1. Summary - University of Michigan



University of Michigan Post-Launch Assessment ReviewProject Wolverine:Butterfly Valve Drag Variation for Mile High Apogee2011-2012University Student Launch Initiative Department of Aerospace EngineeringUniversity of Michigan3012 Francois-Xavier Bagnoud Building1320 Beal AvenueAnn Arbor, MI 48109-2140Table of Contents TOC \o "1-3" \h \z \u 1. Summary PAGEREF _Toc324109433 \h 31.1 Team Name and Location PAGEREF _Toc324109434 \h 31.3.1 Vehicle Size PAGEREF _Toc324109435 \h 31.3.2 Motor Choice PAGEREF _Toc324109436 \h 31.4 Science Experiment PAGEREF _Toc324109437 \h 31.5 Altitude Reached PAGEREF _Toc324109438 \h 32. Vehicle PAGEREF _Toc324109439 \h 42.1 Vehicle Summary PAGEREF _Toc324109440 \h 42.2 Data Analysis and Results of Vehicle PAGEREF _Toc324109441 \h 43. Experiment PAGEREF _Toc324109442 \h 83.1 Experiment Summary PAGEREF _Toc324109443 \h 83.2 Scientific Value PAGEREF _Toc324109444 \h 83.3 Data Analysis and Results PAGEREF _Toc324109445 \h 83.4 Visual Data Observed PAGEREF _Toc324109446 \h 104. Lessons Learned and Experience PAGEREF _Toc324109447 \h 115. Educational Engagement PAGEREF _Toc324109448 \h 126. Budget PAGEREF _Toc324109449 \h 136.1 Budget Overview PAGEREF _Toc324109450 \h 136.2 Project Budget Details PAGEREF _Toc324109451 \h 141. Summary1.1 Team Name and LocationThe University of Michigan Rocket Engineering Association (MREA) from Ann Arbor, Michigan proposes Project Wolverine. 1.3.1 Vehicle SizeOuter diameter: 5.52”Inner diameter: 5.36”Total length: 114.00”Mass with Motor: 40 lbs1.3.2 Motor ChoiceThe motor type for the vehicle is a Cesaroni L-1395 motor (75mm diameter) with 4,895 N-s of total impulse. This determination was made based on Rocksim simulations, hand calculations, and multiple test launches.1.4 Science ExperimentThe projected experiment for MREA’s Project Wolverine is two butterfly valves (separated 180 degrees) that vary pressure drag. During the ascent, the butterfly valves will be actuated based on real time data from GPS and flight computers to vary the pressure drag on the rocket. The goal of this experiment is to put a high impulse motor in the rocket that ensures the rocket will reach at least one mile at apogee, yet have the rocket only reach one mile at apogee due to the increased drag from the butterfly valves.1.5 Altitude Reached5549 ft2. Vehicle2.1 Vehicle Summary965205544820Figure SEQ Figure \* ARABIC 1: Launch VehicleFigure SEQ Figure \* ARABIC 1: Launch Vehicle965203125470Side-can openingAvionics bayMaintenance separationApogee separationSet of four orthogonal fins00Side-can openingAvionics bayMaintenance separationApogee separationSet of four orthogonal finsThe vehicle proposed to answer the requirements of NASA’s USLI is pictured in REF _Ref324107598 \h Figure 1 below. The composite makeup of the rocket includes wound vulcanized fiber for all body tubes, fiberglass for the fins, and standard 6061 aluminum structurally supporting the drag induction mechanism where appropriate. The 36 pound (dry) vehicle is 114 inches long with a 5.5 inch main body tube. Side-cans with a 2.5 inch diameter run just over half of the body length and house two butterfly valves. Electronics contained within the avionics bay send pulse width modulated control signals to a single servo, actuating the counter-rotating butterfly valves located at the maintenance separation. The angle of these valve alter the drag profile of the rocket and hence can be used in order to achieve a given target altitude. One of the most important artifacts of this system that must be considered is the boundary layer progression in the side cans given differing velocities and butterfly flap angles. This boundary layer progression can induce phenomena such as choked flow, causing false pressure sensor readings as well as negating the controller as we will see later in test result analysis. Another important consideration is that this design is single axisymmetric, necessitating robustly assessed stability given lateral wind vectors over at least half the vehicle body. The vehicle has been built to fly at least five full launches and recoveries over a wide variety of environmental conditions to build up confidence in the design and ability to achieve goals.2.2 Data Analysis and Results of VehicleSuccessesThe choice to use Blue Tube 2.0 as the material for the airframe as well as the cans was determined to be an extremely good decision. In multiple launches