Adaptive greedy algorithms based on parameter-domain decomposition and reconstruction for the reduced basis method

The reduced basis method (RBM) empowers repeated and rapid evaluation of parametrized partial differential equations through an offline–online decomposition, a.k.a. a learning-execution process. A key feature of the method is a greedy algorithm repeatedly scanning the training set, a fine discretization of the parameter domain, to identify the next dimension of the parameter-induced solution manifold along which we expand the surrogate solution space. Although successfully applied to problems with fairly high parametric dimensions, the challenge is that this scanning cost dominates the offline cost due to it being proportional to the cardinality of the training set which is exponential with respect to the parameter dimension. In this work, we review three recent attempts in effectively delaying this curse of dimensionality, and propose two new hybrid strategies through successive refinement and multilevel maximization of the error estimate over the training set. All five offline-enhanced methods and the original greedy algorithm are tested and compared on two types of problems: the thermal block problem and the geometrically parameterized Helmholtz problem.     

A parallel multiselection greedy method for the radial basis function–based mesh deformation

Greedy algorithm has been widely adopted for the point selection procedure of radial basis function–based mesh deformation. However, in large deformation simulations with thousands of points selected, the greedy point selection will be too expensive and thus become a performance bottleneck. To improve the efficiency of the point selection procedure, a parallel multiselection greedy method has been developed in this paper. Multiple points are selected at each step to accelerate the convergence speed of the greedy algorithm. In addition, 2 strategies are presented to determine the specific selecting number. The parallelization of the greedy point selection is realized on the basis of a master‐slave model, and a hybrid decomposition algorithm is proposed to address the load imbalance problem. Numerical benchmarks show that both our multiselection method and the parallelization could obviously improve the point selection efficiency. Specifically, total speedups of 20 and 55 are separately obtained for the 3D undulating fish with 10⁶ cell mesh and the 3D rotating hydrofoil with 11 million cell mesh.     

Adaptive reduced basis strategy for rare-event simulations

Monte Carlo methods are well suited to characterize events of which associated probabilities are not too low with respect to the simulation budget. For very seldom observed events, these approaches do not lead to accurate results. Indeed, the number of samples is often insufficient to estimate such low probabilities (at least 10ⁿ⁺² samples are needed to estimate a probability of 10⁻ⁿ with 10% relative deviation of the Monte Carlo estimator). Even within the framework of reduced order methods, such as a reduced basis approach, it seems difficult to predict accurately low-probability events. In this paper, we propose to combine a cross-entropy method with a reduced basis algorithm to compute rare-event (failure) probabilities.     

A new boundary element method for resonance of an acoustic enclosure

Abstract

This paper presents a new boundary element formulation in which the eigenvalue appears outside the integral operator, which distinguishes it from the Helmholtz integral equation. Thus, the formation of global matrices need only be assembled once. Since the kernel of the operator used in the new formulation is real‐valued, all calculations can be carried out in a much simpler way in the real domain. The complex acoustic pressure amplitude is considered herein to deivate by a certain amount from a harmonic function. It is an important contribution that an exact relation between the deviator and the complex acoustic pressure amplitude is constructed locally and thus no more approximations are introduced except conventional boundary discretizations. Several examples are given to illustrate the feasibility of an accurate, effective prediction of resonance.     

Towards error bounds of the failure probability of elastic structures using reduced basis models

Structural reliability methods aim at computing the probability of failure of systems with respect to prescribed limit state functions. A common practice to evaluate these limit state functions is using Monte Carlo simulations. The main drawback of this approach is the computational cost, because it requires computing a large number of deterministic finite element solutions. Surrogate models, which are built from a limited number of runs of the original model, have been developed, as substitute of the original model, to reduce the computational cost. However, these surrogate models, while decreasing drastically the computational cost, may fail in computing an accurate failure probability. In this paper, we focus on the control of the error introduced by a reduced basis surrogate model on the computation of the failure probability obtained by a Monte Carlo simulation. We propose a technique to determine bounds of this failure probability, as well as a strategy of enrichment of the reduced basis, based on limiting the bounds of the error of the failure probability for a multi‐material elastic structure. Copyright © 2017 John Wiley & Sons, Ltd.     

