Transient modelling of pumped twophase cooling systems:

Transcription

1 NLR-TP Jue 2016 Traset modellg of pumped twophase coolg systems: Comparso betwee expermet ad smulato CUSTOMER: Netherlads Aerospace Cetre NLR Netherlads Aerospace Cetre

2 Netherlads Aerospace Cetre NLR s a leadg teratoal research cetre for aerospace. Bolstered by ts multdscplary expertse ad urvalled research facltes, NLR provdes ovatve ad tegral solutos for the complex challeges the aerospace sector. NLR's actvtes spa the full spectrum of Research Developmet Test & Evaluato (RDT & E). Gve NLR's specalst kowledge ad facltes, compaes tur to NLR for valdato, verfcato, qualfcato, smulato ad evaluato. NLR thereby brdges the gap betwee research ad practcal applcatos, whle workg for both govermet ad dustry at home ad abroad. NLR stads for practcal ad ovatve solutos, techcal expertse ad a log-term desg vso. Ths allows NLR's cuttg edge techology to fd ts way to successful aerospace programs of OEMs, cludg Arbus, Embraer ad Platus. NLR cotrbutes to (mltary) programs, such as ESA's IXV re-etry vehcle, the F-35, the Apache helcopter, ad Europea programs, cludg SESAR ad Clea Sky 2. Fouded 1919, ad employg some 650 people, NLR acheved a turover of 73 mllo euros 2014, of whch three-quarters derved from cotract research, ad the remag from govermet fuds. For more formato vst:

3 EXECUTIVE SUMMARY UNCLASSIFIED Traset modellg of pumped two-phase coolg systems: Comparso betwee expermet ad smulato Problem area Two-phase pumped coolg systems are appled whe t s requred to mata a very stable temperature a system, for example the AMS02, whch was lauched wth a space shuttle ( May 2011) ad subsequetly mouted o the Iteratoal Space Stato. However, a two-phase pumped coolg system ca show complex traset behavor respose to heat load varatos. For example, whe the heat load s creased, a large volume of vapor s suddely created, whch results a lqud flow to the accumulator ad a crease the pressure drop. Ths wll result varatos the temperature the system, whch are udesred. It s ecessary to calculate these temperature varatos before a applcato s beg bult. For ths reaso, a software tool for traset two-phase systems has bee developed by NLR. REPORT NUMBER NLR-TP AUTHOR(S) H.J. va Gerer N. Braaksma REPORT CLASSIFICATION UNCLASSIFIED DATE Jue 2016 KNOWLEDGE AREA(S) Rumtevaarttoepassge Results ad coclusos A tool has bee developed that umercally solves the oe-dmesoal tmedepedet compressble Naver-Stokes equatos, ad cludes the thermal masses of all the compoets. The tool has bee used for dfferet projects, ad the umercal results show a excellet agreemet wth expermets. I ths paper, several pumped two-phase coolg systems are dscussed, ad a comparso betwee smulatos ad expermets s preseted DESCRIPTOR(S) pump two-phase traset modellg coolg Applcablty The software toolca be used for all pumped thermal cotrol systems.

4 EXECUTIVE SUMMARY UNCLASSIFIED GENERAL NOTE Ths report s based o a paper ICES , preseted at ICES, July 10-14, 2016, Vea, Austra NLR Athoy Fokkerweg CM Amsterdam p ) f ) e ) fo@lr.l )

5 NLR-TP Jue 2016 Traset modellg of pumped twophase coolg systems: Comparso betwee expermet ad smulato CUSTOMER: Netherlads Aerospace Cetre AUTHOR(S): H.J. va Gerer NLR N. Braaksma ASML NLR - Netherlads Aerospace Cetre

6 Jue 2016 NLR-TP Ths report s based o a paper ICES , preseted at ICES, July 10-14, 2016, Vea, Austra. The cotets of ths report may be cted o codto that full credt s gve to NLR ad the author(s). CUSTOMER Netherlads Aerospace Cetre CONTRACT NUMBER OWNER NLR DIVISION NLR Aerospace Systems DISTRIBUTION Ulmted CLASSIFICATION OF TITLE UNCLASSIFIED APPROVED BY : 2 AUTHOR REVIEWER MANAGING DEPARTMENT H.J. va Gerer J. va Es H. Slot DATE DATE DATE

