ANALYSIS AND TESTING OF AN INTEGRATED REFRIGERATION AND STORAGE SYSTEM FOR LIQUID HYDROGEN ZERO BOIL-OFF, LIQUEFACTION, AND DENSIFICATION

Transcription

1 ANALYSIS AND TESTING OF AN INTEGRATED REFRIGERATION AND STORAGE SYSTEM FOR LIQUID HYDROGEN ZERO BOIL-OFF, LIQUEFACTION, AND DENSIFICATION By WILLIAM USILTON NOTARDONATO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 26

2 Copyight 26 by William Usilton Notadonato

3 To the men and women of NASA KSC and thei opeations contactos fo thei dedication and commitment in pefoming an often-unappeciated ole of pepaing and launching spacecaft.

4 ACKNOWLEDGMENTS I have a numbe of people to thank fom NASA Kennedy Space Cente, including Roy Bidges, who set up the Kennedy Gaduate Student Fellowship, Joe Pota, my boss, who let me take the time to do the eseach wok; and a numbe of KSC cyogenic enginees who lent thei expetise, including but not limited to Robet Johnson, James Fesmie, Zoltan Nagy, Cal Exline, Diane Stees, and Maia Littlefield. In addition, a numbe of industy souces, including people at PHPK, Paxai, Siea Lobo, Ai Liquide, and NIST have contibuted thei expetise. I am also paticulaly indebted to D. Jong Baik of the Floida Sola Enegy Cente, my patne, who did the laboatoy set up and much of the testing, and D. Glen McIntosh of CTS, fo significantly contibuting to the design and fabication of the cyostat. I expess my sincee gatitude to all my committee membes fo ageeing to take the time to help my wok and eview this document. Specifically, I wish to thank D. Peteson and D. Chung fo all of thei help and guidance while I was evising this dissetation, D. Chow fo a numbe of discussions and pojects woked between KSC and UCF, and D. Sullivan and D. Ingley fo ageeing to step in late and take ove fo othe depated membes. I also would like to thank all the pofessos I had at UF fo ekindling my desie fo leaning about advanced engineeing concepts. Finally, and most impotantly, I acknowledge and thank D. Sheif fo his time, patience, and guidance ove the past five yeas. iv

5 This wok would not be possible without the suppot of my immediate family, and fo this I am vey gateful. I thank my paents May-Lou and Joe fo all thei suppot ove the yeas and fo stessing the value of a good education. I also ecognie Loetta Paenteau fo he suppot and help with my family duing the yea I was in Gainesville. Finally, and most impotantly, I thank my wife Celeste fo all of he patience, undestanding, suppot, paise, motivation, and help ove the past five yeas. I could not have done it without he. v

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS... iv LIST OF TABLES... ix LIST OF FIGURES...x ABSTRACT... xiii CHAPTER 1 CURRENT STATE OF THE ART IN HYDROGEN LIQUEFACTION, ZERO BOIL-OFF, AND DENSIFICATION...1 Hydogen Liquefaction...2 Linde-Hampson Cycle...6 Pe-Cooled Linde-Hampson Cycle...8 Linde Dual Pessue Cycle...11 Claude Cycle...12 Lage Scale Hydogen Liquefaction Plants...14 Zeo Boil Off...15 Gound Applications...15 Space Applications...19 Hydogen Densification...21 Launch Vehicle Opeations...25 Summay PROPOSED INTEGRATED REFRIGERATION AND STORAGE SYSTEM...28 Integated Refigeation and Stoage Concept...28 System Behavio...31 Behavio Issues ANALYSIS OF GOVERNING EQUATIONS...4 Themodynamic Analysis of Liquefaction Cycle...41 Cycle Desciption...41 Model Development...44 Analysis...46 vi

7 Themodynamic Analysis of Stoage System...55 Definitions and Assumptions...55 Consevation of Mass...58 Consevation of Enegy...59 Equation of State...62 Opeational Simplifications...62 Closed Stoage With Heat Tansfe To the System (Self Pessuiation)...63 Open Stoage With Heat Tansfe To The System (Boil off)...69 Closed Stoage With Zeo Net Heat Tansfe (Zeo Boil Off)...72 Open System With Heat Tansfe Out Of The System (Liquefaction)...73 Closed Stoage With Heat Tansfe Out Of The System (Densification)...76 Coected model...79 Stoage Fluid Analysis...81 Consevation of Mass...89 Consevation of Momentum...9 Consevation of Enegy...94 Summay...95 Solution Pocedue...96 Bounday Conditions...97 Dimensionless analysis HYDROGEN SAFETY...15 Hydogen Popeties...15 Combustion Haads...18 Hydogen Embittlement Cyogenic Haads Facility Design Management and Opeations Conclusion TEST SYSTEM DESIGN AND ANALYSIS Peliminay Design and Analysis Hydogen Stoage...12 Cyogenic Refigeato...12 Instumentation and Data Acquisition Fluid Distibution Vacuum System Safety and Leak Detection System Cyostat Specification Cyostat Detailed Design Cyostat Constuction Cyogenic Refigeation Acceptance Testing Suppot Systems vii

8 6 LABORATORY TESTING AND DATA ANALYSIS Liquid Nitogen Chilldown LN 2 Dain and Puge Hydogen Liquefaction Hydogen Densification Zeo Boil Off Measuement Uncetainty Analysis CONCLUSIONS AND RECOMMENDATIONS APPENDIX A CRYOSTAT SYSTEM SPECIFICATION Configuation and Pefomance Requiements Design, Fabication, Inspection and Testing Requiements Submittals Shipment B C HEAT LEAK CALCULATION DENSIFICATION SAMPLE MODEL D ACRONYMS AND SYMBOLS LIST OF REFERENCES BIOGRAPHICAL SKETCH viii

9 LIST OF TABLES Table page 1-1 Ideal wok of liquefaction fo cyogenic fluids Ideal wok input and density fo liquefaction vs. pessuiation of hydogen Opeational scenaios...63 ix

10 LIST OF FIGURES Figue page 1-1 Ideal liquefaction pocess T-s diagam and system schematic Linde Hampson schematic Linde Hampson T-s diagam Pe-cooled Linde Hampson schematic Pe-cooled Linde Hampson T-s diagam Cascade cycle schematic Linde dual pessue cycle schematic Linde dual pessue cycle T-s diagam Claude cycle schematic Claude cycle T-s diagam Zeo loss stoage economic analysis Satuated liquid hydogen density, enthalpy, and vapo pessue Poposed Liquefaction Cycle Schematic Acceptable Intemediate Tempeatues as a Function of Helium Compession Compession Wok vs Intemediate Tempeatue (P2=12 kpa) Compession Wok vs Intemediate Tempeatue (P2=12 kpa) Tubine Mass Flow Rates and Cycle Wok Heat Rejection at HX3 and HX Cycle Wok vs Tempeatue fo a Range of Hydogen Pessues Cycle Wok vs Hydogen Pessue...54 x

11 3-9 Cycle Wok vs Helium Pessue Simplified system epesentation Self pessuiation tempeatue and pessue vs. time Self pessuiation tank quality and phase volume vs. time Self pessuiation ates fo vaiable liquid level Self pessuiation (model vs. data) Liquid and vapo mass duing boil off Boil off time dependence on pessue Cyocoole pefomance and heat leak vs. tempeatue Compaison of pedicted vs. actual liquefaction ate Pedicted densification tempeatue and pessue Densification ates fo vaiable liquid levels Pedicted vs. actual densification ates Coected model vs. densification data Coected model vs. pessuiation data Poposed system epesentation Pedicted Fee Convection Velocity Pedicted Fee Convection Reynolds Numbe Pedicted Gashof Numbe Hydogen T-s diagam Cyostat design detail Cyocoole AL33 capacity cuve Liquid hydogen cyostat test aangement Liquid nitogen chilldown Hydogen dewa ovepessuiation xi

12 6-3 Liquid nitogen chilldown detail Completion of chilldown Liquid nitogen dain and puge Tank dain and puge pessue cycles Tank dain and puge tempeatue pofile Cyocoole chilldown in vacuum Hydogen liquefaction on 6/24/ Tempeatue tansients duing initial gas flow Mass flow ate and pessue duing liquefaction Total hydogen volume in tank Heat pipe satuation Ovenight densification System elocation wam up Continuing liquefaction Liquid tempeatue and pessue satuation cuve Ullage collapse Liquefaction on Liquefaction on Zeo boil off opeations Non-dimensional themal statification duing self pessuiation ZBO opeations with P and T eo bas Expanded section plots of P and T with eo bas A-1 System Specification Schematic...17 xii

13 Abstact of Dissetation Pesented to the Gaduate School of the Univesity of Floida in Patial Fulfillment of the Requiements fo the Degee of Docto of Philosophy ANALYSIS AND TESTING OF AN INTEGRATED REFRIGERATION AND STORAGE SYSTEM FOR LIQUID HYDROGEN ZERO BOIL-OFF, LIQUEFACTION, AND DENSIFICATION By William Usilton Notadonato Decembe 26 Chai: S. A. Sheif Majo Depatment: Mechanical and Aeospace Engineeing While hydogen was fist liquefied in the laboatoy by Si James Dewa in 1898, it was not until the beginning of the space age in the 195s that lage-scale poduction of liquid hydogen was common. Since then, the methods used by NASA to poduce, liquefy, stoe and distibute hydogen fo launch vehicle applications have not changed. Specifically, gaseous hydogen is poduced fom natual gas, liquefied in lage scale plants by pefoming extenal wok on the gas and then expanding it, tanspoted to the launch site via tanke tucks and stoed in lage gound tanks in the satuated state until it is loaded into the vehicle duing launch countdown. Duing the pocess, heat leak and tank pessuiation ceate boil-off losses, and chill-down of the tanspot lines cause moe poduct losses. Othe handling issues, including low liquid density, lage themal tansients, leakage and safety concens, and two phase flow poblems have given liquid hydogen a eputation fo being a difficult fluid to stoe and contol. xiii

14 This dissetation poposes a novel method of liquefying and stoing hydogen by incopoating a closed-cycle helium efigeato into the stoage tank. Thee ae numeous advantages to this system. Localied poduction and liquefaction eliminates the need fo tanspotation of haadous liquid hydogen, minimies heat flow into the system, and educes the numbe of pesonnel equied at the launch site. Pope design of the efigeato also allows fo densification of the liquid, inceasing the amount of popellant loaded into the flight tank. In addition, subcooling below the nomal boiling point allows the liquid to stoe moe efigeation enegy, leading to less boil off losses o eliminating boil off completely, and possible allowing fo ecovey of chill down losses. Subcooled popellants also povide geate themal magin befoe onset of evapoation and two-phase flow. Details of these pefomance and economic benefits ae povided in Chapte 2. While thee ae benefits, efigeated and subcooled cyogens behave in a diffeent manne than satuated liquids. These behavio issues must be investigated befoe lage-scale incopoation in futue launch systems. The consevation equations have been pesented, and simplification of these equations in a 2-dimensional tansient mass and enegy model has been developed. Chapte 3 pesents some esults of the pedicted behavio. A small testbed has also been poposed, designed, and fabicated fo expeimental validation of this model, and initial testing has occued to validate the poposed concepts of liquefaction, eo boil off and densification. Results of this initial ound of tests ae povided in Chapte 6. xiv

15 CHAPTER 1 CURRENT STATE OF THE ART IN HYDROGEN LIQUEFACTION, ZERO BOIL-OFF, AND DENSIFICATION Hydogen has many favoable popeties that make it attactive as a seconday enegy caie. Hydogen is vey abundant. It is believed to make up 9% of the mass in the univese, and is the ninth most common chemical element on Eath. Combustion of hydogen is clean, with the majo combustion poducts being wate and heat. (1) With a highe heating value (HHV) of 142, J/g, hydogen caies moe enegy pe unit mass than othe fuels. (2) This is paticulaly useful fo space applications as liquid hydogen/liquid oxygen ockets have the highest specific impulse of any combination of chemical popellant cuently in use. (3) Fo this eason, NASA and the US Ai Foce (USAF) have been inteested in hydogen as a popellant fo space vehicles since the late 195s. Moe ecently, thee is much inteest in using hydogen as an enegy caie in tanspotation applications, including automobiles, busses, and aicaft. (4-5) Use of hydogen does have some negative aspects howeve. Hydogen gas has vey low mass density, so volumetic concens limit it use to small applications. Hydogen can be stoed as a gas in a solid matix such as metal hydides, but these systems ae not mass efficient fo space use. The most pactical stoage method fo hydogen fuels has been as a low tempeatue liquid. Hydogen has a nomal boiling point (NBP) of 2.27 K, and the density at this point is 7.79 kg/m 3. Its citical point has a pessue of 1315 kpa and tempeatue of 33.2 K, and the tiple point has a pessue of 7.2 kpa and a tempeatue of 13.9 K. (6) While NASA and USAF have pioneeed methods 1

16 2 fo lage scale poduction, stoage, and distibution of liquid hydogen, thee ae issues that must still be addessed befoe liquid hydogen gains moe widespead use as a pactical enegy caie. Hydogen equies lage quantities of enegy to liquefy, and since it is stoed fa below ambient tempeatue, heat leaks into the stoage vessels ceating evapoation and poduct losses. In addition, although liquid hydogen density is 789 times geate than gaseous hydogen, it is still a vey low-density fuel and equies lage, well-insulated stoage vessels. These issues will be addessed in the pesent chapte. Hydogen Liquefaction One facto delaying development of widespead use of hydogen in the economy is the difficulty of stoing and distibuting lage quantities. Gaseous stoage equies extemely lage volumes and stoage in a metal hydide equies heavy stoage tanks and addition of enegy to dive the hydogen out of the stoage state. Liquid hydogen has the potential to eliminate these concens, but the pocess of liquefying hydogen is enegy intensive. Many diffeent cycles have been poposed and used fo hydogen liquefaction, anging fom small-scale laboatoy use to lage liquefaction plants capable of poducing 6 tons pe day of liquid hydogen. (7) A challenge associated with liquid poduction and stoage is the need fo development of small- to medium-scale distibuted liquefaction systems that have efficiencies of the same ode of magnitude as the lage-scale liquefaction plants. This allows fo localied poduction of hydogen gas optimied fo the specific location coupled with localied efficient liquefaction fo highe enegy stoage densities. (8) Genealied desciptions of pocedues fo hydogen liquefaction exist in cyogenic technology liteatue. (1,7,9) These methods all ely on taking a puified gas at oom

17 3 tempeatue and cooling it to the extemely low tempeatue of 2.3 K, the nomal boiling point of hydogen. In ode to compae these systems, the themodynamically ideal pocess fist must be identified. The optimum cycle fom the themodynamic pespective is the Canot Cycle, which consists of two evesible isothemal pocesses and two evesible adiabatic pocesses. This cycle would be the ideal efigeation cycle, howeve, the liquefaction pocess is not a closed cycle but an open cycle whee mass is continuously liquefied at the cold end and must be e-supplied at the wam end. Theefoe the ideal liquefaction pocess can be taken as the fist two steps of the Canot cycle, namely a evesible isothemal compession of the gas to a suitably high pessue, followed by an isentopic expansion whee the gas completely condenses in the expande. Refe to Figue 1-1 fo details on this pocess depicted on a T-s diagam. (9) Howeve, the high pessue equied fo complete liquefaction of hydogen, 1 5 ba, makes this pocess technically not feasible. The minimum wok equied fo this ideal liquefie can be detemined and used as a benchmak fo compaison with eal pocesses. m& f COMPRESSOR 1 2 Q & 2 1 T W & c p=const W & e EXPANDER f f m& f f LIQUID s RESERVOIR Figue 1-1 Ideal liquefaction pocess T-s diagam and system schematic

18 4 The Fist Law of Themodynamics fo steady flow with one inlet and one outlet, neglecting kinetic and potential enegy change between the initial and final state, can be witten as Q& W& = m& ( h 3 h1) 1-1 Fom the Second Law of Themodynamics fo an ideal evesible pocess, an entopy balance on the heat exchange pocess gives Q & = m& T ds o Q & = mt & 1( s2 s1) 1-2 Combining equations 1-1 and 1-2 and knowing s 2 is equal to s 3 we find the minimum wok equied fo liquefaction can be expessed as W& m & = T1 ( s1 s3 ) ( h1 h3 ) 1-3 Fom equation 1-3 we find the minimum wok equied to liquefy a gas (assuming the initial and final pessue ae equal) depends on the initial state of the gas and the type of gas to be liquefied. Some common cyogenic fluids ae listed in Table 1-1 along with thei NBP and thei ideal wok of liquefaction. Note that hydogen equies the highest wok input fo liquefaction, moe than even liquid helium, which has the lowest NBP of any fluid. Table 1-1 Ideal wok of liquefaction fo cyogenic fluids Gas Nomal Boiling Point (K) Ideal Liquefaction Wok (J/g) Helium Helium Hydogen ,19 Neon Nitogen Agon Oxygen Methane

19 5 As an inteesting side note egading hydogen stoage fo automotive use, US manufactues tend to favo high-pessue hydogen gas stoage ove liquid stoage, which diffes fom some Euopean auto manufactues. Stoage pessues ange fom the pesent 5-psi systems to poposed 1-psi systems. (1) Reasons cited fo this pefeence include longe stoage times, since thee is no boil as with liquid systems, less systems complexity, and less enegy input equied. The advantages to liquid stoage ae geate enegy density and hence geate ange, and inheent puity of liquid hydogen as opposed to gas. (9,11,12) Table 1-2 shows the elative enegy costs associated with ideal liquefaction systems as compaed to ideal gas compession systems fo 5 and 1 psi. Ideal isothemal compession is given by the equation W = PoV o ln( P 1 ) 1-4 P although this is not attainable in eal life. A moe ealistic pocess is the adiabatic compession pocess, expessed as o γ 1 = ( γ P W ) P {( ) 1} 1 1 γ γ ovo 1-5 P This table also shows the density of the poduct. Fom this table it is appaent that thee is still significant enegy cost associated with high-pessue gas stoage. Table 1-2 Ideal wok input and density fo liquefaction vs. pessuiation of hydogen Ideal Wok Input Density kj/kg kg/m 3 Liquid psi fluid psi fluid o

20 6 Now that a baseline liquefaction wok equiement fo ideal systems has been identified, actual liquefaction pocesses can be compaed. A paamete called the Figue of Meit (FOM) is defined as ideal wok equied divided by the actual wok equied, and vaies between and 1. Linde-Hampson Cycle The most simple liquefaction system is the Linde-Hampson system, which uses a Joule Thomson (J-T) valve fo isenthalpic expansion of the gas. The basic Linde- Hampson system is unacceptable fo use in hydogen systems because of that isenthalpic expansion, since the J-T coefficient fo hydogen at ambient tempeatue is negative. This means the expansion of hydogen gas at ambient tempeatue will ceate a heating effect. It is not until the hydogen gas eaches 25 K, its maximum invesion tempeatue that the J-T coefficient becomes positive and efigeation may occu upon expansion. Howeve, the Linde Hampson cycle will be discussed hee because many of the same pinciples ae used in othe cycles. A simple schematic of the Linde-Hampson cycle is shown in Figue 1-2 (9), and the cycle state points ae epesented in the T-s diagam shown in Figue 1-3 (9). Fist, the gas is compessed isothemally fom a low pessue P 1 to a high pessue P 2. The highpessue wam gas steam then passes though an isobaic heat exchange (HX), being cooled fom the cold low-pessue gas steam. Next, the gas undegoes an isenthalpic expansion back to the oiginal low pessue though a flow estiction, and entes the twophase egion. Hee, some faction of the pocess steam is withdawn as a liquid, and the emaining vapo is used to cool the high-pessue gas steam befoe being e-compessed

21 7 fo anothe ound though the cycle. Make-up gas, equal in mass to the liquid withdawn, is added pio to the compession step. COMPRESSOR 1 2 Q & HEAT EXCHANGER 3 m& W & c JT VALVE m& f & - m& f m g 4 Make up gas m& f f LIQUID RESERVOIR Figue 1-2 Linde Hampson schematic 2 1 T P=const 3 h=const f 4 g s Figue 1-3 Linde Hampson T-s diagam

22 8 A contol volume enegy balance fo the heat exchange, expansion valve, and liquid tank, using the same assumptions as befoe, allows one to solve fo the liquid yield. y = m& f m& = h1 h2 h h 1 f 1-6 Applying a simila enegy balance to the compesso gives the wok pe unit mass liquefied to be W& m & f h1 h f = ( )[ T1 ( s1 s2 ) ( h1 h2 )] h h This is an ideal analysis that does not take into account impefect heat exchange, pessue dops in the system, heat leak into the system, and assumes isothemal compession. Nevetheless, this analysis will give a liquid yield of about 7% fo liquid nitogen systems, and has a typical LN 2 FOM of.115. Pe-Cooled Linde-Hampson Cycle The basic Linde-Hampson cycle can be used as a hydogen liquefie if the highpessue gas steam is cooled below the J-T invesion tempeatue pio to undegoing expansion in the valve. This system, efeed to as the pe-cooled Linde-Hampson, is shown schematically in Figue 1-4 (9) and the cycle state points ae shown in Figue 1-5. (9) A seconday efigeant, most commonly liquid nitogen in eithe an open o closed cycle, is used to decease the tempeatue of the gas pio to enteing the ecupeative heat exchange. Again using an enegy balance fo the contol volume including the two heat exchanges, expansion valve and eceive tank, a elation fo the liquid yield can be found to be y = h1 h2 h h 1 f m& + m& ( h a ( h 1 h h c f ) ) 1-8

23 9 and the ideal wok equiement pe mass of gas liquefied becomes W& m& f = 1 [ T y ( s s ) ( h m& h ) + m& ( h h b a )] 1-9 The final tem epesents the additional wok equied by the efigeant compesso. Thee ae added complexities with this type of system, and caeful design is equied to find the optimum efigeant flow ate. Typically this type of system will aise the FOM fo a nitogen liquefie to a value of.168. Vaiations on the pe-cooled Linde-Hampson include the Cascade cycle, whee the efigeant is pe-cooled by anothe cycle, which in tun could be pe-cooled by anothe cycle. A simple schematic of a cascade system is shown in Figue 1-6. (9) b c W & c2 REFRIGERANT LOOP a d COM PRESSOR Q & HEAT EXCHANGER HEAT EXCHANGER m& W & c1 JT VA LVE m& f m & & m f g 5 Make up gas m& f f LIQUID RESERVOIR Figue 1-4 Pe-cooled Linde Hampson schematic

24 1 2 1 Refigeant BP T 3 P=const 6 4 h=const f 5 g Figue 1-5 Pe-cooled Linde Hampson T-s diagam s AMMONIA ET HYLENE METHANE NITROGEN Figue 1-6 Cascade cycle schematic

