1 1 DESIGN AND DEVELOPMENT OF A COMPOSITE DOME FOR EXPERIMENTAL CHARACTERIZATION OF MATERIA

DESIGN AND DEVELOPMENT OF A COMPOSITE DOME FOR EXPERIMENTAL

CHARACTERIZATION OF MATERIAL PERMEABILITYHector Estrada, Assistant ProfessorDepartment of Civil and Environmental EngineeringVanderbilt UniversityBox 44, Station BNashville, TN 37235Ph: (615) 343–4562, e–mail: estrada@vuse.vanderbilt.eduStanley S. Smeltzer, III, Research EngineerMS 190, Structural Mechanics BranchNASA Langley Research CenterHampton, VA 23681–2199ABSTRACT

This paper presents the design and development of a carbonfiber reinforced plastic dome, including a description of the domefabrication, method for sealing penetrations in the dome, and asummary of the planned test series. This dome will be used forthe experimental permeability characterization and leakage val-idation of composite vessels pressurized using liquid hydrogenand liquid nitrogen at the Cryostat Test Facility at the NASA Mar-shall Space Flight Center (MSFC). The preliminary design of thedome was completed using membrane shell analysis. Due to theconfiguration of the test setup, the dome will experience someflexural stresses and stress concentrations in addition to mem-brane stresses. Also, a potential buckling condition exists for thedome due to external pressure during the leak testing of the cryos-tat facility lines. Thus, a finite element analysis was conductedto assess the overall strength and stability of the dome for eachrequired test condition. Based on these results, additional pliesof composite reinforcement material were applied to local re-gions on the dome to alleviate stress concentrations and limitdeflections. The dome design includes a circular opening in thecenter for the installation of a polar boss, which introduces a geo-metric discontinuity that causes high stresses in the region nearthe hole. To attenuate these high stresses, a reinforcement systemwas designed using analytical and finite element analyses. Thedevelopment of a low leakage polar boss system is also investiga-ted.

INTRODUCTION

Cryogenic fluids have been used successfully for space pro-pulsion systems since the 1950s. In fact, they are the main fuel

components of most propulsion systems in the United Statestoday. The traditional containment vessel for these cryogenicfluids has exclusively been metallic. However, because of theneed to reduce the overall weight of space vehicles, compositematerials are now being considered as a possible alternative tometallic materials. In the pressure vessel industry, compositeshave successfully contained low-pressure fluids and gases. Aconcern associated with storing cryogenic fluids in a compositepressure vessel is the diffusion of the fluid through the tank wall,since composites are more porous than metals. Typically, thismaterial permeability problem has been addressed by using me-tallic liners, since limited information on composite materialpermeability at cryogenic conditions exists. Although linedcomposite pressure vessels offer significant weight savings overtheir metallic counterparts, the liner adds unnecessary weight andreduces potential payload. Therefore, a program was initiated toinvestigate the material permeability characteristics and methodsfor sealing penetrations for liner-less composite vessels contain-ing pressurized cryogenic liquids.This paper describes the design and development of a carbonfiber reinforced plastic (CFRP) pressure vessel to be used as a testarticle in the experimental permeability characterization of com-posite materials. The composite dome test article will be sub-jected to thermal and pressure cyclic loading using liquid hydro-gen and liquid nitrogen. A complete summary of the researchproject is given with a focus on the design of the structural systemand the development of the access system. The paper is orga-nized as follows. First, we describe the dome and the cryostat set-up. Then, we describe the failure modes for the dome. We coverthe analysis and design process next, followed by a brief descrip-tion of the dome polar boss sealing system.1

