3 Rimfire .55 4260 480 kv Outrunner Brushless DC Motor. B. Castle Creations Electronic Speed Controller. A Castle Creations Phoenix Edge HV 120 electronic ...
Performance Testing and Modeling of a Brushless DC Motor, Electronic Speed Controller and Propeller for a Small UAV Andrew Gong∗ , Rens MacNeill† and Dries Verstraete‡ The University of Sydney, Sydney, NSW, 2006, Australia Small UAVs predominantly use electric power and are almost exclusively driven by brushless DC motors and electronic speed controllers. The efficiency of these components varies depending on the input voltage as well as the physical load required to turn the propeller for the desired thrust. This paper presents the performance evaluation and testing of a motor, electronic speed controller, and propeller used in a UAV propulsion system. Models of the performance of each component are generated to allow prediction of overall performance under any expected operating conditions.
I. Nomenclature AC BLDC DC ESC PWM RC UAV
= Alternating Current = Brushless Direct Current = Direct Current = Electronic Speed Controller = Pulse Width Modulated = Remote Control = Unmanned Aerial Vehicle
II. Introduction he use of electric unmanned aerial vehicles (UAVs) is increasing in a range of defence and civilian applications [1, 2]. Electric propulsion enables the use of low cost and quiet components including electric motors and electronic speed controllers, in addition to novel energy systems such as hybrid and fuel-cell power [2–8]. For most electric UAVs, a brushless DC (BLDC) motor and electronic speed controller (ESC) is used. In these systems, the BLDC motor is essentially a permanent magnet AC motor with the ESC used to electronically convert the DC power from a battery (or other power source) to three-phase electricity. The ESC usually generates a square wave or trapezoidal PWM signal for the motor, and both the BLDC motor and ESC are typically derived from hobby equipment with considerable variation in efficiency and performance between different manufacturers and models. Furthermore, there is limited data on the performance of motors and ESCs, and this makes it difficult to predict the overall performance of the motor and ESC combination for a given application [9]. This work builds on the earlier work by the authors in testing and modeling ESCs [10]. A motor dynamometer has now been coupled with the ESC testing equipment to enable evaluation of both the motor and ESC concurrently at a user selectable range of rotational speeds and torque values. The maximum current limitation on the ESC testing has also been increased from 20 A to 150 A to enable testing of higher power motors to cover the vast majority of motors used in all electric powered UAVs. First, an overview of the experimental set-up is given. Results for a BLDC motor combined with an ESC are given along with performance measurements of a typical propeller matched to this motor. Finally, the combined performance characteristics of this motor, ESC, and propeller combination are provided when used as the electric powertrain of a fuel-cell powered UAV [11].
T
∗ Doctoral Candidate, School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney, NSW, Australia, 2006, AIAA Student Member † Doctoral Candidate, School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney, NSW, Australia, 2006, AIAA Student Member ‡ Senior Lecturer, School of Aerospace, Mechanical and Mechatronic Engineering, The University of Sydney, NSW, Australia, 2006, AIAA Senior Member
1
III. Hardware and Experimental Set-Up The hardware set-up used here is based on the ESC testing work previously outlined by the authors [10]. The earlier setup consisted of the Magtrol 6530 3-phase power analyser between the ESC and the motor, with a propeller used to provide the mechanical load on the motor. For this work, additional equipment has been integrated into this system so that the rotational speed and torque load on the motor (and by extension the ESC) can be freely controlled. A Magtrol HD 715-NA hysteresis dynamometer is connected to the motor, able to measure up to 6.2 N·m and 25,000 rpm with a maximum power rating of 3.4 kW. The dynamometer is linked to a DSP7000 high speed dynamometer controller to enable control of the rotational speed and load (torque), as well as to provide logging. The dynamometer and controller can be seen in Figure 1.
