MECHANICAL ENGINEERING DESIGN PROJECTS FINAL STATUS REPORT

Transcription

1 SUBMITTED BY Greg Near, Sam Verdugo, Paul Baranano, Kevin Boreen May 7, P a g e

2 TABLE OF CONTENTS PROJECT OVERVIEW... 3 OVERALL DESIGN... 4 TESTING/PROTOTYPING RESULTS... 6 PROPOSED IMPROVEMENTS/LESSONS LEARNED... 8 REQUIREMENTS COMPLIANCE... 8 COST P a g e

3 PROJECT OVERVIEW Our project, STaRe-UAV (Short Takeoff and Reconnaissance) is an unmanned aerial vehicle (UAV) inspired by the cancelled Boeing Heliwing project, to be used for sustained ground surveillance in areas of rough terrain where a rolling takeoff is undesirable or impossible. The aircraft is designed to be capable of vertical takeoff and transition into horizontal flight, with a flight duration goal of fifteen minutes. Landing is to be assisted by a parachute deployed from the nose. A camera and transmitter system provides live video of the terrain below, with a receiver on the ground where the operators will stream and save high definition video to a laptop. The UAV is designed to be deployed by a team of three (pilot, video monitor, and spotter), and is easily separable into components for each member to carry. With this in mind, the UAV will be a reasonable size and weight- approximately 10lbs and a wingspan of 6 feet. 3 P a g e

4 OVERALL DESIGN Our team began the creation of our design by performing powertrain sizing calculations to ensure our aircraft would have the capability to fulfill the mission requirements outlined above. Of primary concern was the ability to generate sufficient thrust to achieve vertical takeoff while still carrying a battery large enough to supply power both for the takeoff and for sustained flight. A popular rule-of-thumb among RC aircraft enthusiasts is to measure performance with a metric called Watts per pound, which compares total motor power on the aircraft to the aircraft weight for traditional, small, horizontally-flying RC aircraft, the recommendation is about 50W/lb, while for 3D flying aircraft, including our vertical-takeoff aircraft, a more suitable number is around 200W/lb. During our sizing calculations, we felt it would be safest to include extra power and designed to have 250W/lb by estimating an aircraft weight of 10lbs, yielding a suggested power requirement of 2500W, which we achieved using two 1250W outrunning (and counterrotating) motors. There are also recommendations in the hobby community for propeller characteristics, including a relationship between the pitch rate of the prop and the torque output of the motor, which we followed to select a pair of counter-rotating propellers that fit our motors. We also performed rudimentary momentum theory analysis on the streamtubes of the propellers to determine their diameters, and came up with a propeller diameter of 14 inches, with a pitch rate of 4.5. Based on the motor characteristics, we were able to select batteries that would be able to serve the motor the 60A current it needs at max throttle during takeoff and still have energy left to supply the motors during horizontal cruise for at least 15 minutes. After making the selections for our powertrain system, we began designing outwards, starting with the wing. Our wing design was initially fully tapered from a chord length of 12 inches at the center to 6 inches at the ends, with a full wingspan of 6 feet. Again, sizing was the first design task completed, when we used an airspeed estimate from our powertrain system to determine the rough wingspan needed to generate sufficient lift using a typical airfoil. This gave us the 6 foot span, while the taper was used to try to distribute the lift forces on the wing in an elliptical manner, thus decreasing stresses on the wing and airframe. Using XFoil program to perform the panel method and vortex lattice method on several airfoils, we selected a slightly cambered NACA 5-series airfoil from which we would make our wing. With the shape of the outside of the wing selected and designed, we needed to change the solid body into a skeleton that we would be able to lasercut and cover, so we used the typical industry method of ribs and spars the ribs were essentially cross-sections of the wing with holes that allowed wires to pass through and weight to be cut down, while spars were long, thin members that ran across the span of the wing to hold the ribs together and provide longitudinal rigidity. This is all covered by a heat-shrink mylar film, which we are able to attach to the ribs using a soldering iron, stretching it tight as we attach it, and then we are able to remove wrinkles from the surface by applying heat with a heat gun. When we began to design and try to implement an aileron into the wing, we realized there would be a substantial amount of part customization to construct each section of the aileron due to the taper, and the effectiveness of the surface would be decreased due its decreased size. A goal for fabrication was to make it as rapid as possible, and so this design was abandoned and the wing taper stopped after 18 inches from the center, and ailerons with identical sized ribs were placed along the remaining length of the wing. These are now actuated by a center rib that has a raised portion for a control rod to be connected. 4 P a g e

