Conerence on Modelling Fluid Flow (CMFF 6) The 13th International Conerence on Fluid Flow Technologies Budapest, Hungary, September 6-9, 6 GENERATOR COOLING USING HEAT PIPES Bert de LEEUW 1, Harry HAGENS 2, Steven BRAND 3, Mart GROOTEN 4, Frank GANZEVLES 5, Cees van der GELD 6, Erik van KEMENADE 7 1 VDL Klima bv Eindhoven, the Netherlands E-mail: bdleeuw@klimacom 2 VDL Klima bv Eindhoven, the Netherlands E-mail: hhagens@klimacom 3 VDL Klima bv Eindhoven, the Netherlands E-mail: sbrand@klimacom 4 Department o Mechanical Engineering, Technische Universiteit Eindhoven E-mail: mhmgrooten@studenttuenl 5 Department o Mechanical Engineering, Technische Universiteit Eindhoven E-mail: laganzevles@tuenl 6 Corresponding Author Department o Mechanical Engineering, Technische Universiteit Eindhoven Postbus 513, 56 MB Eindhoven, The Netherlands Tel: +31 4 2472923, Fax: +31 4 2475399, E-mail: cwmvdgeld@tuenl 7 Department o Mechanical Engineering, Technische Universiteit Eindhoven E-mail: hpvkemenade@tuenl ABSTRACT This paper presents the study o a heat pipeequipped heat exchanger with two illing ratios o R134a 19% and 59%, respectively The airlow rate varies rom 4 to 2 kg/s The temperatures at the evaporator side o the heat pipe vary rom 4 to 7 C and at the condenser part rom to C The perormance o the heat exchanger has been compared with two pool boiling models, Cooper and Gorenlo, and two ilmwise condensation models, Nusselt and Butterworth A airly good agreement is ound or measurements and the model o Cooper at low airlow Reynolds number at the evaporator side o the heat exchanger The result o this study is that a heat pipe equipped heat exchanger can replace a water-cooled heat exchanger without loss o perormance The tested process conditions are typical or warmer countries Keywords : inned tube, heat exchanger, heat pipe, R-134a, thermosyphon NOMENCLATURE A [m 2 ] surace area D [m] diameter D h [m] hydraulic diameter F [N] orce F e [-] illing degree H [m] in distance L [m] length M [kg/kmol] molecular weight Nu [-] Nusselt number Pr [-] Prandtl number Q [W] heat low rate R [K/W] heat resistance Re [-] Reynolds number T [ C] temperature W [m] distance between pipes c p [J/kgK] heat capacity at constant pressure d i [m] inner pipe diameter g [m/s 2 ] acceleration due to gravity h g [J/kg] enthalpy o evaporation mɺ [kg/s] mass low rate p [Pa] pressure p r [-] reduced pressure q [W/m 2 ] heat lux r [m] radius α heat transer coeicient λ [W/mK] thermal conductivity δ [m] thickness η in [-] in eiciency µ [Pas] dynamic viscosity ρ [kg/m 3 ] mass density χ [-] geometric correction actor Subscripts and Superscripts cond condenser evap evaporator luid i inner lm logarithmic mean max maximum min minimum o outer tot total v vapour w wall x, y Cartesian coordinates 1 INTRODUCTION Stand-alone electricity power generators are usually cooled with ambient air Standard practice is air to air heat transer or using a tube in plate heat
exchanger with water as an intermediate medium In some situations water is not available or ambient temperatures are too high to use ambient air In those cases heat pipes may provide an alternative or cooling powers in excess o kw Multiple heat pipes then connect two plate heat exchangers The heat transer in the system is based on the continuous cycle o the vaporization and condensation process The thermosyphon, or heat pipe i equipped with a wick inside, is heated at the evaporator, which causes evaporation o a part o the luid The vapour lows to the condenser, where the luid condenses while giving o its latent heat, caused by cooling rom the outside The condensate lows back to the heated section along the wall by gravitation, which closes the cycle Thermosyphons can be used to transer heat between two gas streams The advantages are the high heat recovery eectiveness, compactness, no moving parts, light weight, relative economy, no external power requirements, pressure tightness, no cross-contamination between streams and reliability [1, 2] The heat transer being based on evaporation and condensation, the latent heat o the luid is an important parameter The higher the latent heat o a luid is the higher the transer o heat at a lower pressure The working principles o the thermosyphon imply that the luid should evaporate and condense within the temperature range Taking the possible application o cooling an electricity generator with ambient air into account, the working luid R-134a is an option, considering the expected temperature ranges in practice The hot air will be in a range o 4 8 C, the ambient air will be in a range o - C R-134a sublimates at - 4 C and 51 kpa, so phase change rom liquid to gas only occurs above this temperature [3] The critical temperature o R-134a is 16 C [4], which deines the extremes o the temperature range