Numerical Research on Aerodynamic Efficiency of a ...

AIAA 2015-1678 AIAA SciTech 5-9 January 2015, Kissimmee, Florida 53rd AIAA Aerospace Sciences Meeting

Numerical Research on Aerodynamic Efficiency of GFS UAV

Downloaded by TSINGHUA UNIVERSITY on February 8, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2015-1678

Yifei Zhang1, Lijun Xu2 and Haixin Chen3 School of Aerospace Engineering, Tsinghua University, Beijing, 100084, China

GFS UAV is formed by installing a large cowling at the exit of a duct fan. The cowling is expected to make the UAV able to carry relatively large payload without destroy the flow field. Compared with a same sized unmanned helicopter, GFS is simple in structure which can make its maintenance costs lower; GFS has all its rotational parts protected by duct which makes it safer. This paper gives a brief introduction on GFS history. Then, The hover efficienfy of the GFS UAV is numerically studied. Firstly, a simplified model of duct fan and GFS with different cowlings are compared, mainly focused on the hover efficiency influenced by the type of cowlings. Moreover, a duct fan and a series of GFS with 6 real rotating blades are computed and compared, to discuss how the size and shape of a cowling influence the hover efficiency of GFS. The conclusion is that the cowling of GFS jams air flowing out from the exit of a duct fan, which has a negative influence on the propulsion efficiency. However, the cowling decreases the average jet speed of the whole system, in this way the propulsion efficiency is increased. Overall, GFS has an acceptable hover efficiency penalty than traditional duct fan.

Nomenclature η Vin Vout Power PT

ρ 𝑚̇ FANout FANin T 𝑐𝑇 Q 𝑐𝑄 𝛺 R

= = = = = = = = = = = = = = =

propulsion/hover efficiency average inlet velocity for a jet engine average outlet velocity for a jet engine engine power total pressure fluid density mass flow rate of an engine boundary condition, outlet of the duct fan boundary condition, inlet of the duct fan thrust (Lift) of an GFS thrust coefficient torque on blades torque coefficient rotational speed blade radius

I. Introduction

C

OANDA Effect was presented by Henri Coanda in 1934 at France. It is a natural fluid dynamics phenomenon that if a stream goes passing near a curve solid surface with no too large curvature, the stream will tend to leave its original derection and follow the surface[1]. Coanda Effect has been adapted to many aircrafts, such as YC14(Fig.1), AN-74(Fig.2) and MD Explorer(Fig.3).

1

Graduate student, [email protected], Student Member AIAA. Graduate student, [email protected], Student Member AIAA. 3 Professor, [email protected], Senior Member AIAA. Corresponding Author. 1 American Institute of Aeronautics and Astronautics 2

Copyright © 2015 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved.


Figure 1. YC-14

Figure 2. AN-74

Figure 4. Geoff’s Flying Saucer

Figure 3. MD Explorer

British enginner Geoff Hatton successfully designed new concept VTOL UAV(Fig.4) named GFS(Geoff’s Flying Saucer). According to Hatton, a GFS can gain extra lift by taking advantage of Coanda Effect, which makes it a better aircraft than traditional duct fan or helicopter[2]. As shown in figure.4, a GFS can be constructed by adding a large cowling under a duct fan. For the reason of Coanda Effect, the flow driven by duct fan will follow the curvature of the cowling, thus lower the average velocity of the jet stream. And a lower velocity 𝑉𝑜𝑢𝑡 usually means a higher propulsion efficiency, as shown in Eq.(1) for the propulsion efficiency. Besides, as the air flew faster above cowling than below it, Hatton thought the static pressure is lower above the cowling for the reason of Bernoulli’s principle, thus the cowling was generating additional lift. 2 𝜂= (1) 1 + 𝑉𝑜𝑢𝑡 /𝑉𝑖𝑛 However, the cowling will also increase friction to the flow. Apparently, factors that will decrease the efficiency also exits[4]. Many researchers are doing researches on GFS in recent years. Yu Sun and Haixin Chen did a detailed research[3] on GFS in 2008. According to the result of Sun(Fig.5), the same input shaft power could generate higher lifting thrust on a GFS, or in another word, the cowling produced higher propulsion efficiency, since the cowling enlarged the overall mass flow. The cowling could generate positive extra lift(Fig.6), but it ultimately had a negative value when the input power was large enough. Julian and Tan Kok Ping[5] did an experiment(Fig.7) on GFS in the year 2010. Julian’s result(Fig.8) showed that Coanda Effect can be used to generate lift, but it did not mean that the lift of GFS comes from Coanda Effect. Julian did another experiment[6] in 2011, showing that the cowling shape of GFS could affect the lift, but he gave no further discussion on that. Meanwhile, C.Barlow also did an experimental research on GFS[7], which showed that at a same RPM, the aircraft produced less lift with the cowling in place(Fig.9). This suggested that the cowling of GFS was generating negative lift. 2 American Institute of Aeronautics and Astronautics


