Smart Materials and Structures

PAPER

Design, fabrication, and characterization of multifunctional wings to

harvest solar energy in flapping wing air vehicles
To cite this article: Ariel Perez-Rosado et al 2015 Smart Mater. Struct. 24 065042

Manuscript version: Accepted Manuscript

Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”

This Accepted Manuscript is © © 2015 IOP Publishing Ltd.

During the embargo period (the 12 month period from the publication of the Version of Record of this article), the Accepted Manuscript is fully
protected by copyright and cannot be reused or reposted elsewhere.
As the Version of Record of this article is going to be / has been published on a subscription basis, this Accepted Manuscript is available for reuse
under a CC BY-NC-ND 3.0 licence after the 12 month embargo period.

After the embargo period, everyone is permitted to use copy and redistribute this article for non-commercial purposes only, provided that they
adhere to all the terms of the licence https://creativecommons.org/licences/by-nc-nd/3.0

Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions will likely be
required. All third party content is fully copyright protected, unless specifically stated otherwise in the figure caption in the Version of Record.

View the article online for updates and enhancements.

This content was downloaded from IP address 123.231.125.188 on 16/08/2018 at 07:36

Page 1 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
Design, Fabrication, and Characterization of Multifunctional Wings to Harvest Solar
4
5 Energy in Flapping Wing Air Vehicles
6
7
8 Ariel Perez-Rosado, Rachel D. Gehlhar, Savannah Nolen, Satyandra K. Gupta, Hugh A. Bruck
9
10 Department of Mechanical Engineering
11 University of Maryland
12 College Park, MD 20742
13
14
15
16
17 Abstract
18 Currently, flapping wing Unmanned Aerial Vehicles (UAVs) (a.k.a., ornithopters or robotic birds)
19
20
sustain very short duration flight due to limited on-board energy storage capacity. Therefore,
21 energy harvesting elements, such as flexible solar cells, need to be used as materials in critical
22 components, such as wing structures, to increase operational performance. In this paper, we
23 describe a layered fabrication method that was developed for realizing multifunctional composite
24 wings for a unique robotic bird we developed, known as Robo Raven, by creating compliant wing
25
structure from flexible solar cells. The deformed wing shape and aerodynamic lift/thrust loads
26
27 were characterized throughout the flapping cycle to understand wing mechanics. A multifunctional
28 performance analysis was developed to understand how integration of solar cells into the wings
29 influences flight performance under two different operating conditions: (1) directly powering
30 wings to increase operation time, and (2) recharging batteries to eliminate need for external
31 charging sources. The experimental data is then used in the analysis to identify a performance
32
33
index for assessing benefits of multifunctional compliant wing structures. The resulting platform,
34 Robo Raven III, was the first demonstration of a robotic bird that flew using energy harvested from
35 solar cells. We developed three different versions of the wing design to validate the multifunctional
36 performance analysis. It was also determined that residual thrust correlated to shear deformation
37 of the wing induced by torsional twist, while biaxial strain related to change in aerodynamic shape
38
correlated to lift. It was also found that shear deformation of the solar cells induced changes in
39
40 power output directly correlating to thrust generation associated with torsional deformation. Thus,
41 it was determined that multifunctional solar cell wings may be capable of three functions: (1)
42 lightweight and flexible structure to generate aerodynamic forces, (2) energy harvesting to extend
43 operational time and autonomy, and (3) sensing of an aerodynamic force associated with wing
44 deformation.
45
46
47
48 Keywords: Multifunctional wing structures, solar cells, energy harvesting, multifunctional
49 performance analysis, digital image correlation, flapping wings
50
51
52
53
54
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 2 of 35

1
2
3
1. Introduction
4
5 Unmanned Aerial Vehicles (UAVs) are emerging as an important tool in a wide variety of
6
7
8
defense and civilian applications [1-3]. Flapping wing UAVs have the potential to combine the
9
10 positive aspects of both fixed-wing and rotary flight, while eliminating many of the negative
11
12 aspects. Inspiration for flapping has been derived from bats, insects, and birds, and many different
13
14
15 platforms have been constructed, characterized and modeled [2-22]. For example, the robotic
16
17 birds, Small Bird and Big Bird, constructed at the University of Maryland, as well as the
18
19 ornithopter constructed at the University of Delaware [22-25]. Though the size and chassis of
20
21
22 each of these UAVs may differ, the wings were all similar in nature. They were constructed using
23
24 stiff lightweight rods as structural materials and a thin Mylar-based film as the wing surface.
25
26
27 As a part of previous work, we have developed a highly maneuverable robotic bird named
28
29 Robo Raven [26,27]. Robotic birds are vehicles that rely on flapping wings and their deformation
30
31
32 to generate the aerodynamic forces necessary for flight. The size of these vehicles are comparable
33
34 to the size of an actual bird found in nature. This platform features independently controlled
35
36 programmable wings. In robotic birds such as Robo Raven, flight endurance is one of the primary
37
38
39 concerns. To perform missions in remote regions, the UAV cannot charge batteries using electrical
40
41 outlets. A possible way to overcome this challenge would be to charge batteries using on-board
42
43
44
solar cells. Since Robo Raven has a large wing area, solar cells can be integrated into the wings.
45
46 The resulting compliant wing structure with integrated solar cells can be considered both
47
48 multifunctional and smart because it not only provides lift and thrust, but also acts as a method of
49
50
51 harvesting energy and sensing changes in deformation caused by aerodynamic loading, which can
52
53 be used to determine changes in the flapping profile to improve flight control. This combination
54
55 allows for increased flight time while decreasing the payload contribution of a large power source,
56
57
58 thus potentially allowing for either: (1) size reduction with the same performance capability, or (2)
59
60
Page 3 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
an increase in overall payload capacity. It also introduces new capabilities for control schemes
4
5
6 through new sensing capabilities. Successful development of multifunctional compliant wing
7
8 structures through the integration of multiple functions can be expanded to other aspects of all
9
10
11
UAVs, including fixed wing and rotary craft.
12
13
Multi-functional structures combine multiple functional requirements into a single
14
15
16 structural component to create better efficiency in the overall design [28,29]. For example, a Micro
17
18 Air Vehicle (MAV) constructed with MEMS technology has a membrane made of a PVDF skin,
19
20
21
allowing it to act as a real time load sensor to directly analyze flight performance [13,14]. Ma et.
22
23 al. developed another MEMS-based insect-inspired flapping wing platform known as RoboBee
24
25 uses artificial muscles to achieve novel controlled flight dynamics [30]. Thomas et. al. described
26
27
28 the combination of structure and battery in the design of an electric-propelled UAV as an example
29
30 of a multi-functional material system [29,31]. More recently at the University of Maryland,
31
32 elastomeric strain gauges were placed on the wings of a flapping wing Micro Air Vehicle (MAV)
33
34
35 [32]. These sensors captured deformations caused by flapping. The outputs from these sensors
36
37 were directly correlated to thrust production which essentially made the wing into a skin-like
38
39
structure. For remote applications, endurance will be an important performance metric as
40
41
42 evidenced through flight time for UAVs. Flight time is directly related to the energy supplied by
43
44 the battery, the weight of each component of the UAV, and aerodynamic parameters. The ability
45
46
47
to harvest energy and sense changes in wing deformation increases endurance and improves
48
49 system efficiency, which directly impacts all areas of the flight envelope.
50
51
52 Integrating solar cells in the wings present the following three challenges. First, a new
53
54 manufacturing process is needed to integrate solar cells into wings without substantially increasing
55
56
57
weight. Second, we need to ensure that wings with integrated solar cells maintain the appropriate
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 4 of 35