7 NLR-TP Jue 2016 Cotets Summary 5 Nomeclature 5 I. Itroducto 6 II. Two-phase Mechacally Pumped Flud Loop 6 III. 2Φ-MPFL for space ad terrestral applcatos 7 IV. Why s a umercal model requred? 7 V. Physcal-mathematcal model 7 A. Tme-depedet 1D Naver-Stokes equato 7 B. Smplfcato 8 C. Calculato of heat put Q ad compoet temperature 9 VI. Numercal model ad mplemetato 9 A. Ethalpy ad mass equato 9 B. Pressure drop equato 10 C. Dscretzato of heat put Q ad compoet temperature 10 D. Equato of state 10 E. Maxmum tme step for umercal stablty 11 F. Accumulator models 11 VII. Comparso wth expermets 12 A. Model descrpto 12 B. Expermetal setup 14 C. Comparso betwee smulato ad expermet 15 VIII. Methods to reduce the temperature varatos a two-phase system 17 A. Reduce the dameter (or legth) of the trasport tube 17 B. Icrease coolg power the accumulator 18 C. Icrease the sze of the accumulator 18 IX. Cocluso 19 Refereces 19 3

8 Jue 2016 NLR-TP Ths page s tetoally left blak. 4

9 NLR-TP Jue 2016 Summary Traset modellg of pumped two-phase coolg systems: Comparso betwee expermet ad smulato Nomeclature Hek Ja va Gerer 1 Natoal Aerospace Laboratory NLR, Amsterdam, The Netherlads ad Nels Braaksma 2 ASML, Veldhove, The Netherlads Two-phase pumped coolg systems are appled whe t s requred to mata a very stable temperature a system, for example the AMS02, whch was lauched wth a space shuttle ( May 2011) ad subsequetly mouted o the Iteratoal Space Stato. However, a two-phase pumped coolg system ca show complex traset behavor respose to heat load varatos. For example, whe the heat load s creased, a large volume of vapor s suddely created, whch results a lqud flow to the accumulator ad a crease the pressure drop. Ths wll result varatos the temperature the system, whch are udesred. It s ecessary to calculate these temperature varatos before a applcato s beg bult. For ths reaso, a software tool for traset two-phase systems has bee developed by NLR. Ths tool umercally solves the oe-dmesoal tmedepedet compressble Naver-Stokes equatos, ad cludes the thermal masses of all the compoets. The tool has bee used for dfferet projects, ad the umercal results show a excellet agreemet wth expermets. I ths paper, several pumped two-phase coolg systems are dscussed, ad a comparso betwee smulatos ad expermets s preseted. Δt - Tme step (s) Δx - Legth terval (m) - Flud desty (kg/m 3 ) Nomeclature - Vscous stress tesor (N/m 2 ) A - Surface area (m 2 ) C - CFL codto costat (-)- c p - Specfc heat (J/(kg K)) d - Ier dameter tube (m) E - Specfc total eergy of the flud (J/kg) g x - Gravtatoal accelerato the travel drecto (m/s 2 ) H - Specfc total ethalpy flud (J/kg) H teral - Specfc teral ethalpy of the flud (J/kg) h - Heat trasfer coeffcet (W/(m 2 K)) h lv - Specfc latet heat of vaporzato (J/kg) - Idex for elemet a compoet (-) L - Legth of tube or evaporator (m) m - Mass flow (kg/s) m compoet - Mass of a compoet (kg) - Idex for tmestep (-) NE - Number of elemets a compoet (-) 1 R&D Egeer, Space Systems, Hek.Ja.va.Gerer@lr.l, R&D Egeer, TE RES Research F&T, Nels.Braaksma@asml.com,