Figure 5. The Thrust V.S. Input Power

Figure 6. Total Thrust and the Additional Thrust on the Cowling V.S. Power

Figure 8. Julian’s Result

Figure 7. Julian’s Equipment

Figure 9. Performance of the GFS-UAV N-01A body and AXI 2217/20 motor & GWS 1060 prop 3 American Institute of Aeronautics and Astronautics

No matter the efficiency, GFS has several advantages. GFS makes a small duct fan able to carry a bulk payload, which can be used to carry large volume staff without destroying the aerodynamic performance. Compared to a small size unmanned helicopter, GFS use only one fixed fan to realize pitch, yaw and roll, hence is much simper in structure, which makes its maintenance easier and cheaper. Moreover, GFS has all the rotational parts protected by the duct and cowling, which makes it safer when being used in especially urban area in the case of collision. Since there are yet unclear mechanisms on GFS’s lift generation, this paper studies the hover efficiency of GFS, mainly focusing on the influence of the cowling.


II. Modeling and Calculation The commercial software CFX is used for the CFD (Computational Fluid Dynamics) simulation. RANS(ReynoldsAveraged Navier-Stokes equations) is chosen as the mumerical method in this paper and SST(Menter’s Shear Stress Transport) model is used for Reynolds stress closure. A. Simpified Models The cowling plays the most important role in aerodynamic efficiency for GFS and makes it different from the duct fan. In order to simulate the GFS’s primary flow features with small number of grids, the blades are neglected from the duct fan first. Simplified GFSs with three different kinds of cowlings (Fig.10), as well as a simplified duct, are computed and compared. The inlet of a duct is set as the outlet of the ambient flow field, while the outlet of the duct is the inlet(Fig.11). Model S-GFS1(Fig.10 b) has a cowling with a flat bottom surface, while the cowling of model SGFS2(Fig.10 c) has no bottom surface and like a bowl, model S-GFS3(Fig.10 d) has a streamlined long tail cone. The cowlings’ upper parts as well as the ducts are identical. Some geometry and mesh details are shown in Table.1.

a), model S-FAN

b), model S-GFS1

c), model S-GFS2

d), model S-GFS3

Figure 10. Simplified models of duct fan and GFS

4 American Institute of Aeronautics and Astronautics


Figure 11. Boundary conditions for simplified model

Table 1. Geometry and Mesh Details of Simplified Models MODEL DISCRIPTION VALUE Duct Height, mm 60 Duct Inlet Diameter, mm 280 S-FAN Structured Mesh 337K Cowling MAX Diameter, mm 650 Cowling Height, mm 225 S-GFS1 Structured Mesh 845K Cowling MAX Diameter, mm 650 Cowling Height, mm 225 S-GFS2 Structured Mesh 849K Cowling MAX Diameter, mm 650 Cowling Height, mm 936 S-GFS3 Structured Mesh 940K To simulate the blades’ effect of adding work to the flow, and make the velocity field as correct as possible, static pressure is chosen as the boundary condition of the duct fan inlet and total pressure as that of duct fan outlet. The static pressure is fixed and the total pressure is slightly changed to ensure the massflow conservation. The geometry and the rotation of the fan blades are not simulated. B. Complex Models The classical geometry of GFS, like S-GFS2 , without bottom surface for the cowling, is further studied. For the purpose of comparison, a duct fan configuration named FAN (Fig.12a) is constructed with a one-meter-diameter 6blade real fan. Unstructed mesh(Fig.12d) is used for the space inside the duct about the blades (Fig.12d) and structured mesh for the other places(Fig.12c).