1
2
3
deformation during the flapping cycle to ensure production of adequate aerodynamic lift and
4
5
6 thrust. Finally, we need to make sure the modified version of Robo Raven with multifunctional
7
8 wings produces enough thrust and lift to compensate for the heavier wings and enable flight.
9
10
11 In this paper, we describe a new layered fabrication method for integrating commercial off-
12
13
the-shelf solar cells into wings for a new solar-powered FWAV (Flapping Wing Aerial Vehicle):
14
15
16 Robo Raven III. Different wing designs were tested to observe how adding different quantities of
17
18 solar cells affects flight performance through wing deformation, and how this leads to changes in
19
20
21
power output that can be measured for potential onboard sensing and control. A new
22
23 multifunctional performance analysis is also developed to quantify the effects of solar cell
24
25 integration on recharge time and flight time to determine trade-offs from the multifunctional
26
27
28 effects of solar cell integration to be considered by examining the impact of lift and thrust on power
29
30 requirements versus the gains from recharging by harvesting solar energy.
31
32
33 2. Design and Layered Manufacturing Process for Compliant Multifunctional Wings
34
35
36 2.1 Design of Compliant Multifunctional Wing
37
38
39 Multifunctional wings were created for Robo Raven III (Figure 1). The design of the
40
41 multifunctional wing, also seen in Figure 1, was adapted from a design we used previously for
42
43 successfully realizing flapping wing UAVs, which has been shown to be effective in generating
44
45
46 lift and thrust forces across a variety of applications and size scales [23-26]. The parameters of the
47
48 wing are as follows: S is the semi-span, C is the chord, and t ୬ are the diameters of carbon fiber
49
50
51
stiffening rods. Table 1 presents values of the wing parameter used in the design reported in this
52
53 paper. The wing membrane is a 0.001” thick film of biaxially-oriented polyethylene terephthalate
54
55 (Mylar) which provides flexibility and toughness while remaining lightweight. Table 2 lists the
56
57
58 properties of the base Robo Raven platform with batteries.
59
60
Page 5 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18 Figure 1. (left) Robo Raven III, the first solar powered robotic bird using multifunctional wings,
19 and (right) parameters for the multifunctional wing design: S is the semi-span, C is the chord,
20 and ࢚࢔ are the diameters of carbon fiber stiffening rods
21
Youtube Video: https://www.youtube.com/watch?v=t1_mPe8Y0V4
22
23
24
25
26
27
Parameter Value Units
28
29
30
S 605.8 mm
31
32 C 362.0 mm
33
34 t1 3.18 mm
35
36 t2 1.63 mm
37
38 t3 1.63 mm
39
40 t4 1.63 mm
41
42 θ1 0.358 Rad
43
44 θ2 0.750 Rad
45
46 Table 1. Parameters for multifunctional wing design.
47
48
49
50 Parameter Robo Unit
51 Raven
52 Total Mass 0.29 kg
53
54 Length 0.554 m
55
56 Wingspan 1.168 m
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 6 of 35

1
2
3
4
Average Chord 0.248 m
5
6 Aspect Ratio 2.01
7
8 Flight Speed 6.7 m/s
9
10 Table 2. Properties of Robo Raven.
11 Youtube Video: https://www.youtube.com/watch?v=mjOWpwbnmTw
12
13 2.2 Multifunctional Wing Fabrication
14
15
16 To produce appropriate aerodynamic lift and thrust forces, many ornithopters rely on large
17
18
19
deformations using compliant wings at lower flapping frequencies to achieve airfoil shapes [26].
20
21 The basic compliant wing structure weighs 16.8 g with a total area of 1420 cm2. To maintain
22
23 compliance when creating a multifunctional wing with a similar structure, Powerfilm’s© MPT6-
24
25
26 75 flexible solar cell modules were chosen. These flexible 7.3 x 11.4 cm solar cell modules are
27
28 reported by the manufacturer to produce 50 mA of current at 6 V at 100% sunlight flux, which
29
30 represents their maximum power point. However, the bending stiffness and mass of the solar cells
31
32
33 as packaged and was much higher than the Mylar, and therefore would not allow the wing to
34
35 deform enough to maintain flight. Therefore, modifications had to be made to the solar cells to
36
37
reduce the mass and bending stiffness to be more compatible with the Mylar. By heating and
38
39
40 peeling off the protective encapsulation on the solar cells, the bending stiffness of the solar cells
41
42 and mass was substantially reduced. Then, solar modules were glued and soldered together in
43
44
45
parallel to produce more current. Creating the multifunctional wings from the de-encapsulated
46
47 solar cells modules involved a layered manufacturing process (Figure 2), and the completed
48
49 multifunctional wing structure integrated into Robo Raven III can be seen in Figure 3.
50
51
52
53
54
55
56
57
58
59
60
Page 7 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
4 (a) (d)
5
6
7
8
9
10
11
12
13
14
15
16
17 (b) (e)
18
19
20
21
22
23
24
25
26
27
28
29 (c) (f)
30
31
32
33
34
35
36
37
38
39
40
Figure 2. Layered manufacturing process for multifunctional solar cells wings (bottom side of
41 wing is shown).
42 To fabricate the wing, a layered manufacturing process was developed to provide precise
43
44
45 control over the location of each element of the wing (Figure 2). The layered manufacturing
46
47 process consisted of the following steps:
48
49
50 (a) A sheet of mylar is secured to a work table with the use of magnets.
51
52 (b) The wing shape including the hole for the solar modules is cut from the secured sheet
53
54
55 of mylar.
56
57 (c) The solar modules are applied to the wing.
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 8 of 35

