical systems with multiple timescales. The main ideas are to retain the structure of the original model, which is lost in the original POD procedure, while producing a competitive reduction in the number of equations and computational time, and to determine the best structure for the reduced system automatically, via a data-driven analysis of the original model data. For these novel techniques, we present some numerical tests for various behaviors of three different neural network models with multiple timescales, which support the use of these new methods.

Data-Driven Reduced Order Modelling for Patient-Specific Hemodynamics of Coronary Artery Bypass Grafts with Physical and Geometrical Parameters

In this work the development of a machine learning-based Reduced Order Model (ROM) for the investigation of hemodynamics in a patient-specific configuration of Coronary Artery Bypass Graft (CABG) is proposed. The computational domain is referred to left branches of coronary arteries when a stenosis of the Left Main Coronary Artery (LMCA) occurs. The method extracts a reduced basis space from a collection of high-fidelity solutions via a Proper Orthogonal Decomposition (POD) algorithm and employs Artificial Neural Networks (ANNs) for the computation of the modal coefficients. The Full Order Model (FOM) is represented by the incompressible Navier-Stokes equations discretized using a Finite Volume (FV) technique. Both physical and geometrical parametrization are taken into account, the former one related to the inlet flow rate and the latter one related to the stenosis severity. With respect to the previous works focused on the development of a ROM framework for the evaluation of coronary artery disease, the novelties of our study include the use of the FV method in a patient-specific configuration, the use of a data-driven ROM technique and the mesh deformation strategy based on a Free Form Deformation (FFD) technique. The performance of our ROM approach is analyzed in terms of the error between full order and reduced order solutions as well as the speed-up achieved at the online stage.

Nonintrusive parametric NVH study of a vehicle body structure

A reduced order model technique is presented to perform the parametric Noise, Vibration and Harshness (NVH) study of a vehicle body-in-white (BIW) structure characterized by material and shape design variables. The ultimate goal is to develop a methodology which allows to efficiently explore the variation in the design space of the BIW static and dynamic global stiffnesses, such that the NVH performance can be evaluated already in the preliminary phase of the development process. The proposed technique is based on the proper generalized decomposition (PGD) method. The obtained PGD solution presents an explicit dependency on the introduced design variables, which allows to obtain solutions in 0.1 milliseconds and therefore opens the door to fast optimization studies and real-time visualizations of the results in a pre-defined range of parameters. The method is nonintrusive, such that an interaction with commercial software is possible. A parametrized finite element (FE) model of the BIW is built by means of the ANSA CAE preprocessor software, which allows to account for material and geometric parameters. A comparison between the parametric NVH solutions and the full-order FE simulations is performed using the MSC-Nastran software, to validate the accuracy of the proposed method. In addition, an optimization study is presented to find the optimal materials and shape properties with respect to the NVH performance. Finally, in order to support the designers in the decision-making process, a graphical interface app is developed which allows to visualize in real-time how changes in the design variables affect pre-defined quantities of interest.

Reduced order modeling for coupled thermal-hydraulics and reactor physics problems

The development of a new generation of reactor presents several modelling challenges that cannot be effectively addressed by traditional tools used for Light Water Reactors. Multi-physics approaches allow for a comprehensive modeling of reactor cores, as they intrinsically couple the involved physics (i.e. fluid-dynamics, heat transfer, neutronics, thermo-mechanics), but require intense computational efforts that preclude their use in reactor control applications. This work aims at applying a Reduced Order Model (ROM) technique to multi-physics modelling. Such objective is achieved through a ROM technique previously used for Navier-Stokes equations and thermal-hydraulics problems (POD-FV-ROM). The technique used in this article, is built on a modeling framework developed ad-hoc for the Finite Volume (FV) scheme and relies on the Proper Orthogonal Decomposition (POD) and the Method of Snapshots. The reduced order multi-physics approach outlined in this work has been tested with success on a Lid-Driven-Cavity-based homogeneous reactor model. Reduced-order simulations have reproduced accurately the full order velocity, temperature, neutron flux and precursor concentration fields, resulting in relative L2 norms of the differences between full order and reduced order simulations below 1 % for all the considered fields.

A new alternating iteration strategy based on the proper orthogonal decomposition for solving large-scaled transient nonlinear heat conduction problems

In this paper, the application of Free Element Method (FREM) is extended to the transient nonlinear heat conduction problems, and the characteristics of anisotropic, heterogeneous, temperature-dependent thermophysical properties and heat generation are also involved. In order to improve the efficiency during the nonlinear iterations, an alternating iteration strategy, FREM-ROM, is proposed, in which the proper orthogonal decomposition (POD) technique is used to establish a sectionalized extrapolation algorithm with lower dimensions and sufficiently high accuracy. This iteration strategy reduces the computational scale by using the Reduced-Order Model (ROM) technique and can guarantee the accuracy by continually correcting the database and POD bases every few time steps during the process of iterations. It also has the potential to be applied to other numerical methods for improved the efficiency. The accuracy and efficiency are examined by two numerical examples, in which the 2D example illustrates the relationship between truncated basis and accuracy in detail, and the heat transfer performance of a large-scale 3D cross-fin heat sink(CFHS) is investigated to highlight the efficiency advantages of solving large-scaled practical problems.

