EM14 he Second SME Intenational Confeence on Mechanical Engineeing Development, Study, and esting of Small-Scale Refigeation System fo Electonics Cooling Using Cold late Evapoatos hanut Ratchatapattaakul 1, Woasid utassanawin 1,* 1 Mechanical Engineeing, Faculty of Engineeing, Mahidol Univesity, Nakhon athom, hailand 73170 *Coesponding Autho: egsts@mahidol.ac.th, 02-889-2138 Ext. 6401-3, 02-889-2138 Ext. 6429 Abstact A small-scale efigeation system (SSRS) fo electonics cooling was designed, developed, and tested. he efigeation system consisted of fou majo components: a compesso, a mini-channel condense, an expansion valve, and a mini-channel cold plate evapoato. he effects of the mini-channel evapoato dimensions and the channel numbes on the efficiency of the small-scale efigeation system wee studied. hee diffeent channel dimensions of the cold plate evapoato wee tested. he cold plate evapoato had a efigeant-side cooling aea of 1650 mm 2. he hydaulic diametes of Evapoato 1, 2, and 3 wee of 1.075 mm, 1.253 mm, and 2.151 mm, espectively. he channel numbe of Evapoato 1, 2, and 3 wee 27, 27, and 18, espectively. he efigeant mass flow ates wee vaied fom 1.175-1.596 g/s, the efigeant suction and dischage pessues of R-134a wee 143-360 ka and 760-1220 ka, espectively. he compesso speed was between 4500 and 5500 pm, the inlet ai tempeatue of condense was fom 23 to 29C, and the heate input powes of the simulated electonics wee anged fom 200-310 W. Fom the test esults at heat dissipation of 200 W, the oveall system Coefficient of efomance (CO) of Evapoato 2 wee highe than those of Evapoato 1 and Evapoato 3 by 2.60-9.80% and 7.38-12.50%, espectively. While, the evapoato themal esistances of Evapoato 2 ae lowe than those of Evapoato 1 and Evapoato 3 by 38.70-50.02% and 37.04-50.92%, espectively. he evapoato and oveall system themal esistances of Evapoato 2 ae anged fom 0.3803 to 04107 K- cm 2 /W and fom -0.1071 to -0.2646 K-cm 2 /W, espectively. he expeimental esults indicated that Evapoato 2 with a hydaulic diamete of 1.253 mm and aspect atio of 5.06 had the best pefomance because of highe condense effectiveness and oveall system CO and lowe evapoato and oveall system themal esistances. Keywods: electonics cooling, efigeation system, cold plate evapoato.
EM14 he Second SME Intenational Confeence on Mechanical Engineeing Nomenclatue k Conduction heat tansfe coefficient, (W/m ๐ C) CO Coefficient of pefomance, ( - ) h Enthalpy, (J/kg-K) m Mass flow ate, (kg/s) Q Heat loss o heat tansfe, (W) essue, (a) R hemal esistance, (K/W) empeatue, (C) W owe input, (W) Volume flow ate, (m 3 /s) Geek symbols Heat exchange effectiveness, (%) Efficiency, (%) Density, (kg/m 3 ) Subscipts a Ambient ai o ai comp Compesso cond Condense evap Evapoato i Inlet, inne max Maximum o Outlet o oveall Refigeant s Isentopic sys System th hemal tot otal 1. Intoduction In these ecent decades, the compute development has been vey apid. his development is accompanied with the incease of heat dissipation and the miniatuization of the compute chips. he Intenational echnology Roadmap fo Semiconductos [1] the maximum heat dissipation fom a single chip package will ise to 525 W (75 W/cm 2 ) in 2014 fo highpefomance systems. Meanwhile, the maximum junction tempeatue must continue to be maintained at o below 80C in 2014. he cooling methods, such as heat sink and heat pipe, [2, 3] ae widely used nowadays. hese methods ae aimed at fast dissipating heat and hence thei cooling capabilities ae limited. It is estimated that as the chip powe eaches 180 W, the conventional cooling methods would become incompetent [4]. heefoe, new active cooling methods ae needed. Fom the expeiments of vaious active cooling appoaches [5], such as the vapo compession efigeation system [6], the themoelectic efigeation and pulse tube, ae compaed. Seveal poblems ae not esolved including heat capacity, efficiency, eliability, size and cost. Howeve, it is believed that among all these methods, the vapo compession efigeation is the best. Fom the pevious studies, thee is a development of the small-scale efigeation in ode to decease the chip suface tempeatue by using a cold plate evapoato. Howeve, it has not been the expeiments on the minichannel cold plate evapoato to study the effect of channel dimensions on the efficiency of the small-scale efigeation. heefoe, the study of cold plate evapoato dimensions will be pefomed to obseve the diffeences on the chip suface tempeatue and effectiveness. he
