CFD

CFD

A Trend in Automobile Aerodynamics Technology

Abstract :

Aerodynamics in automobile plays a crucial role, how will they behave on road & how efficient they performs the task they are designed for. For all vehicles ranging from small passenger vehicles to

commercial busses & trucks, reducing air drag is one of the most efficient ways of improving fuel economy.

Optimum aerodynamic design has to produce the best balance between drag that allows the car to

be stable & drivable at high speed. In this paper we study the optimum aerodynamics design & basic steps

to be followed during the design using Computational Fluid Dynamics.

Computational Fluid Dynamics (CFD) gives you a means of visualization and understanding

your design as it is a tool for predicting what will happen under the given set of circumstances before

physical prototyping.

CFD helps you to design better and faster, it provides numerical approximations to the equations

that govern the fluid motions. In CFD, governing equations are solved with the help of computer software.

What Is CFD?

Computational fluid dynamics (CFD) is the computational science of simulating

fluid flows by using computers. CFD is the simulation of fluids engineering systems

using modeling (mathematical physical problem formulation) and numerical methods

(discretization methods, solvers, numerical parameters, and grid generations, etc.) CFD

made possible by the advent of digital computer and advancing with improvements of

computer resources. CFD provides numerical approximation to the equations that govern

fluid motion.

Application of the CFD to analyze a fluid problem. First, the mathematical

equations describing the fluid flow are written. These are usually a set of partial

differential equations. These equations are then discretized to produce a numerical

analogue of the equations. The domain is then divided into small grids or elements.

Finally, the initial conditions and the boundary conditions of the specific problem are

used to solve these equations. The solution method can be direct or iterative. In addition,

certain control parameters are used to control the convergence, stability, and accuracy of

the method.

Why To Use CFD?

Basically, the compelling reasons to use CFD are these three:

1. Insight

There are many devices and systems that are very difficult to prototype. Often,CFD analysis shows you parts of the system or phenomena happening within the system that would not otherwise be visible through any other means. CFD gives you a means of visualizing and enhanced understanding of your designs.

2. Foresight

Because CFD is a tool for predicting what will happen under a given set of circumstances, it can answer many ‘what if?’ questions very quickly. You give it variables. It gives you outcomes. In a short time, you can predict how your design will perform, and test many variations until you arrive at an optimal result. All of this is done before physical prototyping and testing. The foresight you gain from CFD helps you to

design better and faster.

3. Efficiency

Better and faster design or analysis leads to shorter design cycles. Time and money are saved. Products get to market faster. Equipment improvements are built and installed with minimal downtime. CFD is a tool for compressing the design and development cycle.

A Case Study On Opel Astra:

Optimum Design Procedure:

CFD flow simulation can thus be used to readily make statements about flow circulation around a car to the point where models or prototypes are not available.Offering comprehensive information to designers about the entire flow field can therefore accelerate the process of aerodynamic design. The flow field around a vehicle is physically very complex.

The efficiency of an aerodynamic CFD simulation depends on the time span required to achieve the first set of results, as well as the accuracy of the simulated flow quantities. The turn around time and confidence level in the predictions is two major criteria for success that compete with one another. Creation of the model geometry,discretization of the physical domain, and choice of a suitable numerical computing scheme are significant factors that can determine the level of success of such an effort.

The underlying simulation process is divided into the following steps: CAD

surface preparation, mesh generation, CFD solution of the fluid flow, mesh adaption, and

visualization of the results. The software packages used for these steps include,

UNIGRAPHICS (CAD), ANSA (CAD/mesh generation), TGRID and GAMBIT

(additional mesh generation), and FLUENT (solver and post processing). An elapsed

processing time of between 5 and 11 days, without CAD modeling, is usually required for

the base case simulation of a complete model, comprised of about 3 million cells. This

time scale depends on the complexity of the model. Once the base case has been built,

individual modifications can be realized in a fraction of this time span 

CAD-Model:

To begin the process, the CAD body shell data for the ASTRA is downloaded

from a common CAE database at OPEL, from which all of the vehicle parts and

components can be accessed. For initial concept studies, the vehicle exterior data is made

available by the design department. Underbody data can also be downloaded from the

database. For early theme studies there is usually a simplified generic underbody model

available, which has already been integrated and meshed. For subsequent aerodynamic

studies, individual engine parts are not modeled. In this study, the air cooling vents and

the engine space above the sub frame are closed.

Using this finished base case, numerous types of vehicles can be generated on the

same template for simulation in a short time period. A half model with a centerline

symmetry plane is sufficient for initial design studies, but the ASTRA in this study is

constructed as a full model with an asymmetric geometry (where the asymmetries are

mostly confined to the underbody). This is done to enable detailed predictions of the flow

field around and under the vehicle. The wind tunnel geometry around the model is a

rectangular enclosure. It is of such a dimension that the adverse pressure effects between

the vehicle and the wall are minimized. The three CAD components, namely the outer

body of the vehicle, the underbody with tires, and the wind tunnel together constitutes the

CFD model

Surface Grid:

The integration, preparation and clean up of the CAD components described

above was done with the program ANSA .To do this, the CAD data was exported in

IGES format and read into ANSA, where production of a closed topology of CAD areas

and definition of macro areas was performed. Additionally, auxiliary surfaces were

generated, so that separately meshed fluid zones could later be linked or delinked from

one another. These areas and surfaces were then used to generate the surface grid. The

following description of the task of grid creation is limited to the styling surfaces, since

the wind tunnel and the underbody have already been meshed.