Toward a multifrequency quasi‐static Ritz vector method for frequency‐dependent acoustic system application

Computational issues concerning the calculation of acoustic responses of a complex finite element (FE) model for various noise and vibration inputs have become prevalent. Such a model requires a significant amount of computation time because of repeated inversions of dynamic stiffness matrices. Thus, even state‐of‐the‐art computer hardware and software often face limitations where a model order reduction (MOR) scheme can help. The established MOR schemes such as Ritz vector or quasi‐static Ritz vector methods are efficient for general engineering systems, but these MOR methods become inaccurate for frequency response analyses in some acoustic systems with frequency‐dependent mass and stiffness matrices and force vectors (hereinafter frequency‐dependent acoustic systems). To cope with the inaccurate prediction by these methods for frequency‐dependent acoustic systems, this research presents and applies the multifrequency quasi‐static Ritz vector method. Unlike the Ritz vector or quasi‐static Ritz vector methods, the present multifrequency quasi‐static Ritz vector method employs direct Krylov subspace bases without an orthonormal procedure at multiple center frequencies. In comparison with the existing MOR scheme, a significant gain in computational efficiency is achieved, as well as enhanced accuracy. A comparison of these methods based on criteria such as efficiency, accuracy, and reliability was also conducted. Copyright © 2011 John Wiley & Sons, Ltd.     

A greedy MLPG method for identifying a control parameter in 2D parabolic PDEs

In this paper, we consider coefficient inverse problems, which are associated with the identification of unknown time dependent control parameter and unknown solution of two-dimensional parabolic inverse problem with overspecialization at a point in the spatial domain. After suitable finite difference approximation of time variable, an MLPG method is used for spatial discretization. To improve the efficiency of the MLPG method, a greedy algorithm is used. In fact, using the greedy algorithm, we avoid using more points from the data site than absolutely necessary and therefore, the method becomes more efficient. Comparison of the different kind of point selection and the effect of noisy data are performed for four test problems while our last test problem considers a problem with unknown solution. The results reveal that the method is efficient.     

Hybridization of stochastic reduced basis methods with polynomial chaos expansions

We propose a hybrid formulation combining stochastic reduced basis methods with polynomial chaos expansions for solving linear random algebraic equations arising from discretization of stochastic partial differential equations. Our objective is to generalize stochastic reduced basis projection schemes to non-Gaussian uncertainty models and simplify the implementation of higher-order approximations. We employ basis vectors spanning the preconditioned stochastic Krylov subspace to represent the solution process. In the present formulation, the polynomial chaos decomposition technique is used to represent the stochastic basis vectors in terms of multidimensional Hermite polynomials. The Galerkin projection scheme is then employed to compute the undetermined coefficients in the reduced basis approximation. We present numerical studies on a linear structural problem where the Youngs modulus is represented using Gaussian as well as lognormal models to illustrate the performance of the hybrid stochastic reduced basis projection scheme. Comparison studies with the spectral stochastic finite element method suggest that the proposed hybrid formulation gives results of comparable accuracy at a lower computational cost.     

Proper orthogonal decomposition‐based reduced basis element thermal modeling of integrated circuits

Various reduced basis element methods are compared for performing transient thermal simulations of integrated circuits. The reduced basis element method is a type of reduced order modeling that takes advantage of repeated geometrical features. It uses a reduced set of basis functions to approximate the solution of a PDE on subdomains (blocks); then these blocks are coupled together to perform a simulation on an entire domain. As the simulations are transient, a proper orthogonal decomposition basis is used, and the proper orthogonal decomposition eigenvalues from each block are used to derive error bounds for the entire simulation. These bounds are used to examine choices of block decompositions for a simplified integrated circuit structure. A decomposition that uses a single block for each transistor device is compared with a decomposition that uses one block for multiple devices. It was found that larger blocks are more computationally efficient; however, the advantage decreases if the devices within a block receive independent signals. Continuous and discontinuous methods of coupling the blocks were also compared. The coupling methods lend themselves to different solution approaches such as static condensation (continuous coupling) and block‐based inversion (discontinuous). Static condensation yielded the best convergence rate, accuracy, and operation count. Copyright © 2017 John Wiley & Sons, Ltd.     