25 11 Linde Dual Pessue Cycle Anothe modification to the basic Linde-Hampson cycle is to add an intemediate pessue compession to the system, so not all the gas is compessed to the final pessue. This educes the total wok equiement somewhat and allows some of the heat to be emoved fom the gas at a highe tempeatue, making the pocess moe efficient in tems of wok equied pe unit mass liquefied. A schematic of the dual pessue cycle is shown in Figue 1-7 (9) and the T-s diagam is shown in Figue 1-8. (9) The liquid yield is found to be y = h h h h f m& i m& ( h ( h 1 1 h2 ) h ) f 1-1 And the ideal wok pe mass liquefied is W& m& f = 1 m& i [ T1 ( s1 s2 ) ( h1 h2 ) { T1 ( s1 s2 ) ( h1 h y m& 2 )] LOW PRESSURE COM PRESSOR 1 HIGH PRESSURE COM PRESSOR m& W & c1 Q & W & c2 Q & 2 HEAT EXCHANGER m& i 4 8 JT VA LVE 5 m& f Make up gas m& m& f m i 6 JT VA LVE g 7 m& f LIQUID f RESERVOIR Figue 1-7 Linde dual pessue cycle schematic

26 P=const T 4 H=const f 7 g s Figue 1-8 Linde dual pessue cycle T-s diagam Claude Cycle Anothe typical cyogenic liquefaction cycle, commonly used in hydogen liquefies, is the Claude cycle. This system is shown schematically in Figue 1-9 (9) and the state points ae shown in Figue 1-1. (9) The majo diffeence between this and othe cycles discussed so fa is the fist stage of expansion is done though a wok-extacting tubine. This expansion is typically appoximated as isentopic, and this allows the hydogen to be cooled moe efficiently than isenthalpic expansion, and cooling will occu no matte what the initial tempeatue. The efigeation poduced in the fist stage expande is used to pe-cool the emained of the gas, so heat can be emoved moe efficiently at a highe tempeatue. A second stage J-T expansion is used fo final

27 13 liquefaction, fo simplicity since tubines typically cannot toleate lage liquid flow. An additional benefit of lage scale Claude systems is the wok poduced in the tubine expande is used to help compess the gas, educing the total wok equied. Liquid yield can be appoximated as y = h1 h2 h h 1 f m& e + m& ( h 3 ( h 1 h h e f ) ) 1-12 And the ideal wok pe unit mass liquefied is W& m& f = 1 [ T y ( s s ) ( h m& e h ) m& ( h h e )] 1-13 Thee ae many vaiations of the Claude cycle in use, not just fo liquefaction of hydogen but in many cases efigeation of helium as well. This cycle can be combined in many ways with dual-pessue systems and pe-cooled systems. In some cases wok is ecoveed in the outlet of the tubine but in othes with wok is dissipated in a baking system. Q & COM PRESSOR HEAT EXCHANGER HEAT EXCHANGER HEAT EXCHANGER m& W & c 3 m& e JT VA LVE m& f m & & m f W & e EXPANDER g 6 Make up gas e m& f f LIQUID RESERVOIR Figue 1-9 Claude cycle schematic

28 P=const 9 T S=const H=const 8 f 6 g e s Figue 1-1 Claude cycle T-s diagam Lage Scale Hydogen Liquefaction Plants Seveal lage-scale hydogen liquefaction plants ae in opeation aound the wold and supply the majoity of liquid hydogen cuently consumed. Details of the exact cycles used ae consideed popietay but in geneal the plants use vaiations on the Claude cycle with a dual pessue system, and include seveal stages of isentopic tubine expansions befoe the final J-T expansion step fo liquefaction. (1,7,13) These plants have been designed and optimied fo the specific conditions of the local economy, and compomises between capital costs (typically the numbe of stages of tubine expansion and heat exchange effectiveness) and opeating costs (electical input and maintenance) have been analyed fo economic efficiency. Geneally, plants in the US also include a

29 15 liquid nitogen pe-cooling stage while this pactice has not been favoed in Euope as much. Suveys of existing lage scale liquefaction plants as well as themodynamic analysis has found that these systems typically opeate at efficiencies appoaching 4% of the Canot ideal cycle. (14) This calculates that the total wok equied pe mass of gas liquefied is typically aound 3, kj/kg. Obviously this is significant since this is appoximately 22% of the HHV of the hydogen. Anothe facto to conside is the otho-paa convesion, and the enegy associated with this exothemic eaction. Details of the otho to paa hydogen convesion ae given in Chapte 4. The optimum method of emoving the heat of convesion is to use a catalyst to speed up the convesion pocess and emove the heat at the highest tempeatue possible. Cuent plants pefom the catalyst step and heat emoval in a numbe of discete tempeatue points, but pototypes have been built and tested that pefom convesion and heat emoval in a continuous pocess. Zeo Boil Off Gound Applications One common featue of liquid hydogen systems, indeed of cyogenic systems in geneal, is the fact that they ae neve in themodynamic equilibium with thei suoundings. Heat tansfe fom the ambient tempeatue to the cyogenic stoage tempeatue will always occu, no matte how good the themal insulation systems ae. If the liquid is subcooled, the heat leak inceases the sensible heat of the liquid, and if the liquid is satuated the heat leak is absobed by the latent heat of vapoiation and boil off occus. This boil off inceases the vapo pessue in the tank until the maximum opeating pessue is eached, and poduct losses occu as the excess vapo is vented

30 16 though the pessue elief system. As a side note, although the tem boil off is in geneal use in the cyogenic industy to descibe this poduct loss, heat leaks into the tank ae typically so small that no nucleate boiling is actually occuing. A moe accuate tem would be suface evapoation. Due to the cost of poducing hydogen fom natual gas feedstock, liquefying it at a cental plant, and then shipping it to the final destination, it appeas economically attactive to ensue that boil off losses ae minimied o pehaps even eliminated. NASA has been investigating eo boil off (ZBO) systems fo many yeas at the Kennedy Space Cente (KSC). Matin Maietta poposed a hydogen eliquefie fo the LC-39 complex in 1977 that would emove the heat that leaked into the tank by compessing the cold vapo at the top of the tank and then expanding it isenthalpically to obtain cooling in a modified basic Linde-Hampson system. (15) The compessos would opeate at ambient tempeatue, with the compesso inlet steam cooling the compesso outlet steam, afte the heat of compession was emoved at ambient tempeatue. The compesso and coldbox wee to be installed at the top of the LH2 tank at LC-39. Review of the themodynamics and economics of the system indicated the appoach was feasible, howeve the idea was neve implemented because NASA management was concened the wok at the pads would impact the upcoming maiden launch of the Space Shuttle. In 1991, Egenics Inc. poposed a system that would captue boil off losses in a metal hydide bed. (16) The captued gas would then be compessed in a hydide compesso, pe-cooled with liquid nitogen, and expanded in a J-T expansion system. The analysis done indicated the system could be sied to captue not just the nomal boil-off of the hydogen tank, but also losses duing tanke offload opeations, fom chill down of the

31 17 tansfe lines as well as losses fom tanke venting. This appoach equied the development of a lage metal hydide stoage system as well as a 2-ton pe day liquefie that would only opeate a few weeks pe yea. This poposal was ejected patly fo the eason of complexity of a tansient efigeation system opeating in batch pocesses, and concens ove eliability in such a design. At this same time, a Phase II SBIR contact was awaded to Hydogen Consultants Inc. to pove the concept of a metal hydide compesso design that could un a J-T efigeation system. (17) This system was sied to povide just enough efigeation to ovecome the steady heat leak into the tank, and the concens ove themal tansients was eliminated since the system was designed to opeate continuously, using dual egeneative hydide beds. The pogam did poduce woking hadwae, but thee wee technical issues with the J-T expansion device feeing and sticking duing opeation. In addition, the measued efficiency of the system was just 8.5% Canot, poo by compaison with othe cyogenic efigeatos. In all thee cases discussed above, the hydogen was allowed to vapoie, and then wok was pefomed on the vapo to compess it pio to expansion and eliquefaction. Despite the lack of success in ceating a eo boil off system at KSC in the past, thee ae still sound economic advantages to ecaptuing hydogen losses fom boil off. In the case of liquid hydogen at Kennedy Space Cente a quick analysis shows the potential payoff fo such a system. (18) Assume KSC pays a ough cost of $5.4 pe kg of LH2 (not a tue cost). This includes the cost to poduce the hydogen fom the natual gas, cost to chill the hydogen fom its ambient tempeatue to a satuated vapo, cost to liquefy the vapo, the cost to ship the hydogen to KSC in the tanke (plus losses in the tansfe pocess), the cost to offload it to the KSC stoage vessel, and pofit fo the

32 18 vendo. Now, afte all the enegy and effot that went into that pocess, evey 2 J of heat that leaks into the stoage vessel ceates a loss of 1 gam of poduct. Fom an enegy atio standpoint, the equation is Ratio = E H poduction 2 + m h m h fg 3K 21K + m h fg + E shipping + E tansfeloss 1-14 It is difficult to estimate all the enegy put into the entie pocess, especially the poduction pocess, but 88% of the enthalpy emoved fom the hydogen duing the liquefaction pocess occus between ambient tempeatue and the satuated vapo state. Howeve, when the liquid boils off and vents fom the tank, this stoed efigeation is vented to atmosphee. This heat leak can be intecepted without ceating boil off losses and the only enegy cost to the system is the enegy that goes into efigeation. Assuming a efigeation tempeatue of 2 K, an efficiency of 35% Canot, and an electical enegy cost of $.9 pe kw-h, a ZBO system equates to buying hydogen fo $.5 pe kg. Thee would be capital costs to be amotied ove the life of the system, and these ae not included in this simple analysis. Any additional opeations (manpowe) cost associated with this efigeation will be offset by opeational eductions in tanke offloads. Thee ae safety benefits fom the elimination of tank venting as well as educing the numbe of tansient opeations. Some estimates of the economic savings associated with ZBO gound systems ae shown in Figue Thee ae a vaiety of cases analyed, fom cuent Shuttle launch opeations to pojected Cew Launch Vehicle and Heavy Launch Vehicle launch ates. The options include only ecovey of boil off in the pad tanks, ecovey of boil off and tanke losses, and finally ecovey of boil off, tanke losses, and chill down losses. This figue shows the estimated payback time fo the system is between 2 and 6 yeas.

33 fixed hydogen pice 24 labo ates 25 Total cost ($M) 2 15 HLV beakeven point HLV ZBO HLV boiloff CLV ZBO CLV boiloff 1 5 CLV beakeven point Time (yeas) Figue 1-11 Zeo loss stoage economic analysis Space Applications In addition to developing ZBO systems fo lage-scale gound use, NASA has actively woked on developing a ZBO system fo in space cyogenic stoage. In this instance, the pimay concen of poduct loss is magnified by the penalties imposed by the ocket equation. The cyogen in this case, fuel and oxidie fo a futue populsion stage, is consideed payload fo the launch vehicle and is subjected to the same small payload mass faction as othe payloads. As an example, conside an in space cyogenic depot situated in low Eath obit (LEO). Fo evey kilogam of poduct deliveed, 6.5 kilogams of popellant ae equied fo the launch vehicle to delive it to LEO. The atio becomes 12.9 to 1 fo a depot at the L1 point, and 22 to 1 fo hydogen deliveed to the suface of Mas. (19) Thus poduct loss fom heat leak can make the use of cyogenic popellants pohibitive fo many missions, unless eo boil off systems ae developed. Lockheed Missiles and Space fist poposed the use of cyogenic efigeatos fo long-tem space missions in 1971 (2), and the USAF investigated simila concepts in

34 (21). At the time, cyocoole development was not at the equied level of development in tems of low mass and flight quality eliability to waant inclusion in systems at that time. Advances in flight quality cyocooles in the 198 s and 199 s, especially using Stiling and pulse tube cycles, have made thei use moe attactive. Analysis of poposed hydogen stoage systems as feedstock fo In Situ Resouce Utiliation (ISRU) systems fo Mas exploation detemined that ZBO made sense if the mission duation lasted longe than 45 days. (22) That is, less mass was added by the incopoation of a cyocoole and its associated powe geneation and heat ejections systems than was lost by boil off and the associated incease in tank sie if the mission lasted longe than 45 days. Patially due to this analysis, NASA funded a seies of expeiments to detemine the optimum integation methods of cyocooles in micogavity cyogenic stoage systems. Initial testing at NASA Glenn Reseach Cente (GRC) in 1999 was a poof of concept demonstation using an existing off the shelf cyocoole and a condensing heat exchange in the vapo space of the tank. Test esults wee positive, showing constant o negative hydogen vapo pessue slopes ove the duation of 77 hous, but this configuation was not epesentative of actual space conditions, as thee ae no gavity foces that would sepaate the liquid and vapo phases. (23) Two phase flow handling in eo-g is the key technology that must be demonstated. A moe epesentative flight like test, pefomed at NASA Mashall Space Flight Cente in 21, used a ciculatoy system with a hydogen pump dawing liquid fom a liquid acquisition device and flowing it though a heat exchange coupled with a cyocoole cold head. The cooled liquid then exited out a spay ba vent tube designed to povide destatification independent of ullage

35 21 and liquid positions in eo-g. The bulk hydogen was maintained in the satuated liquid state, and efigeation enegy povided was exactly balanced by heat leak into the tank. Again, this cyocoole was not a flight quality unit but a commecial unit puchased off the shelf. (24) Testing ZBO concepts with a flight like coole was accomplished in 23 at NASA GRC. This test was pefomed with liquid nitogen and a Nothop Gumman High Efficiency Cyocoole developed fo the USAF. A submeged mixe pump was included to povide foced liquid nitogen flow acoss a heat exchange suface, which was coupled to the cyocoole by a heat pipe and a flexible conductive link. Unfotunately, degadation in the system insulation pefomance ove time led to lage than expected heat leak, and tue ZBO conditions wee neve achieved. Howeve, impotant infomation egading integation of flight like cyocooles with stoage vessels was poven. (25) In all the above cases, ZBO systems wee poposed and tested that depended on the incopoation of a closed cycle efigeation system to emove heat that had leaked into the tank. NASA Ames Reseach Cente has poposed and pefomed a fist ode efficiency analysis on a system that uses the vapo fom boil off as the woking fluid in a efigeation cycle. (26) This analysis shows it is advantageous fom an efficiency point to diectly pefom wok on the fluid to be maintained, pimaily due to the fact that thee is no tempeatue diffeence equied to pomote heat exchange at the low tempeatue end of the system. This concept, simila to gound based studies mentioned ealie, has not been poven in a test envionment. Hydogen Densification One pefomance enhancement unde consideation fo the next geneation of space launch vehicles is densification of the cyogenic popellants. Popellant densification

36 22 efes to cooling the cyogens below thei nomal boiling point. Deceasing the tempeatue of liquid hydogen fom 2.3 K to 15 K inceases the density fom 7.8 kg/m3 to 76. kg/m3, an impovement of 7.3%. This density incease has a coesponding decease in vehicle popellant tank volume and mass, educing the oveall dy mass of the vehicle. In addition to eduction in tank sies, popellant densification has othe pefomance benefits. Liquid hydogen has a vapo pessue of 13 kpa at 15K, compaed to 11 kpa at the nomal boiling point. Lowe vapo pessues mean lowe tank opeating pessues while still meeting the engine inlet net positive suction pessue equied to pevent cavitation. This lowe tank pessue can esult in thinne tank walls, futhe educing dy mass. Highe popellant density also esults in smalle engine tubomachiney fo a given mass flow ate, o inceased safety magins by educing the otational speed on existing sied tubopumps. Finally, subcooled popellants can povide geate cooling powe to the engine noles and combustion chambes due to the inceased enthalpy gain pio to boil off, possibly making cooling passages smalle o minimiing chill down losses. All of the above eductions in vehicle dy mass have a cascading effect on the est of the vehicle subsystems, esulting in mass eductions in aifames and aeodynamic sufaces, obital maneuveing systems, themal potection systems, landing geas and othe systems. Studies by NASA contactos have estimated that popellant densification can esult in the eduction of Goss Lift Off Weight by 12% to 2%, depending on the vehicle design and numbe of stages. (27) Figue 1-12 plots liquid hydogen density, enthalpy, and vapo pessue as a function of tempeatue between the citical point and tiple point. (6)

37 23 Pessue (psia) Enthalpy (J/g Density (g/cm3) P (psia) h (J/g) ho (g/cm3) Tempeatue Figue 1-12 Satuated liquid hydogen density, enthalpy, and vapo pessue Because of the advantages that densified popellants offe, NASA and the Depatment of Defense have been inteested in thei potential use fo many yeas. The National Bueau of Standads pefomed densified popellant popety studies in the 196 s, usually poducing the necessay efigeation by evapoative cooling. Evapoative cooling efes to the technique of vacuum pumping the ullage space in a LH2 tank to educe the vapo pessue, leading to evapoation of some of the liquid. The heat of vapoiation needed fo this evapoation is povided by the emaining liquid, ceating a cooling effect. Union Cabide analytically investigated seveal slush hydogen poduction techniques in the same timefame. Duing the 197 s, Matin Maietta studied the concept of using a 5% slush LH2 mix with tiple point oxygen in a single stage to obit launch vehicle. Using slush hydogen has inceased benefits ove subcooled liquids, pimaily due to a density incease of almost 16% ove nomal boiling point hydogen, and one of the ecommendations fom the epot was to concentate on hydogen slush development only. (28) At this point, popellant densification was consideed an immatue

38 24 technology and NBP liquids wee chosen as the popellants on the Space Shuttle. Futhe slush hydogen wok was accomplished duing the National Aeospace Plane pogam in the late 198 s, with batch poduction of slush hydogen being accomplished in 2- lite quantities by the feee-thaw method of evapoative cooling at NASA Glenn Reseach Cente. (29) Opeational and handling issues associated with pessuiation, tansfe, mixing and sloshing, and instumentation was investigated. Although slush hydogen offes significant pefomance benefits ove subcooled liquids, technical issues with its use (including filteing, mixing to ensue homogeneous states, and mass gauging) has led NASA to pimaily conside subcooled liquids above the tiple point in most cuent studies. Moe ecently, NASA and aeospace contactos have consideed using subcooled hydogen on a modified Shuttle system, the X-33, and othe 2nd Geneation Reusable Launch Vehicles (RLV). Many of the funded pogams in this aea have concentated on development of a densification poduction system. NASA GRC and Lockheed Matin have developed and tested subscale liquid hydogen and liquid oxygen densification units based on the evapoative cooling method. (3) The lagest hydogen unit was capable of cooling 3.6 kg/sec of NBP hydogen to 15 K, and the oxygen unit cools 13.6 kg/sec of NBP LOX to 66.6 K. In addition to poduction testing, tanking tests of the X-33 stuctual test aticle tank wee completed. Tank loading pocedues, including eciculation of wam popellants, wee tested using subcooled LH2. Howeve, safety and opeational concens with using subatmospheic boiling bath heat exchanges and cold compessos led NASA to solicit altenate technologies to poducing subcooled popellants fo the 2nd Gen RLV pogam. Refigeation cycles that wee chosen fo

39 25 futhe development included oifice pulse tube efigeatos, a mixed gas J-T cycle efigeatos, and a packed column cooling towe design based on evapoation of liquid into a non-condensable gas. All thee technologies wee chosen based on the pomise of simplicity and eliability of opeation once development issues had been addessed. (31) Afte two yeas of development, these pojects wee not extended when the 2nd Gen RLV pogam tansitioned into the Next Geneation Launch Vehicle pogam and densified hydogen fell out of favo compaed to RP-1. Howeve, opeability assessments by NASA KSC duing this pogam led to questions egading the manne in which densified popellants would be implemented at the launch site. (32) This opeability assessment is the basis of the poposed integated efigeation and stoage system discussed in this dissetation. Othe popellant studies have shown the mass savings associated with using densified oxygen and methane fo ascent vehicles on the suface of Mas, coupled with an ISRU poduction facility. (33) Launch Vehicle Opeations NASA has developed techniques fo sevicing spacecaft and launch vehicle cyogenic populsion systems since the late 195 s. Techniques have evolved as hadwae and softwae capability has developed, and each cuent pogam has some vehicle and pad specific systems and opeations equied. Howeve, the basic appoach emains simila, and sevicing capabilities (in tems of quality of popellant loaded) ae nealy identical. These systems, o deivatives of them, ae capable of meeting the needs of an in space cyogenic depot, povided this depot uses popellant at o above the nomal boiling point, and fee venting of boil off in space is pemitted. Conditioning of popellants via advanced gound stoage systems has the potential to minimie cost and safety isks, while maximiing launch pefomance.

40 26 The cuent method of lage-scale cyogenic stoage and distibution is vey simila acoss all pogams at Kennedy Space Cente and Cape Canaveal Ai Station. Cyogens (Liquid Hydogen and Liquid Oxygen) ae poduced off site, deliveed via tanke tucks, and tansfeed to gound stoage tanks days o weeks pio to launch. Cyogens in the tanks ae stoed as a satuated liquid, and boil off gas is not ecoveed. Duing launch countdown, as late as possible into the count, the cyogens ae tansfeed to the flight tank, and in the event of a launch scub, ae dained back into the gound stoage tanks. Details on how this is accomplished vay acoss pogams. Hydogen fo the Space Shuttle Pogam is puchased fom Ai Poducts New Oleans plant and deliveed via 13-gallon oad tankes. Peiodic sampling of tankes is done to ensue the popellant meets puity specifications. Waves of up to five tankes can be offloaded at a time, and two waves can be done in a day. Pio to offload, tansfe lines ae puged with gaseous helium and sampled. The tank is vented and valves ae opened to stat chill down. Poduct losses fom tank venting and tansfe line chill down ae fee vented at the top of the pad stoage tank. Afte offload is complete, tansfe lines ae puged of hydogen. The pad stoage tank holds 85, gallons of liquid, with 1% ullage on top. The tank has a vacuum jacket and pelite bulk fill insulation. The cosscounty lines 1 ID, 15 feet long and ae vacuum jacketed (VJ) with multi-laye insulation (MLI). The stoage tank is pessuied using vapoie heat exchanges. Pio to launch, the tank must be filled to 7, gallons, which is enough fo thee launch attempts. Loading of the STS extenal tank (ET) begins at T-6 hous on the countdown clock. Puges to the vaious disconnect cavities is initiated and tansfe line blanket pessue is vented. The ET vent valve is opened, chill down line valve is opened,

41 27 and chill down of coss-county lines begins. Then the stoage tank is self-pessuied, and slow fill (1 gpm) to the lowe ET liquid level sensos is completed. The main tansfe valve is then opened and fast fill to 98% initiated, with flow ate of 85 gpm. When the ET ullage pessue ate eaches a limit, LH2 topping at 775 gpm is initiated until the uppe liquid level senso eads 1% wet. Then the eplenish valve contols the flow to maintain 1%, usually less than 3 gpm. At T-1:57 minutes, eplenish mode teminates. Oveall, 383,4 gallons is loaded into the ET, with 48, gallons lost duing chill down and 4, gallons lost duing eplenish. These vapo losses ae buned in a flae stack. If the launch is scubbed, dainback pocedues ae initiated. Summay The state of the at in gound pocessing of cyogenic popellants is consideed matue, and KSC opeatos have ove 5 yeas of expeience in this type of opeations. Thee ae pefomance enhancements that can be made. Of these, local poduction and liquefaction of hydogen offes benefits of eliminating tanke opeations, and eo boil off stoage can eliminate wasteful poduct losses. Hydogen densification has pefomance benefits fo the flight vehicle. NASA has invested significant funds to investigate these systems ove the past 3+ yeas, but has neve fielded an actual opeating system. Thee ae many easons fo this, but the most poweful of these has always been a lack of confidence that the benefits would outweigh the opeational impacts, and consevative foces in management wee unwilling to ty something diffeent. This wok is an attempt to change some of these positions.

