Narrow Results

In this paper the two-dimensional Euler equations, with a simple chemical reaction model, are used as the governing equations for the detonation problem. The spatial derivatives are evaluated using the fifth-order WENO scheme, and the third-order TVD Runge-Kutta method is employed for the temporal derivative. The characteristics of the two-dimensional detonation in an argon-diluted mixture of hydrogen and oxygen are investigated using Adaptive Mesh Refinement (AMR) method. From computational accuracy point of view, AMR enables the detonation front to be clearer than the method with basic meshes. From the other point of computational time, AMR also saves about half the time as compared with the case of refining the entire field. It is obvious that AMR not only increases the resolution of local field, but also improves the efficiency of numerical simulation.

High-resolution extensions to six Riemann solvers and three flux vector splitting schemes are developed within the framework of a reconstruction-evolution approach. Third-order spatial accuracy is achieved using two different piecewise parabolic reconstructions and a weighted essentially nonoscillatory scheme. A three-stage TVD Runge–Kutta time stepping is employed for temporal integration. The modular development of solvers provides an ease in selecting a reconstruction scheme and/or a Riemann solver/flux vector splitting scheme. The performances of these high-resolution solvers are compared for several one- and two-dimensional test cases. Based on a comprehensive assessment of the solutions obtained with all solvers, it is found that the use of the weighted essentially nonoscillatory reconstruction with the van Leer flux vector splitting scheme provides solutions for a variety of problems with acceptable accuracy.

In this paper, the specific expression for pressure and sound speed in chemical reaction zone of condensed explosives are theoretically deduced, and a new method for deriving the partial derivative of pressure in respect of every conserved quantity is proposed. Combined with the third-order TVD Runge–Kutta method, we develop a parallel solver using the fifth-order high-resolution weighted essentially non-oscillatory (WENO) finite difference scheme to simulate detonation diffraction for two-dimensional condensed explosives. The numerical simulation results revealed the forming reasons of the low-pressure region, the low-density region, the "vortex" region and the "dead zone" in the vicinity of the corner. Furthermore, it demonstrated that the retonation will generate along the inner wall, and it plays an important role in the process of detonation diffraction.

A new numerical multiphase model is presented to study wave propagation over arbitrary shaped submerged obstacles. The high order space accurate weighted essentially non-oscillatory (WENO) method is adopted along with relatively coarse Cartesian uniform grids. Viscosity effects are included and the free surface is tracked using the level set method. The model is validated via application to solitary and progressive wave motion over rectangular, trapezoidal and semicircular obstacles. The complex flow field features induced near a large sized obstacle, including separation vortices and large free surface deformations, are accurately reproduced. Compared to other relatively complicated models, the present model is efficient and produces enhanced results.

In this paper we address the problem of producing an enlarged picture from a given digital image. We propose a zooming technique based on ENO and WENO schemes that are high order accurate finite difference schemes designed for problems with piecewise smooth solutions containing discontinuities. ENO and WENO schemes have been quite successful in applications, especially for problems containing both shocks and complicated smooth solution structures, such as compressible turbulence simulations and aeroacoustics. The algorithm works both on monochromatic images and RGB color pictures. Our experiments show that the proposed method is better than classical simple zooming techniques (e.g. pixel replication, bilinear interpolation). Moreover our algorithm is competitive both for quality and efficiency with bicubic interpolation.