DESCRIPTION OF THE DOME AND CRYOSTAT SETUPThe test article is a one-piece carbon/epoxy composite hemi-spherical dome cap and flange. The dome cap has an eighty-inch(2,032 mm – millimeters) radius and an axial height, H, ofapproximately six inches (152 mm). The flange extends fiveinches (127 mm) outward from the outside edge of the dome capand is oriented perpendicular to the dome center line shown in thedome cross-section in Fig. 1. A cross-sectional view of the testarticle is given in Fig.1. Figure 2 depicts an exploded view of theCryostat Test Facility (CTF) assembly containing the test article.The test article’s geometry and dimensions were chosen to ac-commodate the existing CTF, as shown in Fig. 2, as well as simu-late geometric scale of Reusable Launch Vehicle (RLV) hard-ware. The test article was fabricated with Hercules IM7/8552preimpregnated (or prepreg) carbon/epoxy, plain weave fabricusing the hand lay-up fabrication method. A single piece alumin-ium tool was used for both fabrication and curing of the test ar-ticle. Forty-five degree gore sections of fabric that began at thecenter of the dome and extended to the outside edge of the flangeDome Cross Sectioncontainment shell(permeability volume)test articlering spacercontainment shell(permeability volume)CL

80 in. (2,032 mm)21.76°Figure 2: Cryostat Assembly.

D5 in. (127 mm)containment shelltest articlering spacerDetail Dcontainment shellFigure 1: Geometry of Intact Dome.were cut and aligned on the tool. Subsequent plies were”clocked” at fifteen-degree increments around the tool to keepfrom aligning cuts through the thickness of the laminate. Thelaminate configuration for the test article is [0°ń90°,\"45°]3S,which produced a twelve ply, 0.12 inch (3 mm) nominal thick-ness, quasi-isotropic laminate for the entire dome cap. The nextfabrication step for the test article involved placing reinforcingplies along the flange to dome interface. Details of this reinforce-ment design will be described in the analysis section. A typical350 °F (176.7 °C) curing cycle was employed to cure the test ar-ticle. The final step in the fabrication of the test article involveddrilling a hole pattern along the circumference of the flange. Thishole pattern matches that of the upper and lower domes (contain-ment shells) of the CTF, which are bolted together to perform thetesting. The intact test article will be used to perform the firstphase of the testing for material permeability. After the firstphase of testing is complete, the dome will be modified to includean access hole in the center of the dome, which is meant to simu-late a typical tank sump or vent opening. A polar boss sealing sys-tem, shown in Fig. 3 and described in a later section, will be usedto seal the access hole in preparation of the second phase of thetest series.2

CL

spring-typedevicesK2K1primary gasketPolar Boss AssemblyFigure 3: Polar Boss Assembly and Cross-Section.Polar Boss Cross–SectionMaterial permeability and leak testing of the compositedome test article and polar boss assembly is scheduled formid-1999 in the CTF. The CTF is capable of testing tank domeswith cryogenic liquids at pressures of up to 150 psi (1.03 MPa –megapascal). The major component of the CTF, the CTF assem-bly shown in Fig. 2, is supported on a five foot (1524 mm) highby ten foot (3048 mm) wide and ten foot (3048 mm) long I-beamframe that suspends the entire CTF assembly for easy access. Thetest article is sandwiched between two containment shells, creat-ing two volumes. The pressurized fluid is contained in one vol-ume, while the amount of fluid that diffuses through the test ar-ticle wall over a period of time is measured in the other volume.A schematic of the cryostat test setup and facility piping is shownin Fig. 4. The test series for the composite dome test article con-

sists of two parts. Phase I testing will verify the test setup at theCTF and provide baseline material permeability data for an intacttest article. However, since pressure vessels include openings foraccess, the permeability characterization testing is also per-formed on a test article with a polar boss system installed. There-fore, after the testing on the intact test article is performed, anopening for the polar boss will be machined. Phase II will investi-gate the attachment and resulting permeability/leakage for thepolar boss system. Therefore, the analysis and design are carriedout for the intact dome first; then, the perforated dome is consid-ered. In either case, an example of a typical test sequence for theentire test series is now given. A test sequence consists of fivesteps in pressure from fifteen psi (0.10 MPa) to seventy-five psi

HOV = hand operated valvetest articlecontainment shellResidual GasAnalyzer(RGA) unitHOVHOVHOVHOVpressuretransducerpurge gas inletrelief valve, 15 psi (0.10 MPa)drainhand operated reliefpressure transducerinstrumentation feed throughGas Hepanel

RGA sample line150 psi(1.03 MPa)CLFigure 4: Cryostat Test Setup and Facility Piping.3

