Você está na página 1de 9


Degree in Aerospace Engineering

Marc Rovira - Cayetano Marco - Ignacio Viciano - Daniel Vidal - Alvaro Ortiz


1. Introduction

In this laboratory session for the subject Aerodynamics we studied the ow around sev-

eral car models. This will be done by taking the measurements for drag force in order to calculate the drag coefcient inside a low-speed wind tunnel (M < 0.3).

2. Experimental Results !

2.1 !

Part 1: Drag in car models in under axial wind

In this section we took simple measurements for the drag coefcient of the Porsche

Cayenne and 911 models and compared them. The results are the following:
Porsche Cayenne Velocity (m/s) 12 14 16 18 20 22 23 24 Resistance (N) 0,060 0,080 0,140 0,180 0,240 0,260 0,280 Cx 0,337669677 0,344704462 0,476628393 0,496374426 0,546969064 0,54214388 0,536206942 SCx 0,000511232 0,000521883 0,000721615 0,000751511 0,000828111 0,000820806 0,000811817

Porsche 911 Velocity (m/s) 12 14 16 18 Resistance (N) 0,065 0,100 0,140 Cx 0,434380039 0,511649565 0,565972852 SCx 0,000553835 0,000652353 0,000721615

Lab Session 1: Flow around car models


20 22 23 24

0,180 0,230 0,250 0,280

0,589420299 0,622436495 0,619008925 0,636719458

0,000751511 0,000793607 0,000789236 0,000811817


!"&$ !"%$ !"#$ (#$ (&$ ()$ *!$ **$ *+$ *#$ $%&'()*+#,-./0#

!"&$ !"%$ !"#$ (%$ ('$ ()$ *!$ **$ *#$ *%$ $%&'()*+#,-./0#

If we did a preliminary study we could suppose that the drag of Porch Cayenne would be

higher than the drag of 911. This is because the back of Cayenne is a vertical wall, that is, has not a rounded body. In this zone there exist an abrupt change in geometry where the ow cannot remain attached to the surface and so turbulent ow separation takes place. Meanwhile, in the 911 the ow stays better attached to the surface. ! We have two tables in which we collect these measurements of drag, these measurements

have been obtained at different velocities and its corresponding drag coefcients and frontal area drag can be seen. After analyzing these graphs, we can surprisingly see how the drag coefcient of the 911 model is always higher than the one coming from the Cayenne. This incoherence can only be explained by how inaccurate our results are and as such cannot be taken seriously.


2.2 !

Part 2: Drag in car models in under lateral wind

For this section we took the measurements of the Porsche Cayenne model. For this

experiment we took measurements with the model at 10, 20 and 30 relative to the incident wind in order to see how CD varied with lateral wind. The results are the following:

Lab Session 1: Flow around car models


Porsche Cayenne Velocity (m/s) Resistance (N) "=10 12 18 22 24 0,08 0,20 0,32 0,38 "=20 12 18 22 24 0,10 0,26 0,42 0,49 "=30 12 18 22 24 0,12 0,32 0,53 0,64 0,9192119 1,089436326 1,207890017 1,225615866 0,001391687 0,001649407 0,001828745 0,001855582 0,766009917 0,885167015 0,957195863 0,938362148 0,001159739 0,001340143 0,001449195 0,00142068 0,612807933 0,680897704 0,729292086 0,727709421 0,000927791 0,001030879 0,001104148 0,001101752 Cx SCx

Lab Session 1: Flow around car models


In the graph we can see that the drag coefcient is higher when we increase the yaw angle

(with respect to the principal direction of the winds velocity). This happens because of two reasons. Firstly because the frontal area is higher when the yaw angle is higher. Furthermore when we have a yaw angle different from 0 the geometry of the car does not allow the air to follow soft streamlines so the ow is separated most of the times and the form drag is really high, thus generating a lot of turbulence.

2.3 ! !

Part 3: Drag in car models without ground effect

In this section we took the measurements of the Porsche 911 without a plate under-

neath it which generated ground effect. The results were the following:
Porsche 911 Velocity (m/s) 12 14 16 18 20 22 23 24 Resistance (N) 0,07 0,08 0,13 0,16 0,19 0,20 0,23 Cx 0,4677939 0,4093197 0,5255462 0,5239292 0,5141867 0,4952071 0,5230196 SCx 5,96E-04 5,22E-04 6,70E-04 6,68E-04 6,56E-04 6,31E-04 6,67E-04

Lab Session 1: Flow around car models


As we can observe in the table data and in the graph, the drag coefcient CD starts with a

value of 0.4678 with a velocity of 14 m/s and increases to reach a constant value of around 0.52 for the velocities from 18 m/s to 24 m/s. The CD obtained for the velocity of 16 m/s is not valid, as we can see that it does not t the ascendant curve. This may be due to a precision error measuring that data. ! If we compare it with the one with ground effect, we can see that at low velocities (to 18 m/

s approximately) the CD are similar but, at higher velocities, in this case the CD remains constant and, with ground effect, it continues increasing. That is because ground effect has a greater impact on drag at higher velocities so, as the velocity increases below the car, the downforce increases but also the drag.


2.4 !