1
2
3
(d) A mylar frame is adhered around the solar modules that holds the solar cells in place.
4
5
6 (e) The spars are held in place using magnetic holders with notches while they are adhered
7
8 to the wing.
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27 Figure 3. (Left) Assembled multifunctional wing with 6 solar cell modules, (Right)
28 multifunctional wings integrated into Robo Raven III.
29
30
31
32 Once Robo Raven III was completed, a flight test was conducted and it was determined
33
34 that the platform could achieve flight. The first version of Robo Raven III used 6 solar cell modules
35
36 in each wing to generate 300 mA at 6V, so a second row of solar cells consisting of 5 modules
37
38
39 were used to replace as much of the original wing material as possible (Figure 4). This version of
40
41 Robo Raven III was also flight tested, and it was determined that the new wing design was
42
43
incapable of continuous flight for more than 10 seconds due to increase in mass and decrease in
44
45
46 thrust and lift force generation. Based on our previous experience, additional compliance at the
47
48 trailing edge of the wing can compensate for the increase in stiffness over the area of the wing
49
50
51 when solar modules are integrated, so the wing was redesigned accordingly. The modified wing
52
53 design compared to the original wing design can also be seen in Figure 4. There are three main
54
55 differences between the new design and the previous. The first two involve extending carbon fiber
56
57
58 tubes in the inside part of the wings to permit increase in wing area and compliance of the wing at
59
60
Page 9 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
the trailing edge. The final modification involved changing the shape of the Mylar skin into a
4
5
6 “teardrop”. It was determined that the modified wing design was capable of restoring flight
7
8 capability to Robo Raven III (see https://www.youtube.com/watch?v=a8x8P5F3qTI).
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Figure 4. (left) original Robo Raven III wing design with 11 solar cell modules, and (right) the
25
26 modified wing design.
27 2.3 Integrating Multifunctional Wings into Robo Raven for Energy Harvesting
28
29
30 Robo Raven uses a two-cell Lithium Polymer battery rated at 7.4 V and 370 mAh. To
31
32
33
maintain the balance of the battery cells when charging with the multifunctional wings, a charging
34
35 circuit was design as seen in Figure 5. Each module produces 50 mA at 6 V at 100% sunlight flux
36
37 with the voltage from each wing depicted in Figure 5 as V1 and V2. A zener diode with a
38
39
40 breakdown voltage of 4.3 V is used to regulate the voltage so it does not exceed the maximum of
41
42 4.2 V for each cell. The resistor in the circuit was chosen to achieve the appropriate voltage drop
43
44 from the modules based on their current, reducing the power generation by 25%.
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 10 of 35

1
2
3
4 4.2 V Reference wire
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23 8.4 V Reference wire
24
25 Figure 5. (left) Schematic of battery charging circuit used for multifunctional wing structures,
26
and (right) the actual wiring of the solar cells for the circuit.
27
28 For direct powering of the servomotors, the solar cells can be directly connected to the
29
30
31
servomotors instead of the battery. This is possible since servos can operate at up to 7.2V and the
32
33 solar cell output has been measured up to 7.8V. This would optimize performance of the solar cells
34
35 for powering the UAV in series with the battery pack, as opposed to the 25% reduction in power
36
37
38 experienced when utilizing the recharge circuit for the battery. This has an additional benefit of
39
40 prolonging the life of the battery by allowing the solar cells to assist the battery in powering the
41
42 UAV, thereby reducing the current draw on the battery and extending the discharge time.
43
44
45
46
47 3. Experimental Characterization of Wing Mechanics
48
49
50 3.1 Measurement of Lift and Residual Thrust Forces
51
52
53 Since existing computational models are inadequate for accurately predicting aerodynamic
54
55 loads acting on compliant flapping wings, direct measurement of these loads during the flapping
56
57 cycle was selected as the method for gaining insight into the effects of wing design parameters on
58
59
60
Page 11 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
the wing mechanics. For this study, we used a new test stand we developed with a 6 DOF ATI
4
5
6 Mini40 load cell mounted on a wood and Delrin frame for measuring aerodynamic lift and thrust
7
8 loads simultaneously, as well as the moments generated (Figure 6). The test stand also allowed
9
10
11
the UAV to be set to any angle of attack from 0 to 20 degrees, which was the angle of the bird
12
13 body relative to the wind direction. To simulate the actual flight conditions, the test stand is placed
14
15 at the end of a wind tunnel operating at 6 m/s, which is near the actual flight speed.
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41 6 Degree of
42 Freedom Load Cell
43
44
45
46 Figure 6. Test stand used in wind tunnel to characterize aerodynamic lift and thrust loads
47 on the compliant multifunctional wings.
48
49
50
51
52
53
54
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 12 of 35

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Figure 7. Time resolved load cell results for all four wing designs: (top left) regular, (top right) 6
31 module, (bottom left) 11 module, (bottom right) modified 11 module.
32
33 The wings were testing at a flapping frequency of 4 Hz and a range of 60 degrees. The
34
35 wings flapped symmetrically orthogonal to the body of the UAV. Time resolved load profiles for
36
37
38
wings with and without solar cells can be seen in Figure 7 for Robo Raven. These thrust and lift
39
40 profiles are consistent with previous measurements and models of flapping wings where the lift
41
42 produces a sinusoidal profile consistent with aerodynamic drag while the residual thrust exhibits a
43
44
45 double peak consistent with a “blowback” effect from the rear of the wing during the flapping
46
47 cycle [34]. As a result, the peaks appear 180° out-of-phase when they overlap on the time-resolved
48
49 plot.
50
51
52 Comparing these profiles for each wing design, it can be seen that there is a slight change
53
54
55 in performance caused by the addition of solar cells on the wings. Because the solar cells stiffen
56
57 the wings and reduce compliance in sections of the wing structure, it was predicted that the solar
58
59
60
Page 13 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
cell wings would underperform the regular wings. However, from the profiles it seems that the 6
4
5
6 module wings actually have slightly larger values for lift compared to the regular wings. For the
7
8 11 module wings, the values for thrust decreased significantly compared to the original wings with
9
10
11
only a slight increase in lift, which was consistent with the observed loss of flight capability. The
12
13 modified wing design had an increase in lift force generation compared to the original wings,
14
15 consistent with the observed restoration of flight capability. The average values of lift and residual
16
17
18 thrust load for each trial can be seen in Table 3.
19
20
21 Robo Raven 12 Module 22 Module Modified 22
22 Robo Raven III Robo Raven III Robo Raven III
23 Residual Lift (g) Residual Lift Residual Lift Residual Lift
24 Thrust (g) Thrust (g) (g) Thrust (g) (g) Thrust (g) (g)
25 Trial 1 111 218 105 201 70.6 240 77.4 247
26
27
Trial 2 104 220 98 219 75.2 242 101 229
28 Trial 3 109 221 98 218 91.9 237 74.8 268
29
30 Average 108 220 100 212 79.2 240 84.5 248
31
32 Std. Dev. 3.36 1.79 3.86 10.1 11.2 2.53 14.7 19.5
33
34 Table 3. Lift and residual thrust loads generated by each wing design
35
36 In a wind tunnel, a residual thrust value of 0 g would correlate to steady-state flight
37
38 conditions. However, our low speed wind tunnel has a maximum velocity of ~6 m/s, while the
39
40
41 actual flight velocity for Robo Raven is 6.7 m/s. Since this meant the aerodynamic force would be
42
43 approximately 25% greater during flight, a scaling factor of 1.4 X was determined using this value
44
45
combined with a measured maximum payload of 40 g for Robo Raven. Thus, this scaling factor
46
47
48 made it possible to predict the corresponding payload capacity of Robo Raven from the wind
49
50 tunnel data. The corresponding results, seen in Table 4, clearly explain why the original 22 module
51
52
53 UAV did not fly given the predicted payload of -4 g, which was recovered to 21 g with the modified
54
55 wing design.
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 14 of 35

