AIAA 2002-4139

Page 1

DESIGN AND MANUFACTURE OF THEMESSENGER PROPELLANT TANK ASSEMBLY

Walter TamPressure Systems, Inc.

and

Sam Wiley and Katherine DommerAerojet

and

Larry Mosher and David PersonsJohns Hopkins University Applied Physics Laboratory

ABSTRACT

An ultra-lightweight propellant tank wasrequired for the MESSENGER spacecraft, anda new tank development program wasconducted to design, fabricate, and test thistank. The development program wasconducted in three phases: trade study,analysis and design, and hardware fabricationand test.

Phase 1 trade study was conducted todetermine the most weight efficient tankdesign. This phase was done primarily byanalysis with multiple iterations. Over 50 tankconfigurations were considered before a finalselection was made.

Phase 2 design and analysis included efforts todesign and analyze a vortex suppressor, anti-slosh baffle for nutation control during launch,and tank shell. The effort included subscaledrop tower simulations to determine thenumber, size and location of anti-slosh baffle,analytical determination of the loads on thevortex suppressor and baffle, baffle structuralanalysis, and tank shell stress and fracturemechanics analyses.

Phase 3 fabricated a qualification tank and fourflight tanks (3 flight and a spare). The tankshell components were fabricated fromsolution treated and aged (STA) 6AL-4Vtitanium alloy. The anti-slosh baffles weremachined from annealed 6AL-4V titanium ringforgings, and the vortex suppressors werefabricated from 6AL-4V titanium sheets. Thetank shell was assembled with 4 girth welds-two of which were made to have STAproperties, and the remaining two wereannealed closure welds that were also baffleinstallation welds. All 5 tanks were fabricatedusing identical processes and procedures.

The qualification tank must undergo aqualification test program that includes loadedsine and random vibration testing. Thequalification test program is to conclude with adestructive burst pressure test. All flight tanksare protoflight tested prior to precision cleanand delivery.

The completed flight tank has a mass of lessthan 20 pounds, including attachmenthardware. This ultra-lightweight tank will play acritical role toward the success of theMESSENGER Program.

Copyright 2002 by Pressure Systems, Inc. Publishedby American Institute of Aeronautics & Astronauticswith permission.

Page 2

INTRODUCTIONIn 2000 Pressure Systems, Inc. (PSI) wascontracted to provide a full complement ofpropulsion system tanks for the MESSENGERspacecraft. These tanks include a 12-inchdiameter elastomeric diaphragm tank, a 16-inchdiameter by 30-inch long COPV heliumpresurrant tank, and three 24-inch diametermain propellant tanks-two fuel and one oxidizer-of the same configuration. See Figure 1. Boththe diaphragm and pressurant tanks arederivatives of existing PSI tank designs whichrequired no development. However, the mainpropellant tank required significant effort fordevelopment and qualification.

The MESSENGER propellant tank developmentprogram drew heritage from the NEAR program

propulsion system tank development1. Therewere many similarities between the twoprograms, such as the initial trade study, thedevelopment of a vortex suppressor, and thepropellant management philosophy of using adiaphragm tank to allow settling burns to settlepropellant at the tank outlet prior to the start ofmain engine burns. However, there were alsonew challenges, such as the analysis anddevelopment of anti-slosh baffles, and thedesign, analysis, and test of a tank that'slaunched in an upside down orientation (outletside on top).

The main propellant tank development programwas separated into three phases: (1) tradestudy, (2) analysis and design, and (3)fabrication and test.

Figure 1, MESSENGER Propulsion System Tanks

Fuel Tank#2Oxidizer Tank

Helium TankAuxiliary

Diaphragm Tank

Fuel Tank#1

Page 3

PHASE 1, TRADE STUDY

The goal of the phase 1 trade study was todetermine a tank configuration that offers thelowest tank mass. This was a 3-month longeffort involving several analytical iteration andrepeated reviews. Factors considered in thetrade study included tank diameter, resonant

frequency, tank shell thickness, tank shelltransition, types of tank mounts (flange, tabs,struts), location of tank mounts, size (length,width and thickness) of mounting tabs, risk, andcost. Over 50 tank configurations wereconsidered. Figure 2 below provide a sampleof tank configurations considered in the tradestudy.

