On uniformly accurate high‐order Boussinesq difference equations for water waves

A new accurate finite‐difference (AFD) numerical method is developed specifically for solving high‐order Boussinesq (HOB) equations. The method solves the water‐wave flow with much higher accuracy compared to the standard finite‐difference (SFD) method for the same computer resources. It is first developed for linear water waves and then for the nonlinear problem. It is presented for a horizontal bottom, but can be used for variable depth as well. The method can be developed for other equations as long as they use Padé approximation, for example extensions of the parabolic equation for acoustic wave problems. Finally, the results of the new method and the SFD method are compared with the accurate solution for nonlinear progressive waves over a horizontal bottom that is found using the stream function theory. The agreement of the AFD to the accurate solution is found to be excellent compared to the SFD solution. Copyright © 2005 John Wiley & Sons, Ltd.     

A generalised finite difference scheme based on compact integrated radial basis function for flow in heterogeneous soils

In the present paper, we develop a generalised finite difference approach based on compact integrated radial basis function (CIRBF) stencils for solving highly nonlinear Richards equation governing fluid movement in heterogeneous soils. The proposed CIRBF scheme enjoys a high level of accuracy and a fast convergence rate with grid refinement owing to the combination of the integrated RBF approximation and compact approximation where the spatial derivatives are discretised in terms of the information of neighbouring nodes in a stencil. The CIRBF method is first verified through the solution of ordinary differential equations, 2–D Poisson equations and a Taylor‐Green vortex. Numerical comparisons show that the CIRBF method outperforms some other methods in the literature. The CIRBF method in conjunction with a rational function transformation method and an adaptive time‐stepping scheme is then applied to simulate 1–D and 2–D soil infiltrations effectively. The proposed solutions are more accurate and converge faster than those of the finite different method used with a second‐order central difference scheme. Additionally, the present scheme also takes less time to achieve target accuracy in comparison with the 1D‐IRBF and higher order compact schemes.     

Approximate kinetic equations in rarefied gas theory

The fundamental kinetic equation of gas theory, the Boltzniann equation, is a complex integrodiffcrential equation. The difficulties associated with its solution are the result not only of the large number of independent variables, seven in the general case, but also of the very complicated structure of the collision integral. However, for the mechanics of rarefied gases the primary interest lies not in the distribution function itself, which satisfies the Boltzmann equation, but rather in its first few moments, i.e., the averaged characteristics. This circumstance suggests the possibility of obtaining the averaged quantities by a simpler way than the direct method of direct solution of the Boltzmann equation with subsequent calculation of the integrals.It is well known that if a distribution function satisfies the Boltzmann equation, then its moments satisfy an infinite system of moment equations. Consequently, if we wish to obtain with satisfactory accuracy some number of first moments, then we must require that these moments satisfy the exact system of moment equations. However, this does not mean that to determine the moments of interest to us we must solve this system, particularly since the system of moment equations is not closed. The closure of the system by specifying the form of the distribution function (method of moments) can be considered only as a rough approximate method of solving problems. First, in this case it is not possible to satisfy all the equations and we must limit ourselves to certain of the equations; second, generally speaking, we do not know which equation the selected distribution function satisfies, and, consequently, we do not know to what degree it has the properties of the distribution function which satisfies the Boltzmann equation.A more reliable technique for solving the problems of rarefied gasdynamics is that based on the approximation of the Boltzmann equation, more precisely, the approximation of the collision integral. The idea of replacing the collision integral by a simpler expression is not new [1–4]. The kinetic equations obtained as a result of this replacement are usually termed model equations, since their derivation is usually based on physical arguments and not on the direct use of the properties of the Boltzmann collision integral. In this connection we do not know to what degree the solutions of the Boltzmann equation and the model equations are close, particularly since the latter do not yield the possibility of refining the solution. Exceptions are the kinetic model for the linearized Boltzmann equation [5] and the sequence of model equations of [6], constructed by a method which is to some degree analogous with that of [5].In the present paper we suggest for the simplification of the solution of rarefied gas mechanics problems a technique for constructing a sequence of approximate kinetic equations which is based on an approximation of the collision integral. For each approximate equation (i.e., equation with an approximate collision operator) the first few moment equations coincide with the exact moment equations. It is assumed that the accuracy of the approximate equation increases with increase of the number of exact moment equations. Concretely, the approximation for the collision integral consists of a suitable approximation of the reverse collision integral and the collision frequency. The reverse collision integral is represented in the form of the product of the collision frequency and a function which characterizes the molecular velocity distribution resulting from the collisions, where the latter is selected in the form of a locally Maxwellian function multiplied by a polynomial in terms of the components of the molecular proper velocities. The collision frequency is approximated by a suitable expression which depends on the problem conditions. For the majority of problems it may obviously be taken equal to the collision frequency calculated from the locally Maxwellian distribution function; if necessary the error resulting from the inexact calculation of the collision frequency may be reduced by iterations.To illustrate the method, we solve the simplest problem of rarefied gas theory-the problem on the relaxation of an initially homogeneous and isotropic distribution in an unbounded space to an equilibrium distribution.The author wishes to thank A. A. Nikol'skii for discussions of the study and V. A. Rykov for the numerical results presented for the exact solution.     