in various humidity and temperatures ranging from below freezing to near 80 degrees, the vehicle was solid, yet able to absorb a fair share of wear. Even in the extreme case of falling under only a drogue parachute all the way from apogee to the ground (at a rate of nearly 70 ft/sec), the vehicle sustained minimal structural damage. The tubes did not exhibit any great tendency to warp or crack, and they were very easy to work with (epoxying together, cutting, sanding). The vehicle performed consistently and as designed during boost phase in all launches, exhibiting a high level of stability and the ability to handle different launch angles. This may be attributed to the addition of the Mass Ballast System that added about 3.5 lbs of weight to an otherwise completed rocket in the form of 5 concrete rings located inside the nose cone. Adding this mass brought the pre-launch stability margin of the rocket up to about 2.0, a safe margin that provided a little cushion for any instabilities generated by the unorthodox “side-can” design of the vehicle. Without these concrete rings it is doubtful that the rocket would have displayed such robustness and stability in all flights. At apogee, the vehicle successfully separated every time, deploying a drogue chute and beginning its descent. The size of the chamber used to house the drogue parachute and associated protective packing material (“dog barf”) was repeatedly proved to be adequate – the parachute deployed each launch and sustained no major scorching or other damage as a result of the ejection charge explosion. During descent, the rocket’s recovery system worked flawlessly for 3 out of 4 launches. During the second test launch, the main parachute failed to deploy, causing the rocket to fall from apogee to ground under only the drogue chute. Luckily, due to the drogue and some soft mud, the rocket sustained no major damage. Instead, the most important consequence of this failure was the creation of a new set of rules for folding and packing the parachute, as well as an increase in the length of shock cord in the rocket. Both of these fixes ensured the successful deployment of the main chute at both the third and fourth launches. Quantitatively, actual descent rates under both drogue and main chutes gathered from the final launch were very close to predicted:Predicted (ft/sec)Actual (ft/sec)36” Drogue 7470Cert-3 XL Main Chute1821FailuresThere were two repeated failures that resulted from the design of the avionics bay (AvBay), concerning specifically the system that equalized the pressures inside and outside the avionics bay so that the altimeters could accurately measure pressure. The first failure that occurred more than once was a downward spike in the altitude curve at apogee, as visible in REF _Ref324107934 \h Figure 2. This anomaly was attributed to the improper sealing of the avionics bay from the aft ejection charge that separated the vehicle at apogee. Despite the addition of gaskets and another fiberglass plate inside the avionics bay after the second test launch, this problem persisted through our final launch in Alabama. Although it was never determined that this failure was actually detrimental to any of the electronics in the avionics bay or the vehicle in general, the inability to totally seal the AvBay from the ejection charges produced an incorrect altitude curve that repeatedly marred the flight data. Figure SEQ Figure \* ARABIC 2: Huntsville, AL launch showing erroneous spike in pressure-based altitude curve (blue) at apogeeThe second failure related to the AvBay was a repeated disparity in altitude measurements between the accelerometer in the Raven flight computer and both altimeters (one in the Raven, the other being the Stratologger). At motor burnout, only some 3-4 seconds into flight, there was at least a 200-meter difference between the two altitude measurements, with the pressure-based data from the altimeter indicating the lower altitude of the two – probably the more incorrect altitude. On a larger scale, the altitude curve derived from the altimeters looks very strange up to and at motor burnout, as there is a sudden increase in slope at burnout ( REF _Ref324108021 \h Figure 3 & REF _Ref324108023 \h Figure 4, blue line). This sudden increase in slope would normally indicate a positive acceleration, however the vehicle should not be positively accelerating at all after the engine has ceased to burn. On the other hand, the altitude data from the accelerometer looks much nicer and agrees with calculations for an approximate altitude at burnout ( REF _Ref324108021 \h Figure 3 & REF _Ref324108023 \h Figure 4, black line). 