EM14 he Second SME Intenational Confeence on Mechanical Engineeing expeiments ae designed by using thee diffeent evapoatos and the COs ae calculated and compaed fo the most efficient cold plate evapoato. 2. Expeimental setup A small-scale efigeation system fo electonics cooling as shown in Fig. 1 was designed, constucted, and tested at vaious opeating conditions. he system consists of five main components: a coppe block to simulate the heat fom the chip, a mini-channel cold plate evapoato, a small-scale compesso, a minichannel condense, and an expansion device. he bushless DC compesso is a vaiablespeed compesso with 20-30 V powe input and includes an invete to vay the speed fom 2000 to 6500 pm. he swept volume of the otay compesso is 1.4 cm 3 /evolution. A sepentine condense had suface aea equal to 20x18 cm 2 and thickness of 5.5 cm. A needle meteing valve was used as the expansion device. Heate H hemocouples D Staightene hemocouples D Axial DC Fan Ai Inlet Nozzle owe Supply Ai Outlet Capillay tube Needle valve Filte & Die Sight Glass Heating coppe block F Expansion Device kw Vaiac Heates Micochannel Condense Cold late Evapoato Compesso Cooling Fan c3 c2 c1 Heat Speade c0 Oil sepaato Compesso Contolle Compesso Suction accumulato Sight Glass Fig. 1 Schematic of the SSRS system kw kw F H owe Mete essue ansduce hemocouple Flow mete Humidity Senso CU Size of 19 mm x 19 mm hee themocouples and thee catidge heates of 150 W wee inseted into the coppe block with dimensions of 20 mm on a side. he cold plate evapoato included a mini-channel heat exchange integated with a heat speade distibuted the heat fom the small coppe block suface to the lage heat exchange suface. hee cold plate evapoatos wee tested in this study. he detail dimensions of the evapoatos ae shown in able. 1. he condense was installed in a wind tunnel accoding to ANSI/ASHREA 41.2-1987 Standad [7]. empeatues and pessues wee used to measue both on ai- and efigeantsides. All tempeatue measuements wee -type themocouples. Diffeential pessue
EM14 he Second SME Intenational Confeence on Mechanical Engineeing tansmittes wee installed on the ai-side acoss a condense and a nozzle. he condense ai flow ate was detemined fom the measued pessue dop of a nozzle. Fo the efigeantside, a high pessue tansduce anged fom 0 to 25 ba was installed at the outlet of the condense and a low pessue tansduce anged fom 0 to 16 ba was installed at the inlet of the evapoato. wo diffeential pessue tansmittes wee installed on the efigeant-side acoss the condense and evapoato fo pessue dop measuement. he coiolis mass flow mete was installed between the condense and needle valve. A data acquisition system fom Agilent VEE was used in combination with a compute to collect and ecod all measuement data. Duing the tests, the outputs wee displayed on the compute sceen fo monitoing. able. 1 Detail dimensions of cold plate evapoatos Evapoato 1 2 3 1. Aea of Channels, (mm 2 ) 1,650 1,650 1,650 2. otal aea of cold plate evapoato, (mm 2 ) 2475 2475 2475 3. Evapoato base thickness, (mm) 10 10 10 4. Numbe of Channels, ( - ) 27 27 18 5. Hydaulic diamete, (mm) 1.075 1.253 2.151 6. Channels depth, (mm) 1.90 3.80 3.80 7. Fin width, (mm) 0.75 0.75 0.75 8. Channels width, (mm) 0.75 0.75 1.50 9. Aspect atio (depth/width) 2.53 5.06 2.53 3. Data eduction he heat dissipation ate fom the coppe block to the cold plate evapoato was calculated fom the axial heat conduction in the coppe block: kcoppe Ac Q (1) CU c3 c1 Lc1 c3 whee, c 1and c 3 is the tempeatues at the top and bottom positions in the coppe block. Lc1 c3 is the distance between the top and bottom themocouples in the coppe block as shown in Fig. 1. he themal conductive paste wee used between the coppe block and cold plate evapoato intefaces. he cooling capacity of the efigeation system was calculated by multiplying the efigeant mass flow ate with the efigeant enthalpy diffeence acoss the evapoato: Q m h h (2) evap, evap, o evap, i Fom the measuements of the efigeant outlet pessue and tempeatue in the single-phase supeheat egion, the outlet evapoato enthalpy. he evapoato inlet enthalpy was obtained by assuming an isenthalpic pocess acoss the expansion device. Similaly, the efigeant-side heat ejection ate of the condense was calculated by multiplying the efigeant mass flow ate with the efigeant enthalpy diffeence acoss the condense: Q m h h (3) cond, cond, i cond, o he ai-side heat ejection ate of the condense was detemined as the poduct of the ai volume flow ate measued at the nozzle, the ai density, and the ai-side enthalpy diffeence acoss the condense: Q h h (4) cond, ai ai ai cond, ai, o cond, ai, i