42 CHAPTER 2 PROPOSED INTEGRATED REFRIGERATION AND STORAGE SYSTEM The cuent state of the at in hydogen liquefaction, stoage and distibution fo space launch systems has successfully seved its intended pupose fo the past 5 yeas. Howeve, thee ae possible impovements that can be made that will make liquid hydogen use moe economical, eliable, and safe than cuent systems. This chapte will descibe the basic concept of such a system, will explain the significance of the development, and will qualitatively descibe the themodynamic behavio of the system. Late chaptes will detail designs of an expeimental system built to test these concepts, and data analysis of initial liquefaction, eo boil off and densification tests will be pesented. Integated Refigeation and Stoage Concept The optimum design of a liquid hydogen stoage and distibution system is highly dependent on the intended use of the poduct. Fo example, most industial uses of liquid hydogen ae in a continuous o semi-continuous pocess, such as a steady flow of hydogen to hydogenate oils in the poduction of magaine o cooking oils o steady flow of hydogen in an anhydous ammonia plant. In these cases, hydogen is liquefied only because of the tanspotation issues associated with supplying lage quantities of gaseous hydogen make the pocess impactical. The hydogen is ultimately used as a gaseous poduct, although much of the cooling powe is ecupeated elsewhee in the pocess. Most of the time, thee is not a lage quantity of liquid tansfe lines, the hydogen is vapoied immediately downsteam of the stoage tank. In these cases, lage 28

43 29 scale use of a continuous flow of gaseous hydogen fom a liquid stoage system, heat leak into the tank that ceates boil off is not consideed an issue, thee ae not many issues associated with chill down and two phase flow in the supply lines, and thee is no eason to incease the density of the liquid in the stoage tank. Cuent state of the at in hydogen stoage and distibution is acceptable fo these applications. The typical usage scenaio fo a liquid hydogen system in suppot of the space pogam is vey diffeent than that in industy. The hydogen is used in a batch pocess, often with many months in between launches. Duing this time heat leak into the tank leads to significant poduct losses, in fact only 65% of the hydogen deliveed to Kennedy Space Cente is eve actually launched aboad the Space Shuttle. (34) Anothe majo diffeence is the hydogen is equied to be a liquid at the use point as opposed to a gas as in most majo industial opeations. Not only is the hydogen equied to be a liquid, thee ae stict limits on the state of that liquid that ae dictated by the design of the tank and the engine stat box. Fo the Space Shuttle main engines, tempeatue measuements on the tubopump exit and eciculation lines as well as powe limits on the eciculation pump (to detect cavitation) ensue thee is liquid flowing though the engine passages pio to stat. (35) In this manne, the usage equiements between industy and aeospace vay geatly, and cuent state of the at, while acceptable, is not optimied fo space use. These usage equiements ae even moe stict when consideing liquid hydogen poduced In Situ on the Moon o Mas. Theefoe, liquid hydogen systems designed to minimie o eliminate heat leak and boil off, opeate in a numbe of batch pocesses with lage themal tansients, and still delive good quality

44 3 liquid when equied at a use point many hundeds of metes away, ae equied fo the next geneation of space launch systems. What is tuly needed is a novel appoach to gound pocessing of cyogenic popellants, as opposed to incemental impovements in existing philosophies. The new philosophy should confom to the KISS pinciple, by minimiing opeations, especially ones that cause themal tansients o equie opening the system to outside contamination. This new philosophy should take advantage of advances in cyogenic engineeing ove the past 2 yeas including efficient and eliable efigeation systems, health monitoing of vital components, and advanced insulation systems. Most impotant, the system should offe economic and enegy efficiency by minimiing boil off and chill down losses as well as venting of high cost puge gasses like helium. One possible appoach to this issue is to integate a closed cycle helium efigeation system into the gound stoage tank. This integated efigeation system would offe the advantage of execising moe contol of the popellant themodynamic state while still in gound stoage. Such an advanced system could seve as a liquefie, a eo boil off stoage system, and possibly a popellant densification system. Popely designed, this could povide cost, safety, eliability, and pefomance benefits ove cuent state of the at. This wok poposes to integate a cyogenic efigeato into a liquid hydogen dewa. Advantages to this poposal ae numeous. Fist, the efigeato can be used to emove the heat that leaks into the vessel fom the ambient envionment, so thee is no pessue ise and associated boil off fom the heat leak. This has obvious economic savings, howeve, thee ae safety benefits as well since the tank is not venting gaseous hydogen most of the time. Second, if the efigeato is sied popely, it can seve as a

45 31 local liquefaction system. This minimies opeations cost since fewe pesonnel ae equied fo tanke offload opeations. Thee ae less puge and sampling opeations as well since thee ae no tanke supply lines to inet. An added benefit is elimination of tanke offload losses fom venting the tankes and chilling down the tansfe lines. Again, thee ae safety and eliability benefits due to minimiation of tansient opeations and venting opeations. A thid advantage of the integated efigeation and stoage is the ability to contol the state of the popellant. Cuently, liquid hydogen is ceated and stoed as a satuated liquid, although some degee of subcooling is possible with pessuiation of the liquid. This poposal allows fo both subcooling fo an extended peiod of time as well as densification of the liquid below the nomal boiling point. This is a fundamental change in the way liquid hydogen is stoed. Cuently, thee is no tue themodynamic equilibium in a cyogenic system as thee is always some degee of heat leak between the ambient tempeatue and the liquid tempeatue. This poposal mechanically emoves this heat leak, so the cyogen can be stoed in a manne simila to themodynamic equilibium. System Behavio By integating a cyogenic efigeato into a stoage vessel, themodynamic contol of the cyogenic popellants can be achieved. The contol of the system will be based on the ability to change the enthalpy of the stoed liquid, and depends on the design of the system. Specifically the diffeence between the amount of heat emoved at the stoage tempeatue by the efigeation system compaed to the enegy enteing the system fom eithe heat leak into the system fom the wam suoundings o enegy enteing the system fom a mass input will detemine the ate of change of the stoed enthalpy. Mathematically this is expessed as

46 32 δh m = Qef + QHL + m& h 2-1 δt Thee ae thee possible types of behavios associated with such a system. If the fist tem on the ight is less than the othe two, thee will be a net positive flow of enegy into the system. This leads to a tempeatue incease, o if the fluid is satuated, an evapoation of some of the liquid combined with a pessue incease. Eventually the pessue will incease to the limit on the tank elief valve, and tank venting will lead to poduct loss. This type of system is what is cuently used at KSC. A second type of system behavio is to balance the enegy enteing the tank with the efigeation poduced, so thee is no net enegy input into the system. This quasi steady state behavio is chaacteied by constant system pessue and tempeatues, despite the fact that the system is not tuly in themal equilibium with the suoundings. NASA has been studying simila systems ecently fo the application of long tem eo boil-off stoage fo space missions. Most of these studies ae fo closed systems with no mass input, but some poposals have sied the efigeato to allow fo liquefaction of some small amount of incoming popellant fom an ISRU eacto. The thid type of behavio is exhibited when the efigeato emoves moe enegy fom the system than entes via eithe mass input o heat leak. This esults in a decease in system pessue and tempeatue and a coesponding condensation of some of the ullage gas. Assuming a homogeneous system, the state will follow satuated liquid line to the tiple point and then eventually the liquid completely feees and the path follows the sublimation cuve. Fo systems of inteest to spacepot popellant sevicing, the solid phase most pobably will be avoided and efigeation will be contolled to keep the state above the tiple point. In addition, to avoid contamination the system pessue would

47 33 pobably need to be kept above atmospheic pessue by using a non-condensable pessuiation gas such as helium. Combined with the fundamental change in behavio of the integated efigeation and stoage system as descibed above, thee ae a numbe of opeational enhancements to be consideed. Fist and foemost, the system can be developed to be a eo loss stoage system. Bette than a eo boil off system, a tue eo loss system would neve have any poduct loss, including duing chill down and tansfe opeations. While this idealied case will pobably neve be fully achieved, captue and ecovey of 5% of cuent chill down losses duing countdown will have substantial savings. This implies to less venting and flaing opeations at the launch pad, which is a potential safety impovement, and equies a smalle stoage vessel with less capital costs and heat leak than the cuent altenative. Anothe ambitious goal would be elimination of boil off fom heat leak into the flight tank duing loading opeations. This would lead to a stoable fom of cyogenic popellant, which has geat benefits fom an opeational standpoint. Stoable cyogens will have extended countdown hold times, and can be useful as a apid esponse type of launch vehicle. Since flight tanks ae geneally insulated with foam, the heat leak is an ode of magnitude highe than what is found in the gound supply tank. This necessitates a efigeation system with highe capacity, but this is a benefit in itself. Lage capacities imply geate efficiencies, and the added capacity can be used as a liquefaction system when not in launch countdown. If liquefaction capability is built in to the stoage system, thee is no need fo tanke delivey of liquid, and the only supply connection can be a pue gaseous hydogen input at the top of the tank. This leads to less gound suppot equipment (GSE), no tanke pots and tansfe lines, less haadous

48 34 opeations, less themal tansients, and less potential fo contamination fom line connections. Finally, when the efigeation system is not pefoming eo boil off o liquefaction opeations, the capacity can be used to densify and subcool the popellant. Densification has advantages fo launch vehicle pefomance, as will be discussed in late sections, howeve, even if launch vehicles use NBP popellants thee ae advantages in densifying and subcooling on the gound. Densified, subcooled popellant acts as a souce of stoed efigeation enegy that can be used as a themal mass in liquefaction and eo boil off applications, may allow fo faste chill down times, and ae easie to tansfe due to geate density and less tendency to exhibit two phase flow and cavitation. These thee new opeational capabilities, ZBO o eo loss, liquefaction, and densification/subcooling, will be discussed in geate detail below. The added capability will come at some cost, mainly capital impovement of the pads by adding a cyogenic efigeation system. This capital cost will be offset by savings in pocuement by adding ZBO and liquefaction capability. If popely designed, thee should be little added opeations equied as a esult and should be offset by opeational savings esulting fom elimination of tanke opeations. Close cycle helium efigeatos using the evese tubo-bayton cycle ae cuent state of the at, and need no advanced development to be modified fo this application. These systems ae in use at many lage-scale magnetic labs and paticle acceleatos and ae poven to be efficient and eliable. Many components ae also in use at open cycle liquefaction plants aound the wold. Helium scew compessos ae at ambient tempeatues and ae oil lubicated. The only cold moving pats ae the tubines, and at hydogen tempeatue anges these do not equie expansion of helium into the two-phase egion. The tubines ae also

49 35 mounted outside the cold box and can be eplaced LRU style with no inteuption in sevice. The efigeatos ae designed with moden contols systems ae self-egulating to povide optimum opeation at a ange of output conditions. (36) In a full scale Spacepot system, the efigeation system may be sied to emove the heat leak in the flight tank. Flight tanks geneally use Spay on Foam Insulation, so oveall heat leak may be on the ode of 2 kw. When the flight tank is not loaded, the efigeato will be ovesied. This excess capacity can eithe be tuned down o the system can be used as a hydogen liquefie. On site liquefaction has opeational advantages. Thee ae safety benefits, since thee is no tanspotation acoss public highways, no tansfes of cyogenic liquids fom tankes to stoage tanks, and less haadous venting. Cost savings will occu due to less opeations and tanspotation, especially if the hydogen poduction facility can be sied to povide economy of scale fo many sites. The entie spacepot can consist of one centalied hydogen poduction facility supplying a numbe of modula, self-contained liquefaction and stoage sites at each pad thu pipelines not unlike the natual gas lines in common use. Thee is possible synegy with the futue hydogen economy, and the need fo localied hydogen poduction and stoage, and the system seves as a pototype of a futue Luna o Mas popellant poduction and stoage facility. The poposed testbed will have the capability to delive hydogen gas to the liquid system. Afte pessue eduction, the gas will be meteed thu a mass flow contolle. The egulated gas steam can be deliveed in thee possible pots. Fist, the gas can be added to the ullage space on top of the liquid, condensing at the liquid/vapo inteface. Second, the gas can bubble up thu the bottom of the tank, with the bubbles exchanging heat with

50 36 the subcooled liquid. Finally, the gas can exchange heat diectly with a condensing heat exchange line at the cold head. Fo all thee methods, the liquefaction ate must be measued. To incease efficiency of the liquefaction pocess, the gas may be pe-cooled with a liquid nitogen HX o an intemediate stage of helium efigeation, so the high tempeatue enthalpy changes can be emoved with less expensive efigeation. In addition, pe-cooling allows the addition of an otho to paa convesion catalyst to be added. This will help emove some of the heat of convesion befoe the hydogen entes the dewa. Behavio Issues The poposed novel opeational system will behave diffeent than the NBP countepat and thee will be a leaning cuve associated with thei use. Reseach into opeational behavio must be addessed. The topics to be exploed include pessuiation and venting of stoed subcooled liquids, integated efigeation systems contols to povide optimum pefomance duing diffeent opeational phases of densification, eo boil-off and liquefaction, detemination of heat and mass tansfe coefficients, and methods of handling statification layes. Advanced instumentation should be developed to accuately detemine the state of the popellant quality The vapo pessue of liquid hydogen stoed at 14 K is only 7 kpa. Having a subatmospheic pessue inside the dewa may ceate safety concens since any leak path in the tank will daw in outside ai, which will immediately feee. While the inne tank must be designed to hold a vacuum, that should not be the nomal mode of opeation. Anothe issue is the tendency of the tank pessue to decease as heat is emoved fom the tank, as opposed to nomal heat leak into a tank causing a pessue incease. The tank will opeate at two pessue settings, nomal (extended) opeation with a small positive

51 37 pessue diffeential, and tansfe opeations pessue with a highe positive pessue to ovecome line esistance and elevation changes. Nomal opeational pessue will be appoximately set at 14 kpa. The inside of the tank will not be in themodynamic equilibium, as the cold liquid will continuously condense the vapo at the top of the tank. To maintain positive pessue, thee diffeent souces of ullage pessue may be consideed. Fist, gaseous helium, non condensable at 14 K, will be used. Testing will be pefomed to detemine the ate of pessuiation of the system using vaious flow ates of gaseous helium. This will depend on the heat tansfe ate between the liquid/vapo inteface and the convection heat tansfe in the gaseous helium. Techniques to pevent the collapse of the ullage pessue will be investigated, including bubbling the helium up though the liquid hydogen to incease heat exchange ate and pe-cool the helium. The othe option fo pessuiation gas is hydogen. Liquid hydogen fom the tank will be diected to a vapoie consisting of a heat exchange coil, isolation valve, and pessuiation egulato. The dynamics of the system heat exchange ates will be ecoded, in an attempt to balance the ate of heat tansfe between the liquid/vapo inteface and the vapoie and atmosphee. Again, gaseous hydogen will also be bubbled up though the liquid, although this is not advantageous since the amount of heat enteing the tank will incease using this method. Finally, a bottle of oom tempeatue hydogen will be used as a pessuiation souce. This gas will liquefy at the inteface, so a continuous souce of gas must be used to pevent ullage collapse. This pessuiation method will be discussed moe when hydogen liquefaction is consideed.

52 38 Duing opeations that equie tansfe of liquid fom one tank to anothe, tank pessue will need to be inceased to ovecome fiction and elevation head losses. All thee pessuiation souces discussed above will also be evaluated fo thei suitability fo this pupose. In this case, the ullage space in the tank will be inceasing as the tank liquid level is deceasing, so the mass flow ate equied will be geate than just tank pessue maintenance. This mass flow ate must equal the ate equied to fill the exta volume plus the ate equied to make up the contaction as the gas cools down. Depessuiation without venting the tank will be achieved by teminating the flow ate of pessuiation into the tank while maintaining the cooling of the hydogen. Howeve, this appoximation depends on pefect heat exchange between the cyocoole and the hydogen, and it assumes thee ae no tempeatue gadients in the liquid (lumped capacitance method). In eality, the oveall heat tansfe coefficient of the fee hydogen convection must be detemined. The tank will be instumented to allow fo monitoing of the tempeatue gadients in the liquid, so the total themal enegy in the system can be calculated. Knowing the pefomance capabilities of the cyocoole as a function of tempeatue, the ate of heat tansfe fom the liquid to the cold heat exchange can be calculated, and the heat tansfe coefficient can be detemined. The cold head of the cyocoole will be designed to allow fo a vaiety of heat exchanges to be used. The initial cold heat exchange will be bundled of OHFC coppe extending downwad fom the cold head. Futue designs will include hoiontal and vetical sufaces with a vaiety of extended fins. The vetical position of the heat exchange will be vaiable by emoving o eplacing the heat pipe in the system. One enhancement unde consideation will be the addition of a mixing o stiing device in the tank. This

53 39 could help the ate of heat tansfe in two ways; ensuing a unifom tempeatue in the tank, and ceating a foced convection cuent acoss the cold HX.

54 CHAPTER 3 ANALYSIS OF GOVERNING EQUATIONS In Chapte 2, the concept of the integated efigeation and stoage system was discussed, and the behavio of the system was addessed fom a qualitative standpoint. This chapte will look at the themodynamic behavio fom a quantitative pespective. Fist, a themodynamic model of the poposed liquefaction cycle will be developed. This model will integate the closed cycle helium efigeato into the hydogen input steam. Details on the system opeating chaacteistics will be pesented, and a ange of acceptable intemediate cycle tempeatues will be found. Optimiing the cycle paametes, namely the hydogen and helium compession atios and the intemediate tempeatue, to minimie the total wok input will be completed. Next, a simplified model of the stoage system will be made using a two-dimensional tansient mass and enegy balance appoach, stating with the integal fom of the consevation equations. This model will then look at specific opeational situations and pedict system behavio. Whee applicable, this behavio will be compaed to expeimental data obtained in Chapte 6. Coective changes to this model will be poposed to moe accuately pedict the system behavio. These changes incopoate coection factos accounting fo the vaiable liquid level in the system as well as non-isothemal tempeatue pofiles in the ullage space that lead to vaiations in the system heat leak. Finally, since the bulk fluid mass and enegy balance appoach still falls shot of a tue physical model of the system, the full consevation equations will then be pesented in diffeential fom based on an analysis of expected flow and heat tansfe egimes. Initial conditions and bounday 4

55 41 values will be pesented to fully pose the poblem. This poblem fomulation is an impotant fist step in the geneation of a full solution, as CFD expets may not have the equisite expeience in undestanding the behavio of the system to be modeled. These consevation equations will then be conveted to dimensionless fom, using the initial system enegy and heat leak ate as dimensional constants. This will povide insight into the system behavio fom the pespective of how much heat is enteing the system with espect to how much enegy was initially in the system. Themodynamic Analysis of Liquefaction Cycle Cycle Desciption A novel cycle fo hydogen liquefaction is pesented in this section. This cycle takes advantage of the equied helium efigeation cycle needed fo maintaining the hydogen in the contolled stoage state and uses excess efigeation capacity to pecool the incoming hydogen steam as well as emove the latent heat of vapoiation of any satuated hydogen vapos downsteam of the expansion valve. A schematic of this poposed cycle is show in Figue 3.1. Pue hydogen fom a poduction plant entes the cycle at point 1 and is compessed to pessue P2 by a hydogen compesso. HX3 emoves the heat of compession and sends the ambient tempeatue, high-pessue hydogen gas to the ecupeative heat exchange HX3. In HX3, a cold helium efigeation system emoves heat fom the hydogen and cools it down to the intemediate tempeatue T4. Then, the cold compessed hydogen is expanded though a Joule-Thompson valve to the stoage pessue P5. At this point, depending on the stoage pessue P5 and the unexpanded hydogen state (T4, P4) the hydogen is eithe a cold vapo o a two-phase flow. The integated heat exchange HX4 inside the stoage tank emoves the emainde of the heat

56 42 to convet the hydogen to 1% liquid that can be stoed o dained fom the tank at location HX1 HX2 7 8 W hyd Hydogen Compesso HX Tubine Helium Compesso JT Valve Tubine 5 LH HX4 6 Figue 3-1 Poposed Liquefaction Cycle Schematic The closed cycle helium efigeato in this model is a two-stage Bayton cycle with extenal heat loads at HX3 and HX4. The helium is compessed fom the low pessue to an undefined high pessue P8. Heat of compession, as well as extenal heat loads, is emoved in HX2. The helium flow is split downsteam of this heat exchange, and some mass flow ( m& 14 ) is split off to be expanded though the fist (high tempeatue) tubine. The emainde of the flow ( m& 1 ) is pe-cooled in HX3 and then expanded in the

57 43 second (low tempeatue) tubine to a tempeatue equied to emove the leftove heat of the hydogen vapos in the tank fo that paticula mass flow ate. Both expansion tubines opeate at the same pessue inlets and outlets and it is not a dual pessue cycle. The two tubine flow steams ecombine immediately downsteam of HX3 and flow togethe to pe-cool the high pessue helium steam and incoming hydogen steam. Some of the unique featues that diffeentiate this cycle fom othe hydogen liquefaction cycles ae now discussed. Note how the hydogen input steam is not a continuous cycle with cold vapos etuning to the hydogen compesso. The hydogen flow is in one diection only, fom the compesso to the tank, and the ultimate liquid yield is 1%. This is due to the pesence of HX4, which emoves the emaining heat of vapoiation fom hydogen downsteam of the JT valve. In fact, in some cases the JT expansion may not even be needed and no cooling of hydogen comes intenally fom expansion. The final pessue of the JT expansion is not fixed, but vaiable depending on the mass flow ate of the hydogen and the amount of cooling povided by the helium efigeato at HX4. In some cases this may be subatmospheic pessue if the desied stoage state is densified hydogen. The ecupeative heat exchange HX3 is a key component in this cycle. It is a counteflow heat exchange with two wam inlet steams and one cold inlet steam. Thee is a net heat load on the helium cycle that is detemined by the hydogen mass flow ate multiplied by the enthalpy diffeence between state 3 and state 4. The final hydogen tempeatue T4 pio to expansion is a vaiable to be optimied in the model. This optimiation is done in a late section with HX3 assumed to be a pefect heat exchange with no pessue dop. The advantages to using HX3 to pecool the hydogen ae