Figure 2, Some Tank Configurations Considered in the Trade Study

∞

∞

∞

∞

∞

∞

∞

∞

∞

∞

∞

∞

∞

∞

Page 4

At the conclusion of the trade study, a tank configuration as shown in Figure 3 was selected for analysisand detail design.

The propellant tank specification requirements are listed below in Table 1:

Table 1: Propellant Tank Assembly Design Requirements

Parameters RequirementsOperating Pressure MEOP is 325 psia @ 50°C (122°F), 100 cyclesProof Pressure 406 psia @ 50°C (122°F), 16 cyclesBurst Pressure 488 psia minimum @ 50°C (122°F),Material of Construction Membrane: 6Al-4V titanium, solution treated and aged

Inlet/outlet ports: 3Al-2.5V titaniumVortex suppressor: 6AL-4V titaniumBaffles: 6AL-4V titanium

Expulsion Efficiency 99.75% minimumPropellant Weight 611 lbm (277 kg) maximum nitrogen tetroxidePropellant Fill Fraction 85% minimum, 95% maximumTank Capacity 12,150 in3 (199.1 liters) minimumInternal Dimension 22.14” ID x 39.175” longPropellant Flow Rate 8.7 gpm minimumOverall Length 40.95”, boss to bossTank Weight 20.9 lbm (9.5 kg) maximumPropellant Hydrazine and Nitrogen tetroxideFluid Compatibility N2H4, N2O4, GAr, GHe, GN2, D.I. water, Isopropyl alcoholShell Leakage 30 Hz in lateral direction, > 35 Hz in thrust directionFailure Mode Leak Before BurstOn-Orbit Temperatures 50 to 122 °F (10 to 50 °C)Shelf Life 5 years minimumOperating Life 8 years minimumRange Safety Per EWR 127-1 October 1997

Upper Struts(2 ea.)

Lower Struts(2 ea.)

Figure 3, The MESSENGER Tank

PressurantHemisphere

Cylinder

PropellantHemisphere

Bearing Tab

Side Tabs(2 ea.)

Page 5

PHASE 2, ANALYSIS AND DESIGN

Several design analyses were performed todesign the MESSENGER Propellant Tank.These analyses include vortex suppressorstructural verification, baffle design andstructural load analyses, and stress andfracture mechanics analyses for the tank shell.

VORTEX SUPPRESSOR AND VORTEXSUPPRESSOR LOAD ANALYSIS

The MESSENGER vortex suppressor is basedon the NEAR vortex suppressor design andconfigured as a four-vane cruciform. It ispositioned directly above the propellant outlet,as shown in Figure 4.

A structural analysis was performed to validatethe vortex suppressor design. As notedbefore, the MESSENGER propellant tank isoriented upside down during the launchsequence, thus the vortex suppressor isalways dry during launch. The only factorsconsidered during vortex suppressor analysiswere pad slosh and vibration. It was found thatpad slosh exerts the highest load on the vortexsuppressor, but the vortex suppressor designstill provides a positive margin of safety.

BAFFLE DESIGN ANALYSIS

It was anticipated that an unbaffledMESSENGER propellant tank wouldexperience severe propellant sloshing duringlaunch, resulting in unacceptable nutationgrowth of the launch vehicle spin stabilized 3rdstage. During the course of tank design, it was

decided that baffle (or baffles) must beinstalled to provide nutation control.

Prior to detail baffle design, subscale droptests were conducted to determine the numberand the location of the baffle(s) for theMESSENGER propellant tank. Severalsubscale models were constructed for this test,including spacecraft centerline oxidizer tankand off-axis fuel tanks, as seen in Figure 5.

Additionally, several sizes of baffle rings weremanufactured and tested, as shown below inFigure 6.

The data collected from the drop tower testsupported a tank design with two annular ringbaffles. The baffles are located on the tankcylindrical section approximately 8 inches apartand equal distance from the tank mid-plane.Both baffles are identical in size -approximately 8 inches wide with a 7-inch"hole" in the center of the baffle.

