Standard Proper Orthogonal Decomposition Research Articles

Model order reduction for optimality systems through empirical gramians

In the present article, optimal control problems for linear parabolic partial differential equations (PDEs) with time-dependent coefficient functions are considered. One of the common approach in literature is to derive the first-order sufficient optimality system and to apply a finite element (FE) discretization. This leads to a specific linear but high-dimensional time variant (LTV) dynamical system. To reduce the size of the LTV system, we apply a tailored reduced order modeling technique based on empirical gramians and derived directly from the first-order optimality system. For testing purpose, we focus on two specific examples: a multiobjective optimization and a closed-loop optimal control problem. Our proposed methodology results to be better performing than a standard proper orthogonal decomposition (POD) approach for the above mentioned examples.

Space-time POD and the Hankel matrix.

Time-delay embedding is an increasingly popular starting point for data-driven reduced-order modeling efforts. In particular, the singular value decomposition (SVD) of a block Hankel matrix formed from successive delay embeddings of the state of a dynamical system lies at the heart of several popular reduced-order modeling methods. In this paper, we show that the left singular vectors of this Hankel matrix are a discrete approximation of space-time proper orthogonal decomposition (POD) modes, and the singular values are square roots of the POD energies. Analogous to the connection between the SVD of a data matrix of snapshots and space-only POD, this connection establishes a clear interpretation of the Hankel modes grounded in classical theory, and we gain insights into the Hankel modes by instead analyzing the equivalent discrete space-time POD modes in terms of the correlation matrix formed by multiplying the Hankel matrix by its conjugate transpose. These insights include the distinct meaning of rows and columns, the implied norm in which the modes are optimal, the impact of the time step between snapshots on the modes, and an interpretation of the embedding dimension/height of the Hankel matrix in terms of the time window on which the modes are optimal. Moreover, the connections we establish offer opportunities to improve the convergence and computation time in certain practical cases, and to improve the accuracy of the modes with the same data. Finally, popular variants of POD, namely the standard space-only POD and spectral POD, are recovered in the limits that snapshots used to form each column of the Hankel matrix represent flow evolution over short and long times, respectively.

Experimental investigation of flow unsteadiness over four types of spiked-blunt body at Mach 2.2

Flow unsteadiness over four types of spiked-blunt body are investigated through wind tunnel experiments at a freestream Mach number of 2.2 at 0° and 4° angle of attack. The flow pulsation is visualized with high spatiotemporal resolution schlieren and then the shock-related unsteadiness is investigated through proper orthogonal decomposition (POD) method. Firstly, two special states in the suppressed pulsation mode are observed, namely the phase of withhold and the phase of smash. Flow over the standard aerodome spike will continuously transform between the two states, which can be distinguished from the distinct flow structures. Spectral analysis of unsteady pressure measurement detected the dominant frequencies of 2.2 kHz and 3.2 kHz for pointed spike and standard aerodome spike, respectively, and POD analysis further finds that the signal originates from the periodic motion of the aftershock. The large-amplitude flow pulsation (amplitude 0.23D) in typical pulsation mode is transformed into the high-frequency small-amplitude pulsation (amplitude 0.09D) in the suppressed mode. In addition, it is found that the variation of angle of attack makes little influence on the dominant frequency. And it is hard to find the main frequency of double/inverse aerodome spike, which indicates that the flow unsteadiness of different aerodome cases is not the same.

POD-Based reduced order methods for optimal control problems governed by parametric partial differential equation with varying boundary control

In this work we propose tailored model order reduction for varying boundary optimal control problems governed by parametric partial differential equations. With varying boundary control, we mean that a specific parameter changes where the boundary control acts on the system. This peculiar formulation might benefit from model order reduction. Indeed, fast and reliable simulations of this model can be of utmost usefulness in many applied fields, such as geophysics and energy engineering. However, varying boundary control features very complicated and diversified parametric behaviour for the state and adjoint variables. The state solution, for example, changing the boundary control parameter, might feature transport phenomena. Moreover, the problem loses its affine structure. It is well known that classical model order reduction techniques fail in this setting, both in accuracy and in efficiency. Thus, we propose reduced approaches inspired by the ones used when dealing with wave-like phenomena. Indeed, we compare standard proper orthogonal decomposition with two tailored strategies: geometric recasting and local proper orthogonal decomposition. Geometric recasting solves the optimization system in a reference domain simplifying the problem at hand avoiding hyper-reduction, while local proper orthogonal decomposition builds local bases to increase the accuracy of the reduced solution in very general settings (where geometric recasting is unfeasible). We compare the various approaches on two different numerical experiments based on geometries of increasing complexity.