(0.52 MPa), with the temperature being cycled by allowing thetest article to warm to room temperature following each pressurestep. First, a vacuum is pulled on the volume between the test ar-ticle and lower containment shell to allow sampling of that vol-ume with a residual gas analyzer. The residual gas analyzer mon-itors gases permeating or leaking through the test article. Then,the volume between the upper containment shell and the test ar-ticle is filled with a liquid cryogenic fluid. Finally, the test articleis pressurized to an amount between fifteen psi (0.10 MPa) andseventy-five psi (0.52 MPa) based on the current load step of thetest sequence. The pressure is held at the given level until asteady-state measurement of the permeability is taken. Once themeasurement is complete, the pressure is released and the tankdrained to allow the tank to warm to ambient conditions in prepa-ration for the next cycle.DISCUSSION OF FAILURE MODESWe considered four possible failure modes in the design ofthe dome: (1) leakage at the joint of the cryostat assembly, Fig.1; (2) the failure of the composite dome under internal pressure;(3) buckling of the dome during the line checkout condition; and(4) buckling of the dome during transportation. The first failuremode is addressed by the double o-ring gasket system, Fig. 1, De-tail D, and is not covered in this study. The remaining failuremodes are caused by three different loading conditions that willbe addressed in this study. Since the test article is tested beforeand after the polar boss is installed, the remaining failure modesmust be checked for the intact dome as well as after it has beenperforated and the polar boss installed.Intact Dome CaseFor the intact dome, two possible loading conditions are con-sidered, pressurization of the cryogenic fluid and a loading thatmay cause buckling of the dome. The loading condition associat-ed with failure mode 2 is the maximum nominal internal pressureloading placed on the dome during the test sequence to determinematerial permeability, which is 75 psi (0.52 MPa). The loadingcondition associated with failure mode 3 occurs when the volumebetween the test article and the lower containment shell becomespressurized, Fig. 4. The purpose of this pressurization is to testthe cryostat facility lines for leaks. This loading condition ismodeled with an external pressure of 15 psi (0.10 MPa), and wewill refer to it as the “checkout condition”.Perforated Dome with Polar Boss CaseThe aforementioned two loading conditions for the intactdome also apply to this case. Also, the dome may experience athird loading condition during transportation, if the dome is car-ried on its flanged surface. This loading is due to the dead weightof the polar boss shown in Fig. 3; however, no pressure acts onthe dome for this condition. Thus, each of these three loading

conditions must be considered in the design of the perforateddome.

ANALYSIS AND DESIGN OF COMPOSITE DOMEThe material properties used in the analysis were determinedat a temperature of –65 °F (–54 °C), which is much higher thanthe cryogenic operating temperatures. However, elastic proper-ties, such as strength and stiffness, generally increase as tempera-ture decreases to cryogenic conditions (Barron, 1985). For thisinvestigation, we use lamina elastic properties provided by themanufacturer for the dome design. These properties were usedto obtain the effective isotropic properties listed in Table I. Lami-nation theory can be used to derive these effective isotropic prop-erties (Agarwal and Broutman, 1990).The only design variables needed to complete the design ofthe test article are the dome thickness and composite materiallay-up. The thickness for the dome and the material lay-up aredetermined using the analysis procedures presented in the fol-lowing sections (unlike isotropic designs, both variables are partof the design process). Even though there is an infinite numberof lay-up combinations, we limit this design to a symmetric qua-si-isotropic lay-up, (Agarwal and Broutman, 1990). This simpli-fies the analysis for the preliminary design. Since the intact domeis tested first and then perforated to install the polar boss system,the design of the perforated dome entails determining a secon-dary lamination system to reinforce the hole to attenuate thestress concentrations caused by the geometric discontinuity.That is, the design of the intact dome, thickness and lay-up, isused in the design of the perforated dome. In the next two sec-tions we describe the details of the design for the dome and thereinforcement.Intact DomeFor the preliminary design of the dome, we considered thefailure of the dome under internal pressure, loading condition as-sociated with failure mode 2. The analysis for this preliminarydesign of the dome was completed using membrane shell analy-sis. The stresses in the dome are uniform and are given by the fol-lowing equation:sq+sf+PR+s,

2h(1)

where, P is the internal pressure, R is the dome radius, and h isTable I: Effective Isotropic Properties for IM7/8552[0°ń90°,\"45°]3S Lay-up.