Chaotic Oscillation of Saddle Form Cable-Suspended Roofs under Vertical Excitation Action

The purpose of this paper is to investigate the dynamic behaviour of saddle form cable-suspended roofs under vertical excitation action. The governing equations of this problem are system of nonlinear partial differential and integral equations. We first establish a spectral equation, and then consider a model with one coefficient, i.e., a perturbed Duffing equation. The analytical solution is derived for the Duffing equation. Successive approximation solutions can be obtained in likely way for each time to only one new unknown function of time. Numerical results are given for our analytical solution. By using the Melnikov method, it is shown that the spectral system has chaotic solutions and subharmonic solutions under determined parametric conditions.     

基于DDA的弹性力学全高阶多项式位移逼近方法及其实例验证

基于维尔斯特拉斯多项式函数的逼近定理,通过DDA高阶全多项式位移函数条件下的弹性力学推导,提出了一个逼近弹性力学连续位移函数真解的全多项式位移函数逼近方法。该方法采用完整的高阶多项式位移函数,以不同阶次条件下的多项式系数为未知数,以单纯形积分为解析积分方法,通过建立和求解平衡方程,逐步逼近弹性体真解。在对单纯形积分计算过程研究的基础上,给出了三维空间单纯形计算图解法,该图解法诠释了三维空间单纯形积分公式中各变量间的逻辑关系及计算过程的图形表达。基于上述方法,编写了相应计算程序,并以一个三维简支梁受均布荷载及一个四周固定的弹性薄板受集中力作用两算例为实例,验证了所提方法的可行性。实例计算结果表明,随着逼近函数阶次的提高,数值方法获得的多项式函数计算值均单调地逐步逼近解析解。在文中所用的6阶多项式函数逼近中,简支梁实例位移计算误差小于0.2%,弹性薄板实例位移误差小于0.91%,并且,两算例与解析解位移差值都在微m级。     

Mathematical Models and Methods of the Mechanics of Stochastic Composites

The basic approaches used in mathematical models and general methods for solution of the equations of the mechanics of stochastic composites are generalized. They can be reduced to the stochastic equations of the theory of elasticity of a structurally inhomogeneous medium, to the equations of the theory of effective elastic moduli, to the equations of the theory of elastic mixtures, or to more general equations of the fourth order. The solution of the stochastic equations of the elastic theory for an arbitrary domain involves substantial mathematical difficulties and may be implemented only rather approximately. The construction of the equations of the theory of effective moduli is associated with the problem on the effective moduli of a stochastically inhomogeneous medium, which can be solved by the perturbation method, by the method of moments, or by the method of conditional moments. The latter method is most appropriate. It permits one to determine the effective moduli in a two-point approximation and nonlinear deformation properties. In the structure of equations, the theory of elastic mixtures is more general than the theory of effective moduli; however, since the state equations have not been strictly substantiated and the constants have not been correctly determined, theoretically or experimentally, this theory cannot be used for systematic designing composite structures. A new model of the nonuniform deformation of composites is more promising. It is constructed by performing strict mathematical transformations and averaging the output stochastic equations, all the constants being determined. In the zero approximation, the equations of the theory of effective moduli follow from this model, and, in the first approximation, fourth-order equations, which are more general than those of the theory of mixtures, follow from it     