5 We then designed our other aerodynamic surfaces, including the horizontal and vertical stabilizers to ensure we would have good pitch control and would be stable in flight. After talking with Bruce Kothmann about what constituted a good airfoil for stabilizers, we decided on the NACA a thin, symmetric airfoil. The size of each stabilizer was calculated using volume coefficients. This is a value calculated by the ratio of stabilizer area to wing area multiplied by the ratio of distance between aerodynamic centers of the stabilizer and wing to the mean aerodynamic chord of the wing for horizontal stabilizers. For vertical stabilizers, the distance between aerodynamic centers is replaced by the wingspan. Research into typical values for aerobatic planes lead to us designing the area of each stabilizer for a coefficient of about 0.5 for horizontal, and 0.12 for vertical. Since there are four vertical stabilizers, the coefficient was divided by four, and each was sized for a volume coefficient of The design of the horizontal stabilizers was created in a slightly different method than that of the wing. Since the airfoil is only 6% of the chord length thick, with a chord length of 6 inches it is 0.36 inches thick. As such, it only has two quarter inch balsa spars running through the thickest part of the ribs for support, as opposed to the wing, which has at least six spars attaching each successive rib. There are 16 ribs, 14 of which are cut short at the half chord to allow for the elevator to be attached at the rear. The elevator is constructed using a 1/8 inch diameter aluminum rod for support and 14 ribs designed as the rear half of the airfoil shape. The vertical stabilizers are similarly designed, but are trapezoidal, with a height of 12 inches, lower base of 10 inches, and upper edge of 6 inches in chord length. There are 9 total ribs, each having quarter inch balsa sticks run through it for support where structurally stable. There are spots on the side bodies where each of these components mounts with some adhesive. The bodies of the aircraft hold all of the necessary components for flight, control and reconnaissance, and were designed to meet those needs. The three-body design was selected as we needed the parachute to be centered on the aircraft, which precluded a single-motor design, and thus with two motors we needed two extra bodies to hold them. For these reasons, we decided to house the parachute in the center body, along with the RC receiver and the camera and transmitting system, while placing a motor, ESC and battery in each of the side bodies. The general body design is laser cut circular balsa formers spaced 4 inches apart and connected by 3/8 inch balsa spars. After four formers, the design tapers towards the back to save on material and weight, as well as to make the design more streamlined, although the taper on the side bodies is much more significant than that of the main body. At the point where the taper and straight portions meet, we placed an extra former to allow for a flat surface to attach the two separate parts securely. The tip of the side bodies is a removable nose cone, which guards the DC motor powering the propellers, but allows access for repairs and maintenance. The interior of the straight section of each side body is left open to the top such that the batteries can be installed and the servo cables running to the receiver in the main body can be attached from the wing. In the main body, rather than square formers with filleted corners, as in the side bodies, the front section of the main body has a rounded interior to allow for a stiff paper to be placed on the surface of the interior. This serves to prevent the parachute from snagging on the formers as it deploys. The nose cone for the main body is hinged and contains a servo pin release system inside. The hinge is loaded with torsional springs such that when the pin is pulled, nose cone swings out of the way and the parachute is released. The camera system is rather basic, consisting of a 2.4GHz transmitter and 720p camera that rides in the rear of the main body, with a ground receiver that connects to a USB port through which we are able to capture and record video 5 P a g e