Effective ModulusE, ksi (GPa)10,500 (68.9)Poisson’s Ratio,ν0.363 Strengthσu, κsi (MPa)80 (551)4

the dome thickness. The subscripts q and f represent the spheri-cal coordinates of the dome. In this study, the maximum distor-tion energy theory or von Mises Theory is used to check for theonset of failure in the dome. Primarily, the maximum distortionenergy failure criterion is used to predict failure in ductile materi-als. In our case, this theory was used mainly as a method forcomparing the quasi-isotropic composite dome with other knownisotropic candidate domes. Those comparisons are not docu-mented in this report. According to the maximum distortion en-ergy theory, failure by yielding occurs when the distortion strainenergy per unit volume for combined stress is equal to the maxi-mum elastic distortion energy per unit volume in simple tension(Ugural and Fenster, 1995). The von Mises stress, a stress equiva-lent quantity, can be utilized to predict the onset of failure. There-fore, the von Mises stress can be computed using the stresses giv-en by Eq. (1), since these are equal to the principal stresses. Todetermine dome failure, the maximum von Mises stress iscompared with the strength of the composite. During the prelimi-nary design, we considered three different configurations or lay-ups, [0°,\"30°]4S, [0°,\"60°]4S, and [0°ń90°,\"45°]3S. Thecapacity provided by each lay-up was assessed by comparing itscorresponding strength with the von Mises stress. The final de-sign is based on a [0°ń90°,\"45°]3S lay-up of HerculesIM7/8552 (carbon/epoxy) prepreg fabric, which has a strength of80,000 psi (551 MPa), Table I. The results from this analysis arelisted in the first column of Table II. This twelve ply lay-up re-sults in a nominal dome thickness of 0.12 inches (3 mm). Thisis a quasi-isotropic lay-up which possesses isotropic in-planestiffness properties, i.e., the extensional stiffness matrix, [A], isinvariant with respect to in-plane rotations of the coordinate sys-tem. The bending properties, however, are still orthotropic. Nev-

ertheless, the dome can be analyzed using the effective isotropicmaterial properties without compromising practical accuracy.The material properties used in the analysis are given in Table I.These properties were derived from experimental data providedby Hercules, the material fabricator.

In addition to membrane stresses, the dome will experienceflexural stresses and stress concentrations due to the geometry ofthe test article and configuration of the test setup. Thus, a finiteelement analysis was conducted using MSC/NASTRAN to deter-mine the bending stresses, the stress around the geometric dis-continuities, the overall deformations, and to assess the overallstrength. The flange portion of the dome is partially unsupported,as shown on the cryostat assembly sketch in Fig. 1. This causesthe radius of the flange-dome intersection to straighten uponloading, causing further stressing of the material. For this portionof the dome and other regions where stress concentrations causeexcessive deformation and overstressing, instead of increasingthe thickness of the entire dome, local reinforcement was de-signed to reduce these localized stresses. The finite elementmesh for the dome model is depicted in Fig. 5 and consists of 437CTRIA3 three node shell elements and 13,547 CQUAD4 fournode shell elements (32 elements in the meridional direction and437 elements in the circumferential direction). The loading forthis case is the internal pressure loading associated with failuremode 2. The von Mises stress distribution in the dome is shownin Fig. 6. This figure clearly shows the stress concentration at theflange-dome intersection, and that this stress concentration islocalized near the geometric discontinuity. This over-stressedpart was reinforced with additional layers of carbon/epoxy cloth,Table II: Results Summary for Intact Dome.No Flange ReinforcementMembraneanalysis

sf+sq,psi(MPa)

åf+åqw,in(mm)vonMises,psi(MPa)

25,081 (172.9)1.52E–030.1221 (3.1)25,081 (172.9)Finite element analysisNo bending *25,083 (172.9)1.52E–030.1697 (4.3)25,027 (172.5)Maximum