(a) DSP7000 Dynamometer Controller
(b) HP 715-8NA Dynamometer
Fig. 1
Test Equipment
The electrical power measurement is augmented by the use of high bandwidth high current Hioki 3274 current probes on each of the three electrical phases to enable an increase in the maximum current from 20 to 150 A. The signal from the Hioki probes is fed into the Magtrol 6530 3-phase power analyser. The probes have a bandwidth of 10 MHz for an amplitude accuracy of ±1.0%. An averaging mode in the Magtrol 6530 is used to smooth out the data at each of the test points and to calculate electrical power between the ESC and motor so that motor and ESC efficiency can be measured individually. The dynamometer is integrated with the Magtrol 3-phase power analyser to enable simultaneous measurement of electronic speed controller (ESC) and motor performance for a range of electrical input and mechanical output conditions, including input voltage and output torque loadings. A diagram of the power analyser and motor dynamometer set-up can be found in Figure 2.
3-Phase Power Analyzer
Motor Dynamometer
A1 V1
ADC
Power Supply
VDC
ESC
V2
Motor
A2
Brake
Torque Sensor RPM Sensor
V3 A3
Fig. 2
Diagram of Test Set-up
The following subsection presents the motor and ESC tested and modeled in this paper. Both components are RC hobby grade and representative of consumer grade model aircraft parts. 2
A. Rimfire .55 480 kV Motor A Rimfire .55 480 kV motor is evaluated in this paper. This motor weights 270 g, with a peak power rating of 1820 W and a continous power rating of 830 W. The motor is 42 mm in diameter with a length of 60mm. A picture of the motor can be found in Figure 3.
Fig. 3
Rimfire .55 4260 480 kv Outrunner Brushless DC Motor
B. Castle Creations Electronic Speed Controller A Castle Creations Phoenix Edge HV 120 electronic speed controller (ESC) is used to drive and control the brushless DC motors (Figure 4). This ESC weighs 150 g and can handle up to 120 A of current at a maximum input voltage of 50 V. A PWM switching frequency of 12 kHz is selected with a motor timing of 0◦ for maximum efficiency.
Fig. 4
Castle Edge Phoenix 120 HV Electronic Speed Controller
IV. Test Results A. Motor Motor efficiency is determined using the dynamometer outlined above. With the motor and electronic speed controller at full throttle, a target torque value is chosen, and the dynamometer brake increased until the torque value is achieved. The throttle is then decreased, yielding a decrease in rotation speed, and the brake increased until the torque value is again reached. This is continued until the lower rotational speed threshold is reached, and repeated for a range of torque values to generate the required motor data. Motor performance maps are measured at a range of rotational speeds, torque values, and input voltages. Input voltage variations are required as the full throttle motor torque and rotational speed is dependent on the input voltage. 3
These parameters are determined to adequately define the motor efficiency, given by the ratio of the input electrical power to the output mechanical power. The brushless DC motor operates with three phases in a similar manner to an AC motor, so the motor efficiency is defined as: η=
τ·ω τ·ω = 3 3 P Σn=1 Σ n n=1 An · Vn
(1)
where τ is the motor torque (Nm), ω is the rotational speed (rad/s), Pn is the input electrical power calculated between two of the three phases using voltage and current sensors as configured in Figure 2. The efficiency of the motor as a function of rotational speed and torque can be found. The measured motor efficiency at voltages from 13–18 V is presented in Figure 5. This range is the expected operating voltage range of the fuel-cell system that powers the UAV [11, 12]. The Rimfire motor is tested at rotational speed values from 2500 to 6500 RPM and torque values from 0.1 to 0.5 N·m. Maximum rotational speed is influenced by the input voltage, and at lower voltages the motor is not able to reach 6500 RPM. Furthermore, as the torque load on the motor increases, the maximum rotational speed decreases. This limitation is visible as the white space at the top right hand side of each subfigure, and is particularly noticeable at 13 and 15 V. As the voltage increases, this area decreases and by 17 V this region has shifted above 6000 RPM and is above the rotational speed limits tested here. For all input voltages, there is a distinct efficiency peak at high rotational speeds and moderate torque values. Torque at peak efficiency is typically in the 0.2–0.4 N·m range and at maximum rotational speed, with a peak efficiency of 0.81–0.84. At low torque values, the efficiency drops rapidly while the efficiency also gradually declines as torque is increased above 0.4 N·m. For all input voltages, efficiency decreases at low rotational speeds whilst maximum efficiency occurs at higher rotational speeds for high input voltages.
