Datasets:

ai-aerospace

/

ams_data_full_2000-2020

like 0

SoftRide Vibration and Shock Isolation Systems that Protect Spacecraft from Launch Dynamic Environments Conor D. Johnson*, Paul S. Wilke* and Scott C. Pendleton* Abstract Reduction of the vibration and shock loads seen by spacecraft during launch greatly reduce the risk that the spacecraft and its instruments will be damaged during their ascent into orbit, and also allow more sensitive equipment to be included in missions. Protecting the satellite from these loads by whole- spacecraft vibration and shock isolation systems has now been demonstrated. The basic concept of whole-spacecraft isolation is to isolate the entire spacecraft from the dynamics of the launch vehicle. This paper discusses two different systems: the SoftRide system and the ShockRing system. This paper discusses each of these types of systems and presents flight data that demonstrates their effectiveness. Introduction Satellites are perhaps among the most amazing products in use today, used for many purposes from communications to reconnaissance to weather prediction, and much more. Like all other products, satellites undergo design, fabrication, test, and shipment. However, the shipment of a satellite to its final destination in orbit is far more complicated than for all other products. Since the launch of the world's first satellite in 1957, the capability and reliability of launch vehicles have improved dramatically. What has not improved in 48 years of launching satellites is the launch vehicle-induced vibration and shock environment that a satellite must endure on its trip to orbit. Excessive dynamic and shock loads can be a satellite killer causing permanent damage to electronics, optics, and other sensitive equipment. To compensate for the harsh dynamic environment, payloads must be designed and tested to very high dynamic levels, greatly increasing the cost of many payload components. An excellent alternative is to reduce the launch dynamic loads through the use of whole-spacecraft passive vibration isolation. Whole-s acecraft vibration isolation has been developed to attenuate dynamic loads for some launch vehiclest2, has been successfully flown several times, and is in development for other vehicles and loading conditions. Whole-spacecraft vibration isolation systems can be discussed as systems that significantly reduce the dynamic loads in both the low frequency range (coupled loads analysis range) and in the high frequency range (shock loading range) or systems that only reduce the high frequency shock loads (Figure 1). W hole-spacecraft shock 001 I I I 10 Hz 70 Hz Figure 1. Frequency ranges for attenuation for shock and vibration isolation systems W hole-spacecraft vibration isolation systems have typically been designed to date to attenuate launch dynamic loads from about 12 Hz and upward. This is very useful for mitigation of vibration loads on launch * CSA Engineering, Inc., Mountain View, CA Proceedings of the 38'" Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 89

Document

AMS_2006.pdf

vehicles and also functions to isolate higher frequency shock loads. Whole-spacecraft shock isolation systems, on the other hand, are being designed to attenuate launch shock loads from about 70 Hz and upward. Most damaging shock loads have their most significant magnitude in the 1000 Hz and upward frequency range, however dynamic loads between 100 Hz and 1000 Hz are still considered to be shock loads and this frequency range has been known to be a real source of problems for some launch vehicles. So the logical question is: Why not always use a vibration isolation system, as opposed to a shock isolation system, and get the isolation benefit from the lowest possible frequency? The answer lies in the relationship between the isolation system, the launch vehicle guidance, navigation, and control (GNC) system, and coupled loads analysis. The whole-spacecraft vibration isolation system is lower in frequency and stiffness than a shock isolation system and is carefully sized, for each mission, using coupled loads analysis. The first bending modes of the satellite are typically reduced by the introduction of a vibration isolation system and therefore must be designed in concert with the GNC system such that control instabilities are not introduced. Again, this is possible, and has been done successfully on several flights. The whole-spacecraft shock isolation system, on the other hand, is relatively high in frequency and stiffness and has little or no effect on the GNC system or on coupled loads analysis. For missions that do not anticipate any problems with lower frequency vibration loads, a shock isolation system will be easier to include with only minimal effort. Under a number of contracts from the Air Force Research Laboratory, Space Vehicles Directorate, CSA Engineering has been working on the concept of whole-spacecraft vibration isolation and shock systems (hereinafter referred as the SoftRide system) since 1993. A number of design and performance analyses were performed on a variety of liquid-fueled and solid-fueled launch vehicles, all of which showed great promise. However, it was not until the launch of the GFO spacecraft on Orbital Science’s Taurus launch vehicle in February 1998 did an isolation system designed to vibration-isolate the complete spacecraft actually fly. Since that time, two different types of systems have flown, and design work has been performed on several additional !aunch vehicle/spacecraft combinations. The following sections discuss each type of system, show hardware pictures, and present flight results. SoftRide Vibration Isolation Systems Typical vibration isolation systems work by connecting the isolated structure (payload) to the base structure (launch vehicle) by means of a resilient mount or mounts. The resilient mounts have low relative stiffness as compared to the base and payload, and some degree of structural damping. The stiffness of the resilient mounts is tuned so that the frequency of vibration of the supported payload on the resilient mounts is a specified value (isolation frequency). Damping in the resilient mounts reduces the amplitude of response of the payload at the isolation frequency when the system is under external excitation. The resilient mounts must allow relative motion between the vibrating base structure and the payload at the isolation frequency, which is referred to as the isolator stroke. Because the spacecraft is a major structural component of the launch vehicle/spacecraft dynamic system, variations in the isolation frequencies greatly effect the dynamics of the launch vehicle/spacecraft system. Any unpredicted changes in the dynamics could have an adverse effect on the control system of the launch vehicle and cause instability and thereby loss of the mission. Therefore, the stiffness properties of the isolation system must be predictable for the duration of the flight. This requires a linear isolation system under all load cases, including preloads from -2g’s to +6g’s accelerations of the launch vehicle. This eliminates using an elastomeric material (Le., rubber mounts) as the stiffness component of the isolation system. Owners of spacecraft, which costs tens to hundreds of millions of dollars, demand a metallic connection between the spacecraft and the launch vehicle. This connection, which is the SoftRide system, must also provide a fail-safe connection, must be able to handle, without overstressing, the deflections due to the sum of the dynamic and quasi-static acceleration loads of the spacecraft, and must be of minimal height (reduces payload volume) and weight (reduces payload weight). On expendable launch vehicles, spacecraft are attached to the launch vehicle at their base either at discrete points or by a band clamp. If the attachment stiffness is made soft in the axial or thrust axis, then we refer to that type of isolation system as an axial system. Axial systems can provide isolation in the axial and two rocking directions and therefore can isolate against both axial and bending modes of the launch vehicle. If the attachment stiffness is made soft in the in-plane directions at the attachment points, then 90

Document

AMS_2006.pdf

that type of isolation system will be referred to as a lateral or shear isolator. Whole-spacecraft vibration isolation systems may also be a combination of these. The SoftRide whole-spacecraft vibration isolation systems have flown on two different launch vehicles to date: Orbital Science Corporation’s Taurus launch vehicle and the Air Force Minotaur launch vehicle (Figure 2). This paper discusses axial and combined axial + lateral SoftRide systems designed for and flown on these launch vehicles. Even though these systems were designed to reduce transient vibration loads below 80 Hz, they performed extremely well at reducing high- frequency loads. W hole-spacecraft vibration isolation systems are now offered as a launch option in the Taurus Launch System Payload User’s Guide’ and in the Minotaur Payload User’s Guide2. Minotaur is a four-stage, ground-launched solid propellant, inertially guided spacelift vehicle from Orbital Sciences Corporation. It uses the first two stages from the Minuteman II intercontinental ballistic missile (ICBM) combined with the upper two stages, structure, and fairing from the orbital Pegasus XL air-launched space vehicle3. Figure 2. I rie I uurus and Minotaur launch vehicles A -rL, Passive Whole-Spacecraft Vibration Isolation Systems Two types of passive whole-spacecraft vibration isolation systems have been flown. These are (1) a patented uniaxial damped flexure system called SoftRide UniFlex, and (2) a patented multi-axis damped flexure system called SoftRide MultiFlex. These systems are intended to attenuate low-frequency launch vibration loads from about 20 Hz and higher. The SoftRide vibration isolation systems seek to reduce dynamic loads on a payload by blocking the transmission of dynamic loads present in a base structure to which the payload is attached. The design of classical vibration isolation systems typically assumes that the base is rigid and the isolated payload has dynamics only well above the isolation frequency. Contrary to this, the design of a SoftRide system must be done with full knowledge that the structures on either side of the isolation system, namely the launch vehicle and the spacecraft, are both very rich in dynamics. This necessitates that the SoftRide system must be approached from the perspective of system-level dynamics. Some of the typical design constraints are weight, volume, and strength. Two other major constraints on the design of the isolation systems are: 0 Do not introduce excessive spacecraft to fairing relative displacement. Do not introduce modes that are too low in frequency or high in amplitude such that they interfere with the LV attitude control system. The design of the isolation system therefore requires coupled-loads analysis (CLA), along with detailed design analysis. The basic procedure involves the following steps: 0 Preliminary CLA with worst load cases to optimize system-level isolator performance and get component-level requirements 0 0 Isolator concept design to meet component-level performance requirements Isolator loads analysis to determine design loads for isolator strength design 91

Document

AMS_2006.pdf

Isolator detailed design to arrive at a design that meets all strength and performance requirements Complete CLA using final detailed isolator models in the system model to verify system-level performance The CLA must be performed with actual launch vehicle and spacecraft models. The typical procedure at CSA is to obtain LV models and loads for worst case conditions from the LV manufacturer and perform CLA with the latest model of the spacecraft supplied by its manufacturer. Once the detailed isolator design analysis is completed, then a model of the isolation system is delivered to the LV manufacturer for a complete and final CLA. The following sections describe the isolation systems and present flight telemetry data. H.LW*d.IRnpanr SoftRide UniFlex The patented SoftRide UniFlex whole-spacecraft vibration isolation system is intended to reduce dynamic launch loads that are predominantly axial (thrust-direction) in P..tRmo-O 14 -1 nature. The design procedure requires CLA to accurately predict the responses of the spacecraft with an isolation system. As an example, CLA showed that the isolation system significantly reduced spacecraft responses due to the resonant burn load. For example, the spacecraft net C.G. response in the axial direction was reduced by a factor of seven by using the isolation system (Figure 3). I L I Figure 3. Spacecraft axial C.G. response, resonant burn load I The stiffness and damping of the isolators are sized to mission-specific requirements for reduction of these dynamic loads. This system consists of a set of damped flexure elements that connect the spacecraft to the launch vehicle. Figure 4 shows a UniFlex isolator. This consists of a titanium flexure and a constrained layer damping treatment. The metallic load path of this isolator allows a strong, predictable, stable connection between the spacecraft and the launch vehicle. The damping treatment provides sufficient damping to control resonant amplification of loads. The typical application of this isolation svstem is to replace each bolt at a Figure 4. SoftRide UniFlex and its installation SoftRide MultiFlex 92

Document

AMS_2006.pdf

The patented SoftRide MultiFlex whole-spacecraft vibration isolation system is intended to reduce dynamic launch loads that are both axial (thrust-direction) and lateral in nature. Again, CLA is required to design the MultiFlex for the particular LV / spacecraft. Sample results for the design analysis showing responses non-isolated and isolated is given in Figure 5. The stiffness and damping of the isolators are sized to mission-specific requirements for reduction of these dynamic loads. Similar to UniFlex, this system consists of a set of damped flexure elements that connect the spacecraft to the launch vehicle. Figure 6 shows a MultiFlex isolator. -7 hM M WI(U - -bh YI-7 -YII 0s. w OUR.- --, I-- , -,- i - r-- :i1 i: to the SoftRide system Figure 6. SoftRide MultiFlex and its installation This consists of a pair of UniFlex isolators separated from one another by a central post. The axial isolation is achieved by virtue of the UniFlex isolators in ries with one another. The lateral achieved by the shearing of the assembly with bending oc rring in the flexures. The typical ap lation system is to replace each bolt at a field joint with a MultiFlex isolator element, a Figure 6. Flight Heritage of Whole-Spacecraft Vibration Isolation There is significant flight heritage for whole-spacecraft vibration isolation. These systems have, to date, flown on six separate missions. Flight telemetry data indicating the flight performance of the isolation systems is available from all missions except the MightySat mission and will be presented in the following sections. Table 1 summarizes the missions on which SoftRide has flown. 93

Document

AMS_2006.pdf

side of the isolation system. The remaining spacecraft histories. The isolation system significantly reduces the -151 0 10 XI 30 40 50 60 70 60 R58Z55-hlrrllliS vibration level to the payload by 50% for all load events. Time (sec) - Above iwlaton Figure 7. GFO flight data - below and Taurus 3 / STEX .. The Taurus/STEX SoftRide isolation system was very similar to that of GFO but "tuned for this mission. The sm MDh. meIdWMl2 H2 wjpn ''1 I--&hVbhtm II STEX spacecraft was heavier than the GFO and therefore the isolation system was larger. With one successful flight of this system, the program offices allowed a slightly more aggressive design (lower in frequency) to be flown. Finite element models of the LV - and spacecraft were obtained and full coupled-loads analyses were performed to design the isolation system. While the first mission (TaurudGFO) required both component-level and system-level testing of the isolation system, only component-level tests were For the Taurus/STEX mission, data from two accelerometers, again one below and one above the isolators, was obtained. An overplot of this data is shown in Figure 8 (this data has been high-pass filtered to eliminate the quasi-static accelerations). This data shows a factor of five reduction in the broadband acceleration levels above the isolators. 10- ,. i a-- -I- performed on the TaurudSTEX system. -10 - 0 io Za sa a 5n e4 B57E-hn-1888 -lo Figure 8. STEX flight data - below and above isolators It is of great interest to examine the performance of the SoftRide isolation system in the frequency domain. This allows inspection of the broadband attenuation characteristics of the SoftRide system. The dynamic system made up of the launch vehicle and spacecraft is non-stationary due to continual propellant depletion and stage separations. Also, the highly transient nature of most launch load events precludes digital signal processing of the flight data averaged over the entire launch window. Therefore, the frequency content of the transient flight data is best observed by creating waterfall PSD plots. These plots show the PSDs of 2-second windows of transient data, overlapped by 1 second, and stacked up next to each other. Figure 9 shows the waterfall plots for the axial acceleration below and above the isolators from the STEX flight. Note that the sample rate of 4000 Hz only allows data to be examined up to 2000 Hz. Examination of these plots shows that the SoftRide system provided significant reductions in the acceleration levels across the broadband spectrum. The high frequency accelerations below the isolators may be due to structural-borne acoustic energy. The SoftRide system has greatly reduced the structural-borne acoustic vibration on the spacecraft. 94

Document

AMS_2006.pdf

Figure 9. Waterfall PSD of STEX data - below and above the isolators Taurus 5 / MTI obtained below and above the isolators in both the axial and the radial directions. The instrumentation AblPn IMlstaS and data processing were done similarly to the GFO and STEX missions. Transient data and waterfall PSD plots for the axial direction are similar to the previous flight data. Transient data plot for the radial direction is shown in Figure 10. Note that the UniFlex axial direction, but also provides significant reduction in dynamic responses in the radial direction. For the TaurudMTI mission, flight telemetry data was MTI ngm w.. FIW stape. Ramd 15 ~ wowlsdalom 5 Lateral isolation system not only provides attenuation in the -* r ' I 10 20 30 40 50 Tim (sac) -15' 61 Figure 10. MTI flight data - below and above isolators Taurus 6 / QuickTOMS & OrbView4 The Taurus 6 /QuickTOMS & OrbView 4 mission (see Figure 11) was the first dual mission where both satellites were protected by separate SoftRide systems. Flight telemetry data was obtained below and above the isolators in the axial direction for both satellites. Transient data plots for the axial direction are shown in Figure 11 for both QuickTOMS and OrbView4. This data shows that the isolators provided excellent response reductions, similar to other Taurus flights. 95

Document

AMS_2006.pdf

Figure 11. QuickTOMS and OrbView4 satellites in the launch configuration and flight data Minotaur 1 / JAWSAT The Air Force funded a significant data acquisition system on Minotaur for the purpose of assessing the performance of the SoftRide vibration isolation system. A total of 18 accelerometers were flown for measuring accelerations on both the launch vehicle side and the spacecraft side of the isolation system. These accelerometers were arranged into 6 triaxial sets: three triaxial sets on the hard side and three triaxial sets on the soft side. Flight data was examined and the trends observed agreed very well with the predictions of coupled loads analyses. An example of some SoftRide acceleration flight data from the JAWSAT mission is shown in Figure 12. Figure 12. Typical SoftRide flight data from the Minotaur/JAWSAT mission The auasi-static acceleration measurements have not been filtered out of this data. Note that excellent vibration isolation was achieved in both the axial (thrust) and the lateral directions. Data showing the fairing separation shock event from the Minotaur/JAWSAT flight is shown in Figure 13. The flight accelerometers were not shock accelerometers and therefore some clipping of the high-level “hard side’’ shocks has occurred. However, the isolated “soft side” shows greatly reduced shock inputs to the base of the spacecraft. Minotaur 2 / MiahtvSat The MinotaurIMightySat mission was the second flight of the Minotaur launch vehicle. The auxiliary data acquisition system for collecting SoftRide performance data was not flown on this mission so telemetry data is not available. The launch was a complete success and the MightySat spacecraft, after its soft ride to orbit, began operation as planned. .m 1 I I 116h I188 1111 117 1172 117A 1176 1178 116 m* (-1 I<Dlrn*rn Figure 13. Fairing separation shock flight data showing SoftRide attenuation SoftRide Shock Isolation Systems (ShockRing) 96

Document

AMS_2006.pdf

W hole-spacecraft shock isolation systems have been designed, analyzed, tested, and flown and others are currently in development. These isolators are optimally located in the stack just aft of the satellite in order to attenuate all shock loads from the launch vehicle. Candidate locations include (1) at the top of the payload attach fitting (PAF), just below the satellite separation system (Figure 14), (2) at the bottom of the PAF, or (3) integrated somewhere within the PAF. One patented design for a whole-spacecraft shock isolation system is shown in Figure 14. This is a continuous ring made of a series of highly damped flexures. The designed-in compliance, the high damping, the contorted shock path, and the assembly joints all combine to make this an effective light-weight isolation system. This isolator design, along with several other designs that are in development, have been tested and results will be presented in the following discussion. Flight results from the first flight are also presented. Figure 14. Whole-spacecraft shock isolator at top of PAF W hole-spacecraft shock isolation systems are currently under development for the purpose of attenuating shock inputs from the launch vehicle to the spacecraft. The major source of these shock inputs is typically fairing separation shock, dual payload attach fitting (DPAF) separation shock, stage ignitions, and stage shutdowns. While UniFlex and MultiFlex vibration isolation systems are tailored to mission-specific requirements for low frequency isolation, the shock isolation system is planned to be more of an “off-the-shelf” component. It is envisioned that, for each class of launch vehicle, the shock isolator will be a \ I riyuic IP. ratciircu wiiui~~pa~.cc.iai~ 3iiuc.n isolation systems (ShockRing) “couple sizes fits all” type of system for the purpose of attenuating launch dynamic loads from frequencies of about 70 Hz and higher, depending on the design. Several patented shock isolators are shown in Figure 15. The largest is over 180 cm in diameter. 97

Document

AMS_2006.pdf

During the initial development phase of the whole-spacecraft shock isolation system, prototypes were IA I, d -m’ -nm,I-hg.-y..Dllm TmMiat 4 e a ’5 0.I 0, 0,s D1 o_ 01 TI?. -1 Figure 16. Transient responses and shock response spectra from impact test fabricated and shock tested using a pneumatic impact gun. The shock isolator was attached to two rigid steel blocks and suspended from a test frame. The impact occurred on the steel block referred to as the “base” and the accelerations were measured on both the base and the “payload steel blocks. Acceleration time histories and their corresponding shock response spectra for a typical test are shown in Figure 16. Above the isolation frequency of this shock isolator (100 Hz in the axial direction and 250 Hz in the lateral direction) over a magnitude of attenuation is achievable. This testing is very useful for development of shock isolators. However, this type of testing is missing two essential ingredients to prove the shock isolator’s ability to attenuate shock loads. The launch community will place more credibility on the shock isolator testing if it includes (1) flight-like pyrotechnic excitation and (2) flight-like flexible adjoining structures as opposed to rigid blocks: - Shock tests were subsequently conducted using primacord for pyrotechnic excitation, launch vehicle components, and a spacecraft emulator. The amount of primacord was experimentally adjusted until flight-like shock acceleration levels were measured at the spacecraft interface. Accelerations were measured in all directions at several locations. Figure 17 shows acceleration time histories and shock response spectra from the test of a whole-spacecraft shock isolator. Data is shown for accelerometer locations both forward and aft of the isolator. The excellent attenuation performance of the shock isolator can be seen in both the time and frequency domains. 98

Document

AMS_2006.pdf

m s lo 1 0' 1 0- 1 0' - - m f 10) t d lol 1 on 10.) 10' 1 0' Figure 17. Pyrotechnic shock test results of a shock ring VALPE Fliaht ExDeriment The Air Force Research Laboratory and the Air Force Space VALPE, Vibro-Acoustic Launch Protection Experiment. VALP reduction techniques on board two flights of a NASA sounding and hybrid (passive-active) whole-spacecraft vibration isolation was one component of this program. For the first flight, only a passive SoftRide ShockRing isolation system was flown. For the second flight, an active vibration isolation system was also implemented using the ShockRing as the passive stage. This launch vehicle produces very high quasi-static acceleration loads (up to 20 g's) and very high dynamic loads (up to 40 g's). Therefore, designing the isolation system for strength while maintaining the required flexibility was a challenge. The first flight was launched from NASA Wallops Flight Facility at Wallops Island, VA in November 2002. Figure 18 shows the flight one payload hardware prior to integration into the launch vehicle. Due to the small payload envelope for this experiment, the ShockRing was a compact design. The ShockRing, mounted just below the payload electronics, has a diameter of 32 cm and a height of 4.5 cm. The ShockRing was designed to protect the mission Test Program sponsored a program called E demonstrated several vibration/acoustics rocket, the Terrier Improved-Orion. Passive specific electronics for monitoring the experiment launch environment. Figure 19 shows the launch vehicle and the lift-off. 99

Document

AMS_2006.pdf

L 5 i ___. Figure 19. VALPE Terrier Improved-Orion on launch rail and November 02 launch Flight data was recovered by telemetry from the entire flight. Flight data for the lateral direction for the second stage burn is shown in Figure 20, where the data in red is below the isolation system and data in yellow is on the payload side. This data shows that the ShockRing reduced the acceleration levels even in the lateral direction, which is not as compliant and highly damped as the axial direction. Figure 20. VALPE flight data from last 20 seconds of flight, below and above the isolation system Falconsat-3 In 2006, under a program sponsered by the Air Force Research Laboratory, a whole-spacecraft ShockRing developed for the Air Force Academy's FalconSAT-3 satellite will provide high frequency attenuation during launch on an Atlas V launch vehicle. FalconSAT-3 is a 50 kg experimantal satellite that will be one of the first secondary payloads to fly on ESPA. FalconSAT-3 is a unique satellite that has a boom that protrudes into the center of ESPA and through the center of the ShockRing. The ShockRing provides an open center for the FalconSAT-3 boom, attenuate in both the satellite axial and lateral direction,s and provides a rigid connection to ESPA. The ShockRing isolation system adds just over five centimeters to the stack height of the system and just over 3 kg to the assembly. Figure 21 below shows the FalconSAT-3 qualification model during testing with the ShockRing mounted directly underneath. Figure 22 is test data from system level vibe tests performed with and without the ShockRing isolation system. Acceleration data was recorded at the top of the satellite during an axial direction high level random test. Overall grms levels were reduced from 21 grms to 8 grms. Currently CSA is on schedule to deliver flight hardware to the Air Force Academy in early 2006. 100