1
2
3
4
5
6 Robo 12 Module 22 Module Modified 22 Module
7 Raven Robo Raven III Robo Raven III Robo Raven III
8
Weight of UAV (g) 290 317 331 346
9
10 Force Magnitude (g) 234 235 232 260.1
11 Total Flight Weight (g) 330 332 327 367
12 Payload (g) 40 15 -4 21
13
14
Table 4. Weight and payload characteristics for each UAV design
15
16
3.2 3D Digital Image Correlation (DIC) Characterization of Wing Shape
17
18 Previously, Stanford et al. used 3D Digital Image Correlation (DIC) to study wing
19
20 deformations in fixed membrane wings for MAVs to optimize their design for aerodynamic forces
21
22
23 [37]. Therefore, 3D DIC was used to study the effects of deformation on the different
24
25 multifunctional wing designs by quantifying differences in shapes and strain and relating it to the
26
27 aerodynamic loads generated during flapping. For our 3D DIC investigation, two Flea3 FL3-FW-
28
29
30 03S1M cameras were used to acquire stereoscopic high speed images at 80 HZ while the wings
31
32 were flapping at 4 Hz. Speckle patterns were applied to the surface of the wings, and the software
33
34
package VIC-3D (Correlated Solutions, Inc) was used to obtain deformation measurements at 20
35
36
37 different angles during a single flapping cycle.
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Page 15 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34 Figure 8. Comparison of out-of-plane displacement (W) for each wing at the horizontal position
35 while flapping downward, which was found to be most representative of the relative
36 deformations between the wing designs for the 20 different wing positions that were measured
37
38 during a single flapping cycle. (top left) regular, (top right) 6 module, (bottom left) 11 module,
39 (bottom right) modified 11 module. Dashed lines indicate approximate location of the modules.
40
41 Representative DIC data associated with wing shape be seen in Figure 8, which was
42
43 obtained by taking the out-of-plane displacement (W) in the z-direction normal to the wing while
44
45
46
it is in a horizontal position while flapping downwards. For these measurements, the x-axis and
47
48 corresponding U displacements were chosen to run along the leading spar of the wings, while the
49
50 y-axis and V displacements runs along the body of the UAV. At the horizontal position, wings
51
52
53 are generating the most aerodynamic lift and exhibit the greatest deformation. It is clear that the
54
55 regular wing has greater deformation towards the trailing edge of the wing than the 6 module wing
56
57 or the 11 module wing, where the deformations are more indicative of bending on the leading edge.
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 16 of 35

1
2
3
These larger deformations are intuitive since the wings are not stiffened by the addition of solar
4
5
6 cells.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36 Figure 9. Comparison of time resolved residual thrust and shear strain: (top left) regular, (top
37
38
right) 6 module, (bottom left) 11 module, (bottom right) modified 11 module.
39 A comparison was also made between the four wings and the time-resolved resolved thrust
40
41
42 and aerodynamic lift loads versus the average shear strain and biaxial strain respectively of the
43
44 entire surface of the wing throughout the flapping cycle (Figures 9 and 10). The aerodynamic lift
45
46 and thrust correlate strongly with the biaxial strain and shear deformation from the DIC results.
47
48
49 The correlation between the lift and thrust forces to the biaxial and shear strains have been
50
51 measured using the correlation coefficient, and are shown in Table 5. Since we are interested in
52
53
the change in biaxial strain, the mean strain was subtracted from the strain observed throughout
54
55
56 the flapping cycle to compare the differences. The correlation of lift with biaxial strain is
57
58
59
60
Page 17 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
consistent with the aerodynamic shape change responsible for lift that is caused by out-of-plane
4
5
6 bending deformation and induces in-plane biaxial strain, while the correlation of thrust to shear
7
8 strain is consistent with the out-of-plane torsional deformation responsible for thrust that induces
9
10
11
in-plane shear strain.
12
13
As mentioned earlier, it is clear that the integration of solar cells has an effect on the wing
14
15
16 shape during the flapping cycle due to the increased stiffness of the solar cell material relative to
17
18 mylar. This in turn reduces the amount of thrust and lift as the solar cells are integrated into the
19
20
21
regular wing design. However, by increasing the wing area in the modified 11 module wing design,
22
23 residual thrust and lift could be recovered, although greater deformation was also observed that
24
25 could influence performance. The 17% increase in wing area provided an additional 7% of thrust
26
27
28 force and 4% of lift force.
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
Figure 10. Comparison of time resolved aerodynamic lift and biaxial strain relative to mean
56
57 biaxial strain for the entire wing: (top left) regular, (top right) 6 module, (bottom left) 11 module,
58 (bottom right) modified 11 module.
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 18 of 35

1
2 Robo 12 Module 22 Module Modified 22 Module
3
Raven Robo Raven III Robo Raven III Robo Raven III
4
5 Thrust/Shear Strain
6 Correlation Coefficient 0.45 0.43 0.65 0.66
7 Lift/Biaxial Strain
8 Correlation Coefficient 0.56 0.35 0.58 0.77
9 Table 5. Correlation coefficient of thrust and shear strain and correlation coefficient of
10 lift and biaxial strain for each wing design.
11
12 As the majority of the wing becomes covered in solar cells, the deformation of the wing
13
14 decreases. By observing the time resolved results from the 6 module and 11 module wings, it is
15
16
17 clear that the shear strain slightly decreases as solar cells are added. Where the 6 module wing
18
19 achieved a strain of 2% the 11 module wing remains under 2%. By increasing the wing size and
20
21
allowing for more deformation, a large increase in shear strain in the modified 11 module wing is
22
23
24 observed. These results are also mirrored in the cyclic results. The shear strain for the 6 module
25
26 wing and 11 module wings have a much lower value than the regular wings. However, the
27
28
29
modified 11 module wings have a much higher shear strain value. This increase in deformation is
30
31 the difference between the original 11 cell wing and the modified 11 cell wing. The increase in
32
33 compliance is what allows the modified 22 module UAV to maintain flight. The increase stiffness
34
35
36 and weight of the solar cells is counteracted by the increase in overall wing deformation.
37
38
39
4. Multifunctional Performance Modeling of Wings
40
41 A new model was developed to characterize the multifunctional performance of the wings
42
43
44 based on the aerodynamics of flapping wing UAVs. Let Ft be the thrust generated by the flapping
45
46 wings, V be the flight velocity due to Ft, and Fl be the aerodynamic lift at flight velocity V for the
47
48
49
baseline UAV. Let Mm be the mass of the baseline UAV, and Mb be the mass of battery on the
50
51 baseline UAV. To maximize the flight time, the largest possible battery permitted by the lift can
52
53 be used, leading to the following condition:
54
55
56 Fl = (Mm + Mb)g [1]
57
58 or
59
60
Page 19 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
Mb = Fl/g - Mm [2]
4
5
6
7 Let U be the energy capacity of battery on the baseline UAV. In general, U is proportional
8
9 to the mass of the battery Mb. So,
10
11
12
13
14 U = kbMb [3]
15
16
17 where, kb is battery coefficient. The flight time for this baseline configuration will therefore be:
18
19
20
21 T = U/P = kbMb/P = kb(Fl/g - Mm)/P [4]
22
23
24 Where P is the power consumed by the UAV during flight.
25
26
27 For multifunctional wings, Ms is the mass of solar cells in the wings. The solar cells have
28
29 the following effects:
30
31
32 • They are expected to alter the thrust due to stiffening of the wings. The changed thrust leads
33
34
35
to a different flight velocity, V’ , due to a relative change in drag force, k1 = V’/V= (F’t/Ft)0.5
36
37
38
• They also change the aerodynamic lift coefficient by a ratio, k2, resulting in a total aerodynamic
39
40 lift k1k2Fl.
41
42
43 Equation [1] becomes,
44
45
46 k1k2Fl = (Mm + Ms + M’b)g [5]
47
48
49 Where M’b is the mass of battery on the UAV with solar cell integrated wings. So,
50
51
52 M’b = k1k2Fl /g - Mm - Ms [6]
53
54
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 20 of 35