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 5
Rimfire .55 Motor Measured Efficiency
4
B. Electronic Speed Controller The performance of the ESC with the Rimfire .55 motor is measured at input voltages from 13–18 V (Figure 6). Efficiency is determined as the ratio of the output AC electrical power to the input DC electrical power. The electronic speed controller is directly controlled by the throttle through a PWM output, and input DC current is proportional to the torque load on the motor. The influence of the throttle setting (PWM) and input current on the efficiency of the ESC is presented in Figure 6. The ESC maintains a high efficiency with efficiency above 90% for most operating conditions. There is a general trend of increased efficiency at low torque values, with lower current corresponding to reduced resistive losses. Peak efficiency is measured at approximately 98%, dropping to 85% at high current values. The operating conditions are bounded at low currents by the minimum current required to continue spinning the motor at no load, and at high currents by the maximum torque load applied to the motor. At higher voltages such as 18 V, the upper throttle settings are not used to limit the maximum rotational speed of the motor.
Fig. 6
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Castle Phoenix Edge HV 120 ESC Efficiency with Rimfire .55 Motor
V. Modeling of motor performance A 3 constant equivalent circuit model is chosen to here represent the performance of the brushless DC motor. There are many models available including the 4 constant model [13] and a polynomial fit [14], however the 3 constant model provides a quick and relatively accurate representation of motor performance [15]. Motor manufacturers often also supply these numbers to provide a comparison with test results. The 3 constant motor model uses the standard DC motor model, with the motor modeled using three parameters, kV , Rm and I0 [15]. The circuit diagram of the 3 constant motor model is given in Figure 7. The motor current I is given by the sum of the no load current I0 and the current required to generate the torque
5
Fig. 7
3 Constant Motor Model
I = I0 + τ · kT 2π · kV τ = I0 + 60
(2)
where τ is the torque required, kT is the torque constant and kV is the voltage constant in RPM/V. The input voltage is equal to the sum of the back emf Ei and the voltage loss due to the internal resistance of the motor Rm . The back emf can be calculated as Ei =
ω , kV
(3)
yielding the input voltage E as E = Ei + I · Rm ω = + I · Rm kV
(4)
where ω is the rotational velocity. The input and output power of the motors can then be calculated as Pin = E · I
(5)
Pout = τ · ω
(6)
One issue with the standard 3 constant motor model is it assumes kV , Rm and I0 remain constant, however these ‘constants’ actually vary with voltage [15]. Thus, these values are evaluated at a range of different voltages to assess if this assumption is sufficiently accurate. The speed constant kV and no load current I0 can be determined from measurements of the motor under no load. Figure 8 shows the rotational speed and current for a range of input voltages. The no load current I0 shows a near-linear dependency with voltage and can be represented by the following equation: I0 = 0.5110 + 0.0636 · E
(7)
The speed constant kV appears to be constant and an average of 482 is used to model the motor. This compares well with the manufacturer specified kV of 480. With the kV and I0 values estimated, the test data is then fitted into the model using a non-linear least square approach. This generates an estimate for Rm for each voltage tested. The variation of Rm with voltage is shown in Figure 9, and there is only a slight variation with voltage. Therefore a value of 0.08 is selected for Rm for modelling. The values used in the model for each voltage are given in Table 1. 6
2.5
10000
No Load Current (A)
RPM
15000
5000 0
0
5
10
15
20
25
kV (RPM/V)
Voltage (V) 490 485
2
1.5
1
480 475
0
5
10
15
20
0.5
25
0
5
Voltage (V) (a) kV Variation with Voltage for Rimfire .55 Motor
Fig. 8
10
15
20
25
Voltage (V) (b) I0 Variation with Voltage for Rimfire .55 Motor
Rimfire .55 Zero Load Characteristics
0.084 0.082
Rm ( )
0.08 0.078 0.076 0.074 0.072 13
14
15
16
17
18
Voltage (V)
Fig. 9 Table 1
Rm Variation with Voltage for Rimfire .55 Motor
Rimfire .55 3 Constant Motor Values from No Load Testing E 13 14 15 16 17 18
kV 482 482 482 482 482 482
I0 1.34 1.40 1.47 1.53 1.59 1.66
Rm 0.08 0.08 0.08 0.08 0.08 0.08
R2 0.8457 0.9227 0.6574 0.9058 -2.099 0.9556
The results of the electric motor modelling are shown in Figure 10, with the difference between the motor testing and modelling (δ) shown in Figure 11. The difference δ is defined as δ = ηmodel − ηactual
(8)
where ηmodel is the predicted efficiency from the model and ηactual is the actual efficiency. The motor model shows maximum absolute δ values below 0.02 for all voltages except 17 V, and within 0.05 for 17 V.