Figure 4, The MESSENGER Tank VortexSuppressor

Figure 5, Off Axis Fuel Tank Model

Figure 6, Simulated Baffle rings

Page 6

BAFFLE LOAD DETERMINATION

Prior to the structural analysis of the baffles, afluid dynamics analysis was conducted todetermine the loads exerted on the baffles.Factors considered were pad slosh, test andlaunch vibration, and ignition of the launchvehicle 3rd stage (AKM ignition). Both theoxidizer tank baffles (centerline on spin axis)and fuel tank baffles (off axis) were examined.Some of the 3-D models produced are shownbelow in Figure 7:

The pressure distribution on the baffles wasplotted and the loads quantified. Some of thepressure load plots are presented below:

The baffle loads analysis showed that baffleloads are low, since the fluid supports thebaffle during launch. The analysis concludedthat the loads on the oxidizer tank during NTOAKM ignition are worst case, and a design loadwas quantified for subsequent structuralanalysis.

Figure 7a, Propellant Movement Due toPad Slosh

Figure 7b, Propellant Movement Due toAKM Ignition, Fuel Tank

Figure 7c, Propellant Movement Due toAKM Ignition, Oxidizer Tank

Figure 8a, Pressure Distribution due toPad Slosh, Upper Baffle

Figure 8b, Pressure Distribution due toAKM Ignition, Fuel Tank

Figure 8c, Pressure Distribution due toAKM Ignition, Oxidizer Tank

Page 7

BAFFLE STRUCTURAL ANALYSES

A baffle structural analyses was performed tovalidate the structural integrity of the ring baffledesign. The analysis took into considerationdesign requirements such as materialproperties, fluid properties, baffle loads asdetermined in the previous study, vibrationenvironment, and design safety factors. Bothaxi-symmetric and asymmetric models wereconstructed to facilitate the analyticalevaluation. Cases analyzed include pad slosh,vibration, and AKM ignition. Finite ElementsModels were constructed to examine elementssuch as pressure loading and displacement, asshown in Figure 9 below:

Figure 9a, FEM, Pressure Loading

Figure 9b, FEM, Displacement

Stress contour plots were also used toexamine surface stress/strain. As shown inFigure 10.

Figure 10, von Mises Stress Contour Plot

The baffle structural analysis concluded withlarge positive margins of safety for thenominally 0.010 inch thick baffle.

TANK SHELL ANALYSES

The tank shell analyses included stressanalysis and fracture mechanic analysis. Allanalyses used assumptions, computer tools,test data and experimental data utilized on amajority of the pressure vessels successfullydesigned, fabricated, tested and qualifiedduring the past three decades. Conservatismwas used throughout the analysis process, andthe worst case scenarios were analyzed.

TANK SHELL STRESS ANALYSIS

A stress analysis was performed to establishthat the tank meets the specificationrequirements. The analysis took intoconsideration the requirements such as:

• Temperature environment;• Material properties, STA titanium;• Material properties, annealed titanium;• Volumetric requirements;• Mass properties of tank shell material;• Mass properties of fluid;• Fluids used by the tank;• Tank pressurization history;• External loads;• Girth weld offset and weld suck-in;• Size of girth weld bead;• Resonant frequency;• Tank boundary conditions;• Residual stress in girth weld;• Load reaction points; and• Design safety factors.

Page 8

This stress analysis established the tank shelldesign and mounting features for theMESSENGER propellant tank. The analysisdynamic model provided resonant frequencypredictions. The first through third modes areshown below:

Figure 11a, Oxidizer Tank 1st Mode

Figure 11b, Oxidizer Tank 2nd Mode

Figure 11c, Oxidizer Tank 3rd Mode

Buckling analysis was also conducted as partof the tank analysis. Select buckling shapesare shown below in Figure 12:

Figure 12, 1st and 2nd Buckling Modes

STRUT DESIGN ANALYSIS

An analysis for all-titanium struts was alsoconducted as part of the overall tank design.As part of the acceptance and qualificationphilosophy, each set of four lightweight strutsaccompanied the flight and qualification tanksthroughout acceptance and qualificationtesting.