Investigation on Pull-in Voltage, Frequency Tuning and Frequency Stability of MEMS Devices Incorporating Casimir Force with Correction for Finite Conductivity

This paper studies the influence of Casimir force on the static and dynamic characteristics of electrostatically actuated fixed–fixed microbeam. The theoretical formulation for the study is obtained using the Euler–Bernoulli beam theory. The formulation of governing equation includes the influence of mid-plane stretching of microbeam, fringing field and Casimir force incorporating correction for finite conductivity. The governing differential equation is solved using the Galerkin discretisation method and reduced order model technique. The results of the methodology implemented in current work are validated through comparison with reported numerical and experimental results. Subsequently, the effect of the Casimir force incorporating correction for the finite conductivity on the pull-in voltage and natural frequency is investigated in the presence of the residual stresses in a microbeam. The investigation shows that neglect of the Casimir force significantly overestimates the pull-in voltage; however, inclusion of corrections for the finite conductivity to the Casimir force indicates reduction in amount of overestimation in the pull-in voltage. The linear free vibration characteristics depict that microbeam with high mid-plane stretching parameter in the presence of the compressive residual stress offers tuning of natural frequency for more than two times. Further, the microbeam with high mid-plane stretching parameters provides resonant frequency stability characteristics at high voltage parameters though there is variation in the residual stress due to temperature variation during application. The present work also shows that these frequency tunability and stability characteristics are significantly influenced by the Casimir force. The results of the current work are expected to be valuable for the design of microbeam devices.

A nonlinear eigenmode solver for linear viscoelastic structures

This article deals with the nonlinear eigenvalue problem originating from the finite element discretization of mechanical structures involving linear viscoelastic material. The material function is assumed to be positive real, which allows a location of the eigenvalues in the left complex half space of the Laplace domain. The solution method for the considered nonlinear eigenvalue problem is based on the contour integral method, where special focus is put on the efficient numerical computation of the linear system along the boundary of the given search area. For this purpose, the reduced order model technique is used and appropriate a priori error estimates are provided. Finally, the validity of the proposed method is illustrated in numerical examples.

High Performance Reduced Order Modeling Techniques Based on Optimal Energy Quadrature: Application to Geometrically Non-linear Multiscale Inelastic Material Modeling

A High-Performance Reduced-Order Model (HPROM) technique, previously presented by the authors in the context of hierarchical multiscale models for non linear-materials undergoing infinitesimal strains, is generalized to deal with large deformation elasto-plastic problems. The proposed HPROM technique uses a Proper Orthogonal Decomposition procedure to build a reduced basis of the primary kinematical variable of the micro-scale problem, defined in terms of the micro-deformation gradient fluctuations. Then a Galerkin-projection, onto this reduced basis, is utilized to reduce the dimensionality of the micro-force balance equation, the stress homogenization equation and the effective macro-constitutive tangent tensor equation. Finally, a reduced goal-oriented quadrature rule is introduced to compute the non-affine terms of these equations. Main importance in this paper is given to the numerical assessment of the developed HPROM technique. The numerical experiments are performed on a micro-cell simulating a randomly distributed set of elastic inclusions embedded into an elasto-plastic matrix. This micro-structure is representative of a typical ductile metallic alloy. The HPROM technique applied to this type of problem displays high computational speed-ups, increasing with the complexity of the finite element model. From these results, we conclude that the proposed HPROM technique is an effective computational tool for modeling, with very large speed-ups and acceptable accuracy levels with respect to the high-fidelity case, the multiscale behavior of heterogeneous materials subjected to large deformations involving two well-separated scales of length.

