Engineering Analysis with Boundary Elements最新文献

There is a growing body of literature that recognizes the importance of Smoothed Finite Element Method (S-FEM) in computational fluid dynamics (CFD) fields and, to a lesser extent, in complex turbulent flow problems. This study evaluates the performance of Reynolds-averaged Navier-Stokes (RANS) turbulence models within the S-FEM framework for predicting incompressible turbulent flows. Our assessment of three turbulence models based on the cell-based S-FEM (CS-FEM) is convincingly supported by testing on three flow problems. It is found that the CS-FEM exhibits superior mesh robustness compared to the Finite Volume Method (FVM) and achieves higher computational accuracy than the Finite Element Method (FEM). Notably, the CS-FEM combined with the standard k-epsilon model (CS-FEM-SKE) and the realizable k-epsilon model (CS-FEM-RKE) demonstrate robust performance in handling severely distorted meshes, with CS-FEM-RKE outperforming in regions of strong flow separation and convection. The Spalart-Allmaras model with CS-FEM (CS-FEM-SA) offers faster computational speed but shows poor mesh robustness. The hexcore mesh based on CS-FEM-RKE is employed to evaluate the aerodynamic performance of High-speed train (HST), resulting in enhanced computational efficiency. The outcomes show good agreement with other numerical studies and experimental data. Overall, it also highlights the latent capability of CS-FEM in solving complex engineering problems.

This paper conducts the dynamic analysis of fractional poroviscoelastic reinforced subgrade under moving loading. Based on the Biot theory and transversely isotropic (TI) parameter expression of the geogrid reinforced subgrade, the governing equations of the poroelastic reinforced subgrade are established in the wavenumber domain by the double Fourier transform. Considering the viscosity of the soil skeleton and the flow-dependent viscosity between the soil skeleton and pore water, the governing equations are extended to the fractional poroviscoelastic medium by introducing the Zener viscoelastic model, fractional calculus theory and the dynamic elastic-viscoelastic correspondence principle. Combining boundary conditions and interlayer continuity conditions, the extended precise integration method (PIM) and double Fourier integral transform are employed to obtain the solution of fractional poroviscoelastic reinforced subgrade in the spatial domain. After the numerical validation, a sensitivity analysis of the relaxation time, permeability, reinforcement ratio and the load velocity are conducted.

To quantify the influence of moving immersed media on acoustic radiation, this study develops an efficient method for acoustic radiation with arbitrary immersed media interfaces based on the extended finite element method (XFEM) and the Dirichlet-to-Neumann (DtN) boundary condition. The XFEM is employed for efficient and accurate modeling of the acoustic field with boundary shape variations. It requires no modification of the computational mesh and accurately captures non-smooth solutions on the interface by constructing enrichment functions. Additionally, the DtN boundary condition simulates the far-field radiation condition by establishing the relationship between the acoustic pressure and its derivatives. Numerical examples show that the proposed method efficiently characterizes changes in the position of immersed media interfaces without re-meshing the mesh. Variations in the thickness of porous material domains alter the acoustic radiation characteristics, with thicker porous material domains resulting in more pronounced noise reduction effects. Compared to changes in the thickness of porous material domains, changes in their position significantly alter the distribution of radiation pressure, indicating that ideal noise reduction effects can be achieved by strategically placing porous materials in specific locations in practical engineering applications.

A singular integral equation method is proposed to analyze the complex plane cracks in one-dimensional (1D) orthorhombic quasicrystals. Using the Somigliana formula, the singular integral equations of the curved crack are derived. Based on the general situation of the curved crack, the singular integral equations of the inclined crack and the arc crack are given. Then the analytical solutions for the singular phonon and phason stresses near the tips of the inclined and the arc crack are obtained. Gauss-Chebyshev quadrature method is introduced to calculate the singular integral equations, and a numerical algorithm for solving the stress intensity factor (SIF) is proposed. Numerical solutions for the phonon and phason SIFs of some examples are solved and discussed.

In this paper, a novel method is developed to solve the free-field motion of the non-horizontally layered half-space subjected to seismic excitation in the time domain. The total wave motions are decomposed into a known and an unknown wave motion. Making use of the fact that the nodal forces at nodes in half-space resulted from the two motions will be zeros, the scattering problem resulted from the seismic excitation is transformed into a radiation problem. The radiation damping of the unbounded layered foundation in the time domain is expressed by the acceleration unit-impulse response matrix obtained using the scaling surface based Scaled Boundary Finite Element Method (SBFEM). In the numerical examples, firstly, the accuracy of the scaling surfaced based SBFEM in simulating the radiation damping is demonstrated by surface excitations in the layered half-space. Then, a time-domain analysis of the free-field motion of a horizontally layered half-space is studied to verify the accuracy and validity of the proposed method. Finally, a study of the free-field motion of non-horizontally layered half-space is investigated, and the results show that increasing the dimensions of the computational domain can significantly improve the accuracy.

