Sei sulla pagina 1di 11

Computational Fluid Dynamics (CFD) can be described as the use of computers to produce information about the ways in which

fluids flow in given situations. CFD embraces a variety of technologies including mathematics, computer science, Engineering and physics, and these disciplines have to be brought together to provide the means of modelling fluid flows. Such modelling is used in many fields of science and engineering but, if it is to be useful, the results that it yields must be a realistic simulation of a f luid in motion. At present this depends on the problem being simulated, the software being used and the skill of the user. Computational fluid dynamics (CFD) is concerned with numerical solution of differential equations governing transport of mass, momentum, and energy in moving fluids.

CFD activity emerged and gained prominence with availability of computers in the early 1960s. Today, CFD finds extensive usage in basic and applied research, in design of engineering equipment, and in calculation of environmental and geophysical phenomena. Since the early 1970s, commercial software packages (or computer codes) became available, making CFD an important component of engineering practise in industrial, defence, and environmental organizations. For a long time, design (as it relates to sizing, economic operation, and safety) of engineering equipment such as heat exchangers, furnaces, cooling towers, internal combustion engines, gas turbine engines, hydraulic pumps and turbines, aircraft bodies, sea-going vessels, and rockets depended on painstakingly generated empirical information. The same was the case with numerous industrial processes such as casting, welding, alloying, mixing, drying, air-conditioning, spraying, environmental discharging of pollutants, and so on. The empirical information is typically displayed in the form of correlations or tables and

nomograms among the main influencing variables. Such information is extensively availed by designers and consultants from design handbooks. The main difficulty with empirical information is that it is applicable only to the limited range of scales of fluid velocity, temperature, time, or length for which it is generated. Thus, to take advantage of economies of scale, for example, when engineers were called upon to design a higher capacity power plant, boiler furnaces, condensers, and turbines of ever higher dimensions had to be designed for which new empirical information had to be generated all over again. The generation of this new information was by no means an easy task. This was because the information applicable to bigger scales had to be, after all, generated via laboratoryscale models. This required establishment of scaling laws to ensure geometric, kinematic, and dynamic similarities between models and the full-scale equipment. This activity required considerable experience as well as ingenuity, for it is not an easy matter to simultaneously maintain the three aforementioned similarities. The activity had to, therefore, be supported by flowvisualization studies and by simple (typically, one-dimensional) analytical solutions to equations governing the phenomenon under consideration. Ultimately, experience permitted judicious compromises. Being very expensive to generate, such information is often of a proprietary kind. In more recent times, of course, scaling difficulties are encountered in the opposite direction. This is because electronic equipment is considerably miniaturised and, in materials processing, for example, the more relevant phenomena occur at microscales

(even molecular or atomic scales where the continuum assumption breaks down). Similarly, small-scale processes occur in biocells. Clearly, designers need a design tool that is scale neutral. The tool must be scientific and must also be economical to use. An individual designer can rarely, if at all, acquire or assimilate this scale neutrality. Fortunately, the fundamental laws of mass, momentum, and energy, in fact, do embody such scale-neutral information. The key is to solve the differential equations describing these laws and then to interpret the solutions for practical design. The potential of fundamental laws (in association with some further empirical laws) for generating widely applicable and scale-neutral information has been known almost ever since they were invented nearly 200 years ago. The realisation of this potential (meaning the ability to solve the relevant differential equations), however, has been made possible only with the availability of computers. The past five decades have witnessed almost exponential growth in the speed with which arithmetic operations can be performed on a computer. We know that the three laws governing transport are the following:
1. The law of conservation of mass (Transport of mass), 2. Newtons second law of motion (Transport of momentum), and 3. The first law of thermodynamics. (Transport of energy).

The basis of computational fluid dynamics is the reduction of continuum deferential equation describing the dynamics of the fluid (Navier stokes, Mass & energy conservation equation) into a system of algebraic equation

of finite number of grid points, and the solving of the equation at these limited number of points only. Computational Fluid Dynamics (CFD) provides a qualitative or quantitative prediction of fluid flows by means of 1. Mathematical modeling (partial differential equations) 2. Numerical methods (discretization and solution techniques) 3. Software tools (solvers, pre- and post processing utilities) Fluid flows and related phenomena can be described by partial differential (or integral differential) equations, which cannot be solved analytically except in few special cases. To obtain an approximate solution numerically, we have to use a discretization method which approximates the differential equations by a system of algebraic equations, which can then be solved on a computer. The approximations are applied to small domains in space and/or time so the numerical solution provides results at discrete locations in space and time. Much as the accuracy of experimental data depends on the quality of the tools used, the accuracy of numerical solutions is dependent on the quality of discretization used.CFD is finding its way into process, chemical, civil, and environmental engineering. Optimization in these areas can produce large savings in equipment and energy costs and in reduction of environmental pollution. 2 INTRODUCTION