Dynamic Mode Decomposition for the Comparison of Engine In-Cylinder Flow Fields from Particle Image Velocimetry (PIV) and Reynolds-Averaged Navier–Stokes (RANS) Simulations

Validation of Reynolds-averaged Navier–Stokes (RANS) simulation results against experimental data such as flow measurements from particle image velocimetry (PIV) remains a challenge for the development of thermal propulsion systems. This is partly due to cycle-to-cycle variations (CCVs) in the air motion and partly due to uncertainties in the PIV measurement technique, complicating the question of what constitutes a fair validation target for the RANS model. Indeed, an inappropriate validation target can misguide subsequent adjustments of a RANS model. In this work, the ensemble-averaged PIV field is first investigated for its suitability as a validation target for RANS simulations. The relevance index and the velocity histogram distance are used as quantitative metrics to assess the similarity of the ensemble-averaged field to the full dataset of individual PIV cycles. While a high similarity is seen between the average PIV flow field and the individual cycles on the tumble plane, the similarity is lower and more variable on the cross-tumble plane, where there are significant CCVs. Standard (space-only, phase-dependent) proper orthogonal decomposition (POD) is employed as an alternative method of data processing with the aim of providing a fairer comparison to RANS simulations. The cycle-dependence of the standard POD modes is shown to be an aspect that results in many validation targets and an excessively broad validation range, limiting its utility in this context. Dynamic mode decomposition (DMD) and sparsity-promoting dynamic mode decomposition (SPDMD) are then proposed as alternative solutions, capable of extracting flow structures at specific frequencies. The background 0 Hz SPDMD modes exhibit an ability to produce more realistic flow fields with velocity magnitudes that are significantly closer to the individual cycles.

A new approach to proper orthogonal decomposition with difference quotients

In a recent work (Koc et al., SIAM J. Numer. Anal. 59(4), 2163–2196, 2021), the authors showed that including difference quotients (DQs) is necessary in order to prove optimal pointwise in time error bounds for proper orthogonal decomposition (POD) reduced order models of the heat equation. In this work, we introduce a new approach to including DQs in the POD procedure. Instead of computing the POD modes using all of the snapshot data and DQs, we only use the first snapshot along with all of the DQs and special POD weights. We show that this approach retains all of the numerical analysis benefits of the standard POD DQ approach, while using a POD data set that has approximately half the number of snapshots as the standard POD DQ approach, i.e., the new approach requires less computational effort. We illustrate our theoretical results with numerical experiments.

Hierarchical Model Reduction Driven by Machine Learning for Parametric Advection-Diffusion-Reaction Problems in the Presence of Noisy Data

We propose a new approach to generate a reliable reduced model for a parametric elliptic problem, in the presence of noisy data. The reference model reduction procedure is the directional HiPOD method, which combines Hierarchical Model reduction with a standard Proper Orthogonal Decomposition, according to an offline/online paradigm. In this paper we show that directional HiPOD looses in terms of accuracy when problem data are affected by noise. This is due to the interpolation driving the online phase, since it replicates, by definition, the noise trend. To overcome this limit, we replace interpolation with Machine Learning fitting models which better discriminate relevant physical features in the data from irrelevant unstructured noise. The numerical assessment, although preliminary, confirms the potentialities of the new approach.

Physics-based manifold learning in scaffolds for tissue engineering: Application to inverse problems

In the field of bone regeneration, insertion of scaffolds favours bone formation by triggering the differentiation of mesenchymal cells into osteoblasts. The presence of Calcium ions (Ca2+) in the interstitial fluid across scaffolds is thought to play a relevant role in the process. In particular, the Ca2+ patterns can be used as an indicator of where to expect bone formation. In this work, we analyse the inverse problem for these distribution patterns, using an advection-diffusion nonlinear model for the concentration of Ca2+. That is, given a set of observables which are related to the amount of expected bone formation, we aim at determining the values of the parameters that best fit the data. The problem is solved in a realistic 3D-printed structured scaffold for two uncertain parameters: the amplitude of the velocity of the interstitial fluid and the ionic release rate from the scaffold. The minimization in the inverse problem requires multiple evaluations of the nonlinear model. The computational cost is alleviated by the combination of standard Proper Orthogonal Decomposition (POD), to reduce the number of degrees of freedom, with an adhoc hyper-reduction strategy, which avoids the assembly of a full-order system at every iteration of the Newton’s method. The proposed hyper-reduction method is formulated using the Principal Component Analysis (PCA) decomposition of suitable training sets, devised from the weak form of the problem. In the numerical tests, the hyper-reduced formulation leads to accurate results with a significant reduction of the computational demands with respect to standard POD.