58 44 obvious. The geate the amount of enegy emoved fom the input steam at high tempeatues, the moe efficient the cycle efficiency will be. This is the pupose fo many of the LN2 pe-cooled systems in use today, except this cycle does not use consumable nitogen as a pe-cooling souce but elies on closed cycle helium efigeation instead. An additional featue is the incopoation of an otho-paa catalyst in the inne passages of the hydogen steam, which allows fo the heat of convesion to be emoved continuously at evey tempeatue in the most efficient manne possible. The depiction of the compession step is shown fo simplicity only and in most cases multiple stages of compession will be used with intestage cooling. The helium compesso is most pobably an oil flooded scew compesso, and exta equipment will be needed downsteam of the compesso to emove oil and othe contaminants pio to flowing into the coldbox. These compessos ae used successfully in lage-scale helium liquefies at majo paticle acceleato facilities. The hydogen can be compessed eithe by a mechanical compesso o as a high pessue output of a hydogen poduction system, such as a steam methane efome o an electolysis unit. If the hydogen is aleady compessed by the poduction system, this will save enegy by taking advantage of the pio compession, and in these cases the wok input fo compession of hydogen will not be factoed into the liquefaction wok input. In some cases the hydogen may not even be compessed, if the JT expansion pocess is not needed. This is the situation pesented in the Chapte 6 expeimentation section, with oom tempeatue hydogen being cooled and liquefied entiely by the heat exchange in the stoage tank. Model Development A themodynamic model based on the consevation of mass and consevation of enegy pinciples has been developed to pedict cycle behavio and optimie the

59 45 opeating conditions. The model uses P1, T1, P7, T7, and m& hyd as input paametes. The heat exchange delta pessue and tempeatue can be specified as well, but in most cases the heat exchange pocess was appoximated as pefect. In the model, an isentopic compession in one stage is used, but multiple stages and intecooling moe closely appoximates an isothemal compession stage and would educe the oveall wok input. An isentopic expansion efficiency of 8% is used fo the calculations fo both the compessos and expansion tubines. The desied liquefaction stoage tempeatue is also equied as an input, and has a geat effect on the outcome. The compession atios of the hydogen and helium steams ae vaiables to be optimied. Wok equied fo compession is calculated, then the heat exchange equied to eject the heat of compession is pefomed pio to flowing though the ecupeative heat exchange. Knowing the desied stoage pessue and the heat exchange pessue dops, the system pessues ae known at all points in the cycle. The high tempeatue tubine exit pessue is set by the low-pessue equiements of the compesso, and a dual pessue system is not appoximated. The unknown paametes at the ecupeative heat exchange ae the outlet tempeatues (T4, T11, and T16) and the equied helium mass flow ates though both the high tempeatue and low tempeatue tubines. Assuming an intemediate hydogen tempeatue at T4, the othe heat exchange tempeatues can be calculated. The helium mass flow ates ae then computed using the enegy balance on the tee and the enegy balance acoss the whole heat exchange by the elations m & + & & ( & + & h h13 m15h15 = m16h16 = m13 m15 ) 16 m& h h ) + m& ( h h ) = m& ( h ) 3-2 3( h7

60 46 Fom thee, an isenthalpic expansion is calculated acoss the JT valve to the desied stoage pessue. The heat equied to fully liquefy the JT poduct steam is calculated, and that detemines the siing of the helium at heat exchange 4. If the hydogen is not satuated at the exit of the JT valve (a condition that can occu if T4 is not low enough o P2 is not high enough), then the helium in HX4 must emove the emaining sensible heat pio to hydogen liquefaction. The low tempeatue expansion tubine pessue atio is aleady detemined by the cycle paametes, so the tubine exit tempeatue is known. The equied mass flow ate is calculated using an enegy balance acoss HX4. The model optimies the cycle in tems of minimum mass flow ate equied fo the desied cooling in HX3 and HX4. Seveal constaints ae placed on the optimiation suboutine. A minimum tempeatue at the outlet of the low tempeatue tubine is set at 14K to eliminate the potential fo feeing hydogen in the vessel. The mass flow ates fo all points on the helium cycle must be positive o eo. The low-pessue helium steam into the ecupeato is constained to be lowe than the hydogen exit intemediate tempeatue and the high-pessue helium exit tempeatue. Analysis Fom the development of the model, it becomes appaent the independent vaiables to be used in the optimiation pocess ae the hydogen compession pessue P2, the intemediate tempeatue downsteam of the ecupeative heat exchange T4, and the helium compession pessue P8. The combined wok of the hydogen and helium compessos ae the dependent vaiables to be minimied. This pocess can be epeated fo a ange of inlet conditions and final stoage pessues fo hydogen, vaiations in compesso and tubine efficiencies, and inefficiencies in the ecupeative heat exchange pocess.

61 47 Fo a given pai of hydogen and helium high pessues, thee is a ange of intemediate tempeatues that can be used to give a satisfactoy solution to the equations 3-1 and 3-2. The family of cuves shown in Figue 3.2 displays this elation. Fo this figue, the hydogen cycle was assumed to have no hydogen compession and the steam is isobaic at 12 kpa. This means no isenthalpic cooling of the hydogen and the liquefaction is completely pefomed by the heat exchange with the helium in HX3 and HX4. Using a ange of diffeent helium pessues (anging fom 36 kpa to 72 kpa), the tempeatues that give a non-tivial solution ae plotted. If the intemediate tempeatue is colde than the left hand cuve in the figue, the system will not wok because the high tempeatue tubine will be limited to the tempeatues it can achieve, and T16 will not be less than T4. Similaly, using T4 highe than the ight hand cuve equies a geate amount of cooling has to be pefomed at the low tempeatue tubine, and the tubine exit tempeatue deceases. In ode to get the equied cooling inside the stoage tank, the mass flow ate inceases. The ecupeative heat exchange becomes unbalanced and eventually the equied mass flow ate though the high tempeatue tubine becomes eo. Notice the ange of tempeatues incease as the helium compession atio inceases, which allows fo geate cycle flexibility duing opeation. Simila behavio was obseved in the cycle when the hydogen was compessed and expanded in the JT valve. The dependence of the cycle feasibility on a small ange of acceptable intemediate tempeatues places many constaints on the actual system, and cae must be taken to ensue the cycle does not opeate outside these limits. It is also shown late that the oveall wok inceases as the intemediate tempeatue inceases, so opeating on the left hand side of the cuve is desied.

62 P2 = 12 kpa P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = P8 (kpa) T4 (K) Figue 3-2 Acceptable Intemediate Tempeatues as a Function of Helium Compession The figue above gives no infomation on the wok input equiements, only opeating conditions that ae acceptable fo the cycle at those tempeatues and pessues. Since the cost of liquefaction is diectly elated to the enegy input ate, egads fo the dependence of wok input on the system opeating points should be consideed. Below, Figue 3-3 plots the wok input fo liquefaction of 1 gam pe second of gaseous hydogen against the intemediate cycle tempeatue, fo a ange of compession atios in the helium cycle. All othe cycle paametes ae constant and defined below. No wok input was done on the incoming hydogen gas steam and all cooling was done though a heat exchange pocess in HX3 and HX4. Notice thee is a ange of acceptable solutions fo cetain intemediate tempeatues, and points past the endpoints of the isobaic lines do not have solutions that convege in the model. Fo a given helium compession atio, the wok input is educed as the intemediate tempeatue is deceased. This dependence is moe ponounced as the helium compession is educed, as small inceases in T4 lead

63 49 too much lage inceases in wok input. In addition, the equied wok actually inceases fo these smalle pessue atios, due to the incease in equied mass flow ate. Subsequent analysis theefoe concentates on helium compession atios of 1 o geate. Wok (W) P2 = 12 kpa P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = P8 (kpa) T4 (K) Figue 3-3 Compession Wok vs Intemediate Tempeatue (P2=12 kpa) Figue 3.4 shows a simila plot of wok vs T4 fo a cycle that has some compession wok in the hydogen side. The hydogen compession atio is 1, and some isenthalpic cooling is occuing at the JT valve. Note the lowe helium compession atios ae not included in the figue. The plots look vey simila to the case whee thee is no hydogen compession (Figue 3-3), but thee ae some slight diffeences in values of total wok. These diffeences will be exploed a little late in Figue 3.8, when the wok is plotted as a function of the hydogen compession atio. Again, fo the low helium pessues, thee is a steep incease in wok fo vey small inceases in intemediate tempeatue, and these cycles should be avoided if possible. The highe helium compession cycles also exhibit geatly inceased wok input equiements.

64 5 P2 = 12 kpa P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = P8 (kpa) Figue 3-4 Compession Wok vs Intemediate Tempeatue (P2=12 kpa) Meanwhile, it is infomative to look at some of the cycle paametes in this case to see how the cycle functions. Figue 3-5 plots the equied helium mass flow ate though each tubine as a function of intemediate tempeatue fo a cycle with both the hydogen and helium compession atios set at 1 to 1. Note how the oveall mass flow ate inceases as the intemediate tempeatue inceases, and the high tempeatue tubine flow ate deceases fom appoximately 1% of the flow at the lowe T4 to eo at the uppe end of the T4 cuve. The seconday y-axis plots the wok input fo both the hydogen and helium. The hydogen wok is constant since the compession atio and mass flow ate ae fixed. The helium wok input inceases due to the incease in equies mass flow ate, even though the compession atio is fixed. Anothe way to evaluate the efficiency of each cycle is to look at the heat ejected by the hydogen in each heat exchange. Figue 3-6 shows the heat ejection in HX3 and HX4 fo a ange of intemediate tempeatues and helium compession atios, with the

65 P2 = 12 kpa P8 = 12 kpa P7 = 12 kpa P5 = 11 kpa η c =.8 η t =.8 P HX = 1 kpa T HX = Mass flow ate (g/s) Wok (W) T1 flow T2 flow Hydogen wok Helium wok T4 (K) Figue 3-5 Tubine Mass Flow Rates and Cycle Wok hydogen pessue fixed at 12 kpa. The uppe cuve shows the majoity of the heat being ejected in HX3, which is the highe tempeatue heat exchange. Howeve, fo wame intemediate tempeatues, the amount of heat ejected is deceased in HX3 and inceased in HX4. This fact does not change with inceases in helium cycle pessue, as is evidenced by the diffeent P8 cuves plotting on top of each othe. Anothe way of looking at this plot is to think the cycle that ejects the geatest amount of heat at the highest tempeatues will be the most efficient, and the wok cuves plotted in Figues 3-3 and 3-4 povide moe evidence of this. In the peceeding paagaphs, it is demonstated that the optimum intemediate tempeatue to have in the cycle is the lowest tempeatue allowed fo the given hydogen and helium pessues. The next step is detemining the wok equied as a function of hydogen pessue. The total amount of hydogen wok is much less than the helium

66 52 45 Q (W) P2 = 12 kpa P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = HX3 (P8=12) HX4 (P8=12) HX3 (P8=24) HX4 (P8=24) HX3 (P8=36) HX4 (P8=36) T4 (K) Figue 3-6 Heat Rejection at HX3 and HX4 wok due to the flowates involve, but an oveall advantage can occu if some level of cooling can be achieved by hydogen expansion. Again, wok input may not matte if the wok is fom the hydogen poduction pocess. Figue 3-7 shows the wok equied as a function of T4 fo a ange of hydogen pessues. The helium cycle is woking at a pessue of 24 kpa. Thee is little movement in the cuves fo the pessue atios less than 1 to 1, then gadually the cuve shifts to the highe tempeatue aeas. Then the cuves stat clusteing again above P2=3 and no futhe advantage can be gained by highe compession. Choosing the lowest tempeatue point in the cuves in Figue 3-7 obtains a plot of lowest allowed wok input fo each specific helium compession atio as a function of hydogen pessue. Figue 3-8 shows thee is an initial small spike in compession wok, followed by a decease to a local minimum, followed by a gadual incease as systems each vey high hydogen pessues. This shows thee exists an optimum hydogen

67 53 16 Wok (W) P8 = 24 kpa P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = P2 (kpa) T4 (K) Figue 3-7 Cycle Wok vs Tempeatue fo a Range of Hydogen Pessues pessue, fo each helium pessue and intemediate tempeatue, fo which the liquefaction cycle opeates at maximum themodynamic efficiency. This hydogen pessue occus at 24 kpa. Similaly, Figue 3-9 shows the cycle wok fo a ange of helium pessue atios. It is evident thee exists a helium pessue that leads to a minimiation of wok fo each set of hydogen pessues, and this pessue is below 12 kpa. This figue also shows the cuve with a hydogen pessue of 24 kpa povides the lowest wok total. Based on the above analysis, it is concluded that the poposed combined hydogen and helium liquefaction cycle is feasible, and a ange of opeating conditions exist. Thee is a minimum and a maximum intemediate tempeatue that can be used downsteam of the ecupeative heat exchange. Tempeatues highe than this ange will not wok due to HX3 becoming unbalanced as the mass flow ate though the high tempeatue tubine

68 54 goes to eo. Tempeatues lowe than the acceptable ange will not wok due to the high tempeatue tubine not being able to povide a low enough tempeatue to 95 9 P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = Wok (W) 85 P8 (kpa) Hydogen pessue (kpa) Figue 3-8 Cycle Wok vs Hydogen Pessue 95 Wok (W) P7 = 12 kpa P5 = 11 kpa η c η t P HX =.8 =.8 T HX = 1 kpa = P2 (kpa) Helium pessue (kpa) Figue 3-9 Cycle Wok vs Helium Pessue

69 55 ensue T16 is less than T4. The most efficient intemediate tempeatue to use fo a given set of hydogen and helium pessues is the lowest tempeatue possible. This maximied the amount of heat ejected at highe tempeatues. The analysis in this section also shows the functional elationship between hydogen and helium pessues and wok. It is shown the most efficient cycle will opeate with the hydogen pessue at 24 kpa, and a helium pessue at 12 kpa. The efficiency of the liquefaction system is 17%, which is espectable fo a small scale LH2 system with only 1 intemediate stage. Themodynamic Analysis of Stoage System An effot will now be made to model the themodynamic behavio of the liquid hydogen in the stoage tank fo a vaiety of opeating conditions. In many pactical applications, simplifications to the full consevation equations can be made to povide geat insight into the physical chaacteistics of a system without sacificing too much accuacy. Such simplifications allow fo closed fom analytical solutions that do not equie sophisticated numeical techniques to solve. This section will focus on applying these simplifications ove a contol volume to obtain geneal expessions fo the mass and enegy balance of the system. These contol volume mass and enegy balances will then be tailoed fo the specific opeational constaints found duing the eo boil off, liquefaction, and densification pocesses, and these specific expessions will be compaed to esults of the expeimental testing discussed in Chapte 6. Definitions and Assumptions The contol volume to be analyed is pesented in Figue 3-1. A cylindical double walled tank of volume 15 lites, with a diamete of 5.8 cm and length of 74 cm, is initially patially full of liquid hydogen in the satuated condition. The tempeatue of the tank is well below the ambient tempeatue and thee is a flow of heat into the tank

70 56 fom the suoundings though the insulation into the system. This heat tansfe can be assumed to be of two foms, the adiation heat tansfe fom the oute vessel to the inne vessel, and conduction heat tansfe down the length of the neck of the vessel and othe solid suppots. In addition, thee is a cold heat exchange integated into the vessel that can emove heat fom the liquid. Thee ae thee pots connecting the inne vessel to the suoundings. The liquid fill and withdaw pot is pesent to allow fo the flow of the hydogen fom the stoage vessel to the use point, in this case the flight tank of a ocket. The vent pot is designed to vent vapo fom the top of the tank to the ambient pessue suoundings, eithe in a contolled vent o as an emegency elief when the system pessue eaches the maximum opeating pessue. The gas supply pot is necessay to allow fo pessuiation of the fluid as well as to intoduce hydogen gas fo liquefaction opeations. The inne tank holds only hydogen in gaseous o liquid fom. The system thus defined constitutes an open, tansient, homogeneous, multiphase system with heat and wok inteactions. Despite the appaent complexity of the system as defined above, numeous simplifications can be made to allow fo faily accuate analysis. Fist, it can be assumed that the time constant associated with defining an equilibium state between the liquid and vapo phases is much less than the time constant associated with any system inteactions with the suoundings. This assumption is valid as long as the heat tansfe into o out of the system is low, as it typically is in well insulated cyogenic vessels, o the mass flow ate into and out of the system is small compaed to the total system mass. (37) This defines the system as being in themodynamic equilibium with espect to the liquid and vapo phases, and equies the liquid and vapo phases to have equal

71 57 Vent Pot (vp) M T P vp vp vp M T P gsp gsp gsp Gas Supply Pot (gsp) Q CON MLI Q RAD M Volume=f(t)=V Mass=f(t)=M Temp=f(t)=T Pess=f(t)=P fg GH2 LH2 g g g g Volume=f(t)=V Mass=f(t)=M Temp=f(t)=T Pess=f(t)=P l l l l HX Q REF U lfw M T lfw lfw Plfw Liquid Fill and Withdawl (lfw) Figue 3-1 Simplified system epesentation tempeatue and pessue. It is impotant to keep in mind that the definition of themodynamic equilibium does not imply thee is no time dependence in the system, and in this case the model is a still a tansient system. A second assumption is the liquid and vapo phase have a defined, impenetable bounday. It is assumed thee ae no bubbles in the liquid and no mist in the vapo. This allows fo modeling the system as a combination of two sepaate single-phase contol

72 58 volumes, with heat and mass tansfe between them at the liquid to vapo inteface. The oveall volume of these contol volumes will vay with time, depending on the level of evapoation o condensation at the inteface as well as the tempeatue and pessue of the liquid and vapo. Next, assumptions about the popeties of the fluids must be made. The fluid velocities ae consideed small. Effects of heat addition o emoval and mass addition o emoval ae consideed ove the entie contol volume instantaneously, and thee ae no bounday laye effects. The fluid can theefoe be assumed inviscid. Since inteactions with the suoundings ae assumed to act instantaneously, the fluid popeties ae assumed to be constant with espect to spatial dimensions in the contol volume. Consevation of Mass The geneal fom of the consevation of mass equation fo a fixed contol volume can be expessed in integal fom as (38) t v ρ dv = ρv ds 3-3 Fo the system descibed in Figue 3-1, with unifom, one-dimensional flow acoss one inlet and two outlet locations, the consevation of mass educes to S ρdv = ρ gspu gsp Agsp ρlfwulfw Alfw ρ vpuvp Avp t 3-4 v While the total contol volume is fixed, thee ae actually two volumes of inteest inside the tank that ae vaiable, the total liquid volume and the total gas volume. Beaking these up into sepaate volumes, and assuming the density is constant in that paticula volume, the consevation of mass equation becomes

73 59 ρlvl t + ρ gv t g = ρ gsp u gsp A gsp ρ lfw u lfw A lfw ρ u vp vp A vp 3-5 Fo this model, it is assumed that the themodynamic state and the total mass flow ate of the gas enteing the gas supply pot is known. The density of the vapo venting out of the tank is assumed to be equal to the density of the vapo in the ullage, and the density of the liquid leaving the tank out the liquid withdaw pot is assumed to be equal to the density of the bulk liquid in the contol volume. Using the assumption that the system can be teated as a bulk fluid, the mass addition is immediately distibuted ove the entie contol volume and the effects of the fluid velocity at the inlets ae outlets ae neglected. Beaking up the system into two contol volumes at the liquid to vapo inteface the following expessions fo the consevation of mass ae obtained ρ V ρ V = ( m& m& + m ) t 3-6 & g 2 g 2 g1 g1 gsp vp fg ρ V ρ V = ( m& m ) t 3-7 & l 2 l 2 l1 l1 lfw fg Consevation of Enegy The geneal fom of the consevation of enegy equation fo a fixed contol volume, assuming inviscid flow and neglecting gavitational potential enegy and assuming thee ae no intenal eactions that ae exothemic o endothemic, can be expessed in integal fom as (38) q ρdv PV ds + ρ( F V ) dv = v & 3-8 S v 2 2 V V [ ρ ( u + )] dv + u + V vt ρ( ) 2 2 S ds

74 6 whee the fist tem is the ate of heat added to the fluid inside the contol volume fom the suoundings, (38) the second tem is the ate of wok done on the fluid by suface (pessue) foces, the thid tem is the ate of wok done on the contol volume by body foces, the fouth tem is the time ate of change of enegy inside the contol volume due to tansient effects, and the fifth tem is the net ate of flow of enegy acoss the contol suface. Beaking down the fist tem, q& is defined as the ate of heat added pe unit mass. Physically accuate modeling of this heat tansfe would account fo the heat flux acoss a naow themal bounday laye nea the suface of the tank, howeve this contol volume model assumes bulk fluid popeties and no bounday laye effects. Futhemoe, the natue of the integated efigeation and stoage system includes two heat tansfe mechanisms, the heat leak into the tank fom the ambient suoundings as well as the heat emoved fom the system fom the efigeato. Integating the fist tem ove the entie volume theefoe gives the following simplified expession fo heat tansfe to the system as ρ v q & dv = Q Q 3-9 HL REF Body foces, namely the gavitational effects equied fo natual convection, ae an impotant physical effect that must be taken into account to accuately pedict the heat tansfe ates acoss the system boundaies. Howeve, fo this simple contol volume mass and enegy balance whee bulk fluid popeties and inviscid flow ae assumed, the wok done by body foces can be neglected, and the thid tem in Equation 3-8 can be dopped.

75 61 Assuming the fluid velocities ae negligible, the effect of kinetic enegy of the fluid can be neglected. Again assuming bulk fluid popeties, the density and intenal enegy of the fluid ae constant in the contol volume and can be taken out of the volume integal in the fouth tem. Afte integating ove the contol volume, the tansient tem the ight hand side becomes v 2 V [ ρ ( u + )] dv = mu 3-1 t 2 t The two suface integals can be evaluated togethe. This integal is evaluated only ove the inlet and outlet locations since V ds equals eo at all othe locations. Assuming bulk fluid popeties, the evaluation of these integals yields V ds + uv ds = S P ρ 3-11 S ( P V A 1 + ρ 1u1) V1 A1 ( P2 + ρ 2u2 ) V2 A2 + ( P3 + ρ3u3) 3 3 Reaanging Equations 3-9, 3-1, and 3-11, substituting in the definitions fo mass flow ate and enthalpy, beaking up this expession fo sepaate contol volumes in the liquid and vapo space, and integating ove time educes the geneal consevation of enegy equation to (39,4)) ρ v h v h = m& h t m& h t + m& h t 3-12 g 2 g 2 g 2 ρ g1 g1 g1 gsp gsp ρ v h v h = Q Q m& h t m& h t 3-13 l 2 l 2 l 2 ρl1 l1 l1 HL whee the mass flow ate between the liquid and vapo phases is found by assuming all heat leak into the tank entes the liquid contol volume and ceates an equivalent amount of boil off to the vapo contol volume. The heat leak tem is the summation of the individual heat tansfe components of conduction down the solid suppots and REF lfw vp lfw vp fg fg g g

76 62 convection and adiation thu the MLI. The selection of the vapo enthalpy as the amount of enegy leaving the liquid contol volume is somewhat abitay, and eflects the tue contol volume bounday as being dawn macoscopically in the vapo space above the liquid level. Equation of State At these low tempeatues, the ideal gas model cannot be assumed. To elate the themodynamic popeties of pessue, tempeatue, and specific volume, the National Institute of Standads and Technology (NIST) Themophysical Popeties Database RefPop was used. This database incopoates the Modified Benedict-Webb-Rubin (MBWR) equation of state fo hydogen, using eight empiical constants deived fom testing at the NIST laboatoy. (41) Once the state of the fluid is detemined, Maxwell s elations can be used to find the othe themodynamic vaiables. Opeational Simplifications Although the above equations can be used fo analysis of opeations with multiple simultaneous mass influx and outflows, such scenaios ae ae. In the vast majoity of opeations, thee will be mass influx o outflow though only one of the pots at a time, if any at all. The opeations can be classified depending on whethe the system is open o closed to mass tansfe fom the suoundings, and the diection of heat tansfe in the system. Table 3-1 shows the diffeent opeating chaacteistics of the system. If the efigeato is opeating and geate the capacity is geate than the heat leak, the heat tansfe is out of the system, and if the efigeation capacity is less than the heat leak the net heat tansfe is into the system. This leads to the following simplifications of the consevation equations, depending on the opeational scenaio.

