4_ AIAA 2001-3631 Compton Gamma Ray Observatory: Lessons Learned in Propulsion G. Dressier, G. Joseph and H. Behrens TRW, Redondo Beach, CA D. Asato NASA GSFC, Greenbelt, MD R. Carlson Rancho Palos Verdes, CA AIAA/ASME/SAE/ASEE Joint Propulsion 37 th Conference and Exhibit 8-11 July 2001 Salt Lake City, Utah I I For permission to copy or to republish, contact the o_l)yright owner named on the first page. For AIAA-held copyright, write to AIAA Permi._ions Department, ISOL Alexander Bell Drive, Suite 500, Reston, %x,, 20191-434-1. \1\_, 2_nl.3h3! Compton Gamma Ray t)bservatory: Lessons Learned in Propulsion G.A. Dressier*, G.W. Joseph*, H.W. b "hrens* r"I'R_V, Redondo Beach, CAb D.I. Asato* (,\.ISA GSFC, Greenbelt, MD) and i¢ _. Carlson (Rancho Palos Verdes. C.t; ABSTRACT The Compton Gamma Ray Observatory was the second of NASAs Great Observatories. At [7_,Z tons. it was the heaviest astrophysical payload ever flown at the time of its launch on April 5, 1991 aboard the Space Shuttle. During initial, on-orbit priming of the spacecraft's monopropellant hydrazine propulsion system, a severe v,aterhammer transient was experienced. At that time. anomalous telemetry readings were received from on-board propulsion system instrumentation. This led to ground analyses and laboratory investigations as to the root cause of the waterhammer, potential damage to system integrity and functionality, and risks for switching from the primary (A-side) propulsion system to the Figure 1. CGRO Deployment from STS-37 redundant (B-side) system. The switchover to B-side was After nine years of exciting scientific disco,,eries _: of ultimately performed successfully and the spacecraft very energetic celestial phenomena (far exceeding its completed its basic and extended missions in this minimum mission life of 27 months). CGRO '*as safely configuration. Nine years later, following a critical de-orbited _ith controlled re-entry into the Earth's control gyroscope failure, Compton was safely deorbited atmosphere on June 4. 2000. Pieces of the spacecraft and re-entered the Earth's atmosphere on June 4. 2000. survived the re-entry, landing in a remote part of the Additional risk assessments concerning viability of A- Pacific Ocean near the equator, approximately' 3.862 km and B-sides were necessary, to provide confidence in (2.-100 miles} southeast of Hawaii. attitude and delta-V authority and reliability to manage The CGRO mission was a NASA cooperative the precisely controlled reentry. This paper summarizes program managed by the NASA Goddard Space Flight the design and operation of the propulsion system used on Center IGSFC} and included co-investigators from the the spacecraft and provides "'lessons learned" from the United States, Federal Republic of German_. Netherlands. system engineering, investigations into the propellant ESA and United Kingdom. The Observatory carried four loading procedures, the initial priming anomaly, mission highly sophisticated instruments capable of making operations, and the commanded re-entry following the simultaneous measurements over six decades of energy g',ro failure. 120 keV-30 GeV? These instruments ,*ere: the Burst and Transient Source Experiment (BATSE t. the Oriented INTRODUCTION Scintillation Spectrometer Experiment (OSSE_. the The Compton Gamma Ray Observato_ (CGRO} was Imaging Compton Telescope (COMPTELh and the a large scientific spacecraft designed for celestial Energetic Gamma Ray Experiment Telescope ,EGRET} observations from low Earth orbit. It was the second The CGRO spacecraft _,,as designed and de_eloped element launched in NASA's deployment of four "'Great b_ TRW in Redondo Beach. CA. Table i presents a Observatories" (HST. CGRO. Chandra-AXAF. SIRTF) summary of the spacecraft subs._ stems. The and carried instruments dedicated to the highest part of monopropetlant hxdrazine propulsam subs'.stem the electromagnetic spectrum. The objective of the CGRO consisted of "'A-side'" and full,:' redundant "'B-s_de" sets of mission was to obtain gamma-ray measurements over the thrusters, feed system c_)mp,ments and pr_)pe]]ant tanks. entire celestial sphere v,ith unprecedented sensitivity. Although norma ly inacti',e during science gathering, the spectral range and resolution. propulsion system ,._as :o be used e;er,, _--, '.ears to CGRO was launched aboard the Space Shuttle reb, n_st CGRO to ,_ttset decay in orbital altitude due to Atlantis (STS-37} on April 5. 1991, and was deployed atmo.,pherw drag April 7 into a 45_ km circular orbit at 28.5 degrees The ',pacccr:tt't'_ propui>l_m sub,;2,stem .had :'._, major inclination IFigure 1h At the time of its deplo.,,ment it set on-,,rbtt an_wnalie,, durine :he mls_,i_m [he ;-_r.: .m,,malv t_o rect_rds for non-m_l_tary spacecraft: tt ,,,,as the largest ,c,:'.,rred durt ny' che.:k_ut -nd actt', atu m. ,t :he .{xwecr:fft spacecraft launched by STS and it had the largest arter being released tn,m 'he Shuttle [mmed_,::c'.', up,,n m_,nt>prapellant propuls_t>n s,,stem ever tlo'an. _pcntng,mc,_tthe pn_pe!lant:ank >_lati,,n,.._['...:.. r_:hc " Member of A[A.