Wind Tunnels Tests On A Model Cessna: Year 1 Lab Report
Wind Tunnels Tests On A Model Cessna: Year 1 Lab Report
Name: Vignesh Palaniappan CID: 00637107 Personal Tutor: Dr. Rafael Palacios-Nieto Due Date: 02-May-2011
The purpose of this experiment was to introduce wind tunnel testing; in particular, we examine how the angle of attack and varying wind velocities affect the lift and drag forces acting on an aircraft. A model Cessna 172 was tested; it had a wing area of 0.038 sq. m in comparison to a real size area of approximately 16 sq. m. At both velocities tested, the lift increases as the angle of attack increases between -2 and 12 degrees and maximum lift is generated at 13 degrees. After this point, drag becomes prominent and causes the model to stall. Further evaluations are made to distinguish the relationship between the lift and drag coefficients and their relationship to the angle of attack. We study the basic forces acting on an aircraft, the concept of downwash and perform calculations to find the induced drag. An estimation of the stalling speed of a full scale Cessna is also made and finally as with all wind tunnels, we look at the errors that affect the quality of the results.
CONTENTS
INTRODUCTION ....................................................................................................................................... 3 APPARATUS & EXPERIMENTAL PROCEDURE .......................................................................................... 4 THEORY ................................................................................................................................................... 4 WIND TUNNELS ............................................................................................................................... 4 FORCES ON A PLANE ....................................................................................................................... 5 RESULTS & DISCUSSION .......................................................................................................................... 6 INDUCED DRAG ESTIMATION........................................................................................................ 10 STALLING SPEED ESTIMATION ...................................................................................................... 11 SOURCES OF ERROR ...................................................................................................................... 11 CONCLUSION......................................................................................................................................... 12 REFERENCES .......................................................................................................................................... 12 APPENDIX .............................................................................................................................................. 13 75%................................................................................................................................................ 13 50%................................................................................................................................................ 14
NOMENCLATURE = = = = = = = = = Lift Coefficient Drag Coefficient Induced Drag Coefficient Aspect Ratio Drag-Lift Ratio Gradient from specified figure Air Density Velocity of flow Area
INTRODUCTION
Wind tunnel testing is a vital stage in the design process of many industries, regardless of whether an object is moving or is stationary. It is an excellent way of scrutinizing the airflow round a test model. For aircraft, it is an opportunity to assess the airworthiness of the plane and identify any improvements that can be made to optimise performance. Despite CFD and new simulation techniques, wind tunnel testing has proved over the years to be an aeronautical engineer's most prized asset. We are able to look at how to increase the lift generated by wing sections or equally how to reduce the amount of drag acting on the plane. Road vehicle aerodynamics are of increasing importance as we look to reduce fuel consumption and we know that the more streamlined the vehicle the less power required to generate movement. Formula 1 teams rely heavily on using aerodynamic features of their cars to gain leverage against opponents be it only a few milliseconds and are willing to invest shedloads of money into wind tunnels and research. Another beneficiary of wind tunnel testing is high-rise buildings and other large structures such as bridges and stadiums. For example, a skyscraper like the Burj Khalifa presents a huge surface for the wind to act upon, if it were not designed properly the building would suffer from violent oscillations causing the structure to physically deteriorate and possibly collapse. However, by repeating a process of tunnel testing and modifying the model, a good solution is drawn to reduce the effects of these winds.
THEORY
WIND TUNNELS There are several types of tunnel but this particular tunnel is a closed-return type where the model is placed in the working section and the flow goes round in a complete circuit. The working section is the section around which the tunnel is built, the tests are run on the model here and this is where you want close to perfect airflow. The 'settling chamber 'is in place to straighten the air and to minimise the effect of turbulence. This usually has screens (wire meshes) and a honeycomb which remove eddies and cross flow components of the air. When testing, we want the air to come towards the model in only one direction and the meshes and honeycomb adequately do the job.
The 'contraction' before the working section makes the airflow uniform and faster than in the rest of the circuit. This saves a lot of money since there is no need for a constant high-energy flow over the whole tunnel. The design of this is crucial since it ultimately determines the flow in the working section. In this tunnel, two reference static tappings are present in the contraction and the dynamic pressure is calculated by finding the difference between these two points. This value is also calibrated against a Pitot-static tube in the test section The 'diffuser' is where the air coming out of the test section slows down prior to recirculating. The recovery of static pressure from the kinetic energy is important since it reduces the power required to drive the wind tunnel. The advantages of using this type of tunnel compared to an open circuit or open jet is that no dust is drawn in, the flow isnt sensitive to external disturbances and the pressure in working section is fine(less than the atmosphere due to speedy flow). However, this is significantly more expensive and space consuming. Another point of note is that open jet tunnels have no walls to constrain the flow in the working section this is a more 'real' case scenario.
