Pre-processors: preBFC, GAMBIT, Tgrid

Main codes: Fluent4.5, Fluent5, FIDAP, Rampant, Nekton, MixSim

Post-processors: in-built in above codes, Flpost

Pre-processors: CFX-Build

Main codes: CFX4, CFX5, CFX-ProMixus

Post-processors: CFX-Visualize


Main code: PHOENICS

Post-processor: PHOTON

Pre- and post processor: PROSTAR

Main code: STAR-CD

Pre-processor: FAME

Main codes: FIRE, SWIFT

Post-processor: in-built in above codes

Pre-processor: CFD-GEOM Main code: CFD-ACE6, CFD-ACE (U) Post-processors: CFD-VIEW www.cfdrc.com





AVS/ AVS Express www.avs.com Ensight, EnsightGold www.ceintl.com

of the major tasks of any CFD pre-processor are:

• to enable the user to model the geometry of the problem under consideration;

• to generate a suitable computational grid for the modeled geometry;

• to compile all the necessary data and information about the grid in a form suitable to the CFD solver;

• to accept relevant input data from the user;

• to check the consistency of the input data (as far as possible);

• to store all this information in a form suitable for reading into CFD solvers for further processing.

Geometry modeling and grid generation are the major elements of CFD preprocessors. For simple geometry conforming to a standard co-ordinate system (for example, Cartesian or cylindrical-polar co-ordinates), geometry modeling and grid

TABLE 8.2 Key Issues for Evaluating CFD Pre-processors, Solvers and Post-processors


Geometry modeling approach: solid modeler/surface modeler; top-down/bottom-up Geometry import facilities: CAD packages, general formats like IGES

Geometry repair facilities: gaps/trimmed surfaces, removal of coincident entities Visualization: internal grids, multiple views Grid types: single/multi-block; structured/unstructured/mixed

Grid generation tools: automatic/parametric generation, recovery from error (UNDO facilities) Boundary layer capability

Mesh control: clustering, aspect ratios, skewness Tools for assessment of grid quality

Grid refinement: smooth-ing/orthagonality/clustering

Setting fluid properties/input data: databases/consistency checks Setting boundary conditions/defining cell types Exporting information to different solvers

Future developments: new technologies


Grid types: import from different pre-processors; co-located/ staggered; (un) structured Automatic grid refinement tools, addition of grid elements Geometry modifications (change scale/cell type etc.) without re-meshing

Memory: 1 million cell problem:~ 0.35-1 GB

Compressible/incompressible; primary variables/stream function

Transient simulations: automatic control on time steps/efficient storage

Turbulence models: user-defined model (UD)? Wall functions: constraints on near wall cells

Simulation of rotating flows: sliding mesh/multiple reference frames

User-defined scalar equations: constraints on form/algorithm

Multiphase flows: Eulerian-Eulerian (EE) capabilities: closure/drag laws/additional forces EE-granular flows: model options/Eulerian-Lagrangian (EL): true/psuedo? particle models/UD? VOF: surface forces/ adhesion/contact angle; UD?

Porous media models: isotropic/non-isotropic; pressure drop model; UD?

Rheological models: non-Newtonian fluids/UD? Algorithm?

Reactive-flows: Phenomenological models-EB, ESCIMO, multigroup E model PDF-based models: presumed/ full PDF algorithm? Surface reactions: options for rate controlling steps/UD? Multiphase reactive flows: mass transfer/reactions in all phases?

Boundary conditions: profile/transient/UD? Special/user-defined BCs for multiphase flows? Consistency checks for BCs/input data

Importing physical property and kinetics data from external databases

Discretization schemes: space/time; higher order/user defined?

Limiter functions to avoid non-physical results/UD? Special discretization procedures for multiphase flows

Facility to provide internal traps/limits

Algorithms: pressure-correction/density based; multiple pressure corrections?

Multiphase flows: partial/full elimination? Pressure correction?

Multiphase flows: calculation of volume fractions/internal traps

Segregated/coupled solver? Option?