Document

AMS_2006.pdf

I - 1 IYUI G 1L I. YUc.IlIl~UbIWII LGSU VI F. ShockRing stack on the shaker 10- Urns wlth ShockRirp7.7858 IO' Figure 22. Axial direction test data with and without the ShockRing I 1 1 o2 1 o3 Frsquancy (W Conclusion There is a need to reduce launch loads on spacecraft so that spacecraft and their instruments can be designed with more concentration on orbital performance rather than launch survival. A softer ride to orbit will allow more sensitive equipment to be included in missions, reduce risk of equipment or component failure, and possibly allow the mass of the spacecraft bus to be reduced. These benefits apply to military as well as commercial spacecraft. For all of the missions flown to date, the patented SoftRide UniFlex, MultiFlex, and ShockRing whole- spacecraft vibration and shock isolation systems have proved to be a very effective means of reducing spacecraft responses due to the broadband structure-born launch environment. From both the transient data and the waterfall PSDs, it is clear that the SoftRide whole-spacecraft vibration isolation systems performed very well to reduce structure-borne vibration levels transmitted to the spacecraft. The isolation system was designed specifically to reduce the effects of solid motor resonant burn in the 45 Hz to 60 Hz frequency range, which it did very well. It should also be noted that the SoftRide vibration isolation system provided extreme reductions of shock and structure-borne acoustics at higher-f requencies. The isolation system hardware design was elegant in its simplicity, which ultimately played a great part in its acceptance by both the spacecraft and launch vehicle manufacturers. The SoftRide isolation systems are simply inserted at an existing field joint. No flight hardware changes were required. The only change was to the guidance and control algorithms to account for bending frequency changes introduced by the isolation system. In the end, the choice to fly the isolation system proved to be a tremendous risk- reduction for the spacecraft by drastically increasing the spacecraft margins. Because of the success of these flights, this isolation system design is being used on several upcoming flights. 101

Document

AMS_2006.pdf

References 1 . Orbital Sciences Corporation, “Taurus Launch System Payload User’s Guide”, Release 3.0, September, 1999 2. Orbital Sciences Corporation, “Minotaur Payload User’s Guide”, Draft version, April 26, 2001 3. Schoneman, S., Buckley, S., et. ai, “Orbital Suborbital Program (OSP) “Minotaur” Space Launch Vehicle: Low Cost Space Lift For Small Satellites Using Surplus Minuteman Motors”, AlAA Paper AIAA-2000-5068, AlAA Space 2000 Conference, September 19-21,2000, Long Beach, CA. 102

Document

AMS_2006.pdf

Summary of the New AlAA Moving Mechanical Assemblies Standard Brian W. Gore’ Abstract A new American Institute of Aeronautics and Astronautics (AIAA) standard entitled “Moving Mechanical Assemblies for Space and Launch Vehicles,” AIAA-S-114-2005, has been created. It is based on Military Specification MIL-A-835776, which was cancelled by the Department of Defense in the mid-1 990’s. The new standard supersedes a Technical Operating Report (TOR) (prepared by Brian W. Gore of The Aerospace Corporation with support from the Air Force Space and Missile Center and the National Reconnaissance Office, which was a “cleaned-up,’’ same-format version of MIL-A-83577B) and has already been used as a compliance document in several recent acquisitions and Requests For Proposals. This paper outlines some of the more significant changes and additions made in the new AlAA MMA Standard since the previous TOR and MMA specification were released. Introduction The U.S. Air Force (USAF) Space and Missile Center (SMC) and the National Reconnaissance Office (NRO) have recently established policies supporting and requiring government, industry, and professional society specifications and standards for new acquisitions. NAn S414-20@5 -1 Standard Moving Mechanical Assemblies for Space and Launch Vehicles The Aerospace Corporation, El Segundo, CA Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 103

Document

AMS_2006.pdf

The new AlAA standard, AIAA-S-114-2005, is one of the first products of the recent Specifications and Standards Revitalization Program undertaken by the USAF’s SMC. They asked the AlAA to engage appropriate subject matter experts to develop five consensus standards to be used as compliance documents for SMC (and potentially NRO) acquisition activities. These five standards were: 1. AIAA-S-110-2005, “Space Systems - Structures, Structural Components, and Structural Assemblies 2. AIAA-S-111-2005, “Qualification and Quality Requirements for Space-Qualified Solar Cells” 3. AIAA-S-112-2005, “Qualification and Quality Requirements for Space-Qualified Solar Panels” 4. AIAA-S-113-2005, “Space Systems - Criteria for Explosive Systems and Devices Used on Launch and Space Vehicles” 5. AIAA-S-114-2005, “Moving Mechanical Assemblies for Space and Launch Vehicles” Three other standards are currently beginning the same development process, as of the time of publication of this paper; these involve 1. Space System Electrical Power System Design, 2. Space System Electromagnetic Compatibility Requirements, and 3. Space System Mass Properties Control A memorandum from USAF Lt. Gen. Brian Arnold to SMC dated 14 January 2003 and titled “Policy Letter on Specification and Standards Usage at SMC outlines the framework for using specifications and standards as a key part of their acquisition, contracting, and program management. Some key excerpts from that memo include: “The unintentional result (of reducing the use of specifications and standards as compliance documents through “Acquisition Reform”) was that technical baselines and processes were compromised.” “There is no intent to return to the pre-acquisition reform approach of using an excessive number of specs and standards. A list of high priority critical specs and standards is being.. .established for appropriate use.” “The baseline list of specs and standards will be used in a less prescriptive manner than in the past.” With these ideas in mind, the Moving Mechanical Assembly Standard was created with the utmost thought toward incorporating those requirements that are common to most MMAs for space and launch vehicles. The requirements stated are a composite of those that have been found to be cost-effective for high reliability space and launch vehicle applications. The standard is the result of contributions received from many individuals, most notably those on the AlAA MMA Committee on Standards (Cos). Although the committee started out about twice as large, at the time of approval the actively participating members of the AlAA MMA COS were: Stephen Brock, Liaison Ken Emerick, Co-Chair Brian Gore, Co-Chair Michael Pollard, Co-Chair Dave Putnam Dave Richman Paul Reynolds Bert Timmerman AlAA Space Systems/Loral The Aerospace Corporation Lockheed Martin Corporation (Denver) Lockheed Martin Corporation (Sunnyvale) The Boeing Company Northrop Grumman Corporation Hi-Shear Technology Corporation 104

Document

AMS_2006.pdf

The above consensus body approved the document and the AlAA Standards Executive Council accepted the standard for publication in June 2005. It can be downloaded for purchase ($39.95, or $31.95 for AlAA members) from the AlAA website at httu://www.aiaa.ora/content.cf m?~aueid=363&id=l366&Tvue=StoreProduct&LaverlD=51 Description The new AlAA MMA standard specifies general requirements for the design, manufacture, quality control, testing, and storage of MMAs to be used on space and launch vehicles. It is applicable to the mechanical or electromechanical devices that control the movement of a mechanical part of a space or launch vehicle relative to another part. The requirements apply to the overall MMA as well as to the mechanical components and instrumentation that are an integral part of these mechanical assemblies. Not all requirements in the standard are of equal importance or weight. They have been divided into three categories of importance, ranging from requirements that are imposed on all applications to examples of acceptable designs, items, and practices. The relative weighting of requirements is an important consideration when tailoring the standard to specific applications and in making trade studies of alternatives. Three weighting factors are incorporated in the standard: 1. “Shall” 2. “Shall, where practical” 3. “P ref e rred/S hou Id/May” Note that the old MMA specification designated four separate weighting factors, with this standard essentially combining the lowest two in the hierarchy. The use of the weighting factors in the standard is intended to assist in the tailoring of requirements to specific appiications and to assist contractors in the design process. Detailed definitions and scope of these weighting levels are described in the standard. Unlike the uniform, one-format-fits-all of many military specifications, this particular document reflects the general consensus of the industry in what is required for today’s MMAs from a design, build, inspection, and test perspective. The document was designed to be more “user-friendly” than the old MMA specification as it was consciously re-organized to flow simultaneously from general to specific, as well as along the design and development life of MMAs. To provide an illustration of this new organization, the chapters and major subheadings are listed in Table 1. 105

Document

AMS_2006.pdf

Table 1. Organization of MMA Standard 1 Scope 2 Tailoring 3 Applicable Documents 4 Vocabulary 5 General Design Requirements 5.1 Performance Requirements 5.2 Environmental Design Requirements 5.3 Physical Requirements 5.4 Electrical and Electronic Requirements 5.5 Structural Requirements 5.6 Reliability 6 Component Design Requirements 6.1 Fasteners 6.2 Retention and Release Devices 6.3 Pivots and Hinges 6.4 Cable Systems 6.5 Springs 6.6 Dampers 6.8 End-of-Travel Latches 6.9 Gearings 6.1 0 Electric Motors 6.1 1 6.1 2 Switches 6.1 3 Gears 6.1 4 Pressurized Components 7.1 7.2 Materials 7.3 Lubricants 7.4 Hard Coatings 7.5 Contamination 8.1 8.2 Test Fixtures 8.3 Test Instrumentation 8.4 Test Plans and Procedures 8.5 Development Tests 8.6 Qualification and Proto-qualification Tests 8.7 8.8 Vehicle Level Acceptance Tests 8.9 8.10 Modifications, Rework, and Retesting 6.7 stops Power and Signal Transfer Components 7 Parts, Materials, and Processes Requirements General Parts, Materials, and Processes 8 Testing and Inspection Requirements Parts, Materials, and Process Controls Component and Subsystem Level Acceptance Tests Pre-launch Validation Testing and Inspection 9 Bibliography Annex A (Informative) Static Torque or Force Margins at Different Coordinate Points 106

Document

AMS_2006.pdf

Highlighted Changes Overall flow - easier to uselfind information As stated above, it was an early and deliberate action by the COS to reformat the AlAA MMA Standard to be easier to use and find information than the previous military specification document. Before any work was done on the technical content, the co-chairs spent a notable - and ultimately worthwhile - amount of time determining the current organization, deleting obvious subject items and inserting placeholders that were later filled in regarding new areas of interest and importance. The co-chairs then rearranged the remaining text into the corresponding sections of the new format, and the committee went to work to generate and compile a list of discussion comments. It was this list that formed the focus of the COS efforts. Teleconferences on a bi-weekly, then weekly, then almost daily basis toward the end were the main forum of COS communication, augmented by a members-only, AIAA- hosted website which stored and organized pertinent and necessary documentation. ADDroximatelv 400 comments aenerated, submitted. and disDositioned. After the COS reviewed the base document and compiled the aforementioned list, they used the teleconferences and two, multiple-day, face-to-face “summit” meetings to work through the list. Several of the committee members passed the working document [initially issued as The Aerospace Corporation Technical Operating Report TOR-2004(8583)-1] to their colleagues or certain specialists to solicit additional insightful comments. The COS spent the majority of these meetings discussing the merits of the particular comments submitted, grouped as either general, editorial, or technical; a breakdown of the comments in the list is shown in Figure 1. Breakdown of Submitted MMA Std Comments %- General 8% Figure 1. Breakdown of Submitted MMA Standard Comments Moreover, after the final draft of the standard was prepared, AlAA initiated a 30-day public review period, which yielded still more comments, for a grand total of approximately 400 submissions by members of Aerospace, many contractors, NASA, etc. Figure 2 illustrates the approximate number of comments by submitting individuals. Some of the people listed were points of contact from their entire organization. 107

Document

AMS_2006.pdf

Comments by Contributor Figure 2. Number of Comments by Contributor Conciseness/deletion of DeriDheral areas There were many areas of the old MMA specification that were deemed by consensus to be out of scope for a modern moving mechanical assembly standard. Since we were no longer bound by the standard template, there was freedom to make this document more relevant overall. As such, several sections were deleted altogether in recognition that there were virtually always other documents or specifications, either governmental or internal to contractors that existed to govern those areas. Some of these subject areas included identification and marking, certain details on processes and controls, structural requirements, specific environmental conditions, etc. However, many such sections were minimized and maintained as a consideration guide for younger engineers, or simply as a reference pointer to the appropriate document, such as the new AlAA Structures Standard, MIL-STD-1540E (the environmental testing standard), etc. This scope is one of the main utility features for which the COS was striving. One metric of this conciseness may be indicated by the relative word counts of the documents, as illustrated in Table 2. Table 2. Word Count of Evolving MMA Documents TOR-2004( 8583)-1 I 24,000 AIAA-S-114-2005 19,000 Areas rechecked for modern validity There were several peripheral areas relating to MMAs that were also verified for modern validity, such as contamination, lubrication, electrostatic discharge, etc. For these particular areas, Aerospace and contractor experts were consulted to obtain an insight as to what governing documents existed for their respective disciplines. These were, in turn, referenced in the MMA standard where appropriate, leaving the details to those other documents, unlike the original MMA specification that contained many of the details. 108

Document

AMS_2006.pdf

ADDlicable documents One evident change that appears early in the document is the far fewer number of Applicable Documents called out in the AlAA standard. The only documents in this section of the standard are the ones specifically mentioned in the body of the text. In a parallel effort, each one of the documents named in the original MMA specification was validated regarding its status (active, cancelled, superseded, etc.). Many of them, if not mentioned in the text, were moved to the new Bibliography section at the end, provided the COS consensus agreed that they still contained useful information. One important note to point out is that, simply because the government decided to cancel a document during acquisition reform, it does not mean the contents of that specific document are automatically null and void. After all, the MMA specification falls into the “cancelled category and the industry still recognized it as valid, just not eligible as a contractual compliance document. Acceleration term removed from static toraue marain After significant debate, the COS decided to remove the acceleration term from the equation to calculate static torque margin. Since the true meaning of this particular margin is to show how much force exists above and beyond all static resistances, the acceleration term was deemed irrelevant. A moving mechanical assembly must have sufficient motive force to begin motion, as evaluated with a free-body diagram. Forces required for a given acceleration are deemed performance requirements, and are thus captured in the dynamic force margin requirement. New section on stepper motor marain in forcehoraue marain section One of the areas without clear definition in previous forms of this document was the lack of attention given to forcehorque margins with the use of stepper motors. A section was added which describes two ways to calculate stepper motor margin: 1. using motor available torque (pull-in torque) and comparison to friction loads, and 2. using a step stability analysis Guidelines and conditions are also given regarding which of the two methods should be used in various applications. Expansion of related electrical and electronic reauirements Although seemingly contradictive of the section above describing the deletion of peripheral areas, it was recognized that there are a significant number of eIectricaVelectronic subjects that are particularly germane for design and testing of MMAs. A short paragraph exists for each of the following subjects: Cables and Wiring Connectors Cable Supports and Strain Relief Cable Loops Current Draw Grounding Electromagnetic Interference/Electromagnetic Compatibility (EMVEMC) Electrostatic Discharge Flight Instrumentation For some subjects, little more is given than a statement of consideration to ensure an area is not overlooked, but for others, pertinent reference documents are cited, and lessons learned are described. Again, the COS was focused on MMA-related requirements, but felt at least this level of detail was justified. Additional recoanition of newer. common devices Since the last release of the MMA specification in 1988, many new technologies have been developed and/or are more widely used, and thus warranted recognition. Non-explosive devices, including wax actuators, shape memory actuators, and split spool release devices are examples of these “next- generation” items. There is actually a reduction of information in this standard regarding pyrotechnic 109

Document

AMS_2006.pdf

devices, instead deferring to the new AIAA Standard for Ordnance, AIAA-S-113-2005. Clampbands and retention cables are also included in a new section. Although these hardware items were used prior to 1988, no considerations or requirements were provided in the former MMA specification. Bearinu stresses for new steel materials The development and increased use of hybrid bearing material combinations such as Si3N4 (silicon nitride) balls and M62 (bearing steel) races required a certain level of attention. Allowable stress levels in prior documents only corresponded to 440C, but the new standard recognizes and provides requirements for 52100, M50, and M62 steels as well. Varying levels of allowable stresses for all of these materials are provided for quiet-running as well as non-precision, short duration applications. Life testinu for lonu life mechanisms In one of the more significant changes, and one that easily generated the most discussion and debate by the COS, life test requirements were revisited, motivated by the increasing duration of today’s missions as compared to those typical of two and three decades ago. A distinction is now made between “low-cycle MMAs” (such as release devices, spring driven “one-shot” deployables, etc.) and “high-cycle MMAs” (such as solar array drives, momentum wheels, tracking gimbals, etc.), with different corresponding life requirements now given for each. Run-in testinu modifications Another moderately noteworthy change in the testing requirements involves run-in testing. Run-in testing can be expressed in terms of cycles or a percentage of expected life. A slight reduction in the potential number of minimum run-in cycles has been incorporated. This was a result of some common sense being applied, particularly in the area of release devices, or other “one-shot’’ MMAs. It is generally accepted that the design life for these types of mechanisms is about 50 cycles, given the number of expected tests on the ground, plus on-orbit use(s), plus margin. The current minimum number of required cycles in the AlAA standard brings it closer to correspondence with the minimum percentage of life for these types of MMAs. “Informative” annex on static toruue/force maruins at different coordinate Doints This was the subject of a paper at the 37th Aerospace Mechanism Symposium by R. W. Postma of The Aerospace Corporation, and it was chosen to be included as an informative appendix, not necessarily subject to contractual compliance. This section describes the basic methodology overview for MMAs that have drive forces (or torques) and resisting forces (or torques) applied to mechanical elements that do not all move at the same velocity. A common example would be the ratio of the rotation of a jackscrew relative to its translation (e.g., rad/in). Summary US. Air Force (and effectively NRO) policy is swinging back to previous practices of requiring the use of specifications and standards. Subsequently, several subject matter documents have been reviewed, edited, and updated. Incorporation of those documents has started in the proposal phase of new acquisitions. One of the first new standards to be developed under these new guidelines is the AlAA Standard, “Moving Mechanical Assemblies for Space and Launch Vehicles,” AIAA-S-114-2005. It is based on the former MMA military specification, but reorganized and technically scrutinized by a committee of industry experts to reflect the current state of the art in designing, fabricating, and testing space mechanisms. This standard will begin to be required as a compliance document for future Air Force space acquisitions, and can be purchased through the AlAA website. 110

Document

AMS_2006.pdf

References AIAA-S-114-2005, Movinq Mechanical Assemblies Standard for Space and Launch Vehicles, American Institute of Aeronautics and Astronautics standard, July 2005. MIL-A-835776, Militarv Specification - Assemblies, Movina Mechanical, For Space And Launch Vehicles, General Specification For, 1988. Gore, Brian W., Movinq Mechanical Assemblies Standard for Space and Launch Vehicles (Draft 11, The Aerospace Corporation Technical Operating Report, TOR-2004(8583)-1, July 2004. Acknowledgements The author recognizes several individuals who not only made this paper possible, but were supportive of the entire effort behind its subject matter. Mr. Dave Davis of SMC/AX, through Valerie Lang, John Ingram- Cotton, and Joe Meltzer from the Aerospace Office of the Chief Engineer, provided the funding necessary to complete all of the work in getting the TOR and AIM Standard prepared and released. Ken Emerick of Space Systems/Loral and Mike Pollard of Lockheed-Martin (Denver) were loyal and admirable co-chairs on the AIAA MMA Committee on Standards. Their leadership work is to be commended. Stephen Brock of AIAA served as COS Liaison, and without him, the COS would still be trying to figure out how to use the COS website, and probably still be editing changes. Certainly not to be forgotten, Mike Hilton, Tom Trafton, Leon Gurevich, and AI Leveille (retired) of The Aerospace Corporation helped get the ball rolling in the right direction by providing some historical perspective on the MMA specification and its contents. 111

Document

AMS_2006.pdf

Lessons Learned From the Development, Operation, and Review of Mechanical Systems on the Space Shuttle, International Space Station, and Payloads Alison Dinsel', Wayne Jermstad', and Brandan Robertson' Abstract The Mechanical Design and Analysis Branch at the Johnson Space Center (JSC) is responsible for the technical oversight of over 30 mechanical systems flying on the Space Shuttle Orbiter and the International Space Station (ISS). The branch also has the responsibility for reviewing all mechanical systems on all Space Shuttle and International Space Station payloads, as part of the payload safety review process, through the Mechanical Systems Working Group (MSWG). These responsibilities give the branch unique insight into a large number of mechanical systems, and problems encountered during their design, testing, and operation. This paper contains narrative descriptions of lessons learned from some of the major problems worked on by the branch during the last two years. The problems are grouped into common categories and lessons learned are stated. Introduction The Mechanical Design and Analysis Branch at JSC is responsible for the technical oversight of over 30 mechanical systems flying on the Space Shuttle Orbiter and the ISS. The branch houses the MSWG, which has the responsibility to review all mechanical systems on all Space Shuttle Program (SSP) and ISS payloads to verify compliance with the fault tolerance requirements as part of the payload safety review process. These responsibilities give the branch unique insight into a large number of mechanical systems, and problems encountered during their design, development, testing, and operation. This paper describes some of the recent problems worked by the branch, and lessons that can be learned from them to improve future mechanical systems. The paper contains narrative descriptions of some of these problems. The problems are grouped into common categories and lessons learned are stated. The categories used are derived from the Mechanical Systems Safety memorandum [l]. The letter has 11 key design implementation and verification provisions to be followed to help ensure that credible failure modes have been reliably and effectively controlled as a result of a thorough design, build, and test process for mechanical systems. For this report, some categories will be combined, some omitted, and others expanded upon, to arrive at the list of categories: binding, jamming, and seizing; fastener locking and preload; strength; positive indication of status; and testing. Following each example, lessons learned are stated. Binding/Jamming/Seizing This provision addresses the prevention of mechanism binding, jamming, and seizing. Appropriate design features include dual rotating surfaces or other mechanical redundancies, robust strength margins such that self-generated internal particles are precluded, shrouding and debris shielding, proper selection of materials and lubrication design to prevent friction welding or galling, and others. Adequate dimensional tolerances on all moving parts are needed to ensure that functional performance will be maintained under all natural and induced environmental conditions. Tolerances associated with mechanical adjustment (or rigging) must also be taken into account. Mechanical system designs must ensure compatibility of lubricants used with interfacing materials, other lubricants used in the design, and the natural and induced environments. Designs must also ensure that appropriate quantities of lubricant are specified. * NASA Johnson Space Center, Houston, TX Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 113

Document

AMS_2006.pdf

Nose Landina Gear Udock Mechanism The Space Shuttle Orbiter’s nose landing gear, nose landing gear door, and nose landing gear uplock mechanism, shown in Figure 1, are interconnected, and must be rigged and operated together. Following replacement of the door environmental seal, rigging was performed to achieve proper seal compression. During nose landing gear cycling, the gear uplock indication did not illuminate because the mechanism did not reach the full uplock condition. Binding in the rotational fitting between the uplock fitting and the bellcrank prevented the uplock mechanism from going to the full over-center position for gear uplock. Measurements of the width of the bellcrank and the internal width of the fitting showed an interference fit between the two assemblies. Rework on the bushings per specification requirements removed the interference condition, allowing the bellcrank to move freely. Lesson Learned: Proper tolerancing and inspection are critical to preventing interferences in mechanical systems. Bungee location, not shown Environmental door seal, both doors Shock strut Fr/ Figure 1. Nose Landing Gear Mechanisms During subsequent nose landing gear retract operations, there was an early indication that the gear uplock mechanism was in the gear-up position. As the shock strut was entering the wheel well and bringing the doors closed, the gear stalled prior to being fully up and locked. After an immediate halt to operations the gear fell freely to the down position. It was observed that the uplock hook was in the gear-up position, thus preventing the uplock roller from engaging. Upon investigation, it was discovered that when hydraulic pressure was applied to retract the gear, the uplock actuator immediately drove the uplock hook closed to the gear-up position. Normally, the mechanism is in a gear-down over-center condition and cannot move prior to gear uplock roller engagement. When the gear uplock roller enters the hook, the roller pushes the mechanism out of its over-center position and allows the uplock hook to engage with the strut and bring the gear to the up-and-locked over-center position. A new source of binding in the mechanism had prevented the hook from being in the full down position. During the rework of the mechanism for the binding described above, inadvertent damage was imparted on the bungee spring. Tooling used to assist in the rework efforts described above is believed to have caused benthaised metal on the bungee end cap. The resulting binding in the bungee prevented the mechanism from freely going to the full down over-center position. A replacement bungee was installed, and rotational pins and linkages in both the 114

Document

AMS_2006.pdf

gear uplock and in the door uplock mechanisms were inspected with no signs of damage observed. The mechanism now properly “snaps” into both the gear-up and gear-down over-center positions. Lesson Learned: Repair and rework of mechanical assemblies can cause collateral damage. Video Camera SUDDO~~ Assembly The Video Camera Support Assembly (VCSA) is used to mount video cameras in any of 14 potential positions on the truss segments, nodes, laboratory module, or habitation module on the exterior of the ISS. As shown in Figure 2, the VCSA is mechanically attached to ISS primary structure with a threaded bolt that is installed by an astronaut during an extra-vehicular activity (EVA). Bolt Assembly \ Lu Figure 2. Video Camera Support Assembly During testing, the VCSA bolt experienced extremely high running torques. The recorded running torque values were higher than the Pistol Grip Tool, which was to be used to drive the bolt during an EVA, was able to generate. Investigation revealed that no lubricant thickness was specified on the bolt drawing, and that the dry film lubricant thickness on the bolt was 10 times thicker than the requirement by the application specification, causing an interference with the female threads. In addition, the surface of the bolt was not bead-blasted prior to lubricant application, as required by the application specification. To resolve the issue, the dry film lubricant was removed and the bolts were returned to print by first bead- blasting the surface, and then reapplying the lubricant. Lesson Learned: Dry film lubricant thickness is important and the appropriate surface preparation must not be overlooked. Fasteners Fasteners remain one of the most problematic areas on mechanical systems. The problems can generally be grouped into two subcategories, the first related to secondary locking features (or lack thereof), and the second related to fastener installation torque and the resulting fastener preload. 115