\ ('_p_,rIght ')2001 bv TRW [nc Published b', Amcrwan Institute ,t \c>'f_at_t:,- _,,_rz_i \,rr,,r:a::_:_,. [*;c. '._th _c::_:..,, _ \1 _\ 2(1(|1-3631 IJ,l_c 2 ,,t [" l'ableI. CGR() Spacecraft Subsystem,,; .\-',I,..tC pr_pclla[lt i11,tilll,_ld. {c[cil:ctr'. Lnd_c,tlcd that t,.:o ,q the iN_}tatlt',n _,al;es had un_.,,mm,tr'dcd changer, in the Science Instruments • Four main m_,trun'Lcnts _B\ I'SE. OSSE. COMPTEL. opcn/ch;:,e pt>.ltlt_tl s[,.ltU>. ,llld t_fle _1 the pressure transducers indicated "'o:er-Iimit'" pressure. Since the EGRET_comprlsmg approxmlatel: 6.300 kg 17tons_ • Conunuous dctccmm o_er 20 Ke: to 30 Gev range science act:vlties were not affected, re-activation of the pr_>pulsion system was postponed until April 1993 at Structure ebolted aluminum box-girder framework) ,._,hich time the Observatory orbit had decayed to 350 km. • Masslloadcdl: 15.S76kg_35.0OOlbm) This tv,o-year interval allowed a comprehensive • Body Size: 4.6 m _ 55m xq.I m (21.3 roSA span) analysis of the anomaly and detailed development of corrective actions. Ground tests and analyses concluded Power Subsystem that. in spite of being "'fully" loaded with hydrazine, the • Solar Array Po_*,er: two accordion-style, deployable A-side propellant manifold had been exposed to very high arrays generating 4300 w BOLJ3980 w EOL with surge pressures (a "waterhammer'" transient). To prevent 396 if: area reoccurrence, a method was successfully developed to • Battery. Power: six Ni-Cd batteries at 50 A-ha"each safely prime the B-side manitbld by opening the isolation Thermal Subsystem valves for very short durations to slowly raise the • Uses coatings, blankets, louvers, radiators and heaters hydrazine pressure in the downstream manifold to design • Science instruments thermally isolated from spacecraft operating pressure. The B-side Attitude Control Thruster and each other (ACT) manifold was successfully primed April-July 1993 • Redundant thermostats and heater elements in preparation for restoring the orbital altitude of CGRO to 450 km. Communications & Data Handling Subsystem The second propulsion subsystem on-orbit anomaly • Standard NASA modular design based on Solar Max occurred during the calibration burn segment of the orbit and Landsat 4 & 5 spacecraft raising operation. The plan was to raise orbit using only • S-band telecom using 1.52 m (60 inch) HGA the four B-side ACTs. During the test burns, one of the • Two omnidirectional LGAs ACTs (designated "'ACT-B2"') produced unacceptably • Two second generation TDRSS transponders low thrust. Ground tests determined that the low thrust • Uplink at .125 or 1.0 Kbps: downlink at 32 Kbps (256- was most likely related to flexing of the thruster valve 512 Kbps viaTDRSS) • Two NASA standard tape recorders for playback at up seal at high propellant mass flow rate. to 512 Kbps via HGA and TDRSS To compensate for the low thrust, two of the four • Advanced clock for time accuracy to .0001 second Orbit Adjust Thrusters (OATs) would be needed to maintain attitude control during the orbit raising burns. Attitude Control & Determination Subsystem The B-side OAT manifold ,,,,as successfully pressurized • 3-axis stabilized, zero momentum biased control system using procedures previously employed to prime the B- using reaction wheels with magnetic unloading side ACT manifold. • attitude sensors The first orbit reboost of the CGRO was completed in o Fixed head star trackers (3) December 1993 and restored the orbit to 450 km circular o Inertial reference gyros (4) from a low point of 350 kin. As predicted, the o Coarse & fine sun sensors performance of ACT-B2 became nominal as the operating • Attitude control flow rate decreased, a natural result of the propellant tank o Reaction wheel assemblies (4) o Monopropellant rocket thrusters (8) pressure decay in blowdown mode. A second orbit • Single target pointing control for up to 14days reboost was performed in March-June 1997 with nominal • Pointing control to __.0.5°: measurement to -,-0.03° performance on all B-side ACTs and OATs. The propulsion subsystem performed fla_lessly Propulsion Subsystem during four critical, controlled re-entr', burns that ended * H>drazine propellant: 192--1.kg 142--t0 Ibm). High Punty the CGRO mission on June a. 2000 • GN2pressurant: 172kg(381bm) • Four 4-40 N i 100 Ibf IOrbit Adjust Thrusters OVERVIEW OF PROPULSION SUBSYSTEM • Eight 22 Ni5 lbt') Attitude Control Thrusters Figure 2 shov, s the propulsion subsystem complete • "'BIo_:down" operating pressure: 2760 kPa t400 psia) on a buildup fixture prior to transfer to the spacecraft BOL to 600 kPa (87 psia) EOL structure. Figure 3 is a schematic of this subsystem. • Four 'itanium propellant tanks, each with AFE-332 CGRO :,,as the first scientific ,pacecrat't designed for on- dtaphragm orbit refueling of propellant. The ,.m-,_rbit refueling • On-orbit refueling module m_dule contained a NASA-suppt_ed pr_,pellant coupling. • Safe'.'. compliant with NI-IB 1700 7A Pa_c"_,d t2 p i .... -.... - .... & - _T.-_7.__ & I__a..l..$+_ 1I.+¢_a I1_ II + ,+,,., I t__1,,o+1 I ,I + +I _ + .... '_'' + + ,' A _ + ' ,,,--=4_,_ ,, /ill"" _f ___/ "'. ............ := .... Figure 2. CGRO Flight Propulsion Subsystem Figure 3. CGRO Propulsion Subsystem Schematic The propulsion subsystem consisted of A-side and torques about any of the three spacecraft axes. They fully redundant B-side sets of thrusters, feed system provided primary yaw control and secondary pitch and components and propellant tanks. As will be seen, it is roll control during OAT firings. Thrust levels and significant that this design featured functional redundancy moment arms about the spacecraft center-of-mass were _ith fault isolation capability. Crossover isolation valves such that the ACTs could provide complete three-axis permitted full utilization of propellant and provided control at all times. They were to operate primarily in a capability for center-of-mass management by control of pulse mode, but were designed for steady-state the quantity of propellant used from each tank. Details of operation--a design feature that was used when they were the design and development of the CGRO propulsion required to backup the OATs for altitude raising. subsystem ,sere given in a previous paper 3. Of particular significance to following discussions, The mission tasks for the propulsion subsystem were the subsystem employed multiple latching isolation valves to provide: (indicated in Figure 3) to direct and lock off propellant • Orbital altitude restoration (drag make up) flow. The design of the isolation valve is shown in Figure • Attitude control during reboost 4. It is a pressure-balanced, dual coil, solenoid-operated • Descent for refueling and on-orbit servicing latching shutoff valve with downstream backpressure relief capability. The design features a closure spring to • Ascent (from STS servicing orbits) hold the valve closed and a permanent magnet latch to • Descent for STS retrteval or controlled reentry hold the valve open without continuous power drain. The • Provide sate hold operating mode in event of loss inner solenoid coil is powered to open the valve and the of gyroscope stabilization. outer coil is powered to close the valve. The critical The system was designed to operate in a pressure blo_do_,n mode over a range of 400 to 87 psia. The four 4_t() N OATs were to be fired simultaneously to provide AV impulse tbr orbit altitude change, orbit maintenance, descent for refueling, ascent and controlled reentry. The OATs '_ere placed on the spacecraft X and Y axes with thrust vectors parallel to the Z-axis (see Figure 2). The OATs _sere to be off-modulated to provide primary attitude control about the spacecraft pitch and roll axes during the AV firings. The 22 N ACTs were to provide primar? yax_ attitude control during operation of the OATs. In the e_ent that one of the OATs failed during tiring, tt.,,geomemc ,_pp,_.,_te _outd be automatically shut d_),.sn and Hnpulse '._->uid c_,nttnue to be prostded by the remamln._, O-\g parr Fhe \CF_ ,,_ere c.m_cd ,tf the spacecraft Z-axis and. .,_ncn fired appr(_prx_tc[) _r: pa_rs, could prt,;td¢ c_)ntrt)l Figure 4. CGRO Latching Isolation Valve Design %1%.k21)ill-3631 ,,. !ace _I the p_ppct Is ,,phcr,call) -,hapcd tt_c!t_e _n a dl_llMMIn_ ,_f .l qrtaI1:4nk prc:-,nurl,:ccl t_ ";l) p,,i7 _04p',lA). T'hi_tank',, -,h!raL_'ctlrrl¢ AIM h_.lndllilL_'',,,ere ,,utficlcnt tk_r :c_hm Nng .,eal rhc _dcno,d and mm (ch_,,urei ,prIng arc hermcticall,, sealed tmm propellant b,, _'elded the prc_pellclnt to bcct_mc full', saturated ,.'ilk nltr(/_en ,,as hclh-.'s assemblies. The spring assemhl? ctmtrolling the at 6J psla. Nitrt)gcn _:1:,,bubbles _hich e;entually fi_rmed ba_:kpressure relief function is exposed to the fluid. A in the propellant teed lines came from tv,o sources: tOO-micron absolute wire mesh tilter is installed in each residual nitrogen in the s,,stem pra_r to tilling (imperfect of the fluid ports. A position switch assembly, v,hich ,,acuuml, and nitrogen that came out of solution from the electrically indicates valve position, is located at the top hydrazine due to depressurizing the lines to 25 psia until and is integral with the valve. Other than being damaged launch and deployment. by above-specification overpressures during the A-side The system vacuum betbre loading hydrazine was priming attempt, the latching isolation valves performed estimated at 10 torr (gauge read "'30 inches ""mercury). nominally during the nine years of on-orbit life. This yields a ',olume ratio of 1.3% ,,,,hen re-pressurized to Moreover, it was this valve's beyond-specification 14.7 psia. The volume of nitrogen coming out of solution capability to respond to millisecond-level pulse from the hydrazine is less clearly defined. First, available commands that ultimately provided a means to safely references tbr nitrogen solubility in hydrazine give prime the subsystem B-side. thereby enabling the CGRO different values. It is also unclear how much can be re- mission to continue with low risk. dissolved. The resultant bubble volume ratio ranges form 0.72 to 1.88%. Overall, the combined bubble volume PROPELLANT LOADING SEQUENCE ratio is somewhere between 2.0 and 3.2%. This yields a AND ISSUES bubble length of 5.9 to 9.3 cm in the lines between the The propulsion system