––0.199 (5.0),500 (444.7)Reinforced Flange

Finite element analysisNo bending *

––0.160 (4.1)25,000 (172.4)Maximum

––0.160 (4.1)43,400 (299.2)Buckling Analysis for Check-out ConditionAnalytical

“Linear”

lcr

* the region of the dome where no flexure occurs

Non-linear

1.4

Finite element analysis1.9605

1.939

5

Figure 5: Dome Finite Element Mesh.see Fig. 7. The lay-up for the reinforcement is similar to that forthe dome, i.e., [0°ń90°,\"45°]. A summary of the results for thefinite element analyses of the unreinforced and reinforced intactdome, is reported in Table II. This table shows that the membraneanalysis results are in good agreement with the results of the finiteelement analysis for the portion of the dome where no bendingoccurs, the “no bending” results column in Table II. Although the.5 ksi (445 MPa) maximum stress for the unreinforced domedoes not exceed the 80 ksi (551 MPa) strength allowable, we de-cided to reinforce the dome to attenuate the excessive deforma-tion and stresses as well as to obtain a factor of safety against fail-ure of approximately two. Therefore, the results for thereinforced dome in Table II provide a factor of safety of nearlytwo, based on the maximum von Mises stress for the reinforced

flange of 43.4 ksi (299 MPa) and the 80 ksi (551 MPa) strength.The other possible loading for the intact dome, the loadingcondition associated with failure mode 3, is the external pressureof 15 psi (0.10 MPa) applied during the CTF line checkout proce-dure. To assess the stability (buckling) of the dome under thisloading, analytical and finite element analyses were conductedwith the results listed on the bottom of Table II. These results arepresented in terms of critical eigenvalues. The critical externalpressure can be determined by multiplying the critical eigenvalueby the external pressure, i.e., lcr+PcrńPappl. The ”linear” ana-lytical solution was obtained using the following equation(Harvey, 1991):vonĂMisesmin+0Ăksi

.vonĂMisesvonĂMisesmaxYZ

X

Figure 6: von Mises Stress for Intact Dome.vonĂMisesmax+.5Ăksi(445ĂMPa)1.000.930.870.800.730.670.600.530.470.400.330.270.200.130.070.006

5 in. (127 mm)4 in. (102 mm)[0°ń90°,\"45°]2

0.12 in. (3 mm)5 in. (127 mm)[0°ń90°,\"45°]6 in. (152 mm)7 in. (178 mm)5 in. (127 mm)[0°ń90°,\"45°]3S[0°ń90°,\"45°]5 in. (127 mm)

7 in. (178 mm)CLFigure 7: Dome and Reinforcement Lay-up.h,2Elcr+21ń2RP[3(1*n)]

ǒǓ2

(2)

tion of stress in the region of the circular hole can be computed

using the following equation (Harvey, 1991):sc+s1)a2

r

where, E is Young’s modulus and ν is Poisson’s ratio, the otherquantities were defined after Eq. (1). This solution does not ac-count for prebuckling nonlinearity or deformation; therefore, wecategorize the analysis as ”linear”. Although Eq. (2) was derivedfor a clamped shallow spherical shell, the results seem to be ingood agreement with those obtained from the linear static buck-ling finite element analysis. The prebuckling nonlinearity or de-formation was not taken into account in the finite element analy-sis either. The nonlinear buckling analytical solution wasobtained from Fig. 37 in Bushnell (19). This figure in Bushnelldepicts a plot for the stability of a shallow spherical dome gener-ated using nonlinear numerical analysis and experimental testdata. Based on a cap shallowness parameter,2[3(1*n2]1ń4(Hńh)1ń2, of approximately 17 for our compositedome test article, the critical pressure for the nonlinear case is de-termined using the test data on the plot, i.e.Pcrń1.2E(hńR)2+0.74. Therefore, using the experimental re-sults to determine the critical eigenvalue for the nonlinear case,the shell is considered safe against buckling during the linecheckout condition, i.e., PcrńPapplY1.4.Perforated Dome with Polar BossIn order to access the interior of a pressure vessel, an openingis required. Therefore, a circular opening will be machined at thepolar region of the intact dome described in the last section. Aboss or a stub flange is used to connect the pressure vessel to theline that carries the fluid to and from the pressure vessel. In com-posite pressure vessels, this boss is typically metallic, see Fig. 3.The geometric discontinuity that the hole presents causes a stressconcentration near the opening. As shown in Fig. 6, the stressstate in the polar region of the intact dome is uniform, and the twoprincipal stresses are equal. Based on this observation, the varia-