FORCES ON A PLANE
Figure 3 The 4 main forces acting on a plane (Ref.2) Moving onto the model, there are two main forces we analyse: the lift and the drag. The lift is the force acting normal to the resultant of the free stream velocity. The drag is the force acting normal to the lift opposite to the direction of flight. Airflow around an aerofoil tends to be quicker over the top and slower over the bottom, this difference in velocity and consequently pressure is the reason why we get lift. Another thing is that the airflow over the top has less pressure than the airflow outside the wing therefore meaning the air flows inwards towards the fuselage. The inverse happens on the bottom where flow is out away from the fuselage. Combining these two flow concepts together, we are able to understand the formation of wing tip vortices. So in the jet figure above, the vortex on the left wing will rotate clockwise and the right wing vortex will rotate counterclockwise.
Diagram 1 - the centreline is the fuselage; the flows proceed as shown past the aerofoil. (Ref.4) Acknowledging the existence of wing tip vortices and their ability to create 'downwash' (a backward tilting motion of the plane), we know that a part of the lift force acts backwards, this is a component that contributes to the overall drag on the model and this component is called the induced drag. As we keep increasing the angle of attack, we should keep increasing the lift coefficient until a certain critical angle where a further increase would result in a loss of lift, this angle is called the stalling angle and it varies from aerofoil to aerofoil. The wing disrupts the flow since it produces a large enough surface for the air to 'hit' against, the air separates and drag becomes more pronounced.
LIFT (75%)
1.2 1 0.8 Lift Coefficient Lift Coefficient 0.6 0.4 0.2 0 -0.2 0 -0.4 -0.6 5 10 15 1.2 1 0.8 0.6 0.4 0.2 0 -0.2 0 -0.4
LIFT (50%)
-5
-5
10
15
Incidence (deg)
-0.6
Incidence (deg)
Stalling angle
Stalling angle
DRAG (75%)
0.3 0.25 Drag Coefficient 0.2 0.15 0.1 Drag Coefficient 0.3 0.25 0.2 0.15 0.1 0.05 0 0 5 10 15 -5 0
DRAG (50%)
0.05
0 -5
5 Incidence (deg)
10
15
Incidence (deg)
Figs 4-7 (from left to right) Lift and Drag coefficient curves vs. the angle of attack for 75% and 50% wind tunnel capacity. The dashed lines show the stalling angle 7
Cd vs. Cl (75%)
0.3 0.25 Drag Coefficient Drag Coefficient 0.2 0.15 0.1 0.05 0 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 -0.5 Lift Coefficient 0.3 0.25 0.2
Cd vs. Cl (50%)
0.15
0.1 0.05 0 0 0.5 Lift Coefficient 1 1.5
D/L vs Cl (75%)
1 0.8
0.6 D/L (Drag to Lift Ratio) 0.4 0.2 -0.4 -0.2 -0.2 0 -0.4 -0.6 -0.8 -1 -1.2 Lift Coefficient -1.5 -1 0.2 0.4 0.6 0.8 1 1.2 D/L 0 -0.5 -0.5 0 0 0.5 1
0.5
1.5
-0.6
Lift Coefficient
Cd vs Cl^2 (75%)
0.19 0.18 0.17 0.16 Drag Coeff 0.15 0.14 0.13 0.12 0.11 0.1 0 0.2 0.4 0.6 Lift Coefficient^2 0.8 1 1.2 0.1 0 0.2 y = 0.0559x + 0.1082 R = 0.8535 Drag Coeff 0.24 0.22 0.2 0.18 0.16 0.14 0.12
Cd vs Cl^2 (50%)
y = 0.0716x + 0.1163 R = 0.7401
0.4
0.6
0.8
1.2
1.4
Lift Coefficient^2
Wind Tunnel Tests on a Model Cessna' by Vignesh Palaniappan Figs. 8 and 9 show the drag coefficient versus the lift coefficient and it is evident that they are analogous to the drag curves, this is because the lift-angle of incidence relationship is mostly linear for most of the data points. Only the latter part of this graph is different in that the stalling of the aerofoil (penultimate point) causes drag to increase rapidly and lift generation to decline. The wool tufts attached to the model on the aerofoil were oscillating randomly and rapidly instead of being flat suggesting that the flow had separated at that point. Figs. 10 and 11 show the drag-lift ratio versus the lift coefficient. The ratio is dependent on the actual forces and not the coefficients. From any designer's point of view, a major goal is achieving a low drag coefficient as well as a low drag-lift ratio. Good values influence the green aviation vision of the future by improving fuel economy and aerodynamic efficiency. From a glider's point of view, a minute drag-lift ratio is crucial to staying in the air for long periods. For the most part, the ratio lies