7
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 10
Rimfire .55 Motor Model Efficiency
VI. Modeling of Electronic Speed Controller To model the performance of the ESC, a bi-linear equation was fitted to the efficiency data as a function of throttle setting (in µs) and current. This model allows the prediction of the efficiency given the demands of the motor and the throttle commanded. The efficiency is modeled using the equation η = a · δT + b · I + c
(9)
where δT is the throttle setting, I the input current to the ESC, and parameters a, b, and c are model parameters. The model parameters determined for each voltage is given in Table 2. Table 2
Castle Phoenix Edge HV 120 ESC Model Parameters
Voltage 13 14 15 16 17 18
a 1.244e-4 1.253e-4 1.355e-4 1.434e-4 1.425e-4 1.729e-4
b -0.006204 -0.006820 -0.005618 -0.006497 -0.006327 -0.006345
c 0.7714 0.7907 0.7560 0.7666 0.7815 0.7260
R2 0.9774 0.9439 0.9457 0.9542 0.9660 0.8816
To make this model applicable for all voltage ranges, the values of a, b and c are determined as functions of voltage (Figure 12). The value for b is relatively independent of voltage and chosen as 0.0063. Values for a and c are made 8
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 11
Rimfire .55 Motor Model and Test Results δ
linear functions of input voltage. This yields the values for a, b, c, and R2 seen in Table 3. The linear fits are shown by the orange lines in Figure 12. Table 3 V 13 14 15 16 17 18
a 1.191e-4 1.277e-4 1.364e-4 1.450e-4 1.536e-4 1.622e-4
ESC Model Fit
b -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3
c 0.7824 0.7754 0.7684 0.7614 0.7544 0.7474
R2 0.9756 0.9172 0.8728 0.9519 0.8827 0.856
Using this model generates the efficiency predictions for the ESC at each of the input voltages (Figure 13). There is slight variation between each of the input voltages, with a minimum efficiency of 75% predicted at low throttle settings (PWM) and high current (corresponding to low motor speed and high torque). At 17 and 18 V, high throttle settings (PWM) and low currents overpredict the efficiency. However, this is not a feasible operating condition due to the current drawn by the motor at no load and thus can be ignored. To compare the accuracy of the ESC model with the measured results, the difference of the ESC testing and modeling is given in Figure 14. The difference δ for the ESC is given by
9
1.8
10-4 -5.6 -5.8
1.6
-6
1.5
-6.2
b
a
1.7
10-3
1.4
-6.4
1.3
-6.6
1.2
-6.8
1.1 13
14
15
16
17
18
-7 13
14
15
Voltage (V)
16
17
18
Voltage (V)
(a) ESC ‘a’ Constant
(b) ESC ‘b’ Constant
0.8
c
0.78
0.76
0.74
0.72 13
14
15
16
17
18
Voltage (V) (c) ESC ‘c’ Constant
Fig. 12
ESC Voltage Constants
δ = ηmodel − ηactual
(10)
where ηmodel is the predicted efficiency from the ESC model and ηactual is the measured ESC efficiency. Maximum δ of 0.02 is observed at 14 V and 18 V, corresponding to an absolute error in the efficiency model of 2%. The throttle commands can be re-based to a more logical 0–1 (0–100 %) throttle range. This is useful for the hardware-in-the-loop simulation [11] as the autopilot outputs throttle commands from 0 to 100 % throttle, and yields the following equations for each of the model parameters. a = 0.0086 · V + 0.0069
(11)
c = 0.00163 · V + 0.8803
(12)
A Table showing the re-based model parameters and the corresponding R2 value is given in Table 4.