A new,high-resolution shock-capturing hybrid scheme of flux vector splitting-Harten's TVD

The paper presents a new high-resolution hybrid scheme combining implicit flux vector splitting with Harten's TVD, which is proved suitable for shock-capturing calculation in gasdynamics. Fluxsplitting procedures are applied to discretize the implicit part of the Euler equations whereas Harten's numerical fluxes are used to calculate the residual of steady-state solutions. It ensures good shock-capturing properties and produces sharp numerical discontinuities without oscillations. It excludes expansion shocks and leads only to physically relevant solutions. The block-line-Gauss-Seidel relaxation procedure (block-LGS) is used to solve the resulting difference equations. The time step and the CFL number are much larger than those in the linearized block-alternating-direction-implicit approximate factorization method (block-ADI). Numerical experiments suggest that the hybrid scheme not only has a fairly rapid convergence rate, but also can generate a highly resolved approximation to the steady-state solution. Hence scheme seems to lead to an effective nonoscillatory shock capturing method for steady transonic flow. Project Supported by National Natural Science Foundation of China     

Free convection from an arbitrarily inclined plate in a porous medium

A mathematical model for the flow and heat transfer in the free convection from an arbitrary inclined isothermal flat plate embedded in a porous medium is presented, in which the Darcy–Boussinesq approximation is adopted to account for bouyancy force. A novel inclination parameter ξ is proposed such that all cases of the horizontal, inclined and vertical plates can be described by a single set of transformed boundary layer equations. Moreover, the similarity equations for the limiting cases of the horizontal and vertical plates are recovered from the transformed equations by setting ξ=0 and ξ=1, respectively. Detailed results for the skin friction coefficient and Nusselt number as well as for the dimensionless velocity and temperature profiles are presented for a wide range of the parameter ξ. A comparison with similarity solution shows excellent agreement.     

THE INTERACTION BETWEEN MULTIPLE CURVED RIGID LINE AND ANTIPLANE CIRCULAR INCLUSION

The interaction between multiple curved rigid line and ciruclar inclusion in antiplane loading condition is considered in this paper. By utilizing the point force elementary solutions and taking density function of traction difference along curved rigid lines, a group of weakly singular integral equations with logarithmic kernels can be obtained. After the numerical solution of the integral equations, the discrete values of density functions of traction difference are obtainable. So the stress singularity coefficient at rigid line tips can be calculated, and two numerical examples are given.     

流体饱和两相多孔介质拟静态问题的混合有限元方法

针对基于混合物理论的两相多孔介质模型,采用Galerkin加权残值有限元法,导出求解所静态问题的基于us-uF-P变量的混合有限元方程,由于系统方程的系数矩阵非定,进而针对该方程组提出了一种失代求解方法,并由分片试验得出节点压力插值函数的阶须低于固体相节点的位移插值函数的阶的结论,算例结果表明,采用基于u2-uF-p变量的混合法计算所得的固体相和流体相速度以及固体相的有效应力与罚方法一致,而压力值的粗度高于罚方法。     

The second order approximation theory of three dimensional elastic plates and its boundary conditions without using kirchhoff-love assumptions

The first order approximation theory of three dimensional elastic plates and its boundary conditions presented in the previous paper^[1] establishes six differential equations for the solutions of six undetermined functions u_o, u_a, A_(o) and S_(2)a defined in the x, y plane. They can be divided into two groups, each constitutes three equations to calculate u_o, S_(2)a and u_a, A_(o) respectively. Their boundary conditions as well as these equations are derived from the stationary conditions of variations of a functional for this problem based on the generalized variational principle. The solutions given by this theory are close to those given by the classical theory of thin plates as the ratio of thickness h to width a is small. For large ratio, say h/a=0.3 a considerable difference arises between the two theories. It has not been made clear that in what range of the ratio such difference is reasonable to give more precise solutions. In order, to solve this problem, we must study the second order approximation theory. In this paper following the previous one, we shall establish the second order approximation theory by applying the, stationary condition of variations of a functional for this problem based on the generalized variational principle, to derive nine differential equations and the relate boundary conditions, which are used to calculate nine undetermined functions u_o u_a, A_(o), A₍₁₎, S_(2)a and S_(3)a. And the range of the validity of the first order approximation theory can be found out by comparing the second order theory with the first order theory and the classical theory. It should be pointed out here that the equations of, the second order theory can also be divided into two groups to be solved separately, and the procedure of solution is not too complicate to perform as well. Here, we will use the same notations adopted in the previous paper, and not repeat their definitions.     