6 footage. The transmitter comes with a large and unwieldy antenna almost three inches long. We had to make space for this in the rear of our main body, and thus extended the body slightly from our previous design iteration. TESTING/PROTOTYPING RESULTS After analyzing our project problem and determining the team s execution plan, we decided that given the scope of our project, we would be better off building quickly and performing large amounts of field testing rather than large amounts of prototyping and relatively less field work. This is due in large part to budge constraints, as we used almost all of the $1200 budget on our large aircraft, and thus did not have the funds to buy parts for a scale model before moving to fullscale. As such, our testing methodology was to perform subsystem tests on the ground to ensure systems were working in the manner which we expected, then integrated system tests on the ground to ensure they were working together properly, and then finally field testing of the aircraft to test controlled flight and the functioning of all integrated systems together. We began by performing load testing of the structures that we had built, to ensure they would be able to withstand the loads placed on them during flight. For the wing, this meant we placed distributed weights along its span to represent the distributed weight of the aircraft being felt by the wing as lift over its surface. We determined that the center of the wing was not strong enough, as the balsa in that section had begun to deform, and thus we attached several 1/8 balsa sheets to give more longitudinal stiffness to the wing. This increased weight slightly, which is one of the primary reasons we included extra power in our powertrain selection we wanted to be able to overbuild the frame to ensure it had the necessary strength, without worrying about weight. One of our primary concerns was the proper functioning of the electronic systems, as none of our team members had mechatronics or robotics experience. To this end, we began breadboarding the circuits for our motor-esc-battery systems and performed static thrust and spin testing of the motors. To test the circuits, we initially spun up a motor to ensure connectivity, then attached the motor to its motor mount and ran thrust tests, plotting throttle position versus rotation speed of the propeller, measured using a tachometer. From these tests, we were able to measure how the gain of the controller affected the motor thrust response, which we would have been able to use for tweaking of the controls had we had more time for testing. The parachute system was also vital for the successful flight of the aircraft, and thus needed significant testing on the ground. Again, we ran through a series of tests that successively increased in scope the first test was just to ensure the parachute would inflate at low speed, by running with the packaged parachute until it deployed. From this test, we were also able to determine a generous estimate of about 75 feet for the distance the parachute would need to deploy. We then integrated the parachute system into the main body and tested the deployment methods, including the hinge and servo-pin system used for the nosecone pivoting. We noticed when using just the hinge with torsional springs, the nosecone would often open but hang in the way of the opening, blocking the parachute from deploying. We developed a system using a tensioner attached to the nosecone, which applied a force to the nosecone pulling it out of the way. When we were able to ensure the nosecone would function properly every time, we began testing the deployment of the parachute using the plunger system to push the folded chute out. These tests consisted of holding the center body 6 P a g e