VII. Propeller Performance With the motor and ESC performance determined for a range of input voltages, rotational speeds, torque values and throttle settings, the motor and ESC are combined with a propeller to yield the complete electric powertrain. An APC 10
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 13
Castle Phoenix Edge HV 120 ESC Efficiency Model Table 4 V 13 14 15 16 17 18
a 0.119 0.128 0.136 0.145 0.154 0.162
ESC Model 0–1 Throttle b -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3 -6.30e-3
c 0.901 0.903 0.905 0.906 0.908 0.910
R2 0.9756 0.9172 0.8728 0.9519 0.8827 0.856
17x12 thin electric propeller is used here as a typical two bladed composite propeller used with electric motors. The propeller has a diameter of 17" and a pitch speed of 12" and weighs 70 g. Performance is tested in a propeller test stand installed at the 7’x5’ wind tunnel at the University of Sydney. The thrust coefficient, power coefficient, and efficiency as a function of propeller advance ratio is given in Figure 15.
VIII. Combined Powertrain Performance Using the propeller performance data with the models of the motor and ESC, the operating lines of the electric powertrain can be plotted in both static operating conditions as well as a steady cruise condition for a particular UAV [11]. Combining the motor and ESC efficiency yields the overall electrical efficiency for a range of input voltages as a function of the rotational speed and torque load (Figure 16). As shown, there is a distinct peak efficiency of 75% that 11
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 14
Castle Phoenix Edge HV 120 ESC Efficiency Difference 1 3000 RPM 4000 RPM
3000 RPM 4000 RPM
0.05
0.8
0
0.6
0.05
0.4
CP
CT
0.1
0.2 0
0 0
0.2
0.4
0.6
0.8
1
(a) Thrust and Power Coefficient vs Advance Ratio
Fig. 15
0
0.2
0.4
0.6
0.8
1
Advance Ratio J
Advance Ratio J
(b) Efficiency vs Advance Ratio
APC 17x12E Propeller Performance
occurs at high rotational speeds and moderate torque values, and this region expands with increasing voltage. At 13 V, the region occurs above 4000 RPM and approximately 0.15–0.25 N·m, whilst at 18 V it occurs above 5000 RPM and 0.15–0.35 N·m. The static performance of the APC 17x12 propeller on this motor and ESC efficiency plot is given by the solid black line. At 2500 RPM, the propeller requires 0.2 N·m, increasing to 0.5 N·m at 4000 RPM. Based on these results, the propeller operates most efficiently when the input voltage is 13 V, but will be limited to 4100 RPM at full throttle. At 13 12
V, the motor and ESC combination operate at 65-70% efficiency with this propeller when stationary. The operating line can also be mapped assuming a steady state cruise condition for a particular UAV. In this case the drag characteristics for a 6.2 kg electric UAV [11] are used to determine the thrust and rotational speed requirement to maintain steady level cruise for speeds from 16–24 m/s. The results are this are shown by the solid blue line. At cruise speed the propeller unloads, requiring less torque to maintain the same rotational speed with the fixed pitch propeller. This shifts the operating line to the right, which is benefical as it moves the electric motor and ESC into a higher efficiency operating condition. At 13 V, the combined motor and ESC efficiency is always 70-75%, and even at 18 V the combined efficiency is above 70 % for most cruise speeds. Thus, given the propeller performance the motor and ESC efficiency can be predicted for any operating condition. This enables the designer to check that the propeller is well matched to the torque and rotational speed characteristics of the motor and ESC, and ensure that all components operate efficiently.
(a) 13 V
(b) 15 V
(c) 17 V
(d) 18 V
Fig. 16
Overall Electric Powertrain Efficiency with Propeller
13
IX. Conclusion This paper has presented the performance testing and modeling of a motor, electronic speed controller, and propeller used in a UAV propulsion system. The methodology used for hardware testing of the motor and electronic speed controller have been presented and used to generate detailed efficiency maps at a range of input voltages, rotational speeds, torque values, and throttle settings. Empirical models of these components have been developed that enable prediction of efficiency, and combined with propeller performance and aircraft drag data to enable estimation of overall powertrain efficiency at any operating condition including static performance and steady level cruise. This can enable the selection of propellers to suit the rotational speed and load characteristics of the motor and ESC.
