CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 1

CTU CARTECH CFD REPORT

Team CTU Cartech has taken part in this year Formula Student competitions with third car-prototype

called FS.03. Our intension was to reach the top ten at least in one competition. In order to achieve

our goals we aimed to improve our overall team performance trough disciplines.

We are sure that CFD simulations are important part of the design process, because they can

improve the car performance at the track and team presentation during the design event. CFD can be

used in Formula Student for the following areas

• Aerodynamics

• Engine performance

• Heat transfer.

With the help of STAR-CCM+, we have been able to perform CFD simulations for the first time in our

teams history. We will now review the work done with STAR-CCM+ software during the last year.

AIR INTAKE SYTEM

The design of Formula Student air intake system has to fulfill specific Formula Student/SAE rules,

which have significant effect on the engine performance, because they influence negatively the

amount of air aspired by cylinders compared to unrestricted engine. Therefore it is important to

simulate airflow in air intake system in order to optimize engine performance.

CTU Cartech cars are powered by YAMAHA R6 four cylinder engine with displacement volume of

599 ���.

RESTRICTOR FLOW

Starting with STAR-CCM+ we performed simulations to explore its possibilities, settings and

preferences. The purpose of the simulation was to set the mass flow/ pressure drop dependence in

order to calibrate the 1D engine simulation model in GT-Power.

In STAR-CCM+ we focused mainly on mesh, solver and different turbulence settings. In CFD

simulations it is very desirable to achieve mesh independent solutions and that was exactly our goal

in that particular problem. In mesh independency test we created meshes containing from 50 000 up

to 3 million polyhedral cells. Figure 1 shows plot of mass flow throughout the restrictor/number of

cells in chocking conditions. The solution is mesh-independent when the domain contains about 1.5

million cells. The most important and useful fact is that the results are very accurate even when the

domain contains about 400 thousand cells. This number of cells was used later in air intake system

domain for unsteady flow simulations.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 2

Figure 1

Figure 2 shows flow through the restrictor in chocking conditions (Velocity Mach 1 in the throat) with

shock wave in the diffuser. The coupled flow solver can accurately predict chocking conditions and

shock waves, which affect maximal engine performance.

Figure 2

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 3

Table 1 summarizes the setup of restrictor flow and also air intake system simulation.

Table 1 – Simulation setup

Mesh

Type Polyhedral

Number of cells 50 000 – 3 milion

Number of prism layers 3 – 6

Solver Type Coupled Implicit, Flow, Energy

Physics

Turbulence model Realizable k-e, SST k-w, RSM

Wall functions All �� wall treatment

Max wall �� ~ 50

AIR INTAKE SIMULATION

Putting of throttle body, restrictor, airbox, injection and runner pipes together results in a complete

assembly of air intake system (see figure 3).

Figure 3

Because the nature of engine flow is inherently unsteady, we decided to perform unsteady

computation in order to understand the flow behavior in air intake system. Firstly boundary

conditions and domain geometry had to be set. Some simplifications in geometry have been made.

The throttle plate was removed in order to reduce mesh size because fully opened throttle body´s

wake should have minimal influence on engine performance. Cylinder runner pipes have been

modeled up to engine head, therefore no valve opening and closing needed to be simulated. In

addition time dependent boundary conditions have been gained from 1D engine model in GT-Power,

because the flow at the inlet to the induction ports can be assumed as one-dimensional. Static

pressure, temperature and turbulence parameters have been applied to the domain outlets and

stagnation pressure and temperature on the inlet, simulating suction from atmosphere. The

simulation has been carried at 12000 crankshaft rpm.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 4

The domain has been divided into approximately 380 000 polyhedral cells and one engine cycle

period into 540 intervals. Second order discretization schemes have been used for both flow and

time. 6 cycles needed to be run in order to achieve converged solution.

The outputs of the simulation are mass flows at domain outlets throughout one engine cycle. The

mean mass flows at outlets during one cycle and suction phase have been computed. The former

serves as verification the simulation is right, because the sum of mean mass flows per cycle at outlets

has to be the same as the mean mass flow through inlet or restrictor throat. The latter serves as

performance parameter of particular cylinders.

Output of the unsteady simulation of FS.03 air intake system and different airbox shape as a possible

concept for next season are summarized in table 2.

Table 2 – Unsteady simulation

Mean mass flow during suction phase � ∙ ���

Cylinder 1 Cylinder 2 Cylinder 3 Cylinder 4 Sum

FS.03 50 53,3 53,3 50,3 206,9

Different airbox 52,7 51,6 52 51,5 207,8

Table 2 shows that there is potential to improve the uniformity of mass flow distribution for

particular cylinders (firs line). Simulations done for the next race season with different airbox shape

show, that it is possible to achieve more uniform distribution of mass flows. But the total sum of

mass flow doesn´t increases. We have also worked out that there is some connection between

steady and unsteady simulation. In steady state case we opened just one runner pipe, applied

backpressure to its outlet and measured mass flow. Inner cylinders (cylinder 2 and 3) have greater

mass flow both in unsteady and steady state for the FS.03 while for different airbox shape mass flows

are more equal for both states (see table 3).

Table 3 - Steady state

Mass flow [g/s]

Inner cylinder Outher cylinder Difference [%]

FS.03 59,5 57 4,2

Different airbox 57,5 58 0,8

In addition for the coming year we would like to do 1D/3D engine simulation and explore the airbox

flow parameters (response) in transient engine simulations (car acceleration).