1
2
3
Flight time for the multifunctional wings will therefore be as follows when directly providing
4
5
6 power during flight:
7
8
9
10 T’ = kbM’b/ (P-ksMs). [7]
11
12
13
14 In this expression, the power produced by the solar cells is assumed proportional to the mass of
15
16 the solar cell, and ks is the solar coefficient influenced by factors such as the conversion efficiency
17
18
19
of the solar cell and the solar energy flux. Flight time for the multifunctional would be as follows
20
21 if the solar cells do not provide power during the flight (i.e., baseline flight time):
22
23
24 T’ = kbM’b/ P [8]
25
26
27 Using Equation [6], we can determine the flight time when the solar cells provide power as follows:
28
29
30
31
32 T’ = kb (k1k2Fl /g - Mm - Ms)/(P-ksMs). [9]
33
34
35 Provided the current output of the solar cells does not exceed the recharge limit of the battery, the
36
37
38
battery recharge time for an UAV with integrated solar cells can also be determined as follows:
39
40 Tr = fckb (k1k2Fl /g - Mm - Ms)/ksMs. [10]
41
42
43
44 Where fc is the fraction of the battery charge that was consumed before recharging.
45
46
47 For most situations, flight time is considered a system design problem, resulting in a
48
49 constraint on the minimum value of flight time,
50
51
52 Tflight ≥ Tmin [11]
53
54
55 If T ≥ Tmin, then there is no benefit in integrating solar cells into the wings such (i.e., T’= Tmin).
56
57
58 Therefore, the objective in that case would be to select Ms such that Tr is minimized without T’
59
60
Page 21 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
exceeding Tmin. Since there are complex interactions between the baseline wings and solar cells,
4
5
6 solar cells should be placed such that k1 and k2 are maximized in order to maximize T’.
7
8
9 Equation [7] enables the multifunctional criterion ks to be determined that allows for the
10
11 mass of the solar cells to generate the same amount of current as consumed by the UAV (i.e.,
12
13
infinite flight time), and can be considered a critical multifunctional criteria, ks*, under the
14
15
16 following condition,
17
18
19 ܲ [12]
20 ݇௦∗ =
‫ܯ‬௦
21
22
23 Therefore, a comparison between the multifunctional criterion for the solar cells, ks*, and the mass
24
25
26 of the UAV, Mm, can also be made for design purposes based on: (1) area of solar cell coverage,
27
28 and (2) power consumed by the UAV.
29
30
31 5. Experimental Results for Multifunctional Performance
32
33
34 Inputs for the multifunctional performance model were obtained from experimental results.
35
36 Battery energy storage is typically stated as Ampere-hour (A-hr). Hence for the power calculations,
37
38
39
energy storage is multiplied by the operating voltage of the battery and get power in terms of Watt-
40
41 hour (W-hr). The value kb is obtained by dividing battery capacity by battery mass, resulting in
42
43 106.54 W-hr/kg. To determine flight time, a fully charged battery was used to power the UAV
44
45
46 until depleted by flapping, resulting in 4.53 min. Using this in Equation [8], the average current
47
48 draw from the batteries was found to be 4.9 A. The UAV mass, battery mass, and solar cell module
49
50 mass were determined using a DigiWeigh model DWP-1001 scale with 0.1 g resolution, resulting
51
52
53 in 263.3 g for the UAV, 25.7 g for the battery, and 1.7 g per module. Therefore, the minimum total
54
55 mass without solar cells is 289 g, and the Fl is 2.83 N.
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 22 of 35

1
2
3
Due to the change in wing design for the modified 22 module UAV, the Mm increases to
4
5
6 278.3 g, resulting in a Fl of 2.98 N. The actual lift force generated by the UAV was previously
7
8 measured to be 3.23 N. To compare the aerodynamic lift changes caused by the integration of
9
10
11
solar cells, the raw lift forces were compared for all wing designs. Thus, the k2 was found to be
12
13 1.023 for the 12 solar cell UAV, 1.029 for the original 22 module UAV, and 1.194 for the modified
14
15 22 module UAV. Similarly, k1 had to be calculated for each of the new wing designs since a
16
17
18 change in thrust production was observed on the load cell. Using the thrust forces observed on the
19
20 load cell, k1 was 0.970 for the 12 solar cell UAV, 0.912 for the 22 module UAV, and 0.859 for the
21
22 modified 22 module UAV.
23
24
25 5.1 Direct Powering of Motors
26
27
28 To calculate T’, ks must be found first. Since ksMs gets subtracted from P, ksMs is equal to
29
30
31
the power being supplied by the solar panels, which was measured to be 4.10 W for the 12 module
32
33 UAV. Therefore, using the previously reported mass of the solar cells of 27 g, ks equals 0.152
34
35 W/g. This makes the new predicted maximum flight time 5.05 minutes. This represents an 11.5%
36
37
38 increase in operation time using the solar cell wings. Thus, the overall effect of using the solar cell
39
40 wings on flight time turns out to be positive despite the additional mass and rigidity that it adds to
41
42 the wings.
43
44
45 The calculation is slightly different for the 22 module UAV. Since the body of the UAV
46
47
48 was modified to accommodate the new wing design, a different flight time can be expected since
49
50 the weights of the UAVs are different. To lift the difference in weight, the servos must pull more
51
52 power from the battery, shortening the flight time. The new flight time for this wing was 4.32
53
54
55 minutes, which is consistent with the increased weight and area requiring more energy to power
56
57 the wings at the same frequency and amplitude. The average current draw was calculated to be
58
59
60
Page 23 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
5.14 A, which results in an average power consumption of 37 W. Next, the new ks for the modified
4
5
6 22 module UAV must be calculated. The solar cells add an additional 42 g to the UAV and were
7
8 found to generate 7.41 W, making ks a value of 0.176 W/g. Using these numbers, we can predict
9
10
11
the new maximum time of flight of the 22 module UAV using Equation [9]. The new time was
12
13 calculated to be 5.23 minutes. This is 12.5 seconds more than the 12 module UAV and a 15.4%
14
15 increase in operational time compared to the original Robo Raven design.
16
17
18 The prediction of the multifunctional performance model demonstrates the potential gain
19
20
21
from the solar cells. However, it does not take into account the power generation variations
22
23 introduced by flapping the UAV. While flapping we expect a deviation from perfect conditions
24
25 because the solar cells are constantly changing their position relative to the sun. We
26
27
28 experimentally measured the increase on the operation time by using both battery and solar cells
29
30 to power motors. The actual operation time for the 12 solar cell UAV increased by 10.2% (flight
31
32 time of 5.00 minutes) which was close to predicted. The same test was done for the modified 22
33
34
35 module UAV. The actual operational time for the 22 module UAV was 5.17 min, representing a
36
37 14.1% increase in operational time, which was also close to the prediction. Results for changes in
38
39
power and flight time are summarized in Table 6.
40
41
42 It is important to note that while increasing the modules from 12 to 22 on the UAV only
43
44
45 increased the operational time by 3.9%, the low increase was primarily due to the extra power
46
47 required for the new wing design that reduced some of the solar cell benefit. This further reinforces
48
49
the trade-off that is assessed when using more solar cells. Therefore, in addition to the time of
50
51
52 flight, we also determined the critical ks* according to eqn. [12] which would require improving
53
54 the solar cell production output instead of adding more solar cells or redesigning the wing. Given
55
56
57
Ms and P for the 12 module UAV, ks* equals 1.34 W/g. Given the value of 0.152 W/g for the
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 24 of 35