Document

AMS_2006.pdf

Secondary Lockincr Features For decades, dating back to the development of aircraft engines, the response of fastened joints to vibration environments has been a critical issue. Failures of critical fastened joints led to the development of “secondary” locking features that served as a method to guarantee that fasteners would not rotate and back out under vibration environments. Preload was recognized as a “good method to prevent bolt rotation, but it was neither highly reliable nor predictable. So, “positive” locking methods were developed for highly critical joints, such as lockwire, cotter pins, locking tab washers, and safety cable. These locking methods are very reliable, but are also very labor intensive and thus expensive. To decrease labor, prevailing torque locking features were developed that rely on friction to help decrease propensity for bolts to rotate. Prevailing torque locking features include lock nuts, deformed thread keenserts, helicoils, and various types of locking patches and pellets. Liquid Locking Compounds (LLCs) including epoxy have also been used as secondary locking features, but these have proven to be unreliable. LLCs are very sensitive to application process and environmental factors and cannot be verified after installation without breaking the bond. At JSC, the Engineering Directorate has been critically evaluating use of LLCs and has implemented policies restricting its use. Many of the problems with secondary locking features can be traced to a lack of well-defined requirements. Using explicit language such as this can help: “Each bolt, screw, nut, pin, or other fastener used in a safety critical application shall incorporate two separate verifiable locking features. Preload may be used as one of the features combined with a conventional aerospace secondary locking feature that is positive locking and vibration rated [2].” Lesson Learned: Clear secondary locking feature requirements need to be specified in program and project requirements documents. Over the past two years a significant number of problems have been encountered on many different pieces of hardware relating to either the lack of a secondary locking feature or the improper use of such features. External Stowaae Platform 2 The External Stowage Platform 2 (ESP-2) was a payload mounted in the payload bay of the Space Shuttle Orbiter on STS-114. ESP-2 had several space station Orbital Replacement Units (ORUs) attached to it. After docking, the ESP-2 was removed from the payload bay and mounted to the external airlock on the ISS. Following a test failure caused by migration of an uncured LLC, NASA Materials and Processes (M&P) and Structures and Mechanisms personnel advised both the ISS and Shuttle programs to closely restrict the use of LLCs and to develop application procedures for its use. As a result of the policy, an investigation of all uses of LLCs on ISS began. It was determined that an LLC had been used in some locations to assemble the ESP-2. The NASA ISS M&P System Manager advised the ISS Program that the installation procedures for ESP-2 did not call out the use of primer on the titanium inserts as recommended and that the design should be corrected by using a standard, verifiable locking feature. It was decided to attempt qualification of the hardware with the LLC as configured, while following a parallel path to fix the hardware if the test was not successful. A conservative vibration test used to generically qualify secondary locking features was adapted to the ESP-2 LLC configurations. The test was performed at lower-than-flight preloads to help isolate performance of the secondary locking feature from preload. Five samples each of five different configurations were tested. All samples were assembled using the same process as on the ESP-2 flight hardware. All 25 samples showed various degrees of rotation following testing and were therefore determined to be failures. Upon disassembly of several of the samples, it was observed under microscopic inspection that the LLC had cured, but had failed to adhere to the titanium properly. The bolts with LLC were removed and replaced with bolts that had a verifiable locking feature, a Mylar locking patch, to resolve the issue. 116

Document

AMS_2006.pdf

Lesson Learned: Liquid locking compounds are very application process-sensitive and not verifiable, and therefore are not recommended for use as a secondary locking feature. An interesting observation during the testing was that there appeared to be a relationship between fastener preload and LLC adhesion, which may warrant further study. For details, refer to JSC-62850 [3]. Resumlv Stowaqe Rack The Resupply Stowage Rack (RSR) is used to carry pressurized cargo to the space station in the Multi- Purpose Logistics Module (MPLM). The RSR consists of various sizes of locker compartments. These compartments, which accommodate individual stowage trays or bags, are bolted into the rack structure. The compartments have structural doors with latches, shown in Figure 3. Figure 3. RSR Latch Mechanism Only one month prior to the scheduled launch of STS-114, an issue was discovered with the locker door mechanism on the racks installed into the MPLM. The locker doors were held closed by a 90-degree latch that was held in the closed position by a single thumb screw. The thumb screws thread into locking inserts, but all had zero running torque due to numerous installation cycles. In addition, the engineering drawings required the thumb screws to be torqued “hand-tight”. Testing was quickly performed to determine that a maximum of approximately 0.34 N-m could be achieved, far less than a fastener of this size would be nominally torqued. The fasteners essentially had no secondary locking features, and very low preloads. One of the lockers contained a 36.3 kg component with a high-pressure tank, which created a safety concern if the fasteners backed out and allowed the component to fall out of the open locker door. Any fix to the lockers required opening the Payload Canister and the MPLM hatch, which had already been closed out for flight. Because of this, a decision was made to temporarily fix all of the lockers (about 40) with Permacel Tape. The tape worked because it did not have to prevent the loss of preload, only locker latch rotation. While not very elegant, this was an acceptable fix for STS-114. Prior to any future flights, a secondary locking feature is being added to this system. The proposed concept consists of a slider, which fits over the existing handle and prevents the thumb nut from backing out, and can be slid away to permit operation. Lesson Learned: Measure running torque during each fastener installation cycle and keep a record for verification. Replace locking features that do not exhibit running torques within design specifications. Lesson Learned: Fasteners and locking inserts are not the proper locking feature solution for latch designs requiring many cycles. 117

Document

AMS_2006.pdf

Bearina Motor Roll Rina Module SDanner Nut The Bearing Motor Roll Ring Module (BMRRM) is part of the space station Beta Gimbal Assembly (BGA), which provides mounting and gimbaling for the space station solar arrays. The BGA and BMRRM are shown in Figure 4. Figure 4. BGA and BMRRM The BGA BMRRM assembly was experiencing anomalous behavior after acceptance vibration testing: it would rotate smoothly in one direction, but would rotate slightly then seize when rotated the opposite direction. A layer of debris was also noted on the exterior of the assembly, as shown in Figure 5. Disassembly and inspection revealed that a large spanner nut, which was used to capture and preload the internal components of the roll ring and housed inside the BMRRM, had loosened and rotated out. As the nut rotated out, it closed up the clearance in a labyrinth gap as shown in Figure 5, resulting in metal to metal contact between the rotating and fixed parts of the housing components and generating the debris. I r A Lt' I1 FiS-." ". .... .. .... ---. .- and Labyrinth Gap and Spanner Nut 118

Document

AMS_2006.pdf

Pre-delivery photographs and post-installation photographs taken with a borescope were evaluated to determine if multiple units had experienced a similar loosening problem. No other units were found with this problem. A review of the drawings and build processes revealed that the nut design used a light preload and a small amount of LLC on its threads as its primary and secondary locking features. Due to a general concern for the locking design and the inability to define the exact cause of the failure, a mechanical lock was devised to fix the problem. The lock will be installed on all seven serviceable units. The three on-orbit units cannot be serviced. While the vendor had what appeared to be a properly documented and controlled application process, the LLC had failed, allowing the nut to loosen. When the situation was discussed with the LLC manufacturer, they noted that the application fell outside of the recommended usage for their product because of the large size of the spanner nut and low preload and therefore would not guarantee its function. Lesson Learned: Liquid locking compounds cannot be depended upon when used in applications outside of manufacturer’s guidelines. Pre-test, post-test, and post-installation photos were used as a non-destructive method to identify any other instances of nut movement. Since the units are not visible after being fully assembled into their next higher assembly, a borescope was used to obtain the photographs needed to clear the units. Lesson Learned: Maintain photo documentation of mechanical systems during assembly and subsequent testing and usage with similar viewing angles to allow for comparisons. Lesson Learned: Provide access for inspection of mechanical systems. Torclue/Preload A significant number of issues have been worked during the past two years relating to fastener installation torque and the resulting fastener preload. Many of these issues were the result of the improper installation torque being applied to the fasteners. Others were the result of an improper torquing sequence applied to a fastener pattern. A recurring theme has been a lack of understanding of the torque-tension relationship. InsDection Boom Assemblv and Shuttle Remote ManiDulatincl System The Inspection Boom Assembly (IBA) supports components of the Orbiter Boom Sensor System, including sensors and video cameras used by the crew for situational awareness and inspection of the Orbiter. The IBA and Shuttle Remote Manipulator System (SRMS) are mounted in the payload bay of the Orbiter. During flight the IBA is removed from the payload bay and operated as an extension of the SRMS, as shown in Figure 6. 3 Orbiter Payload Bay , 119 L Figure6. IBAo.. _...-._ An installation drawing review of the IBA handrails revealed that unusually high torque values had been specified for some of the IBA fasteners, for which the vendor was not able to supply any supporting test data. An analysis assessment showed that several fastener groups within the IBA could potentially be

Document

AMS_2006.pdf

torqued above the yield strength of the fasteners. Similarly, a stress assessment for the SRMS bolts revealed that several SRMS fasteners could potentially be torqued above their yield strength as well. A series of tests was conducted at JSC to evaluate the suitability of this condition. Torque-tension tests were performed on four different fastener groups, two from the IBA and two from the SRMS, to directly measure the relationship between torque and preload. The tests were able to show that the IBA and SRMS were acceptable to fly in their current condition. For more details refer to JSC 63083 [4]. Lesson Learned: Ensure torque tables are substantiated by relevant test data, accounting for the materials, lubricants, and installation process. Lubricants or sealants can significantly alter torque-tension relationships. Fliaht Releasable Attachment Mechanism The Flight Releasable Attachment Mechanism (FRAM) provides a generic structural and electrical interface between spare ISS hardware components, called ORUs and either the Shuttle or ISS. The FRAM consists of an active half, which is mounted on the ORUs, and a passive half, which is connected to Shuttle payload bay carriers and on-orbit stowage locations. This provides interchangeability between storage locations. After acceptance vibration testing of the FRAM, a post-test functional test and inspection revealed that a locking collar was loose and minor damage to an ACME thread was noted. The locking collar is shown in Figure 7. Figure 7. FRAM Locking Collar . It was discovered that improper torquing of the locking collar clamping fasteners resulted in a loose collar and that the ACME thread on the shaft was damaged during the testing. The fasteners had been torqued to their full level individually and were not alternated or checked a second time around to ensure that the tightening of one fastener had not reduced preload in another. Unfortunately in this case, the clamping force of the collar was very sensitive to the order and manner of torquing the fasteners. To correct the problem, the fasteners were incrementally torqued to their full torque, while alternating between the two fasteners. Lesson Learned: Use proper torque sequence in multi-fastener patterns. 120

Document

AMS_2006.pdf

Strength Mechanical system components and linkages need to be designed with sufficient strength to tolerate an actuation forcehorque stall condition at any point of travel and maintain a positive margin of safety with an ultimate factor of safety applied. Mechanical systems that incorporate end of travel mechanical stops need positive strength margins for worst case dynamic loading conditions, considering variables in inertia properties, actuation force/torque, drive train resistance, and other environmental conditions. Exposed mechanical system components, protective shrouds and covers, and mounting structure need to accommodate inadvertent impact loads from manipulator systems, payload operations, and crew activity. Main Landina Gear Door Retract Mechanism The Space Shuttle Orbiter main landing gear (MLG) door retract mechanism is shown in Figure 8. The door retract mechanism is a four bar over-center linkage, with the orbiter structure forming the fixed link. A spring-loaded bungee is also part of the mechanism and helps to hold the mechanism over-center when the door is open. Door /I Hockey stick Figure 8. MLG Door Retract Mechanism The door retract link on the starboard main door retract mechanism of Atlantis (OV-104) was found to be bent and cracked during a routine inspection. Figure 9 shows the damaged link in comparison with its counterpart on the port side. 121

Document

AMS_2006.pdf

Figure 9. Comparison of Starboard and Port 452 Links Extensive failure analysis including metallography of the failed part, historical data retrieval, dimensional verification, loads analysis, and borescope inspection of Discovery, were conducted to understand the scope of the problem. It was concluded that the damaged part had adequate design properties, and that this part was damaged during replacement of the O-ring seals in the piston axle assembly of this main landing gear strut. This ground operation involved using a hoist to support the lower part of the main gear from overhead by attaching ground support equipment to the hockey stick. The over-center link was overloaded during this procedure, causing the damage. The vehicle flew two flights in this condition. Discovery was inspected to ensure that it did not have a similar problem, and the damaged over-center link on Atlantis was replaced with a good link borrowed from Endeavour. Ground servicing procedures will be changed to prevent this problem in the future. Lesson Learned: Ensure that strength analyses are performed for planned ground operations of mechanisms, and prior to any unplanned ground operations. Positive Indication of Status All movable mechanical systems should provide positive indication that the mechanism has achieved its desired position (i.e., ready-to-latch, latched, open, closed, etc.) and end of travel stops should be provided for all movable mechanical systems. Limit Switches Limit switches are often used to provide positive indication of status for mechanisms. Limit switches are small electronic devices not capable of sustaining high mechanical loads, and are typically protected by an actuating lever mechanism. Limit switches require ground rigging to set the actuation lever in the proper location. Due to their size, there is usually a small adjustment window in which the switch will indicate the proper status. One example of an SSP mechanism that incorporates limit switches is the payload bay door drive mechanism. This system has switches that sense the position of its rotary actuators, and also indicate the location of the door. The rotary actuator limit switches are internal to the actuators, and are therefore protected from extreme thermal gradients between the hardware that is being sensed. The limit switches 122

Document

AMS_2006.pdf

that indicate the position of the door are incorporated into the bulkhead switch module, which is installed on the payload bay bulkheads, and are exposed to space. There have been numerous failures during missions with the bulkhead limit switch module and other limit switch applications. The switches do not accurately change status as the mechanism is operated, and take a few seconds to hours to flip to the proper indication. These failures do not always repeat themselves on the ground. The phenomenon is not currently understood, and despite tearing down the limit switch assembly, rebuilding, reinstalling, and ground rigging, the failures tend to repeat on orbit, and are attributed to thermal effects. Extra effort should be made to locate limit switches so they are not subjected to extreme thermal environments. In the event that the switches must be installed in these environments, redundancy should be bui!t into the limit switch system and extensive testing should be done to understand the interaction between the various components of the system under the applied thermal gradients. Lesson Learned: Ensure that thermal effects have been considered in the design and analysis of limit switches, and have been reproduced during environmental testing. External Tank Door Mechanisms Following the External Tank (ET) iettison, the ET doors are used to close out the aft areas where the Space Shuttle Orbiter was attached to the ET. The doors, shown in Figure 10, are open for launch and ascent and then must be closed while on-orbit and remained closed for entry. Figure 10. ET Doors A centerline latch mechanism holds the doors open. Once the latches have been released, the ET door drive mechanism is used to drive the doors into a nearly closed, ready-to-latch position. The uplock mechanism then operates to pull the doors into a closed position. For the ET door drive and uplock mechanisms, mechanical limit switches are incorporated into the actuators. This provides an indication that the actuator has rotated the proper amount, but does not directly indicate the status of the doors. If, for example, debris became wedged between the door and the frame, the actuator might be able to turn the correct number of degrees but the door opening might actually not be fully sealed. Given the external tank foam-shedding problem, this was a concern during the STS-114 return to flight mission. A better way to determine the true door status is being developed. Lesson Learned: Provide true indication of status for all safety critical mechanisms. 123

Document

AMS_2006.pdf

Testing Proper testing of mechanical systems is extremely important, and cannot be over-emphasized. A comprehensive test program, including run-in tests, functional and environmental acceptance tests, qualification tests, and design life verification tests, is recommended to ensure that mechanical systems operate properly on-orbit. In addition, development testing, done early with prototype hardware, is extremely valuable and has proven to be cost effective by catching problems early when there is time to fix them. One aspect of testing that is often overlooked, which will be discussed below, is the interaction of hardware inspections and functional verifications. Pavload Bav Door Mechanisms Each Space Shuttle Orbiter Payload Bay Door (PLBD) is comprised of multiple sections that are permitted to float along dry film lubricated shear pins. There are multiple requirements to inspect these expansion joint shear pins after a predetermined number of flights. At the same time, there are multiple door functional verifications for the operation of the PLBD mechanical systems (drive mechanism, bulkhead latch mechanism, and centerline latch mechanism). The shear pins are critical to the proper operation of the doors. In the event that the pins are replaced following a door functional, the door functional test must be repeated with the newly installed pins in place. Over the life of a vehicle, unanticipated actuations like these can add many cycles to mechanisms that were not necessarily accounted for during the original design phase. In an integrated vehicle of various systems, there will be some overlap between hardware inspections and functional verifications. Numerous examples of this interrelation of maintenance requirements and functional verifications exist within the space shuttle program, including the interaction between tile replacement work and the cycling of the landing gear doors, star tracker doors, air data probe doors, vent doors, external tank door, etc. Lesson Learned: Design life of mechanisms should consider maintenance cycles of interfacing systems. Lesson Learned: Processing work should be planned such that the fewest number of functional tests are performed, and cycles on mechanical systems should be tracked. Summary Several lessons learned have been presented in this report. A complete summary of the lessons are presented in Table 1. These lessons may prove to be of benefit for anyone who designs, develops, evaluates, or operates mechanical systems, regardless of the program. Categon, Binding, Jamming, i and Seizing Table 1 : Summary of Lessons Learned Lesson Proper tolerancing and inspection are critical to preventing interferences in mechanical systems. Repair and rework of mechanical assemblies can cause collateral damage. Dry film lubricant thickness is important and the appropriate surface preparation must not be overlooked. Fasteners: Secondary Locking Features Clear secondary locking feature requirements need to be specified in program and project requirements documents. Liquid locking compounds are very application process-sensitive and not verifiable, and therefore are not recommended for use as a secondary locking feature. 124

Document

AMS_2006.pdf

for verification. Replace locking features that do not exhibit running torques within - design specifications. Fasteners and locking inserts are not the proper locking feature solution for latch designs requiring many cycles. Liquid locking compounds cannot be depended upon when used in applications outside of manufacturer’s guidelines. Maintain photo documentation of mechanical systems during assembly and subsequent testing and usage with similar viewing angles to allow for comparisons. Provide access for inspection of mechanical systems. Strength Positive Indication of Status Ensure that strength analyses are performed for planned ground operations of mechanisms, and prior to any unplanned ground operations. Provide true indication of status for all safety critical mechanisms. Ensure that thermal effects have been considered in the design and analysis of limit switches, and have been reproduced during environmental testing. Design life of mechanisms should consider maintenance cycles of interfacing Testing I I systems. Acknowledgements The authors wish to acknowledge the team members that contributed to the resolution of the problems that have been described in this paper: Mechanical Systems Working Group, ISS Structures and Mechanisms System Problem Resolution Team, and SSP Mechanical System PRTs. References 1. Mechanical Systems Safety, MA2-00-057, NASA JSC, Sept. 28, 2000. 2. Materials Control Plan for JSC Flight Hardware, JSC-27301 D, NASA JSC, Feb. 2000. 3. Use of Liquid Locking Compound on ESP-2, JSC-62850, NASA JSC, Jan. 2005. 4. Inspection Boom Assembly and Shuttle Remote Manipulator System Fastener Test Report, JSC- 63083, NASA JSC, Aug. 2005. 125

Document

AMS_2006.pdf

Reliability and Fault Tolerance in ISS Thermofoil Spaceflight Heaters Victor J. Bolton' Abstract Extra-vehicular avionics systems and mechanisms used on the International Space Station (ISS) typically require redundant survival heaters due to cold thermal extremes. Two such survival heater systems have completely failed to date. These designs were not truly fault tolerant because a failure within the heater patch would bring down both the prime and redundant systems. This report is intended to make designers and operators of mechanisms requiring survival or operational heaters for spaceflight applications aware of common issues that contributed to in-flight failures. Introduction A dual element thermofoil heater is in essence set of two resistive elements laminated together. Failures in either element can easily propagate to the other by damaging other element or the laminate structure. Figure 1 illustrates a layered, dual element design. El nentsM -I r Kapton J Teflon Figure 1. Layered Design Flexible Heater Exploded View Heaters are often treated exclusively as electronic components; however, standard electronics qualification and acceptance testing is insufficient to certify a heater design. Flight history lessons learned lead directly to design and testing recommendations for mechanism heaters. Survivability in particular was not thoroughly examined in the design and testing process. Failures are presented with a short description to provide data points that suggest areas to pursue alternative designs. Recommendations are meant as solutions to these known problems and other sensitivities of heater patch design and installation. History and descriptions presented here were developed through in depth analysis of failure telemetry, and by direct interview of hardware designers, manufacturers, and operators. Boeing ISS Structures and Mechanisms, Houston, TX Proceedings of the 3d' Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 127

Document

AMS_2006.pdf

By reviewing the lessons learned from these failures we can create more robust designs that significantly reduce risk and are tolerant of severe conditions: high watt density, over-voltage, complex installation and unknown stress. These lessons are consolidated in the design and testing recommendations section. Failure History Requirements for survival or operating heaters were determined after much of the hardware design process, leaving this critical system as an afterthought. Flexible heaters have a long flight heritage within the US space program. In the ISS program to date we have seen at least three failures of flexible heater systems on-orbit: Auxiliary Power Converter Unit discharge resistor (not used for heating purposes), Nitrogen Tank Assembly (NTA) survival heaters, and the Segment-to-Segment Attach System (SSAS) Capture Latch Assembly survival heaters. We have also seen one failure in on ground testing, the Flex Hose Rotary Coupler Flight Support Equipment survival heaters. The Nitrogen Tank Assembly and Segment-to-Segment Attach System Capture Latch Assembly cases are presented in this paper. Nitroaen Tank Assemblv [NTA) The NTA uses two separate heater patches with layered elements. Each layer within the patches was wired in series to form two circuits that spanned both patches (Figure 2). As a result, failing both layers in a single patch would completely fail both strings of heaters. String 1 Pwr String 2 Pwr P 1- I 31 W “““I 15 W I ^A/?^^^ I I String 2 Rtn String 1 Rtn 4----- 4 Figure 2. NTA Heater Strings On January 17, 2003 (GMT 2003-017:16:00) both heater strings failed. The NTA heater assembly cannot be inspected until it is replaced with a spare and returned for analysis. After the failure of the NTA heaters, all similar ISS heaters were identified for further analysis. SusDect Condition Action Notice [SCAN) 044 After the failure of the NTA heaters, a Suspect Condition Action Notice (SCAN) was issued across the entire ISS program to check for similar issues. Many heaters were identified by the search criteria: 1. Multiple elements in a single patch. 2. Greater than 0.46 W/cm2 (3.0 W/in2) watt density. Once identified, the true redundancy of the elements, software control, and possible pimpacts were evaluated. The majority of station heaters have watt densities under 0.46 W/cm (3.0 W/in2); unfortunately, several systems contained multiple elements in a single patch. The critical impact of the suspect condition to ISS operations was mitigating the hazard presented by activating both elements in a layered patch simultaneously. Simultaneous activation effectively doubles the watt density of a heater, greatly increasing the risk of failure. Suspect software controlled heater activation set-points were offset to reduce the possibility of this type of failure. Set-point adjustment was not possible for thermostatically controlled heaters on orbit, and operational procedures were modified to minimize this risk during power up. 128

Document

AMS_2006.pdf

PI -P3 Seament-to-Seament Attach Svstem (SSAS) Capture Latch On May 25, 2004 (GMT 2004-146:20:01), both heaters on the Port side, segment 1-3 SSAS capture latch failed. This heater was a single patch with dual element, layered design. Figure 3 gives a general view of the capture latch mechanism. The heater patch wraps around the gearbox at the base and would not be readily visible in this view. Telemetry was available at the time of failure and automated warning alarms were activated when cold limits were approached. Figure 3. ISS P1 Truss SSAS Capture Latch Failure of both elements was simultaneous. The last short peak in both traces in Figure 4 represents the failure point. The temperature of the system began to decay, and system temperatures did not begin 129

