A study of implicit difference schemes for inviscid equations

The analysis presented in this paper relates to the numerical difficulties encountered in obtaining a solution of the Full Viscous Shock Layer (FVSL) equations for flow past blunt bodies. A simple model problem consisting of two first order inviscid equations are shown to exhibit all the numerical difficulties encountered by the FVSL Set. In an effort to alleviate these difficulties, several numerical schemes for simultaneously solving the two first order equations have been investigated. It has been found that a staggered coupled implicit difference scheme for solving these two equations eliminates the numerical problems.     

Accelerating technique for implicit finite difference schemes for two dimensional parabolic partial differential equations

In this paper we define a new accurate fast implicit method for the finite difference solution of the two dimensional parabolic partial differential equations with first level condition, which may be obtained by any other method. The stability region is discussed. The suggested method is considered as an accelerating technique for the implicit finite difference scheme, which is used to find the first level condition. The obtained results are compared with some famous finite difference schemes and it is in satisfactory agreement with the exact solution.     

Compact alternating direction implicit method to solve two-dimensional nonlinear delay hyperbolic differential equations

In this article, a high-order compact alternating direction implicit method combined with a Richardson extrapolation technique is developed to solve a class of two-dimensional nonlinear delay hyperbolic differential equations. The solvability, stability and convergence of the method are analysed simultaneously in L²- and H¹-norms by the discrete energy method. Numerical experiments are provided to demonstrate the accuracy and efficiency of the schemes.     

An implicit,compact, finite difference method to solve hyperbolic equations

At present most approximate (discrete) solutions of time dependent hyperbolic equations are obtained by explicit finite difference methods, where the maximal allowable time step is given by a condition of numerical stability (i.e., the CFL condition). This report contains the development and the analysis of an implicit method of high order accuracy which is unconditionally stable, thus allowing to progress much faster in time. Furthermore, the presence of additional artificial boundary conditions does have an influence on the accuracy of the solution but not on the stability of the method. The numerical scheme has been checked considering two examples: the solution of the (linear) wave equation and the non linear Euler equations of fluid dynamics.     

An alternating direction implicit fractional trapezoidal rule type difference scheme for the two-dimensional fractional evolution equation

An alternating direction implicit (ADI) fractional trapezoidal rule (FTR) type difference scheme is formulated and analysed for a two-dimensional fractional evolution equation. In this method, standard central difference approximation used for the spatial discretization and the time stepping – an ADI scheme based on FTR, combined with chosen second-order fractional quadrature rule suggested by Lubich, are considered. The L², H¹-stability and convergence are derived. Numerical experiments in total agreement with our analysis are reported.     

On an alternating direction implicit finite element method for flow problems

An alternating direction implicit finite element method (ADIFEM) has been developed for flow problems in which the convective terms dominate. Application of the method to the thermal entry problem produces an unconditionally stable algorithm that is computationally more efficient than a corresponding ADI finite difference method. Application of the method to viscous compressible flow produces an algorithm that is only conditionally stable. The off-axis contributions to the convective terms are shown to limit the stability. Illustrative results are presented for the flow past a rectangular obstacle and over a step.     

On the convergence of compact difference schemes

Difference schemes that are compact in space, i.e., schemes constructed on a two- or three-point stencil in each spatial direction, are more efficient and convenient for boundary condition formulation than other high-order accurate schemes. Originally, these schemes were developed primarily to obtain smooth solutions. In the last two decades, compact schemes have been actively used to compute gas dynamic flows with shock waves. However, when a numerical solution with guaranteed accuracy is desired, the actual properties of difference schemes have to be known in the calculation of solutions with discontinuities. For some widely used compact schemes, this issue has not yet been well studied. The properties of compact schemes constructed by the method of lines are examined in this paper. An initial-boundary value problem for the linear heat equation with discontinuous initial data is used as a test problem. In the method of lines, the spatial derivative in the heat equation is approximated on a two-point stencil according to a fourth-order accurate compact differentiation formula. The resulting evolution system of ordinary differential equations is solved using various implicit one-step two- and three-stage schemes of the second and third order of accuracy. The relation between the properties of the stability function of a scheme and the spatial monotonicity of the numerical solution is analyzed. In computations over long time intervals, the compact schemes are shown to be superior to traditional schemes based on the second-order accurate three-point approximation of the spatial derivative.     