Part 4: Drag and lift of an aircraft model

This rst graph shows the drag coefcient of the plane in function of velocity and in

the next one we can see the lift coefcient being plotted as a function of velocity:

("&$$$$% ("#$$$$% (")$$$$% ("!$$$$% !"'$$$$% !"&$$$$% !"#$$$$% *% '% !$% !!% !(% !)% !+% !#% !,% $%&'()*+#,-./0#


$"!!!!!# !"!!!!!# %# $!# $&# $%&'()*+#,-./0# $'# $("(#

Lab Session 1: Flow around car models


As we can observe, both the drag and the lift coefcient of the plane decrease almost line-

arly in function of velocity. However, we should take into account that due to the characteristics of our model it has been quite difcult to obtain proper measurements in order to plot the graphics (the measures of the lift were 0.02 in several measurements), but we can draw some conclusions: Both the drag and lift coefcient decrease as we increase velocity. This is due to the fact that although drag increases as the velocity increases, when it comes to calculating the coefcient, the drag is in the numerator and the velocity is in the denominator and to the square, so in the end we have lower values of CD as we increase velocity despite the fact that the drag decreases. A curious case is observed in the lift coefcient graphic, since it is reaching values really close to 0, what apparently does not make much sense. However, it can be explained as we look to out plane model. Its design is not the optimal one for a plane. The surface is way too low for the model to obtain decent values of lift, and since these values of lift are already small, once we use them in order to calculate the lift coefcient we obtain really low values of the lift coefcient, which is telling us that this plane is not t for ying. ! Below we can see the two graphs showing the evolution of CD with velocity for both the

plane model and the 911 car.

("&$$$$% ("#$$$$% (")$$$$% ("!$$$$% !"'$$$$% !"&$$$$% !"#$$$$% *% '% !$% !!% !(% !)% !+% !#% !,% $%&'()*+#,-./0#

Comparing both graphs, there is an obvious statement: The drag coefcient of the car in-

creases with velocity while in the plane it is exactly the opposite thing, it decreases the higher the velocity is. This apparent contradiction is explained due to the fact that the car and plane have two different shapes according to the needs of each transport: a car needs to have a not very aerodynamical shape in order to have space enough for both the occupants themselves and their luggage or whatever they need to carry in the trunk. This makes us have a high value of drag coming from the vehicle shape that we do not have in the plane, since when we design a plane we do it in order to have the most streamlined shape as possible, which is possible to be done since we do not have the same high shape requirements as in a car.

Lab Session 1: Flow around car models



! 1. !

2.5 !

Part 5: Aftermath questions

Do you think the results you obtained in the wind tunnel are directly transpos-

able to the real cars? Explain why. ! The results obtained from the wind tunnel should be handled with care. On the one

hand it undoubtedly gives us some data to work with in order to use it comparatively and understand the diverse elements and several effects that inuence drag, yet on the other hand the lack of precision and absence of adequate similarities should be more than enough to not consider this measures transposable to real cars. 2. ! Explain what similarity needs to be veried before we can conclude one way or

another (think in terms of non-dimensional numbers). ! When we want to perform experiments in a wind tunnel of a model before produc-

ing the nal product we need to take into account several factors if we want our model to reproduce exactly the same conditions as our product would have. For this purpose we use nondimensional analysis, so that we are able to generate non-dimensional numbers to help us with our experiment, obtain scaling laws for predicting results and predict tendencies in the relations between the different parameters. Therefore our model needs to be geometrically, kinematically and dynamically similar to our plane, meaning that the shape, the velocity and the force of the uid in any point of the model must be proportional in value and direction to the one of the real plane. As an example of non-dimensional numbers, taking Reynolds number we observe that for having a model X times smaller in size that the real plane we would need to have a velocity in the tunnel also by X times, which is not always possible to achieve and therefore we cannot take the results obtained in the wind tunnel to a real vehicle. 3. ! we used? ! There were several reasons that could affect the measurements precision. For examWhat do you think could affect the measurements precision in the wind tunnel

ple, the way we put the car in the wind tunnel was not very accurate so the car could move a little due to the force of the air, producing an error in the data. Because of that way to put the car with the bar, that bar produced an effect in the ow that affected the measurements. Related to the wind tunnel, it could have little zones at the bottom where we introduced the car which was not totally closed so part of the ow could go out by there. Furthermore, there is the error of the measurement devices for the lift, the drag and the velocity. For lift and drag, the measurement devices were very simple so it is probable that there was a little error. To measure
Lab Session 1: Flow around car models

the velocity, there was a difference between the level of the liquid attached to the tube and its level at the center so, although we always tried to measure it in the same way, it could be an error there. Finally, there is a human error which is always important when we talk about precision. 4. ! Can you think of a way to estimate the drag coefcient of your car without using

a wind tunnel, but just plain everyday instruments like a speedometer, a clock or stopwatch? ! In order to calculate the drag coefcient of our car we may follow some steps. First

we accelerate until some velocity (100 kph for example) and then we stop accelerating, so the velocity of the car will decrease. During this decreasing of the speed we have to write the value of the velocity along the time (for example every 5 seconds). Until certain period we have enough information. We may repeat this experience some times in order to nd the average result of the velocities. With the data obtained we can plot velocity vs time, and the slope of the curve (it should be more or less linear) will be a negative acceleration. ! If we assume that this negative acceleration is due only to the drag, we can calculate

it using Newtons second law and knowing the mass of our car. Then using the drag formula seen in theory lectures. Where D is the force we have already calculated, rho is the density of the air (we can estimate it for our conditions), V is the initial velocity from which we have start the experiment and A is the frontal area of our car. 5. ! Seek out information about how drag measurements are performed in the auto-

motive industry. ! In the industry drag, the rst real attempts on measuring drag were done with

models as in our case. Model scale testing is ideal for rapid evaluation of the inuence of different body styles and features on the vehicle's aerodynamics. By its very nature model scale testing uses smaller facilities with lower running costs than their full scale counterparts. Sophisticated test methods such as employing a moving ground plane can be done without great expense. Yet industry has evolved into using both computer modeling (CFD) and large scale wind tunnel testing. For the most accurate results from a wind tunnel test, the tunnel is sometimes equipped with a movable oor which moves at the same speed as the air ow. This prevents a boundary layer from forming on the oor of the working section and affecting the results. ! !
Lab Session 1: Flow around car models