A BEM sensitivity and shape identification analysis for acoustic scattering in fluid–solid problems

In this paper a boundary element formulation for the sensitivity analysis of structures immersed in an inviscide fluid and illuminated by harmonic incident plane waves is presented. Also presented is the sensitivity analysis coupled with an optimization procedure for analyses of flaw identification problems. The formulation developed utilizes the boundary integral equation of the Helmholtz equation for the external problem and the Cauchy–Navier equation for the internal elastic problem. The sensitivities are obtained by the implicit differentiation technique. Examples are presented to demonstrate the accuracy of the proposed formulations. © 1998 John Wiley & Sons, Ltd.     

Adaptive reduced order modeling for nonlinear dynamical systems through a new a posteriori error estimator: Application to uncertainty quantification

Multiquery problems such as uncertainty quantification (UQ), optimization of a dynamical system require solving a differential equation at multiple parameter values. Therefore, for large systems, the computational cost becomes prohibitive. This issue can be addressed by using a cheaper reduced order model (ROM) instead. However, the ROM entails error in the solution due to approximation in a lower dimensional subspace. Moreover, the ROM lacks robustness over a wide range of parameter values. To address these issues, first, an upper bound on the norm of the state transition matrix is derived. This bound, along with the residual in the governing equation, are then used to develop an error estimator for general nonlinear dynamical systems. Furthermore, this error estimator is used in conjunction with the modified greedy search algorithm proposed by Hossain and Ghosh (Int J Numer Methods Eng, 2018;116(12-13): 741-758) to adaptively construct a robust proper orthogonal decomposition-based ROM. This adaptive ROM is subsequently deployed for UQ by invoking it in a statistical simulation. Two numerical studies: (i) viscous Burgers' equation and (ii) beam on nonlinear Winkler foundation, showed an improved accuracy of the error estimator compared to the current literature. A significant computational speed-up in UQ is achieved.     

A dual-reciprocity method for acoustic radiation in a subsonic non-uniform flow

In this paper, the dual-reciprocity boundary-element method is used to model acoustic radiation in a subsonic non-uniform-flow field. The boundary-integral formulation is based on a direct boundary-integral equation developed very recently by the authors for acoustic radiation in a subsonic uniform flow. All the terms due to the non-uniform-flow effect are taken to the right-hand side and treated as source terms. The source terms result in a domain integral in the standard boundary-integral formulation. The dual-reciprocity method is then used to transform the resulting domain integral into boundary integrals. Numerical tests show reasonably good agreement with an analytical soution for a pulsating sphere submerged in a potential-flow field.     

Optimization of a variable mouth acoustic horn

By using boundary shape optimization on the end part of a semi‐infinite waveguide for acoustic waves, we design transmission‐efficient interfacial devices without imposing an upper bound on the mouth diameter. The boundary element method solves the Helmholtz equation modeling the exterior wave propagation problem. A gradient‐based optimization algorithm solves the resulting least‐squares problem and the adjoint method provides the necessary gradients. The results demonstrate that there appears to be a natural limit on the optimal mouth diameter. Copyright © 2010 John Wiley & Sons, Ltd.     

Isogeometric FEM‐BEM coupled structural‐acoustic analysis of shells using subdivision surfaces

We introduce a coupled finite and boundary element formulation for acoustic scattering analysis over thin‐shell structures. A triangular Loop subdivision surface discretisation is used for both geometry and analysis fields. The Kirchhoff‐Love shell equation is discretised with the finite element method and the Helmholtz equation for the acoustic field with the boundary element method. The use of the boundary element formulation allows the elegant handling of infinite domains and precludes the need for volumetric meshing. In the present work, the subdivision control meshes for the shell displacements and the acoustic pressures have the same resolution. The corresponding smooth subdivision basis functions have the C¹ continuity property required for the Kirchhoff‐Love formulation and are highly efficient for the acoustic field computations. We verify the proposed isogeometric formulation through a closed‐form solution of acoustic scattering over a thin‐shell sphere. Furthermore, we demonstrate the ability of the proposed approach to handle complex geometries with arbitrary topology that provides an integrated isogeometric design and analysis workflow for coupled structural‐acoustic analysis of shells.     