Page 9

The stress analyses concluded with positivemargins of safety for all design parameters, assummarized in Table 2.

Table 2: Propellant Tank Safety Margins

Characteristics M.S.

Pressure, proof, sphere, yield +0.06

Pressure, proof, cylinder, yield +0.05

Pressure, burst, sphere, ultimate +0.02

Pressure, burst, cylinder, ultimate +0.09

Girth weld, yield +0.01

Girth weld, ultimate +0.14

Inlet tube, yield +3.60

Inlet tube, ultimate +6.30

Outlet tube, yield +4.14

Outlet tube, ultimate +7.00

EB weld, yield +0.07

EB weld, ultimate +0.18

Baffle, yield +0.17

Baffle, ultimate +0.19

Strut, yoke, yield +1.80

Strut, yoke, ultimate +2.30

Strut, yoke, global buckling +1.01

Strut, yoke, local crippling large

Strut, yoke, bolt, ultimate +2.74

Strut, yoke, weld, ultimate large

Strut, side, yield +1.60

Strut, side, ultimate +2.10

Strut, side, global buckling +1.52

Strut, side, crippling large

Strut, side, bolt +3.90

Vortex suppressor, screws, yield +3.43

Vortex suppressor, screws, ultimate +3.48

Vortex suppressor, legs, yield +1.01

Vortex suppressor, legs, ultimate +1.05

Axial tab (+X side), external load, yield +0.50

Axial tab (+X side), external load, ult. +0.53

Shell, axial tab, external load, yield +0.97

Shell, lateral tab, external load, yield +1.09

TANK SHELL FRACTURE MECHANICSANALYSIS

A fracture mechanics analysis was performedto establish whether the growth of an initial flawin the anticipated cyclic and sustainedpressure environment may cause a failure inthe tank shell. The analysis was performedusing external and internal stresses from thestress analysis, and using NASA/FLAGRO withminimum thicknesses as parameters. Specialfracture critical dye-penetrant and radiographicinspections are required to detect flaws. Theminimum flaw size that can be detected bysuch special fracture critical inspections wasused as initial flaw size for this fracturemechanics crack propagation analysis. Theanalysis was performed at:

• Girth welds and heat affected zones;

• Maximum pressure stress location in thehemisphere;

• Maximum stress location in the cylinder;

• Maximum stress location in the cylinderside boss;

• Maximum stress location in thehemisphere/cylinder transition;

• Intersection between the hemisphere andthe pressurant boss;

• Intersection between the hemisphere andthe propellant boss; and

• Maximum external load stress in thehemisphere near the pressurant and thepropellant bosses.

The fracture mechanics analysis establishedthe leak-before-burst (LBB) characteristics ofthe propellant tank. This analysis concludedthat the MESSENGER tank shell meets all thefracture mechanics requirements. The specialfracture critical NDE flaw screening required bythis fracture mechanics analysis include:

• Fracture critical dye-penetrant inspection;and

• Fracture critical radiographic inspection.

These requirements were instituted as part ofthe tank fabrication requirements.

Page 10

TANK FABRICATION

The MESSENGER propellant tank shellconsists of two hemispherical heads, twocylinder extensions, and a cylindrical centersection. All 5 shell component pieces weremachined from 6AL-4V titanium alloy forgings,as shown in Figure 13. All raw forgings haveannealed properties at the time of receipt.

Figure 13, Hemisphere & Ring Forgings

The outlet hemisphere and its cylinderextension forgings were rough machined andelectron beam (eb) welded together, as shownin Figure 14.

Figure 14, EB welded Outlet Hemisphereand Cylinder Extension

This assembly was solution treated, aged, andfinish machined to the tank shell thickness asrequired by the stress analysis. This processallowed the eb weld joint to have STAproperties. At final machine, the eb weld jointwas machined as part of the shell membraneand became indistinguishable from the rest ofthe membrane material. After final machine,the eb weld joint has the same membranethickness (0.020 ± .003) as the rest of themembrane material. There is no weldreinforcement at or near the eb weld joint. A

finish machined outlet hemisphere/cylinderextension is shown in Figure 15.