10 Jue 2016 NLR-TP Q - Volumetrc heat put (W/m I. Itroducto 3 ) II. P, P out - Power put compoet, Power out of codeser (W) p, p sat - Pressure, Saturato pressure (N/m 2 ) Re - Reyolds umber (-) T, T sat - Temperature flud, Saturato temperature (K) t - Tme (s) u - Flud velocty (m/s) X - Vapor mass fracto (-) x - Dstace (m) V - Volume (m 3 ) Two-phase Mechacally Pumped Flud Loop I. Itroducto wo-phase pumped coolg systems are appled whe t s requred to mata a very stable temperature a Tsystem, or whe the tubg for a coolg system must have a small dameter. Ths paper dscusses a software tool to calculate the behavor of two-phase pumped coolg systems. The software tool has bee used for dfferet systems ad customers, e.g. for Thales Alea Space (for a space applcato) ad ASML (for several terrestral systems). The expermetal work ad part of the modellg work ths paper has bee carred out by Nels Braaksma at ASML as part of a MSc thess. ASML s the largest suppler the world of photolthography systems for the semcoductor dustry. II. Two-phase Mechacally Pumped Flud Loop I a Two-phase Mechacally Pumped Flud Loop (2Φ-MPFL), a pump s used to crculate a workg flud. Dowstream of the pump, the flud frst flows to a evaporator, where lqud s evaporated, whle heat from the payload s beg absorbed. The vapor the flows to a codeser where t s codesed back to lqud. The saturato pressure (ad thereby the saturato temperature) the system s cotrolled by the accumulator. Oe of the advatages of a two-phase system s that the temperature of the lqud/vapor mxture s the same the etre system (assumg that the pressure drop s small), ad depedet of the heat put. Ths cotrast to a sglephase (e.g. lqud water) coolg system, where heat put results a temperature crease of the lqud. Furthermore, the requred mass flow for a two-phase pumped system s much smaller tha for a sgle-phase coolg system, because the heat of evaporato h lv of a flud s much larger tha the specfc heat capacty of a flud tmes the allowed temperature gradet (.e. c p ΔT). Ths results a much smaller tubg dameter for a two-phase system tha for a sgle-phase pumped loop. Because of these reasos, a two-phase pumped system was selected for the thermal cotrol system of the tracker strumet of the Alpha Magetc Spectrometer (AMS02, see ext chapter). The subcooled lqud from the codeser ca have a very low temperature. I most applcatos, a heat exchager s appled whch heat from the vapor/lqud s used to warm the cold lqud to ear saturato temperature. A schematc drawg of a 2Φ-MPFL wth heat exchager s show Fgure 1. vapor vapor 3 + lqud 4 + lqud P evaporator heat exchager accumulator codeser P out lqud ear saturato temperature pump Fgure 1 Schematc drawg of a 2Φ-MPFL wth heat exchager. The red umbers 1 to 5 correspod to the umbers the pressure-ethalpy dagram Fgure 3 6

11 NLR-TP Jue 2016 III. IV. 2Φ-MPFL for space ad terrestral applcatos III. 2Φ-MPFL for space ad terrestral applcatos A two-phase pumped thermal cotrol system has bee developed by the NLR for the Alpha Magetc Spectrometer (AMS02, see Fgure 2 for a photo). AMS02 s a large (8500 kg, 2 bllo dollar) partcle detector that has bee lauched wth the space shuttle May 2011, after whch t was mouted o the Iteratoal Space Stato 1. Sce the, the two-phase thermal cotrol system keeps the AMS02 partcle detector at a very stable Why s a umercal model requred? temperature (fluctuatos less tha 0.3 C) a strogly fluctuatg thermal evromet 2. The thermal cotrol system uses CO2 as refrgerat because t has the best thermal performace 6. V. Physcal-mathematcal model A. Tme-depedet 1D Naver-Stokes equato Fgure 2 Photo of the AMS02 strumet o the teratoal space stato 1 Two-phase pumped thermal cotrol systems are also beg developed for terrestral applcatos. For example, the lthography maches of the compay ASML, large heat loads have to be removed wth very lght-weght ad small systems. Several two-phase thermal cotrol prototypes have bee bult ad modelled, some of whch acheve a temperature stablty of C wth varyg heat load. IV. Why s a umercal model requred? A 2Φ-MPFL s usually appled whe a uform system temperature s requred. Ths ca be acheved easly wth a steady-state heat load. However, whe the heat load o the evaporator of a 2Φ-MPFL chages, lqud wll flow to or out of the accumulator. As a result, the pressure the accumulator wll chage, ad therefore the system saturato temperature. The accumulator ca respod by heatg/coolg sde the accumulator order to retur to the desred temperature. I prcple, the accumulator ca mata exactly the desred temperature the system whe the accumulator coolg capacty s very large or whe the accumulator s very bg. I practce however, the coolg capacty ad accumulator sze are lmted ad the system temperature wll vary. A accurate model of the complete system s requred to calculate how much the temperature wll vary. V. Physcal-mathematcal model A. Tme-depedet 1D Naver-Stokes equato The flud flow a 2Φ-MPFL ca be modelled wth the oe-dmesoal tme-depedet compressble Naver- Stokes equatos: u u E ue u 0 2 p pu u Q mass equato (1) g x mometum equato (2) ug x eergy equato (3) 7