Document

AMS_2006.pdf

recovery until the Integrated Motor Controller Assemblies (IMCAs) were activated to provide self heating. The IMCAs are rectangular motor assemblies that bolt to the gearbox at the base of Figure 3. They will actuate the capture latch to berth the P3/P4 truss. Until the P3/P4 truss arrives on flight 12A, the IMCAs will run continuously to mitigate this failure. At the time of failure only one of the two layers was powered; however, software changes to avoid simultaneous activation of both strings may have helped to prevent this failure. The most likely cause of the failure is heater patch burn-through. This heater assembly is permanently installed to structure and cannot be returned for failure investigation. -33 -35 -37 f - d f -39 Figure 4. Failure Temperature Traces Proaressive Failure in a Thermofoil Heater Progressive failure or burn-through is the result of overheating within an element. Many things cause overheating: over-voltage, necking in the element trace, bubbles under the element, incomplete or improper installation, or delamination within the patch. Once the overheating begins, the stages of failure typically follow these steps: 0 0 0 0 Temperatures increase within the element, increasing resistance. Melting temperature for the Teflon is reached and the patch delaminates. In this process the Kapton film may char. Once delamination occurs, there is no direct conductive path for heat transfer and radiation becomes the dominant mode of heat transfer. Unable to release the heat, the heater element foil continues to increase in temperature until it melts or fractures and loses electrical continuity. At this point the heater has failed irrecoverably. 130

Document

AMS_2006.pdf

* In dual element heater patches the local delamination will likely cause failure to propagate between both elements, in both side-by-side and layered conditions. A heater is typicaliy considered failed when it delaminates. The remaining steps lead to failure of the element and an open circuit. There is no data available to determine a critical flaw size. As a result, any visible delamination would be considered unacceptable. Design and Testing Recommendations Desian Recommendations Due to their physical proximity, elements in a dual element heater patch are highly likely to fail simultaneously, even if they are not layered directly on top of one another. They do provide system redundancy for other failures that occur in the power system leading up to the heater patch, but any failure within the patch itself will fail all elements. Multiple element heater patches do NOT provide fully redundant thermal conditioning. Dual element heaters are inappropriate for situations that require true redundancy and fault tolerance unless survivability is addressed. Individually, high watt density, layered design, and installation difficulty would not necessarily be a significant driver towards failure for a flexible heater; however, the combination of multiple higher risk design elements into an individual heater patch creates sensitivity to conditions that increase the risk of failure. Because layered design has been common across all discovered failures to date it should be approached with a high level of caution. Each individual issue further erodes the margin for error. Heater systems should be integrated as early in the design process as is possible. Earlier inclusion in the design process would allow for specifications of system components that would permit a large, flat area for heater installation. Heater patch effective area should be maximized and watt density should be minimized within the bounds of required heat generation: An ideal heater system will be as large, simple, flat and low-watt density as possible. As watt density increases, the importance of effective heat transfer from the element to the heat sink increases. Installation conditions must be carefully examined and fully understood to mitigate overheating. Testina Recommendations Acceptance and qualification testing of thermofoil heaters is specified similarly to standard electronic components; however, this approach is not sufficient to determine all three crucial criteria: Suitability of the design for the intended use, Capability of the design to maintain the desired temperature under minimum power conditions, or Survivability of the design under extended use and maximum power conditions. Survivability verification under worst-case use conditions must be part of acceptance testing. Survivability in particular has not been part of heater acceptance and qualification testing programs. ISS voltages historically have been running near the upper end of the acceptable range of 11 3-126 volts. Heaters were typically tested with the minimum voltage (or ambient equivalent) to verify they would produce sufficient heating in a worst-case environment, but never tested to verify they could survive the highest available voltages indefinitely. A 25% increase in heater power can lead to failures not found in qualification testing at 11 3 V. Testing under vacuum conditions in particular will create a more realistic environment for verifying survivability of a heater system design. Ideally, all hardware would be fully tested in a thermal vacuum environment, but cost and schedule constraints can prohibit this level of testing. When planning the testing regime, additional planning and analysis should be spent to ensure that the heater has seen an equivalent duty cycle. Because vacuum conditions drastically change the thermal environment, testing of a heater design under vacuum attached to an appropriate heat sink would be a reasonable compromise. 131

Document

AMS_2006.pdf

Conclusions Flexible thermofoil heaters have failed several times in the ISS program. We must look at and seriously consider the processes going into design, testing, and installation of heater patches on spacecraft. Suitability and capability of the designs to fulfill their functions have generally been understood and analyzed in depth; however, survivability has not always been treated with sufficient rigor. Qualification and acceptance testing must include steps to verify heater survival. Overestimation of the ability of the system to survive high voltages, extended use, or inadequate installation undermines all design effort spent on the system. If the heater itself does not survive, failure of the thermal system jeopardizes mission objectives. If no resolution can be found, the system could be a total loss. Mitigating the impacts of these failures consumed substantial time and resources within the ISS program. The most critical lesson learned is to do everything possible to maintain the survivability of the mechanism via heater patch design by incorporating knowledge from previous failures during initial design. References 1. Bolton, Victor J “SCAN 044 Response for Structures and Mechanisms,” Boeing memo A92-J383- STN-M-VJB-04-032, February 2004. 2. Bares, Geoffrey “PRACA 3661 : P1 Nitrogen Tank Assy (NTA) Heater Failure,” Boeing presentation, February, 2003. 132

Document

AMS_2006.pdf

Development, Pre-qualification and Application of an Active Bearing Preload System Simon Lewis* and Martin Humphries** Abstract This paper describes the development to pre-qualification status of a novel and widely applicable technology development known as a Bearing Active Preload System (BAPS). The BAPS can be thought of as a “smart bearing housing” which replaces a conventional bearing or mechanism housing and whose function is to permit ball bearing preload variation on command. An overview of typical BAPS requirements, its design, the range of actuation options and some performance data are provided. A number of historical and recent bearing applications are reviewed and the benefits realizable by a capability to vary preload are highlighted. These can include order of magnitude improvements in lubricant lifetime, reductions in bearing torque, mechanism mass, cost and complexity. Introduction Ball bearings are preloaded to provide adequate rotor location and bearing stiffness, as well as to protect the bearings themselves from damage due to “hammering” during launch. However, increased preload also has some undesirable effects discussed below. Grease- or oil-lubricated bearings have essentially two main torque components, namely a speed and temperature dependent “Viscous” component and a load dependent “Coulombic” torque component. At low- to moderate-speeds, the Coulombic component can be dominant, such that mean bearing torque is approximately proportional to preload4/“. This relationship is also true at all speeds for solid- and self- lubricated bearings. Furthermore, peak Hertzian ball-raceway contact stress is proportional to preload’I3 [1 I. In ESTL much work has been done in the past concerned with characterization of bearing torque and lifetime, particularly for self-lubricating (e.g., Duroid or PGM-HT) and solid-lubricated (e.g., thin films of MoS2 or lead) bearings. A review of some of the highlights of this experimental work [2-51 demonstrates that preload (or peak Hertzian contact stress) has a very significant effect, both on film lifetime for solid lubricated bearings, and on separator (cage) wear for self-lubricating bearings as shown in Figure 1. For liquid lubricated bearings too (especially PFPE-based oils and greases), though the lifetime may be often be longer, ultimately lubricant degradation due to shear or chemical reaction at balVraceway asperities and separator stability/wear issues are aggravated by high bearing preload. In summary, the use of a high preload throughout life not only increases the Coulombic torque and therefore motor mass/power requirements for the bearing, but also decreases the potential operational lifetime significantly. Furthermore, since in most applications the high bearing stiffness required for launch is no-longer essential once in-flight, the capability to operate in-flight at relatively low preload is highly desirable, particularly in long-lifetime applications and those with challenging thermal requirements (e.g., those having a particularly wide operational temperature range, large or adverse thermal gradients). Historically bearing “off-load devices” or “launch-locks” have been used to protect bearings during launch by providing an alternative load path in those relatively few applications where the launch loads necessarily exceed the capacity of the bearings, or more often where the low-torque or long-life requirement dictated a moderate to low preload should be selected (e.g., Giotto de-spin bearings [6]). However, bearing off-load devices are relatively mass inefficient and need to be tailored to the application. * ESTL (European Space Tribology Laboratory) - ESR Technology Ltd., Warrington, Cheshire, U.K. ** Sula Systems Ltd., Wotton-under-Edge, Gloucestershire, U.K. Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 133

Document

AMS_2006.pdf

The BAPS concept provides a bearing cartridge structure that is of appropriate stiffness and load capacity to enable the maximum loads (combination of preload and externally applied launch loads) for the particular bearing size to be applied and a factor of approximately 10 highlow preload ratio. This ratio permits at least a halving of Hertzian contact stress, with commensurate bearing torque and life benefits. 1 .E+11 Duroid/PGM-HT lubricating Cages MoS2 (Only on races) \ ‘h E ‘E l.E+07 Q, !e -I b 1 .E+06 1 .E+05 1 .E+04 I I I I I I I h SI \ 31 \\ 0 500 1000 1500 2000 2500 3000 3500 4000 Peak Hertzian Contact Stress (MPa) Figure 1. Lifetime v Contact Stress for Bearings Lubricated by DuroidPGM-HT or MoS2 BAPS Requirements The BAPS device requirements were determined via a comprehensive Market Survey of current or near term applications in which more than 34 space companies were contacted. Responses to the questionnaire indicated that over 20 applications could benefit from BAPS technology, for example deployment hinges, steering devices, APM’s, various instruments, actuators, reactionhomentum wheels, CMG’s, solar array drives and scan mechanisms for microwave sounding and radiometry. Design drivers from the Survey were ranked in order or priority as Reliability, Envelope, Mass and Cost. The Survey suggested most potential users would require a bi-stable BAPS device capable of switching from a single high-preload (State 1) to a single low-preload (State 2), however a significant minority of respondents also foresaw applications where the ability to continually vary preload could be valuable. A wide range of bearing sizes was identified, from 25-190 mm bore, but the most popular size range was 50-100 mm bore. The number of operations required of a bi-stable BAPS (assumed to exclude qualification margins) was enveloped as e20 on ground and 1 in-flight. For a variable type device, up to 2 million operations were envisaged. Envelope constraints were also deduced from the Survey, together with thermal requirements. Excluding the relatively infrequent cryogenic or high-temperature applications, the majority of applications were enveloped by an operational temperature range of -40°C to +80”C. BAPS Concept Derivation and Description By combining the results of the Market Survey and the experience of the team, a generic requirements specification was generated together with a number of concepts potentially capable of meeting these requirements. The concepts were subjected to an initial filtering using “killer criteria” (Le., criteria which all viable concepts MUST meet) after which surviving concepts were then subjected to a formal trade-off against criteria again reflecting the aspirationskoncerns expressed in the Market Survey returns ranked using a rigorous paired comparison method. 134

Document

AMS_2006.pdf

As part of the trade-off process an assessment was made of the flexibility of each candidate design for adoption with 4 candidate design cases ranging from small (25-mm bore IS0 10 section, e.g., SNFA EX25 with State 1 preload of 200 N), to medium (50-100 mm bore, thin section) and large (>150 mm bore thin section, e.g., Kaydon KA075AR0, with State 1 preload of 3000 N) bearing envelopes. From these design cases, a single so-called “enveloping design case” was established as a most-challenging case for the Breadboard Model (BBM). The enveloping case was based on the thin-section SNFA SEA55 (7CE3) bearing, having a preload ratio of 3000 N (State 1) / 200 N (State 2) generating a peak bearing Hertzian contact stress ratio of approximately 2000 MPa / 850 MPa. In the trade-off winning concept, which is based on the monolithic titanium structure shown in Figure 2 below, the bearings (not shown) are located by upper (floating) and lower (interfacing) thrust rings. These rings are linked via a series of “suspension beams” which provide the required structural stiffness characteristics for launch. By monitoring the strains induced in these features due to changing relative axial displacements of the two rings the bearing preload can be determined. Rotation of a central “synchro-ring” by a few degrees adjusts the spacing of the upper and lower rings by some microns and so modifies bearing preload. The ‘upper and lower rings remain parallel despite the rotation of the synchro-ring because they are linked through a series of flex-struts to the upper and lower rings preventing misalignment. These features also provide an over-center toggle action such that the preload is stable un-powered in two synchro-ring positions. For the high preload state the central synchro- ring is retained just away from top-dead-center (TDC) in a stable configuration (in contact with an end- stop). To set the bearings to low preload, a “Self-Release” force is applied to displace the synchro-ring such that the flex-struts are pushed over TDC and to a stable low-preload position used for flight. This position is maintained by application of a “Holding Force”. For the BBM, the high (3000 N) preload state is at a synchro-ring position of +lo0 prn, and low (200 N) preload at -600 pm as shown in Figure 3. A maximum “Reset Force”, around 100 N on the BBM must be applied. Figure 2. Monolithic Titanium Alloy Structure Concept and Hardware (includes liners to interface with bearings) 135

Document

AMS_2006.pdf

Holding Force Displacement Curve 1 / 50 - ielf Release Forc v- I -200 400 10 E Drive Force Curve High Preload Posit ion -1 00 Low Preload Position Reset Force - IdU Synchro-ring Displacement (micron) Figure 3. Predicted Drive Force and Thrust Ring Displacement Curves for SEA55 BAPS Practical Experience with Hardware Highlights of the performance of the breadboard model (BBM) BAPS, which contains bearings lubricated by a PFPE-based grease thickened by PTFE particles are provided below. Dimensional inspection showed the BBM to have been manufactured to tolerances in-line with requirements, especially when considering bearing seating tolerances. No significant misalignment of the bearings on assembly or during BAPS operation was measurable. On assembly the BBM structure introduced 400 pad of angular misalignment and when operated at most a further 20 prad. These values are well within even the permissible groove-face misalignment for the bearings themselves (e.g., for ABEC 9 grade bearings of this size, 200 pad PER BEARING is allowed). This figure is also much less than the known face-face target misalignment for good performance of solid-lubricated bearings, which is around 300 prad. As expected, because the BAPS is a purely elastic system, the strain gauges used to monitor displacement of the suspension beams were found to be very linear (sensitivity typically 20 @pn of BAPS total axial displacement) and repeatable (within 1-2 p when BAPS is repeatedly switched from high- to low-preload), demonstrating accurate, repeatable preload switching even after vibration. The target preload ratio of 3000 N (State 1) / 200 N (State 2) which would have provided bearing mean torques at room temperature of 1400 gcm and 110 gcrn respectively was not quite reached due to the effects of elastic strain in the bearing seats. The achieved preload values were 2500 N and 300 N for which the measured torques were around 1100 gcrn and 200 gcrn (Figure 4). However a positive effect of this bearing seat elasticity was that because strain energy is stored in the BAPS, bearings AND the bearing seats, on switching from high to low preload much less force is needed to hold the device in against the low-preload end-stop position than is predicted based on a model which excludes bearing seat effects, as can be seen in Figure 4. 136

Document

AMS_2006.pdf

mnn 1400 LOW PRELOAD Measured Drive Force Predicted Drive Force L 4 rn 0 Synchro-ring Position (microns) Figure 4. Measured torque and Drive Force Performance of BBM BAPS Unit Actuation Options The BAPS can be actuated by several means depending on the application. Most commonly a direct linear actuation will be used for which, all that is required is to provide up to around 60-70 N force and around 700 pn of total linear displacement. For the majority of applications where a typical operational temperature range of -40 to +80°C is demanded, then actuation will be via a Shape Memory Bender or High Output Paraffin Actuator. Both devices have the advantage of reset-ability and ease of redundant implementation. However, the BAPS can also be entirely passively actuated, for example by exploiting differential thermal expansion for applications that have high or low temperature requirements, or indeed implemented in a fully-variable form. In the latter case, use of a rotary piezomotor is an attractive option. How Designs Could Change Post-BAPS A number of historical and more recent applications are reviewed (and summarized in Table 1). For each we identify some of the main potential design changes and resulting engineering benefits of use of a BAPS device in similar future applications. Optical Pointincl and Trackina (ea. GOMOS SFA on ENVISAT) The GOMOS Steering Front Assembly (SFA), currently flying on ENVISAT, is an optical acquisition and tracking mechanism for stars. The mechanism includes a coarse pointing azimuth stage that provides a wide pointing range by allowing a mirror and turntable to rotate. The bearing configuration would ideally support the mirror and turntable during launch. However, given the need for low torque noise emission and reliable operation the bearings required a soft, low preload, and a relatively small outside diameter. However, high strength and stiffness were required during launch, and these conflicting demands could only be met by use of a massive and complex motorized launch protection system (bearing offload device or BOLD). Using BAPS, a more elegant design using large, thin-section bearings could have been adopted, capable of meeting the launch design case, yet also providing low and soft bearing preload on-orbit. Preliminary analysis suggests that a pair of back-to-back thin-section bearings having a bore of 200 mm could have met the launch load and stiffness requirements. This approach would have yielded very significant savings in mass, cost and complexity. 137

Document

AMS_2006.pdf

Application BB orsuper- Duplex GOMOS SFA Giotto De-Spin cantilevered -but with 3d hold-down loads - 1.5-3 kN > 5.4~1 O5 Nm/rad I -50-85 points (Launch 1200 N (min) for launch stiffness mm Optical Terminal BB or FF & Spring/Diaphragm APM with RF Feedthrough (Bepi- Colombo) I 2o ym Up to 1-2 kg >loo-150 HZ (typ. SADM (BAPTA) 1 >loo I >60 kg mm BB ReactionlMomentum WheeVCMG >70 Hz Swash Plate APM Push-Broom Scanner (EGPM) Table 1. Summary HistoridRecent Application Details Face-Face:FF FFIBB >160 Hz / Minimize I (e.g., 30-300 N) I KA075 I 2kg1120mm plus 102 Nm) BB&Diaphragm 1 25mm I Unknown >lo0 Hz / 45 N in flight I 177 mm 1 Unknown 4 Pt Contact BB Hard Preload I >IO0 Hz/ 1500 N Special Constraints BOLD used BOLD used Large bore, severe masslpower constraints Bore to accommodate RF joint High launch loads and moments BOLD used, released on-orbit stiffness must support solar array Life target 5x1 0’” revs Stiffness on-orbit adequate for payload 6x1 O3 Nm/rad. On orbit Despin Mechanism (ea. GIOTTO De-spin) In common with many deep space missions, GIOTTO the last European deep space mission using a spin stabilized spacecraft, required a de-spun antenna to maintain a high-data-rate communications link to Earth. The key issues for this type of mechanism are to support the antenna during launch yet provide long life, low friction and reliable operation in adverse thermal conditions (e.g., with wide fluctuations in temperatures and temperature gradients). To maximize bearinghbricant life while maintaining insensitivity to adverse thermal gradients a low preload was required and applied by means of compliant diaphragm. But launch loads and frequency constraints for the application could only be met by employing a separate bearing off-load device (BOLD). Had BAPS technology been available this application could have avoided the need for the BOLD even if a bearing pair with virtually the same diameter were selected. Optical Communications Laser Communications Terminals for inter-satellite communications links are a major application for pointing mechanisms. One device currently under development is a Course Pointing Assembly. This device is required to gimbal a high precision mirror through large angles. The bearing must be almost free of stick-slip effects and must provide a through-bore sufficiently large to allow the passage of an optical beam (typically 15-mm dia.) through the azimuth gimbal. To achieve both of these requirements in a mass- and cost-effective manner ideally requires high bearing preloads during launch and very low preloads once on-orbit. A preliminary assessment suggests that launch requirements could be fully met at low risk if launch preload could be set at 2000 N, provided preload is released to less than 200 N once on orbit. The use of BAPS could provide an even larger range than this and would provide a lower risk solution than the use of a fundamentally lower fixed preload and snubbers to limit rotor motion during launch vibration. The latter conventional approach allows hammer like contact at the bearing and drive gear contacts which increase risk of degraded life-time and in-orbit performance. 138

Document

AMS_2006.pdf

Two Axis Antenna Pointina Mechanisms (APM's) with RF Feedthrouah There are a number of currenthear future applications for two-axis APMs that require integrated R/F rotary joints within their azimuth actuators. The applications cover both commercial LEO constellations through to dedicated science missions such as the Bepi-Colombo mission to Mercury. These applications need to accommodate internally mounted rotary RF joints that call for high precision and repeatable co- alignment of R.F. choke faces. The bearings themselves are required to be stiff during launch to support the antenna but must provide low friction torques once on-orbit even when subjected to inverted temperature gradients caused both by R.F. losses at the shaft and in some cases by the operational environment. The major risks associated with these mechanisms are their sensitivity to thermal gradients and installation tolerances given the lack of elastic preloading. In the unique Bepi-Colombo case, an operational temperature range of +50"C to +250"C is envisaged with 70% of the expected life at the hot case and with adverse temperature gradients which could be up to 30°C across the bearings. It is foreseen that the use of a BAPS for these applications would significantly enhance performance and reduce risk. It may be an advantage for applications to control bearing stiffness at actuator level in order to maximize the overall structural efficiency of the combined APM and Antenna Launch Configuration, with the Antenna in its stowed configuration. To implement BAPS in for example the Bepi-Colombo application, a pair of angular contact bearings would replace the baselined hard-preloaded super-duplex bearings (without significant impact on radial envelope) and titanium housing. For launch, the BAPS could be set to provide a high-preload in excess of the currently specified 1200 N (enhancing launch stiffness, eliminating bearing gapping during launch and further minimizing risks of fretting damage to gears). On-orbit the preload could be reduced, and more importantly the preload-stiffness achievable with BAPS is low, compared to a conventional hard-preloaded system thus rendering the bearings much less sensitive to the significant and adverse thermal environment. Medium Power SADM (BAPTA) The BAPTA was developed for early satellite programs, e.g., OTS, ECS, Marecs. It became the baseline SADM for a later series of defense satellites transferring up to 900 W of array power and associated signal lines. The SADM design uses a pyrotechnically released internal bearing off-load system providing bearing protection and adequate launch stiffness to the solar array. In orbit, the bearings are compliantly preloaded. Use of a BAPS for this type of SADM would replace a complicated system of compliant preloading, hard bearing-offload and pyrotechnic release mechanisms. Final preload could be set on station for required torque performance and maximum power efficiency. Such a device would be suitable for small to medium mass arrays on small satellites. Reaction/Momentum W heel/Control Moment Gvros Reaction/momentum wheels and Control Moment Gyros run continuously and allow transfer of momentum/torque as required by the orbital control system. Ideally the operating case requires a minimum of bearing preload to minimize power loss, torque disturbances and motor mass/volume. The rotating elements supported on bearings have significant mass and require correspondingly high stiffness, both to prevent damage during launch and in the case of the CMG to provide structural efficiency during slew maneuvers. Cage instability and wheel whirling instability are also factors to be considered. Use of BAPS for wheels would prevent gapping on launch yet allow minimum preload and hence torque loss value for in-orbit operation thus improving power consumption and wheel performance per unit mass. Such a capability is particularly important for solid-lubricated wheels sometimes used in Smallsat and Microsat applications for which life benefits may be significant. With a variable type BAPS, preload could be varied to alleviate cage or resonant frequency problems in reaction/momentum wheels or to permit tuning of CMG stiffness characteristics and energy dissipations. 139