ǒ

2

Ǔ

(3)

where, σ is the membrane stress,a is the radius of the circular po-lar opening, and r is the meridional distance from the center ofthe hole. Notice that the stress decreases rapidly as the distancefrom the edge of the hole increases. At the edge of the hole, r =a and the maximum stress is 2σ, Eq. (3). At a distance from theedge of the hole equal to the radius, r = 2a, the stress has fallento 1.25σ. The von Mises stress at the edge of the hole can easilybe determined since one principal stress is zero and the other isgiven by Eq. (3); the results are listed in Table III. The reinforce-ment around the circular hole is designed to reduce this stressconcentration. The reinforcement lay-up is depicted in Fig. 7.This lay-up eliminates the symmetry of the overall lay-up; how-ever, the bending-extension coupling effect should be small sincethe polar boss clamps this region.We again conducted finite element analyses of the dome us-ing MSC/NASTRAN, with and without the reinforcementaround the circular opening. The mesh for this analysis consistedof 6,976 CQUAD4 four node shell elements (32 elements in themeridional direction and 218 elements in the circumferential di-rection). The results are listed in Table III. Figure 8 shows thedistribution of the von Mises stress. In this part of the design, wealso had to take the weight of the polar boss into account. Theweight of the polar boss was estimated at 40 pounds (18.14 Kg)from a previous design. This weight was applied as a ring load,equivalent to 1.28 pounds per inch (0.22 N/mm) at the edge of thehole. Also, the boundary conditions at the edge of the hole arenot well defined. We considered two cases, shown in Fig. 9. Inthe first case, the polar boss provides complete clamping of theedge of the hole, allowing only vertical displacements of theedge. For the second case, the edge of the hole is allowed to slide7

Table III: Results Summary for Dome with Circular Opening.No Circular Opening Reinforcement

Finite–Element AnalysisAnalytical(Harvey)

w,in.(mm)

–

Clamped hole edgeNo Bending* 0.150 (3.8)Maximum 0.151 (3.8)Semi–Free hole edgeNo Bending*

Maximum

0.150 (3.8) 0.467 (11.9)vonMises,psi(MPa) 50,162 (345.9) 27,000 (186.2) 43,400 (299.2) 27,000 (186.2) 98,400 (678.4)Reinforced Circular Opening (finite–element analysis)

Clamped hole edgeNo bending*

w,in(mm)

0.160 (4.1)

Maximum0.160 (4.1)43,200 (297.9)Semi–Free hole edgeNo bending*0.160 (4.1)27,000 (186.2)Maximum0.270 (6.9)62,400 (430.2)vonMises,psi(MPa) 27,000 (186.2)Buckling Analysis for Check-out Condition (finite–element analysis)

Pcr

* the region of the dome where no flexure occurs

29.27 psi (0.202 MPa)freely between the clamping system of the polar boss, restrainingonly rotations but not displacements. We believe the true condi-tion is somewhere in between, but closer to the clamped hole con-dition. However, from Table III, we can see that even the mostconservative case results in a maximum possible stress for the re-inforced hole of 62.4 ksi (430 MPa) which does not exceed thestrength of the composite, 80 ksi (551 MPa).