12 Jue 2016 NLR-TP B. Smplfcato equato of state (wth E ( p, Eteral ) 1 2 teral E 2 u ) (4) I these equatos, t s assumed that the lqud ad vapor a mxed flow have the same temperature ad velocty (.e. the homogeeous flow model s assumed). The eergy E ad ethalpy H are related va E=H-p/ρ. The eergy coservato equato ca therefore be rewrtte to the ethalpy coservato equato: H uh whch ca be rewrtte to: H p u Q u H H H u ug x ethalpy equato (5) B. Smplfcato Solvg the complete Naver-Stokes set of equatos ca be very tme cosumg. For ths reaso, the set of equatos s smplfed: Oly the mass equato (1) ad ethalpy equato (6) are tally solved. The mometum equato (2) s ot solved drectly. As a result, pressure waves (travellg wth the speed of soud) caot be smulated wth ths method. Ths greatly reduces the computato tme for the model, sce the maxmum tmestep s ow determed by the flud velocty, stead of the soud velocty (see secto VII.E). For two-phase pumped loops, pressure trasets ad vscosty oly have a lmted fluece o the ethalpy equato (e.g. vscous heatg s oly a few percet of the heat put). For ths reaso, pressure trasets ( p/ t) ad the vscosty stress tesor ( ) are gored the ethalpy equato (6). Gravty s gored. For the system descrbed ths paper, gravty has very lttle fluece o the calculated behavor of the system, sce o large heght dffereces are preset the system ad the pressure dffereces caused by gravty are eglgble compared to the absolute pressure ad other pressure dffereces (e.g. caused by frcto). I other smulated systems (ot descrbed ths paper), gravty effects are cluded. Wth these smplfcatos, the ethalpy ad mass equato become: H H Q u u u ethalpy equato (7) mass equato (8) All flud propertes ca be derved whe two thermodyamc varables are kow: 2 p, H ) equato of state (wth H H 1 u ) (9) ( teral For example, Fgure 3 shows the pressure-ethalpy dagram for CO 2. Ths fgure shows how the desty (blue les), temperature (red les), etropy (gree les), ad vapor mass fracto (black les) of CO 2 are related to the pressure ad ethalpy. I the example of a coolg cycle the fgure, the CO 2 s evaporated to a vapor mass fracto of 0.7. A heat exchager s used to preheat the cold lqud from the pump to ear saturato temperature (betwee 1 ad 2) by usg eergy from the vapor that leaves the evaporator (betwee 3 ad 4). The uderlyg data for ths dagram (whch s obtaed wth Refprop 3 ) ca be used to calculate the desty, temperature, etropy, ad vapor mass fracto at a certa pressure ad ethalpy. The pressure drop ( p/ x) Eq.(2) cossts of a vscous term ad a term caused by the mometum chage of the flud. The vscous pressure drop s calculated wth (emprcal) pressure drop correlatos: p vscous =0, see Eq.(1) p u Q teral f ( u,, H, d) ug x Wth the use of Eq.(1), the pressure drop due to the mometum chage ca be wrtte as: 2 (6) (10) 8

13 NLR-TP Jue 2016 C. Calculato of heat put Q ad compoet pmometum u u temperature u VI. Numercal model ad mplemetato A. Ethalpy ad mass equato V C. Calculato of heat put Q ad compoet temperature The flud exchages heat wth the compoet (tube, evaporator, codeser, etc.) through whch t flows. The volumetrc heat put Q to the flud s calculated wth: Aflud-compoet terface Q h( T T (12) compoet ) The temperature of the compoet s calculated wth: T compoet ( P h( T m flud compoet T) A compoet flud-compoet terface compoetc p, compoet The dscretzato of these equatos ad the mplemetato Matlab s dscussed the ext chapter. ) (11) (13) pump subcoolg codesato Small ΔT due to Δp Fgure 3 Pressure-Ethalpy dagram of a pumped two-phase cycle wth CO 2 (R744). The red umbers 1 to 5 correspod to the locatos fgure 1. VI. Numercal model ad mplemetato I ths chapter, Eq.(7) to Eq.(13) are dscretzed to make them sutable for umercal mplemetato. A. Ethalpy ad mass equato The ethalpy equato (7) s dscretzed by the MacCormack predctor-corrector scheme. Ths explct dscretzato scheme s very effcet ad s oe of the smplest stable schemes to mplemet. Each compoet (tube, evaporator, codeser etc.) the 2Φ-MPFL s dvded a umber of elemets NE. Frst, a predctor value of the ethalpy of elemets =1 to NE s calculated usg the varables at tmestep ad backward dfferecg: H p H after whch a predctor desty s calculated: Q u ( H H 1 ) predctor step (14) 9