270764016002000 Figure SEQ Figure \* ARABIC 3: 3rd test launch data Figure SEQ Figure \* ARABIC 4: Huntsville, AL launch data Due to the above considerations, it was determined after the third launch that these measurements from the altimeters were faulty and contained significant error during the initial portion of the ascent. Besides this failure having dramatic and extremely detrimental effects on the payload portion of the rocket (discussed in 3.2) it also prompted a look and redesign of the pressure equalization system between the AvBay and the outside of the vehicle. For the first three launches, the vehicle had four, orthogonal quarter-inch holes drilled in both the airframe and the AvBay internal body tube to equalize pressure for the altimeters, which were open to the environment inside the AvBay. It was suggested that perhaps there was airflow either over or through these holes during boost that was causing errant measurements from the altimeter. Before the launch in Alabama, wind tunnel testing was performed on the actual AvBay to try and relocate these holes in order to ensure accurate measurements. In reaction to the wind tunnel work the holes were made smaller and moved as far away from any obstacles as possible. Despite these adjustments, the problem persisted and actually worsened during the flight in Alabama ( REF _Ref324108023 \h Figure 4). There has not been any definitive conclusion about why this failure occurred, however, there is no doubt that the vehicle did not perform exactly as intended as a large consequence of the inability to accurately determine altitude based on static pressure during and immediately after boost – the most dynamic part of flight and most critical for adjusting drag. 3. Experiment3.1 Experiment SummaryOur payload will use the principles of PID control theory to govern the aforementioned mechanism designed to induce pressure drag as a means of regulating vehicle altitude. The control system will be actuated at a pre-designated trigger velocity to ensure that our flight speed has passed the highly unstable transonic regime, where shock formation on our control surfaces could lead to instabilities. The most significant objective of our controller is induction of pressure drag in the mean energy solution path such that both apogee-amplifying and apogee-depreciating perturbations are recoverable through the majority of flight. The controller should be robust enough to recover altitude goals over various launch environment conditions expected during operation in testing and in competition. Drag should be calculated dynamically during flight, and the controller should respond to physical system changes in no more than 50 milliseconds. Of utmost importance in payload success is that drag is actuated exclusively opposite to the vehicle velocity vector, and that no moments are created around any other axis than the longitudinal body axis of our vehicle. The control system must demonstrate sufficient disturbance rejection over the range reasonable environmental perturbations encountered during flight at any stage, and recover within 2% of the goal altitude.3.2 Scientific ValueDue to the nature of our flight goals, and the high level of a prior knowledge we have regarding the nature of our flight, we have implemented a PID algorithm with a slight modification. As opposed to direct gain scheduling, we modify our target altitude dislocating it from our nominal target altitude to ensure that the mean energy path solution is attained as opposed to the controller quickly damping to the goal altitude. The extent and means by which this dislocation occurs is described in detail below, and is based off of several vehicle state variables. This design was selected due to the fact that the control system has no means to add energy back into the system, making energy management an extremely important aspect of the controller design. If the nominal flight path is attained within a few moments of the controller activating, there would be no means for the controller to reject any disturbance which would reduce the apogee altitude. The risk inherent with this classical PID design in failing the criterion of being able to recover within 2% of the goal altitude was determined to be too great to allow an unmodified PID algorithm to control the ascent of our rocket. 