Augmented reduced order models for turbulence

The authors introduce an augmented-basis method (ABM) to stabilize reduced-order models (ROMs) of turbulent incompressible flows. The method begins with standard basis functions derived from proper orthogonal decomposition (POD) of snapshot sets taken from a full-order model. These are then augmented with divergence-free projections of a subset of the nonlinear interaction terms that constitute a significant fraction of the time-derivative of the solution. The augmenting bases, which are rich in localized high wavenumber content, are better able to dissipate turbulent kinetic energy than the standard POD bases. Several examples illustrate that the ABM significantly out-performs L2-, H1- and Leray-stabilized POD ROM approaches. The ABM yields accuracy that is comparable to constraint-based stabilization approaches yet is suitable for parametric model-order reduction in which one uses the ROM to evaluate quantities of interests at parameter values that differ from those used to generate the full-order model snapshots. Several numerical experiments point to the importance of localized high wavenumber content in the generation of stable, accurate, and efficient ROMs for turbulent flows.

Efficient algorithm for proper orthogonal decomposition of block-structured adaptively refined numerical simulations

Adaptive mesh refinement (AMR) is increasingly being used to simulate fluid flows that have vastly different resolution requirements throughout the computational domain. Proper orthogonal decomposition (POD) is a common tool to extract coherent structures from flow data and build reduced order models, but current POD algorithms do not take advantage of potential efficiency gains enabled by multi-resolution data from AMR simulations. Here, we explore a new method for performing POD on AMR data that eliminates repeated operations arising from nearest-neighbor interpolation of multi-resolution data onto uniform grids. We first outline our approach to reduce the number of computations with examples and provide the complete algorithms in the appendix. We examine the computational acceleration of the new algorithms compared to the standard POD method using synthetically generated AMR data and operation counting. We then use CPU times and operation counting to analyze data from an AMR simulation of an axisymmetric buoyant plume, finding that we are able to reduce the computational time by a factor of approximately 2−5 when using three levels of grid refinement. The new POD algorithm is the first to eliminate redundant operations for matrix multiplications with repeated values in each matrix, making it ideal for POD of data from AMR simulations.

Investigation of local severe suction on the side walls of a high-rise building by standard, spectral and conditional POD

Local severe suction greatly determines the safe and economic wind-resistant design of cladding system in the façade of buildings. This study focuses on two localized regions with severe suction of a building, i.e., the lower windward corner of side wall and the region immediately upstream of trailing edge. The aspect ratio of building is 3. The characteristics and mechanism of wind peak pressures are investigated by standard, spectral and conditional proper orthogonal decomposition (POD). Spectral POD extracts the space–time modes which oscillate at the single frequency of the anti-symmetric vortex shedding, where the pressure distributions have the shape of ‘dragonfly’ wing and opposite signs on two side walls. Conditional POD is demonstrated to be a powerful tool to study the average evolution of peak pressures. By inspection of conditional POD modes, high positive fluctuating pressure is observed in the trailing portion of the lower side wall, accompanied by the minimum pressure in the windward corner. The identified pressure behaviors provide the statistical evidences for flow reattachment and shear-layer roll-up which were observed by instantaneous flow visualization in the literature. The trailing-edge peak pressure in the middle-height region is induced by trailing-edge vortex of great vorticity concentration. The flow mechanism and occurrence phase of trailing-edge peak pressures are similar to those of infinite-length cylinders.

Error analysis of proper orthogonal decomposition data assimilation schemes with grad–div stabilization for the Navier–Stokes equations

The error analysis of a proper orthogonal decomposition (POD) data assimilation (DA) scheme for the Navier–Stokes equations is carried out. A grad–div stabilization term is added to the formulation of the POD method. Error bounds with constants independent on inverse powers of the viscosity parameter are derived for the POD algorithm. No upper bounds in the nudging parameter of the data assimilation method are required. Numerical experiments show that, for large values of the nudging parameter, the proposed method rapidly converges to the real solution, and greatly improves the overall accuracy of standard POD schemes up to low viscosities over predictive time intervals.