14 Jue 2016 NLR-TP B. Pressure drop equato c C. Dscretzato ( p p H H u H of H ) heat put t Q ad compoet 1 p The ethalpy temperature tmestep +1 s the average betwee the predctor ad corrector value: D. Equato of state p p f ( H, p ) (15) A corrector value of the ethalpy s calculated usg predctor values ad forward dfferecg: H ( H 1 p c Q corrector step (16) H ) / 2 averagg step (17) After whch the desty at tmestep +1 s calculated: 1 1 f ( H, p ) (18) The velocty s calculated by dscretzg the mass equato (8) usg forward dfferecg: u 1 u 1 u 1 1 ( ( ) ) 1 (19) B. Pressure drop equato The frctoal pressure drop s calculated usg the emprcal Muller-Stehage & Heck pressure drop correlato whch the frcto factor s calculated wth the Colebrook equato whe the flow s turbulet (Re>4000), ad wth 64/Re whe the flow s lamar (Re<2400). I the termedary rego (2400<Re<4000), a smoothg fucto betwee the turbulet ad lamar frcto factor s used. p vscous f ( u 1 1 1,, H, d) emprcal pressure drop relato (20) The mometum pressure drop equato (11) s dscretzed as: p mometum ( u u ) 1 1 ( u 1 u ) u (21) The ew pressure s foud by addg the vscous ad mometum pressure drop: ew ew p 1 p pmometum pvscous (22) C. Dscretzato of heat put Q ad compoet temperature The flud exchages heat wth the compoet (tube, evaporator, codeser, etc.) through whch t flows. The volumetrc heat put Q to the flud s calculated wth: A, flud-compoet terface Q h( T, compoet T ) (23) V The temperature of the compoet s calculated wth:, flud compoet ( P h( T T ) A T T ) 1, compoet, flud-compoet terface compoet, compoet m C, (24) compoet p compoet For the smulatos dscussed ths paper, the heat trasfer coeffcet h s calculated wth Cooper s pool bolg correlato wth dryout model for evaporato 4, ad the Shah correlato for codesato 5. Other heat trasfer correlatos are also mplemeted. D. Equato of state Whe the ethalpy ad pressure the flud have bee calculated, the other flud parameters, lke desty, temperature ad vapor mass fracto ca be derved va the equato of state. The program Refprop 3 cotas the equatos of state for a large umber of fluds, ad ths program ca terface wth Matlab. However, callg Refprop for each compoet, at each tme step, s computatoally very expesve. I order to avod excessve overhead, flud tables (whch ca be mported at the start of a smulato) are created usg Refprop Ths flud table cossts of matrces whch the desty ρ, vapor mass fracto X, ad temperature T s stored for dfferet values of 10

15 NLR-TP Jue 2016 E. Maxmum tme step for umercal stablty p ad H. Durg the smulato, the desty, vapor mass fracto, ad temperature of all elemets ca be obtaed very fast by lear terpolato from ths table. E. Maxmum tme step for umercal stablty F. Accumulator models The Courat Fredrchs Lewy (CFL) codto s a ecessary (but ot always suffcet) codto for covergece whe solvg hyperbolc partal dfferetal equatos wth a explct umercal scheme. The reasog behd ths codto s that perturbatos must ot travel further tha oe legth terval Δx durg oe tme step Δt. For the compressble Naver-Stokes mometum equato, perturbatos travel wth the soud velocty ad a ecessary codto for covergece of a explct scheme s: u soud 1 CFL stablty crtero for compressble Naver-Stokes (25) However, the smplfed scheme that s used for the dyamc model, perturbatos do ot travel wth the soud velocty, but wth the flud velocty [see the ethalpy equato Eq.(7)]: u flud 1 CFL stablty crtero for smplfed scheme (26) I a 2Φ-MPFL, the flud velocty s typcally ~1 m/s, whle the soud velocty CO 2 at 22 C s ~450 m/s. Ths meas that whe the compressble Naver-Stokes equatos are solved, the tmestep Δt has to be ~450 tmes smaller tha the tmestep wth the smplfed scheme. Ths mples that t would take ~450 tmes loger to solve the compressble Naver-Stokes equatos. I the mplemeted model, the tme step Δt s determed automatcally wth the CFL codto: compoet compoet max( uflud, compoet) t C, t m( ) (27) Where the CFL codto costat C s usually 0.9 the smulatos. compoets F. Accumulator models The pressure a 2Φ-MPFL s cotrolled by the accumulator. For a two-phase flud, the pressure s related to the temperature ad the accumulator therefore cotrols the saturato temperature the system. Two types of accumulators are possble; Pressure Cotrolled Accumulators (PCA) ad Heat Cotrolled Accumulators (HCA). I a PCA, oly subcooled lqud ad o vapor s preset the accumulator. The pressure of the lqud the accumulator ca be cotrolled mechacally, for example by a bellows that s pressurzed wth a actuator or by pressurzed ar. The ma advatage of a PCA s that t respods very fast. The ma dsadvatage s that t s relatve complex. The pressure/temperature the HCA ca be creased by creatg vapor or decreased by creatg lqud (.e. by codesg vapor). Ths ca be acheved for example by a heater that s always submerged lqud, ad a coolg matle at the top of the accumulator (see Fgure 4 for a schematc drawg of the HCA). The ma advatage of a HCA s that t s relatve smple. The ma dsadvatage s that the coolg capacty s usually lmted, whch results a slow respose to chages evaporator heat put. Besdes the heat puts that are appled to the HCA (.e. P heat HCA ad P cool HCA ), there s also teral heat exchage betwee the vapor ad lqud (P vapor to lqud ) ad betwee the wall ad the flud (P wall to flud ) whe the HCA s ot thermal equlbrum. Both PCA ad HCA are modelled ad have bee used smulatos. The smulatos ad expermets descrbed ths report have bee carred out wth a HCA. Note that for terrestral applcatos, accumulator desg s relatve straghtforward, sce the lqud ad vapour are separated by gravty. For space applcatos, capllary forces are used to esure that lqud s always preset at the let/outlet tube of the accumulator. For example for AMS02, a staless steel mesh scree that s folded to a fa lke structure (where the gap betwee mesh layers becomes arrower towards the lqud etrace ppe of the accumulator) has bee used sde the accumulator 2. 11