1
2
3
flexible solar cells used in this investigation, we are at only 11.3% of the value needed for infinite
4
5
6 flight. Thus, ~8.8X improvement is needed to reach infinite flight time. Doing the same
7
8 calculation for the Modified 22 module UAV, ks* equals 0.787 W/g. With a ks value of 0.176 W/g,
9
10
11
we are generating 22.4% of the power necessary for infinite flight. Only ~4.5X improvement is
12
13 needed for infinite flight. Since the efficiency of the flexible solar cells we are using is only 5%,
14
15 infinite flight time would only require increasing the efficiency to 22.5%, which would obviate the
16
17
18 need for batteries and render them a secondary power source.
19
20
21
22
23 Power Solar Power Predicted Increase Measured Increase
24 Consumption Generation in Flight Time in Flight Time
25
26
27 W W Time % Time %
28
(sec) (sec)
29
30 Regular UAV 35.8 n/a n/a n/a n/a n/a
31
32 12 module UAV 36.4 4.10 29.4 10.8 27.6 10.2
33
34 22 module UAV 36.8 7.41 dnf dnf dnf dnf
35
22 module UAV 37.0 7.41 41.9 15.4 38.1 14.1
36
37 (modified wings)
38 Table 6. Comparison of predicted and measured flight time for regular and multifunctional
39
40
wings (dnf denotes “did not fly”).
41
42
43
44 5.2 Recharging of Batteries
45
46
47 To determine recharge performance to compare with the multifunctional performance
48
49 model, the UAV was placed in sunlight, and measurements were taken to see how long it would
50
51 take to completely recharge a depleted battery. The 2 cell Lithium Polymer battery is completely
52
53
54 recharged when it reads 8.4V. The results for both the 12 solar cell and 22 module UAV are shown
55
56 in Figure 11. It took 149 minutes for the 12 solar cell UAV to completely recharge the battery and
57
58
59
60
Page 25 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
90 minutes for the 22 module UAV. These results are compared to the fastest theoretical results
4
5
6 in Table 7 below. Differences can be attributed to the passive recharge circuit that minimized
7
8 weight and power consumption. A more active circuit using maximum power point tracking could
9
10
11
optimize recharge time, but at the expense of adding more weight and increasing the power
12
13 requirements for the UAV.
14
15
16
17 8.5
18
19 8.3
20 8.1
21
22 7.9
23
24 7.7
Voltage

25
26 7.5
27 7.3
28
29 7.1
30
31 6.9 22 Cells
32
33
6.7 12 Cells
34 6.5
35 0 50 100 150
36
37 Time (min)
38
39 Figure 11. Recharging time profiles for the 12 and 22 module UAVs.
40
41
42
43 Theoretical Fastest Actual Recharging Time
44 Recharging Time (Min) (Min)
45
46 12 module UAV 74.0 149
47
48 22 module UAV 40.4 90
49
50 Table 7. Comparison of theoretical charging to actual charging results.
51
52
53
54
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 26 of 35

1
2
3
5.3 Impact of Aerodynamic Forces of Power Output Solar Cells for Sensing
4
5
6 The time-resolved power generated by the multifunctional wings was determined by
7
8
9 measuring the voltage and current during flapping. In Figure 12a, results for the 22 module UAV,
10
11 generating an average of 7.42 W, are compared with the 12 module UAV, generating 4.10 W. The
12
13
power output was expected to change sinusoidally while flapping, due to its position relative to
14
15
16 the sun. However, the response is not sinusoidal, which indicates the deformation of the wings
17
18 must have an effect on power output as well. In Figures 12b and 12c, it was found that changes
19
20
21
in the power output appear to directly correlate with the thrust generation of the wing, as indicated
22
23 by the correlation coefficients in Table 8. Since it was previously determined that the thrust
24
25 correlates with in-plane shear strain attributed to the out-of-plane torsional deformations of the
26
27
28 wing responsible for the thrust, it is postulated that the change in power output is due to out-of-
29
30 plane rotation associated with torsional deformations that can reorient cells to increase or decrease
31
32 their angle of incidence relative to the sun. The expected relationship is also verified by directly
33
34
35 comparing the shear strain with the change in power output for each wing design, as seen in Figure
36
37 13 along with their correlation coefficients that have been added to Table 8. Therefore, the solar
38
39
cells wings are not only multifunctional in being able to harvest solar energy and serve as skin to
40
41
42 generate aerodynamic force during flapping, but it can also be used to sense those forces.
43
44
45 The implications of sensing thrust using solar cells has many applications going forward.
46
47 Gusts of winds can be detected during flights using the same structure that is used to help power
48
49
the UAV. Thus, this information can be used to change the flapping profile in reaction to the
50
51
52 changes in aerodynamic loads. These changes can be potentially automated to allow for correction
53
54 while it is being piloted or flown autonomously, in which case the wings would be used as smart
55
56
57
structures. Because the solar cells are being used to achieve longer flights, being able to adjust to
58
59
60
Page 27 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
flight conditions using solar cell sensing can be a very powerful new tool for further increasing
4
5
6 flight time.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31 (a)
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54 (b)
55
56
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 28 of 35

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28 (c)
29
30
31 Figure 12. (a) Electrical power generated by the 12 and 22 module UAVs while flapping,
32 and thrust versus % change in power output for (b) 22 module and (c) 12 module UAVs
33 indicating that the solar cells can be used to sense aerodynamic forces
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57 (a)
58
59
60
Page 29 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
4 0.035 35
5
6 Shear Strain Power Output
7 0.025 25

% Change in Power Output

8
9 0.015 15
10 Shear Strain
11
12 0.005 5
13
14
-0.005 -5
15
16
17 -0.015 -15
18
19
20 -0.025 -25
21
22 -0.035 -35
23
0.5 0.7 0.9 1.1 1.3 1.5
24
25 Time (sec)
26
27 (b)
28
29 Figure 13. Shear strain versus % change in power output for (a) 22 module and (b) 12
30 module UAVs directly verifying the expected relationship.
31
32
33 12 Module UAV 22 Module UAV
34
35 Thrust/% Change in Power
36
37 Output Correlation 0.87 0.79
38 Coefficient
39 Shear Strain/% Change in
40 Power Output Correlation 0.69 0.67
41 Coefficient
42
43
44 Table 8. Correlation coefficient of the thrust and % change in power output and of the shar strain
45
and % change in power output for the 12 module and 22 module UAVs.
46
47
48
49 8. Conclusions
50
51
52 This paper investigates the mechanics that affect the potential benefits of introducing
53
54 multifunctional structures to harvest solar energy on the wings of a flapping wing UAV. We used
55
56
Robo Raven as our base platform for this investigation. Three different wing designs were initially
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 30 of 35