EXTERNAL AERODYNAMICS

In Formula Student the aerodynamic forces don´t play so high role like in Formula 1, because cars

speed at the track is significantly lower. Nevertheless aerodynamics plays important part of car

performance, because while the car is moving the drag is still present and the engine has to be

cooled. In our aerodynamics calculations we primarily focused on resolving the flow through heat

exchanger in an entire aerodynamic model of the car.

Our cooling assembly consists of a heat exchanger, an upstream and downstream duct with a fan.

The addition of upstream duct to present car prevents the air leakage throughout the system while

the downstream duct improves the flow uniformity through heat exchanger honeycomb (see fig. 5).

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 5

Figure 5

The computational domain for the aerodynamic calculation consist of a large box (40 × 6 × 6 �),

where half of the car is placed (see appendix). Velocity is prescribed at Velocity Inlet boundary while

zero gauge pressure is defined at Pressure Outlet boundary. Vehicle ride is simulated by means of

moving ground and rotating wheels. The fan on the car is switched on when the water temperature

crosses some specified value (e.g. 85℃). Simulations have been carried out with the fan switched off

by higher velocities and switched on by lover velocities and when the car is standing. The fan

pressure rise was modeled the simplest way possible as an internal interface and it´s characteristics

was defined by polynomials. The heat exchanger flow was modeled by porous media just setting the

inertial and viscous resistance coefficients gained from pressure drop curve measured by our heat

exchanger manufacturer.

For aerodynamic simulations we have use both polyhedral and trimmed meshes. The final results are

obtained on meshes containing 5.5 million polyhedral cells (see table 4).

Table 4-Aerodynamic simulation setup summary

Mesh

Type Polyhedral, Trimmer

Number of cells (Max) 5,5 millions (Poly) 6 (Trim)

Number of prism layers 3

Solver Type Segragated

Physics

Turbulence model SST k-w

Wall functions All �� wall treatment

Max wall �� ~ 45

The main outputs of the simulations are aerodynamic forces acting on the car and mass flow through

heat exchanger. Low value of downforce coefficients �� in table 5 confirms that our car doesn´t

produce significant downforce. This is due to lack of diffuser and wings. The usage of diffuser can

increase the amount of downforce and also reduce the drag of the car because it changes the wake

structure behind it. The fitting of the diffuser is the biggest problem because the rear suspension is

quite low, there is also exhaust system and jacking rod in possible diffuser location.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 6

Table 5 -Aerodynamic forces

Car velocity

[kmh]

Drag

[N]

Downforce

[N]

�� ��

30 18.6 0.6 0.54 0.017

50 50 4.6 0.48 0.044

80 120 8 0.46 0.033

Mass flows through heat exchanger in different regimes are summarized in table 6.

Table 6 - Cooing system mass flows

For the next season we would like to improve simulations by adding heat transfer, more realistic fan

model, more cells in the domain, e.g.

Compared to previous years we didn´t have any problems concerning the cooling of the car. We have

been able to finish every endurance race even in very hot temperatures like in Italy and Spain. We

are sure we can improve the cooling system further.

Conclusion

Using STAR-CCM+ we have been able to design parts of cooling and air intake system. STAR-CCM+

allow us to understand and analyze the car´s parameters linked to fluid flow systems. It is helping us

to apply the results during the design stage of the car components and for presentation within

competitions program. For the next year we intend to use STAR-CCM+ even more for the design of

the car´s components in order to improve the performance of the car.

This year we have attended the races in Silversone (United Kingdom), Varano (Italy) and Barcelona

(Spain). In Silverstone we finished 7 th, in Varano 6

th and in Barcelona 2 th. The first application of

STAR-CCM+ in actual season is one of the factors which generated the positive results related to the

passed races in 2011.

Car velocity

[kmh]

Cooler mass flow

[kg/s]

Fan switched on

0 0,21 Yes

30 0,23 Yes

30 0,101 No

50 0,2 No

80 0,36 No

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 7

APPENDIX

Figure 1 - Streamlines in FS.03 airbox during suction of cylinder 1.

Figure 2 – Polyhedral mesh for unsteady simulation (approximately 380 000 cells).

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 8

Figure 3 – Absolute pressures gained from GT-Power at air intake outlets.

Figure 4 – STAR-CCM+ mass flows through one engine cycle.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 9

Figure 5 – Streamlines in different airbox concept during suction of cylinder 4. Restrictor is put into

the airbox which results in more equal mass flow distribution.

Figure 6 – Computational domain and boundary conditions for aerodynamic simulations.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 10

Figure 7 – Pressure distribution on the car and streamlines in 80 kmh.

Figure 8 – Streamlines while the car is standing and fan switched on. Fan is modeled as internal

interface, therefore no swirl is included.

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 11

Figure 9 – Streamlines containing mass flow through cooling system. It is possible to analyze flow

interaction with a car.

Figure 10 – Pressure distribution on plane cut through computational domain. Pressure rise can be

observed before upstream duct of cooling system after that pressure drop through cooler (porous

medium). Fan is switched off (otherwise there would be pressure rise).

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 12

Figure 7 – Polyhedral mesh (approximately 5.5 million cells)

Figure 8 – Polyhedral mesh (approximately 6 million cells)

CTU Cartech, Czech Technical University in Prague, Formula Student/SAE Team 13

Figure 9 – Velocity distribution on cooler honeycomb. Car is standing, fan is switched on. More

uniform velocity distribution compared to fan mounted directly on cooler honeycomb.