77 63 Table 3-1 Opeational scenaios HEAT IN ADIABATIC HEAT OUT MASS IN Gas Pessuiation Liquefaction CLOSED Self Pessuiation ZBO Densification MASS OUT Boil Off o Tansfe Liquid Tansfe Liquid Tansfe Closed Stoage With Heat Tansfe To the System (Self Pessuiation) Cuent stoage systems often have peiods of time whee they ae vented and then closed fom the suoundings. Duing these peiods, the valves emain closed and the heat leak into the tank ceates an evapoation of liquid that causes a pessuiation of the system. The pessue will ise until the maximum opeating pessue is eached and the elief valve opens. Duing this time, the mass inside the entie tank is constant, but the elative amounts of liquid mass and vapo mass will change. The consevation of mass equations fo the vapo and liquid space can be expessed as ρ V V = m& t 3-14 g 2 g 2 ρ g1 g1 fg ρ V V = m& t 3-15 l 2 l 2 ρ l1 l1 Duing this peiod it is assumed the efigeato is off and the only heat tansfe in the model is the heat leak. The consevation of enegy equations fo the vapo and liquid space can be witten as fg ρ V h V h = m& h t 3-16 g 2 g 2 g 2 ρ g1 g1 g1 fg g ρ V h V h = Q m& h t 3-17 l 2 l 2 l 2 ρ l1 l1 l1 The above equations have seven unknowns ( ρ V, h,, V, h, and _ m& ). HL fg g g 2, g 2 g 2 ρ l 2 l 2 l 2 The initial tempeatue and pessue in the tank ae known giving the est of the initial fg

78 64 themodynamic state vaiables, and the heat leak is assumed known by eithe design data on the insulation system effectiveness o by expeimentation. Howeve the liquid and vapo volume ae elated by the following expession, V = V + V 3-18 system l g and the mass flow between the liquid and vapo depends on the heat tansfe by Q HL = m& h 3-19 fg fg Combing equations 3-14 thu 3-19 with the equation of state, the system of seven equations and seven unknowns can then be solved. A Micosoft Excel based pogam has been witten to model this behavio. Fist, the heat leak into the system is estimated fo the given initial conditions. This heat leak model includes the effects of adiation and convection thu the multi laye insulation as well as the effect of conduction down the solid wall of the neck and vaious inteface tubing. The adiation and convection acoss the tank suface aea follows the modified Lockheed coelation fo multi laye insulation. (42) This takes into account the numbe of layes of insulation, the density of the layes, and the bounday conditions of tempeatue. This also assumes a vacuum in the annulus of less than 1-3 To. The conduction heat leak down the length of the neck o tube can be found by the Fouie equation, except the themal conductivity will not be constant ove that wide ange of tempeatues. The themal conductivity can be integated ove a ange of tempeatues if the govening equation is known, and the themal conductivity integal has been calculated fo a ange of mateials and tempeatues by McModie. (43) These elations ae then added into the model to ceate an estimate fo heat leak into the tank as a function of geomety and themal bounday conditions. Fo this model, it is assumed the uppe bounday

79 65 tempeatue is constant at 3K, while the lowe bounday tempeatue is vaiable and equal to the fluid tempeatue at that instant. Once the heat leak is known, the total inteactions with the suoundings ae known since this is a closed system. Next, the initial tempeatue, pessue, and mass ae set, and the est of the initial conditions ae detemined. Then, the equation of state is used to find the liquid and vapo densities. Then the total volume of liquid and gas ae found using the consevation of mass and the volume elation expessed in Equation The liquid and vapo specific enthalpies ae found using the RefPop pogam, and the total liquid and vapo enthalpy is then computed. At this point all state popeties of the liquid and vapo phases ae known fo the initial condition. The total amount of enegy tansfeed into the system is then calculated by multiplying the heat leak ate by the chosen timestep. Ideally, knowing the new enthalpy in the tank, RefPop should be able to calculate the new pessue since the specific enthalpy and the density of the system ae known. Unfotunately, thee ae some themodynamic states fo cetain fluids whee convegence of the enthalpy/density specifications cannot be met by the RefPop pogam, and this poved to be the case in this situation. An altenate pocedue was developed, whee a new pessue is assumed, then new liquid and vapo enthalpies and total quality ae calculated, and the total system enthalpy is compaed to the pevious timestep plus the known heat leak. The Excel add in Solve was used to eliminate the diffeences in these system enthalpies by vaying the pessue, and a new system pessue was detemined. This model was used to pedict system pessue and tempeatue inceases ove a peiod of time fo a liquid hydogen system with a known heat leak. Figue 3-11 shows

80 66 the tempeatue and pessue fo a 15-lite hydogen dewa initially filled with liquid hydogen at the 5% mak. Notice the system tempeatue ate Pessue V=5% (5412 g) Q=f(T) Tempeatue P (kpa) T (K) Time (min) 2 Figue 3-11 Self pessuiation tempeatue and pessue vs. time of incease begins to fall while the pessuiation ate of incease gets lage as time pogesses. As the time continues, the system tempeatue inceases so the heat of vapoiation deceases, and moe of the heat leak is contibuting to latent heating of the hydogen as opposed to sensible heating. Although liquid hydogen is being conveted into a vapo, the oveall liquid level continues to ise ove time. Figue 3-12 shows the system quality and the liquid and vapo volumes espectively. Notice the faction of the vapo in the system inceases in a simila manne to the pessue, again this is due to the decease in the heat of vapoiation. But due to a decease in density in the wame liquid, the oveall liquid volume is inceasing. The coesponding decease in vapo volume also plays a ole in the pessuiation ate incease, although the vapo density is also inceasing. This incease in liquid volume is citical to the design of dewas especially when they ae nea

81 P1=11 kpa T1=2.2 K Liquid Volume Volume (cm3) Quality.6.4 Quality Vapo Volume Time (min) Figue 3-12 Self pessuiation tank quality and phase volume vs. time thei 1% full level, since futhe inceases in tempeatue will lead to themal expansion of the incompessible liquid, and a apid ovepessuiation may occu. A compaison of system pessuiation ates is shown in Figue 3-13 fo thee diffeent liquid levels. It seems intuitive that the system that is most full will pessuie moe quickly since thee is less vapo space to fill. Howeve, this poves not to be the case. This is explained by the fact that the geate the liquid quantity in the tank, the moe of the heat leak is absobed by sensible heat of the liquid and less goes to vapoiation. Fo example, the liquid absobed 97% of the heat leak into the system as sensible enegy fo the tank that was 75% full, but only 78% of the heat leak fo the 25% full case. Figue 3-14 shows the behavio of the system as pedicted by the model compaed to the expeimental data. The data is fom a eo boil off test conducted ovenight with the cyocoole tuned off, afte liquefaction opeations intoduced 1.25 kg of hydogen into the tank. The initial pessue was measued at 33 kpa. As is evident in the gaph,

82 Pessue (kpa) % full 5% full 75% full 1 5 T1=2.3 K Q=f(T) Time (min) Figue 3-13 Self pessuiation ates fo vaiable liquid level 3 25 T1=17.6 K Liquid volume =16 L Pessue (kpa) Model Expeiment Time (min) Figue 3-14 Self pessuiation (model vs. data) the pedicted system pessuiation ate is within 15% of the actual test data. This discepancy is explained by the assumptions that the tank is modeled as an isothemal system, with the tempeatue at the top of the tank being equal to the tempeatue at the bottom (liquid tempeatue). The heat leak estimate assumes the tank is nealy full, so it

83 69 gives the heat leak ate fo tanks in this condition. In this instance, 1.25 kg of hydogen only had a liquid level of 16 lites, and thee was a lage tempeatue gadient down the length of the tank, educing the heat leak compaed to the model pediction. Coections to this model will be made in late sections to account fo vaiable liquid levels. Open Stoage With Heat Tansfe To The System (Boil off) Thee may be peiods of time when the heat leak into the tank ceated a pessue ise that causes the tank elief valve to open, ceating a boil off loss. Duing these times, it is assumed the system is closed to gaseous mass input o fom liquid mass output. Typically duing boil off situations, the tank pessue will emain constant, at the set pessue of the elief valve. Some systems with no backpessue vent diectly to atmospheic pessue. This implies the liquid tempeatue inside the tank will also emain constant, at the satuation tempeatue coesponding to the tank pessue, and so the liquid density is also constant. Likewise, the vapo tempeatue and density can assume to be constant, although thee will be some small incease in density as the liquid boil off ceates a foced convection cuent that cools the uppe pat of the vapo volume. The consevation of mass equations then educe to ρ ( V V 1) = ( m& + m ) t 3-2 g & g 2 g vp fg and the consevation of enegy equations educe to ρ ( V V 1) = m& t 3-21 l l 2 l fg ρ h ( V V 1) = m& h t + m& h t 3-22 g g g 2 g vp vp ρ h ( V V 1) = Q m& h t 3-23 l l l 2 l HL fg g fg g

84 7 The above equations have fou unknowns and can be solved easily. Note the equation of state is not necessay since the state of the fluid is assumed constant with the isobaic venting pocess. A Micosoft Excel speadsheet has been developed to model this boil off pocess. Initial conditions and system constaints ae fist detemined. Fo this example, the initial system mass must be input, and the elief valve set pessue is detemined. Using these paametes with the system volume, the initial tempeatue, liquid and vapo mass, density and volumes, and heat of vapoiation ae calculated. The system heat leak is calculated fo this tank tempeatue. Then, the mass of the liquid boil off is calculated, and the new liquid mass and volumes ae found. Next, the new vapo volume is calculated and using the constant vapo density, a new vapo mass is found. Knowing the initial and final vapo and liquid masses, the mass of the vent loss is calculated. This mass will always be slightly less than the mass of the liquid boil off, since some boil off vapo must fill in the additional volume lost by the liquid. Figue 3-15 shows the values of liquid and vapo mass in a 15 lite dewa filled halfway with liquid hydogen at atmospheic pessue. Notice the liquid mass dops at an appoximately linea ate, since the heat leak is a function of the constant liquid tempeatue. The vapo mass inceases slightly ove the boil off pocess. Figue 3-16 compaes the time it takes to completely evapoate a given volume of liquid depending on the boil off pessue. Hydogen stoage losses occu moe quickly when the vapo pessue is elevated, mainly due to the decease in heat of vapoiation, but also patly due to the geate mass faction of hydogen that is in the vapo phase initially.

85 P1=11 kpa T1=2.3 K Mass (g) 3 2 Liquid Vapo Time (h) Figue 3-15 Liquid and vapo mass duing boil off 5 Liquid Volume = 75 L 4 Liquid Mass (g) 3 2 P=1 atm P = 3 atm P = 5 atm P =.5 atm Time (h) Figue 3-16 Boil off time dependence on pessue The simplification made that the liquid and vapo tempeatue and pessue wee unifom thoughout the volume makes this boil off model less pecise, in actuality thee will be tempeatue gadients in the ullage space and down the tank walls that will decease the total heat tansfe to the liquid as the liquid volume dops. Thee is no expeimental data

86 72 in this wok to compae this model to, since no boil off occued duing the liquid hydogen expeimentation potion discussed in Chapte 6. Closed Stoage With Zeo Net Heat Tansfe (Zeo Boil Off) The ZBO opeation is a closed stoage system with efigeation povided to emove heat leak into the tank. The state of the fluid will depend on the capacity of the efigeation system. Thee exists a point on the efigeato capacity vs. tempeatue cuve whee the capacity is exactly equal to the heat leak into the tank. As such the system can be modeled as steady state and the consevation equations can be educed to V = V 3-24 g 2 g1 V = V 3-25 l 2 l1 Q HL = Q REF 3-26 As given by Cyomech, the AL33 pefomance cuve is plotted with the heat leak vs. tempeatue cuve below in Figue Afte cuve fitting the pefomance cuve, 1 Heat Tansfe Rate (W) ZBO Data Points Pedicted Cyocoole Pefomance (63) Pedicted Heat Leak (59) Tempeatue (K) Figue 3-17 Cyocoole pefomance and heat leak vs. tempeatue

87 73 the equations can be set equal to each othe and solved fo the intesection, which is K. This system tempeatue can be consideed analogous to a cyocoole no-load tempeatue. The esults of this closed system enegy balance deived in this section agees vey well with the expeimentation value of no load system tempeatue, anging fom K to K, obtained duing seveal densification tests. These data points ae plotted on the cyocoole cuve and ae labeled below Open System With Heat Tansfe Out Of The System (Liquefaction) Duing liquefaction opeations it is pefeable thee is no mass outflow fom the tank, so thee is always an incease in oveall system mass. The efigeation system must be opeating (except when stoed efigeation enegy in a subcooled liquid can tempoaily liquefy a small quantity of gas) to emove the enthalpy flowing into the gas supply pot. The consevation equations can be simplified as follows; ρ V ρ V = ( m& + m ) t 3-27 & g 2 g 2 g1 g1 gsp fg ρ V V = m& t 3-28 l 2 l 2 ρ l1 l1 fg ρ V h V h = m& h t + m& h t 3-29 g 2 g 2 g 2 ρ g1 g1 g1 gsp gsp fg ρ V h V h = Q Q m& h t 3-3 l 2 l 2 l 2 ρ l1 l1 l1 At fist it is tempting to conside the liquefaction model to be equivalent to the boil off model as they both ae open systems with net heat exchange, but instead of heat tansfe into the tank and mass tansfe out of the tank, the liquefaction system opeates in the opposite diection. Howeve this is not accuate, as thee ae additional simplifications in the boil off model imposed by the elief valve, namely that the pessue in the tank is constant. Depending on the gas supply pot mass flow ate and the net amount of efigeation, the tank pessue may incease, decease, o emain constant. But HL REF g fg g

88 74 duing opeations whee the ullage pessue is inceasing, gas is enteing the tank at a faste ate than the cyocoole capability to emove its enegy, and a non-satuated vapo condition occus. This phenomenon is not descibed accuately by this model, which assumes themodynamic equilibium between the liquid and vapo phases. The model has been modified to accept these conditions and pedict the system liquefaction ate to maintain themal equilibium fo a given set of initial conditions. The net heat tansfe out of the system fom the combination of heat leak and cyogenic efigeation is then calculated fo that given tempeatue condition. To maintain equilibium this must match the incease in system enthalpy fom the gas supply pot. Knowing the conditions of the inlet steam and the final liquid state, the change in enthalpy equied fo liquefaction is calculated. The appopiate mass and vapo volumes ae then found. Figue 3-18 shows pedicted liquefaction ates fo a ange of stoage states. This chat shows the liquefaction ate inceases as expected, when the stoage pessue is highe. This is due to the satuation tempeatue coesponding to this pessue being highe, which has thee effects. Fist, the cyocoole opeates moe efficiently at highe tempeatues and thee is geate cooling powe available. Second, the amount of enthalpy equied to be emoved fom the incoming gas is less. Finally, the net heat tansfe fom the suoundings is slightly less since thee is a smalle themal gadient. Figue 3-18 also shows the expeimental data fo thee peiods whee the mass flow ate was constant and sufficient to maintain a constant pessue in the tank. This compaison shows a geneal coelation in the shape of the pofile, but the esults vay by as much as +/- 28%. The discepancies between the pedicted vs. actual liquefaction

89 75 ates can be patially explained by the expeiment not taking into account the heat of convesion between otho and paa hydogen. The model assumes equilibium hydogen but thee is a finite time equied to change the nomal hydogen to equilibium hydogen. Othe factos include the actual expeiment was not an isothemal system and thee wee tempeatue statifications that the model did not include. An additional cause of the discepancy is the uncetainty in the measuements, since the exact state of the incoming gas was not known and thee was no expeimental method to contol the Flowate (slm) Pessue (kpa) model expeiment 1 expeiment 2 expeiment 3 Figue 3-18 Compaison of pedicted vs. actual liquefaction ate pessue and mass flow ate. In these cases it can be seen that while the pessue and mass flow ate was elatively constant, thee wee vaiations in the measuements that implied the system was not tuly in themodynamic equilibium. These uncetainties ae plotted as x and y eo bas in the individual liquefaction expeiment data points. In some cases, vaiations in the constant mass flow ate wee as high as 8.7%, and pessue fluctuations wee up to 2.2%. Clealy, moe data needs to be collected, and with bette contol and fidelity.

90 76 Closed Stoage With Heat Tansfe Out Of The System (Densification) A closed system with heat tansfe out of the system by efigeation is simila to the closed self-pessuiation case with the exception the pessue and tempeatue will decease until the system eaches steady state and the efigeation capacity balances out the heat leak into the tank. In the meantime, the mass of the liquid and vapo will change as some vapo is condensed out of the ullage space. The consevation equations can be witten as ρ V V = m& t 3-31 g 2 g 2 ρ g1 g1 fg ρ V V = m& t 3-32 l 2 l 2 ρ l1 l1 fg ρ V h V h = m& h t 3-33 g 2 g 2 g 2 ρ g1 g1 g1 fg g ρ V h V h = Q Q m& h t 3-34 l 2 l 2 l 2 ρ l1 l1 l1 Again this is a system of seven unknowns and seven equations (including the volume elation, the mass tansfe between the vapo and liquid elation, and the equation of state), and can be solved in a simila manne. The same Excel pogam was used to estimate densification ates as the self pessuiation ates, with the exception that the heat tansfe ate peviously given by the heat leak estimation is now the diffeence between the heat leak and the efigeation capacity of the cyocoole at the system tempeatue. Results of this model ae shown below. Figue 3-19 shows the tempeatue and pessue of the system with a tank initially 5% full of LH2 at the nomal boiling point. The pessue and tempeatue decease moe apidly at fist since the cyocoole is poducing moe efigeation at highe tempeatues. Eventually the pessue and tempeatue each the steady state condition HL REF fg g

91 Liquid Volume = 75 L Pessue (kpa) Tempeatue (K) Pessue Tempeatue Time (h) Figue 3-19 Pedicted densification tempeatue and pessue defined by the eo boil off model. Figue 3-2 demonstated the pedicted densification times fo thee diffeent initial liquid levels anging fom 25% to 75% full. As expected the smalle the mass in the tank, the less time it takes fo densification. Some launch 21 2 P1=11 kpa T1=2.3K Tempeatue (K) Volume 25% 5% 75% Time (h) Figue 3-2 Densification ates fo vaiable liquid levels

92 78 scenaios may depend on on-boad densification, and pedictive models will need to be developed to estimate countdown hold times fo densification Figue 3-21 compaes tempeatue data obtained duing expeimentation with the simplified model. Note the pedicted densification time pofiles vay widely; this is explained by two factos. Fist, non-equilibium pocesses ae occuing in the tank whee the ullage space is developing significant tempeatue statifications. Some cyocoole capacity must be used to chill this vapo back to the satuated state. Refe to Figue 6-14 on page 149 fo details on the themal pofiles in the tank duing this densification pocess. Second, simila to the pessuiation model, heat tansfe pocesses ae modeled using a unifom tempeatue in the tank and the heat leak and cyocoole pefomance will vay in those conditions P1=18 kpa M L =.98 kg 22. Tempeatue (K) Model Data Time (h) Figue 3-21 Pedicted vs. actual densification ates Based on the above discussions, it appeas the model has two majo components of eo. Fist, the isothemal model is not entiely accuate as themal statification in the tank will cause vaiations in heat leak. Second, thee ae non-equilibium pocesses

93 79 happening duing many opeational scenaios, especially when the liquid and vapo states ae not fully satuated. An attempt to ceate a coection facto fo these two situations will be made in the following section. Coected model Vaiations in the pedicted heat leak will occu since the model used assumes an isothemal tank and the physical system has tempeatue statifications inside the vapo egion. These statifications will cause wame tempeatues at the top of the tank, minimiing the tempeatue gadient and educing heat leak fom the outside. A coection to the isothemal model developed above will be poposed to account fo this vaiation. The coection facto will consist of two components. Fist, an adjustment must be made to account fo vaiations in the tank liquid level. The modified Lockheed Matin coelation fo MLI pefomance depends on the tank being at 1% full and at a constant tempeatue. Actual tanks will have a tempeatue gadient up the vetical walls of the tank between the liquid level and the top of the tank. This means the delta T between the hot and cold tempeatues will be less than 1% fo a potion of the tank, especially any wam suppots and tubes that connect to the top of the tank. Boil off testing at KSC has shown heat leak is appoximately a linea elationship with liquid level, so the fist component of the heat leak coection will be of this fom. The second component of the coection facto must take into account tansient effects associated with the geneation o destuction of themal statification layes inside the system. This coection will be applied duing opeational models when the state of the liquid and vapo in the tank is changing. Fo the cases above, this includes pessuiation and densification opeations. Duing boil off venting, steady state eo boil off, and steady liquefaction the themodynamic vaiables exhibit no time

94 8 dependence, since thee is no net heat tansfe to o fom the fluid at these times. Duing the steady boil off o liquefaction the enthalpy flow into o out of the tank cancels the heat leak o efigeation, and duing eo boil off the efigeato is balanced by the heat leak. Duing the pessuiation case, a net heat leak into the system ceates a tempeatue change is the fluid, while duing densification the net efigeation powe has the opposite effect. Looking at an enegy balance in the system duing this tansient opeation shows a geneal elationship fo the heat leak into the tank and a tempeatue change accoding to the elation h ρ = k T 3-35 t Integating this equation with espect to time gives a function fo heat leak that is dependent on the time and the initial volume. To detemine exact tue function of this elationship, this diffeential equation can be solved. The bounday conditions fo this case would be a vaiable tempeatue at = (liquid level), vaiable tempeatue along the walls of the tank and a vaiable tempeatue at the top of the tank. This would also need to be modeled with a vaiable themal conductivity and specific heat, and enthalpy would need to be expessed in tems of the tempeatue and pessue in the tank. This execise is beyond the scope of this simplified model coection. To detemine a geneal fom of this coection tem, it is useful to examine the pofiles of ullage tempeatues in Figue Data taken duing self pessuiation tests that show themal statifications in the liquid level give an indication a logaithmic elationship of heat leak with espect to time. Adding a time dependent logaithmic coection in combination with the linea dependence on liquid level gives a coected heat leak of the fom