On the harmonic balance with linearization for asymmetric single degree of freedom non-linear oscillators

This paper deals with analytical approximation of non-linear oscillations of conservative asymmetric single degree of freedom systems, using the method of harmonic balance with linearization. This technique which consists of linearizing the governing equations prior to harmonic balance permits us to avoid solving complicated non-linear algebraic equations. But it could be applied only to symmetric oscillations for which it proves to be very simple and effective. This restriction is due to the fact that the method requires an appropriate initial approximate solution as input. Such a solution could not be readily identified for nonsymmetric oscillations, contrary the symmetric case where the fundamental harmonic works well. For these nonsymmetric oscillations, we propose in this paper to consider an initial approximation which consists of a small bias plus the fundamental harmonic. By expanding the corresponding harmonic balance equations respectively to first and second order in the bias, we are able to easily determine the bias and thus the required initial approximate solution that yields consistent solution at higher order. We use three examples to illustrate the proposed approach and reveal its simplicity and its very good convergence.     

Some methods of training radial basis neural networks in solving the Navier‐Stokes equations

The purpose of this research is to analyze the application of neural networks and specific features of training radial basis functions for solving 2‐dimensional Navier‐Stokes equations. The authors developed an algorithm for solving hydrodynamic equations with representation of their solution by the method of weighted residuals upon the general neural network approximation throughout the entire computational domain. The article deals with testing of the developed algorithm through solving the 2‐dimensional Navier‐Stokes equations. Artificial neural networks are widely used for solving problems of mathematical physics; however, their use for modeling of hydrodynamic problems is very limited. At the same time, the problem of hydrodynamic modeling can be solved through neural network modeling, and our study demonstrates an example of its solution. The choice of neural networks based on radial basis functions is due to the ease of implementation and organization of the training process, the accuracy of the approximations, and smoothness of solutions. Radial basis neural networks in the solution of differential equations in partial derivatives allow obtaining a sufficiently accurate solution with a relatively small size of the neural network model. The authors propose to consider the neural network as an approximation of the unknown solution of the equation. The Gaussian distribution is used as the activation function.     

Analyzing the 2D fracture problems via the enriched boundary element-free method

In this paper, the enriched boundary element-free method for two-dimensional fracture problems is presented. An improved moving least-squares (IMLS) approximation, in which the orthogonal function system with a weight function is used as the basis function, is used to obtain the shape functions. The IMLS approximation has greater computational efficiency and precision than the existing moving least-squares (MLS) approximation, and does not lead to an ill-conditioned system of equations. Combining the boundary integral equation (BIE) method and the IMLS approximation, a boundary element-free method (BEFM), for two-dimensional fracture problems is obtained. For two-dimensional fracture problems, the enriched basis function is used at the tip of the crack, and then the enriched BEFM is presented. In comparison with other existing meshless boundary integral equation methods, the BEFM is a direct numerical method in which the basic unknown quantity is the real solution of the nodal variables, and the boundary conditions can be implemented easily, which leads to a greater computational precision. When the enriched BEFM is used, the singularity of the stresses at the tip of the crack can be shown better than that in the BEFM. For the purposes of demonstration, some selected numerical examples are solved using the enriched BEFM.     

Exponential dichotomies of nonlinear discrete systems and its application to numerical analysis and computation

In this pepar we consider the upwind difference scheme of a kind of boundary value problems for nonlinear, second order, ordinary differential equations. Singular perturbation method is applied to construct the asymptotic approximation of the solution to the upwind difference equation. Using the theory of exponential dichotomies we show that the solution of an order-reduced equation is a good approximation of the solution to the upwind difference equation except near boundaries. We construct correctors which yield asymptotic approximations by adding them to the solution of the order-reduced equation. Finally, some numerical examples are illustrated.     