7 and running with it while a teammate deploys the parachute from the controller, continuing until the parachute fully deploys. From the ground tests, we again determined that the system we designed was unable to apply a large enough force to deploy the parachute, and decided to implement a pilot chute that would hang out of the nosecone and pull the full seven foot diameter parachute out. Further ground tests proved this system to be successful. Range was the primary reason for testing the camera and RF systems, and thus the testing regime for them was simple: after ensuring the signal was correctly received at close range, we continued to increase the range until the signal began to become distorted. This was simple for the camera system, as it was obvious when the signal was significantly degraded, and we managed to successfully achieve a range of over 1000 feet, farther than the desired altitude of flight for the aircraft. For the radio system, we performed a similar test, walking with a servo attached to the receiver, while actuating it constantly with the controller. It was more difficult to determine when the signal had degraded sufficiently to render the aircraft inoperable, but it was significantly farther than the 1000 feet of the camera system. After integrating the electronic circuits and components into the airframe, we performed range-of-motion testing of the control surfaces and managed to achieve 30 degree deflection in both directions of the ailerons, and about 20 degrees for the elevator, as the connection point was closer to the servo. We determined from research of other model aircraft control surfaces that this was more than sufficient for controlling the aircraft at its cruise velocity, although as we found during field testing, it was not enough for low-velocity flight. With the full integration of airframe and electronic systems, we were ready to begin flight tests. We began with a hover test, ensuring that we had sufficient thrust from the two motors to lift the aircraft into the air. While the test showed that we did indeed have enough power, it also showed there were some serious issues with low-speed control of the aircraft, as it was difficult to maintain steady hover. As this test was performed late in the semester with little time left on the project, we were not able to drastically change the design of the aircraft to try to remedy this problem, and instead decided to continue with testing the horizontal flight capabilities of the design. We performed five field tests with varying results. The first test highlighted the both the importance of care and of safety in the project, as we accidentally hit the throttle while setting up the plane for launch, sending it up about five feet and then crashing it back into the ground and destroying several formers in the bodies as well as the connection points between them. However, the crash also showed that the design was promising in the sense that it could be rebuilt quickly, as we were able to perform another four tests with another four crashes in the span of two weeks. The second test was again a lesson in care, as we had the motors trimmed incorrectly on the controller, leading to one motor providing significantly more thrust than the other and causing extreme yaw in the first few seconds of the flight, again sending the plane crashing into the ground. After another rebuild, the third flight test provided us with more knowledge about how to fly the plane, as it was launched at an angle that was past stall for the wing as well as having weight distributed too far back compared to the center of pressure of the wing. In our next rebuild, we moved the batteries forward slightly, as they were our heaviest component and thus had the greatest effect on the center of gravity of the aircraft. We also knew to launch the aircraft at a smaller angle, which we demonstrated in our fourth flight test. However, communication issues between team members meant the aircraft was launched at half-throttle instead of full-throttle, and it glided to a smooth crash that only destroyed the 7 P a g e

8 bottom vertical stabilizer tails. Our fifth test was by far the most promising, as we achieved a height of about fifteen feet before losing control of the aircraft and sending it into a spin into the ground, at which point we had to conclude testing and prepare the aircraft to present. By far the most convincing lesson we learned from the tests was the degree of difficulty there is in controlling low-speed flight, as there is so little time to correct using the controller and control surfaces before the aircraft comes crashing down. Although we were unable to achieve the vertical or horizontal flight we had hoped and set as requirements, we managed to perform some promising tests that showed with careful thought about low-speed flight dynamics, the aircraft would certainly fly. PROPOSED IMPROVEMENTS/LESSONS LEARNED As with many projects, there is a great amount of room for improvement with this UAV. Most importantly would be the addition of PID control in pitch, roll, and yaw. The control system design for this aircraft was a minimalist proportional control, which is what the radio control system purchased was capable of delivering. While this is what is found on most model airplanes, this one could use a bit more stabilization purely due to its size and operating realm. Since it is designed to operate at moderate to low Reynolds numbers, low airspeed dynamics became an important factor during flight, and some aspects of airplane response in this realm are not very desirable. As such, adding a system of sensors and a processor such as an Arduino or Ardupilot, and programming a PID algorithm could reduce undesirable responses, and make the airplane easier to fly. However, since this is a large amount of work, we think that this could be done as another senior design project in the future. The other major improvement for the project is the addition of tilt rotor control or variable-pitch propellers. Tilt rotors or variable-pitch propellers allow a plane to achieve thrust vectoring, and more helicopter-like characteristics for takeoff and landing. Successful implementation of either of these and proper controllers would allow the airplane to achieve stable hover, eliminating the need for a parachute as the primary landing system. This means that vertical takeoff and landing is also possible, and controllable. If that were attempted with the current design, instabilities would arise and the plane would most likely crash. Furthermore, the resulting elimination of the parachute would save space and reduce weight, allowing for more control equipment to be put into the design. As with the implementation of a flight stabilization system, we believe this could be implemented as another future senior design project. REQUIREMENTS COMPLIANCE The first customer requirements set for us were to design a UAV that could achieve a lift to drag ratio greater than or equal to 15, and be capable of vertical takeoff. Analysis of the former was completed using XFOIL code to gather lift and drag data on many different candidate airfoils. Eventually, it was decided that we would move forward with the NACA airfoil for the wing because of its desirable lift characteristics for the UAV s mission. It would give us reasonably high lift at low angles of attack, which would be ideal for keeping the camera level in the rear of the aircraft. When extruded to become a 3D wing design, a vortex lattice CFD code was run on the design to ensure the proper lift force could be achieved and the lift to drag ratio was in the range we were looking for. In the end, a lift force of 15 lbs could be achieved at a speed of 25 mph, and the lift to drag ratio was about P a g e