The vehicle body is meshed with triangles having a side length of about 20 mm,

in as uniform a manner A possible. A minimum side length of about 5 mm used in the

areas of the side mirror and the A-pillar. The window frames are covered with regular

quadrilateral elements so as to close the stage of the window later using separately

inserted prism blocks. Experience shows that generation of a fine and uniform grid on the

vehicle body is necessary to obtain realistic values for drag and lift coefficients. Fig

shows the shaded surface mesh of the ASTRA.

Generating The 3d Hybrid Mesh:

A hybrid mesh comprised of tetrahedral and prismatic elements was chosen for

the CFD model of the ASTRA. This was done so as to ensure that an extensive automatic

grid generation of the complex geometry could occur. The volumetric grid was built in

TGRID, using the surface grid that was generated by ANSA. It was then exported in a

FLUENT format. On the one hand, this means that a high quality resolution of the

surface mesh is necessary to create a good volume grid. On the other hand, any

modifications must only be performed in the two-dimensional “space” that is the surface

mesh.This therefore simplifies and accelerates the task of grid generation by taking into

account the possibility of further simulations of various base vehicle variants. To improve

the resolution of the oncoming flow boundary layer, the styling surfaces of the vehicle

and the floor of the wind tunnel were made up of multiple prism layers. In this instance,

avoiding three-dimensional corners or steps in the surface to form a grid with prisms was

effective. This help’s generate cells of good quality. The set-in side windows of the

ASTRA model along with their corresponding rectangular windowsills are not well

suited for direct coverage by prism layers, because distorted prismatic elements would be

created. The same is true for geometric steps in the areas of the cowl and A-pillar, and

also for inserts and steps in the front and rear of the vehicle. In ANSA, these areas are

closed in advance, using the auxiliary surfaces mentioned above. The limiting areas are

filled with either tetrahedral or, as in the case of the side windows, regular hexahedral

blocks generated in GAMBIT.

Wherever such auxiliary surfaces could not be used to separate areas of complex

geometry from the surrounding prism layers, the so-called no conformal interfaces option

was used. In the case of the ASTRA, the external mirrors, the entire underbody, and the

wheels are covered using tetrahedral. This means that the prism layers which were

generated on the styling surfaces and on the ground, end at the body sill and at predefined

areas around the external mirror and wheels. A so-called prism side is generated at the

edge of the prism layers. This is a circumscribing zone, which displays quadrilateral

surfaces. To couple the quadrilateral surfaces with the tetrahedral mesh of the wind

tunnel that is yet to be generated, these are copied and triangulated automatically in

TGRID, in correspondence to the distribution of the edge nodes. The triangles that are so

created can be manually modified if necessary. Both areas are later handled as internal

interfaces. This allows the solver to interpolate the flow data between the different types

of elements.

TGRID was used to generate prism layers on the vehicle body shell and ground

plane. The height of each prism element is controlled by entering the aspect ratio of the

first cell and a geometric growth rate. To do this, meshing parameters were chosen to

target a good initial y+ distribution and a smooth transition to the tetrahedral grid in the

external fluid region. The planar sidewalls of the wind tunnel were automatically split

into zones with quad or triangular surface elements, depending upon where the prism

layers join. There they were retriangulated. The regions that had not been made into a

grid using prisms were grouped together to form their own domain. The remaining

volumetric grid generation was completed using TGRID, resulting in a hybrid mesh of

approximately 3 million cells. The maximum skewness was less than 0.8 .

Numerical Solution Of Flow Equations And Grid Adaption :

A 3D steady state, incompressible solution of the Navier-Stokes equations was

performed using FLUENT 5. Turbulence modeling was done with the realizable k model

using non-equilibrium wall functions. The free stream velocity was set to be 140 km/h.

For about the first 300 iterations, a first order upwind discretization scheme was used to

accelerate the convergence.

Thereafter, a second order upwind scheme was used. For simulations of this type,

a total of 5 combined grid adaptions are usually carried out during the solution process, to

satisfy the y+ criteria. Statistical pressure gradient adaption is also executed, acting on

about 1 to 2% of the total cell count. Hanging node adaption is typically used, which

allows for subsequent coarsening of the grid if the need arises later in the solution process.

After each adaption, about 150 iterations are usually required to converge the solution to

the previous residual values. For scaled residuals, the convergence criterion is satisfied if

the residuals fall below a value of 10-3. A more accurate measure is when fluctuations in

the lift and drag coefficients lie within the range of ± 0.001. In this particular case, a total

of 1600 iterations on an eight processor SGI Origin 2000 were performed.

Post-Processing (Flow Visualization):

In a normal wind tunnel test, flow visualization is either too restrictive or too

expensive. In contrast, the CFD simulation process readily generates data for the entire

flow field. This can be combined and presented in the manner the user requires. Flow

field visualization (surface pressure distribution, for example) assisted by path lines can

provide the aerodynamic development engineer with significant insight into the basic

flow features. It also indicates areas for further optimization. In addition, general

statements may be made about the aero-acoustics and contamination of the vehicle. It can

be used to analyze turbulence energy distribution or the progression of wall surface

streamlines in the A-Pillar region and at the side glass. Figure shows examples of flow

field visualization.

CFD flow simulation is a useful tool for providing predictions of pressure

distribution and forces exerted on the vehicle components.

References:

1. www.flowfluent.com/solutions/automobile/aero.html

2. www.cd-adapco.com/apps/automotives.html

3. Tao Xing & Fred Stern,” Introduction to CFD” IIHR-hydro science & Engg., University of Iowa,2002.

4. Ing. Andreas Kleber “ JA123-Simulation Of Air Flow Around Opel Astra Vehicle With Fluent”,2001

5. GRIEBEL M.,DORNSEIFER T. AND NEUNHOEFFER T.”Numerical Simulation in Fluid Dynamics:

A Practical Introduction”. SIAM Monographs on Mathematical Modeling and Computation. SIAM 1998.