16 Jue 2016 NLR-TP VII. Comparso wth expermets Coolg water A. Model descrpto P cool HCA vapour P wall to flud P vapour to lqud lqud heater P heat HCA Fgure 4 Schematc drawg of the Heat Cotrolled Accumulator VII. Comparso wth expermets A. Model descrpto Fgure 5 shows a schematc drawg of the test setup. The preheater (1) s ot used durg the expermets; t s just a thermal mass. I the evaporator (2), a heat load s appled. I the expermets ad smulatos, the heat load s vared betwee 131W ad 331W. I actual applcatos, the object that has to be cooled s ofte located at a cosderable dstace from the thermal cotrol system, ad for ths reaso, a 5 m log trasport tube s used betwee the evaporator ad the codeser both the expermets ad smulatos. The yellow meaderg tube Fgure 5 represets ths trasport tube. The codeser (3) cossts of a tube--tube heat exchager that s cooled wth lqud water. The water flow s also cluded the umercal model. The accumulator (4) s a HCA, wth a coolg matle at the top that provdes a coolg power of approxmately 47W the HCA. At the bottom of the HCA, a heater s used to provde heater power. The thermal capacty of the accumulator vessel has a sgfcat effect o the system behavor, ad s cluded the smulato. The flud s crculated the system by the pump (5). Two massflow meters are cluded the setup (6). The lqud massflow to the accumulator s equal to the dfferece the massflow betwee the two flow meters. The massflow the system s cotrolled to be 3 gr/s. The heatg power for the accumulator s cotrolled wth a PID regulator. Fgure 6 shows the calculated steady-state temperature, vapor mass fracto ad mass flow the test-setup wth a heat put the evaporator of 131W. The saturato temperature s cotrolled by the accumulator to be 22 C. The vapor mass fracto resultg from the heat put s 0.1. Fgure 7 shows the calculated temperature, vapor mass fracto ad mass flow the test setup, 6 secods after the heat put s creased from 131 to 331 W. I the evaporator, the vapor mass fracto s creased to 0.6 due to the creased heat put. Ths creased vapor mass fracto travels through the system wth creased flud velocty (same mass flow rate, lower flud desty) ad s half-way the trasport tube at ths tme stat. Dowstream of ths creased vapor frot, the massflow s early two tmes hgher tha upstream of the frot, as the flud adopts the velocty of the frot whle havg a hgher desty. Ths excess massflow goes to the accumulator, whch results a hgher pressure the accumulator ad therefore a hgher saturato temperature the system. Fgure 8 shows the calculated steady-state temperature, vapor mass fracto ad mass flow the test-setup wth a heat put the evaporator of 331W. 12

17 NLR-TP Jue Preheater 2. Evaporator 3. Codeser 4. Accumulator 5. Pump 6. Mass flow meters Fgure 5 Schematc drawg of the test setup P=131W Fgure 6 Steady-state temperature, vapor mass fracto, ad mass flow for P=131 W P=331W Vapor frot Fgure 7 Temperature, vapor mass fracto, ad mass flow, 6 secods after the heat put s chaged from 131 to 331 W 13

18 Jue 2016 NLR-TP B. Expermetal setup P=331W Fgure 8 Steady-state temperature, vapor mass fracto, ad mass flow for P=331 W B. Expermetal setup Fgure 9 shows a photo of the expermetal setup ad Fgure 10 shows a photo of the HCA. The setup has the same compoets ad approxmately the same layout as Fgure 5. Fgure 11 shows a photo of the 5 m log trasport tube. Coolg matle Vewg glass Fgure 9 Photo of the expermetal setup heater Fgure 10 Photo of the accumulator Fgure 11 Photo of the 5 m log trasport tube 14