Nonlinear difference schemes for linear partial differential equations

We discuss a nonlinear difference scheme for approximating the solution of the initial value problem for linear partial differential equations. At each time step of the calculation the method proceeds by processing the data and determining the best possible scheme to use for that step, according to an optimization criterion to be described. We show that the method is stable and convergent applicating it on the heat equation. In all cases considered the nonlinear method was more accurate than the classical methods.     

Conservative upwind difference schemes for the Euler equations

In a recent paper [1] a number of numerical schemes for the shallow water equations based on a conservative linearization are analyzed. In particular, it is established that the schemes are related through the use of a source term. In this paper this technique is applied to the Euler equations, and further analysis suggests a new formulation of an existing scheme having the same key properties.     

交替方向隐式CFD解法器的GPU并行计算及其优化

交替方向隐格式(ADI)是常见的偏微分方程离散格式之一,目前对ADI格式在计算流体力学（CFD）实际应用中的GPU并行工作开展较少。从一个有限体积CFD应用出发,通过分析ADI解法器的特点和计算流程,基于统一计算架构(CUDA)编程模型设计了基于网格点与网格线的两类细粒度GPU并行算法,讨论了若干性能优化方法。在天河-1A系统上,采用128×128×128网格规模的单区结构网格算例,无粘项、粘性项及ADI迭代计算的GPU并行性能相对于单CPU核,分别取得了100.1、40.1和10.3倍的加速比,整体ADI CFD解法器的GPU并行加速比为17.3     

Conservative upwind difference schemes for the shallow water equations

A study of a number of current numerical schemes for the shallow water equations leads to the establishment of relationships between these schemes. Further analysis then suggests new formulations of the schemes, as well as an alternative scheme having the same key properties.     

A parallel algorithm for solving the implicit diffusion difference equations

This paper presents a parallel algorithm for solving the implicit diffusion difference equations. The basic idea is based on vectorization of the tridiagonal Toeplitz difference equations. This method is superior to the algorithm showed by H. Stone [8]. We computed some examples on an NEC SX-3/44R supercomputer by our method. The results showed a good parallelism with this algorithm.     

A new implicit compact difference scheme for the fourth-order fractional diffusion-wave system

In this paper, a new implicit compact difference scheme is constructed for the fourth-order fractional diffusion-wave system by the method of order reduction. The temporal Caputo fractional derivative is discretized by an L₁ scheme. The spatial derivative of order 4 is reduced to one of order 2 by order reduction. Then, the reduced derivative of order 2 is discretized by a difference formula of order 4. Using order reduction, two simple and accurate formulae of discretization for the derivative boundary conditions are obtained. And a new way of proving the stability and convergence of the scheme is presented in this paper. Some numerical results demonstrate the accuracy and efficiency of our new scheme.     

Numerical study of fourth-order linearized compact schemes for generalized NLS equations

The fourth-order compact approximation for the spatial second-derivative and several linearized approaches, including the time-lagging method of Zhang et al. (1995), the local-extrapolation technique of Chang et al. (1999) and the recent scheme of Dahlby et al. (2009), are considered in constructing fourth-order linearized compact difference (FLCD) schemes for generalized NLS equations. By applying a new time-lagging linearized approach, we propose a symmetric fourth-order linearized compact difference (SFLCD) scheme, which is shown to be more robust in long-time simulations of plane wave, breather, periodic traveling-wave and solitary wave solutions. Numerical experiments suggest that the SFLCD scheme is a little more accurate than some other FLCD schemes and the split-step compact difference scheme of Dehghan and Taleei (2010). Compared with the time-splitting pseudospectral method of Bao et al. (2003), our SFLCD method is more suitable for oscillating solutions or the problems with a rapidly varying potential.     