Figure 15, Finished Machined OutletHemisphere and Cylinder Extension

The mounting hemisphere was also eb weldedto a cylinder extension, solution treated andaged, and finish machined. However, part ofthe finish machine process also included ebwelding three tabs to the mounting hemisphereand final machine these mounting tabs. Afinish machined mounting hemisphere isshown below in Figure 16.

Figure 16, Finished Machined MountingHemisphere and Cylinder Extension

The fifth tank shell component, the centercylinder, also underwent the rough machine-solution treat-finish machine manufacturingprocess. The solution heat treat processincreases the strength of the titanium alloy,thus minimizing the weight of the tank shell.The excellent strength to weight property,coupled with its manufacturability, make thistitanium alloy the material of choice foraerospace application.

EB Weld

Page 11

In addition to the tank shell components, twobaffles were machined from titanium ringforgings. These baffles have a nominalthickness of 0.010 inch. A picture of amachined baffle is shown in Figure 17.

Figure 17, A Machined Baffle

Prior to the start of girth weld, the vortexsuppressor was installed above the outlet port.This four-vane cruciform vortex suppressorwas constructed of titanium sheets andsupported with four support posts. The foursupport posts were welded to the tank shell, asshown in Figure 18.

Figure 18, An Installed Vortex Suppressor

The MESSENGER propellant tank wasassembled with two girth welds. Both girthwelds have the same weld joint design. Thisweld joint is nearly identical to weld joints on allPSI's diaphragm tanks, thus provided withflight heritage of almost 900 tanks.Nevertheless, a Weld Development Programwas conducted to develop a customized weldschedule for the MESSENGER Propellant

Tank. Each girth weld was designed to jointogether three components: ahemisphere/cylinder extension assembly, thecenter cylinder, and a machined baffle. Figure19 shows a tank assembly with the first girthweld complete, which includes the welded-inbaffle .

Figure 19, The Welded Baffle

After both welds are complete, the girth weldswere subjected to fracture critical radiographyand dye penetrant inspection as required byfracture analysis. Figure 20 shows the tankbeing radiographically inspected after closureweld.

Figure 20, Radiographic Inspection,Propellant Tank Assembly Girth Weld

Page 12

After closure the tank assembly is stressrelieved in a vacuum furnace to removeresidual stress from the weld operations. SeeFigure 21.

Figure 21, Vacuum Heat Treat

Following closure weld, a boomerang wasinstalled to the propellant hemisphere asshown in Figure 22. The boomerang is usedfor strut connection.

Figure 22, Installed Boomerang Assembly

TANK WEIGHT

The propellant tank weight per the specificationis not to exceed 20.9 lbs. The actual weight ofthe qualification tank is less than 19.00 lbs.With the addition of 1.01 lbm for the four strutsand mounting hardware, the as-delivered tankweight is less than 20.0 lbs.

QUALIFICATION TEST PROGRAM

The propellant tank is qualified by test. Aqualification tank was fabrication for thequalification test program. The qualificationtest sequence is listed below:

- Preliminary examination- Pre-proof volumetric capacity- Ambient proof pressure test- Post-proof volumetric capacity- MEOP pressure cycle- Proof pressure cycle- Expulsion efficiency- Flow rate determination- External leak test- Qualification vibration test- External leakage test- Penetrant inspection of girth welds- Radiographic inspection of girth welds- Final visual examination- Burst pressure test

Conservatism was exercised throughout thequalification test program, and all pressuretests were temperature adjusted for the worstcase operating temperature. Pass/Fail criteriaconsisted of acceptance type external leaktests and non-destructive evaluationsconducted at intervals throughout the testprogram.

The MESSENGER Program qualificationtesting philosophy was to test associated flightitems along with the tank, including mountingstruts, thermal switches, a Click Bond, and aheater installed around the perimeter of thetank's cylindrical section. These items wereinstalled onto the qualification tank prior to thestart of qualification testing.