a), model FAN

b), model GFS0

c), mesh of FAN

d), mesh of blade

e), mesh of GFS0 Figure 12. Model FAN and GFS0 A GFS model named GFS0(Fig.12b) shares the same blade geometry and mesh with FAN. A baseline cowling is mounted. To study the influence of the distance between the baseline cowling and duct fan, a series of models named GFS-X are constructed based on GFS0. As the distance increases, the models are named GFS-X1, GFS-X2, GFS-X3, GFS-X4. To study the influence of the cowling radius, models named GFS-Z are constructed. Similarly, models are named GFS-Z1, GFS-Z2, GFS-Z3, GFS-Z4, GFS-Z5 as the cowling’s diameter increases.

a), model GFS-X

b), model GFS-Z

Figure 13. model GFS-X and GFS-Z


The meshes of all the GFS models are in high similarity, 389K mesh number for unstructured mesh and about 500K for structured mesh. Varies rotation velocities are given to simulate different duct fan working conditions. Two coordinate systems are used in the computation. The blade mesh is set to rotating at a fixed RPM, while the far-field is set to opening boundary conditions.


III. Results and Analysis Hover efficiency and power consumption are two of the most important physical quantities for GFS. The definition of the hover efficiency is given in helicopter theory[8]. Hover Efficiency: 𝑐𝑇1.5 𝜂= (2) 𝑐𝑄 Where Thrust Coefficient 𝑇 𝑐𝑇 = (3) 𝜌𝐴(Ω𝑅)2 Torque Coefficient 𝑄 𝑐𝑞 = (4) 𝜌𝐴Ω2 𝑅3 A. Simplified Models The power input by the fan to the airflow by the simplified models can be calculated by integrating the total pressure change on massflow, as shown in Eq.(5). Total pressure is used in order to take both momentum and pressure into account when solving shaft power. 𝑃𝑇 𝑃𝑇 𝑑𝑚̇ − ∫ 𝑑𝑚̇ 𝜌 𝐹𝐴𝑁𝑜𝑢𝑡 𝐹𝐴𝑁𝑖𝑛 𝜌

𝑃𝑜𝑤𝑒𝑟 = ∫

(5)

Lift can be easily obtained by integrating total pressure on all the surfaces of simplified models. However, there are no RPM for simplified models. According to dimensional analysis, hover efficiency can be deducted to be as follows.

𝜂=√

MODEL S-FAN S-GFS1 S-GFS2 S-GFS3

𝑇3 𝑃𝑜𝑤𝑒𝑟 2 𝜌𝐴

Table 2. Numerical results of simplified models Duct Fan Inlet -50Pa -100Pa -200Pa -400Pa Static Pressure 10.45 20.61 69.52 187.50 Power, W 2.84 5.05 10.50 20.89 Lift, N 80.66 218.64 603.03 1681.6 Power, W 4.76 8.86 18.56 36.26 Lift, N 80.70 218.61 602.20 1676.5 Power, W 6.28 12.16 24.31 48.95 Lift, N 81.26 219.69 604.53 1682.4 Power, W 14.09 27.49 54.18 107.47 Lift, N


(6)

-1000Pa

716.17 51.95 5916.8 88.32 5895.4 114.01 5900.0 248.33

Total Lift

140 S-FAN

120

S-GFS1

S-GFS2

S-GFS3

lift, n

100 80 60

40


20 0 0

1000

2000

3000

4000

5000

6000

Power, W

Figure 14. Total Lift of Simplified Models

Hover Efficiency 100.00% 80.00% 60.00% 40.00% 20.00% 0.00% S-FAN

S-GFS1

S-GFS2

S-GFS3

Figure 15. Hover Efficiency of Simplified Models, FANout static pressure -400Pa The result(Fig.15) shows that for the hover efficiency, S-FAN > S-GFS3 > S-GFS2 > S-GFS1. It is easy to understand that the efficiency of a bare duct fan is higher than that of GFS as the cowling is generating friction and separation in the flow. As shown in Fig.16, the flow downstream of model S-GFS1 is not stable and becomes unstable, thus causing a lift loss. In the cowling of model S-GFS2, a stable votex is generaged automatically, forming a streamline similar to to the geometry of S-GFS3, making the mainstream of air flow able to pass it smoothly. Compared to duct fan, the cowling could increase the static pressure near the exit of the fan (Fig.17). This pressure increase, enlarges the lift on duct fan and the drag on cowling at the same time.