was designed for a STS tank isolation valve (A1) and the thruster isolation valves launch, incorporating two-fault tolerance for propellant (A3 & A4), based on a 290 cm total line length. It is also leakage and planning to launch 'wet" to the thruster difficult to assess where the bubbles may have collected valves but unpressurized (at atmospheric pressure) during the filling process, further complicating the do,.'nstream of the tank isolation valves A1. A2, BI and definition of bubble size and location. Analyses and tests B2. described below evaluated the effects of bubble size and In conducting normal leakage tests using nitrogen location to span the probable distributions. gas. the leakage rates of isolation valve A2 and crossover In January 1991, the B-side manifold, the tank isolation valve BIA were determined to exceed crossover manifold (on-orbit refueling coupler, isolation specification. The valve leakage problems occurred late valves AIB, A2B, BIA, B2A) and the four propellant tanks were loaded with "High Purity Grade" hydrazine in the program, resulting in a Noncompliance Report. A condition for approving CGRO for launch was using a unique propellant loading system developed by "'demonstration of no continuous liquid leakage" in the TRW. The propellant loading system incorporated an air- launch configuration. This demonstration was driven positive displacement pump to transfer propellant, accomplished by a partial fueling operation in June 1990 an electronic mass flowmeter to monitor the rate of at which time 15 Ibm of hydrazine was loaded into the A- propellant being loaded, and a reservoir propellant tank to side of the system. The system was monitored for liquid remove entrained bubbles before being supplied to the leakage: none was observed and therefore launch was propellant pump. CGRO was loaded with a total of 1924 kg (4240 Ibm) approved. The partial fueling contributed to the surge problem of hydrazine distributed evenly among the four propellant because a different propellant loading system was used tanks. After closing all of the isolation valves in the for this small mass of propellant. propulsion system, the propellant tanks were then The loading operation was as follows. Tank AI fuel pressurized with nitrogen to a flight pressure of 390 psia. and gas fill and drain valves and isolation valves AI. A4. and A3 v,ere opened. The tank was vented and evacuated INITIAL SYSTEM PRIMING EVENT (A-SIDE) to 26 inches of mercury (Hg). The gas fill and drain valve At the completion of the isolation valve leak integrity was closed and the fuel side evacuated to "'30 inches of demonstration and final propellant loading and flight mercury'" (gauge resolution). 15.7 Ibm of hydrazine was pressurization, the propulsion system was left in the loaded and the fuel fill and drain valve (FDV) closed. following configuration for launch: The gas FDV was then opened and pressurized to 15.1 • Hydrazine loaded to the thruster val_es psia. Isolation ,,aires AI. A3 and A4 were then closed • Manifolds bet_'een A-side thrusters and isolation and the tanks pressurized to 430 psia. The system was _atves .-%l, A2, A3 and A4 pressurized to 14.7-15. I monitored for leakage tier 5 days before depressurizing psia _he.&-side tanks to 25 psia. • Cmsst_er manifold ;on-orbit refueling coupler. AIB, The importance of the abo_,e operation is that the A2B, B[.\, B2:\} pressurized at [4.7-15.I psla 'partial fueling" operation ut,lized a hydrazine supply .klAA 2tl4)1-3631 [:':>-:c g ,q [2 • l>:_++'ullatanntk.andmamf_qddt_,,_i,tn>ist+latum ht_,torlc.fl precedents qmckly c_mcluded that the f_:.llc,,._,lng ,.+it,.e-,\.t,..\2..\l B.andA2Bprc:+,:+unzud t,t+ 390 e'.ent> had .,ccurrcd during the priming ',cquence: ?MA • Tank t>t>lation val',e A Lopened as commanded • B ,,lde prc_pellant tanks and mantt\dds loaded and • When valve AI opened, rapid propellant flo,a, pr::,,_.urtzed in similar manner as A-side occurred resulting in a pressure surge or • .-\il A-side and B-side isolation valves, including all "';,,aterhammer'" crossover isolation valves, v,ere closed from launch • The pressure surge overstressed the Tank A I pressure through being released from the Shuttle arm. transducer sensing element, causing a "'zero shift" On April 7. after being released and establishing the that resulted in an off-scale high reading required "'safe distance" from the Shuttle. activation of • The pressure transient, flow surge, and/or resulting CGRO propulsion was started. Following nominal pre- dynamic excitation of the fluid system caused the established planning, the propulsion subsystem activation crossover isolation valve (A1B) to change from a procedure consisted of a series of commands to open A- "'closed" to "open" state, and caused the tank side thruster isolation valves in the following sequence: isolation valve (AI) to change from an "'open" to A3. A4. A1. A2. The commands and telemetry responses "'closed" state. were: Although there were no indications of propellant COMMAND TELEMETRY leakage through any of the A-side thruster valves, there OPEN ISO A3 A3 OPEN were significant concerns as