9 To ensure vertical takeoff was possible, we used momentum theory for propellers to calculate the force that we could achieve with different sized propellers. Since the estimated weight of the plane was about 10 lbs, we aimed to have roughly twice this amount available in thrust force. It was determined that for our chosen motors, two 1250 W outrunners, we could achieve this force using two counter rotating 14 diameter propellers. We tested this by taking the entire UAV to an open area, having two team members support the wings (to prevent destabilization in pitch), and ramping up the throttle until a hover was achieved. Liftoff was achieved at about 35% throttle input, which was backed up by experimental findings about the relationship between throttle input percentage and thrust force. The next requirements came as a result of building any controllable airplane. The weight distribution must be balanced such that there is no resulting moment to negatively affect flight, and to only achieve pitch, yaw, and roll moments from control inputs. The UAV was designed such that the center of mass would lie at the center of the wingspan, and in between 25% and 33% of the mean aerodynamic chord, which is the suggested place for it to lie in the lengthwise direction. Results from the fourth field test confirmed this, as we were able to witness the gliding behavior of the UAV, which was level the entire time. The control system was both successful and unsuccessful in implementation. During the second field test, the controller was not set to the proper trim configuration, causing the left propeller to spin much faster than the right. Upon launch, the aircraft experienced a yaw moment to the right, which was irreversible due to the trim setting, causing the plane to lose control and crash. However, despite the crash, this proved to us that a yaw moment could be achieved using our implemented method of differential thrust rather than a mechanical rudder. Both pitch and roll control were unsuccessful during field tests. Video analysis of each test where a pitch moment was achieved showed that the result was from the aileron configuration. The servo-aileron control rod turned out to have an unanticipated degree of freedom that caused inaccurate actuation of the control surfaces. Because of this, upon launch, the ailerons both deflected up due to the pressure difference created on the airfoil during flight, causing a destabilizing nose up moment. The nose up moment flipped the airplane over at the beginning of the final two tests. The problem was not diagnosed until after the final test because the previous test was launched beyond the stall angle, which had a greater contribution to the issue at the time. Video from two different angles on the final test also helped us diagnose this issue. As a result of the aileron failure, we were never able to test elevator functionality, leading us to say that there was a failure in achieving both pitch and roll moments. Furthermore, due to lack of total control, the integrated system was unable to fly for an extended time before the project deadline. We do believe that given another iteration of the design, this time fixing the servo-aileron connection, controlled flight could possibly be achieved with this system. 9 P a g e

10 COST Item Quantity Unit Cost Total Cost Balsa Sheet 1/4x12x Balsa Sheet 1/8x12x Balsa Stick 1/4x1/4x Balsa Stick 3/8x3/8x Wild Scorpion RC 6S Lipo Battery Gens Ace Mars Brushless Motor with 60A ESC GHz CT6B 6-Channel Transmitter and Receiver Sony WDR770 Camera Plug and Play 1.2GHz Wireless System APC 14x4.7 Counter Rotating Airplane Propellers " Diameter Rescue Parachute Econokote 6' Roll Titebond Wood Glue Elmers Wood Filler 1/2 Pt M Sandpaper 3 pack Total Spending: P a g e