Smoothed particle hydrodynamics (SPH) has attracted significant attention in recent decades, and exhibits special advantages in modeling complex flows with multiphysics processes and complex phenomena. Its accuracy depends heavily on the distribution of particles, and will generally be lower if the particles are distributed non-uniformly. A high-order SPH scheme is proposed in the present work for simulating both compressible and incompressible flows. It uses a Gaussian quadrature rule to perform the particle approximation of SPH by introducing Gaussian nodes. Unfortunately, the Gaussian nodes hardly overlap with SPH particles due to the Lagrangian feature, and thus we use a high-order interpolation method to obtain the corresponding physical quantities at the Gaussian nodes. The accuracy and robustness of the proposed Gaussian SPH are demonstrated by several numerical tests, including the Sod problem, Poiseuille flow, Couette flow, cavity flow, Taylor–Green vortex and dam break flow, and a convergence analysis is also conducted to evaluate the effects of particle resolution and distribution for reconstructing a given function. The simulation results for each test case are in good agreements with the available analytical, experimental or numerical results, showing that the proposed Gaussian SPH method is accurate and reliable but expensive for simulating compressible and incompressible flow problems.

Several strategies for solving a nonlinear eigenvalue problem are evaluated. This problem stems from the boundary integral equation solution of propagation in photonic crystal fibers. The origin and specificities of the eigenvalue problem are recalled before considering the solution of this eigenvalue problem. The first strategy, which is the starting point to illustrate the difficulties, is to solve the problem using Muller’s method. We then look at more recent techniques based on contour integrals or a rational interpolant that can be used to compute several eigenmodes simultaneously and considerably reduce the volume of computations.

Deep learning techniques, particularly neural networks, have revolutionized computational physics, offering powerful tools for solving complex partial differential equations (PDEs). However, ensuring stability and efficiency remains a challenge, especially in scenarios involving nonlinear and time-dependent equations. This paper introduces novel residual-based architectures, namely the Simple Highway Network and the Squared Residual Network, designed to enhance stability and accuracy in physics-informed neural networks (PINNs). These architectures augment traditional neural networks by incorporating residual connections, which facilitate smoother weight updates and improve backpropagation efficiency. Through extensive numerical experiments across various examples—including linear and nonlinear, time-dependent and independent PDEs—we demonstrate the efficacy of the proposed architectures. The Squared Residual Network, in particular, exhibits robust performance, achieving enhanced stability and accuracy compared to conventional neural networks. These findings underscore the potential of residual-based architectures in advancing deep learning for PDEs and computational physics applications.

Pump-jet propulsor excitation transfers to submarine hull along rotor-shaft and duct-stator paths simultaneously. The investigations on the effects of excitation transfer paths on structural vibration and acoustic radiation of submarine are limited. The present work aims to investigate vibro-acoustic characteristics of coupled shaft-submarine hull system utilizing a theoretical wavenumber analysis method and conduct acoustic design. The energy functional of the coupled structure-fluid system of the research object is first developed, and the displacement components of the jointed shell and the acoustic pressure are expanded by the Fourier series along circumferential direction. This allows for obtaining vibro-acoustic responses in the circumferential wavenumber-frequency domain, by which the predominant wavenumbers contributing to acoustic radiation are identified. The discussions reveal that the modes n = 0 and n = 1 respectively dominate the acoustic radiation under axial and vertical rotor loads. The acoustic radiations under duct-stator load are mainly contributed by mode n = 0, and the higher order modes n = 1 and n = 2 determine several acoustic peaks. Furthermore, two acoustic design schemes are proposed to suppress the wavenumbers with high radiation efficiency. It is proven that the design of the symmetric inner foundation and the application of new material are two efficient ways to improve acoustic performance of the submarine.

A novel hybrid boundary element is developed for polygonal holes in finite anisotropic elastic plates based on two different special fundamental solutions for holes. Since these special fundamental solutions satisfy traction-free condition along the hole's boundary, there is no mesh required on the boundary of polygonal holes. Various types of polygonal holes with rounded corners, such as triangles, rhombuses, ovals, pentagons, are considered by adding proper perturbation to an elliptical hole. The developed hybrid element is a mixture of two special boundary elements: one is based on the special fundamental solution derived through nonconformal mapping and the other is based on the solution derived through perturbation technique with conformal mapping. The special boundary element methods are combined through submodeling technique. First, the global model is solved using the perturbation solution. Then, using the displacements obtained from global model, an auxiliary submodel is set up and the results are evaluated with the nonconformal solution. The present method is compared and validated with conventional boundary element method and finite element method. The effect of hole curvature, material anisotropy, and loading condition on the stress distribution around the hole is presented.