Document

AMS_2006.pdf

Push-broom Earth Scanner (ea, EGPMl The now postponed EGPM mission requires an earth-scanner using the push broom scan approach. The main instrument rotor and reflector (mass >70 kg) rotate at a constant speed up to 32 revs per minute over a nominal 15 year mission, resulting in >250 million revolutions. The instrument rotor must be supported rigidly during launch and rotated about a precise axis in-flight. The main issue is related to the high instrument rotor mass and the relatively large C of G offset of the instrument rotor in relation to the scan mechanism. This leads to combination of high lateral and axial forces and very high moment loading during launch. Therefore the current instrument configuration incorporates a dedicated hold-down system that supports the rotor at its outer extremities at four locations for launch. These are released once on-orbit and then the instrument rotor is supported purely by the scan mechanism. While the primary launch protection is able to rigidly secure the bulk instrument structure, due to the size and supported mass of the instrument equipments mounted within the instrument there are also local panel modes between the scan mechanism interfaces with the instrument lower panel and the spacecraft side wall. It is not possible to accommodate these modes by allowing relative internal movements within the mechanism due to a number of critical internal operating gaps (e.g. at the motors, sensors and power/signal transfer devices) Therefore the scan mechanism must react the local forces and moments associated with constraining these local modes, which leads to significant local forces and moments across the mechanism. The bearings of the scan mechanism are required to have a bore of greater than 100 mm due to internal design constraints and so in principle the mechanism is capable of reacting significant loads. The use of BAPS technology allows the bearings to react these launch loads without introducing any significant internal movements and provides a relatively stiff fifth hold down point at a location where it is most mass effective. This is done without compromising the mechanism’s on orbit performance, in particular the long life and the critical on orbit nodding frequency. Note that although a bi-stable BAPS is currently base-lined, there would be some benefits in this application in the use of a fully variable BAPS that would enable on-orbit tuning of instrument rotor preload in order to achieve an ideal rotor stiffness. Conclusions This paper demonstrates that the novel BAPS device can be used in a wide variety of the more demanding spacecraft mechanism applications to improve both system level and mechanism performance and lifetime. The examples used demonstrate that for maximum benefit, the BAPS concept needs to be incorporated within the design from an early stage in the system development. The BAPS is currently at the pre-Qualification stage of development, but it is expected that it will be Qualified during 2006. References 1. Roberts E.W. (Ed.) ESTL Space Tribology Handbook 2. Performance Guide - Self-lubricating Bearings Prepared by NCT, June 1976 3. Anderson M.J. “Duroid Replacement Material Requirements Specification” €STL/TM/20 (July 1998) 4. Lewis S.D. “GERB Phase 3b Post-Test Review” (Unpublished Experimental Work) ESTL (2000) 5. Stevens K.T. & Todd M.J. “Parametric Study of Solid Lubricant Composites as Ball Bearing Cages” Tribology International Oct. 1 982 (pp293-302) 6. Felici F. “The Giotto Mechanisms and Their Functions” Roc. Is‘€SMATS, Neuchatel 7983, ESA-SP- 196 140

Document

AMS_2006.pdf

Development of a Dual Mode D-Strut@ Vibration Isolator for a Laser Communication Terminal Dale T. Ruebsamen’, James Boyd*, Joe Vecera. and Roger Nagel’ Abstract This paper provides a review of the development by Honeywell of a dual mode D-strut@ vibration isolator for long range communication instrument. This paper reviews the basic design requirements for the dual mode isolator, the D-strut’ isolator design drivers, the prototype D-strut’ isolator design, and the results of prototype D-strut’ isolator testing. Introduction Honeywell is developing a dual mode D-strut@ vibration isolator to isolate the sensitive instrument from the spacecraft bus during launch vibration and during in-flight operation (spacecraft bus vibration distubances). This paper will describe the design challenges that led to a new isolator design meeting the stringent requirements of this sensitive instrument, and the issues that were identified as a result of testing a set of prototypes of the new design. Two isolator design options were considered for this application. The first option was to launch lock mount the instrument to the spacecraft and design the instrument to survive the launch loads, and then release the instrument once in flight allowing the isolator to isolate the instrument from the spacecraft bus disturbances. The second (dual mode) option uses the same isolators to isolate the instrument from the launch environment and the spacecraft bus disturbances. In both cases, the isolators are configured in an optimized hexapod configuration (also know as a “Stewart Platform”) to provide the desired isolation. The dual mode isolator configuration was chosen to minimize the loads into the instrument due to launch and still provide the required in flight isolation due to the spacecraft bus disturbances. There are several requirements that define the new isolator design. The requirements for the D-strut@ isolator are that the weight of the isolated system is 48.1 kg (106 Ib) maximum; the quasi-static acceleration is 16.25 g’s maximum during launch. The vibration requirement means that the peak axial load into each D-strut@ isolator is approximately a factor of over 2.0 times the previous dual mode design. The higher loads also required that the stroke in each isolator increase by almost a factor of 1.6. These increased loads and strokes needed to be accommodated without changing the length of the isolator. The D-strut@ isolators are three parameter isolation systems. The parameters defined for the three parameter isolators are static stiffness (Ka), dynamic stiffness (Ka + Kb), and damping factor (Ca). These are the parameters Honeywell uses to size the main machined spring, the tuning spring, and the damper annulus gap filled with the proper damping fluid. Honeywell has submitted a patent application for a new isolator design that meets the load and stroke requirements needed to support the instrument. In the new isolator design, the sealing bellows in the damper assembly are externally pressurized instead of internally pressurized. The new configuration allows for a significant increase in the load capacity and stroke of the isolator without increasing the overall length of the isolator. To demonstrate that the new isolator design would meet the requirements, Honeywell built two complete isolator assemblies to measure the isolafor parameters and fabricated the fixtures needed to test the isolators in a bipod configuration up to the maximum quasi-static accelerations. Honeywell, Defense and Space, Glendale, AZ Proceedings of the 38 Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2OC6 141

Document

AMS_2006.pdf

To characterize a bipod of the new Isolator configuration, we performed a series of sine vibration tests with a 18.1-kg (40-lb) mass attached at 0.25 g, 2.5 g’s, 7.5 g’, 12 g’s, and 16.25 g’s from 5 Hz to 1 kHz; we performed a random vibration test from 10 Hz to 2000 Hz at a level of 9.1 Grms the results of which will not be discussed here be cause of the similarity to the sine test results; we also performed a 900 G- SRS Shock Beam test with a 18.1-kg (40-lb) payload mass and a 31.8-kg (70-lb) payload mass. These tests characterized the isolator and showed how the “Q” (amplification factor) at the isolator resonant frequency compared at the different input “g” levels. The testing also found that the lateral mode of the isolator was very well damped and that there was an undamped machined spring surge mode at 61 5 Hz. All the testing was performed in the Honeywell Vibration Lab. This paper will describe the development of the new isolator D-strut@ for the Laser Communication Terminal and the testing performed on the development bipod. Design Requirements and Analysis The design requirements for the D-strut@ are not finalized at this stage of the program. However, there are some preliminary design requirements that are design drivers for the sizing of the D-strut@ isolator for the instrument. The original desire was to use an existing D-strut@ isolator’ design that Honeywell had qualified for launch and on-orbit isolation for another application. See Reference 1 for detailed description of the existing qualified D-strut@ isolator design. The original qualified system had been developed for a smaller payload, lower environmental loads, and the system isolation requirements were less stringent. A comparison of the requirements for the previous system and the new system is shown in Table 1. The impact of these design requirements differences and how they affect the design will be discussed in the following paragraphs. Table 1. Design Requirements Comparison ImDact of Requirements on the new D-Strut@ Isolator Desian The peak damping force requirement was one of the limiting factors in the previous D-strut@ isolator design. The existing design would have to be redesigned to meet the new design requirements. The new design is a through shaft system with externally pressurized sealing bellows. By externally pressurizing the sealing bellows, there is no squirm pressure limitation. The externally pressurized bellows design is stable, Le., increasing the pressure does not cause the bellows to try to move to one side since the fluid is always pushing in. The limitation of the externally pressurized bellows is the strength of the bellows materials resisting the pressure and external loads. Because of this difference in the design, the new design capabilities exceed 87.02 kPa (600 psi) which is a factor of 2 over the previous qualified isolator design. Another limiting factor in the existing qualified D-strut@ isolator design is that the dynamic stroke of k5.1 mm (0.20 inch) will not meet the new requirement of k 8.13 mm (k 0.32 in). The new D-strut@ isolator design contains externally pressurized bellows and does not have the limiting pressure issue; therefore, the bellows stroke capabilities can be increased by adding more convolutes and increasing the bellows 1 42

Document

AMS_2006.pdf

length. The stroke capability of the new D-strut@ isolator damper assembly is in excess of * 8.13 mm (* 0.32 in), as required. The original D-Strum Isolator is shown in Figure 1 and the new patent pending D- Strut@ Isolator is shown in Figure 2. It needs to be noted that the mounted lengths for both of the isolators are the same, 21.03 cm (8.28 inches). c 1. Original D-Stru The original D-Strut@ lsoktor shown above was used in a previous program to provide launch and on orbit isolation for an instrument which has already flown. The environmental loads for a new payload instrument exceed the capabilities of the existing D-Strut@ isolator. 2- I I I I I I1 FigL. - _. . __ __ - __. ___ .__.___. . _. _____. .._. The new D-Strut0 Isolator shown will be used to provide launch and on orbit isolation for the Laser Communication Terminal Experiment which is under development. The launch loads for the Laser Communication Terminal required the design changes. There is a patent pending on this new design. Desiqn Analvsis The loads from the derived requirements in Table 1 were used to determine the stresses in the components that make up the new D-strut@ isolator. The analysis of the sealing bellows and compensation bellows was performed by the bellows supplier using proprietary methods. The flexure, KB spring, and main spring analyses were performed using finite element analysis modeling methods with l-DEAS@ application software. There was a requirement that the factors of safety be 1.1 to the material yield limit and 1.25 to the material ultimate limit. We calculated the loads, predicted the stresses, and calculated the factors of safeties for all the critical structural parts in the new D-Strut@ Isolator. The limiting part in the new D-Strut@ Isolator, from this set of analysis that needs further evaluation is the KB spring with a minimum factor of safety to the yield strength is 1.14 and the minimum factor of safety to the ultimate strength is 1.22. Once detailed requirements are defined, the loads in the parts will be further analyzed. Isolator Bipod Test The new D-strut' isolator bipod testing was performed in August, 2005 with the isolators mounted to an 18.1 -kg (40-lb) mass. This configuration proved that the isolators were capable of meeting the extreme load case defined for the launch quasi-static environment of 16.25 g's. The testing was performed up to 2000 Hz. The configuration is shown in Figure 3. The suspended mass was 18.1 kg (40 Ib) so that the bipod first mode would align with the predicted system hexapod bounce mode of approximately 40 Hz. We reconfigured the shock test setup to test with a Shock Beam methods using both the original 31.8-kg (70-lb) mass and the new 18.1 -kg (40-lb) mass. We attached the bipod (with the masses suspended) to a shock beam and provided the shock levels by impacting the beam with a weight on a pendulum. The shock beam test setup is shown in Figure 4. 143

Document

AMS_2006.pdf

18.1-kg (40-lb) pay- load mass shown - Figure 3. Bipod Vibration Configuration Electro-Dynamic shaker Input is at Bipod Base on the left with the payload mass placed in-line to the input. This configuration minimized the suspended mass modes in the measured response. Figure 4. Shock Beam Test Fixture Set-up. This set-up was used to provide a shock to the base of the Bipod using the impact mass suspended on the chain. The impact mass is allowed to swing and impact the shock beam to produce the shock required at the bipod base on the far end of the shock beam. Shock Beam Test When performing the shock beam test, we used an 18.1 -kg (40-lb) mass and a 31.8-kg (70-lb) mass. The test setup is shown in Figure 4. Figure 4 shows the actual setup of the shock beam. There was an array of accelerometers placed on the new D-strum isolator bipod and fixtures to measure the shock input and the response of the isolators and payload mass. Only the in-axis input and in-axis response of the payload mass will be discussed in this paper. The set up was the same for both of the masses. The test results for each of the payload masses were much the same; therefore, only the results from the 18.1 -kg (40-lb) payload mass test will be presented. The locations of the accelerometers were selected to provide information on the isolator spring body axial surge modes, isolator spring body lateral modes, and the mass response. The locations of the accelerometers are shown in Figures 5. In order to ensure that the shock input to and response of the payload mass were captured, we used shock accelerometers at the channel 1 location, and the channel 8 location. Base input Figure a. anocK deam Test Accelerometer Lucations From this view, five of the accelerometers can be clearly seen. The accelerometers of interest which indicate how each of the isolator body modes affect the payload mass are channel 1 and channel 8. 144 Channel 1 8 BiDod

Document

AMS_2006.pdf

The SRS's of the shock pulses are shown in Figure 6 and the time histories of the pulses are shown in Figure 7. The response of the payload mass was measured with accelerometers at the locations shown in Figure5. The peak response of the mass was measured at the channel 8 (in axis) accelerometer location. The peak response from the shock pulse was peak input value of 805 g's and a peak response of the payload mass was 21.2 g's. This was a reduction of 31 dB. The amount of isolation due to base shock input can be seen by viewing the base input time history on the same plot as the mass response as plotted in Figure 7. The SRS of the 18.1-kg (40-lb) mass to both of the shock pulses is shown in Figure 6. The SRS of the response reflects the fact that, in the higher frequencies, the first input shock pulse is higher and this is reflected in the response of the mass. One noticeable artifact of the mass response in Figure 6 is that there is a peak between 560 Hz and 630 Hz even though the break frequency of both pulses is around 800 Hz. The 630 Hz peak is very close to known surge frequency of the isolator main springs. The sine vibration data only goes to 2 kHz; therefore, the correspondence to the peaks in the SRS data can only be tracked to 2 kHz. Conclusions about the Shock Beam Test Results The following conclusions can be made about the test results. 0 The shock beam test was able to achieve the input levels required by the potential isolation system. The isolators in the bipod configuration were able to decrease mass responses relative to the maximum input by 30 dB or more. These Isolators eat shock! The isolator surge modes do contribute to the response of the payload mass due to the base shock input; but, the contribution is not as great as previous testing indicated. 10 -Reference - - -LowerLimit - - -Upperliml - -Beam Shock Input (g), Control Chn 1,4W. P1 -Beam Shock Input (g). Control Chn 1,4oW. Pa -Beam Shock Response. (9). Chn 8.40#. Pi - Beam Shock Response. (a). Chn 8.40% P2 Figure 6. Bipod 18.1-kg (40-lb) Mass response SRS, both Pulses The isolated calculated SRS of the payload mass is significantly lower than the input SRS. The spring surge mode at around 620 Hz can be observed in the above plot. 145

Document

AMS_2006.pdf

Dstrutm Bi-Pod Shock Beam Test, Both Pulses, 40 Ib Mass, Base Input compared with Mass Response, Time History 800 700 600 500 -100 -200 -300 -400 -500 0 5 10 15 20 25 30 35 40 Pulse Time (rn-sec) I-Shmk Beam, CHN 1, 4M, Pulse 1 -Channel 8, Pulse 1. 40 Lb Mass, Time History Figure 7. Comparison of Shock Input and Mass Response, 18.1-kg (40-lb) Mass, Shock Beam Test The time history for the recorded shock pulses is shown along with the mass response. This plot shows the amount of reduction in the response of the bipod mass due to the isolation of the shock input. The effective reduction of the mass response due to the base input is over 30 dB. Sine Vibration Test Sine Vibration Test Setup The test setup for the sine and random vibration test is shown in Figure 15. The picture in Figure 16 shows the actual setup of the sine and random vibration testing. The bipod and mass were instrumented with 15 accelerometers. The accelerometers were placed to measure the response of the isolators and payload mass with-in the limitations of the available instrumentation. The locations of the accelerometers were selected to provide information of the in-line spring surge modes, the isolator spring body lateral modes, and what the mass response was at those modes. The locations of the test accelerometers are shown in Figure 8. The testing was performed with only the 18.1-kg (40-lb) mass. This was done to characterize the equivalent system bounce mode and to characterize the isolator spring body lateral mode, spring body in-line surge mode and harmonics, which are not affected by the bipod suspended mass. Figure 8. Sine and Random Vibration Accelerometer Location, South Side The north (top) and south (bottom) isolator with the location of the visible accelerometers are shown. The slip plate and the adapter plate are to the right with the suspended 18.1 -kg (40-lb) mass to the left. 'I 146

Document

AMS_2006.pdf

Sine Test There were a total of six different test input levels of 0.25 g's initial, 2.5 g's, 7.5 g's, 12 g's, 16.25 g's and finally 0.25 g's. The input at the base was limited for the 7.5 g's, 12 g's, and 16.25 g's to achieve a peak response at the payload mass of 16.25 g's. The mass responses for the 0.25 g's and 2.5 g's test levels are not limited. Reviewing the mass response provides an indication of the affect of the isolator structural modes on the payload mass. It is best to determine the performance of the isolators by reviewing the transfer function of the isolator which is a ratio of the response relative to the input. Using the Transfer function the different input levels can be compared to each other. Sine Test Transmissibilitv Data (Transfer function) Prior to performing the vibration test, we identified which response channels we wanted to compare to the reference channels so that the transmissibility of the isolator could be evaluated without consideration to the input level. For the purposes of this paper, we will discuss the calculated transmissibility of mass response accelerometer, channel 8, to the bipod base. The transfer function for all the base input different levels is plotted in Figure 9. In this plot, the bipod break frequency is shown, the lateral mode of both of the isolators can be clearly identified, the spring body surge mode with its very high response can be clearly identified, and there are modes above 1000 Hz that are evident. In order to evaluate the mass response and the bipod structural modes, we will break the plot in Figure 9 up into three different ranges, 10 Hz to 100 Hz, 100 Hz to 1000 Hz, and 1000 Hz to 2000 Hz. First, we will examine the break frequency of the bipod shown in the plot from 10 Hz to 100 Hz. The bipod break frequency shows shifts with the change in the base input level. Care was taken to ensure that the temperature of the isolator was at room temperature before the start of each test. The break frequency and calculated transmissibility changed slightly as the input level increased. This is consistent with the theory for the isolator. An explanation of the three parameter isolator theories that predict shift in frequency and change the isolator break frequency and damping is included in Reference *. The isolator design is such that as the damping (Ca) decreases because of the heating of the isolator, the isolator is more closely tuned to the input level which slightly decreases the break frequency and the isolator calculated transmissibility. We measured the temperature rise of the damper housing during the 16.25 g test and the increase in temperature was 18.8 "C (34 OF). The effect of the temperature change is clear in Figure 9. In the frequency range from 100 Hz to 1000 Hz we see that there is a lateral mode of both of the isolators in the bipod at 21 5 Hz and the isolator spring body modes at 615 Hz. This data is consistent with the data collected in previous testing. The 615-Hz mode can be used to evaluate the transmissibility of the main spring surge mode. The worst-case transfer function peak of the 615-Hz mode is approximately 0.9. If there was no mode at this location, the mass response should be approximately 0.02 therefore, the calculated transmissibility of the mass response due to the spring surge mode is approximately 45. This is likely an optimistic estimate since this measurement is not in-line with the isolator. In the frequency range from 1000 Hz to 2000 Hz, there are some structural modes of interest in the payload mass response transfer function. The first mode of interest is at 1 190 Hz. The transfer function approaches a calculated transmissibility of 0.12 or a factor of 8 below bipod base input. There are additional modes above 1 190 Hz but there levels are even lower. Evaluation of the higher modes is still in progress. 147

Document

AMS_2006.pdf

Transfer Function, 5Hz to 2KHr Sine Vibration Test of Bi-Pod with 40Lb Mass, Mass response at Channel 8 only Relative to the Base Input I I IIIII I I IIIII T--T i 1-rii I I IIIII I 1 IIIII I I IIIII 100 1000 10000 Frequency (Hz) 10 1-8tol,lst.25g -8to1,2.5g’s -8to1,7.5g’s -8to1, 12g’s -8101 (16.25g’s) -8tol,Final.25g’s 1 Figure 9. Transfer Function of the Mass Response only at the Bipod Interface. Since the in-line response of the mass is much the same in both accelerometers, the plot in this figure is just the mass response at the bipod interface to the 18.1 -kg (40-lb) mass. These Isolators work so well that for low inputs (0.25 g’s), the high frequency response is less than the accelerometer noise floor Final Conclusions A new D-Strut@ isolator has been developed for a Laser Communication Terminal experiment. This new D-Strut@ isolator advances the state of the art for dual mode isolators in the space environment. This new D-Strut@ isolator increases launch load capacity by at least a factor of two with the limiting factor now being the structure and not the damper bellows as in previous designs. This new D-Strut@ isolator was exposed to a sine vibration environment of 16.25 g’s and still functions properly afterwards without changing it’s isolation capabilities. This new D-Strut@ isolator reduces the shock environment by a minimum of 31 times (-30 dB). A customer concern with the new D-Strut@ isolator is the main spring surge mode at 615 Hz. This mode may affect the line of sight pointing capability of the Laser Communication Terminal instrument. To solve this problem we have completed testing using constrained layer damping, or the using Tuned Mass Dampers on the spring to reduce the main spring surge mode. At the time of this writing, we have not completed the evaluation of the test results but firs indications is the constrained layer damping has no effects on the surge mode and tuned mass damping has significant effects. We will have some conclusions about the additional damping methods available in a future paper. References 1 “Performance of a Launch and On-Orbit Isolator”, Jim Boyd, T. Tupper Hyde, Dave Osterberg, Torey Davis; Presented at the Smart Structures and Materials Conference; SPIE, March 2001 2 “Advanced 1.5 Hz Passive Viscous Isolations System”; Porter Davis, David Cunningham, John Harrell; presented at the 35th AlAA SDM Conference, April 1994. 148

Document

AMS_2006.pdf

Design and Testing of a Low Shock Discrete Point Spacecraft Separation System Pete Wolf and Daryn E. Oxe" Abstract A separation system was designed for use on a standard Lockheed Martin satellite bus structure to allow the satellite to separate from standard launch vehicles. The separation system had two key design requirements: six discrete point attachments and a low shock separation (e1600 GIs). The design solution was a system utilizing a heritage separation system component and a previously unused low shock release device for booster separation, a Split Spool Release Device (SSRD). This paper describes the overall design of the system as well as the unique challenges encountered during component and system level design and test. The challenges were primarily associated with achieving the low shock requirement imposed on the system and the integration of previously unused components into a new separation system. Introduction The Discrete Point Spacecraft Separation System key requirements are: six discrete point attachments and a low shock separation. These requirements were dictated by the spacecraft bus structural design and the use of shock sensitive electronics. The structure is comprised of hexagonal bulkheads that are connected to a central cylinder via radial panels. The radial panels are the primary mounting location for all component electrical boxes on the spacecraft. They are also the primary load path for all launch loads and the location of the discrete point attachments. In order to meet the low shock requirement, a new low shock device had to be implemented because a standard pyro-actuated separation nut could not meet the low shock requirements. After evaluating several candidate device designs, a Split Spool Release Device (SSRD) was chosen as the restraint device in the separation system. This device has never been used in a spacecraft separation system to date and there were many challenges associated with using this relatively new device. Design Description The LM9OOA spacecraft bus does not mount directly to a standard Payload Attach Fitting, instead using a secondary structure called a Booster Adapter Assembly (BAA). By using a secondary structure, the separation plane as well as all associated separation components can be tested with the spacecraft to minimize the amount of testing that occurs at the launch base. This allows a separation demonstration to be performed prior to shipment to the launch base. The BAA is a large cylindrical ring that bolts on to an industry standard 66 Payload Attach Fitting. Attached to the cylindrical ring of the BAA are 6 A-frame brackets. These brackets are where the six SSRD's are mounted along with six Spring Mechanisms. The SSRD's provide the restraint of the spacecraft during launch and provide a low shock release upon command. The Spring Mechanism provides the separation tip-off force for the spacecraft during separation. Shear loads between the launch vehicle and space vehicle during launch are carried by a cup/cone interface. The A-frame brackets contain the conical half and the cup half is located in the aft bulkhead of the spacecraft. Mounted inside the radial panels of the spacecraft, is a custom designed low shock Retraction System used in conjunction with the SSRD's. The Retraction System ensures the preload bolt clears the separation plane after release and provides positive retention for the life of the spacecraft. The Discrete Point Separation System and the components -have been completely qualified for use in a space environment and several system-level separation demonstrations have been performed. Figure 1 shows the basic components of the Discrete Point Separation System. ** NEA Electronics, Inc., Chatsworth, CA Lockheed Martin Space Systems Company, Sunnyvale, CA Proceedings of the 38'" Aerospace Mechanisms Sympsosium, Langley Research Center, May 17- 19,2006 149

Document

AMS_2006.pdf

Spring Mechanism Bolt Catcher Shell Retraction Nut & Retraction Spring .-.-.-.-.-.-.-.-.-.- Separation Plane Figure 1. Discrete Point Separation System Components The heritage component of the separation system is the Spring Mechanism. The Spring Mechanism has extensive flight history and was also used in the previous iteration of this separation system design. The Spring Mechanism’s primary purpose is to provide the separation tip-off force from the booster vehicle. It delivers this separation force by utilizing a spring-loaded plunger configuration of which there are nine configurations capable of 225 - 525 N of force. The Spring Mechanism also has adjustment capability that allows each unit to exert an exact force and allow for mounting adjustments. These design features allow the Spring Mechanism to be used in a variety of applications and meet a wide range of requirements with only a slight impact to the weight of the unit. Several years ago, NASA and ESA expressed a need for an alternative to traditional explosive actuators. NEA Electronics Inc. satisfied this objective by developing a reliable, fast-acting, sure release, low shock output, redundant, non-explosive separation mechanism called a Split Spool Release Device. The SSRD is an electromechanical separation nut-release mechanism that eliminates the residue and shock produced by explosive devices and are factory refurbishable for extended use. The device has significant flight heritage primarily as a hold down and release mechanism of deployable structures (solar arrays, antennas, etc). This is the first spacecraftlbooster discrete point separation system design utilizing the SSRD as the restraint device. The SSRD is a mechanically and electrically redundant and in this application is capable of 55.6 kN of preload. The typical generated shock output of a single device at maximum preload is less than 500 G’s. Table 1 shows a brief list of capabilities of the device used in the discrete point spacecraft separation system. 150