95 81 Q coected = ( A + ( Q + Q A) B) C Ln( D t) 3-36 LM CON whee A is the heat leak in the tank as the liquid level appoaches eo, Q LM is the pedicted heat leak fom the MLI coelation, Q CON is estimated conduction heat leak, B is the liquid level (%), and C and D ae constants used to fit the model data to the expeimental cuve. The coected model (model B) shown below uses the following values; A=5, B=1% (based on conditions used in expeiment), C=.19, and D= 1,5. Figues 3-22 and 3-23 below show the esults of this coected model fo the densification and selfpessuiation pocesses as compaed to the expeimental data. An additional day of data was added to each figue, and this data is compaed to the model as well. The added pessuiation data has the same initial conditions as the fist, and the cuves plot on top of each othe vey neatly. The added densification data was fom testing seveal days afte the pevious data cuve, and the initial mass and tempeatue change. In this case, the coected model was un again and the cuve is plotted against the data. In both the pessuiation opeation and the densification opeation, the coected model is a bette pediction tool than the oiginal isothemal model. Howeve, in ode to get a tue model of the system behavio the full consevation equations must be discetied and solved numeically. Stoage Fluid Analysis In ode to moe accuately model the behavio of the hydogen in the system, the full consevation of mass, momentum and enegy equations must be consideed. The system descibed by Figue 3-24 is an unsteady, open system that will be appoximated in

96 P1=18 kpa Mass=.98 kg Tempeatue (K) model A (6-25) data (6-25) model B(6-25)) data (7-16) model B(7-16) P1=78 kpa Mass=1.22kg Time (h) Figue 3-22 Coected model vs. densification data 3 25 Pessue (kpa) T1=17.6 K Mass=1.22 kg model A (7-15) data (7-15) model B(7-15) data (7-16) Time (min) Figue 3-23 Coected model vs. pessuiation data thee-dimensional space using cylindical coodinates. Due to symmety about the axis, dependence in the θ diection can be neglected. Gavitational effects must be taken into account. The oveall system bounday is set as the inside wall of the inne tank, defined as =R. The uppe height of the tank is at the coodinate =Z. Thee ae thee modes of

97 83 heat exchange between the system and the suoundings, adiation on the oute sufaces though the MLI, conduction down the length of any pipe o tube suppot between the inne and oute vessel, and efigeation by the cold heat exchange of the cyocoole. Thee is mass tansfe in and out of the tank fom the gas supply line and the vent line. The velocity pofiles at the tank inlet and outlet will depend on the geomety, and fo now a fully developed lamina flow pofile diven by the pessue gadient can be assumed at the inteface. Thee is a longitudinal and adial velocity component in both the liquid and vapo space. The liquid velocity is diven by the fee convection cuents at the heat exchange sufaces, with a ciculation patten of wam updafts along the walls and a cold downdaft in the cente at the cold HX. This diffes fom conventional cyogenic dewas that only have the wam upwad velocity with a laye of wam fluid at the top of the liquid. Theefoe the liquid velocity will have a supeimposed feesteam velocity fom the ciculation cuents as well as the fee convection velocity. One impotant esult of a full numeic analysis is to detemine whethe this feesteam velocity will dive the heat tansfe coefficient highe than othewise, and hence incease the amount of heat tansfe fom the suoundings. If so, anti convection baffles may be added. The vapo velocity will be diven by wam convection cuents as well as the velocity in and out of the supply and vent line. The themodynamic state of the liquid and gas is vaiable, dependent on both time and location inside the tank. The total mass of the liquid and vapo will depend on the initial conditions as well as mass flow into and out of the tank and evapoation o condensation at the liquid to vapo inteface. The liquid to vapo inteface is assumed to be a fee liquid suface, with homogenous phases above and below, and the location of the inteface in the diection and the pofile will

98 84 Vent Pot M T P vp vp vp U vp M T P gsp gsp gsp Gas Supply Pot Q CON U gsp MLI FREE CONVECTION CURRENTS Q RAD Mass=f(t) Temp=f(,,t) Pess=f(,,t) Mfg GH2 LH2 FREE CONVECTION CURRENTS Volume=f(t) Mass=f(t) Temp=f(,,t) Pess=f(,,t) h HX Q REF TEMPERATURE VELOCITY BOUNDARY LAYER U lfw M T lfw lfw Plfw Liquid Fill and Withdawl Figue 3-24 Poposed system epesentation vay. This location will be denoted as =h. Moe details on this inteface will be given when the bounday conditions ae discussed. To undestand the fluid behavio in ode to model it, a consideation of the flow egime must be taken into account. While the exact flow velocity is unknown, ode of magnitude estimates can be made to find a epesentative Reynolds numbe. A potential

99 85 and kinetic enegy balance of the fluid gives a velocity of the magnitude (44) ρ 1 2 V { gl } 3-37 ρ The velocity depends on the chaacteistic length of the system and the elative change in density, which depends on the tempeatue gadient. An enegy balance between the heat leak thu the insulation compaed to the heat tansfe to the liquid allows fo a ough calculation of the tempeatue diffeence between the wall and the bulk fluid. Using the heat tansfe coefficient found in Chapte 6, fo steady state conditions this delta T will be between.1 and.4 K. This estimate compaes well to the expeimental data found duing boil off tests conducted with liquid nitogen at KSC. (45) Duing some tansient opeations this tempeatue gadient will be geate. Vaying the length scale of the heat tansfe and the delta T, the liquid vetical velocity anges between 2.5 and 35 cm/sec. The vapo velocities ae slightly highe. The dependency of the chaacteistic velocity as a function of tempeatue diffeence is shown in Figue The liquid velocities ae plotted fo cooling of the hydogen by the cold heat exchange, while the vapo velocities ae fo the waming of the vapo by the tank walls. The length scale is 35 cm, which is appoximately with the tank half full. At fist, this fee convective velocity seems vey high, but an examination of the fluid popeties shows this is feasible. The volumetic coefficient of themal expansion, β,fo liquid hydogen at 2 K is.164 K -1. This is two odes of magnitude highe than β of wate, which is 2.15 x 1-4 K -1. Similaly, the bulk coefficient of gaseous hydogen is 17 times geate than that of steam. Small tempeatue diffeences in hydogen systems cause lage displacements and hence, high velocities.

100 P1=11 kpa T1=2.3 K L=35 cm Velocity (cm/s) Liquid Vapo Delta T (K) Figue 3-25 Pedicted Fee Convection Velocity Using the density and viscosity of both tiple point and NBP liquid hydogen, a ange of estimates fo Re can be computed. The chaacteistic length used is the height of the liquid and fo the figue below is set at 5% full. Figue 3-26 plots this estimate of Reynolds numbe fo a ange of tempeatue gadients. Fom this figue, it appeas the Reynolds numbes may be quite high at times, like when the tempeatue gadients ae highe duing tansient opeations. Afte the system has had time to appoach steady state conditions, the delta T will be less and the liquid and vapo Reynolds numbe will appoach 8x1 4 and 4x1 4, espectively. In this case the flow inetia is moe modeate, and viscous foces play a geate facto in the flow. Howeve the Reynolds numbe is still high enough that the flow can be modeled with a viscous bounday laye and an inviscid coe. This pemise will be evisited in a few sections when an ode of magnitude analysis on the consevation of momentum equation is addessed.

101 87 6.E+5 P1=11 kpa T1=2.3 K L=35 cm Re 3.E+5 Liquid Vapo.E Delta T (K) Figue 3-26 Pedicted Fee Convection Reynolds Numbe Pehaps a bette epesentation of the flow egime is given by the Gashof numbe. This non-dimensional numbe is defined as 3 gβ ( T T ) x G x = = Re 2 υ 2 x 3-38 And is a measue of the fee convective buoyancy effects vs. the viscous effects. Again, this value will depend on the exact conditions, but a epesentative value at 2K fo the liquid and vapo space fo a ange of tempeatue gadients has been plotted in Figue Again, duing tansient opeations with highe gadients, the flow is moe tubulent, but nomal opeations with smalle delta T the flow egime is lamina, especially in the vapo egion. The high value of the bulk coefficient accounts fo the high G. Typically the tansition to tubulent flow occus aound G=1 9.

102 88 3.E+11 P1=11 kpa T1=2.3 K L=35 cm G 1.5E+11 Liquid Vapo.E Delta T (K) Figue 3-27 Pedicted Gashof Numbe The above analysis is a ough scaling analysis of the type of flow egime expected to occu, and is useful fo late analytical solutions based on heat tansfe coelations. Howeve, to tuly model the system behavio, an exact numeical solution to the full Navie Stokes equations is equied. Fotunately the availability of moden fast compute pocessing makes this endeavo moe feasible than in the past. The following section deives the equations that must be solved. The following consevation equations will be consideed in diffeential fom. Thee ae numeous texts that deive the full consevation equations stating fom an infinitesimal, fixed contol volume, (46-49) and this effot will not be epeated hee. The full two dimensional consevation equations fo unsteady flow of a Newtonian fluid in cylindical ae stated below, followed by addition of some simplifying assumptions. The vaiations in the θ diection ae neglected due to symmety about the axis.

103 89 Consevation of Mass The Consevation of Mass can be expessed in diffeential fom as ρ 1 + ( ρv t ) + ( ρ v ) = 3-39 whee the fist tem is the ate of change of mass inside the element and the next two tems ae the net mass flow into o out of the element fom the and diections espectively. This equation is applicable fo all fluids and assumes the density is vaiable. Typically fo cyogenic applications, a lage ange of tempeatues does not allow fo system analysis using constant popeties. In this case, liquid densities can vay 18.8% between the tiple point and 25 K, and vapo densities vay 46% between the NBP and the ambient tempeatue. Theefoe this system will not be modeled with a constant density as the density will vay with espect to time, dependent on the tempeatue and pessue vaiations. Tempeatue gadients will be pesent in the adial diection diven by heat flux fom the wall and convection cuents and the axial diection due to buoyancy effects. The local velocity field will be diven pimaily by natual convection, although duing peiods of filling o daining the tank the local velocity will be diven by pessue gadients. It is anticipated the velocity will be dependent on both spatial coodinates inside the system. Reaanging the equation to do an ode of magnitude analysis gives ρ ρv = t + v ρ v + ρ v + ρ + v ρ 3-4 In ode fo this equation to be valid, the ight hand side must be the same ode of magnitude as the left hand side, so the LHS ode of magnitude is elevant. Fo this enclosed system, the ode of magnitude of the adial velocity must be the same as the

104 9 axial velocity (ode of V), and knowing the geomety of the tank detemines the ode of magnitude of the adial and axial diections ae the same (ode of L). Beaking down the equation and dividing by V/L, the ode of magnitude of the tems on the RHS become ) ( ) ( ) ( ) ( ) ( ρ ρ ρ ρ ρ O O O O O 3-41 Fo the vapo egion, a eview of themodynamic vaiables as a function of tempeatue shows the ode of magnitude of the density change is the same as fo the total density. Fo the liquid egion, the density changes ae an ode of magnitude less than the total density, and fo a simple model with less accuacy equied, the ρ and ρ tems could be neglected. Howeve, if an accuacy geate than 9% is desied, these tems must be included. Assuming bounday laye scaling (V <<V, <<), the second tem on the RHS can be neglected. In this case even though ρ ρ, the inequality V <<V still applies and the tem is not the same ode of magnitude as the othes. In this case (if bounday laye appoximations ae used), the consevation of mass educes to v v v v t = ρ ρ ρ ρ ρ 3-42 Consevation of Momentum The geneal consevation of momentum of a 2 dimensional viscous, unsteady flow fo the and diections can be expessed in diffeential scala fom as )} ( { ))} ( 3 1 ( {2 1 ) ( v v v v v v P g v v v v t v = + + µ µ ρ ρ 3-43

105 91 v v ρ( + v t 1 v + { µ ( v P + v ) = ρg v v + )} + {2µ( 1 ( 3 v v + v + ))} 3-44 whee the fist tem on the left hand side is the time ate of change of momentum inside the fluid element and the next two tems ae the momentum flow into the element fom the and diections. The ight hand side of the equation is the summation of foces on the element, with the fist tem being the gavitational foce (othe body foces such as electomagnetic ae neglected), the second tem is the pessue acting on the nomal suface, and the thid and fouth tems ae the viscous dissipation foces. At this point it is tempting to simplify this equation by assuming constant fluid popeties, educing the viscous tems to a much simple expession of µ 2 V. In many cases this could be a poo assumption, since the liquid hydogen viscosity is highly tempeatue dependent. Indeed, the viscosity of liquid hydogen will vay 67% between the nomal boiling point and the tiple point. Changes in the fluid popeties will occu ove time, as the tempeatue of the fluid inceases o deceases depending on the level of heat flow into o out of the tank. The vapo density and viscosity exhibit simila dependency on tempeatue. This dependency is known fom themodynamic elations and expeimental data, and the function µ = f ( P, T ) can be numeically diffeentiated in the solution pocedue. The pevious section discussing the consevation of mass also detemined the system couldn t be assumed to have constant density. Anothe common assumption, the Boussinesq appoximation, cannot be used since density changes ae impotant in all tems and the viscosity vaiations ae of the same ode of magnitude as the density vaiations. The fact that the density and viscosity ae vaiable makes these

106 92 Navie Stokes equations much moe complicated, and equies a solution to the enegy equation that is coupled to the momentum equation. An ode of magnitude analysis, simila to the one done fo the consevation of mass, will help validate the assumption that the viscosity cannot be consideed constant. Again assuming O( )~O()~L and O(v )~O(v )~V, the ode of magnitudes of the viscous tems in the LHS can be expessed as 1 { L µ L L V L V + µ L V + µ L L µ L V L V L V L µ µ L } 2 2 L µ V + L L V + µ L 2 µ V + L L V + µ L 2 V Fom this it is appaent that all the viscous tems in this equation ae of the ode µ o 2 L V µ L 2. Looking at the themodynamic and tanspot popeties of hydogen vapo, the µ between the NBP at ambient is 7.68 µ Pa sec while the actual viscosity is 1.7 µ Pa sec. Similaly, µ =12 µ Pa sec and µ =13.3 µ Pa sec fo liquid hydogen between the NBP and 14 K. This means O( µ )~O( µ ) and the assumption of constant viscosity cannot be used to simplify the momentum equation. Anothe simplification that is fequently taken into account ae bounday laye appoximations, whee an ode of magnitude analysis of diffeent tems in the consevation equations eveals some tems that ae so small they can be consideed negligible in that thin egion. The thickness of a bounday laye in fee convection flows can be scaled (5) on the ode of P = 3.93 ( ) G 2 x δ x P 3-45 The pedicted Gashof numbe fo liquid hydogen was peviously plotted in Figue 3-29 fo a ange of conditions. Fom this plot it is appaent the Gashof numbe is

107 93 between the ode of G =1 9 and 1 11., depending on the tempeatue diffeence between the wall and the bulk fluid. The liquid Pandtl numbe at the NBP is 1.22 while the vapo Pandtl numbe is.77, implying the elative magnitude of the viscous and themal bounday layes ae of the same ode. A quick estimate shows that the scale of the bounday laye thickness is δ =.22x, two odes of magnitude less than the length scale of the system. Knowing the bounday laye appoximations ae acceptable fo the thin viscous egion nea the tank sufaces, the consevation of momentum equations fo this egion can be expessed as P = 3-46 v v v P v ρ ( + v + v ) = ρg + ( µ ) 3-47 t Now that viscous effects ae accounted fo in the thin bounday laye egion, an ode of magnitude analysis should be done to detemine if viscosity is impotant in the est of the flow. The balance of foces in the momentum equation shows the inetial foces scale on the ode of V ~, the pessue foces scale to L F I 2 F P P ~, and the ρ L V viscous foces scale to F V ~ υ. Knowing estimates fo the chaacteistic length, L 2 velocity, kinematic viscosity and density, an ode of magnitude analysis can be pefomed. This esults in the appoximation, in the feesteam outside the bounday laye, that the O(F v )<<O(F I ). Of couse, this eveals the same thing as the calculation of the Reynolds numbe pefomed peviously. Neglecting viscous effects in the feesteam gives the following equations fo consevation of momentum.

108 94 P v v v v t v = + + ) ( ρ 3-48 P g v v v v t v = + + ρ ρ ) ( 3-49 Consevation of Enegy The fist Law of themodynamics applied to a diffeential fluid element can witten as ) ( 3 1 ) ( 2 1 ) ( ) ( ) {( 2 ) ( v v v v v v v v T k T k T k P v P v t P h v h v t h = + + µ ρ 3-5 whee the fist tem on the left hand side epesents the ate of change of enegy in the element ove time, and the next two tems ae the convective flow of enegy into o out of the element. The fist thee tems on the ight hand side is the wok done on the system by the pessue foces, and the next thee tems is the ate of heat tansfeed into the system though the contol suface via conduction. The final tems on the ight hand side ae the ate of enegy dissipation via viscosity. Pevious analysis has shown that the viscous tems cannot be neglected, o even simplified by assuming constant viscosity, in the bounday laye egion. Howeve, an ode of magnitude analysis can educe the complexity of the equation in the bounday laye. Knowing v <<v and <<, and fom the momentum equation = P, the consevation of enegy can be educed to 2 ) ( ) ( v T k T k P v t P h v h v t h = + + µ ρ 3-51

109 95 In many applications, if Fouie s law is assumed, and the themal conductivity is assumed to be constant, the heat tansfe tem can be educed to k 2 T. Howeve, like the density and viscosity, the themal conductivity is too dependent on tempeatue to be assumed constant. Thee is no way to simplify this tem, but if the natue of the themal conductivity dependence is known, it can be numeically diffeentiated. This equation adds anothe themodynamic vaiable, namely the enthalpy, to the list of unknowns. Howeve, enthalpy can be expessed as a function of the othe themodynamic vaiables in a numbe of ways. Most often, the ideal gas appoximation is used, but obviously that cannot wok in this instance. Howeve, fo all substances, enthalpy can be expessed as dh = c p dt + ( 1 βt ) dp 3-52 Outside the bounday laye, inviscid flow can be assumed, based on the same assumptions discussed above fo the momentum equations. Howeve, the bounday laye appoximations cannot be used. In this case, the consevation of enegy educes to Summay h h h P P P T k T T ρ ( + v + v ) = + v + v + k + + k 3-53 t t The peceeding section pesented the consevation equations fo unsteady, viscous, compessible flow with vaiable tanspot popeties fo cylindical coodinates in the and diection. The flow is consideed axisymmetic and thee is no dependence in the angula diection. It has been shown though ode of magnitude analysis of individual tems and scaling of competing foces that this system can be modeled as a whole by consideing fou sepaate egions. Fist, the liquid and vapo egions must be consideed sepaately. Then, in each of these egions, thee is a thin bounday laye whee viscous

110 96 dissipation must be consideed but many tems will be negligible due to the nomal bounday laye appoximations. Inside this bounday laye egion, fo both the liquid and vapo flow, the following equations must be solved; Consevation of Mass Eqn 3-41 Consevation of Momentum in diection Enq Consevation of Momentum in diection Enq Consevation of Enegy Eqn Outside the thin bounday laye, the flow can be modeled as inviscid. No bounday laye appoximations ae used. The following equations apply in this egion; Consevation of Mass Eqn 3-4 Consevation of Momentum in diection Enq Consevation of Momentum in diection Enq Consevation of Enegy Eqn Solution Pocedue In geneal the above fou consevation equations contain 6 unknowns in each egion ( P, T, ρ, h, v, v ). This is assuming the tanspot popeties (themal conductivity, viscosity) ae known fom expeimental data as a function of tempeatue and pessue. Two additional elations ae needed to solve this system of equations. Fist, the equation of state can be used to elate the themodynamic vaiables P, T, and ρ. The ideal gas equation of state will not be accuate in this egion. NIST has developed a hydogen themodynamic popety datatbase using the Modified Benedict- Webb-Rubin (MBWR) equation of state fo hydogen, and this can be used fo the cuent analysis. Indeed, in the fist pat of this chapte, this NIST database is used to

111 97 help close the system of simplified equations. A second themodynamic elation elating the enthalpy of a fluid to the tempeatue and specific heat can also be used, but cae must be taken to incopoate vaiations in the specific heat as a function of tempeatue. This elation was pesented by equation 3-52 when the consevation of enegy was discussed. Moe specific, this system is descibed by a set of six patial diffeential equations. This system of equations must be solved simultaneously. Vaiations in fluid popeties ae significant and the continuity and momentum equations ae coupled to the enegy equation. In addition, these coupled equations pobably need to be solved sepaately but simultaneously ove liquid and vapo egions in the system. The oveall geomety of these spaces ae time dependent as well since the liquid level and bounday laye thickness will change as mass is added to the system and the tempeatue of the system will change. The appopiate bounday conditions between these fluid egions must then be equalied. The tanspot popeties will vay ove time and with espect to location, and the assumption that these popeties ae vaiable will add to the complexity of the consevation equations discussed above. Howeve, these popeties ae assumed only dependent on tempeatue and the natue of this dependence is known. Theefoe, a function elating this dependence must be found and input into the solution pocedue, and the deivatives can easily be calculated. Bounday Conditions Fist, since the stoage system to be analyed is unsteady, thee must be initial conditions pescibed fo all the vaiables at all locations inside the system. These initial conditions will depend on the flow situation to be analyed. Most likely, thee will be an initial liquid level equied, and the liquid will be assumed to be at a satuated condition

112 98 with no mass flow in o out of the tank. The initial pessue will be constant in the vapo space, and will have a hydostatic distibution in the liquid space. The initial tempeatue, density and enthalpy will be constant in the vapo space and will vay slightly with pessue in the liquid. All velocities will be assumed eo fo an initial condition. The initial conditions can be expessed mathematically as Vapo egion P(,,)=P sat T(,,)=T sat v (,,)= v (,,)= Liquid egion P(,,)=P sat + ρ g(h-) T(,,)=T sat v (,,)= v (,,)= Bounday conditions must be pescibed fo the fou independent vaiables P, T, v, and v. The density and enthalpy initial and bounday conditions can be calculated knowing the pessue and tempeatue at that location. Fo both the liquid and vapo bounday conditions, fluid velocity components will be assumed eo at the tank wall. This no-slip bounday condition will apply aound the entie geomety, with the exception of the gas supply and vent pots, whee a complete pofile fo the velocity, tempeatue and pessue distibution will have to be assumed. Against the wall the themal bounday condition is given fom a calculated heat flux though the insulation. The pessue bounday condition can be specified by consideing the fluid at the liquid to vapo inteface. Thee will be a thin laye of liquid at the top of the tank that will be at the satuated condition, and hee thee is only one independent themodynamic vaiable. The bounday pessue in the liquid below the inteface will be equal to this satuation pessue plus the hydostatic head pessue. The pessue aound the bounday in the vapo space is assumed to be constant. The most complicated bounday condition in this system is at the liquid to vapo inteface (37). At this infinitesimal inteface, the total velocity, shea stess and

113 99 tempeatue must be continuous acoss the suface. The nomal and tangential velocities must match, and can be expessed in tems of the inteface motion. The tempeatues also match. Pessue equilibium is satisfied by factoing in the suface tension foces. Heat flux is equal acoss the inteface, although the tempeatue gadient will diffe due to diffeent themal conductivity. The bounday conditions ae expessed mathematically as Vapo bounday laye P(δ,,t)= P T(δ,,t)= T P(,h,t)=P liq T T (R,,t)= Q /k T T (,Z,t)= Q /k k gas (,h,t)= kliq v (R,,t)= v (δ,,t)= v v (,,t)= v (,h,t)=v liquid v (R,,t)= v (δ,,t)=v v (,,t)= v (,h,t)= Vapo fee steam P(δ,,t)= P P(,h,t)=P liq BL T(δ,,t)= T BL T (,,t)= T T (,Z,t)= Q /k k gas T (,h,t)= k liq v (,,t)= v (,Z,t)= v (,,t)= v (,Z,t)= The liquid bounday laye and feesteam bounday conditions ae simila. Using the consevation equations shown above combined with the initial conditions and bounday conditions expessed in this section, the solution to the fluid behavio in this system can be modeled numeically. This would be a difficult modeling execise, well beyond the scope of this dissetation, which is intended to focus on themodynamic issues of this system.