A DEM-based method for predicting the wear evolution of structural boundary composed of spherical boundary elements

In the discrete element method (DEM) simulation for wear prediction, structural boundary is now represented extensively by triangular meshes with high resolution, which brings a huge computational cost. A DEM-based method for predicting the wear evolution of structural boundary has been developed for computational efficiency. The structural boundary subjected to wear is represented by the spherical boundary elements in the DEM simulation in combination with the inside triangles and fitting curved surface in wear prediction. Wear prediction is performed through a series of evolution steps. In each evolution step, the collision energies by particles at structural boundary are collected via the DEM simulation and assigned to the boundary elements. Then, the volume losses of structural boundary are predicted across each relevant boundary element. Finally, the new geometry of structural boundary in response to wear is described by moving the boundary elements along the depths of wear individually. Through converting the contact detection between structural boundary and particles into between spherical boundary elements and particles, our method greatly reduces the computational cost in the DEM simulation. Through two numerical tests, our method has been verified to be an efficient and accurate method for the wear prediction of structural boundaries with different resolutions.     

Testing complete and compact radial basis functions for solution of eigenvalue problems using the boundary element method with direct integration

This paper analyses the performance of the main radial basis functions in the formulation of the Boundary Element Method (DIBEM). This is an alternative for solving problems modeled by non-adjoint differential operators, since it transforms domain integrals in boundary integrals using radial basis functions. The solution of eigenvalue problem was chosen to performance evaluation. Natural frequencies are calculated numerically using several radial functions and their accuracy is evaluated by comparison with the available analytical solutions and with the Finite Element Method as well. The standard radial basis functions have presented similar performance to compact radial functions, being even slightly superior.     

A robust time‐space formulation for large‐scale scalar wave problems using exponential basis functions

In this paper, a new effective boundary node method is presented for the solution of acoustic problems, directly in time domain, using exponential basis functions. Unlike many other methods using boundary information, the final coefficient matrix is sparse. The formulation is well suited for domains whose extent is relatively larger than the distance traveled by the acoustic wave in an increment of time. The exponential basis functions used satisfy the time‐space governing equation. This helps to choose a relatively large time increment and a moderate number of boundary points, which leads to reduction of computation time. The computation is performed incrementally using a weighted residual in time. Through a series of numerical examples, it is shown that the method, when combined with a domain decomposition strategy, is effectively capable of solving various 1‐ to 3‐dimensional acoustic problems.     

A fictitious domain decomposition method for the solution of partially axisymmetric acoustic scattering problems. Part 2: Neumann boundary conditions

We present a fictitious domain decomposition method for the fast solution of acoustic scattering problems characterized by a partially axisymmetric sound‐hard scatterer. We apply this method to the solution of a mock‐up submarine problem, and highlight its computational advantages and intrinsic parallelism. A key component of our method is an original idea for addressing a Neumann boundary condition in the general framework of a fictitious domain method. This idea is applicable to many other linear partial differential equations besides the Helmholtz equation. Copyright © 2003 John Wiley & Sons, Ltd.     

An efficient indirect boundary element technique for multi‐frequency acoustic analysis

An efficient indirect boundary element solution procedure for the analysis of multi‐frequency acoustic problems is developed by incorporating techniques that improve the efficiency of the integration and matrix solution phases of the computing process. The integration phase is made efficient by computing the system matrices at few predetermined key frequencies only and then evaluating the matrices at other intermediate frequencies by quadratic interpolation. The matrix solution process is made efficient by iterating the solutions using the factored form of the key frequency matrices. The effectiveness of the present development is confirmed by solving a number of example problems. Copyright © 1999 John Wiley & Sons, Ltd.     

考虑影响声传播因素的车用交流发电机气动噪声预测

为了更准确地预测车用交流发电机的气动噪声,基于计算流体力学及声类比理论,考虑影响声传播的因素,对实验室安装条件下的某型车用交流发电机气动噪声进行研究。利用大涡模拟方法计算了交流发电机内部三维非稳态流场;依据Lighthill声类比思想,将转子表面的压力脉动等效为旋转偶极子源点集;考虑发电机机壳及实验台面对声传播的影响,建立了以机壳内表面为声源边界的半自由声场计算模型,进而预测了发电机的远场气动噪声;最后,利用实测数据对发电机气动噪声仿真结果进行了验证。结果表明:交流发电机气动噪声的辐射声场具有明显的偶极子指向特性;仿真计算结果与实验测试结果具有很好的一致性。所提的研究方法能更准确地预测发电机的气动噪声,同时可为实车安装条件下的车用交流发电机气动噪声预测提供参考。