3.3 Data Analysis and Results29241754457065Figure SEQ Figure \* ARABIC 5: Launch Vehicle Shortly After Liftoff on April 22, 201200Figure SEQ Figure \* ARABIC 5: Launch Vehicle Shortly After Liftoff on April 22, 2012292417527940The primary purpose of our USLI launch was to reach a mile altitude, and perform a full scale vehicle test. As with the previous two test launches, the vehicle weighed in at 16.3 kg and was propelled via a Cesaroni L1395 motor which provided 4900 N-s of impulse over 3.6 seconds.This launch took place on April 22, 2012 in Huntsville, Alabama. In the early afternoon our vehicle began its 18 second ascent. Our pressure port size had been reduced in an attempt to alleviate some of the issues experienced during the last two test launches with pressure equalization lag.However, as the flight data in REF _Ref324109193 \h Figure 6 indicated, our pressure port sizing issue had not been alleviated, but rather aggravated. Our rocket apogee was inconclusive, and lay between 1580 and 1683 meters. What likely contributed most significantly to this pressure deviation was the burnout altitude projected at apogee. According to our barometric sensors, which are very heavily1733550781050Latently regressive BATES burn profileLarge pressure equalization lagAccelerometer (black) vs. Altimeter (blue) altitudeMax velocity at motor burnoutPressure from ejection charge indicates altimeter seal issues00Latently regressive BATES burn profileLarge pressure equalization lagAccelerometer (black) vs. Altimeter (blue) altitudeMax velocity at motor burnoutPressure from ejection charge indicates altimeter seal issues-190503686175Figure SEQ Figure \* ARABIC 6: Raven Flight Computer Data from Flight on April 22, 201200Figure SEQ Figure \* ARABIC 6: Raven Flight Computer Data from Flight on April 22, 2012-19050742950weighted in altitude calculations, we were only at 91 meters of altitude at motor burnout. All simulations and accelerometer based altitude readings indicated that this altitude was closer to 350 meters. Due to our system having the largest ability to influence drag early on in the flight, this effect was extremely detrimental. Our final conclusion is that there is a fundamental flaw in the setup of pressure entrance into our avionics bay. Larger pressure holes in various locations may alleviate this issue although that does come at the cost of a larger chance for dynamic pressure fluctuations to interfere with guidance. In the future it is strongly suggested that a guidance system based on inertial rather than external measurement is used. 3.4 Visual Data ObservedThe major finding during this test and launch process has been that a DART, or any target modifying control system is extremely sensitive to incorrect flight data. The inherent risk of such a system is present in twofold, one being that when the system is modifying the target the most heavily, data must very accurate. The second is that a system such as ours relies on a controller that becomes less effective as we approach our target. A perhaps more thorough revision of the controller would include an undershoot command, that rectifies itself via injection of energy into the flight system to increase apogee. Speaking more directly to the control system itself, gain scheduling effects have been mimicked with general success via a modified target. Had the controller had the correct data during the entire flight, there is a good chance we would have seen altitudes that very nearly achieved our goals, as we can see by the effect of the controller in our final test flight with just a single point of what can be considered consistent data. However by comparing the flight results with what would have been achieved via a simple Zeiger-Nicholas tuned PID controller, we can see the DART controller did provide some substantial gains. Further work regarding a DART controller, or a controller that employs continuous gain scheduling, would involve looking at which state variables can most accurately be used to modify a given arbitrary target, and how those modifications can be based on vehicle characteristics in a general way. Overall this study has shown that there is promise that continuous gain scheduling, if built into the dynamics and via a priori knowledge of a vehicle, can attain better results that a simple PID controller. 