Document

AMS_2006.pdf

Table 1. Specification for the 55.6-kN Split Spool Release Device Requirement Ultimate Load Max. Rated Release Load Source Shock Actuation Circuit Actuation Time Qual. TemDerature Ranae Capability 69,500 N 55,600 N 400 G’s Q 55,600 N preload 4 A Q 6 VDC per circuit 25 msec max. -60°C to +105”C The device consists of a load- bearing spool that is split in half. These spool halves capture a spherical rod end and represent the primary load path of the unit. The spool halves are wrapped with a restraining wire that keep the spool halves from moving laterally when load is applied to the spherical rod end. The restraining wire is fixed to one of the spool halves on one end and restrained with two fuse wires on a toggle on the other. The restraining wire puts Figure 2. External View of 55.6-KN 33~~ a small preload on the-fuse wires when the unit is “set”. Upon receipt of the actuation signal, the fuse wire heats up which decreases its tensile strength. The toggle breaks through one or both of the fuse wires depending on the type of actuation signal and the restraining wire starts to uncoil from around the spool halves. The unrestrained spool halves are then gradually driven apart by the preload until the rod end is no longer captured and “drops” through the device. The actuation of this device is very unique in that it gradually relieves the preload, which is the biggest contributor to shock output, and releases the constrained end of the joint. Refurbishment of the unit is achieved by replacing the restraining wire and fuse wire assemblies after actuation and can be done numerous times to allow for ground test. In the Discrete Point Separation System, the SSRD assemblies are jettisoned with the Booster Adapter Assembly after actuation. Figures 2 and 3 show external and internal views of the 55.6-kN SSRD. Fuse Spherical Rod End u Restraining Wire Figure 3. Internal Components of a Typical SSRD 151

Document

AMS_2006.pdf

In similar separation devices, an external system is utilized to ensure retraction of the preload bolt or rod. However, in this application existing hardware was not compatible with the SSRD nor would it meet the low shock requirements, therefore a new Retraction System was designed. The primary purpose of the Retraction System is to carry the preload in the separation joint and retract the spherical rod end out of the SSRD. The system is made up of a Preload Bolt, Retraction Spring, Spring Retainer, Shock Absorber, and Bolt Catcher Can. The Retraction System is assembled in place after the BAA is mated to the spacecraft. The Preload Bolt is threaded into the spherical rod end in the SSRD through the separation fittings. The Spring Retainer and Retraction Spring are then fed over the Preload Bolt and the spring is compressed. Once compressed, a washer and nut are threaded onto the Preload Bolt and the preload is applied. Preload verification is done via a strain gage within the Preload Bolt to assure proper preload. After applying the preload, the Bolt Catcher Can is installed to contain the assembly inside the spacecraft. When the SSRD is actuated, the spring retracts the Preload Bolt and Spherical Rod End out of the SSRD and across the separation plane, eliminating the possibility of hang up during separation. The Retraction Spring also ensures positive retention of the Preload Bolt and Spherical Rod End throughout the lifetime of the spacecraft. Figure 1 shows the components of the Retraction System. Design Challenges The majority of the design challenges were associated with using an SSRD as the primary restraint device. As stated earlier, this device has never been used in a spacecraft separation system and had a significant impact on the design of the Retraction System. There were two key design drivers on the Retraction System related to the design of the SSRD: torque retention capability and mechanical redundancy. By design, the SSRD has no torque retention capability within the unit except for a key on the spherical rod end. Thic posed a significant challenge on ho\n. the unit is preloaded. The spherical rod end within the unit needs to be held fixed in its orientation such that the friction critical surfaces between the rod end and spool halves are not damaged while preloading the unit. There were two potential solutions to this challenge. The first solution was to react the torque through the key in the rod end. The key is accessible through the SSRD housing but reacting a torque load equivalent to the 55.6-kN preload proved to be risky. The second and chosen solution was a unique hardware design used in conjunction with a torque tool to react the torque through the bus structure and not the SSRD. The preload bolt has an external hex feature on the rod that interfaces with an internal hex feature on the spring retainer. The fit between these two component parts allows relative axial motion but both will rotate as an assembly. The spring retainer also has an external hex feature that allows a tool to be installed and bolted to the aft bulkhead of the spacecraft. This tool constrains the rotation of the spring retainer and preload bolt, but allows the bolt to stretch applying a preload. When torque is applied to the preload nut bearing against the spring retainer, the torque is reacted through the tool and onto the aft bulkhead of the spacecraft and not the SSRD. Figure 4. SSRD Torque Retention Tool A key feature of the SSRD is it is a mechanically redundant device. It achieves this redundancy because only one of the split spool halves needs to move laterally in order for the rod end to be released. However, this is only true as long as the rod end is not constrained from exiting the unit. In order to maintain mechanical redundancy at the system level, the retraction system needed to be designed such 152

Document

AMS_2006.pdf

that its functional motion did not inhibit the rod end from exiting the SSRD. Using the baseline designs of the Retraction System, SSRD, Booster Adapter Assembly, and Spacecraft Bus Structure an in depth analysis was done to determine if redundancy would be achieved at the system level. This analysis took into account tolerances on all components as well as utilizing conservative constraints within the system. The results of this analysis showed the baseline design would not work unless slight modifications were made to the SSRD, Retraction System, and Booster Adapter Assembly. All modifications were incorporated into their respective designs and redundancy testing with the Retraction System was successfully performed during SSRD component-level qualification testing. Test Challenges In addition to design challenges there were several test challenges as well. These included minimizing the shock output from the retraction system as well as characterizing the shock output from the SSRD. During spacecraft separation, there are two distinct shock sources from the actuation of an SSRD. These are the shock output from the SSRD unit and the shock output from the Retraction System. The shock in an SSRD is caused by several components within the unit. These are the restraining wire unwinding, a spool halve hitting the inside of the housing, the rod end hitting the base washer as it exits, or a combination of these. The shock from the Retraction System is caused by the preload bolt hitting the top of the bolt catcher can. These two shock sources dictated a course of action: minimize the shock output from the Retraction System and characterize the shock output from the SSRD. As described earlier, the Retraction System consists of a retraction spring that pulls the preload bolt and spherical rod end out of the SSRD. The spring force is sized to ensure retraction of the Preload Bolt and Spherical Rod End; however, this force also causes a sizeable impact of the Preload Bolt into the Bolt Catcher Can Cover. In order to minimize the impact and associated shock output, a shock absorbing material was incorporated into the Bolt Catcher Can design. A development test effort was undertaken to determine the type of material to be used as well as optimize the design. Initially, two candidate shock absorbing materials were considered: Poron and Aluminum Honeycomb. Preliminary test results showed the Poron to be a poor shock absorbing material; however, favorable results were shown for the aluminum honeycomb. The test setup consisted of a Bolt Catcher Can, Spring Retainer, Retraction Spring and a rod and nut of roughly the same mass as the flight designed hardware. The aluminum honeycomb test specimen was cut to the same inner diameter as the Bolt Catcher Can and placed inside the cover at the end of the can. The test specimen was a 2.5-cm nominal piece of aluminum honeycomb pre-crushed to a thickness of 2.22 cm. Tri-axial accelerometers were mounted at the base in two locations 90" apart and on top of the Bolt Catcher Can. Figure 5 shows the test setup. I. Figure 5. Internal and External View of the Aluminum Honeycomb Shock Test 153

Document

AMS_2006.pdf

Four configurations of the aluminum honeycomb were tested in order to find the configuration to minimize the shock output. Each honeycomb configuration went through two test runs and the data was evaluated. Table 2 lists the aluminum honeycomb configuration, and Table 3 lists the maximum shock levels at the three accelerometer locations, and crush depth in the honeycomb due to the rod of each configuration. - Density (kg/m3) 16.0 32.0 110.5 104.0 Cell Size (mm) 9.53 4.76 4.76 9.53 Crush Strength (kPa) 172.4 482.6 3930 351 6.3 Foil thickness (mm) 0.018 0.018 0.064 0.064 Table 2. Aluminum Honeycomb Configuration Data Information I 1X Honeycomb Configuration -007 I -013 1 -015 I -017 Honeycomb Configuration -007 -01 3 -01 5 -01 7 None 167.631 27.025 1240.894 1233.103 622.312 1 Y 1 Z 2X Table 3. Aluminum Honeycomb Shock Data 138.727 17.238 1465.1 70 11 17.588 382.61 6 233.268 29.820 2143.589 1270.930 51 6.098 170.565 18.846 1367.840 1238.71 9 51 6.545 1- -1 22 3X 3Y 257.538 23.989 1863.343 1664.746 431.61 1 41 0.757 289.466 1985.1 65 1739.393 1635.51 6 1090.883 530.638 1321.61 0 1538.292 191 5.1 43 Run I 2~ i 101.467 i 11.751 j 787.916 j 564.251 j 239.809 1 IG'S\ The data collected during the development testing showed that the -01 3 configuration honeycomb had the best shock absorbing characteristics. This was based on data collected on Accelerometers 1 & 2 which were mounted at the base of the Bolt Catcher Can. As the data in Table 2 shows, the shock levels using the -013 configuration honeycomb are much lower than any of the other configurations for all test runs. Data collected at Accelerometer 3 was used for reference purposes but is also lower for the -013 configuration in all but one test run. It's interesting to point out that when compared to no honeycomb the levels significantly lower but the levels seen for the -015 and -017 configuration were higher. This was 154

Document

AMS_2006.pdf

possibly due to the honeycomb increasing the shock impact surface area. These two configurations had a significantly higher stiffness which lends some validity to the observation. Another testing challenge occurred during the component qualification testing of the SSRD. Since this was the first time an SSRD was used in a discrete point booster separation system, the shock output of the unit needed to be characterized fully and extensive generated shock testing was performed. The test setup consisted of a 75-cm square aluminum plate that had a flight-like interface. The interface included a flight like cup-cone interface as well as the A-frame bracket. The SSRD was mounted to the A-frame bracket and the preload was applied through the center of the aluminum plate. The test was performed with the flight designed Retraction System and the crushable honeycomb shock absorber. Tri-axial accelerometers were used to collect data at locations 7.5 cm, 15 cm, and 30 cm from the center of the plate. Figure 6 shows the test setup for generated shock testing. 45"\ \ J L Figure 6. Generated Shock Test Setup The design of the test setup was meant to mimic the exact flight configuration in the preload path; however, for conservatism, the large aluminum plate was incorporated to collect shock levels at varying locations and did not represent the flight honeycomb panel configuration. This created a major challenge during the component-level shock testing because configuration as well as accelerometer location had some bearing on the data. In addition to the component-level generated shock tests, there was a system-level separation and shock test. The purpose of this test was to demonstrate the separation system and to measure the shock output of the system. The BAA was attached to the spacecraft with all the component parts of the separation system. The spacecraft bus was instrumented with 255 tri-axial accelerometers to measure the shock at various locations on the spacecraft bus. The entire assembly was suspended from an overhead crane for the separation test. The SSRD's were actuated in pairs that were diametrically opposed from each other. Upon actuation of the last pair, the BAA separated successfully from the spacecraft and fell into padding below the suspended spacecraft. Data collected during the three actuation events showed shock readings well within specified limits for the spacecraft. 155

Document

AMS_2006.pdf

Conclusions The Discrete Point Separation System was able to bring together a heritage component with a new low shock restraint device and successfully complete space flight qualification. There were many design and test challenges that needed to be overcome throughout the Qualification Program that show the difficulties of implementing a new component. However, understanding the components interaction and their impacts the design and test challenges are able to be overcome. 156

Document

AMS_2006.pdf

Faying Surface Lubrication Effects on Nut Factors Deneen M. Taylor* and Raymond F. Morrison** Abstract Bolted joint analysis typically is performed using nut factors derived from textbooks and procedures from program requirement documents. Joint specific testing was performed for a critical International Space Station (ISS) joint. Test results indicate that for some configurations the nut factor may be significantly different than accepted textbook values. This paper presents results of joint specific testing to aid in determining if joint specific testing should be performed to insure required preloads are obtained. introduction During review of International Space Station analysis documentation and build paper it was discovered that one of the International Partners had overtorqued the installation bolts on the Common Berthing Mechanism (CBM) Ring on each of their elements. The build paper also raised questions concerning lubrication and whether or not it had been applied to the faying surfaces of the bolt head/washer in addition to the threads. The CBM is a mechanism built and qualified by Boeing and furnished to the other partners for use with the pressurized elements. Too high of a preload could overload the bolts when combined with on-orbit thermal and mechanical loads and potentially cause failure. Too low of a preload could cause the joint to gap and leak atmosphere overboard on orbit. Due to this out of configuration condition, it was decided that joint specific testing would be run to quantify the effects of the torque and lubrication. Background The testing was implemented to determine the nut factor, K, of three different bolt materials, A286, MP35N and lnconel in combination with A286 helicoil inserts and nuts, with and without lubrication on faying surfaces. To remove any uncertainties of previous contamination to the bolts, all bolts were cleaned before testing. All bolts are about 6.3-mm (exactly I4 inch) diameter. Countersunk CRES washers were placed under bolt heads and A286 passivated flat washers under the nuts. Helicoils and nuts came from the manufacturer with standard locking features and lubrication. Lubrication applied during testing to bolts was Braycote 81 52 oil. The bolted material is anodized aluminum. The different permutations of the tests are shown in Table 1. Although the objective of this testing was to verify that the higher torque did not yield the bolts and to obtain actual nut factors for these specific configurations, due to the numerous tests and various configurations some interesting trends became apparent. It is important to note that the lubrication was added between cycles not just for the first cycle, reflective of the flight hardware. In addition, the max/min nut factors cited are inclusive from all previous cycles, not just the max and min from the noted cycle. * NASA Johnson Space Center, Houston, TX ** Boeing, Huntington Beach, CA Proceedings of the 38th Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 157

Document

AMS_2006.pdf

Table 1. Permutations of tested bolt configurations i Min Max Test \ Bolt / Inserthut Lubrication Tightenvia Samples Cycles / :KiK ' Bolthead / 9 10 I .18 I .56 i 10 10 I .14 \ .18 i 10 I 10 , .08 , .18 ....................................................... i ........................................ ................................................................ .............................................................................. ..................................................... ........... j ................................................. i... ........................... .......................... + 4 i + i ............................................................. i ................................................................................. 4 ......................................... ..+ ..................... .i ......................................... + ............................ * .............. Case 1-1 ~ A286 ' Helicoil I Threads only Case 1-2 ' A286 , Helicoil Case2-1 MP35N / Nut ; Threads only Case2-2 1 MP35N I Nut i Threads only Case2-3 I MP35N Nut I surfaces of bolt i Case3-1 , Inconel718 ' Nut I surfaces of nut I Case3-2 I Inconel718 I Nut I surfaces of nut i I Threads and faying ' l Bolthead ~ 5 5 / .ll I .14 Case 4-1 Case 4-2 lnconel 718 Helicoil / Threads only I Bolt head / 5 I 5 1 .21 .51 j Threads and faying I Bolt head / surfaces of bolt i .................................................... .................... ............................................ ....................... .................. ...................................................... ........................................................ ................................................................ ...................................................... ......................... ................................................ ......................... I Nut / Bolthead I 10 / 10 .23 ; .49 i- _____. i : i 4 i i 4. 2 i i I. + 4 4 ................................................................. ............ ...... ........................... ................................................ ............................ .. ............................................ j + i i 4 4 i i .i I Threads and faying Bolt head , ' Threads and faying I 5 ' 10 i .10 ; .21 i Threadsandfaying Nut 141 I 5 ~ .12 .17 ...... ............................................. ........................... ..... .................................................. ................................................................ ....................... ...................................................... ...., L- ! -. 4 i Bolt head ~ 4 1 5 / .13 ~ .45 .............................................. ........................... .................................................. ...................... ............................................. ............................... ;---+ .../ ..i. 4 j 4 4 ..................................................... ............................................................. .......................................................................................................................................... ..................... ......................................... ........................... ........ j + ~ i. ...I 4 i .............................................. ............................. .............. ..................................................... c 4 ................................................................. ?urfaCeS..of.~!t ..................................................... ; i i (.. lnconel 718 1 Helicoil Test Cases 1-1 and 1-2 Test cases 1-1 and 1-2 compared the effects of lubrication on threads only with lubrication on threads and faying surface under bolt head for the A286 bolts. Both cases were tightened at the bolt head. Data indicates that for bolt/insert configurations, lubrication on the faying surfaces under the head of a bolt significantly reduces the maximum nut factor and the scatter (Figure 1). After 10 cycles, the faying-surface lubricated bolts had a &, = 0.18 and bin = 0.14, whereas the threads-only lubricated bolts had a K,,,, = 0.56 and Lin = 0.18. Without lubrication on faying surfaces, increasing bolt cycles also significantly increases nut factor. This is presumed to be because lack of lubrication on the faying surface induces wear under the washer, then additional cycles exacerbates the wear increasing the nut factor (Figure 2). Test Cases 2-1.2-2 and 2-3 Test cases 2-1 and 2-2 compare the affect of tightening the MP35N bolts via nut versus tightening via bolt head. Both cases had lubrication on the threads only. Bolt tightened tests were significantly higher. After 10 cycles, nut-tightened nut factors were K,, = 0.18 and bin = 0.08. Bolt-tightened nut factors were K,, = 0.49 and Gin = 0.23. This is an increase of almost 75% (Figure 3). Test cases 2-2 and 2-3 compared the affects of lubrication on threads only with lubrication on threads and faying surface under bolt head for the MP35N bolts. Both cases were tightened via bolt head. Data indicates that for bolts with lubrication applied only to the threads the nut factor will be much higher than for bolts with lubrication applied to the faying surfaces under the bolt head in addition to threads. After 10 cycles, the threads-only lubricated bolts had a K,, = 0.49 and Lin = 0.23 whereas the faying-surface lubricated bolts had a Lax = 0.21 and Kmin = 0.1 0. This is an increase of over 100% (Figure 4). Test Cases 3-1 and 3-2 Test cases 3-1 and 3-2 are a comparison of tightening lnconel bolts via bolt head with tightening via nut. Both cases had lubrication added to threads and to the faying surfaces at the nut. The bolts tighten via bolt head had similar nut factors as those tightened via nut for the first cycle, but after just 2 cycles bolt- tightened cases had significantly higher maximum nut factors. Bolt-tightened cases produced nut factors of K,, = 0.45 and K,,,,,, = 0.13. Nut-tightened cases produced K,, = 0.17 and Kmin = 0.12 (Figure 5). Test Cases 4-1 and 4-2 Test cases 4-1 and 4-2 compare the affects of lubrication on threads only with lubrication on threads and faying surface under bolt head for the lnconel bolts. Both cases were tightened via bolt head. The bolts that had lubrication both on threads and on faying surface under bolt head had lower nut factors than those bolts lubricated at threads only. After 5 cycles, the faying-surface lubricated bolts had a K,, = 0.14 and Kmin = 0.1 1 whereas the threads-only lubricated bolts had a K,,, = 0.51 and Kmin = 0.21 (Figure 6). 158

Document

AMS_2006.pdf

A286 CRES Bolt, CRES helicoil, CRES countersunk washer 0.6 0.5 0.4 Y i B 0.3 LL 0.2 0.1 0 0 / / /--- - -Lube Thread only, K rnax Lube Thread & Head, K mir - -Lube Thread & Head, K ma / / -.- 2 4 6 8 10 Cycles Figure 1. Comparison of test cases 1-1 and 1-2. 12 , i .. Figure 2. Photo on left from Test Case 1-1, max preload 15 kN (3400 Ib) Photo on right from Test Case 1-2, max preload 8.0 kN (1800 Ib) 159

Document

AMS_2006.pdf

0.6 0.5 0.4 Y 0.2 0.1 MP35N Bolt, A266 CRES nut, CRES countersunk washer Lubrication on Threads Only /- / / ~ .c 0 - -Tghtened via Nut, K max -Tightened via Bolt, K rnin - - Tahtened via Bolt. K max -I-- -++- -+-e- ------- C 0 2 4 6 8 10 12 0.6 0.5 0.4 Y 5 0.3 L s 0.2 0.1 0 Cycles Figure 3. Comparison of test cases 2-1 and 2-2. MP35N Bolt, A266 CRES nut, CRES countersunk washer Bolt lightened via Head ~~ -.-- ____ ~- ._ ~---- __.- - ~~~ ------- ~~~ . /- - -Lube Thread only, K rnax Lube Thread & Head, K rnir / / ~~ 0 0 2 4 6 8 Cycles 10 12 Figure 4. Comparison of test cases 2-2 and 2-3. 160

Document

AMS_2006.pdf

0.6 0.5 0.4 Y 0 0.3 U z ii L 0.2 0.1 0 0 0.6 0.5 0.4 Y L- 2 0.3 B 0.2 0.1 Inconel718 Bolt, A286 CRES nut, CRES countersunk washer Lubrication on Threads and Faying Surfaces of Nut Tightened via Nut, K min Tightened vla Nut, K max -Tightened via Bolt, K min __ - _._ 0 0 -. 0 0 0 0 0 0 0 0 0 // // 0 # / ------------- 0 -4-- _. . 1 2 3 4 5 Cycles Figure 5. Comparison of test cases 3-1 and 3-2. Inconel718 Bolt, CRES helicoil, CRES countersunk washer L - c- - -Lube Thread only, K max Lube Thread 8 Head, K min Lube Thread 8 Head, K ma -e-- /- 0 / / / 0 0 0 1 2 3 Cycles 4 5 6 6 Figure 6. Comparison of test cases 4-1 and 4-2. 161

Document

AMS_2006.pdf

Conclusion In conclusion, bolts that were tightened via nut (test cases 2-1 and 3-2) produced similar nut factor ranges (K = 0.11 to 0.18) as those typically found in literature for “lubricated” steel bolts (K = 0.12 to 0.20) regardless if lubrication was added to the faying surface under the nut or not. In all published sources reviewed, if the term “lubricated was defined it meant lubricated at the threads and “non-lubricated” would mean no lubrication was added to “as-received condition, but bolts were not specially cleaned to remove any residuals from machining/processing. Of course, if lubrication is added to the threads it can very easily migrate to the nut faying surfaces hence explaining the similarity. However, for bolts tightened at the head, there are some significant differences. If lubrication exists on the faying surface of the bolt head (test case 1-2, 2-3 and 4-1), nut factor ranges typical for nut-tightening were produced (K = 0.1 0 to 0.21), but if there was no lubrication on the faying surface of the bolt head (test cases 1-1, 2-2, 3-1 and 4-2), the nut factor ranges were significantly larger (K = 0.18 - 0.56). It made little difference for the bolts tightened via head whether it was threaded into a nut or insert. A fairly comprehensive literature search was performed to look for nut factor data on bolts tightened at the head or effects of additional faying surface lubrication. Nothing specific was found. However, one source had a very brief statement that acknowledged K values could be “up to 50% higher” for bolt head tightening 111. It is very clear from this testing that general “text-book nut factors values can not be assumed to be appropriate for non-lubricated faying surface bolts tightened at the bolt head. Testing demonstrates that as early as the second cycle the nut factor increases significantly when tightened at the head without lubrication at the faying surface. This change in nut factor needs to be considered for reworking of a nonlubricated bolted joint or where a mix of installation cycles exists in a critical joint to maintain the desired preload. Torque tension joint specific testing was not required for the majority of ISS Space Station hardware. Test results indicate that for future programs, testing should be considered for critical joints where it is not possible to apply lubrication at the faying surface. References 1. Engineering Sciences Data Unit “Applying, Measuring and Maintaining Pretension in Steel Bo!ts.” 2. National Aeronautics and Space Administrations. “Standard, Threaded Fasteners, Torque Limits for” 3. Gibson, James N. “Torque Tension Tests for the Common Berthing Mechanism (CBM), International 4. Morrison, Raymond F. Analysis of CBM Bolts for Overtorque Condition” Boeing Report D684-12216- 5. Shigley, Joseph E., and Mischke, Charles R.(1996). Standard Handbook of Machine Design. 2”d ed. 6. Bickford, John H. (1 995). An lntroducfion to the Design and Behavior of BoltedJoints. 3rd Rev ed. New ESDU Item Number 8601 4, July 1986 MSFC-STD-486 Rev B, November 1992. Space Station” Boeing Report M&P-3-1638, June, 2005 01, June 2005 New York: McGraw-Hill. York: Marcel Dekker. 1 62