Volumetric Capacity Examination: Thecapacity of the propellant tank was measuredutilizing the weight of water method, usingclean, filtered deionized water as the testmedium. This test was conducted before andafter the proof pressure test to verify that theproof pressure test did not significantly alterthe tank capacity.

Proof Pressure Test: The proof pressure testwas the first pressurization cycle applied to thetank after fabrication. It was intended tovalidate the workmanship by verifying thestrength and integrity of the tank shell. Thetest was conducted hydrostatically at proofpressure (406 psia, normalized for test

Page 13

temperature) for a pressure hold period of 5-minute minimum. Tank linear and radialgrowths were also recorded to validate theanalytical predictions. Figure 23 below showsthe pressure test setup for the MESSENGERpropellant tank.

Figure 23: The Pressure Test Setup

Pressure Cycles Tests: The propellant tankwas cycle tested with a total of 16 proof and100 MEOP pressure cycles. Pressure testingwas conducted hydrostatically. Both proof andMEOP test pressures were temperatureadjusted to test the worst case conditions.

Expulsion Efficiency: The propellant tankwas drained and the fluid expulsion efficiencydetermined. The MESSENGER propellant metthe 99.75% expulsion efficiency requirement.

Flow Test: A flow test was conducted todetermine flow rate through the propellantoutlet. Figure 24 shows the flow test setup.Note that the flow test was conductedpropellant side down. The tank easily met therequired flow rate of 8.7 gallon per minute.

Figure 24: The Flow Test Setup

Qualification Vibration Test: Thequalification vibration test is designed to verifyworkmanship of baffle and vortex suppressor,as well as the integrity of the tank shell. Thereare only two phases of the qualificationvibration testing: wet random and wet sine. Allthree principal axes are tested. Deionizedwater was used as the test fluid and therequired vibration levels were adjusted toaccount for the fluid density difference betweendeionized water and NTO. For both randomand sine vibration testing, the QualificationTank was loaded with 422 lbm of D.I. waterand pressurized to 243 psig, adjusted fortemperature. The tested vibration environmentis shown below in Table 4.

Two test fixtures were fabricated for thevibration testing. The random vibration testfixture was designed to test the tank in thehorizontal position, while the sine vibration testfixture was designed to test the tank in thevertical position. Both fixtures were designedto mount the propellant with flight struts. SeeFigure 25 for the test setup.

Page 14

Table 4a: Qualification Vibration Levels

Qualification Sine Vibration

Axes Frequency (Hz) Acceleration Sweep RateThrust 10 – 24

24 – 2828 – 100

0.5 in DA14.30 g2.48 g

2 oct/min

Lateral(2 axes)

10 – 2020 – 25

25 – 100

0.34 in. DA6.90 g1.96 g

2 oct/min

Qualification Random Vibration

Axes Frequency (Hz) PSD Input Level OverallAmplitude

(Grms)

Duration(sec.)

Thrust 2080

8002000

0.010.080.080.01

9.7 120

Lateral(2 axes)

2080

8002000

0.010.040.040.01

7.3 120

Figure 25: Vibration Test Setup

Random, X-axis Random, Y-axis

Random, Z-axis, Sine, X-axis

Page 15

External Leak Test: The external leak testverified the integrity of the tank shell and alsoserves to validate the above vibration testing.The tank is placed in a vacuum chamber,evacuated to under 0.2 microns of mercury,and helium pressurized to 325 psia for 30minutes. The helium leak rate must thespecification requirement of

Page 16

ABOUT THE AUTHORS

Mr. Walter Tam is a Program Manager atPressure Systems, Inc., Commerce,California. Mr. Sam Wiley is an IntegratedProduct Team Lead at Aerojet, Sacramento,California. Ms. Katie Dommer is a LeadSystems Engineer at Aerojet. Mr. LarryMosher is a Lead Spacecraft PropulsionEngineer at Johns Hopkins University AppliedPhysics Laboratory, Laurel, Maryland. Mr.Dave Persons is a Lead Spacecraft StructuralEngineer at APL.

NOTES