Each software package aimed at the CFD market has to assist the user in carrying out the tasks that form the analysis process. This is done by providing, typically, three main pieces of software: a pre-processor a solver a post-processor

together with a variety of utility programs. The use of all these programs will be explained below. Pre-Processing Programs All the tasks that take place before the numerical solution process is started are called pre-processing. This includes the first three phases of the analysis process that we have discussed, thinking, mesh generation and flow specification and the part of the fourth phase that defines the numerical control parameters. Whilst the first phase needs considerable thought, and considerable engineering judgement, if the physical flow problem is to be translated into a problem that is solvable by the CFD software; it does not involve any computing. It is only when this first phase has been completed that the computing starts. To assist in the computational part of the pre-processing phase, most software packages have a pre-processing program that can be used to carry out the following operations: define a grid of points and perhaps volumes or elements. define the boundaries of the geometry apply the boundary conditions specify the initial conditions set the fluid properties set the numerical control parameters. In carrying out these tasks the user has to interact with the computer in some way and so the pre-processing program usually has a graphical interface, so that parameters can be set, and the resulting changes seen quickly. This is particularly important when the mesh is being built. Also, datafiles can be read that contain lists of commands so that repetitive sets of instructions, say for a similar, but not identical problem, do not have to be typed too often. Usually, the most difficult task in the pre-processing phase is the generation of the grid of points or mesh. Quite often this task can be simplified by using software that has been especially designed to carry out mesh generation. One example of this is the use of programs written to produce meshes suitable for the finite element analysis of structural problems. Such software is commonly available and can interface with computer-aided design systems. This allows the analyst to access computer models of objects, the surface data of which can form the basis for the geometry around

which the mesh for a CFD simulation can be built.

Solving The Equations Each package has a program that solves the numerical equations for the problem under consideration. This program must be given all the relevant data that has been defined by the pre-processor. To transfer the data between the programs, the pre-processor writes out datafiles that the solver program can read. These files can also bemoved, if necessary, between computers. This is extremely useful as it means that the solver program can run on a machine specifically designed for high-speed numerical work such as a supercomputer, while the interactive tasks are carried out on a smaller machine. This splitting of the tasks between machines enables the hardware to be used in the most efficient manner, keeping graphics-intensive and so-called number-crunching activities separate. Once the datafiles are in place, the solver program is activated and the required solution process carried out. At the end of this phase, further datafiles will be available, which may have to be transferred back to the machine where the pre- and post-processing programs are run. Although the solver program is the core of any CFD software

system, the user sees little of its operation. Post-Processing Programs As large numbers of points have to be created within the flow domain if reasonable simulations are to be obtained and as several variables are stored at these points, computer graphics is often the only means of assessing the data written by the solver program. The post-processing program is used to display the results, and, as with the pre-processor, this program is interactive and so usually run on the same machine as the pre-processor. Typical pictures obtained with the post-processor might contain a section of the mesh together with vector plots of the velocity field or contour plots of scalar variables such as pressure. These pictures enable global trends in the data to be seen.

GAMBIT as a state-of-the-art preprocessor for engineering analysis, it has several geometry and meshing tools in a powerful, flexible, tightlyintegrated, and easy-to use interface. GAMBIT can dramatically reduce preprocessing times for many applications. Most models can be built directly within GAMBIT's solid geometry modeler, or imported from any major CAD/CAE system. Using a virtual geometry overlay and advanced cleanup tools, imported geometries are quickly converted into suitable flow domains. A comprehensive set of highly automated and size function driven meshing tools ensures that the best mesh can be generated, whether structured, multiblock, unstructured, or hybrid. GAMBIT's range of CAD readers allow you to bring in any geometry, error free, into its meshing environment. GAMBIT also has an excellent boundary layer mesher for growing optimum grid cells off wall surfaces in your geometries for fluid flow simulation purposes. GAMBIT is Fluents geometry and mesh generation software. GAMBIT's single interface for geometry creation and meshing brings together most of Fluent's preprocessing technologies in one environment. Advanced tools for journaling let you edit and conveniently replay model building sessions for parametric studies. GAMBIT's combination of CAD

interoperability, geometry cleanup, decomposition and meshing tools results in one of the easiest, fastest, and most straightforward preprocessing paths from CAD to quality CFD meshes. Moreover its functions can be summarized as :

1. Geometry creation 2. Mesh generation 3. Boundary conditions specifications

FLUENT is a computational fluid dynamics (CFD) software package to simulate fluid flow problems. It uses the finite-volume method to solve the governing equations for a fluid. It provides the capability to use different physical models such as incompressible or compressible, inviscid or viscous, laminar or turbulent, etc. Geometry and grid generation is done using GAMBIT which is the preprocessor bundled with FLUENT.

ANSYS FLUENT software contains the broad physical modeling capabilities needed to model flow, turbulence, heat transfer, and reactions for industrial applications ranging from air flow over an aircraft wing to combustion in a furnace, from bubble columns to oil platforms, from blood flow to semiconductor manufacturing, and from clean room design to wastewater treatment plants. Special models that give the software the ability to model in-cylinder combustion, aeroacoustics, turbomachinery, and multiphase systems have served to broaden its reach.

Flow chart of CFD procedure:

Limitations of CFD Physical models. CFD solutions rely upon physical models of real world processes (e.g. turbulence, compressibility, chemistry, multiphase flow, etc.). The CFD solutions can only be as accurate as the physical models on which they are based. Numerical errors.

Solving equations on a computer invariably introduces numerical errors. Round-off error: due to finite word size available on the computer. Round-off errors will always exist (though they can be small in most cases). Truncation error: due to approximations in the numerical models. Truncation errors will go to zero as the grid is refined. Mesh refinement is one way to deal with truncation error. Boundary conditions. As with physical models, the accuracy of the CFD solution is only as good as the initial/boundary conditions provided to the numerical model. Example: flow in a duct with sudden expansion. If flow is supplied to domain by a pipe, you should use a fully-developed profile for velocity rather than assume uniform conditions