a), Streamline of S-FAN

b), Streamline of S-GFS1

c), Streamline of S-GFS2

d), Streamline of S-GFS3

Figure 16. Streamline of Simpilified Models, FANout static pressure -400Pa

a), Static pressure contour of S-FAN

b), Static pressure contour of S-GFS1

c), Static pressure contour of S-GFS2

d), Static pressure contour of S-GFS3

Figure 17. Static Pressure Contour of Simpilified Models, FANout static pressure -400Pa As shown in Table.2, at a same mass flow rate, the lift of S-GFS is larger than that of S-FAN. Coanda Effect can be observed in Figure.17. The outer part of cowling’s upper surface has a lower static pressure than the bottom surface of the cowling. However, a high pressure area always exists below the duct fan. Overall, cowlings are generating negative lift(Fig.18).


Force on Cowlings

0 -20

0

1000

2000

3000

4000

5000

6000

7000

-40

force, n

-60 -80 -100


-120 -140 -160

S-GFS1

S-GFS2

-180

S-GFS3 power, w

Figure 18. Total Lift of Simplified Models

B. Complex Models Different from the simplified models, since the CFD computation is performed on a configuration with rotional blades, more direct formulae are used to calculate the thrust power and hover efficiency. Torque Q could be integrated easily on blades. Thus 𝑃𝑜𝑤𝑒𝑟 = 𝑄 ∙ Ω (5)

𝜂=√

MODEL FAN

GFS0

𝐿3

(6)

𝑄2 Ω2 𝜌𝐴

Table 3. Numerical Results of Complex Models Duct Fan RPM 1000 2000 3000 4000 2245.08 18285.5 63239.5 155238 Power, W 184.62 749.52 1700.36 3034.8 Lift, N 92.86% 91.79% 89.16% Hover Efficiency 92.51% 2379.1 19332.6 66867.7 164767 Power, W 159.17 641.72 1456.00 2616.75 Lift, N 71.30% 70.81% 69.60% Hover Efficiency 71.40%

5000

321293 4730.26 83.83% 351668 4211.34 66.65%

Compare the lift, power and hover efficiency of model FAN and GFS0. Though the hover efficiency of GFS0 is a bit lower, the performance of GFS0 is much better than the expected value according to the analysis of simplified model. As the designed RPM of the FAN blading is 4000, numerical data at RPM 4000 is compared. As shown in Figure.20(b), the velocity contour shows that air flow does follow the upper surface, as a result of Coanda Effect. However, no extra lift is generated by adding a cowling on, and the cowling is still generating negative lift. The high pressure area is still there, and it could add both the lift on blades and drag on cowling. Unfortunately, the neat contribution is negative. In fact, at RPM 4000, hover efficiency of model GFS0 is about 80% that of model FAN. Considering GFS0 could use the volume under cowling to carry payloads, and the diameter of this payload could even be larger than that of duct fan, such a penalty on efficiency seems acceptable. 10 American Institute of Aeronautics and Astronautics

Lift & Hover Efficiency


100.00%

4500

90.00%

4000

80.00%

3500

70.00%

3000

60.00%

2500

50.00%

2000

40.00%

1500

30.00%

1000

20.00%

η

Lift, N

5000

500

Lift-FAN

0 0

50000

100000

Lift-GFS0 150000

η-FAN

200000 250000 Power, W

300000

η-GFS0 350000

10.00%

0.00% 400000

Figure 19. Total Lift and Hover Efficiency of Complex Models

a), Velocity contour of FAN

b), Velocity contour of GFS0

Figure 20. Velocity Contour of Complex Models, Duct Fan RPM 4000

a), Static pressure contour of FAN

b), Static pressure contour of GFS0

Figure 21. Static Pressure Contour of Complex Models, Duct fan RPM 4000 Operating conditions of model GFS-X and GFS-Z at RPM 4000 are calculated and the results are given bellow. As Figure.22 shows, as the distance between duct fan and cowling increases from GFS-X1 to GFS-X4 (Fig.22), hover efficiency grows larger. Imagine an extreme case. The cowling is so far from the duct fan that the cowling’s influence is little, of course the hover efficiency will have a tendency of becoming that of the duct fan. On the contrary, if the distance is too small, flow will be easily “chocked”, thus causing a low hover efficiency. 11 American Institute of Aeronautics and Astronautics