between -1 and 1 and so the lift is always greater than the drag. If the lift-drag ratio were plotted against the angle of attack, it would be evident that the graph rises rapidly up to about 3-4 degrees. After this, as the induced drag increases appreciably, the ratio is lesser. Figs 12 and 13 have been plotted with their first and last points omitted so that the linear portion of the curve is discernable. We use this particular graph to calculate the induced drag acting upon the aircraft. The Reynolds numbers for these tests have been calculated to be roughly 146000 at 75% wind tunnel capacity and 95000 for 50% capacity. These are of importance when we consider scaling up the model back to its full scale. All of the laws and values are dependent on the Reynolds number and this must be the same in the tunnel as in the air otherwise all the other calculations end up useless. This concept of dynamic similarity is extremely valuable and is of utmost importance.
INDUCED DRAG ESTIMATION We know the following two relations, from this we can derive an equation to calculate the induced drag coefficient ( ). ( ) ( ) ( )
The drag coefficient was an average of all the data points for a scenario. This was done for both speeds and substituted in. The gradients were calculated by using Excel's 'trendline plotting' feature. The term AR is the aspect ratio (for the model Cessna AR= 7.52).
, , ,
10
As we now have this coefficient, we can simply find the induced drag force by multiplying by the area of the wing and the dynamic pressure. The area A=16.2 and the dynamic pressure is given in the results gathered, we simply have to use the average value ( )( )
, , ,
The weight of the aircraft is roughly 10000N and we can see that the induced drag is in the range of 10% to 25% of this value. STALLING SPEED ESTIMATION The following equation is used for the estimation of the stalling speed of an aircraft. The term 'n' is equal to the load factor and represents the ratio between lift and weight. In steady level flight lift is equal to the weight and so n=1 which is what we will assume when calculating the stall speed. When aircraft bank or rise sharply, the load factor changes. Weight=W= (1040 x 9.81) N. We use the max lift coefficient obtained in the 75% capacity experiment (although note that both coefficients are within 0.05 of each other). This is equal to 1.05. ( )
The actual value is 26m/s and this is reasonably close estimation. We must bear in mind the sources of error that come with using data from wind tunnel testing. The lift coefficient is the one thing that we had to have determined previously, all the other variables were physical quantities. Instead of using the value of 1.05 that we determined in the tunnel, if we try = 1.5, = 26.6 m/s. SOURCES OF ERROR A difficulty of using models is that they are hard to make accurate, the smaller the more difficult. Also we need them represent the full size object as closely as possible (especially the exterior surfaces and physical geometry). The Cessna used here is 1/20th its actual size and the solution to this problem is to make larger models. One obvious error in the testing is that the flow in the tunnel is not the same as the free flow encountered in the sky above. The flow is constrained by the walls; this in turn affects our measurements and conclusions derived from the experiment. So ideally, we want large tunnels(working sections) in comparison to the model. It's quite apparent that if we kept making everything larger (forgetting huge costs) we would essentially be testing the full scale thing itself!
11
Wind Tunnel Tests on a Model Cessna' by Vignesh Palaniappan Finally, we need to consider the Scale Effect, which says that the simple aerodynamic relationships for lift, drag, speed squared law and the law of dependence on area and density aren't strictly true unless certain conditions apply. These conditions are those founded by the Reynolds experiments. We need to ensure that velocity multiplied by the size value (the 'VL' law) remains the same for both the model and full-scale flight. We test a 1/20th scale model, so we should be testing at speeds that are 20 times larger to maintain similarity.