4. Lessons Learned and ExperienceBefore the competition we attempted three separate test flights, but we were unable to collect data from our control system on any of these test flights. We feel like we could have benefitted from data collection to see if the control system was working correctly. This showed us how important test flights are in the testing of systems.We attempted to control our altitude by using two butterfly valves that vary pressure drag. The valves would be actuated based on real time data from multiple onboard electronics shortly after motor burn out. However at competition it was determined that the butterfly valves did not actuate as expected. After doing some analysis we have determined that it was due to inaccurate number readings. At apogee the accelerometer we were using to read off of said 5185ft when in actuality our Raven altimeter read 5523ft and our competition altimeter read 5549ft. We believe the reason for the valves not performing as expected is due to some fundamental flaw in the pressure-hole system resulting in inaccurate readings from both altimeters. This has been an issue that we have been trying to resolve since our third full scale test launch. Our currently assumption is that having the avionics bay enclosed in another hollow cylinder has allowed pressure to leak into the avionics bay. Even though the system did not work as desired the experience helped prepare us as future engineers. By having these issues we were able to go through the design-fabrication-test cycle multiple times, testing our problem solving skills. Each time we learned more about how the onboard systems work in relation to each other and the environment. 5. Educational EngagementThe Michigan Rocket Engineering Association’s first run of a new outreach project was a deemed a success in a few ways. Our members taught students at Pinckney Community High School the basics of rocketry in a three week seminar in which students built rockets in teams as a part of a TARC-like competition. These students in turn showed a true interest in the subject wanting to start a TARC team and have MREA members mentor the team. The school labeled the seminar a success and invited MREA back to teach again next year. The goals of the USLI educational outreach project were to inspire an interest in rocketry and engineering in general as well as to establish an ongoing project that could continue in years to come. MREA has completed these goals on a small scale and hopes to continue this project on a larger scale. Plans for a manuscript that describes in detail how to run this seminar are in the works and MREA hopes to distribute the information to other schools so that they may run their own seminar without having our members there in person. MREA hopes to continue this project in the fall with a new set of students, perfecting this seminar through trial and error.6. Budget6.1 Budget OverviewMREA has spent just over $6,200 on the entire USLI project. Actual flight hardware has cost MREA just over $3,200, which puts MREA well under the competition rule of $5,000 of flight hardware on the launch pad. Additionally, MREA spent almost $2,400 transporting 12 team members to Huntsville for the competition, and MREA spent about $600 on the outreach project with Pinckney High School. REF _Ref324108281 \h Table 1 and REF _Ref324108325 \h Figure 7 detail all of the expenditures MREA has incurred during the USLI project. SubsystemTotal Cost (USD)Structural Components722.58Controls Structural Components289.43Avionics Components600.05Propulsion Components1216.91Travel Expenses2443.43Outreach Project604.37Shipping Costs387.13 TOTAL 6263.90Table SEQ Table \* ARABIC 1: Project Cost Breakdown by Subsystem and CategoryFigure SEQ Figure \* ARABIC 7: Visual Representation of Project CostsIn order to fund the project, MREA received numerous grants and contributions from organizations around the University of Michigan. These organizations included the Aerospace Engineering Department at the University, student councils, and Raytheon Missile Systems who provided a $1,000 grant in support of MREA’s USLI efforts. Finally, MREA also had its twelve travelling team members contribute to defray the travel costs. Overall, MREA had ample funding to support the USLI. Income SourceCategoryAmountStarting BalanceStarting Balance1956.67UM Aerospace Engineering Dept. GrantGrants2500.00Raytheon Company GrantGrants1000.00UM Central Student Government GrantGrants700.00University