to the ability of the A-side OPEN ISO A4 A4 OPEN isolation and thruster valves to function properly and OPEN ISO AI • AI CLOSED repeatedly following exposure to high transient pressures. Consequently, an extensive program of analytical • A1B OPEN modeling and ground testing was undertaken with the • Tank A1Pressure Transducer objectives of: Reading Full Scale (510 psia) (a) determining whether the CGRO mission should be continued on the A-side or B-side, Review of telemetry data confirmed that the anomaly had not affected the health of the rest of the spacecraft. A (b) determining least-risk method restoring propulsion decision was made to secure the propulsion subsystem subsystem function, and and to postpone activation of the subsystem until 1993, (c) establishing the fundamental ("root") cause of the when orbit raising of CGRO was anticipated to be priming pressure transient. necessary. Suspecting that the propulsion subsystem manifold SYSTEM ANALYSIS AND MODELING may have been subjected to unexpectedly high surge A coordinated analysis effort was undertaken at TRW pressures, isolation valves AI, A3, A4. A1B were to model and understand the dynamic behavior of the commanded closed to minimize the chance of propellant GRO propellant supply system. In conjunction with the leakage from potentially damaged lines and components. analytical effort, experiments were conducted at TRW to All valves except crossover ALB were verified to be validate the analysis model. The test setup mimicked the closed following commanding. The fact that AIB did not flight configuration for critical parameters such as line show a response to a valid command indicated either that diameter, line lengths, number and location of sharp the valve was mechanically damaged and could not bends, and location of valves. Developmental isolation function, or that the valve's position indicator had been valves of the same design as used on flight were damaged and was no longer reliably reporting the valve incorporated in the setup to insure proper transient D)sition state. response to valve opening and closing. Test series were A team was formed at the NASA GSFC and TRW- conducted using both water and hydrazine. The analysis Redondo Beach to investigate the anomaly and to assess model was exercised to predict the pressure spikes and options for safely activating the CGRO propulsion natural frequency response for each test/bubble system +Sa. The results of the investigation were configuration. A key objective of this work was to use the distributed to the NASA centers and to the aerospace empirically-calibrated analysis model of the flight system industry. This included dissemination of the anomaly and to support decisions on pending activation of CGRO mo_,t probable cause via the GIDEP alert system'. propulsion. ANOMALY IMMEDIATE CAUSE Anal,,sis Model. [t is customary to use a lumped DETERMINATION parameter approach to study the transient behavior of liquid flo_,, m a complicated propellant teed system. The [mttal re',te,._, of the propulsion s_stem and popularit.,, ,)f such an approach is due it) its simplicity and c_,mponent design data. spacecraft telemetr,, data. and numerical efficiency [n the lumped parameter approach. _.1_._ 21)0|-343 I Pa,-'c_),_I 12 ,j'.tIt.tl '..|It_tl_n t'-. I_rlt_rdd, rcsuJtltlg it1 .1 -,c[ !'_[ _rdirlarv trt'Ie lt_c;.llltH1 (_. l,cl t;Igttles "q .tlld ()1 ll1 (he _,",lclrl alld 7 rather than pumal diffcrunlbal cqu,itnms, lhe_,e ODE's at an_thcr h_call_,n ,K,,. per Figure 7). .nn i,c ruad_l,. -c,l'.cd by <tam.lard mtegrathm .21gortthms. The same t>pes ,,t tests ,,,,ere done _,tth h,,drazme. ,uch ,> i_.Lmge-Kutta method The peak pressures _,,ere. as expected, slightly higher due Various ,>sumpttons mere made to arrive at the final to the difference m bulk modulus as compared to water .mal,.s_:, model. First. the flow v,as assumed to be (see Figures 8 and 0>. incompressible and isothermal As a result, the energy In the last test using h,,drazine, performed with a AP equation was not considered in the system This of 375 psid. the A3 valve was closed to prevent too high a assumption ',,,as definitetv valid because the Mach pressure peak in this part of the system. Note that at number of the flow (either single phase or two-phase) is station K4 the ratio of peak pressure to initial AP was 5 for t?pically about .O0t. Another key assumption is that the both analysis and test. as it was for other tests having propellant flow was fully turbulent. Again. this lower initial :XP (50 and [00 psid) across the isolation assumption was valid because the typical Reynolds valve. number during the priming event was about 6x[0 +, well exceeding the transition Reynolds number of 2000. GROUND TESTS The lumped parameter approach appears to be sound. The CGRO propulsion subsystem on-orbit anomaly In a system that consists of only a pipe with high pressure investigation included a comprehensive review of at the entrance, it can be shown that closed-form solutions spacecraft-level ground tests, including electrical circuit for the natural frequency and pressure history exist. testing. No