Wind peak pressures on a square-section cylinder: Flow mechanism and standard/conditional POD analyses

This study emphasizes the flow mechanism of negative peak pressures on a square-section cylinder at Reynolds number of 22,000, particularly of those immediately upstream of the trailing edges (trailing-edge peak pressure). The minimum trailing-edge pressure coefficients are observed as low as -5 and have a close correlation in a streamwise length of approximately 0.1D (D is the width of the cylinder). The occurrence of small trailing-edge vortex, which has highly concentrated vorticity, is responsible for the trailing-edge peak pressures. The trailing-edge peak pressure occurs immediately before the shear layer moves far from the cylinder and recovers to the symmetric state. Standard proper orthogonal decomposition (POD) is applied to discover the inherent spatial modes with sharp peak pressures upstream of the trailing edges. Instead of the 1st mode (corresponding to vortex shedding), the higher-order modes contribute significantly to the trailing-edge peak pressures, whose mode coefficients have harmonics of vortex-shedding frequency. More than nine POD modes are necessary to reconstruct the trailing-edge peak pressures. As a variant of POD proposed recently, conditional space–time POD is valid in studying the average evolution of wind peak pressures. It further provides the relative phase and frequency ratio between the trailing-edge peak pressure and vortex shedding.

A non-intrusive space-time interpolation from compact Stiefel manifolds of parametrized rigid-viscoplastic FEM problems

This work aims to interpolate parametrized reduced order model (ROM) basis constructed via the proper orthogonal decomposition (POD) to derive a robust ROM of the system’s dynamics for an unseen target parameter value. A novel non-intrusive space-time (ST) POD basis interpolation scheme is proposed, for which we define ROM spatial and temporal basis curves on compact Stiefel manifolds. An interpolation is finally defined on a mixed part encoded in a square matrix directly deduced using the spacial part, the singular values and the temporal part, to obtain an interpolated snapshot matrix, keeping track of accurate space and temporal eigenvectors. Moreover, in order to establish a well-defined curve on the compact Stiefel manifold, we introduce a new procedure, the so-called oriented SVD. Such an oriented SVD produces unique right and left eigenvectors for generic matrices, for which all singular values are distinct. It is important to notice that the ST POD basis interpolation does not require the construction and the subsequent solution of a reduced-order FEM model as classically is done. Hence it is avoiding the bottleneck of standard POD interpolation which is associated with the evaluation of the nonlinear terms of the Galerkin projection on the governing equations. As a proof of concept, the proposed method is demonstrated with the adaptation of rigid-thermoviscoplastic finite element ROMs applied to a typical nonlinear open forging metal forming process. Strong correlations of the ST POD models with respect to their associated high-fidelity FEM counterpart simulations are reported, highlighting its potential use for near real-time parametric simulations using off-line computed ROM POD databases.

Experimental investigation of turbulent coherent structures interacting with a porous airfoil

The flow field on solid and porous airfoils subjected to turbulence shed by an upstream cylindrical rod and the corresponding far-field noise radiations are studied through particle image velocimetry (PIV) and microphone measurements, respectively. Three different Reynolds numbers based on the rod diameter are considered in a range between $${2.7\times 10^4}$$ and $${5.4\times 10^4}$$ , and two porous airfoil models are tested to analyze the influence of the design elements of the permeable treatment. A standard proper orthogonal decomposition (POD) algorithm is employed to band filter the different length scales that characterize the turbulent flow, making it feasible to determine which turbulence scales are affected by porosity. The aeroacoustic results indicate that the porous treatment of the wing profile leads to a noise reduction at low frequencies and a noise regeneration at high frequencies due to surface roughness. The investigation on the flow field shows that the main effect of porosity is to mitigate the turbulent kinetic energy in the stagnation region, attenuating the distortion of turbulence interacting with the airfoil surface. The application of the POD algorithm indicates that this effect acts mainly on the largest scales of turbulence.

High performance reduction technique for multiscale finite element modeling (HPR-FE2): Towards industrial multiscale FE software

The authors have shown in previous contributions that reduced order modeling with optimal cubature applied to finite element square (FE2) techniques results in a reliable and affordable multiscale approach, the HPR-FE2 technique. Such technique is assessed here for an industrial case study of a generic 3D reinforced composite whose microstructure is represented by two general microcells accounting for different deformation mechanisms, microstrucural phases and geometry arrangement. Specifically, in this approach the microstrain modes used for building the reduced order model (ROM) are obtained through standard proper orthogonal decomposition (POD) techniques applied over snapshots of a representative sampling strain space. Additionally, a reduced number of integration points is obtained by exactly integrating the main free energy modes resulting from the sampling energy snapshots. The outcome consists of a number of dominant strain modes integrated over a remarkably reduced number of integration points which provide the support to evaluate the constitutive behavior of the microstructural phases. It is emphasized that stresses are computed according to the selected constitutive law at the reduced integration points and, therefore, the strategy inherits advantageous properties such as model completeness and customization of material properties. Overall results are discussed in terms of the consistency of the multiscale analysis, customization of the microscopic material parameters and speedup ratios compared to high-fidelity finite element (HF) simulations.