of Michigan Engineering Council GrantGrants1000.00Team Travel ContributionsRefund500.00?TOTAL7851.67Table SEQ Table \* ARABIC 2: Project Income Sources6.2 Project Budget DetailsIncluded in the following tables are details on the individual components that were purchased for the Structural, Controls Structural, Avionics, and Propulsion subsystems. Additionally, outreach project expenses and travel expenses are listed. Structural ComponentsSupplierUnit Cost (USD)QuantityTotal Cost (USD)Upper Body TubeApogee Rocketry56.95156.95Outer Fuselage/Side CansApogee Rocketry26.955134.75Can CouplerApogee Rocketry9.25437.00Upper/Lower Body TubesApogee Rocketry56.952113.90Misc. Lab SuppliesAce Hardware93.05N/A93.05Blue Tube Coupler 2.56"Apogee Rocketry9.25218.50Blue Tube Coupler 5.5"Apogee Rocketry18.95118.95Main ParachuteLOC Precision189.001189.00PlywoodHome Depot$34.97N/A34.97Home Depot SuppliesHome Depot25.51N/A25.51RadioShack SuppliesCircuit City31.17N/A31.17???TOTAL722.58Table SEQ Table \* ARABIC 3: Structural Subsystem Component CostsControls Structural ComponentsSupplierUnit Cost (USD)QuantityTotal Cost (USD)Flat WasherMcMaster-Carr3.3713.37Lock WasherMcMaster-Carr4.0314.03ScrewMcMaster-Carr9.8019.80RodsMcMaster-Carr6.2116.21Hex LocknutsMcMaster-Carr5.0115.01Hex NutsMcMaster-Carr4.4914.49Drag Flap Thrust Plate (Ring)McMaster-Carr51.30151.30Avionics BayApogee Rocketry38.95138.95HS-5645MG Digital TorqueServoCity46.99146.99Voltage Regulator ResistorMouser Electronics0.19122.28DiodeMouser Electronics0.3462.04GearsMcMaster Carr36.483109.44EyeboltsMcMaster-Carr2.7625.52???TOTAL289.43Table SEQ Table \* ARABIC 4: Controls Structural Subsystem Component CostAvionics ComponentsSupplierUnit Cost (USD)QuantityTotal Cost (USD)Drag Flap Thrust Plate (Ring)McMaster-Carr51.30151.30732 Ohm ResistorMouser Elec.0.06120.72240 Ohm ResistorMouser Elec.0.13121.56300 Ohm ResistorMouser Elec.0.13121.56CapacitorMouser Elec.0.31123.72CapacitorMouser Elec.0.981211.76Arduino Female Terminal ConnectorsMouser Elec.2.0648.24Accelerometer Voltage RegulatorMouser Elec.0.6063.60Arduino Voltage RegulatorMouser Elec.1.70610.20ConnectorsMouser Elec.1.0566.30Accelerometer Sparkfun Elec.28.95128.95LED IndicatorMouser Elec.1.3922.78LED IndicatorMouser Elec.0.5921.18LED MountMouser Elec.0.2641.04Fixed Terminal blockMouser Elec.0.4141.64Public Missile Fin-D-07Public Missiles19.15476.60Key SwitchesMcMaster-Carr12.95225.90E-Charge CanistersApogee Rocketry10.00550.00E-Charge CanistersApogee Rocketry9.99549.95Perfect Flight AltimeterPerfect Flight67.96167.96Voltage Regulator ResistorMouser Electronics0.19122.28DiodeMouser Electronics0.3462.04Micro-SD Card Breakout BoardAdrafruit Industries15.00115.00Steel Bore Miter GearMcMaster-Carr18.58355.743"x6"x3" Aluminum BlockMcMaster-Carr44.12144.12Servo HubServo City9.9919.99Coupler HubServo City7.9917.99ScrewsServo City0.3541.40Key SwitchesMcMaster-Carr6.65426.58Drag ComputerMouser Elec.29.95129.95???TOTAL600.05Table SEQ Table \* ARABIC 5: Avionics Subsystem Component CostsPropulsion ComponentsSupplierUnit Cost (USD)QuantityTotal Cost (USD)Motor Mount TubePublic Missiles16.50116.50Motor Mount Centering RingsLOC Precision 7.00428.00Motor RetentionApogee Rocketry52.00152.00Motor Thrust PlateMcMaster-Carr12.47112.475 Grain Motor Retention Giant Leap Rocketry203.451203.45Pro 75 Case SpacerGiant Leap Rocketry24.95124.95Flight Test Motor #1Loki160.001160.00Flight Test Motor #2Cesaroni239.541239.54Flight Test Motor #3Cesaroni240.001240.00USLI Launch MotorCesaroni240.001240.00???TOTAL1216.91Table SEQ Table \* ARABIC 6: Propulsion Subsystem Component CostsOutreach Project ComponentsSupplierUnit Cost (USD)QuantityTotal Cost (USD)Nose ConeLOC Precision14.65573.25Payload BodyLOC Precision6.25531.25Tube CouplerLOC Precision2.85514.25Main BodyLOC Precision6.25531.25Parachute 36"LOC Precision20.955104.75Motor MountLOC Precision4.75523.75Centering ring(2)LOC Precision5.80529.00Bulkhead assemblyLOC Precision2.85514.25Rail ButtonsLOC Precision5.25526.25Shock ChordLOC Precision6.25318.75G-10 FiberglassGiant Leap Rocketry17.99589.95F-Class Motors Apogee42.973128.91Rocket Fair DisplayHome Depot18.76118.76???TOTAL604.37Table SEQ Table \* ARABIC 7: Outreach Project CostsTravel ExpenseUnit Cost (USD)QuantityTotal Cost (USD)Car Rental300.002vehicles600.00Gas0.242000miles482.35Hotel 272.225nights 1361.08???TOTAL2443.43Table SEQ Table \* ARABIC 8: Travel Costs ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download