The line checkout condition for this part of the investigationis also safe against buckling, as shown at the bottom of Table III,lcr+PcrńPappl+1.9514. This result was obtained using finiteelement analysis for the perforated dome with the clampedboundary condition, loaded with the external pressure of 15 psi(0.10 MPa) applied during the CTF line checkout procedure.vonĂMisesmax+43.2Ăksi

(298ĂMPa)

vonĂMisesvonĂMisesmaxvonĂMisesmin+0Ăksi

YZ

X

Figure 8: von Mises Stress for Dome with Circular Opening Clamped.1.000.930.870.800.730.670.600.530.470.400.330.270.200.130.070.008

CLCL

no rotation and no radialdisplacement, onlyvertical displacementClamped Hole Edgeno rotation, onlyradial and verticaldisplacementFree Hole EdgeFigure 9: Boundary Conditions for Finite Element Analysis of Perforated Dome.For this part of the investigation, we also had to consider anadditional loading condition, the weight of the polar boss whenthe dome is transported upside down. The weight of the polarboss is 40 pounds (18.14 Kg), which is equivalent to a 1.28pounds per inch (0.22 N/mm) ring load on the edge of the hole.The resulting stresses determined using finite element analysisare negligible. Therefore, this loading is negligible and does notcause any detrimental instability or stress concentrations.DESCRIPTION OF POLAR BOSS SYSTEMAccess systems are required in all pressure vessels; a polarboss system is employed in the test article presented in this paper.A requirement of all access systems is to contain the fluid withminimum leakage, since all access systems leak at some level.To minimize leakage, this type of polar boss system is oftensealed with two o-ring gaskets, i.e., a second gasket is placed ad-jacent to the primary sealing gasket, see Fig. 3. It is difficult fora flat rigid polar boss assembly to conform to the curvature of thedome. In our case, a flat rigid polar boss assembly does not com-press the two sealing gaskets uniformly or with adequate pres-sure. To obtain the same compression on both gaskets, additionalforce must be applied on the part of the dome near the primarygasket because this gasket is located further from the center lineof the boss assembly. To solve this problem, we developed an in-novative design for the polar boss, Fig. 3. The improved polarboss design has a flexible washer that conforms to the curvatureof the dome, applying uniform compression to both gaskets. Thedesign consists of two interlocking rings that sandwich twospring-type components. These spring-type components areplaced at the same distance from the center line of the boss assem-bly as the o-ring gaskets. A unique feature is that each of thespring-type devices may have a different stiffness, which allowseach spring-type device to transfer a different amount of force.The different spring stiffnesses can be chosen such that their re-sulting forces compress the two o-ring gaskets with the sameamount of force, thus making the sealing system more effective.CONCLUDING REMARKSIn this paper, a method is presented for designing compositepressure vessels with a combination of membrane and finite ele-

ment analyses. The membrane analysis can be used as a prelimi-nary design tool; then, parts of the system where bending andstress concentrations are possible can be investigated using a fi-nite element analysis. We made use of this analysis tool com-bination in the design of a CFRP dome cap and flange for materialpermeability characterization. In regions where stress con-centrations or gradients occur, rather than increasing the thick-ness of the entire dome, additional layers of carbon/epoxy clothare used to attenuate these stress concentrations. Thus, a com-plete numerical investigation of the critical load cases for theCFRP dome test article have provided analytical results whichshow the test article remains below required stress levels.ACKNOWLEDGEMENTS

The authors would like to thank the NASA/MSFC Center Di-rector’s Discretionary Fund office and the NASA/ASEE SummerFaculty Fellowship program for making this research possible.We would like to thank Bill McMahon, Seth Lawson, and person-nel of EH35/MSFC and the Thiokol Corporation for the fabrica-tion of the test article. Thanks to Kathryn Horton and Scott Lauf-fer of ED51/MSFC for providing essential engineering drawingsupport and Russ Abrams and Mat Bevill of EP91/MSFC for testconfiguration drawings.REFERENCES

Agarwal, B. D. and Broutman, L. J., 1990, ”Analysis and Perfor-mance of Fiber Composites,” 2nd edition, John Wiley & Sons,Inc., New York.

Barron, R. F., 1985, ”Cryogenic Systems,” Oxford UniversityPress, New York.

Bushnell, D., 19, ”Computerized Buckling Analysis ofShells,” Kluwer Academic Publisher, Boston.Harvey, J. F., 1991, ”Theory and Design of Pressure Vessels,”Chapman & Hall, New York.

Ugural, A.C. and Fenster, S.K., 1995, “Advanced Strength andApplied Elasticity”, 3rd edition Prentice Hall, Upper Saddle Riv-er, New Jersey.

9