Multiscale proper orthogonal decomposition (mPOD) of TR-PIV data—a case study on stationary and transient cylinder wake flows

Data-driven decompositions of particle image velocimetry (PIV) measurements are widely used for a variety of purposes, including the detection of coherent features (e.g. vortical structures), filtering operations (e.g. outlier removal or random noise mitigation), data reduction and compression. This work presents the application of a novel decomposition method, referred to as multiscale proper orthogonal decomposition (mPOD, Mendez et al 2019) to time-resolved PIV (TR-PIV) measurement. This method combines multiresolution analysis and standard proper orthogonal decomposition (POD) to achieve a compromise between decomposition convergence and spectral purity of the resulting modes. The selected test case is the flow past a cylinder in both stationary and transient conditions, producing a frequency-varying Karman vortex street. The results of the mPOD are compared to the standard POD, the discrete Fourier transform and the dynamic mode decomposition. The mPOD is evaluated in terms of decomposition convergence and time-frequency localization of its modes. The multiscale modal analysis allows for revealing beat phenomena in the stationary cylinder wake, due to the three-dimensional nature of the flow, and to correctly identify the transition from various stationary regimes in the transient test case.

Multi-scale proper orthogonal decomposition of complex fluid flows

Data-driven decompositions are becoming essential tools in fluid dynamics, allowing for tracking the evolution of coherent patterns in large datasets, and for constructing low-order models of complex phenomena. In this work, we analyse the main limits of two popular decompositions, namely the proper orthogonal decomposition (POD) and the dynamic mode decomposition (DMD), and we propose a novel decomposition which allows for enhanced feature detection capabilities. This novel decomposition is referred to as multi-scale proper orthogonal decomposition (mPOD) and combines multi-resolution analysis (MRA) with a standard POD. Using MRA, the mPOD splits the correlation matrix into the contribution of different scales, retaining non-overlapping portions of the correlation spectra; using the standard POD, the mPOD extracts the optimal basis from each scale. After introducing a matrix factorization framework for data-driven decompositions, the MRA is formulated via one- and two-dimensional filter banks for the dataset and the correlation matrix respectively. The validation of the mPOD, and a comparison with the discrete Fourier transform (DFT), DMD and POD are provided in three test cases. These include a synthetic test case, a numerical simulation of a nonlinear advection–diffusion problem and an experimental dataset obtained by the time-resolved particle image velocimetry (TR-PIV) of an impinging gas jet. For each of these examples, the decompositions are compared in terms of convergence, feature detection capabilities and time–frequency localization.

Reduced order models for thermally coupled low Mach flows

In this paper we present a collection of techniques used to formulate a projection-based reduced order model (ROM) for zero Mach limit thermally coupled Navier–Stokes equations. The formulation derives from a standard proper orthogonal decomposition (POD) model reduction, and includes modifications to improve the drawbacks caused by the inherent non-linearity of the used Navier–Stokes equations: a hyper-ROM technique based on mesh coarsening; an implicit ROM subscales formulation based on a variational multi-scale (VMS) framework; and a Petrov–Galerkin projection necessary in the case of non-symmetric terms. At the end of the article, we test the proposed ROM formulation using 2D and 3D versions of the same example: a differentially heated cavity.

Reduced-Order Modeling of Subsurface Multi-phase Flow Models Using Deep Residual Recurrent Neural Networks

We present a reduced-order modeling technique for subsurface multi-phase flow problems building on the recently introduced deep residual recurrent neural network (DR-RNN) (Nagoor Kani et al. in DR-RNN: a deep residual recurrent neural network for model reduction. ArXiv e-prints, 2017). DR-RNN is a physics-aware recurrent neural network for modeling the evolution of dynamical systems. The DR-RNN architecture is inspired by iterative update techniques of line search methods where a fixed number of layers are stacked together to minimize the residual (or reduced residual) of the physical model under consideration. In this manuscript, we combine DR-RNN with proper orthogonal decomposition (POD) and discrete empirical interpolation method (DEIM) to reduce the computational complexity associated with high-fidelity numerical simulations. In the presented formulation, POD is used to construct an optimal set of reduced basis functions and DEIM is employed to evaluate the nonlinear terms independent of the full-order model size. We demonstrate the proposed reduced model on two uncertainty quantification test cases using Monte Carlo simulation of subsurface flow with random permeability field. The obtained results demonstrate that DR-RNN combined with POD–DEIM provides an accurate and stable reduced model with a fixed computational budget that is much less than the computational cost of standard POD–Galerkin reduced model combined with DEIM for nonlinear dynamical systems.