EM14 he Second SME Intenational Confeence on Mechanical Engineeing he Coefficient of efomance (CO) of the efigeation cycle and the CO of the oveall system as follows: Qevap, COefig (5) W CO Q comp, evap, sys (6) Wcomp, e whee the efigeant compession wok and oveall system wok can be obtained by: Wcomp, m hcomp, o hcomp, i (7) he condense effectiveness was defined as the atio between the ai-side heat ejection ate and the maximum heat ejection ate of condense: Q cond, a cond (8) Qcond,max whee the maximum heat ejection ate was: Qcond,max mai cai sat @ cond, i ai, i (9) he pefomance of the cold plate evapoato was defined by the cold plate evapoato themal esistance: chip evap Rth, evap (10) Qevap, whee chip is the chip suface tempeatue obtained by linea extapolation fom thee themocouples in the coppe block. Similaly, the oveall system themal esistance was defined as the atio of the diffeence between the chip suface tempeatue and the condense ai inlet tempeatues to the chip heat dissipation ate: R th, sys chip cond, ai, i (11) Q CU 4. Results and Discussion he expeiments wee conducted fo a condense ai inlet tempeatue of 24-30C and a condense ai flow ate of 0.078 kg/s by vaying two paametes: CU heat dissipation ates between 200 and 310 W; and compesso speeds fom 4500 to 5500 as shown in able. 2. he effects of two paametes on the chip heat dissipation ate, the efigeation cooling capacity, the condense effectiveness, the evapoato and oveall system themal esistances, and the oveall system Coefficient of efomances (COs) ae investigated fo thee cold plate evapoatos. able. 2 est matixes. est Q CU N comp (W) (RM) 1 199.5 4500 2 199.5 5000 3 199.5 5500 4 252.0 4500 5 252.0 5000 6 252.0 5500 7 310.5 4500 8 310.5 5000 9 310.5 5500 able. 3 illustated the measuing data fo Evapoato 1. he pefomance esults wee calculated based on the measuing of efigeant-side tempeatues and pessues, efigeant mass flow ate, and coppe block tempeatues. Measuements fom the pesent expeiments also showed that the efigeation system can maintain the chip suface
EM14 he Second SME Intenational Confeence on Mechanical Engineeing tempeatues anged fom 40C to 50C, which ae below 80C citeia of IRS [1]. able. 3 Measuing data fo Evapoato 1. est comp,o evap,o CU evap cond m [C] [C] [C] [ka] [ka] [g/s] 1 40.56 9.58 43.72 326 865 1.175 2 42.97 9.00 40.90 316 934 1.249 3 47.22 8.87 40.99 316 1062 1.288 4 43.66 11.99 49.74 360 954 1.206 5 47.58 11.57 48.47 354 1074 1.268 6 46.63-2.41 36.47 153 1021 1.458 7 46.83 2.04 45.53 195 1029 1.404 8 47.46 0.40 42.57 176 1046 1.425 9 48.00-1.38 39.76 163 1066 1.510 Fom the calculations by using the equations in the data eduction pat, the chip suface tempeatue, the cooling capacity, the condense effectiveness, the evapoato and oveall system themal esistances ae calculated and shown in able. 4. he esults of Evapoato 2 and 3 ae detemined by the same appoaches of Evapoato 1. able. 4 Data fo Evapoato 1. chip Qevap est cond R th,evap R th,sys [K-cm 2 /W] [C] (W) (W) [K-cm 2 /W] 1 37.73 197.9 197.9 0.142 0.120 2 34.12 209.7 209.7 0.119 0.071 3 34.32 215.5 215.5 0.118 0.071 4 41.29 203.8 203.8 0.143 0.109 5 40.24 213.7 213.7 0.134 0.103 6 28.52 237.4 237.4 0.130 0.020 7 35.85 231.7 231.7 0.145 0.063 8 33.13 234.3 234.3 0.139 0.047 9 30.78 246.7 246.7 0.130 0.033 he condense effectivenesses of thee evapoatos ae demonstated in Fig. 2. he condense effectivenesses of Evapoato 2 ae between 25.42% and 66.41%. Fo the given heat dissipation ate of 200-310 W and compesso speed of 4500-5500 pm, the effectivenesses of Evapoato 1, Evapoato 2, and Evapoato 3 ae anged fom 20.88% to 37.88%, fom 25.77% to 67.22%, and fom 16.66 to 21.58%, espectively. At a given heat dissipation ate of 310 W, the effectiveness of Evapoato 2 ae highe than those of Evapoato 1 and Evapoato 3 by 6.99-13.24% and 33.96-42.04%, espectively. Fig. 2 Condense effectiveness of thee evapoatos. Fig. 3 shows oveall system COs of thee evapoatos. he esutls shows that fo heat dissipation ate between 200 and 310 W, the oveall system CO of Evapoato 2 ae highe than those of Evapoato 1 and Evapoato 3. Fo instance, at heat dissipation of 200 W and the compesso speed fom 4500 to 5500 pm, the oveall system COs of Evapoato 2 ae highe than that of Evapoato 1 and Evapoato 3 by 2.60-9.80% and 7.38-12.50%, espectively.