THE APPLICATION OF A FINITE DIFFERENCE/FINITE VOLUME ALGORITHM TO SOLVE MAXWELL'S EQUATIONS IN THE TIME DOMAIN

Abstract

A finite volume/finite difference method based on Ni's multigrid formulation is introduced for the solution of Maxwell's equations. The scheme is presented for the cases of transverse magnetic scattering from two-dimensional circular and square cylinders, as well as from NACA 0012 airfoil. The codes are validated against the traditional Method of Moments, which is analogous to a panel method in CFD. The circular cylinder scattering is compared to the analytical series solution for better understanding how the roles of numerical dispersion and dissipation errors affect the solution. The reflecting boundary conditions are modeled by the idea of inducing fields inside the conductor and a method of modeling the singularities that arise at a sharp corner is presented. Absorbing boundary conditions are modeled by integrating along the characteristic compatibility equations in the direction of the outgoing wave.     

Modified equations of finite-size layered plates made of orthotropic material. Comparison of the results of numerical calculations with analytical solutions

This paper describes the modified bending equations of layered orthotropic plates in the first approximation. The approximation of the solution of the equation of the three-dimensional theory of elasticity by the Legendre polynomial segments is used to obtain differential equations of the elastic layer. For the approximation of equilibrium equations and boundary conditions of three-dimensional theory of elasticity, several approximations of each desired function (stresses and displacements) are used. The stresses at the internal points of the plate are determined from the defining equations for the orthotropic material, averaged with respect to the plate thickness. The construction of the bending equations of layered plates for each layer is carried out with the help of the elastic layer equations and the conjugation conditions on the boundaries between layers, which are conditions for the continuity of normal stresses and displacements. The numerical solution of the problem of bending of the rectangular layered plate obtained with the help of modified equations is compared with an analytical solution. It is determined that the maximum error in determining the stresses does not exceed 3 %.     

UNSTEADY／STEADY NUMERICAL SIMULATION OF THREE-DIMENSIONAL INCOMPRESSIBLE NAVIER-STOKES EQUATIONS ON ARTIFICIAL COMPRESSIBILITY

A mixed algorithm of central and upwind difference scheme for the solution of steady/unsteady incompressible Navier-Stokes equations is presented. The algorithm is based on the method of artificial compressibility and uses a third-order flux-difference splitting technique for the convective terms and the second-order central difference for the viscous terms. The numerical flux of semi-discrete equations is computed by using the Roe approximation. Time accuracy is obtained in the numerical solutions by subiterating the equations in pseudotime for each physical time step. The algebraic turbulence model of Baldwin-Lomax is ulsed in this work. As examples, the solutions of flow through two dimensional flat, airfoil, prolate spheroid and cerebral aneurysm are computed and the results are compared with experimental data. The results show that the coefficient of pressure and skin friction are agreement with experimental data, the largest discrepancy occur in the separation region where the lagebraic turbulence model of Baldwin-Lomax could not exactly predict the flow.     

A Novel Workflow to Model Permeability Impairment through Particle Movement and Deposition in Porous Media

This article presents a practical transfer function type solution to a complex problem in which variations in a number of parameters can be taken into account. A new mathematical model, which is based on mass balance transfer function of particles movement/retention in porous media, has been derived. It is used to predict permeability reduction as a function of time. The linear forms as well as the radial forms of the model are described. Although the differential equations derived are similar to the general form of diffusion–convection equations, the marked difference is the suitability of the model, for being applied for variation of parameters, such as particle concentration in the fluid, injection rate, density of solid particles, against the depth and time of invasion. This transfer function has been solved, and the results of the simulation run agree reasonably well with the experimental damage data obtained in laboratory. Owing to its simplicity, this model is more practical to describe permeability reduction for the flow of suspended particles in porous media.     

An Enthalpy Control Volume Method for Transient Mass and Heat Transport with Solidification

A single domain enthalpy control volume method is developed for solving the coupled fluid flow and heat transfer with solidification problem arising from the continuous casting process. The governing equations consist of the continuity equation, the Navier–Stokes equations and the convection–diffusion equation. The formulation of the method is cast into the framework of the Petrov–Galerkin finite element method with a step test function across the control volume and locally constant approximation to the fluxes of heat and fluid. The use of the step test function and the constant flux approximation leads to the derivation of the exponential interpolating functions for the velocity and temperature fields within each control volume. The exponential fitting makes it possible to capture the sharp boundary layers around the solidification front. The method is then applied to investigate the effect of various casting parameters on the solidification profile and flow pattern of fluids in the casting process.