Model reduction and mechanism for the vortex-induced vibrations of bluff bodies

We present an effective reduced-order model (ROM) technique to couple an incompressible flow with a transversely vibrating bluff body in a state-space format. The ROM of the unsteady wake flow is based on the Navier–Stokes equations and is constructed by means of an eigensystem realization algorithm (ERA). We investigate the underlying mechanism of vortex-induced vibration (VIV) of a circular cylinder at low Reynolds number via linear stability analysis. To understand the frequency lock-in mechanism and self-sustained VIV phenomenon, a systematic analysis is performed by examining the eigenvalue trajectories of the ERA-based ROM for a range of reduced oscillation frequency $(F_{s})$, while maintaining fixed values of the Reynolds number ($Re$) and mass ratio ($m^{\ast }$). The effects of the Reynolds number $Re$, the mass ratio $m^{\ast }$ and the rounding of a square cylinder are examined to generalize the proposed ERA-based ROM for the VIV lock-in analysis. The considered cylinder configurations are a basic square with sharp corners, a circle and three intermediate rounded squares, which are created by varying a single rounding parameter. The results show that the two frequency lock-in regimes, the so-called resonance and flutter, only exist when certain conditions are satisfied, and the regimes have a strong dependence on the shape of the bluff body, the Reynolds number and the mass ratio. In addition, the frequency lock-in during VIV of a square cylinder is found to be dominated by the resonance regime, without any coupled-mode flutter at low Reynolds number. To further discern the influence of geometry on the VIV lock-in mechanism, we consider the smooth curve geometry of an ellipse and two sharp corner geometries of forward triangle and diamond-shaped bluff bodies. While the ellipse and diamond geometries exhibit the flutter and mixed resonance–flutter regimes, the forward triangle undergoes only the flutter-induced lock-in for $30\leqslant Re\leqslant 100$ at $m^{\ast }=10$. In the case of the forward triangle configuration, the ERA-based ROM accurately predicts the low-frequency galloping instability. We observe a kink in the amplitude response associated with 1:3 synchronization, whereby the forward triangular body oscillates at a single dominant frequency but the lift force has a frequency component at three times the body oscillation frequency. Finally, we present a stability phase diagram to summarize the VIV lock-in regimes of the five smooth-curve- and sharp-corner-based bluff bodies. These findings attempt to generalize our understanding of the VIV lock-in mechanism for bluff bodies at low Reynolds number. The proposed ERA-based ROM is found to be accurate, efficient and easy to use for the linear stability analysis of VIV, and it can have a profound impact on the development of control strategies for nonlinear vortex shedding and VIV.

Nonlinear dynamics of an aeroelastic airfoil with free-play in transonic flow

Nonlinear dynamic behaviors of an aeroelastic airfoil with free-play in transonic air flow are studied. The aeroelastic response is obtained by using time-marching approach with computational fluid dynamics (CFD) and reduced order model (ROM) techniques. Several standardized tests of transonic flutter are presented to validate numerical approaches. It is found that in time-marching approach with CFD technique, the time-step size has a significant effect on the calculated aeroelastic response, especially for cases considering both structural and aerodynamic nonlinearities. The nonlinear dynamic behavior for the present model in transonic air flow is greatly different from that in subsonic regime where only simple harmonic oscillations are observed. Major features of the responses in transonic air flow at different flow speeds can be summarized as follows. The aeroelastic responses with the amplitude near the free-play are dominated by single degree of freedom flutter mechanism, and snap-though phenomenon can be observed when the air speed is low. The bifurcation diagram can be captured by using ROM technique, and it is observed that the route to chaos for the present model is via period-doubling, which is essentially caused by the free-play nonlinearity. When the flow speed approaches the linear flutter speed, the aeroelastic system vibrates with large amplitude, which is dominated by the aerodynamic nonlinearity. Effects of boundary layer and airfoil profile on the nonlinear responses of the aeroelastic system are also discussed.

PGD-VTCR: A Reduced Order Model Technique to Solve Medium Frequency Broad Band Problems on Complex Acoustical Systems

The calculation of vibrational responses of complex systems on frequency bands appears to be more and more important in engineering simulation. This is particularly true in the medium frequency regime where the solution is very sensitive to the frequency. In this work, we propose a new path to determine the frequency response of a system at many frequencies. It is based on the Variational Theory of Complex Rays (VTCR)[1], a mid frequency dedicated numerical strategy, and the Proper Generalized Decomposition (PGD) [2], a model order reduction technique. The VTCR enables one to model the problem thanks to the use of waves, and the PGD expands the VTCR approximation over the frequency band through a separated variable representation. This strategy is illustrated on a 2-D acoustic car cavity example.

NOVEL MULTI-DIMENSIONAL WCAWE TECHNIQUE FOR THE EFFICIENT CALCULATION OF RCS

In this paper, a novel moment-matching reduced order model technique termed the multi-dimensional well-conditioned asymptotic waveform evaluation (MDWCAWE) method is presented. The MDWCAWE method can be used to e-ciently determine the radar cross section (RCS) of arbitrarily shaped objects, in both the frequency and angular domains simultaneously. Numerical results are given in order to demonstrate the accuracy and robustness of the MDWCAWE method. All scattering problems investigated in this work are formulated using the two-dimensional volume-surface electric fleld integral equation (EFIE). We consider problems involving scattering from both dielectric dispersive and conducting objects.

Reduced-order model technique for the analysis of anisotropic inhomogeneous media: application to liquid-crystal displays.

Electromagnetic wave propagation in anisotropic inhomogeneous media is computed by a novel reduced-order model technique, which is based on the restriction of the Marcuvitz-Schwinger equations on Krylov subspaces and on the application of the singular-value decomposition. The model is derived from the standard coupled-wave method and includes both wide-angle diffraction and light scattering at dielectric interfaces. The method, currently implemented for two-dimensional problems, was applied to the analysis of different liquid-crystal test cells. Numerical results are compared with those obtained through the application of the coupled-wave method and the Jones method and with experimental microscopic measurements.