Acknowledgments This research has been supported by the Australian Defence Science and Technology Group through its Strategic Research Initiative on Signatures, Materials, and Energy. A student scholarship was provided to Andrew Gong by Northrop Grumman Australia and The University of Sydney. The authors would also like to acknowledge the support of the School of Aerospace, Mechanical and Mechatronic Engineering of The University of Sydney for the acquisition of the dynamometer set-up.
References [1] Gundlach, J., Designing unmanned aircraft systems: a comprehensive approach, 1st ed., AIAA Education Series, American Institute of Aeronautics and Astronautics, Reston, VA, USA, 2012. [2] Gundlach, J., and Foch, R. J., Unmanned Aircraft Systems Innovation at the Naval Research Laboratory, Library of Flight, American Institute of Aeronautics and Astronautics, Reston , VA, 2014. [3] Herwerth, C., Chiang, C., Ko, A., Matsuyama, S., Choi, S., Mirmirani, M., Gamble, D., Paul, R., Sanchez, V., Arena, A., Koschany, A., Gu, G., Wankewycz, T., and Jin, P., “Development of a small long endurance hybrid PEM fuel cell powered UAV,” SAE Technical Papers, 2007. [4] Gong, A., and Verstraete, D., “Fuel cell propulsion in small fixed-wing unmanned aerial vehicles: Current status and research needs,” International Journal of Hydrogen Energy, Vol. 42, No. 33, 2017, pp. 21311–21333. [5] Bradley, T., Moffitt, B., Mavris, D., and Parekh, D., “Development and experimental characterization of a fuel cell powered aircraft,” Journal of Power Sources, Vol. 171, No. 2, 2007, pp. 793–801. [6] Swider-Lyons, K., Stroman, R., Rodgers, J., Edwards, D., Mackrell, J., Schuette, M., and Page, G., “Liquid hydrogen fuel system for small unmanned air vehicles,” 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition 2013, 2013. [7] Hung, J., and Gonzalez, L., “On parallel hybrid-electric propulsion system for unmanned aerial vehicles,” Progress in Aerospace Sciences, Vol. 51, No. 0, 2012, pp. 1 – 17. [8] Gong, A., and Verstraete, D., “Role of battery in a hybrid electrical fuel cell UAV propulsion system,” 52nd AIAA Aerospace Sciences Meeting - AIAA Science and Technology Forum and Exposition, SciTech 2014, 2014. [9] Green, C. R., “Modeling and Test of the Efficiency of Electronic Speed Controllers for Brushless DC Motors,” MSc thesis, 2015. [10] Gong, A., and Verstraete, D., “Experimental Testing of Electronic Speed Controllers for UAVs,” 53rd AIAA/SAE/ASEE Joint Propulsion Conference, 2017. [11] Gong, A., MacNeill, R., Verstraete, D., and Palmer, J. L., “Analysis of a Fuel-Cell/Battery/Supercapacitor Hybrid Propulsion System for a UAV using a Hardware-in-the-Loop Flight Simulator,” 1st AIAA/IEEE Electric Aircraft Technologies Symposium, 2018. [12] Gong, A., and Verstraete, D., “Design and Bench Test of a Fuel-Cell/Battery Hybrid UAV Propulsion System using Metal Hydride Hydrogen Storage,” 53rd AIAA/SAE/ASEE Joint Propulsion Conference, 2017. [13] Gabriel, D., Meyer, J., and Plessis, F. D., “Brushless DC Motor Characterisation and Selection for a Fixed Wing UAV,” IEEE Africon, 2011, pp. 1–6.
14
[14] McDonald, R., Electric Motor Modeling for Conceptual Aircraft Design, Aerospace Sciences Meetings, American Institute of Aeronautics and Astronautics, 2013. [15] Lundström, D., Amadori, K., and Krus, P., “Validation of Models for Small Scale Electric Propulsion Systems,” 48th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition, 2010.
15