1
2
3
observed: (1) the regular wings without solar cells, (2) wings with 6 solar cell modules that
4
5
6 constitute a 12 module UAV, and (3) wings with 11 solar cell modules that constitute a 22 module
7
8 UAV. Immediately, a deterioration in flight performance was observed when solar cells were
9
10
11
added. The 12 module UAV was still able to fly, but the wings of the 22 module UAV were too
12
13 stiff and heavy to generate enough aerodynamic lift, resulting in a calculated negative payload
14
15 capacity. Knowing that deformation plays a major role in force production, the wing design for
16
17
18 the 11 module wings was altered, and a modified wing design was developed for the 22 module
19
20 UAV to recover aerodynamic forces.
21
22
23 With these four wing designs, the force production of each wing design was compared to
24
25 understand the mechanical effects the solar cells had on the aerodynamic performance of the wing.
26
27
28 Therefore, the lift and thrust forces generated by each wing design were quantified. Next, the
29
30 deformation of each wing surface was quantified while flapping using 3D DIC to determine the
31
32 specific effects of the mechanical properties of the solar cells as the wing design was varied. There
33
34
35 was a clear correlation between the measured DIC deformations and the aerodynamic forces, in
36
37 particular the correlation biaxial strain associated with the change in aerodynamic shape to the lift
38
39
force and the correlation of shear strain associated with torsional deformation to the thrust force
40
41
42 responsible for providing the forward velocity necessary to generate the aerodynamic lift that
43
44 enables the UAV to achieve flight. It was also found that changes in power output directly
45
46
47
correlated to the thrust, indicating that the multifunctionality of the solar cell wings was not limited
48
49 to just harvesting solar energy and serving as skin to generate aerodynamic forces, but that they
50
51 could also be used to sense those forces. Since thrust correlated to shear strain associated with
52
53
54 torsional deformation of the wings, it was postulated that these deformations induce out-of-plane
55
56
57
58
59
60
Page 31 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
rotations that increase or decrease the angle of incidence of the cells relative to the sun causing the
4
5
6 change in power output.
7
8
9 Next, the electrical benefits of adding solar cells were determined. A charging circuit was
10
11 developed so that the UAV can be charged by the solar cells. The solar cells were also directly
12
13
integrated to the electrical power of the UAV to extend the operating time of the UAV. A
14
15
16 performance model was developed to model the change in operating time due to the integration of
17
18 solar cells. It was determined that it takes 149 minutes to completely recharge the battery with 12
19
20
21
solar cells and 90 minutes to recharge with 22 modules. Theoretically we could have a maximum
22
23 13.9% increase in flight time with the 12 module UAV and a 21.2% increase for the 22 module
24
25 UAV. Unfortunately the model does not take into account the flapping motion of the solar cells
26
27
28 nor the heat of running these tests outdoors. We observed a 10.2% and 18.7% increase in
29
30 operational time for the 12 module and 22 module UAVs respectively. The current solar cells
31
32 have an efficiency of 5%; however, with recent developments in flexible PV solar cell technology
33
34
35 provide more efficient solar cells can be integrated to provide for a longer and maybe infinite flight
36
37 time. We found we would need a flexible PV solar cell that is 22.5% efficient to achieve infinite
38
39
flight, and there are flexible PV solar cells that are more than 24% efficient commercially
40
41
42 available.
43
44
45 In summary, the fundamental mechanics of adding solar cells to flapping wing UAVs has
46
47 been established. The added stiffness of the solar cells prevents the wings from deforming resulting
48
49
in a loss in aerodynamic force generation. The shear and biaxial strains on the surface on the wing
50
51
52 were directly correlated to the thrust and lift generated by the wings. This new information is
53
54 utilized in developing a multifunctional performance model to predict the effects on the flight time
55
56
57
of the UAV. Although the stiffness and weight of the solar cells were quantified and determined
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 32 of 35

1
2
3
to have a negative effect on the aerodynamic force production, this was offset by the power
4
5
6 generation of the solar cells to result in significant gains in the flight time. From this model, further
7
8 addition of solar cells to larger wings, the body, or the tail of a UAV can be expected to result in
9
10
11
increased flight time. However, the greatest benefit to flight time is expected to result from higher
12
13 efficiency flexible solar cells, such as thin film GaAs multi-junction solar cells, where a 4.5X
14
15 improvement in efficiency will theoretically result in infinite flight time and obviate the need for
16
17
18 batteries other than as a secondary power source. Finally, it was determined that multifunctional
19
20 solar cell wings may be capable of three functions: (1) lightweight flexible structure to generate
21
22 aerodynamic forces, (2) energy harvesting to extend operational time, and (3) sensing of an
23
24
25 aerodynamic force associated with wing deformation.
26
27
28
29
30
31
Acknowledgments
32
33 This work was sponsored by AFOSR grant FA9550-12-1-0158 with Dr. Byung-Lip “Les” Lee
34
35
36 program manager. Opinions expressed are those of the authors and do not necessarily reflect
37
38 opinions of the sponsors.
39
40
41
42
43 References
44 [1] Gerdes JW, Gupta SK and Wilkerson S (2012) A Review of Bird-Inspired Flapping Wing
45
46 Miniature Air Vehicle Designs. ASME Journal of Mechanism and Robotics.
47 [2] Kumar V and Michael N, (2012) Opportunities and Challenges with Autonomous Micro
48
Aerial Vehicles. The International Journal of Robotics Research.
49
50 [3] Pines DJ and Bohorquez F (2006) Challenges Facing Future Micro-Air-Vehicle
51 Development. Journal of Aircraft.
52
53 [4] Sane, SP and Dickinson, MH (2002) The Aerodynamic Effects of Wing Rotation and a
54 Revised Quasi-Steady Model of Flapping Flight. Journal of Experimental Biology.
55
56
57
58
59
60
Page 33 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
[5] de Croon GCHE, de Clerq KME, Ruijsink R, Remes B and de Wagter C (2009) Design,
4
5 Aerodynamics, and Vision-Based Control of the Delfly. International Journal of Micro Air
6 Vehicles.
7
8
[6] Muijres FT, Johansson LC, Barfield R, Wolf M, Spedding GR and Hedenstrom A (2008)
9 Leading-Edge Vortex Improves Lift in Slow-Flying Bats. Science. 319: 1250–1253.
10 [7] Zhao L, Huang Q, Deng X and Sane S (2009) Aerodynamic Effects of Flexibility in Flapping
11
12 Wings. Interface.
13 [8] Arabagi V, Hines L and Sitti M (2012) Design and Manufacturing of a Controllable
14
Miniature Flapping Wing Robotic Platform. The International Journal of Robotics Research.
15
16 [9] Mahjoubi H and Byl K (2013) Trajectory Tracking in the Sagittal Plane: Decoupled
17 Lift/Thrust Control via Tunable Impedance Approach in Flapping-Wing MAVs. American
18
19
Control Conference (ACC).
20 [10] Keennon M, et al. (2012) Development of the Nano Hummingbird: A Tailless Flapping Wing
21 Micro Air Vehicle. presented at 50th AIAA Aerospace Sciences Meeting. Nashville,
22
23 Tennessee.
24 [11] Pornsin-Sirirak T, Tai Y, Ho C and Keennon M (2001) Microbat: A Palm-Sized Electrically
25
Powered Ornithopter. Proceedings of the NASA/JPL Workshop on Biomorphic Robotics.
26
27 Pasadena, CA.
28 [12] Mueller TJ (2001) Fixed and Flapping Wing Aerodynamics for Micro Air Vehicle
29
30
Applications. American Institute of Aeronautics and Astronautics. Reston, VA.
31 [13] Yang LJ, Hsu CK, Ho JY and Feng CK (2007) Flapping Wings with PVDF Sensors to
32 Modify the Aerodynamic Forces of a Micro Aerial Vehicle. Sensors and Actuators A:
33
34 Physical.
35 [14] Hsu CK, Ho JY, Feng GH, Shih HM and Yang LJ (2006) A Flapping MAV with PVDF-
36
Parylene Composite Skin. Proceedings of the Asia-Pacific Conference of Transducers and
37
38 Micro-Nano Technology.
39 [15] Tsai BJ and Fu YC (2009) Design and Aerodynamic Analysis of a Flapping-Wing Micro
40
41
Aerial Vehicle. Aerospace Science and Technology.
42 [16] Hsu CK, Evans J, Vytla S and Huang P (2010) Development of Flapping Wing Micro Air
43 Vehicles - Design, CFD, Experiment and Actual Flight. 48th AIAA Aerospace Sciences
44
45 Meeting. Orlando, FL.
46 [17] Cox A, Monopoli D, Cveticanin D, Goldfarb M and Garcia E (2002) The Development of
47
Elastodynamic Components for Piezoelectrically Actuated Flapping Micro-Air Vehicles.
48
49 Journal of Intelligent Material Systems and Structures.
50 [18] Yan J, Wood RJ, Avadhanula S, Sitti M and Fearing RS (2001) Towards Flapping Wing
51
52
Control for a Micromechanical Flying Insect. Proceedings ICRA. IEEE International
53 Conference on Robotics and Automation.
54 [19] Fenelon MAA and Furukawa T (2009) Design of an Active Flapping Wing Mechanism and
55
56 a Micro Aerial Vehicle Using a Rotary Actuator. Mechanism and Machine Theory.
57
58
59
60
 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3 Page 34 of 35