credible mechanism was identified that Furthermore. the peak pressure is predicted to be related errors in commands or electrical miswiring to the approximately two times the initial pressure difference. telemetry response observed during the A-side activation Such a simple system corresponds to TRW's Series I anomaly. Icalibration) tests, in which a factor of two of initial Supported by the surge pressure analytical model pressure difference was observed. (which was calibrated by testing) and by test data using a For a more complicated system, such as the flight high fidelity mockup of the CGRO propulsion manifold, subs.,,stem and corresponding ground test simulated the anomaly investigation team reached the following key s?stem, however, numerical analysis (i.e., computerized conclusions'_.s.6: time-marching calculations) is necessary, to predict • Larger-than-expected bubble volume was likely left pressure levels and transients responses. Due to pressure in the manifold downstream of thruster isolation ,,,,ave interactions arising from complex hydraulic valves AI, A2. A3. and A4 during propellant loading. configurations, much larger overpressures were calculated The primary sources for the large bubble(s) were and measured than are estimated from closed form inadequate evacuation of the propulsion system and calculations based on simple pipe approximations. The nitrogen gas coming out of solution from the results of the detailed modeling are discussed below. hydrazine. The analysis indicated that 2-3.2 percent Empirical Calibration of Model and Comparison of (5.9-9.3 cm of line length) of the long, large diameter Results. Comparisons between model predictions and test manifold volume may have contained bubbles. data at various transducer locations were made for tests • Two key errors ,,,,+erecited as the cause of the large in,,olving liquid water and initial pressure differences trapped bubble volume in the manifold. The first f.k.P) across the latching isolation valve of 50. 100 and error was the use of a low resolution vacuum gauge 375 psid. [n the following comparisons, predictions of on the propellant loading ground support equipment. the pressure history were made using pressure drop losses The second error was the oversight in allowing across valves based on scaling from specification values, isolation of the manifold at pressure much Io_,er and pressure drop losses along the lines according to (14.7-[5.[ psia) than the "+pad" pressure 165 psia) of standard Moody diagrams. the hydrazine supply tank used in the leak integrity Comparisons between tests and analysis _,ith water demonstration test. ,,ieided peak pressure and frequency results of the The transient flow model developed by TRW showed decaying pulses t,,pically within 15 percent of each other. reasonable correlation to ground test data. [t '.,,as a This was true for initial _Ps of 50 psid and 100 psid. The useful anah. tJcal tool to assess pressurization options actual deca> rates of the pressure pulses ,,ere quite a bit for the CGRO propulsion system. The model faster than calculated, which is due to a smaller amount of revealed that the resultant pressure ,,,,as highly damping in the model as compared to the tests. sensitivity to the location and size distribution of the Examples of these results are sho,._,n in Figures 5 and bubbles. 6. [t should also be noted that the pressure ratio of peak pre,,.,,ure after opemng the val,,e to the pressure dir'fcrentiat before opemng the valve is approximately 5 at ._I_,'_"Oq)1-3631 P..L,.:,'e,_f12 GRO Series I Tests wtt.h Water i "'" "0 '---o-_ _ .. 'i_O__.._ Predated i P i LaIDT_ L24 I 1 i ! i'" A ]_ f.. 100r- _o_- I I E I , i ' i ', L 100 :- 0 • l I i l i Q-O 02 (1.4 0,6 0.8 1.0 0.0 O_ 014 0.8 0.8 1.0 _me. s_onc_ mle,m Figure 6. Water/l.36 in_Bubble/117 psiaTank Figure 5. Water/1.36 in3Bubble/67 psia Tank GRO Series II Tests with Water [IMuW4_llubl_e J_md _A1_I_m PmI_l. 1171_1_ "!I "l 0._ 0.2 0.4 O.II 0.8 1.0 tlrne, s_con(lI Figure 7. Water/Multiple Bubbles/ll7 psia Tank GRO Series IV Tests with ttydrazine GRO Series IV Tests witl_ Hydrazine I'_A-_,_,_I ,.. ': ._: __Z_l _ -L,, " t i L F /'_! /_ ,_ ,, L 0' 010 0.2 0.4 0.8 O_ t0 Umo. Figure 8. Hydrazine_lultiple Bubbles/117 psia Tank Figure 9. HydrazinefMultiple Bubbles/392 psia Tank t I.\.% _'tlt)I-.tOt3I P:l_c ";,_t {2 \ rncth_td referred t,._.is 'fast <)cling" uf tke isolation tc_,ts ;_,erc run ,._,ith h_, prcs.,,t,res i07 and II_ p,,la) up,,tream and [7 p,,la do_,n.,,tream of the .