Source-dominated flows: handling of user-defined sources/scalars

Convergence behavior: sensitivity to under-relaxation parameters

Algebraic equation solvers: conjugate gradient? Acceleration tools: multigrid/block correction Parallelization: technology? speed-up efficiency

On-line convergence monitoring tools

Data storage/Exporting data to different post-processors

Access to the source code/internal flow

Overall computational performance/bench mark cases

Future developments: algorithms/algebraic solvers


Ease of analysis during simulation: coupling with solver/local integral quantities Error analysis: residue reduction, distribution within domain

Basic presentation capabilities: vectors, contours, streak-lines, iso-surfaces Computation of fluxes, sub-domain balances Automatic feature detection: trailing vortices/re-attachment

Presentation of user-defined derived quantities: constraints/ flexibility

Visualizing results on arbitrary surfaces Overlay capabilities/lighting/

shading Importing tabular data for validation/comparisons

XY plots, Function calculators to compare global results

Post-processing of transient simulations/multiple datasets

Animation/video facilities/different formats Exporting results to other presentation tools (RGB, BMP, MPEG, PS, EPS) Future developments: better integration generation is fairly straightforward and can be executed by accepting relevant data from the user. However, most industrial reactors have complex configurations and therefore require advanced geometry modeling tools. Complex geometries may be developed either by using a bottom to top approach (defining points, lines, faces and so on to construct higher order objects) or by using a top to bottom approach (starting with solid volumes and carrying out Boolean operations on them to arrive at the desired geometry). With the advent of widespread applications of computer aided design (CAD) and solid mechanics analysis, several geometry modeling tools are now available. Most of these tools allow use of a top to bottom approach to define the desired geometry. Most pre-processors of the commercially available CFD codes allow one to import geometry from these design tools. In addition to importing geometry information from these design tools, most CFD pre-processors also have in-built geometry modeling tools. Some tools are also provided to repair 'dirty' geometry (gaps, trimmed surfaces and so on). These capabilities are essential and must be critically evaluated during the selection process.

Meshing or generating a suitable computational grid for the modeled geometry is one of the most important pre-processor tasks. Quite sophisticated algorithms and tools are required to divide the modeled geometry into computational cells based on either a structured or unstructured grid. A structured grid requires that all interior nodes have an equal number of adjacent elements (typically all elements are quadrilateral or hexahedral). This restriction is relaxed in an unstructured grid (triangular or tetrahedral elements may be used). The type of grid is subject to constraints imposed by the discretization method selected and the solution algorithm. Once the type of grid is selected (structured or unstructured), several methods are available to generate the desired grids. Details of these methods will not be discussed here. More information on grid generation may be found in Thompson (1996) and at an excellent website on grid generation maintained by Steven Owen: http://www.andrew.cmu.edu/user/sowen/mesh.html.

It must be mentioned here that geometry modeling and grid generation may account for a substantial percentage of the time required to carry out the total flow modeling task. For example, in aerospace engineering applications, the time spent on geometry modeling and grid generation may account for more than 50% of the total project time. Even for reactor engineering applications, where model development may require most of the time, the time spent on grid generation is not insignificant. It is, therefore, important to evaluate various facilities made available in any grid generation tool, to reduce the time spent on grid generation. Most commercial grid generation tools allow parametric grid generation to facilitate faster grid generation for similar geometries. Facilities to recover from errors, while building the geometry or while generating the grids, are also very useful (e.g. customizable UNDO features). A boundary layer capability to ensure adequate resolution near walls and corners is also useful. Appropriate tools to provide control of clustering, cell aspect ratio and cell skewness, are essential to generate good quality grids. More often than not, some refining operations are needed to make the generated grid better suited to flow simulations. Such refining operations may be classified into (1) smoothing (includes operations which adjust node locations while maintaining the element connectivity), and (2) clean-up (operations which change element connectivity). Capabilities for grid refinement and tools to assess the quality of the generated grid are very important and need to be examined critically.

Although boundary conditions and fluid properties may be set in pre-processors, most commercial codes allow these to be set in CFD solvers.

0 0

Post a comment