19 NLR-TP Jue 2016 C. Comparso betwee smulato ad expermet C. Comparso betwee smulato ad expermet Fgure 12 shows the evaporator heat put for both the expermets ad smulatos. I the left fgure, the heat put s creased from 131 to 331W, ad the rght fgure, the heat put s decreased from 331 to 131W. As a result of a crease the evaporator heat put, lqud wll flow to the accumulator (see Fgure 13). The smulated peak s a bt hgher ad ther tha the expermetal peak. A explaato for ths s that the expermet, the evaporator cossts of a tube that s soldered to a heater. I the model, t s for smplcty assumed that the wall of the evaporator tube has the same temperature as the heater. I the expermet however, there s a thermal resstace betwee the tube ad the heater, ad as a result, the flud the evaporator wll heat up slghtly slower the expermet tha the smulato. Ths slghtly slower respose of the expermetal evaporator results a lower ad wder massflow peak to the accumulator. However, the total amout of lqud that flows to the accumulator s the same. Fgure 14 shows the saturato temperature dowstream of the evaporator. The PID cotroller for the accumulator heater tres to keep ths saturato temperature at 22 C. However, as a result of the lqud flow to the accumulator, the saturato temperature the system wll crease. The correspodece betwee the expermet ad smulato s very good, ad from ths fgure t ca be cocluded that the ma objectve of the model (.e. to calculate the varatos the system saturato temperature as a result of chages the heat load) s acheved. Fgure 15 shows the power of the heater the accumulator. Ths heater s PID cotrolled; whe the saturato temperature the system s lower tha 22 C, the power wll be creased, ad whe the saturato temperature s hgher tha 22 C, the power wll be decreased. I steady-state, the heater power s equal to the coolg power of the coolg matle of the accumulator. Whe flud flows to the accumulator, the accumulator heater power s quckly reduced to zero, makg the accumulator remove heat from the system. The fgure shows that tally after a evaporator power chage, the smulated ad expermetal accumulator heater power s very smlar. However, the expermetal heat power shows much more overshoot tha the smulato. The reaso for ths s ot uderstood, sce the PID parameters are the same for both the expermet ad smulato. A possble reaso could be that the PID cotroller the expermet respods dfferetly whe the lmts of the heater power are reached (.e. P heat accu = 0 or P heat accu = 100W). Fgure 16 shows the temperature upstream of the evaporator. There s a large dfferece betwee the smulated ad expermetal let temperature. Ths dfferece s caused (amog others) by the accuracy of the emprcal correlato (Shah correlato 5 ) that s used to calculate the heat trasfer coeffcet the codeser. However, the dfferece does ot fluece the system behavor. The reaso for ths s that a large dfferece the lqud temperature before the evaporator, oly results a relatve small dfferece the vapor mass fracto after the evaporator: c p T ΔX 0.1for ΔT = 4 C (28) H lv The lqud temperature dfferece could be reduced by adjustg some parameters (.e. tug ), but ths s ot ecessary, because the saturato temperature s accurately predcted wthout ay tug of the results. I actual thermal cotrol systems, a heat exchager s usually used to warm the lqud before the evaporator to ear saturato temperature (see Fgure 1), ad that case, the dfferece betwee the smulated ad expermetal evaporator lqud let temperature would become much smaller. 15

20 Jue 2016 NLR-TP Fgure 12 Evaporator heat put for both the expermet ad smulato Fgure 13 Massflow to the accumulator as a result of a crease ad decrease of the heat put Fgure 14 Saturato temperature after the evaporator 16

21 NLR-TP Jue 2016 VIII. Methods to reduce the temperature varatos a two-phase system A. Reduce the dameter (or legth) of the trasport tube Fgure 15 Heatg power the accumulator. Ths heatg power s regulated wth a PID cotroller Fgure 16 Lqud temperature before the evaporator VIII. Methods to reduce the temperature varatos a two-phase system Oe of the ma reasos to use a two-phase thermal cotrol system s that t ca mata a much more stable temperature tha a sgle-phase thermal cotrol system. However, as Fgure 14 the prevous chapter shows, the temperature a two-phase thermal cotrol system ca vary sgfcatly whe the heat load to the evaporator s suddely chaged. I ths chapter, some smple measures that ca be used to reduce the temperature varatos are dscussed. More complex methods are also possble, but are ot dscussed ths paper. A. Reduce the dameter (or legth) of the trasport tube I the smulatos ad expermets descrbed the prevous chapter, the teral dameter of the trasport tube s 4 mm. Whe ths dameter s smaller, the volume of the trasport tube becomes smaller, ad ths wll reduce the amout of lqud that flows to the accumulator whe the heat load s creased. As a result, the crease saturato temperature whe the heat load s creased wll become smaller, as ca be see Fgure 17. However, the trasport tube dameter caot be made too small. Whe the dameter s just 1.5 mm, the varatos the pressure drop over the trasport tube become too large: The pressure drop over the trasport tube creases from 1.8 bar to 3.5 bar whe the heat put the evaporator s creased from 131 to 331W. As a result of ths rse the 17