\[-equi,,alent _al_e--_)pcning the isoiatltql ',al_,es ["or _,er_, short l.,,olati_m val,,e, and there ,._as a final "'moment c)f truth" duratum--,._,a_, demon_,tr;lted to be a _,iahlc option to ,,aiL'.ly raise the h_,drazine pressure in the manifold to test approximating the actual flight conditions _,_ith 392 design operating pressure. Using breadboard psia upstream and [7 psia downstream The resulting electronics and refurbished isolation ,,alves. TRW pressures ,,_,erehigh enough to damage the isolation ,,alve: showed that the valve could respond to a command it indicated "'closed" only after several commands were sent to close it. and it had gross leakage in the closed sequence consisting of a valve OPEN command tcommand to isolation valve "'open coil"), followed position following this test. This test substantiated the "'N'" ms later with a valve CLOSE command suspected damage to CGRO isolation valves AI and AIB 4command to isolation valve "'close coil"). and indicated the need to switch to B-side and use a modified priming sequence to avoid similar damage to The JSC/White Sands Test Facility fWSTF) played a isolation valve BI. critical role in helping the CGRO anomaly investigation team assess risks of the candidate manifold pressurization The analytical model, calibrated with empirical data, options s. WSTF previously had performed extensive showed that during the on-orbit priming attempt on April testing to characterize the likelihood of "'adiabatic 7 the A-side of the CGRO propulsion subsystem had compression decomposition" (ACD) in hydrazine experienced peak surge pressures that ranged from 1,200 propulsion systems. The basic mechanism of ACD is that to 4,900 psia. These levels are consistent with analysis of a rapid compression of a gas bubble containing hydrazine pressure loads required to damage the bellows assemblies within the isolation valve. ,,apor will heat the gas, which in turn might initiate rapid decomposition of the hydrazine vapor (an exothermic After assessing the risks and benefits, a decision was made to use fast cycling of the isolation valves on the B- process), thereby causing peak pressures far in excess of the already large waterhammer pressure. side manifold. The principal advantage of the B-side manifold was that it had not been exposed to the high The objectives of the WSTF tests were: surge pressure. The isolation valve response I) determine the likelihood of ACD having occurred in characteristics could be assumed to be reasonably close to the initial priming attempt on CGRO A-side acceptance test conditions. Further, the "as-launched'" manifold, pressure conditions in the manifold could be modeled 2) provide additional hydrazine vs. water comparison more accurately in the analysis. data for TRW's analytical model, and The fast cycling method was validated at NASA 3_ expose an ACT valve to surge pressures of the levels GSFC with a CGRO-like command and telemetry system analytically predicted (assuming no ACD) to help and a representative mockup of the CGRO B-side assess the state-of-health of A-side thruster valves propellant manifold. Key hardware in the test setup were and the risks of continuing on A-side versus an engineering model CGRO Electrical Interface attempting priming on B-side. Assembly (EIA) subdecoder, a breadboard isolation valve The WSTF test setup used simple tubing driver and a refurbished isolation valve. For safety configurations that replicated CGRO manifold line reasons, water was used instead of hydrazine. The tests lengths and diameters, but no attempt was made to characterized the effect on pressurization rate due to replicate individual components, tees, bend angles, etc. A factors such as valve pulse width (duration between fast response, low pressure-drop ball valve was used in commands to the isolation valve open and close coils, or the test setup to simulate the CGRO isolation valve. This effectively the "'open time" of the valve), pressure test hardware approach permitted many tests to be rapidly dov_nstream of the isolation valve, and spacecraft voltage. performed without risk to the higher value, high fidelity A linear displacement transducer was used to measure s,,stem mockup used at TRW. movement of the isolation valve poppet as a function of The WSTF tests provided data for understanding how electrical pulse width between commands to the open and surge pressure was affected by factors such as fluid close coils. The tests demonstrated that at 8-12 ms composition (hydrazine versus water), supply pressure, (millisecond) open pulse widths, the valve poppet could bubble size, and tubing diameters and lengths. Conditions be safely constrained to ,*ithin the desired 1-8 percent of that trigger significant hydrazine decomposition were the full 90 rail (2.3 mini stroke. ditf2cuh to define, but the resultant pressures were high Results from tests performed at TRW '.,,ere used to enough to rupture one of the tubes during the tests. determine response &fferences between the CGRO-like command and telemetry simulators used in the tests While the TRW high fidelit', propulsion subsystem versus the CGRO flight s>stem. The TRW flight mL,ckup initially used water ti)r testing, a final series of operatitms team de,.eloped and validated t,>,_ critical te,4s ,_,as perfi_rmed using hydrazine. Nineteen h;drazine pr,_cedures that >,lgnificantly increased the reliabilit.,, and %I.k',,. 201)L-363 l I:'a_c '),_I [2 efficient', ,_t the fa_,t cychng techmque. [he first the priming ,_pcratum in hal/ R,,ughi? 3"q)() val'+e ta,,t pr,,cudurc a,,,.,,ured umnterrupt_ble blocks of electrical cycles were reqmred to prime lhe I3-,idc .