114 1 Dimensionless analysis The oveall usefulness of these govening equations can be made geate if the equations ae pesented in non-dimensional fom. This esult can be obtained when the vaiables in the equations ae changed by dividing them by a constant efeence popety. The choice of efeence constants is abitay, but should have some physical significance to the system being analyed. The following non-dimensional vaiables ae poposed fo this system. = L = L V = V V V = V V ρ = ρ ρ P = P P h = h h T = T T t = t t µ = µ µ k = k k The dimensional length constant L is taken to be the length of the diagonal of the R-Z ight tiangle. This gives a measue of the sie of the system without having to use the both the tank adius and tank height as dimensional constants, and takes into account a vaiety of aspect atios. The choice fo dimensional constants fo all the themodynamic vaiables is just the initial value of the vaiable, most pobably the satuated conditions at the nomal boiling point. Similaly, the themal conductivity and viscosity constants ae also the initial values. The choice of dimensional time constant is unique to this analysis and is taken to be the amount of heat equied to vapoie the total amount of liquid inside the system divided by the ate of heat tansfe to the system. This is simply the time equied to vapoie the contents of the vessel. The dimensional velocity constant V then is defined as the system length divided by this time constant.

115 11 Plugging the new vaiables into the consevation of mass in the bounday laye yields v L V v L V v L V v L V t t = ρ ρ ρ ρ ρ ρ ρ ρ ρ ρ 3-54 Substituting the elation V L t = leaves the equation of mass consevation unchanged when conveting to dimensionless vaiables. This is expected and is a geneal esult fo all consevation of mass equations. The final dimensionless equations fo the bounday laye egion and the inviscid egion ae expessed below. v v v v t = ρ ρ ρ ρ ρ 3-55 v v v v v t = ρ ρ ρ ρ ρ ρ 3-56 Using the same dimensional constants as above, and eaanging tems, the consevation of momentum can be deived in dimensionless tems. The equation fo the bounday laye egion expessed by equations 3-46 and 3-47 can be ewitten as = P ) ( ) ( ) ( ) ( v L t P L P t L g t v v v v t v + = + + µ ρ µ ρ ρ ρ 3-58 Fo the inviscid coe egion, the non-dimensional momentum equations can be simplified to 2 2 ) ( ) ( P L P t v v v v t v = + + ρ ρ ) ( ) ( ) ( P L P t L g t v v v v t v = + + ρ ρ ρ 3-6

116 12 Thee ae thee dimensionless paametes contained in the above equations, shown in paenthesis on the ight hand side of equation These dimensionless paametes show the elative stength of the buoyancy, pessue, and viscous foces with espect to the inetial foces, but in a diffeent manne than the classic dimensionless equations. The choice of dimensionless constants emphasies the elation of the latent heat of vapoiation in the system to the ate of heat leak as a dimensional time constant. Fo small time constants, the heat leak is vey high with espect to the latent heat, and the liquid boils off moe apidly. This implies the heat tansfe to the liquid is geate, leading to highe convection velocities. This makes the dimensionless paametes smalle, and the viscous, buoyancy, and pessue foces ae espectively lesse than the inetial foces. Lage time constants imply a well-insulated system with small heat leak and smalle convection velocities. The dimensionless paametes ae lage, implying the inetial foces become less significant. It is woth noting the buoyancy and pessue foces scale with espect to 2 t, demonstating a geate sensitivity to the time constant than the viscous tems. Pefoming the same dimensional analysis on the consevation of enegy fo the viscous egion and the inviscid egion espectively yields 2 2 ) ( ) ( } ){ ( } ){ ( ) ( v t h T k T k L h t T k P v t P h P h v h v t h = + + µ ρ µ ρ ρ ρ 3-61 } ){ ( } ){ ( ) ( 2 T k T k T k L h t T k P v P v t P h P h v h v t h = + + ρ ρ ρ 3-62

117 13 Again, thee dimensionless paametes show up. They can be intepeted as showing the elative impotance of the flow wok, the heat tansfe, and the viscous dissipation. The tem ρ h is a measue of the initial enegy in the system, and shows up in the denominato of all thee dimensionless paametes. It is inteesting to note the flow wok paamete appeas independent of the time constant, while the heat tansfe paamete has a diect elation with the time constant and the viscous tem is invesely popotional to the time constant. Fo smalle time constants, the viscous tems ae elatively moe impotant than the heat tansfe tems. Howeve, if the time constant inceases, the convection velocities decease and the themal conductivity tems incease in elative value. To fully pose the poblem in dimensionless tems, the bounday conditions and initial conditions must also be non-dimensionalied. Looking at the vapo egion, the initial conditions ae expessed as P (,,)=1 T (,,)=1 v (,,)= v (,,)= The bounday conditions in the vapo bounday laye become ),, ( = P t P δ ),, ( liq P t h P = ),, ( = T t T δ " 2 ),, ( k Q T t k L h t R T ρ = " 2 ),, ( k Q T t k L h t Z T ρ = ),, ( T k k k t h T k liq gas liq gas = ),, ( = t R v ),, ( = v t v δ ),, ( = t v ),, ( liq v t h v = ),, ( = t R v ),, ( = v t v δ

118 14 ),, ( = t v ),, ( = t h v Bounday conditions in the feesteam ae ),, ( BL P t P = ),, ( liq P t h P = ),, ( BL T t T = δ ),, ( = t T " 2 ),, ( k Q T t k L h t Z T ρ = ),, ( T k k k t h T k liq gas liq gas = ), (, = t v ),, ( = t Z v ), (, = t Z v ),, ( = t Z v Two othe dimensionless paametes appea in the bounday conditions. The fist is simply the atio of the dimensionless themal conductivity between the liquid and vapo space. Choosing the same dimensionless constant could have eliminated the paamete. The second elates the heat flux to the tempeatue gadient at the Neumann bounday condition. In this case, the dimensionless paamete can be simplified by emembeing the dimensional time constant is the initial enegy in the system divided by the heat flux.

119 CHAPTER 4 HYDROGEN SAFETY To validate the analysis conducted in Chapte 3, a liquid hydogen test pogam was initiated. Any seious discussions egading use of hydogen, eithe as a liquid o a gas, must include factos petaining to its safe use. Fo this dissetation, such consideations ae taken within two sepaate contexts; fist, safety of liquid hydogen specifically as it elates to the expeimental system designed fo data gatheing fo this wok, and second, geneal hydogen safety pinciples that must be followed fo the evolution to a hydogen enegy based economy to take place. This chapte will focus pimaily on safety pinciples necessay fo the specific expeimental set-up, with geneal safety concens included as applicable. NASA KSC has histoically been one of the wold s lagest consumes of liquid hydogen, and aeospace use of hydogen has helped lead to the development of many of today s systems in use in industial applications. NASA s hydogen safety specification (51) is the pimay efeence fo this chapte, although the National Fie Potection Association (52) (NFPA) and Depatment of Tanspotation (53,54) (DoT) codes wee consideed as well. Thee is also an aeospace industy standad that is in development (AIAA), but this wok daws heavily on the NASA standad and was not efeenced in this chapte. Hydogen Popeties Hydogen is a nontoxic, non-coosive gas that is cololess, odoless, and tasteless, and is theefoe difficult to detect with human senses. It foms a diatomic molecule with a molecula weight of 2.6. In natue, hydogen has two isotopes; hydogen with an 15

120 16 atomic mass of 1, and deuteium, with an atomic mass of 2. The atio of odinay hydogen to deuteium is 32/1. Due to the small molecula sie, gaseous hydogen is difficult to contain in a leak fee manne, and liquid stoage is even moe difficult due to the low tempeatues equied. The NBP of hydogen is 2.3 K, with a tiple point of 13.9 K and a citical tempeatue of 33.2 K. A T-s diagam fo liquid hydogen (9) is shown in Figue 4-1. The NBP of hydogen is the second lowest of any element, only liquid helium is lowe. Hence, NBP hydogen will feee out any contaminants except helium, so cae must be taken in LH2 systems to eliminate any contaminants, especially ai. Hydogen is unique in that it exists in two sepaate molecula foms, depending on the diection of spin of the individual atoms. In otho-hydogen, the atoms spin in the same diection, wheeas paa-hydogen atoms spin in opposite diections. Nomal hydogen, o hydogen gas at oom tempeatue is a mixtue of 75% otho and 25% paa. Howeve, this mixtue vaies with tempeatue, so one efes to equilibium hydogen as the equilibium mixtue pecentage at a given tempeatue. At the NBP, equilibium hydogen is 99.8% paa-hydogen. This convesion fom otho to paa is significant since thee is a heat of convesion associate with the change. This heat of convesion is due to the momentum change equied as one atomic nucleus eveses spin diection, and this is an exothemic pocess. The heat eleased is significant, geate than the heat of vapoiation at the NBP. Enegy efficient handling of liquid hydogen equies accounting fo this heat of convesion, and catalysts ae usually used to speed up the convesion pocess so the heat can be emoved at a highe tempeatue. Duing hydogen system design pocesses, cae must be taken to account fo othe changes in themal and tanspot popeties, depending on the equilibium mixtue.

121 17 Figue 4-1 Hydogen T-s diagam (Flynn)

122 18 Hydogen possesses a vey low mass density; its vapo is buoyant in ai even at a tempeatue of 23K. Combined with a high diffusion coefficient, hydogen leaks dispese eadily in the atmosphee. Knowledge of hydogen convection and diffusion pattens is helpful in designing hydogen stoage facilities, in ode to minimie chances of a detonable mixtue foming. Anothe inteesting popety of hydogen is the low Joule- Thompson invesion tempeatue of 193 K. Hydogen is one of thee gasses with a JT tempeatue this low, helium and neon being the othes. This equies pecooling of a high-pessue gas steam to below this tempeatue befoe an isenthalpic expansion will poduce a efigeation effect. Room tempeatue hydogen heats up duing an isenthalpic expansion pocess. Combustion Haads Combustion theoy descibes a fie tiangle consisting of thee sides necessay fo combustion; fuel, oxidie, and an ignition souce. Obviously in the case of hydogen systems, the pesence of hydogen indicates the pesence of a fuel. One chaacteistic of hydogen fuels is they need vey little ignition enegy to initiate a eaction. The minimum ignition enegy usually descibed is.17 mj, o 1 times less than a spak geneated by static electicity. An ignition souce should always be assumed, and common ignition souces include static electicity, lightning, paticle impacts, metal uptue/factue, fiction, hot sufaces, adiabatic compession, shock waves, and heat fom smoking, open flames and welding. The best pactice fo eliminating hydogen fies is theefoe to minimie mixing with an oxidie. Howeve, anothe chaacteistic of hydogen fies is the lage ange of flammability limits in ai. The lowe flammability limit (LFL) of hydogen in ai is 3.9% and the uppe flammability limit (UFL) is 75%. A small hydogen leak has the potential to build up the concentation to a flammable limit, so cae

123 19 should be taken to design the stoage system to allow fo any leaks to dispese quickly. This is best accomplished by minimiing confinement of the hydogen, fo instance to allow fo a vent path out of a facility at a high point o to ceate a foced flow into and out of a facility. In ode to safely addess concens associated with hydogen combustion, knowledge of the combustion pocess is necessay. Once a spot in a fuel/oxidie mixtue is aised to its ignition tempeatue, combustion begins. This eaction then heats the adjacent mixtue above the ignition tempeatue and the combustion can poceed thu the entie mixtue that is in the combustible ange. The speed of the combustion is a function of seveal vaiables and geneally depends on the ate of heat tansfe in the mixtue. Usually this flame speed is of the ode of magnitude of a few metes pe second, and this is efeed to as deflagation. Howeve, mixtues in the ange of 18% to 59% hydogen can undego a detonation. In this case, combustion occus without the limitation of heat tansfe because a shock wave foms that is stong enough to heat the mixtue above the ignition tempeatue. The combustion speed then becomes supesonic, the speed of the shock wave. Detonations ae caused by sufficiently stong ignition souces within a detonable mixtue ange, o if thee is sufficient confinement o tubulence in an existing flame. While deflagations have pessue atios of the ode of magnitude of 8, detonations have pessue atios that can exceed 1. Nomal pessue elief devices can usually handle the ovepessue esulting fom deflagation, but not detonation. In addition, the pessuiation ate is so much faste in detonations that vent paths cannot keep up with the pessue wave. Theefoe detonations ae much moe destuctive than deflagations and systems should be designed to minimie detonation haads. This is

124 11 accomplished by minimiing confinement, such as intenal ooms with fou walls o intenal pipes with smalle diametes, as well as educing tubulence in intenal flows like that caused by flow estictions and flow path changes. Diluents should be used, especially in vent lines, to educe the possibility of detonable mixtues fom foming. In the event of a hydogen fie, pesonnel esponding must be tained in the best pactice fo fighting hydogen fies. A hydogen flame is invisible and has a low themal emissivity, so pesonnel cannot see o feel the fie until they ente the actual combustion one. Thee is no smoke pesent, and all combustion poducts ise in the atmosphee. Pope taining can ensue pesonnel ae awae of this dange and do not ente into a hydogen combustion aea. While thee have been instances of small hydogen fies being extinguished with chemical suppessants, these methods ae uneliable, and may actually allow the hydogen leak to build up to a detonable mixtue. The best pactice is to isolate the leak fom the hydogen supply, and spay wate on suounding sufaces to pevent the fie fom speading. Eventually the fie will self extinguish once the fuel souce has been isolated. All the existing explosive haads testing fo liquid hydogen has been completed with liquid hydogen at its nomal boiling point, so special consideation must be given fo densified hydogen. This is pimaily due to inceased mass of hydogen in a given stoage volume, howeve the eaction dynamics will also change when the additional sensible heat of the subcooled liquid is consideed. A small-scale simulation was modeled using subcooled LOX and LH2 and the esults indicated an explosive yield of 2-3 % should be expected, which is well below the yield nomally used fo nomal boiling point popellants. Howeve, thee always exists the possibility the subcooled popellants

125 111 could wam up to the satuation state, so the study ecommended no change in the safety standads. (55) Hydogen Embittlement Hydogen can attack metals in a numbe of ways. Geneally, what occus is a diffusion of hydogen atoms in the lattice of the metal, and the esult of this attack is the mechanical popeties of the mateials ae significantly alteed. Collectively this phenomenon is efeed to as hydogen embittlement, although physically thee ae thee diffeent types of embittlement. Fist, envionmental hydogen embittlement occus when metals and alloys ae plastically defomed in a hydogen envionment, such as compession on spings. The effect is magnified in high-pessue systems. Next, intenal hydogen embittlement occus when hydogen is absobed duing some manufactuing pocess, such as welding, electoplating, o acid teating. Finally, hydogen eaction embittlement occus when absobed hydogen chemically eacts with the paent metal, foming a weake hydide. This eaction is inceased duing high tempeatue pocesses. No matte what the physical pocess, hydogen embittlement educes the tensile stength, ductility, and factue toughness of mateials. Cack popagation occus moe eadily, and bittle failues occu moe fequently. Knowledge of the system envionment and mateial popeties is essential in designing fo educed hydogen embittlement haads. High-pessue pocesses incease the haads, and high tempeatues incease eaction embittlement haads. The embittlement haad is inceased with polonged exposue time and the concentation of hydogen pesent has a geat effect. Reducing stess on components educes the haad, and a highly polished suface will pemit less hydogen intusion than a ough suface. Micocacks o stess factues ae to be avoided since these ae embittlement enty

126 112 points. To design fo minimum haad, hydogen components should be manufactued fo a hydogen envionment. Thicke oveall pats educe stess and sufaces should be polished. If possible, oxide layes that estict hydogen absoption can be used. EDM pocesses should be minimied, since dielectic fluids used contain hydogen that is eleased duing ioniation. Pope welding techniques to educe the exposue to hydogen should be followed. Cetain alloys should be avoided, including cast ion, nickel and nickel alloys, and some stainless steels such as 41 SS and17-7 ph SS. Aluminum alloys in the 11, 6, and 7 seies ae not susceptible to embittlement and 3 seies stainless, oxygen fee coppe, and titanium ae geneally consideed safe as well. Fo ou application, we used 34 stainless steel in evey wetted pat, plus ou pessues and tensile stesses wee so low that embittlement was not a concen. Cyogenic Haads Not specifically hydogen elated, geneal safety pinciples duing use of cyogenic systems must be maintained. Cyogenic liquids ae stoed fa below atmospheic tempeatue, so natue equies heat tansfe to occu fom ambient to the inside of the stoage vessel. Such heat leak is unavoidable but pope insulation can minimie the total amount. This heat leak also leads to boil-off of the hydogen, ceating a pessue ise in the system. This pessuiation ate is geneally small if the vessel is popely insulated and can be accommodated by two means. Fist, the quantity of heat leaking into the tank can be emoved by mechanical efigeation befoe boil-off of the hydogen occus. Such eo-boil-off systems may be desiable in many cases and have been discussed at length ealie in this wok. The othe method is to allow fo a small pessue ise befoe boil off gasses ae vented thu a elief valve. This has been the method typically pefeed in the past. In this case, siing of the elief valve is citical. The capacity fo flow in the elief

127 113 valve must be matched to the maximum possible boil-off ate, othewise pessuiation above the maximum opeating pessue (MOP) may occu. NASA safety standads (56) equie the elief valve to be sied to accommodate flow esulting in complete loss of insulation of the vessel, plus any intenal heat geneation fom otho to paa convesion. Typically this means assuming the vacuum intenal to the annulus of a dewa has failed, and convection heat tansfe now occus between the inne and oute vessel. As an added pecaution, edundancy is equied in this function, pefeably using non-mechanical elief. Icing o othe feeing constituents may cause blockage of elief valves, and bust disks may be less susceptible. A pimay elief valve set at 1% of maximum opeating pessue coupled with a bust disk set at 116% MOP satisfies this intent. In addition, single elief valves must be installed in evey flow passage that has the possibility of being isolated fom the vessel elief system. Finally, at the component level, intenal wetted pats must be designed to peclude the possibility of tapping liquid o cold vapo in an intenal space, such as passages of closed ball valves o lobed pumps. Low tempeatues can ceate physical haads to the human body. Cyogenic buns, simila to heat buns, damage tissues. This can occu eithe fom diect contact with a liquid o contact with sufaces exposed to liquid hydogen. Pope equipment fo potection, such as cyogenic gloves and apons, ae essential. In most cases, vacuum jacketed insulation, o othe high quality insulation, eliminates the potential fo accidentally touching cold sufaces. Asphyxiation can occu when pesonnel ente aeas whee oxygen has been displaced by hydogen o othe gasses, such as nitogen o helium puges. In confined spaces, oxygen monitos ae equied. Pooly insulated liquid hydogen components have the capability of foming liquid ai. As this liquid ai wams,

128 114 the nitogen component will evapoate leaving a liquid with eniched oxygen content. This has the potential of ceating a combustion haad, as most substances combust in the pesence of pue oxygen. Ensuing the liquid hydogen lines ae well insulated mitigates this isk. Fo this wok, all pope pesonnel potective equipment was available, minimiing the low tempeatue haads. All elief valves wee sied popely. Continuous oxygen monitoing was available duing peiods of puging opeations, and hydogen detectos wee used to detect leaks. All lines and components wee vacuum jacketed except the vent line, and vent pocesses wee limited in duation and fequency. Facility Design Knowledge egading hydogen haads is essential in the pope design of a hydogen facility. The ode of pefeence in siting systems is always as follows; outdoos, in a sepaate building, in a special oom, inside a shaed oom. Depending on the total quantity stoed, some of these options ae consideed fobidden. Fo instance, quantities above 114 lites ae not pemitted to be stoed in anything but outdoos o a sepaate building. All quantities above 227 lites must be stoed outdoos. The system used fo this wok has a maximum stoage volume of 14 lites, technically too small to fall unde eithe NFPA o NSS specifications, howeve, fo the puposes of safety the system was consideed to be coveed by these codes. The system was allowed to be stoed inside a shaed oom. Liquid hydogen stoage should always be above gade, allowing fo ease of access and quick detection of leaks. Tables coelating the total stoage quantity to safe distance, efeed to as Q-D tables, dictate the minimum sepaation between the stoage aea and facilities o aeas whee pesonnel ae pesent. Access to this exclusion aea is limited to tained pesonnel with the pope pesonnel potective equipment. All equipment inside the exclusion aea must be appoved fo use

129 115 to ensue no ignition souces ae pesent, and any opeation inside the aea must have eviewed and appoved opeational pocedues. Baicades may be used to limit the extent of the exclusion aea, and dikes and impoundments ae necessay if thee ae any oxidies o othe flammable mateials stoed neaby. All stoage vessels should be designed pe the ASME Boile and Pessue Vessel Code, Section VIII, except fo potable stoage used in tanspotation, which must confom to DoT egulations. Stoage vessels should be labeled with the content, capacity and maximum allowable woking pessue (MAWP). The tank should be potected by baies if thee is the pesence of otating o ecipocating mechanical equipment neaby. A emote opeated valve should be povided as close as possible to the stoage tank to allow fo isolation of the stoage content in the event of a leak. All piping systems must confom to ASME B31.3. Electical bonding shall be pesent acoss all joints, and the entie system shall be gounded. Piping shall allow fo flexibility, to account fo coefficient of themal expansion as the system chills down. In most cases, bellows ae povided fo flexibility, although flexhoses, U bends, o low coefficient of themal expansion mateials such as Inva may be used. Suppots shall account fo this flexibility, using olles whee possible to constain the system in only one degee of feedom. Piping shall also be labeled with content, MAWP, and flow diection. Stoage systems must have pope vent and pessue elief systems. Relief system design equiements have been discussed in pevious sections. The pessue elief path shall be connected to an appoved vent stack. Venting to the atmosphee is acceptable if the flowate is below.5 lb/sec, any highe vent flowates should be buned in a flae stack. In eithe case, the vent system shall be designed to pevent ai o moistue