1
2
3
[20] Jones KD, Bradshaw CJ, Papadopoulos J and Platzer MF (2004) Improved Performance and
4
5 Control of Flapping-Wing Propelled Micro Air Vehicles. Proceedings of the AIAA 42nd
6 Aerospace Sciences Meeting and Exhibit. Reno, NV.
7
8
[21] Zdunich P, Bilyk D, MacMaster M, Loewen D, DeLaurier J, Kornbluh R, Low T, Stanford
9 S and Holeman D (2007) Development and testing of the mentor flapping-wing micro air
10 vehicle. Journal of Aircraft.
11
12 [22] Madangopal R, Khan Z and Agrawal S (2005) Biologically Inspired Design of Small
13 Flapping Wing Bird Vehicles Using Four-Bar Mechanisms and Quasi-Steady
14 Aerodynamics. Journal of Mechanical Design.
15
16 [23] Bejgerowski W, Ananthanarayanan A, Mueller D and Gupta SK (2009) Integrated Product
17 and Process Design for a Flapping Wing Drive-Mechanism. Journal of Mechanical Design.
18
19
[24] Mueller D, Gerdes JW and Gupta SK (2009) Incorporation of Passive Wing Folding in
20 Flapping Wing Miniature Air Vehicles. ASME Mechanism and Robotics Conference. San
21 Diego, CA.
22
23 [25] Bejgerowski W, Gupta SK and Bruck HA (2009) A Systematic Approach for Designing
24 Multifunctional Thermally Conducting Polymer Structures with Embedded Actuators.
25 Journal of Mechanical Design.
26
27 [26] Gerdes J, Holness A, Perez-Rosado A, Roberts L., Greisinger A, Barnett E, Kempny J,
28 Lingam D, Yeh C-H, Bruck HA, and Gupta SK (2014) Robo Raven: A Flapping-Wing Air
29 Vehicle with Highly Compliant and Independently Controlled Wings. Soft Robotics.
30
31 [27] Perez-Rosado A, Griesinger AJG, Bruck HA, and Gupta SK (2014) Performance
32 Characterization of Multifunctional Wings with Integrated Solar Cells for Miniature Air
33 Vehicles. ASME 2014 International Design Engineering Technical and Computers and
34
35
Information in Engineering Conference, Buffalo, NY.
36 [28] Nemat-Nasser S, Plaistead T, Starr A and Amirkhizi A (2005) Multifunctional Materials.
37
38
Biomimetics: Biologically Inspired Technologies, Ed. Y. Bar-Cohen. CRC Press.
39 [29] Thomas JP and Qidwai MA (2005) The Design and Application of Multifunctional Structure-
40 Battery Materials Systems. JOM.
41
42 [30] Ma KY, Chirarattananon P, Fuller SB and Wood RJ (2013) Controlled Flight of a
43 Biologically Inspired, Insect-Scale Robot. Science.
44
45 [31] Thomas JP et al (2005) Multifunctional Structure-Plus-Power Concepts. AIAA.
46
47 [32] Wissman J, Perez-Rosado A, Edgerton A, Levi BM, Karakas ZN, Kujawski M, Phillips A,
48 Papavizas N, Fallon D, Bruck HA and Smela E (2013) New Compliant Strain Gauges for
49 Self-Sensing Dynamic Deformation of Flapping Wings on Miniature Air Vehicles. Smart
50 Materials and Structures.
51
52 [33] Mueller D, Bruck HA and Gupta SK (2010) Measurement of Thrust and Lift Forces
53 Associated With Drag of Compliant Flapping Wing Air Micro Air Vehicles Using a New
54 Test Stand Design. Experimental Mechanics.
55
56
57
58
59
60
Page 35 of 35 CONFIDENTIAL - AUTHOR SUBMITTED MANUSCRIPT SMS-101270.R3

1
2
3
[34] Gerdes JW, Cellon KC, Bruck HA and Gupta SK (2013) Characterization of the Mechanics
4
5 of Compliant Wing Designs for Flapping-wing Miniature Air Vehicles. Experimental
6 Mechanics.
7
8
[35] Bauhuis GJ, Mulder P, Haverkamp EJ, Huijben JCCM, Schermer JJ (2009) 26.1% thin-film
9 GaAs solar cell using epitaxial lift-off. Solar Energy Materials and Solar Cells.
10 [36] Lin Q, Huang H, Jin Y, Fu H, Chang P, Li D, Yao Y, Fan Z (2014) Flexible Photovoltaic
11
12 technologies. Journal of Materials Chemistry C.
13 [37] Stanford, B, Ifju, P, Albertani, R, & Shyy, W (2008) Fixed membrane wings for micro air
14
vehicles: Experimental characterization, numerical modeling, and tailoring. Progress in
15
16 Aerospace Sciences.
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60