Document

AMS_2006.pdf

Torque Loss and Stress Relaxation in Constant Torque Springs Robert W. Postma^ Abstract Constant torque springs are manufactured from spring steel strip that in some applications is stressed beyond the yield strength to achieve maximum torque-to-weight ratio. An adverse consequence of a high state of stress is torque loss resulting from stress relaxation, which typically occurs over a prolonged period of time, accelerated by thermal cycles or continuous elevated temperatures. This poster paper discusses a case of torque loss resulting from thermal cycling of a spacecraft hinge spring manufactured from Type 301 corrosion resistant steel strip, cold worked to the extra hard condition. The equations governing the design of the constant torque (Neg’ato?) spring are reviewed. Included is a discussion of ongoing work to better understand the design, manufacturing, and stress relieving of constant torque springs, particularly in regard to stress relaxation and delayed cracking from sustained high stress levels in aerospace environments. Introduction Constant torque springs are sometimes stressed beyond the yield strength, by design, in order to obtain maximum torque-to-weight ratio. The springs are initially fabricated as a tightly wound coil of steel strip, and may have multiple laminates. The material is stress relieved, possibly before and after forming into the coil, at temperatures up to 425°C. When installed on the hinge, the spring is reverse flexed from the take-up spool onto the spool driving the output shaft (Figures 1 and 2). Stow Direction (Shown in Deployed Position) carp. I Figure 1. Two Constant Torque Springs Installed in 90-degree Hinge The Aerospace Corporation, El Segundo, CA Figure 2. Constant Torque Spring Critical Dimensions Proceedings of the 3dh Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 163

Document

AMS_2006.pdf

This work was initiated following loss of torque in spacecraft hinge mechanisms after a number of temperature cycles during thermal vacuum testing. The torque loss at the end of the deployment motion for one of the 180-degree hinges was 10 percent, as shown in Figure 3. (For a 90-degree hinge, as pictured in Figure 1, the torque loss was 7 percent). Note that the torque at the stowed position remains unaffected. Evidently, it is the slope of the deployment curve that is affected by the thermal cycling during this particular test. On tests of other hinges, however, the slope remained approximately the same, but the mean torque level decreased. Empirical Analysis of Torque Loss After considering a number of other possible explanations, it was concluded that the 10 percent thermal cycling torque loss was due to stress relaxation. The most likely other explanation, increase in interlaminate friction, was eliminated by a special test using only one laminate. (The torque loss was about the same amount). Since stress relaxation is a function of time, temperature, and stress level, it can be further surmised that the 37°C elevated temperature portion of the thermal cycles, combined with high bending stress levels, are the primary factors contributing to the observed torque loss. The Figure 3 data also shows 5 7 percent hysteresis and an initial negative slope of 15 percent of the maximum torque. It is also surmised (tentatively) that these adverse effects can be attributed in some measure to high bending stresses. I I I 501 Courtesy Lockheed Martin Gorp. (1 in4b = 0.1 13 N.m) -190 -180 170 -160 150 -140 -130 -120 -110 100 -90 -80 -70 60 50 -40 -30 -20 -10 0 Angle (dog) Figure 3. Torque vs. Angle for 180-degree Hinge As the torque loss analysis proceeded, calculations were made that indicated that the springs were stressed beyond the yield strength (discussed in the following section). This conclusion is supported by examination of springs used for development testing, shown in Figure 4. Yielding of the spring stock from the high operating stress and strain levels after being reverse flexed is evident from observed distortion of the springs. The material is Type 301 corrosion-resistant steel, 0.0229-cm (9-mil) thick, cold worked to the extra hard condition (60 percent cold reduced). It is seen from Figure 4 that after being removed from the hinges, the springs are no longer in a tightly wound coil, but are expanded out of shape into a loose spiral. It can be concluded that being stressed beyond yield, combined with thermal cycling, contributes to stress relaxation and the associated reduction of torque. 164

Document

AMS_2006.pdf

Courtesy Lockheed Martin Corp. Figure 4. Springs Showing Various Degrees of Yielding To find out if performing multiple thermal cycles has resulted in the springs achieving a state of stress stability, a special test was performed where the torque output was measured after every three thermal cycles. It was found that the torque became essentially constant between 9 and 15 cycles at a level that still allowed sufficient torque margin. Additional thermal cycles are planned to confirm this. Stress and Strain Analysis The practice of operating constant torque (Neg’ato?) springs at high stress, sometimes beyond yield, derives from general usage where the design stress level is based on low cycle fatigue requirements and the number of operational cycles. The invention of this type of spring is credited to Frank A. Votta Jr. In his technical paper’, he based the number of allowable cycles on a formula for (what he called) stress factor, fs, shown here in Equation (1). fs = Stress Factor t = Spring Strip Thickness D1 = Fabricated Diameter (Fig.2) D2 = Drive Spool Diameter (Fig. 2) This is identical in form to the usual formula from strength of materials for calculating bending strain in the outer fibers. &b = Bending Strain The first term, t/D1, is the bending strain from the change of curvature during the transition from the original coiled diameter D, to the straight section between the coils. The second term, t/D2, is the strain from further changing the curvature by reverse flexing the spring from the straight transition to the diameter of the drive spool DP. Stress vs. Strain Characteristics of Spring Steel Strip If one were to calculate the quasi-linear bending stress on the assumption that the modulus of elasticity in simple tension were constant and equal to the compression modulus, we would come to the impossible finding that the calculated bending stress would exceed the ultimate stress. The spring represented by Figure 3 has a calculated bending strain of 1.1 percent. Based on a modulus of elasticity of E = 193,000 MPa (28,000,000 psi) and a Poisson’s ratio of 0.3, the linear formula for bending stress 165

Document

AMS_2006.pdf

(Equation 3) would give a value of 2330 MPa (338,000 psi), which is well above the characteristic ultimate tensile strength of 1860 MPa (270,000 psi) for extra hard Type 301. &E a, = (1 - v2) (3) ob = Bending Stress (in Outer Fibers) E = Modulus of Elasticity in Tension v = Poisson's Ratio This formula represents the quasi-linear change in bending stress from the state of residual stress in the fabricated spring coil. (This method of calculating the bending stresses in a constant torque spring is summarily discussed in the Associated Spring Design Handbook2, which also cites Votta's paper). The spring manufacturer can intentionally pre-stress the spring in a process called strain hardening, to induce residual bending stresses of opposite sign, and thus reduce the stresses when the spring is installed on the hinge or spring motor. (It is tentatively believed that this was not done for the particular springs involved in this case of stress relaxation). Quasi-Linear Stress-Strain Curve for E = 193,000,000 kPa (28,000,000 psi), nominal (1 psi = 6.894757 kPa) (Courtesy of the publishers of the ASMH, Reference 3. The original data is from the Allegheny Steel Corp, 1958). 0 0004 0008 0 0004 0.008 Strain. adin. FIGURE 3.0212. STRESS-STRAIN CURVES FOR SHEET AND STRIP COLD ROLLED TO FULL- HARD AN0 EXTRA-HARD TEMPERS (11) Figure 5. StresdStrain Curves for Type 301, Extra Hard Condition Tensile and compression test curves for extra hard Type 301, from the Aerospace Structural Metals Handbook3 (ASMH), are shown in Figure 5. A hypothetical (quasi) linear curve is added to the graph. From comparison of calculated operating strains for constant torque springs with typical published tensile or compression test data (as shown in Figure 5), it can be concluded that the behavior of steel strip at high strain levels is different in bending compared to pure tension and compression. Type 301 extra hard shows higher values of elastic modulus in tension than in compression. The tensile and compression curves start to droop near peak values, such that the ultimate failure stress is less than would be predicted by the nominal assumption that the stress strainkurve is linear all the way to ultimate. Stressktrain curves published in the ASMH for extra hard Type 301 typically terminate at 1.0 percent strain, (although some of the numerical data in the ASMH state elongation to failure greater than 1.0 percent). Another example that infers greater elongation to failure in bending compared to simple tension is the required bending test for a slightly less cold worked material, Type 301 full hard. (Stressktrain curves for 166

Document

AMS_2006.pdf

Type 301 full hard are also shown in Figure 5). Type 301 full hard is required by MIL-HDBK-5F4 to be capable of being bent around a rod of diameter six times the thickness of the spring material, without the occurrence of surface cracking. Using Equation 2, the average bending strain calculated for 0.0254-cm (10-mil) thick strip bent around a 0.1524-cm (60-mil) diameter rod (neutral surface assumed at the 0.1 778-cm (70-mil) mean diameter) is 14.3 percent, compared to the MIL-HDBKdF requirement for minimum elongation to failure in simple tension of 8 percent. No similar bending requirement is given for the extra hard Type 301, as this level of cold work is not controlled by MIL-HDBKd or by ASTM specifications. Accordingly, it is prudent for the end user to provide specifications for the extra hard version, rather than rely entirely on the spring vendor’s judgment and practice. For a limited number of cycles at ambient temperature, strains beyond the yield strength (based on 0.2 percent offset) can be tolerated without failure. From a graph in Votta’s paper, his recommended allowable stress factor (strain) for less than 5000 cycles is 2.0 percent, for what he calls “safe design”, using 1095 carbon steel. In the Associated Spring Design Handbook2, a graph is presented showing an “allowable” (quasi-linear) stress of 2,760 MPa (400,000 psi) for fewer than 260 cycles. In actual practice, however, the design of constant torque springs is based on empirical data, rather than Votta’s formulas. In a couple of other major spring manufactures’ design guides, Type 301 is rated at up to 2000 cycles for strains above 1 .O percent or greater. (Typically 50 to 100 cycles are considered adequate for life testing of deployables that only operate once in orbit, plus the cycles needed during manufacture and testing). Normally, one would not be concerned over delayed cracking in typical spacecraft storage environments. However, for springs stowed at stress levels beyond yield over prolonged periods, this aspect merits some consideration. Type 301 is on the list of materials considered to be resistant to stress corrosion cracking; however, when cold reduced to the full hard or extra hard state, a phase transformation from austenite to martensite occurs, which would logically obviate a totally exempt status. Looking into this further, the storage atmosphere is typically at controlled humidity, and at the launch site the spacecraft is usually in an air-conditioned payload housing. One reference, by Phelps and Loginow , cites an exposure test to atmospheric corrosion at Kure Beach, N.C. of Type 301, 60 percent cold reduced, for periods of 240 to 370 days, during which the test specimens showed “excellent resistance to stress corrosion”. However, these specimens were only stressed to 75 percent of the stated yield strength of 1640 MPa (238,000 psi). Data for springs stressed beyond yield, sustained over typical aerospace storage periods of five years, has not been located. On the other hand, neither has breakage of this type of spring been reported, due to delayed cracking (or for any other reason) in aerospace usage. Torque Determination 5 In Votta’s technical paper, he gives a formula for calculating the torque on a constant torque spring motor. The formula currently derived (Equation 4), based on strain energy, is almost the same as Votta’s except for the factor in the denominator involving Poisson’s ratio. (4) T = Torque W =Width (Figure 2) This factor also appears in the quasi-linear stress formula, Eqn. 3, and is conventionally used to adapt the standard handbook formulas for beam bending of exceptionally wide beams. It results from an elastic state called “plain strain”, which is applicable to a thin, wide beam (strip) in bending. (An interesting comment from A. M. Wahl, a noted authority in spring design of that era, was published in the “Discussion” portion of Votta’s paper. Wahl stated that this factor should have been present in Votta’s formula. Votta’s response was that their test data showed that this factor did not apply because of energy lost in transverse [anticlastic] curvature of the straight length of spring between the two spools.) Equation 4 implies that the torque should be constant with angular displacement. However, as seen from Figure 3, there is a fairly linear, negative slope to the deployment curve. It is surmised that stresses beyond yield contribute to the magnitude of the negative slope. Hysteresis is another component of 167

Document

AMS_2006.pdf

torque loss that adds to the total loss. From tests with single laminates it was found that interlaminate friction is not a significant cause of the hysteresis. (The single laminate has approximately the same hysteresis as multiple laminates). As of now, it is not known how much of the hysteresis is internal to the material from stress related yielding of the spring stock, and how much is due to bearing friction. One broad objective is to better understand why the physics (applied mechanics) behind the theoretical formula for torque (Equation 4) does not account for the downward slope of the torque vs. deflection curves. This will assist in design optimization of the relationships between the variables affecting torque and torque loss. It may be that optimum design would be achieved at lower stress levels, by reduced laminate thickness and increased number of laminates. It is desirable to minimize this decreasing torque slope, because the maximum torque is usually needed at the end of deployment to actuate end-of-travel latches or wind-up cable bundles. Conclusions and Recommendations This poster paper has discussed how general practice for the design and manufacturing of constant torque springs using empirical design data provides springs that may be stressed beyond the yield strength. Typically, highly stressed springs are subject to stress relaxation and related torque loss under prolonged load and elevated temperature conditions. It appears that torque loss due to thermal cycling, for the springs cited herein, had achieved a state of stability, in that the torque output of the unwound spring actuated hinges was not continuing to decrease with each successive cycle. Additional cycles are expected to confirm this. It also remains to be determined whether maintaining stresses above the yield strength over periods of prolonged storage in aerospace environments can be done without excessive loss of torque from stress relaxation and without loss of immunity to delayed cracking. The final outcome from this work will likely be that Type 301 extra hard corrosion resistant steel will continue to be an optimum material for one-time operation in space of constant torque spring actuated hinges. Particularly for the extra hard condition, which is not controlled by ASTM or military specifications, the end user and spring manufacturer should be mutually cognizant of spring material requirements, such as yield and ultimate strength and ductility, and stress relieving and pre-stressing procedures. Acknowledgments This work has been conducted interactively with the Lockheed Martin Corporation, Sunnyvale CA. The Lockheed Martin portion of the investigation is under the cognizance of Stuart Loewenthal. The hinge drawing (Figure l), test data (Figure 3), and sample springs (Figure 4), were provided by Lance Lininger, courtesy of Lockheed Martin. Usage of the single figure from Reference 2 has been verbally approved by telephone discussion with the publisher, Purdue University. References 1. “The Theory and Design of Long-Deflection Constant-Force Springs”, F. A. Votta, Jr., ASME Paper No.51-F-11, June 11,1951. 2. Desiun Handbook, Enaineerinu Guide to Sprinu Desian, Associated Spring, Barnes Group Inc., 1987 Edition. 3. Aerospace Structural Metals Handbook, 39th Edition, Vol. 2; Code 1301, Original Code1 301 Author - W. D. Kloop, March 1988; Published and Updated by Purdue-University. 4. Metallic Materials and Elements for Aerospace Vehicle Structures, MIL-HDBKdF, DOD, Nov. 1, 1990. “Stress Corrosion of Steels for Aircraft and Missiles”, E. H. Phelps, and A. W. Loginow, Feb. 3, 1960, 1 6th Annual Conference of Corrosion Engineers, March 14-1 8, 1960. 168

Document

AMS_2006.pdf

Mechanical Design of a Multi-Axis Martian Seismometer Franck Pecal’, Nicolas Paulin*, Doug Mimoun” and Gabriel Pont” Abstract A planetary seismometer intended to fly onboard future missions to Mars has been developed by SODERN as the industrial partner of IPGP and CNES. This seismometer is a Very Broad Band seismometer (VBB). Its main purpose is to measure the Martian seismic activity. It is included in the Geophysical Package of the ESA ExoMars mission (to be launched in 201 1). This paper describes the mechanical architecture of the seismometer and identifies the main challenges that were overcome. In particular, the paper points out how SODERN managed the requirements for the extremely high sensitivity of the seismometer and coped with the very low size and mass budgets together with the very harsh mechanical and thermal environments specific to Martian missions (especially very low temperatures and severe shock when landing). The lessons learned from the full test program results are presented. Introduction Since 1992, EADS SODERN has been the industrial partner of IPGP (Institut de Physique du Globe - Paris) and CNES (Center National d’Etudes Spatiales) for the development and the industrialization of planetary seismometers. Following the launch failure of the Mars 96 mission and the subsequent loss of the seismometer “OPTIMISM”, in 2002 CNES and IPGP selected SODERN as their partner in the development of a new multi-axes seismometer. This seismometer is a Very Broad Band seismometer (-0 to 20 Hz). It is intended to measure the Martian seismic activity in a network of seismometers. It was initially developed for the joint CNES/NASA “NetLander” mission. Although this mission was stopped in 2003, development went OR until the end of the B phase (preliminary design). The VBB is now part of the geophysical package instrument suite of the ESA Exomars mission, to be launched in 201 1. The aims of this B phase were to demonstrate: 0 * 0 The seismometer detection performance with a Breadboard model The mechanical robustness and stability under thermal and mechanical environments with a Structural and Thermal Model (STM) The compliance of the definition of a future Flight model with regard to the main requirements for the Martian missions The VBB is currently the only planetary seismometer available for a launch towards Mars in 201 1. Its characteristics make it compatible with Mars, Moon and Mercury environments. With some modifications, it may be compatible with the “telluric” satellites of the outer planets such as Europa and Titan. f. EADS SODERN, Limeil Brevannes, France + CNES, Toulouse, France IPGP, Saint Maur, France Proceedings of the 3@ Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 169

Document

AMS_2006.pdf

General Description Princide The seismometer is based on the principle of an inverted pendulum with an angle of 35.25" wrl vertical axis (Figure 1). This kind of configuration gives a high sensitivity for a low mass and reduced volume. It also allows detection in the horizontal and vertical directions with the same sensor. The pendulum equilibrium is realized when the leaf spring moment balances the gravity moment. Figure 1. Seismometer model The leaf spring transmits the movement to the mobile mass when a quake occurs. The mobile mass vibrates and oscillates around the axis of the pivot. The position of the center of gravity is governed by Eq. (1): d2a da dt2 dt J -+ p- + k (a-a,) = -mgD, sin@-a,) + m0, ysof (1 1 with: cx angular position of mobile part J moment of inertia of the mobile part ,O : coefficient of viscous friction k: stiffness of the assembly pivot-spring m: mass of the mobile part g: gravity The transfer function between the displacement of the mobile part X, and ground acceleration ysol follows Eq. (2): k - mgD, cos(@ Q=- 0,z.J P m; = J The displacement of the mobile part is measured by two sensors: a short term sensor called DCS (Differential Capacity Sensor) and a long term sensor called OCS (Oscillating Cavity Sensor). 170

Document

AMS_2006.pdf

A feed-back system, based on a magnetic coil actuator, is used to control the mobile part position and set it at the equilibrium position. A closed-loop system is necessary to widen the measurement bandwidth of the system and to give a representative output voltage of the ground acceleration. Another digital feedback loop is used to compensate for the long term and daily thermal effects. The principle of the seismometer is illustrated by the diagram block Figure 2: CoiVMagnet Feed-back Figure 2. Block Diagram of the VBB seismometer VBB sensor The Breadboard model of the Very Broad Band sensor developed during the R&D program is described in Figure 3. OCS sensor (shielding) \ Inclined fixedpart 1 CoiVMagnet actuator '/- (shielding) A Counter / mechanism balance Fixed plate AB Pivot Leaf spring F7 c c -P Connectors 1 Earth compensation gravity mass L Mobile part Figure 3. Breadboard model of the VBB sensor In addition to the main parts of the seismometer, two other components are essential to operate the seismometer. A counter balance mechanism is used to adjust the equilibrium position of the mobile part. This adjustment is necessary according to the uncertainty of the local Martian gravity or the drift due to the thermal sensitivity. An additional mass is also used to balance the VBB sensor, designed for Mars gravity, during Earth testing. Except the design of the housing structure, this breadboard model is very close to the flight model. 171

Document

AMS_2006.pdf

SDhere “Sphere” is the name given to the seismic sensor part of the multi-axes Martian seismometer. The sphere, in flight model configuration, consists of the following components: 0 0 Getters 0 Flex-rigid cables and feed-through 0 0 Radiative screen 0 Spherical covers 0 Sealing pipe 2 VBB seismometers in opposite sensing directions Inclinometer, vacuum sensor, thermal sensors Inner plate with insulating blocks Structural ring (diameter of about 140 mm) Vacuum is set in the sphere in order to reduce the viscous damping and the buoyancy (Archimede thrust) on the mobile part. Figure 4 presents an overview of the flight model of the sphere, without the upper cover. VBB2 VBB 1 Insulating block 7 plate J Inner screen ring Figure 4. Flight model of the Sphere The breadboard model of the sphere offers the same functions as the flight model. The breadboard sphere, intended to test on the Earth ground, is composed of the breadboard models of the VBB sensor. Consequently, the breadboard sphere volume and mechanical interfaces are a little bigger. There is no radiative screen on this model. 172

Document

AMS_2006.pdf

Figure 5. Breadboard model of the Sphere Technical Challenges Overcome Struaalina for a low mass The total mass allocated to the Martian seismometer, including sphere, installation device and electronics, is 2.4 kg. The mass budget allocated to the sphere is only 540 g. At the end of this Phase 6, we managed to get very close to the mass budget with 570 g, which almost ensures matching the specification mass in a phase C. It is the result of a trade-off between seismic performance, mechanical resistance and thermal sensitivity. Huge efforts were made to miniaturize all functions and match with the mass specification while matching at the same time with the necessary robustness. Low mass vs. seismic Performance Theoretically, the bigger the mobile part mass is, the better the seismological performance is for an open-loop seismometer (the sensitivity of a seismometer varies with m/k, with m the mass and k the stiffness). For example, the seismometer STS2 which is the current reference on Earth weighs 11 kg. The use of a feed-back loop on the mobile part position is a way to reduce its mass. Concerning the fixed part, some rather heavy but, unfortunately compulsory components - the getter and the shielding for instance - makes the “fight” for a low mass even more challenging. Low mass vs. mechanical resilience As an other big issue with regard to a low mass, the structural part has to be stiff enough to not interfere with the oscillations of the ground and resistant enough to withstand the landing on Mars. The mass of the structural part has been reduced by calculations on a finite element model and especially thanks to the use of stuck glued and welded assemblies (instead of bolts and screws). Low mass vs. thermal sensitivity Finally, concerning the materials, the choice was led by the thermal sensitivity and the need of vacuum. Titanium (TAGV or T40) was selected for all the parts participating in the oscillations measures (ring, plate, fixed part and mobile part) except for the insulating blocks. Titanium is of course an additional difficulty with regard to the low mass objective. For the covers, an alternative to the titanium was explored. With a composite material in carbon covered with a metal deposit, the mass of the covers can be reduced by 40% (35 9). Unfortunately the current technologies of deposit on composite materials do not guarantee a sufficient level of 173

Document

AMS_2006.pdf

tightness for the vacuum desired inside the sphere. Consequently the covers are made of titanium (T40) by hydroshaping with a 0.3-mm thickness only. This thickness represents the very limit of this kind of technology. The mass of the DCS and OCS shielding has been optimized by the use of aluminium with a 0.4-mm thickness. Hiah thermal insulation from Mars environment The thermal environment on Mars is very tough. The minimum temperature can be as low as -120°C in the polar region and the daily amplitude is about 110°C. This environment imposes a very high thermal insulation to allow the operation of the electronics and to reduce the thermal sensitivity of the sensors. A thermal regulation inside the sphere is not possible because of the perturbations it would cause on the seismic signal. The internal insulation of the sphere has been reached by a drastic reduction of the conductive and radiative heat transfer. In fact, there are mainly three thermal paths between the inside and the outside of the sphere. The first thermal path is the conduction through the structure. Although the titanium TA6V has a rather low conductivity, it is not sufficient. The high thermal resistance expected has been raised up thanks to four insulating blocks in TorlonTM (composite material with short fibers). This assembly (Figure 6) gives a huge thermal resistance of about 2000 WW. Such value also requires that the block be fixed by an adhesive joint on the structural ring. Insulating dr Figure 6. Insulating block assembly The second thermal path is the conduction through the internal harness. Indeed, many electrical signals have to be shielded and these shields constitute a real thermal path. The thermal resistance on this path has been increased thanks the use of flex-rigid cables. Also interesting for the integration, this solution allows an optimal section of the wires and especially the thermal insulation of the electronic shielding. The electrical connection of the shielding is made in a very local way with a thin track. Finally, the third main thermal path is the radiative exchange with the covers. This exchange was reduced by the set-up of radiative screens and by the use of a gold coating. The 2 radiative screens (upper and lower) are made from a sheet of titanium of only 50 pm in thickness. Then, this sheet is shaped and welded by point to recreate the spherical shape of the covers. This part cannot be obtained by machining or hydroshaping. Little insulating blocks in TorlonTM are also used for the screen assembly (Figure 7). They are stuck on both sides on the covers and the screen. 174