\('T mamtt_ld pub,c _,ldths to the _solatitm ,,alve _pen/ch>:,e ctnls, up to a max_rnt,m of i4 ms duration, using commands issued A sectmd pr_pulsum subs',stem on-orbtt anomaly frum the On-Board Computer _OBC). The second occurred in May-June [o93 during the calibration burns procedure was a high rate telemetry patch (b4 ms segment or" the orbit raising operatu>n, usmg the sampling} tot the pressure transducer. The capability to successt'ull? primed B-side. During the initial test firings. monitor transient pressures during the fast cycling one of the four ACTs I"ACT-B2"') produced unacceptably operation significantly decreased the time to prime the B- low thrust. side propellant manifold. While several potential causes were investigated. evidence led to suspecting the thruster vat,,e, in particular B-SIDE ACTIVATION & SUBSEQUENT the valve's AFE-4I I seat area. It v,as noted that the AVs OPERATIONS performed with relatively cool thruster start temperatures The least risk assessment by the anomaly resolution resulted in nominal performance while the kVs initiated team concluded that only the B-side ACT manifold with warmer start temperatures were anomalous. Review should be primed to support the orbit raising operation. of the test history for all the valves in the CGRO lot The ACTs alone could perform the required orbit raising showed a tendency of the valve to change flow without exceeding their qualification limits. Furthermore, characteristics with varying temperature and exposure to it was determined that the B-side OAT manifold should certain fluids. In particular, the acceptance test data show only be primed just prior to required use of the OATs for increased ..M:' with elevated temperature. Further, it was the controlled re-entry firings at the end of mission life. discovered that the valve used on thruster ACT-B2 was The baseline plan tbr priming of the B-side ACT listed on a SIR (Supplier Information Request) for manitbld assumed that the 6.4 ms/sample pressure anomalous "non-flow'" during acceptance testing at transducer telemetry patch would not provide sufficient elevated temperature. This valve was reworked and later sampling of transient pressure data to assess the progress accepted. As further evidence, early in the valve of the priming operation. The baseline procedure was as procurement a valve failed flow test (no-flow) due to follows" elevated temperature and the use of alcohol. Both TRW • Alternately fast cycle isolation valves BL and B2 for and the vendor investigated the SIR problem and 500 cycles increasing the valve "'open" pulse widths concluded the alcohol flow test fluid had caused the seal in I ms increments from 8 to 12 ms to swell and close off the seat opening. The investigation • Fast cycle isolation valve B3 for 500 cycles indicated the seal was not sensitive to exposure to increasing the valve '+open" pulse widths in I ms hydrazine or water at elevated temperature, so it was increments from 8 to 12 ms. On the final 12 ms set, concluded that the valve design was acceptable as long as open B3 for 1minute alcohol was not used as the elevated temperature flow test fluid. • Repeat fast cycling sequence for isolation valves B 1, Extensive consultations with the thruster valve B2 and B3 at 300 cycles, 150 cycles, 100 cycles supplier and ground tests conducted at TRW determined • Open isolation valves B 1. B3, and B2 that the low thrust observed on-orbit was most likely • Fast cycle cross-over isolation valves BIA and B2A related to flexing of the thruster valve seal at high for 100 cycles increasing valve "'open" pulse widths propellant mass flow rate. This was confirmed by a hot in I ms increments from 4 to 12 ms fire test series of sixteen test runs at fixed inlet pressure • Open isolation valves BIA. B2A. ALB, and A2B. (350 psia) and gradually increasing valve and propellant Since about ten thousand valve open/close commands temperature (from ambient to 220 "F). For reference, the were anticipated, a 30 second wait was imposed between valve temperature reported from the spacecraft during the each fast cycle to reduce the thermal stress on the anomaly period ,,,,as in the range of 95"F to 120°F. A electronics. ground spare CGRO thruster ,.,,as tested, using two While performing the first set of fast cycling with separate, exchangeable val,,es that ,.,,ere similar to the isolation valves B I and B2, the 64 ms sampling of anomalous flight valve with regard to flow AP as pressure telemetry was configured to increase the measured during valve acceptance test. likelihood of capturing the initial cycles of the pressure At ele,,ated val,.e temperatures, the thruster exhibited oscillations. Having established this capability, the degraded performance for the first 14 seconds of a steady pressure telemetry was used _o determine when to stop the state firing, revealing that the seat material caused a flow fast c,,cling and to command open the isolation valves. restriction. The initial thruster chamber pressure was Implementing this methud reduced the time to complete approximately 10% of nominal dur,.no_, the first six seconds of the run. Thruster temperatures and pressures confirmed that _h_s was n_t a catal,,,,t bed 'v_a:,h_mt'"