As the cowling radius increases from GFS-Z1 to GFS-Z5(Fig.23), hover efficiency decreases. One can imagine the extreme cases that the cowling radius is so large that the fan-cowling aircraft cannot even hover. On the other end, if the cowling radius is very small, model GFS-Z turns into model FAN, and the hover efficiency also recovers to its value.

η of GFS-X 71.50%

71.46%

71.47%

GFS-X3

GFS-X4

71.00% 70.48%


70.50% 70.00% 69.50%

69.48%

69.00% GFS-X1

GFS-X2

Figure 22. Hovering Efficiency of Models GFS-X, Duct fan RPM 4000

η of GFS-Z 71.50%

71.31%

71.14% 70.89%

71.00%

70.49%

70.50%

70.17%

70.00% 69.50% 69.00%

GFS-Z1

GFS-Z2

GFS-Z3

GFS-Z4

GFS-Z5

Figure 23. Hovering Efficiency of Models GFS-Z, Duct fan RPM 4000 To confirm the conclusions related to cowling size, another model GFS-Z0(Fig.24) is build. The cowling size of model GFS-Z0 is even smaller, and the hover efficiency is 73.25%.



Figure 24. Contours of Models GFS-Z0, Duct fan RPM 4000

IV. Conclusion From the numerical results, the following conclusions can be drawn: A. The results of simplified models and complex models are in good agreement. The hover efficiency of GFS is lower than that of a bare duct fan. B. The cowling of GFS can enlarge the lift on duct fan while generating a negative lift on itself. C. The hover efficiency grows larger while the distance between the duct fan and the cowling increases. D. The hover efficiency grows smaller while the diameter of cowling increases.

Acknowledgments This work was supported by National Key Basic Research Program of China (2014CB744806) and National Natural Science Foundation of China (11102098 and 11372160).

References Coanda, France, United States Patent Office, “Propelling Device”, 2108652[P], Date of Filing: 15.01.1935 2Geoffrey Hatton, Simon Mclntosh, GFS Projects Limited, Peterborough, United Kingdom, UK Patent Application, “Craft having aerofoil surface for controlling its spin”, GB 2424400 A, Date of Filing: 23.03.2005. 3Chen Haixin, Sun Yu, “Numerical Experiments on the Lift Generating Mechanisms of the GFS UVA”, 44th AIAA/ASME/SAE/ASEE Joint Propulsion Conference & Exhibit, 21-23 July 2008, Hartford, CT, USA. 4Lewis, Darren, et al. “Investigating the use of the coanda effect to create novel unmanned aerial vehicles”, University of Southampton [online database], URL: http://eprints.soton.ac.uk/id/eprint/343525 [cited 04 Oct 2012]. 5Ping, Julian Tan Kok, “Preliminary design of vertical take-off and landing (VTOL) UAV with steerable vertical thrust effect”, IEEE Conference on Robotics Automation and Mechatronics, 28-30 June 2010, Singapore. 6Ping, Julian Tan Kok, “Coanda Effect Test Bench (CoETB) - Design enhancement of the CoandaJLT craft”, IEEE Conference on Sustainable Utilization and Development in Engineering and Technology, 20-21 Oct 2011, Malaysia. 7 Chris Barlow, “Investigating the use of the Coanda Effect to create novel unmanned aerial vehicles”, International Conference on Manufacturing and Engineering Systems, 17-29 Dec 2009, Huwei, Taiwan. 8Weiqin Chen, “Experimental Investigation of Influence of Rotor Solidity on Hover Efficiency”, Journal of Nanjing University of Aeronautics & Astronautics, Dec 1997, Vol.29, No.6. 1Henri