Document

AMS_2006.pdf

radiative screen block Figure 7. Upper radiative screen assembly Withstandina he shock when landing When the surface module that contains the seismometer, lands on Mars a very violent shock is generated. This shock, equivalent to a half-sine 200 g - 20 ms, is relatively long and thus very energetic. The seismometer resistance to this shock is a real challenge for all the subassemblies but especially for the mobile part because of the absence of a locking mechanism. The power and mass budgets do not allow using any locking devices. Good performance was obtained by stiffening the structure and by placing mechanical final stops. Much work was carried out on the pivot because it is the most sensitive element as far as mechanical requirements. The pivot Figure 8 is based on a flexible blades assembly. The flexible blades are stuck on two rigid parts (fixed and mobile part). Consequently, the pivot supports all the loads of the mobile part and permits only a limited movement. Final stop L- - I - Lowerbody \ Flexible (mobile part) - _- blade Figure 8. Pivot 20 blades In order to have acceptable stresses in blades, mechanical final stop are designed to limit the deformation of the blades at f50 vm. 175

Document

AMS_2006.pdf

Thermal sensitivitv avoidance Because of the great amplitude of temperature and the small stroke of the mobile part (200 pm), the VBB sensor has to be insensitive as possible to temperature variations. Our requirement is a mobile part displacement of 2 pm for a temperature variation of 1°C (measured at the DCS level). To reach this very critical requirement, material homogeneity is essential in order to avoid any shift of the center of gravity of the mobile part. All structural parts of the VBB sensor are titanium TA6V. However, this is not enough and the use of a very specific material for the leaf spring, the THERMELAST, is necessary. The advantage of this material is the possibility to adjust the thermoelasticity coefficient j3 (Eq. (3)) with an appropriate heat treatment. (3) p=-.- 1 AE E AT with: E: Young modulus T temperature For the THERMELAST, /3 values can be adjusted in the * 16.1 0-6 K’ range. For example, when the temperature is cold (lower than the integration temperature), the dilatation makes the distance to the center of gravity smaller and pulls the mobile part upward. If the leaf spring is less rigid when the temperature is cold (/bo), a moment balancing can be found. Obtainina and keepina vacuum To reduce the viscous damping and the buoyancy (Archimede thrust) on the mobile part of the seismometer, vacuum is necessary inside the sphere. The vacuum level at the end of the nominal mission (2 years) must be less than lo-* mbar. This requirement imposes a sealed volume, a very low outgassing for the materials placed inside the sphere, and a passive pumping after the sphere sealing. The tightness is obtained by laser welding for the covers and by brazing for the feed-throughs and sealing pipe assemblies. The sealing pipe (copper tube) is used for the pumping and the closing of the sphere. Structural ring w Figure 9. Section of the sealing pipe assembly A low outgassing requires adapted material (plastics have to be avoided or minimized) and a preparation before sphere sealing. A preliminary outgassing under vacuum is performed for all the sphere components in order to evacuate the solvents molecules. Then an oven drying of the sphere at 120°C during the pumping is done to evacuate the water molecules. The oven drying at 120°C is extremely challenging for the material and adhesive joint choice. 176

Document

AMS_2006.pdf

In spite of all these preparations, the materials keep on outgassing inside the sphere which is not acceptable. Getters are necessary to absorb these gasses and thus ensure vacuum (the getter material works as a sponge with the gas molecules). Unfortunately, the use of this technology leads to some very tough thermal constraints. When activated, the getter’s temperature raises up to 900°C (current of 8 A). As a consequence, the getter has to be strongly thermally insulated. The getter assembly (Figure 10) was designed with insulating braces made out of ceramic, and surrounded with radiative screens. Screens are opened and placed so as to mask the getter from the outside. Getter connection Inner radiative screen Outer radiative screen screw Figure 10. Getter assembly Hiqh level of intearation Beyond the studies, the seismometer integration is also a big challenge. For example, the OCS and DCS assemblies (Figure 11) are made with M1.6 screws. The gaps between the fixed and mobile electrodes of these sensors (6 electrodes) are adjusted at 250 pm with an accuracy of +lo pm. These adjustments are reached thanks to measurements on an optical bench. Fixed DCS Setting shims Figure 11. DCS and OCS assembly 177

Document

AMS_2006.pdf

The wiring is also an example of the high level of integration. First, the pivot is used to make the electric interconnection between the fixed and mobile parts. It requires a precise number of blades for the pivot and also requires that the blades be insulated from the structure. The need of 20 blades for the VBB sensor makes the pivot assembly very hyperstatic and thus leads to a relatively wide distribution of the pivot stiff ness. On the other hand, the wiring is made by using flex-rigid cables. The flex-rigid cables make their way through the structure towards each component (OCS, DCS, coil, mechanism.. .). These flex-rigid cables are composed of 4 layers of 12.5 pm in thickness for the electric signals and their shielding. Figure 12 shows the wiring of the mobile part with the pivot. b- 1 - Flex-rigid cable Pivot wiring LA Figure 12. VBB mobile part wiring Finally the wiring requires specific feed-throughs to take out the electric signals of the sphere under vacuum. There are 2 feed-throughs for the getter power supply and 2 feed-throughs of 42 pins for the VBB sensors wiring (Figure 13). Figure 13. VBB Sensor feed-through Modelling VBB sensor modellinq The mechanical modelling of the VBB sensor is essential to reach the performance required for the pendulum resonance, the mechanical sensitivity, the thermal sensitivity and the shock withstanding. The VBB sensor was modelled on ABAQUS because of non-linear calculations (contacts on the final stop and large displacements for the leaf spring). Great care was taken to achieve the simplified model of the VBB (Figure 14) in order to give the best predictions. The model is based on shells, beams and concentrated masses. Each component was updated on tests or on detailed models. 178

Document

AMS_2006.pdf

Figure 14. Finite element model of the VBB sensor Concerning the performance, Table 1 presents the modelling results. After the stiffening of the pivot (upper and lower body), the shock calculations give positive margins on stresses and efforts in the assemblies. These results validated the final stop setting at 50 vrn. Table 1. Modelling results of the VBB sensor Pendulum 0.35 Hz 0.70 Hz Spring 81 Hz 81 Hz Pivot 90 Hz 96 Hz Fixed Dart 137 Hz sensitivity I Earth config. Mars config. 1.39 1 o-2 s2 2.38 s2 1.99 1 O2 s2 3.38 102 s2 Vertical sensitivity Horizontal sensitivity 0.9 pm/OK B= +16.1 O6 K-' Mechanical desicln The mechanical design of the sphere structure was verified and optimized based on a finite element model of the sphere (Figure 15). The model was developed on I-DEAS. 179

Document

AMS_2006.pdf

Figure 15. Structural model of the Sphere The following calculations were applied: 0 0 0 0 Modal analysis to check the stiffness of the structure Quasi-static load at 2009 to simulate the shock (pessimistic case) Loading under pressure at 1 bar on the covers to simulate the vacuum set up Thermo-elastic to simulate the storage (AT of -1 40°C) and oven drying (AT of +1 OOOC) Several iterations were carried out to obtain the best compromise between mass, rigidity and thermal resistance. For example, much work was done on the insulating block. To withstand the shock in the transversal direction a bigger section was necessary but this decreased dramatically the thermal resistance. A compromise was found by widening the block in the upper part and especially by optimizing the blend radius (Figure 16). m I Initial design Final design Figure 16. Insulating block optimisation Specific tests were made to validate the use of the composite material (TorlonTM) for the insulating block as a structural part. Static tests on the whole insulating block assembly showed good results even after thermal cycles at +120"C and -1 20°C. Finally, the results of the mechanical modelling gave positive margins for all structural parts. Thermal desicln A modelling of the thermal design of the sphere was developed on I-DEAS TMG (Figure 17). 180

Document

AMS_2006.pdf

Figure 17. Thermal model of the Sphere The analysis gave the thermal behavior inside the sphere for hot and cold dynamic cases. Figure 18 shows the efficiency of the insulation: in the cold case with only 70 mW inside the sphere (electronics), the minimum temperature is -93°C for an ambient temperature at -122°C and the thermal amplitude during a Martian day is reduced from 1 10°C to 45°C. However, Figure 19 shows higher thermal amplitude inside the sphere with 58°C. This is a potential problem because of the thermal sensitivity of the VBB sensor. With such thermal amplitude, the mobile part could come on the final stop and not allow measurement for several days. Dynamic cold case +Ambient I Figure 18. Dynamic cold case Dynamic hot case -,m Time (s) Figure 19. Dynamic hot case 181

Document

AMS_2006.pdf

Testing Seismoloaical performance The first Breadboard model of the VBB sensor (Figure 20) was tested in a seismic cellar at St Maur near Paris by the IPGP team. Figure 20. Breadboard model of the VBB sensor With several earthquake detections, for example the Sandwich Islands one on September 6, 2004 (Figure 21), the IPGP characterized the VBB sensor performance. Figure 21. Sandwich Islands earthquake detection The measured performance is presented in Table 2. They show results relatively close to the modeling. Furthermore, the comparison measurements between the VBB sensor and the STS2 seismometer, the current reference on earth, confirmed the high performance of this Martian seismometer. Thermal & Mechanical aualification Shock test on a pivot mock-up A preliminary mechanical qualification test with random vibrations and shocks was performed on a pivot mock-up (Figure 22). The aim of this test was to validate as soon as possible the pivot design and the use of the final stop on the mobile part. 182

Document

AMS_2006.pdf

Table 2. Testing results of the VBB sensor Leaf spring TModelino I Testing -[0.50- I Testing - -. iensitivitv 4.1 pmPK 4.8 pmPK p = +I.IO-~ K-' p = +i.ia6 K' Pendulum 0.35 Hz 3 I 1.39 10"s2 I 1.73 10-2s2 I I sensitivity 1 m Figure 22. "Pivot" mock-up during shock testing The shock test performed at CENCESTA was successful: the pivot performance was conserved and no degradation was noticed. We did learn from the shock test that the torque of the M3 fixation screw of the pivot had to be increased to avoid any slip. Qualification on a StructuraVThermal Model of the SDhere (STM) The STM is representative of the flight model except for the functional aspects. It includes the structural parts and all the critical assemblies like the pivot, the getters, the sealing pipe, the radiative screen and the feed-throughs. The STM cavity is not under vacuum but the covers are welded onto the structural ring. A complete qualification test of the seismometer design was performed on the STM. The STM was opened (cut) after mechanical and thermal environment tests for visual inspection. 183

Document

AMS_2006.pdf

Figure 23. Sphere STM during integration Figure 24. Section of the STM CAD model of the sphere Figure 25. STM on shaker 184

Document

AMS_2006.pdf

During the mechanical qualification, the STM had to undergo a 9g RMS random vibration test and a 200g-20ms shock on all axes (2 shocks per axis). The STM was instrumented to measure the structural eigenmodes frequencies and the dynamic responses. The random vibration test was performed in SODERN’s lab on a 40-kN (9000-lb) shaker. The low sine tests performed before and after the random test did not show any discrepancy on the eigenmode frequencies which gave a first clue that there was no damage, before sphere opening and visual inspection. The main concern was the evolution of the eigenmode amplification factors after the test; from -35% to +50% depending on the eigenmode frequency and measurement location. There is still no reliable explanation of this phenomenon apart maybe from the use of non-metallic materials (glueings and composite material of the insulating blocks or the presence of an unlocked mobile part) and mechanisms. The 6 shock tests were performed at CEA CESTA in France. The Low Sine tests performed after the shock test did not show any frequency drift. After the mechanical environment tests, the sphere was opened for visual inspection (cut by machining). No damage was noticed. r Figure 26. STM Sphere opening Figure 27. STM Sphere ready for thermal balance In addition to the mechanical tests, thermal cyclings in the extreme specified temperature range [-120;+120°C] were successfully performed at the IPGP in the Martian environment simulator and in Sodern facilities. No damage was observed. Thermal balance test A thermal test of the STM Sphere was performed in SODERN’s EV5 vacuum chamber in order to update the mathematical thermal model and the thermal studies of the Sphere. Fifteen (15) thermocouples were used in order to measure the temperature at various locations and correlate with the mathematical model predictions in static and dynamic cases. Heaters were used to simulate the power dissipated by the real electronics (70 mW). The test showed that the thermal insulation of the sphere was lower than predicted. The representativity of the mathematical model w.r.t the test conditions was analyzed and confirmed. In particular, an additional test with only two thermocouples was performed and confirmed that the wires of the 15 thermocouples were not doing an undesired thermal shunt. A sensitivity analysis was also performed with the thermal mathematical model (I-DEAS) showing that a refinement of the model assumptions (more accurate material data, more representative geometry, radiative exchanges modeling improvements) and few corrections were able to suppress most of the discrepancies 185

Document

AMS_2006.pdf

between test and mathematical result. This correlated thermal model will be of course very useful for any further analysis. An improvement of the thermal insulation of the Sphere has still to be worked out in Phase C in order to reduce the sensitivity of the seismometer to temperature variations between night and day. It is however not really an issue and the use of MLI around the sphere should help a lot. Getter activation tests Functional tests were carried out on the getters. Firstly, preliminary tests were performed on a getter mock-up to determine the best getter activation parameters (power supply, activation duration). Secondly, a fully representative test was carried out with the STM. Both tests were performed under vacuum of course. Getters Figure 28. Getter Breadboard Figure 29. Getters in Sphere STM The getter Breadboard was made of the getter itself and the parts of the seismometer located close to the getter in order to check they were not damaged during the getter heating. This test helped us to determine the maximum acceptable temperature of the getter in order to prevent any damage of the surrounding parts, actually 600°C. The activation duration needed to obtain the expected vacuum level will be defined precisely in Phase C with a fully representative model of the Sphere. Indeed it is very dependant of the material amount and outgassing properties. A second getter activation test was performed on the STM. The goal temperature of 600°C was not reached on the 2 getters and the insulating blocks were locally damaged because of the temperature of getter power supply wires. The cause of this problem is clearly understood and it can easily be fixed. It is due to the electrical resistance of the feed-through. A larger feed-through solves the problem. This has been successfully confirmed by complementary tests. Conclusion At the end of this phase B, the most critical points of the design of the very core of the seismometer, the detection sphere, have been successfully verified. The few last points to be improved or adjusted are clearly identified. The expected high performance was confirmed on the Breadboard and the global consistency of the design with regard to the requirements was also confirmed. SODERN is now ready to start a C/D phase for any future seismological planetary mission and will be a very serious candidate on the ESA EXOMARS 201 1 mission. 186

Document

AMS_2006.pdf

Commercial off the Shelf Components in Reaction Wheels Andrew Haslehurst' and Guy Richardson* Abstract This paper presents the approach adopted by Surrey Satellite Technology Ltd (SSTL) for development of reaction wheels incorporating commercial components and the accelerated life test / qualification philosophy. The paper focuses on the mechanical development and re-qualification of reaction wheel which has flown successfully on many missions having conducted years of in orbit operation. The scope of the requalification was to verify the wheels performance with ground testing for new missions where an increased life was required, with some re-design where applicable. Introduction The Microsat Reaction Wheel (MRW) or Micro Wheel was developed at SSTL in the mid-nineties and is unique in its design from the simple drive electronics which are enclosed in the base of the wheel to the commercial brushless DC motor using dry lubrication, to the wheel volume at approximately 100*100*100 mm3 all weighing in at under 1 kg, this wheel was succeeded by the Enhanced Microsat Reaction Wheel (EMRW) in the year 2000. The aim of the new Superior Microsat Reaction Wheel (SMRW) was to build on the heritage and success of the MRW and EMRW increasing the performance and extending the life in terms of revolutions of the wheel. Figure 1 shows the MRW, EMRW and SMRW wheels. At the time of writing this paper the SMRW is integrated onto the protoflight model of 5 imaging micro satellites currently being built at SSTL and due for launch towards the end of 2006. The Microsat wheels have led the way forward for the development of the next generation of larger more powerful wheels which use similar electronics and lubrication systems. Although this wheel is not discussed here in the paper, four smallsat reaction wheels also incorporating dry lubricated bearings have been integrated onto the GIOVE A spacecraft ready for launch at the end of December 2005. Figure 1. SSTL Microsat reaction wheels Background The small satellite market (400 kg) is particularly demanding not only for the small budgets involved or the mass and power requirements but also for the fact that a typical program for SSTL from contract to in orbit operation is normally from 18 - 24 months which leaves a very short period of time for development and qualification. ' Surrey Satellite Technology Ltd., Surrey Space Centre, Guildford, Surrey, U.K. Proceedings of the 3d' Aerospace Mechanisms Symposium, Langley Research Center, May 17- 19,2006 187

Document

AMS_2006.pdf

One approach that SSTL uses builds on its philosophy of "affordable access to space" by flying new developments as back-up or experimental units on missions. Hence, flight qualification can be achieved without the need for lengthy and costly ground qualification. The market SSTL is addressing is changing though and there are now fewer opportunities to qualify in this manner with customers demanding proven reliability and heritage. To achieve the rapid development and qualification with wheels a different approach needs to be taken from the typical wet (oil or grease) bearing lubricated systems. This simplifies the design, not requiring reservoirs for the oil and anti-creep barriers but allows for an accelerated life test. Due to the properties of wet bearing lubricated systems preventing acceleration of life tests with either speed or bearing preload, wet lubricated systems have to perform real time life tests (at flight speed). This leads to very long, potentially costly wheel qualification programs, which can lead to launch occurring before wheel qualification is complete. All of SSTL early spacecraft were microsats approximately 50-kg spacecraft 360-mm square by 750-mm long in launch configuration. The AOCS system of these microsats did not rely on reaction wheels to provide attitude stability. The spacecraft design used a gravity gradient boom to keep it nadir pointing. Yaw motion was controlled using magnetorquers. Augmenting the AOCS with wheels allowed much better control of yaw and the potential to off point from nadir to increase the imaging opportunities of a particular target. Early SSTL wheels were flown as experiments providing enhanced performance above the baseline mission requirements. This allowed development cost to be kept down by reducing ground testing to a minimum. All SSTL wheels to date and the current EMRW, SMRW and SRW use self-lubrication in the form of a PGM-HT cage. Fortunately for such self-lubricating bearings, there is an increasing body of life test data, which shows there to be a quite strong correlation between wear rate (i.e. bearing lifetime) and peak Hertzian ball- raceway contact stress. Much of this data was recently generated by ESTL (European Space Tribology Laboratory) as part of an ESA-funded campaign to source, fully characterize and qualify a replacement material for the Duroid 5813 material at the time it ceased production. Taken as a whole, this data includes both in-vacuum long lifetime application data with PGM-HT and Duroid 581 3 materials and an in-air test campaign of approximately 3000 bearings aimed at generating design guidelines for industrial applications of this type of bearing. Further information can be found in [I], [2]. The graph presented in Figure 2 shows the relationship between preload and life which comes from experimental data from tests and predictions from design analysis, Test terninatiozls . NOT end of life Figure 2. Preload V's life graph Use of solid lubricants in reaction and momentum wheel bearings is clearly quite rare. One reason for this is the higher torque noise of some solid lubricated bearing systems, their perceived short lifetime and intolerance to misalignment, competing against the long-established heritage of liquid lubricated solutions

Document

AMS_2006.pdf

which offer high lifetime (though often this is achievable only with the addition of a wheel re-lubrication system). The disadvantage of the conventional approach for an accelerated wheel development program is that to qualify a liquid-lubricated wheel, no strict tribologically valid method exists which can accelerate the life test without modification to the lubricant regime in a manner which risks significant under-test. For a solid-lubricated wheel however testing can be relatively straightforwardly carried out at high speed without significant impact on the lubricant wear behavior. SSTL's approach with such bearings can be thought of using the cage as providing a lubricant reservoir with solid lubricant being transferred from cage to balls and ultimately onto the raceways as a so-called "transfer film". Bearing lifetime is limited by ultimate wear-out of the cage pockets or by excessive or lumpy transfer, which may in some applications create unacceptable torque noise. For SSTL's first generation of reaction / momentum wheels, which were successfully flown, the lifetime requirement was not particularly high and the self-lubricating bearings were adopted and used in a relatively commercial manner. However, for this latest generation of wheels, there was a requirement to demonstrate a much higher lifetime nominally 2.2 billion revs for a 7-year DMC type mission (including a factor of 2) with low torque noise, combined with a programmatic need to demonstrate the life within an approximate 12 month testing window. With this change in requirements for lifetime there is an on going investigation into extending life by using novel hybrid lubricant combinations, which are not presented in this paper. MRW Backaround / Development The first experimental MRW flew in 1995 on FASat-Alpha which unfortunately failed to separate from the launch vehicle upper stage. FASat-Bravo was built to replace FASat-Alpha and was successfully placed into orbit in 1998. The experimental reaction wheel worked well and encouraged development and use of similar reactions wheel on future missions. Between 1997 and 1999 three reaction/moment wheels were developed: a nanosat momentum wheel, a minisat reaction wheel and a modified microsat reaction wheel. All of these wheels were based on commercial motors and with the exception of the nanosat momentum wheel, which used vacuum grease lubrication, all used dry lubricated bearings. The most successful of these wheels was the modified microsat reaction wheel, which flew on Thai Phutt (TMSat), Tiungsat-1. TMSat and Tsingsat-1 each had a single wheel that was used for experimental operations only. Tsinghua-1 had 3 reaction wheels and was operated in three-axis reaction wheel control for months before its gravity gradient boom was deployed, during this time one wheel biased at 500 rpm conducted over - 6.59~108 revolutions. The modified microsat reaction wheel used for Thai Phutt, Tiungsat-1, Tsinghua-1 was the basis of the EMRW. The EMRW is a primary attitude control component on six SSTL spacecraft currently in orbit. The first of the spacecraft to use the EMRW was AISat-1 this was the first spacecraft of the DMC constellation, launched in 2002. This spacecraft used two wheels in combination with a gravity gradient boom. Since the lifetime requirements for the AISat-1 wheel were largely met by the on orbit operation of the Tsinghua- 1 wheels and the design of the wheels were similar, extensive ground testing and requalification was not carried out. UK-DMC, NiSat-1 and BilSat-1 were the next spacecraft to use the EMRW. These three were launched together in 2003 and also form part of the DMC constellation. The core platform design of UK-DMC and NiSat-1 was the same as AISat-1. However, BilSat-1 was first of a new generation of SSTL spacecraft designed to be more agile and have a longer mission life (seven years instead of five). It was not conclusive from the available data at that time that the EMRW design would have sufficient life to meet the mission requirements of BilSat-1. Hence, as a contingency, Bilsat-1 also included a deployable boom to allow gravity gradient ADCS operation in the event of a wheel failure. 189

Document

AMS_2006.pdf

At this time ESTL performed a life test on a MRW motor miss aligning and loading the bearings to their maximum capacity in an attempt to purposefully cause an early failure. The wheel conducted 0.6-billion revolutions (proving mission life taking into account an acceleration factor) showing no signs of failure at which point the test was stopped and the bearings inspected. Some pitting was observed on some balls and these were examined under a SEM, which can be seen in Figure 3, the damaged areas showing delaminating are typical with high sub surface stress and although not desired the wheel was operating nominally. The cages were weighed pre / post test and seen to have negligible cage weight loss. Assuming cage wear limits the life then it is predicted that lifetime in the order of tens of billions revs could be achieved. Although these tests fundamentally proved the motor was fit to meet current mission life requirements it was thought the life could be extended even further. Hence, it was determined that bearing misalignment should be taken out by some small design changes, which are presented in the development the SMRW. Figure 3. MRW (SR4) Bearing inspection after life test Since their launch the EMRW wheels in these spacecraft have performed millions of revolutions without anomaly. On orbit data can be seen from two wheels of the UK-DMC spacecraft, in this case of operation mode operating around zero. Table 1 summarizes the wheel parameters for the MRW, EMRW and SMRW. t . . , . ... a.. I w Lt 3mi8 tm wll 11- Figure 4. Snap shot of in orbit data from a UK-DMC spacecraft showing speed (RPM) against time 190

Document

AMS_2006.pdf