130 116 intusion into the vent. Moistue can fom ice blockage duing venting of cold gasses, obstucting the vent path. Ai intusion allows fo the fomation of combustible mixtues. The oveall flow velocity of the vent should be kept to a minimum, to peclude the possibility of geneating enough ignition enegy and to pevent tubulent mixing, which may aid in fomation of detonations. The vent stack must have a puge capability to keep oxygen out of the system. Gound vent stacks shall be 25 feet above gade, and oof vent stacks shall be 16 feet above the oof level. In all cases, vent stacks should be positioned away fom ai intakes o othe mechanical systems, and should have lightning potection in the fom of a 3-degee cone aound the top of the vent. In the event the hydogen is stoed inside a special oom o building, povisions fo the design of that facility must be made. Use of lightweight, non-combustible mateials is necessay, and walls, ceilings and floos should have a minimum 2-hou fie potection ating. All windows must be shattepoof glass o plastic, and doos shall open outwad to facilitate easy egess. Emegency venting shall be accommodated in the event of a detonation, and the vent must open with a maximum pessue of 25 lb/ft2. Adequate ventilation is equied to pevent the accumulation of hydogen above a concentation of 1%. The minimum vent aea shall be 1 ft2/1 ft3 of oom volume, and shall be located at the top of the oom. The oom shall be designed to avoid ceiling peaks and tapped pockets such as false ceilings. The hydogen system shall be 1% coveed with wate spay systems, but cae must be taken to avoid wate deluge into a vent line. Any electical systems within 3 feet of hydogen connections must be cetified as Class 1, Goup B, Division 1 pe NFPA Section 7, and Class 1, Goup B, Division 2 within 25 feet. Any system containing hydogen shall have adequate instumentation and contols

131 117 to ensue that opeation is within acceptable limits, and shall have adequate alams and wanings if opeation falls outside of acceptable limits. Means of detection hydogen leakage must be employed. Fixed hydogen detectos shall be used at likely leak points such as tanke connections, and potable hydogen detectos should be used duing opeations whee hydogen may accumulate. The detectos must have shot esponse times and must be capable of detection concentations of 1%, o 25% of the LFL. Management and Opeations Histoical data egading causes of past hydogen accidents eveal system and hadwae failues ae seldom the sole cause of the incident. Often, thee ae a numbe of individual events that fom a chain of events leading to an accident, and if any of those events was pevented the chain would be disupted and the incident could be avoided. Moe often than not, impope management o opeations is one of the key events leading to an accident, so cae should be taken to ensue this cause is avoided. The most impotant peventative measue is pope management of hydogen systems and the opeations. Safety eviews must be held at a numbe of points in the poject life cycle. Design eviews to achieve a fail safe design must be held, and the design must incopoate all the equiements discussed above. The system should be single fault toleant befoe failue that can lead to loss of life, injuy o majo system damage. A detailed haads analysis by an enginee knowledgeable in hydogen systems should be pefomed pio to fist opeation. All opeations shall equie appoved witten pocedues, and the pocedues must be maintained at the opeational site. Once in the opeational phase, good housekeeping must be maintained. Any combustible mateials, including weeds, ags, wood, and cadboad, must be kept 25 feet away fom any vessels o lines that may stoe hydogen. Egess outs must be kept fee and clea,

132 118 and must be maked popely. Pactice of emegency planning o pocedues should peiodically be pefomed. It is impotant that aea contol be enfoced to ensue people not knowledgeable about the opeation ae kept away. Conclusion Once the pope eseach and education was completed, these safety pinciples wee used to design, constuct, and opeate a small liquid hydogen stoage system. This stoage system, discussed in Chapte 5, contained an integated efigeation system. Although this type of system had not been developed peviously, and the sie of the system developed technically did not equie adheence to the safety standads, the pinciples leaned in this chapte wee invaluable in the following chaptes. If used popely, liquid hydogen systems ae as safe as othe flammable hydocabon systems.

133 CHAPTER 5 TEST SYSTEM DESIGN AND ANALYSIS The next majo step was to design and fabicate a sub-scale system capable of testing concepts intoduced in Chapte 2 and validating the models fom Chapte 3 while confoming to safety standads addessed in Chapte 4. Funding was obtained in collaboation with Floida Sola Enegy Cente though the NASA Hydogen Reseach fo Floida Univesities Pogam. The development pocess included a peliminay design and analysis phase, followed by a geneation of a system specification used fo pocuement of the test cyostat, eviews of the detailed design dawings made by the cyostat vendo, acceptance testing of the cyostat afte delivey, and finally the detailed design and fabication of the cyostat suppot systems. Peliminay Design and Analysis Peliminay design equiements consisted of the need to effectively demonstate the opeational concepts discussed in Chaptes 2 and 3. Majo components needed would be a hydogen stoage vessel, a efigeation system to emove heat, instumentation and data acquisition, fluid supply and distibution, vacuum system, and safety and leak detection. The geneal design path was to detemine the system sie in tems of stoage volume, then estimate the heat leak in that stoage system. The efigeation system is then sied to accommodate that heat leak combined with any heat emoval equied by the gas liquefaction ate. Once the siing was complete, suppoting systems such as instumentation, pneumatics, and vacuum systems could be selected. Constaints on system sie wee dictated by budgets as well as safety specifications. 119

134 12 Hydogen Stoage The equied volume of the hydogen stoage was set at 15 lites. This volume is simila to familia lab scale dewas that hold 18 lites, but was educed to bing the total volume down below the minimum NFPA 5B guideline of 15 lites. This classified the cyostat as a lab scale eseach vessel, not a definite stoage site. Howeve, the volume was still expected to be lage enough to pefom long time liquefaction opeations and to obseve statification in the densified liquid. The stoage vessel was to be made as efficient as possible as detemined by poven design chaacteistics. (57) Many opeations wee to be pefomed below atmospheic pessue, so thee was a equiement fo the inne tank to be vacuum tight. Maximum opeating pessue in the dewa was set at 358 kpa. Pimay and seconday elief fom ovepessuiation at 1% and 116% of the system MOP (equied by ASME code) was satisfied by a elief valve and bust disk espectively. In addition, a vacuum elief valve fo vacuum annulus potection was specified. The minimum tempeatue expected fo the liquid hydogen was 14 K, just above the tiple point tempeatue of 13.9 K. Satuated LH2 at this tempeatue has a density of.769 g/cm3. Theefoe, the maximum expected mass stoed in the system was 11.5 kg. Cyogenic Refigeato Once the stoage system has been sied, a cyogenic efigeato, o cyocoole must be identified that is capable of cooling the system. Based on a ule of thumb that the poduct loss ate of a new, well designed lage dewa is less than 5% pe day, an estimated heat leak can be calculated. The heat equied to vapoie 7.5 lites of liquid

135 121 hydogen at the NBP ove 24 hous is found to be appoximately 3 W. At a minimum, to achieve ZBO, a cyocoole that can poduce 3 W at 2 K is needed. If liquefaction is also equied, efigeation to emove the sensible heat of the gas between atmospheic tempeatue to the liquid tempeatue (3652 J/g) plus the heat of vapoiation (443 J/g) must also be supplied. Additionally, the heat of convesion between otho and paa hydogen must emoved (73 J/g). Assuming the need to fill the dewa evey 4 days, a mass flow ate of.33 g/s of hydogen gas is equied. The efigeato must then be capable of emoving the following amount of heat; Q =.33 g/s (3652 J/g J/g + 73 J/g) = 16 W 5-1 Combining the liquefaction equiement with the heat leak equiement, a cyocoole capable of at least 19 W at 2 K must be used. Liteatue and web seaches identified seveal candidates, mostly Giffod-McMahon style pulse tubes o stiling cycles, so no equiement was imposed fo any paticula manufactue. The cyocoole also needed to be integated into the stoage system in an efficient manne, to exchange heat between the cyocoole cold head and the liquid hydogen at the bottom of the tank. Details fo this inteface design, including penetation at the bottom of the tank vs. at the top using a heat pipe o coppe ba to tansfe heat to the bottom, wee left to the vendo building the cyostat. One citical popety that needed to be estimated was the fee convection coefficient of liquid hydogen. A simple heat exchange had to be designed to effectively tansfe the efigeation fom the cyocoole to the liquid, and the only mechanism was fee convection. Liteatue eviews did not eveal any expeimental values, so heat tansfe coelations had to be used. The simplest HX to incopoate was flat staps of

136 122 oxygen fee coppe anchoed to the coldhead, and this could be appoximated as a vetical flat plate. Chuchill developed a coelation fo this geomety, which gives (58).357 ( Ra 1/ 6 L 2 Nu L = [ ] /16 8 / 27 (1 + ( P The Pandtl Numbe fo liquid hydogen at 2 K is found fom the efigeant popety database efprop to be P = The Rayleigh Numbe is found fom ) ) ) Ra L 3 g β ( Tw T ) L = 5-3 α ν The volumetic themal expansion coefficient is defined as 1 ( δρ β = T ) P 5-4 ρ δ and using efprop to find the state vaiables is found to be 1 β =.14 aveage between K 15 K and 2 K. The aveage kinematic viscosity between 15 K and 2 K is ν =.18 cm 2 s cm 2 and the aveage themal diffusivity is found to be α =.15 s. The total length of the coppe staps is set as 1 cm, long enough to each within 5 cm of the bottom of the tank. Assuming a cold head tempeatue of 14 K and a liquid tempeatue of 2 K, the Rayleigh Numbe is estimated as 3.4 x This high of a Rayleigh numbe implies the dominant fom of heat tansfe in this system will be convection and the natual convective flow will be tubulent. Late opeational scenaios whee the tempeatue gadient is not as lage will educe the oveall heat flux and the flow egime will tansition to lamina. Fom this the Nusselt numbe is calculated to be Finally, the convection coefficient is found, knowing the themal conductivity of LH 2 and using the following elation

137 123 Nu k h = 5-5 L L to be W = 436. This is a significantly high heat tansfe coefficient and means the m K h 2 heat exchange should be an efficient pocess. A calculated total suface aea of oughly 1 cm2 is equied fo the heat exchange. Instumentation and Data Acquisition Data acquisition needs wee dictated by the type of planned opeations. On the cyostat itself, a pobe with five silicon diode sensos was necessay to measue tempeatue of the liquid as a function of liquid level. Othe diodes wee used to monito the health of the cyocoole and to contol tempeatue. The cyostat also needed pessue tansduces fo both the inne vessel and the vacuum jacket. It was also anticipated than mass flow of the poduct gas to be liquefied would be measued by a mass flowmete. The lage themal mass and the low heat tansfe ates in the system eliminated the need fo a high-speed data acquisition (daq) system. Labview 7. was used as the daq softwae, utiliing a pogam witten by D. Baik at FSEC. Fieldpoint netwok modules wee the inteface between the instumentation and the daq compute. The fieldpoint modules used have a 16-bit analog to digital convete. Fluid Distibution The following fou fluid intefaces wee equied fo the cyostat; a liquid fill and withdaw pot (½ VJ), vent pot (½ tube), liquefaction supply pot (3/8 tube), and a gas supply pot (3/8 tube). The liquefaction pot and the gas supply pot intoduced gas at two sepaate locations, one wapped aound the cyocoole cold head fo conduction heat exchange, and the othe to intoduce vapo bubbles at the bottom of the tank. The

138 124 tank elief valve and bust disk wee flow though type with the outlet plumbed to the vent system. All shut off valves wee specified as manual globe valves. In addition a pessue building leg was added to allow fo self-pessuiation, with egulation fo pessue contol. Gaseous nitogen, helium, and hydogen K bottles would be used as a commodity supply system. Appopiate K bottle egulatos wee puchased, including high and low pessue gauges and shut off valves. Liquid nitogen could be used to facilitate chilldown. Vacuum System The pefomance of any good cyostat depends diectly on the quality of its vacuum system. To ensue good pefomance and to enhance flexibility in the laboatoy, a high capacity, dual pump vacuum system was pocued fom Leybold (59). The fist stage of this system is a otay vane mechanical pump capable of pumping speeds of 35 m 3 /h, and an ultimate low pessue of 1-3 to. The second stage consists of a TuboVac tubomolecula pump with pumping speed of 55 l/s and an ultimate low pessue of 1-9 to. Vacuum pessues ae measued using a Combivac CM31 univesal vacuum gauge, combining a Themovac piani instument fo highe pessues (atmospheic to 1-3 to) and a Penningvac cold cathode senso fo lowe pessue ( to). A built in RS232 inteface allows data to be sent to the daq compute. Safety and Leak Detection System To measue any hydogen leakage, a catalytic potable hydogen leak detecto was puchased fo the laboatoy. The detecto woks by sensing the heat geneated by combustion of hydogen and oxygen on the suface of a palladium catalyst. Duing nomal opeations the detecto senso was positioned above the cyostat to detect leaking hydogen gas that is buoyant, but the senso could also be used as a point detecto duing

139 125 some tansient opeations to measue leakage at a specific inteface. A potable O 2 monito was used to ensue any helium o nitogen leakage would not ceate an oxygen deficient atmosphee in the enclosed laboatoy. All electical connections on the cyostat wee specified to be built explosion poof pe Class 1 Goup B Division 2 of NFPA 7. Vendo supplied equipment, such as tempeatue and pessue monitos that wee not explosion poof, wee specified to be enclosed in a puged envionment, meeting the intent of NFPA 7 Cyostat Specification Following completion of the peliminay design, a cyostat specification was poduced and distibuted to a vaiety of vendos fo bids. This specification used the siing and opeations equiements peviously analyed, and incopoated all the necessay design and fabication equiements imposed by cyogenic standads. All constuction was to be stainless steel, with the inne vessel confoming to ASME Boile and Pessue Vessel Code Section VIII, and the vacuum jacket to Section VIII and Section IX. Tubing was specified by ANSI B31.3. A numbe of test equiements, such as x-ay of welds, cold shock, leak test, and pessue test wee also specified. The vendo was equied to submit a numbe of test epots and cetifications as well as the fabication and assembly dawings. A final cleaning equiement and shipping details wee included as well. The full system specification, as well as a peliminay test schematic, is included in Appendix A. Cyostat Detailed Design Cyogenic Technical Sevices, of Longmont, Coloado was selected to fabicate the cyostat, pimaily due to the expeience of the owne Glen McIntosh with liquid hydogen. Glen added seveal innovative design featues, especially in the aea of the

140 126 pessue buildup system (PBU). His expetise in designing and fabicating low heat leak vessels eliminated the need fo a shield aound the inne tank. The cyocoole chosen was a G-M, connected to the top flange of the dewa with a heat pipe to tansfe heat to the bottom of the tank. A detailed design is shown in Figue 5-1. T6 T7 T1 T2 T3 T8 T4 T5 Figue 5-1 Cyostat design detail (Cyogenic Technical Sevices)

141 127 Cyostat Constuction The inne vessel is 2 inch SS 34 tubing, with a cylindical section 25.5 inch long and two elliptical end caps. The total volume is 157 lites, and has a suface aea of 1.56 m2. 45 layes of dexteous pape and aluminied myla, 1 ½ inches thick, seve as the multi-laye insulation (MLI). The inne vessel is pimaily suppoted by the 12-inch long, 8-inch diamete 34 SS neck. Thee is also a low conductivity G-1 bottom suppot. The cyocoole cold head is installed down the length of the neck, offset fom the cente, and thin stainless steel baffles suppess and adiative heat tansfe in the neck of the dewa. One design featue added is the incopoation of a flange at the top of the vacuum shell, to allow the entie inne vessel to be emoved. Anothe flange at the top of the neck allows emoval of the cyocoole coldhead, heat pipe, and inne vessel instumentation. A total of fou lines wee connected to the inne vessel. The liquid fill and dain line, VJ outside the dewa, was 3/8 inches ID,.49 inch wall thickness, and had a conduction path of 12 inches long. This line had a bayonet connection inteface, and an to the bottom of the inne vessel. The vent line was 5/8 inches OD, had.49-inch wall thickness and had a path length of 12 inches. It connected to the high point of the inne vessel, and had extenal tees to the pessue tansduce, the vent stack, the PBU leg, and a shut off valve. The liquefaction inlet line is 3/8 inch OD tube, with.49 wall thickness and a conduction path of 12 inches. This line is baed onto the cold end of the heat pipe fo good conduction heat tansfe, and has a shut-off valve at the wam inteface. The PBU system was a unique challenge. In most stoage vessels, the PBU line is connected to the bottom of the tank, has a numbe of coils wapped in the VJ annulus, and has a shut off valve and egulato outside the dewa at the top of the system. In this case, the PBU system uses a liquid tap to minimie heat leak when not in opeation. In deeply

142 128 subcooled systems, thee is no vapoiation occuing immediately in the PBU line, so thee is no vapo backpessue peventing the liquid fom flowing ove the liquid tap and into the uninsulated potion of the PBU line. This would damatically incease the total heat leak in the cyostat to the extent that the cyocoole would not be able to ovecome this heat leak, and subcooling would not be achievable. The simple solution to this is to move the PBU isolation valve to the bottom of the tank, but this design is complicated since the valve stem would be inveted and liquid could be pesent at the valve stem packing. This situation is unacceptable since heat leak would incease and thee could be leakage poblems at the dynamic seal at the top of the valve stem. CTS poposed an innovative solution to this poblem, placing the valve body and poppett at the bottom of the tank and using a stainless steel cable as a link between the valve seat and the manual actuato at the top of the system. The PBU line, 3/8 inches OD and.49 inch wall thickness, then uns downsteam of this valve a total of 23.5 inches until it exits the vacuum annulus and entes the vapoie section at the bottom of the tank. The PBU egulato is at the top of the tank. Thee is some additional heat leak associated with this cable, but this small heat leak is negligible. Knowing the details of constuction, a moe exact calculation of the total tank heat leak can be made. This numbe is estimated by adding the contibutions fom the adiation aound the aea of the dewa with the conduction down the dewa neck and all tube penetations, and adding in a cyocoole heat leak contibution given by the vendo Q hl ( T Tc ) = ha( T Tc ) + ka + Qcc 5-6 L whee h is the oveall heat tansfe coefficient acoss the MLI calculated fom the modified Lockheed coelation (42).

143 129 The total heat leak fo the cyostat is now estimated to be 8.7 W, moe than doubled compaed to the oiginal estimate based on the 5 % pe day ule of thumb. This is due to the fact that the small vessel has moe paasitic heat loads than lage dewas due to the scaling of heat leak being elated to aea ( squaed) compaed to volume ( cubed). Details of this heal leak calculation, made using an Excel speadsheet, ae included in Appendix B. The final instumentation design fo the cyostat included 8 Lakeshoe DT-4 seies silicon diode tempeatue sensos, with 4 wies each, calibated at the factoy. The sensitivity of these diodes is +/- 22 mk (6,61). Five of these diodes wee installed on a G- 1 suppot down the height of the tank, spaced evey 5 inches, with T1 eading the poduct (usually vapo) tempeatue at the top of the vessel and T5 eading the poduct (usually liquid) tempeatue at the bottom of the vessel. T6 is the cyocoole cold head tempeatue, T7 is the tempeatue at the top of the heat pipe, and T8 is the tempeatue at the bottom of the heat pipe. This aangement allowed fo visibility into liquid tempeatue statifications, as well as cyocoole pefomance, heat pipe pefomance, and the effectiveness of the cyocoole to heat pipe themal connection. A Tabe Industies pessue tansduce, factoy calibated with a esolution of.5 psia, monitos the inne tank pessue fom a tee off the vent pot (62). An Ameican Magnetics capacitance based liquid level pobe was included to bette estimate the total volume of liquid in the tank, but it neve seemed to function consistently and the data was not used. Cyogenic Refigeation A Cyomech AL33 Stiling type G-M cyocoole was selected fo the efigeation system. The AL33 is capable of poducing 4W at 2 K, 15 W at 15K and has a no load tempeatue of 12 K. A capacity cuve is shown in Figue 5-2 (63). The cold

144 13 head package, consisting of a otay valve, displace, egeneative heat exchange, and coppe cold mount, weighs 44 lbs and is connected to the compesso by two gaseous helium lines and an electical powe feed. The CP97 compesso is oil lubicated and wate cooled. The compesso powe equiement is 22 VAC, 3 phase, 6 H, and uses a total of 7.5 KW. At 2K, this gives a specific powe of W/W, not vey efficient but the cyocoole is vey eliable Powe (W) Tempeatue (K) Figue 5-2 Cyocoole AL33 capacity cuve (Manufactue Spec) The heat pipe is a 2 1/2 inch diamete tube, 26 inches long with an intenal volume of 127 in3. It is themally anchoed at the top to the cold head, using inva scews fo low themal expansion and Apeion N vacuum gease to enhance heat tansfe between the two polished oxygen fee coppe sufaces. The heat pipe coppe heads ae machined on the inside suface with paallel fins, to incease the condensation and evapoation suface aea. The heat pipe has a chage pot that has an inteface outside the dewa, fo changing the woking fluid if the tempeatue ange equies it, o to efill any

145 131 leakage that may have occued. The bottom coppe mass on the heat pipe has the liquefaction line soldeed to the outside suface, and fou flexible coppe staps,.75 inch wide by 5 inches long, designed to exceed the equied heat tansfe aea by giving 194 cm2 of suface. Acceptance Testing The cyostat was deliveed to FSEC in Octobe, 23. Following unpacking and visual inspection, the system was set up in the Chemisty Laboatoy. The instumentation was poweed up and the vacuum level on the vacuum jacket was veified to be good. The vacuum pump was connected to the inne vessel and this was evacuated to 1 micons, following which a vacuum decay check was pefomed. The pessue ose 7 micons in 6 hous. Vacuum leak ates can be quantified by the equation (64) Q = 79{273/(273 + T )} V P / t 5-7 Whee Q is the leak ate in sccm, V is the volume in lites, and P is the pessue in To. The estimated leak ate becomes.36 sccm. Afte the vacuum decay, the cyocoole was stated and a tempeatue of 14 K was eached on the cold head. At this point the heat pipe had not been chaged. All diodes functioned as expected. Suppot Systems Once the acceptance testing was complete, the cyostat was positioned into the lab whee the testing was to take place. Peviously, a hole was cut into the oute wall and a 1 vent pipe was installed and placed seveal feet above oof level. A K-bottle ack was puchased as well as nitogen, helium and hydogen K-bottle egulatos. Fo the gaseous hydogen supply system, a MKS mass flow mete was installed to contol and measue the quantity of hydogen enteing the system. The flowmete had a ange of -5 slm and an accuacy of.5 slm (65). Flexible coppe tubing was used to connect diffeent

146 132 elements, and swagelok compession fittings wee used fo all gas connections. Check valves wee used to minimie the potential fo system contamination. A liquid nitogen laboatoy scale dewa was puchased to aid in the initial chilldown, and a foam insulated, metal bellows flexhose was used to connect the LN2 to the cyostat. A pictue of the completed test set up is shown in Figue 5-3.

147 133 Figue 5-3 Liquid hydogen cyostat test aangement