Catálogo de publicaciones - libros

A finite volume method to solve Maxwell’s equations on partially unstructured grids for the media having jump discontinuities in the electric permittivity coefficient has been presented. Test numerical simulations prove the second order accuracy of the method proposed.

The apparent resistivity for vertical profiling above local heterogeneity is calculated. The problem is solved by integral equations method. The vertical profiling inverse problem is formulated and some numerical calculations methods are suggested.

Simulation on Supercomputers has become a standard way of getting insight both in research and in industrial applications like product design and engineering. The process of computing, however, is only a small part of a chain of steps that lead us from reality to simulation results. In recent years the level of performance has dramatically increased and has forced us to reconsider the foundations of supercomputing. In this paper we set out to discuss the state of the art in supercomputing, the theoretical foundations of simulation and the implications that modern supercomputing architectures have on these foundations.

Two hydro turbine fluid flow problems are solved numerically in the paper. Both problems are solved in frames of Euler model using CFD code developed by the authors. The code is based on finite volume artificial compressibility approach. The accuracy of numerical scheme is second order in time and third order in space. For the solution of unsteady problems dual-time-stepping algorithm is used. Complex geometry of a turbine passage is handled using domain decomposition. The first problem considered is a simulation of precessing vortex rope downstream the turbine runner. Numerical results showed that this phenomenon can be captured by Euler model. The second problem is an automatic design (optimization) of runner blade shape with help of Breeder genetic algorithm. Different formulations of the objective function are considered and their influence on the blade shape is demonstrated.

The effectiveness of the parallelization of a 3D fluid flow simulation code using Software Engineering principles is considered. The problem on three-dimensional stationary ideal fluid flows through channels with complicated geometry has been investigated. The realization of the numerical algorithm has led to the sequential program, which has to be parallelized in order to reduce the execution time. The analysis of the sequential code is done in order to choose the parallelization approach, specify the kind of parallelization method, make the decision about the parallelization approach for the solver and define the target computer architectures. The results of parallel computations are provided for several hardware platforms.

Documentation is one of the key issues in Component-Based System Development that emphasizes the creation of models of testing information. Hence, testing based on these models might point out missing entities — such as methods — or inconsistencies in the treatment of entities for testing. Documentation influences component specification as well as specific information for testing. For example, Hybertson in the chapter ‘Using a Specification Approach to Facilitate Component Testing’ declares that more precise and complete specifications will enable and simplify component testing, and Memon in the chapter ‘A Process and Role-Based Taxonomy of Techniques to Make Testable COTS Components’ presents some examples in the use of meta-information applied to regression testing. There are many other examples of techniques that enrich documentation by structuring and classifying information. Along those lines, aspect information has been used to help implement better component interfaces and to encode knowledge of a component’s capability.

In this chapter, we introduce an aspect-based categorization of information for testing. This information is added as a component’s metadata to generate aspect-dependent test cases. We will illustrate the proposed technique and its supporting tool, which allows us to add testing metadata to Enterprise JavaBeans (EJB) components.

For a large class of non linear evolution problems we derive an abstract formulation that is based on writing the original model as a system of first order partial differential equations. Starting from this reformulation, a set of interface classes are derived that allow a problem independent implementation of various temporal and spacial discretization schemes. In particular, the abstract framework is very well suited for discretizing evolution equations with the Local Discontinuous Galerkin ansatz []. The implementation of the proposed framework is done within the Distributed and Unified Numerics Environment DUNE [], [].

Large-eddy simulations (LES) of two complex flow problems, a continuous tundish flow and the flow around multi-element airfoils are presented. The numerical computations are performed by solving the conservation equations for compressible fluids. An implicit dual time stepping scheme combined with low Mach number preconditioning and a multigrid accelerating technique is developed for LES computations. The method is validated by comparing data of turbulent pipe flow at =1280 and cylinder flow at Re=3900 at different Mach numbers with experimental findings from the literature. Finally, the characteristics of the flow in a one-strand tundish is analyzed and a solution for a flow around a two-element airfoil as well as a zonal solution for the trailing edge region are discussed.

We give an overview of some contemporary processor developments and draw conclusions for developments of numerical algorithms. Multicore CPUs are developed by nearly all hardware manufactures. The bandwidth peak performance relation is decreasing. Both have consequences for the efficiency of todays computer codes. We propose a Finite Difference discretization technique which is decreasing the data intensity.

The prediction of unsteady vortex dominated flows show is a challenging task, applying classical RANS simulations often fails and leads to poor results. Here a VLES approach is shown. For this approach an adaptive turbulence model based on the k- model of Kim and Chen is developed. Applying this approach to vortex dominated flows - vortex rope in a straight diffuser and tip vortex at a ship propeller - leads to an increase on computational accuracy. Therefore this approach seems to be suitable for this types of flows.