|What is CFD2k ?
What is CFD2k ?
CFD2k is a CFD (computational fluid dynamics) simulation software, which incorporates a principal study of an innovative lagrangian particle model and an advanced numerical approach to solve the particle interactions effectively. Principal study means here, that the models and numerics are very sophisticated from the physical and mathematical point of view. But the range of application is so far very limited. CFD2k is a fully compressible solver for ideal gases.
This approach eliminates some weak points of the established particle methods and shows high potential for more accurate CFD solution, especially in areas where conventional CFD tools and models like LES still show inaccuracies. Here one can mention mainly the issues related to aero-acoustics, laminar-turbulent transition behaviour and relaminarisation. The model, which is presented here, deals with discrete circular particles and calculates the detailed collision processes (micro scale). With thermodynamic, statistical and similarity laws global (macro scale) flow variables can be derived. The model is so far 2d and the particles can move in every direction in the 2d plane depending on the forces acting on it. Overlapping particles and multi-particle interactions are important features of this model. A freeware software is developed and the executables, the user guide and several exemplary cases can be downloaded here. The following animation visualises the model conception of the current approach using overlapping circular particles.
The animation shows the movement of the circular sub-grid particles. It can be seen as a zoom inside a ppr-cell up to the sub-grid particle level. The particles are moving in space and time due to the multiple particle collision laws which are implemented in the model. Since difficulties are arising in the exact describtion of the turbulent transport processes, a further extension of the model was developped. In the next upgrade of CFD2k instead of circular overlapping particles, so-called moving Voronoi-particles are used. With this approach it is possible to describe correct turbulent transport processes. The following animation visualises the model conception of the new Voronoi-particle approach..
Both particle approaches, which are implemented in CFD2k (circular- and Voronoi-particles approach) use functions which insert or merge particles depending on local conditions. This is necessary to ensure a better accuracy of the diffusion processes.
The main constraint of this released version of CFD2k is the limited range of geometric shapes, which can be simulated i.e. CFD2k accepts only simple hexahedral 2D grids so far. But there are also some limitations regarding the operating conditions, e.g. only ideal gas law is available with constant material properties like heat capacity (Cp - even independent from temperature variations). But apart from these strong limitations of the current version of CFD2k, the new model concept reveals high potential in important areas, which is reason enough to enhance the research activities on this approach.
CFD2k represent also a new philosophy of CFD applications. In conventional CFD simulations, the grid quality is essential for the quality of the flow results and there are always the uncertainties regarding grid independence of the final solution. Theoretically in CFD2k these problems are obsolete. The self-adaptive mesh (particles) on which CFD2k solves the equations is not visible to the user. The user has only to provide a so-called post-processing grid (ppr-grid) on which CFD2k interpolates the results. The particles, their movement and interactions, can be regarded as a kind of simulation on a self-adaptive sub-grid-level. This so-called self-adaptive sub-grid (or solver-grid) is always much finer than the ppr-grid and the user should be aware of the resolution, which is really needed for the post-processing. The general rule is to use as coarse as possible ppr-grids and only use the spatial resolution, which is really needed. The flow values of each ppr cell represent the density-weighted average over the whole ppr-cell volume. The ppr cell contains in general several sub-grid cells (i.e. particles). The flow results on the sub-grid level should be in theory grid-independent so also the result of the ppr-grid should be grid-independent. Of course by increasing the ppr-grid resolution, one resolves more details, which occurs inside a certain ppr-cell but the average value over the refined ppr-cells should stay constant as the original coarse ppr-cell value.
Model conceptions of CFD2k (zoom inside a ppr-cell directly at the backward-facing step):
Copyright © 2007 Dr. Blazenko Ivancic