A new way for constructing high accuracy shock-capturing generalized compact difference schemes

The generalized compact (GC) schemes and some of their important properties are presented. And a new way for constructing high order accuracy and high-resolution GC schemes is presented. The schemes constructed by using this way could satisfy some principles and demands prescribed in advance to ensure some desired properties to the schemes, such as the principle about suppression of the oscillations, the principle of stability, the order of accuracy and number of scheme points, etc. As two examples, a three-point third-order compact scheme and a three-point fifth-order GC scheme satisfying the principle about suppression of the oscillations and the principle of stability are described in this paper. Numerical results show that these schemes are shock-capturing. The time-dependent boundary conditions proposed by Thompson are well employed when the algorithm is applied to the Euler equations of gas dynamics. Fourier analysis shows that the resolution characteristics are spectral-like.     

Exponential difference schemes for solving boundary-value problems for convection-diffusion type equations

The paper considers the numerical solution of boundary-value problems for multidimensional convection-diffusion type equations (CDEs). Such equations are useful for various physical processes in solids, liquids and gases. A new approach to the spatial approximation for such equations is proposed. This approach is based on an integral transformation of second-order one-dimensional differential operators. A linear version of CDE was chosen for simplicity of the analysis. In this setting, exponential difference schemes were constructed, algorithms for their implementation were developed, a brief analysis of the stability and convergence was made. This approach was numerically tested for a two-dimensional problem of motion of metallic particles in water flow subject to a constant magnetic field.     

Fourth-order runge-kutta schemes for fluid mechanics applications

Multiple high-order time-integration schemes are used to solve stiff test problems related to the Navier-Stokes (NS) equations. The primary objective is to determine whether high-order schemes can displace currently used second-order schemes on stiff NS and Reynolds averaged NS (RANS) problems, for a meaningful portion of the work-precision spectrum. Implicit-Explicit (IMEX) schemes are used on separable problems that naturally partition into stiff and nonstiff components. Non-separable problems are solved with fully implicit schemes, oftentimes the implicit portion of an IMEX scheme. The convection-diffusion-reaction (CDR) equations allow a term by term stiff/nonstiff partition that is often well suited for IMEX methods. Major variables in CDR converge at near design-order rates with all formulations, including the fourth-order IMEX additive Runge-Kutta (ARK₂) schemes that are susceptible to order reduction. The semi-implicit backward differentiation formulae and IMEX ARK₂ schemes are of comparable efficiency. Laminar and turbulent aerodynamic applications require fully implicit schemes, as they are not profitably partitioned. All schemes achieve design-order convergence rates on the laminar problem. The fourth-order explicit singly diagonally implicit Runge-Kutta (ESDIRK4) scheme is more efficient than the popular second-order backward differentiation formulae (BDF2) method. The BDF2 and fourth-order modified extended backward differentiation formulae (MEBDF4) schemes are of comparable efficiency on the turbulent problem. High precision requirements slightly favor the MEBDF4 scheme (greater than three significant digits). Significant order reduction plagues the ESDIRK4 scheme in the turbulent case. The magnitude of the order reduction varies with Reynolds number. Poor performance of the high-order methods can partially be attributed to poor solver performance. Huge time steps allowed by high-order formulations challenge the capabilities of algebraic solver technology.     

A comparison of conservative upwind difference schemes for the euler equations

Numerical results are presented and compared for three conservative upwind difference schemes for the Euler equations when applied to two standard test problems. This includes consideration of the effect of treating part of the flux balance as a source, and a comparison of different averaging of the flow variables. Two of the schemes are also shown to be equivalent in their implementation, while being different in construction and having different approximate Jacobians.