CONCLUSION
The wind tunnel testing entailed taking measurements of the lift and drag forces acting on a 1/20th scale model Cessna at different speeds and angles of attack. It was observed that a linear relationship existed between the angle of attack and the lift coefficient for low angles of attack. The maximum lift was achieved at around 13 degrees after which stalling occurred. Beyond this point, the wool tufts identify the separation of the airflow and the consequent increase in drag. Figs 10 and 11 illustrated how the drag-lift ratio was always below 1 for positive angles of attack (which created positive lift coefficients). This indicates how the aerofoil always generates more lift than drag until stalling. Nevertheless, we already know that at more extreme angles of attack, induced drag plays a more important role by contributing significantly to the overall drag, the draglift ratio will eventually be greater than 1. Another conclusion that can be drawn from the induced drag estimations is that as the airspeed goes up, the induced drag goes up. The estimated stalling speed of 31.8m/s is a guide to the accuracy of this experiment; it is a useful value since the running conditions were not exactly optimal. It underlines the fact that whatever model we test, the errors of using constrained flow and scaled objects need to be accounted for before drawing conclusions from these experiments. In reality, the induced drag is lower than what we calculated. We need to go through a process of refinement to get the two values closer together. Also as we can't increase the size of the tunnel, the alternative is to make an even more accurate representation of the model or to test individual sections of the plane but this can often be costly and erroneous when we piece information back together.
REFERENCES
1. Motor Sport Circuit Guide [Online], Viewed on 31/03/2011, Available at:
http://www.motorsportcircuitguide.com/cms_images/Lotus_F1_Racing_wind_tunnel_model_C.jpg 2.
3. Greenhalgh, E. S., 01/03/2011, 'Wind tunnel tests on a model Cessna', Aeronautics Department, Imperial College London 4. Kermode, A.C (1995) Mechanics of Flight, 10th ed, 'Wind Tunnels' pg 45-51, 'Scale effect and Reynolds Number' pg464-469 5. Cavcar,M., 30/13/2006, 'Stall Speed', Anadolu University, School of Civil Aviation Available at : http://home.anadolu.edu.tr/~mcavcar/common/Stall.pdf
12
Wind Tunnel Tests on a Model Cessna' by Vignesh Palaniappan AVG Cd 0.1485 0.0559 0.3533 446.75 Pa 2556.6 N
APPENDIX
Cessna gross wing area (S): Barometric pressure: Temperature: Air Density: Weight: 75% INCIDENCE (deg) NO WIND -2 0 2 4 6 8 10 12 13 PRESSURE (Pa) 0.000 446.301 442.280 447.105 447.105 448.713 448.713 445.496 449.517 445.496 -2.338 -0.118 2.148 4.345 6.555 8.766 10.934 13.048 14.015 0.002 0.000 0.000 0.002 0.005 0.010 0.015 0.019 0.018 Incidence (corr) Cd (tunnel) Velocity (m/s) 0.00 27.43 27.31 27.46 27.46 27.51 27.51 27.41 27.53 27.41 -5.749 -1.988 2.514 5.874 9.487 13.095 15.857 17.952 17.230 2.495 2.067 1.897 1.843 1.901 1.996 2.393 2.755 4.101 -0.339 -0.118 0.148 0.346 0.556 0.768 0.937 1.051 1.018 0.149 0.123 0.112 0.111 0.117 0.127 0.157 0.181 0.260 0.038 998.9 20.26 1.186 1040 sqr.m m.bar Drag correction FAI correction 0.0174 0.997 0.82 16.2 0.0000178 kg/ms
Lift (N)
Drag (N)
Cl
Cd
D/L
L/D
Cl^2
Reynolds No. Re=(pVL)/ visc 146232.368 145572.174 146364.050 146364.050 146627.058 146627.058 146100.568 146758.386 146100.568
13
PRESSURE Incidence (Pa) (corr) 0.00 186.562 188.170 187.366 188.170 187.366 186.562 186.562 188.170 188.492 -2.404 -0.107 2.154 4.378 6.571 8.746 10.918 13.099 14.037
Cd (tunnel)
Velocity (m/s) 0.00 17.737 17.813 17.775 17.813 17.775 17.737 17.737 17.813 17.829
Lift (N)
Drag (N)
Cl
Cd
D/L
L/D
Cl^2
Reynolds No
-0.405 0.158 0.383 -0.107 0.137 1.280 0.154 0.126 0.818 0.379 0.122 0.314 0.572 0.130 0.217 0.748 0.139 0.172 0.921 0.153 0.151 1.102 0.231 0.190 1.040 0.286 0.257
-2.608 0.164 -0.781 1.223 3.183 4.617 5.803 6.642 5.259 3.898 0.011 0.024 0.144 0.328 0.560 0.848 1.214 1.081
928.7 N
14