22 Jue 2016 NLR-TP B. Icrease coolg power the accumulator pressure drop, the saturato temperature the evaporator creases. For ths reaso, the crease evaporator saturato temperature s larger for a 1.5 mm dameter tube tha for a 2 mm dameter tube. C. Icrease the sze of the accumulator Fgure 17 Saturato temperature the evaporator for dfferet trasport tube dameters B. Icrease coolg power the accumulator Fgure 18 shows the calculated temperature for a smulato whch the coolg power the accumulator s creased from 47 to 94W. The heatg power the accumulator s also creased wth a factor of two. The fgure shows that the crease the accumulator coolg power reduces the varatos the saturato temperature. Fgure 18 Calculated saturato temperature wth 47W ad 94W of coolg power the accumulator C. Icrease the sze of the accumulator Whe the heat put to the evaporator s creased from 131 to 331W, lqud CO 2 flows to the accumulator. Ths flow of lqud compresses the vapor the accumulator, ad as a result, the pressure ad saturato temperature the system rses. I the expermet descrbed the prevous chapter, the volume of the accumulator s 300 ml. Whe the accumulator volume s 900 ml, the crease saturato temperature due to a crease the evaporator heat put s smaller, see Fgure

23 NLR-TP Jue 2016 IX. Cocluso Refereces Fgure 19 Saturato temperature wth two dfferet accumulator volumes IX. Cocluso Whe the heat load o a 2Φ-MPFL chages, lqud wll flow to or out of the accumulator. As a result, the pressure the accumulator wll chage, ad therefore the system saturato temperature. A accurate model of the complete system s requred to calculate how much the temperature wll vary. For ths reaso, a software tool that umercally solves the tme-depedet mass ad ethalpy equato for a two-phase flud has bee developed. The model also cludes the thermal capacty of all the sold compoets. Numercal results have bee compared wth expermets for dfferet systems (e.g. wth a system that keeps a payload at C temperature stablty), ad the correspodece s very good. For other projects, smulatos have bee carred out wth other fluds tha CO 2 (e.g. wth R134a, R152a ad R245fa). However, tests have oly bee carred out wth CO 2. Tests wth R134a are plaed order to vestgate whether the assumptos that are made the model (e.g. a homogeeous flow s assumed) are also vald wth other fluds. Refereces 1 AMS-02, The Alpha Magetc Spectrometer expermet, URL: [cted 20 Jauary 2014] 2 va Es, J., Pauw A., va Dok G., Laud E., Gargulo C., He Z., Verlaat B., Ragt U., va Leeuwe P., AMS02 Tracker Thermal Cotrol Coolg System: Test Results of the AMS02 Thermal Vacuum Test the LSS at ESA ESTEC, AIAA (2012) 3 Lemmo, E.W., Huber, M.L., McLde, M.O. NIST Stadard Referece Database 23: Referece Flud Thermodyamc ad Trasport Propertes-REFPROP, Verso 9.1, Natoal Isttute of Stadards ad Techology, Stadard Referece Data Program, Gathersburg, Zeyep Ozdemr, M, Expermetal Ivestgato of CO2, Two-Phase Heat Trasfer Characterstcs Ad Predcto Of CO2 Dryout Vapor Qualty, MSc thess, Techcal Uversty of Edhove, WET , Shah, M.M., A Improved Ad Exteded geeral Correlato for Heat Trasfer Durg Codesato Pla Tubes, HVAC&R Research, vol. 15, o. 5, pp , va Gerer, H. J., va Bethem, R. C., va Es, J., Schwaller, D., Lapesée, S., Flud selecto for space thermal cotrol systems, 44th Iteratoal Coferece o Evrometal Systems, ICES

24 Jue 2016 NLR-TP Deze paga s opzetteljk blaco. 20

25

26 NLR Athoy Fokkerweg CM Amsterdam, The Netherlads p ) f ) e ) fo@lr.l )