EM14 he Second SME Intenational Confeence on Mechanical Engineeing Fig. 3 Oveall system Coefficient of efomance (CO) of thee evapoatos. Fig. 4 compaes the evapoato themal esistance chaacteistics of thee evapoatos. he evapoato themal esistances of Evapoato 2 ae lowe than those of Evapoato 1 and Evapoato 3 by 38.70-50.02% and 37.04-50.92%, espectively. Fo a given compesso speed anged fom 4500-500 pm, the evapoato themal esistance inceases as the heat dissipation ate inceases, since the incease of chip suface tempeatue is geate than the incease of heat dissipation ate. he evapoato themal esistances of Evapoato 2 ae between 0.3803 and 04107 K-cm 2 /W. Fig. 4 Evapoato themal esistance of thee evapoatos. he oveall system themal esistances of thee evapoatos ae illustated in Fig. 5. he oveall system themal esistances of Evapoato 1, Evapoatos 2, and Evapoato 3 ae anged fom 0.0764 to 0.2805 K-cm 2 /W, -0.1071 to -0.2646 K-cm 2 /W, and -0.1354 to 0.0133 K-cm 2 /W, espectively. It is found that the Evapoato 2 has the lowest oveall system themal esistance and negative values, since the chip suface tempeatues ae lowe than the condense ai inlet tempeatues. Fig. 5 Oveall system themal esistance of thee evapoatos. 5. Conclusion A bead boad small-scale vapo compession efigeation system (SSRS) using R-134a as the efigeant was designed, built, and tested. he expeimental esults showed that the efigeation system using Evapoato 2 can dissipate chip heat fluxes of appoximately 49.15-65.25 W/cm 2 and maintain the chip suface tempeatue between 10.12C and 19.85C fo a chip size of 2.0 cm 2. Evapoato 2 has a hydaulic diamete of 1.253 mm and aspect atio (depth/width) of 5.06.
EM14 he Second SME Intenational Confeence on Mechanical Engineeing 6. Acknowledgements he authos ae gateful fo the suppot of the hailand Reseach Fund (RF) fo this study. 7. Refeences [1] he Intenational echnology Roadmap fo Semiconductos (IRS). (2010). Intenational echnology Woking Goup, Assembly and ackaging, URL: http://www.its.net/epots.html. [2] astukhov, V.G., Maidanik, Yu.F., Veshinin, C.V. and Koukov, M.A. (2002). Miniatue loop heat pipes fo electonics cooling, Applied hemal Engineeing, vol. 23 (9) 2003, pp. 1125 1135. [3] Nguyen,., Mochizuki, M., Mashiko, K., Saito, Y., Sauciuc, I. and Boggs, R. (2002). Advanced Cooling System Using Miniatue Heat ipes in Mobile C, IEEE ansaction on Component and ackaging echnologies, vol. 23, 2002. [5] Wu, Z. and Du R., (2011). Design and expeimental study of a miniatue vapo compession efigeation system fo electonics cooling, Applied hemal Engineeing, vol. 31 (2-3), 2011, pp. 385-390. [6] utassanawin S., Goll E.A., Gaimella S.V., Cemaschi L., (2006). Expeiment investigation of a miniatue-scale efigeation system fo electonic cooling, IEEE ansaction on Component and ackaging echnologies, vol. 29 (3) 2006, pp. 678-687. [7] Standad Methods of Laboatoy Ai Flow Measuement, (1992). ANSI/ASHREA Standad 41.2-1987 (RA92).