o R-134a, with a critical pressure o 46 MPa Other possible working luids are ammonia, pentane or water [1] All these luids have the advantage over R-134a that they have a higher latent heat, which enables higher heat transer Unortunately, the maximum practical temperature limit o ammonia is C [5], which is too low or the situation at hand Water has a risk o reezing at the lower temperature range Pentane could be a useul alternative or R-134a, considering its temperature range rom - 1 C, the higher latent heat and the higher surace tension [1, 6, 7] A higher surace tension has the beneit o lowering the risk o entrainment, which is the most likely occurring limit in the application o the thermosyphon [8] Possibly, other hydrocarbon rerigerants mentioned by Lee et al[9] are possible working luids as well The type o illing luid, and the operational limits will be subject o later research This paper presents experimental data o airheat pipe-air heat exchangers with two illing ratios The results are compared with those o a model that is based on existing correlations o the literature Results o this study show which conditions oster application o this novel type o heat exchanger 2 EXPERIMENTAL SET-UP AND METHOD OF ANALYZING 21 Test setup A laboratory scale test rig was designed and built to compare the perormances o conventional plate-type exchangers (with water as intermediate medium) and heat pipe equipped plate heat exchangers A range o mass low rates o ambient air o 2 25 kg/s is possible Temperature dierences between hot and cold sides o the heat pipe o 6 C are possible Two illing degrees, F e see Eq (1), o the heat pipe have been examined liquid _ ill _ volume F = e evaporator _ volume (1) oriice m in m out temperature sensors manometers heat pipe heat exchanger Figure 1 Schematic o the test rig air heater In this study, the overall heat transer and temperature distribution are assessed under mass low rates o ambient air varying rom 4 kg/s to 2 kg/s The ambient air temperature varies rom C, whereas the hot air low has temperatures in the range rom 4 7 C The heat pipe is illed with R-134a at ratios o 19% and 59% A schematic overview o the setup is shown in Figure 1 The upper side is the cold side, where ambient air enters Up- and downstream o the heat exchanger temperatures are measured with 16 Pt s (IC Istec ME 9), with an accuracy o 1 C The temperatures o our sensors are averaged and they are denoted as T 1, T 2, T 3 and T 4 (see Figure 2) respectively The sensors are mounted at ¼ and ¾ o the length o the diagonal cross cut o the 645 5 mm 2 rectangular duct The air stream velocity proile was measured and ound to have a homogeneous proile Downstream
the hot section, ten Pt- temperature sensors are mounted to investigate the temperature variation in the height o the pipe at the evaporator section They are mounted vertically at mm apart and 117 mm o the sidewall The Pt- sensors are all calibrated with accuracy better than 1 C or the temperature range o C The measurement section is thermally insulated minimize errors in the heat luxes deduced with A = 52, B = 2484, C = 2631 22 Heat transer perormance To analyze the perormance o the heat pipe equipped heat exchanger, the heat low rate is given by Eq (3) Q = mc ɺ T (3) p ambient air m in m out T 1 T 2 T 4 heat pipe heat exchanger T 3 hot air with the temperature dierences o the air lows up- and downstream the 4 rows o the heat pipe heat exchanger The eectiveness o the heat transer at both the hot and cold side o the heat pipe heat exchanger is expressed in the overall heat transer coeicient and it is deined by Eq (4) [11]: Figure 2 Deinition o temperatures in air streams At the entry, the dynamic pressure measurement with an oriice results in the air mass low rate, at an accuracy o 2% The uncertainties o all measured and calculated parameters are estimated according to [] The air heater is a water-air heat exchanger, with 3 mm spaced vertical ins, which allows a uniorm velocity proile upstream the evaporator This neutralizes the induced swirl in the airlow caused by the radial an The heat exchanger consists o 4 rows o alternating 14 and 13 copper pipes The pipes have an outer diameter o 16 mm and a wall thickness o 8 mm The total length o a pipe is 15 m, with 64 m in the condenser section and the evaporator section each The adiabatic length is 22 m This is the distance between the two sections o the airlow in the wind tunnel The inner surace o each pipe has small spiral grooves, to enhance the heat transer in evaporation and condensation The grooves are 2 mm wide and 2 mm deep each, separated 1 mm, under an angle o 25 with the vertical The distance between the pipes in a row is 365 mm Each row is illed with R-134a separately The rows are 275 mm apart and the total depth o the aluminium ins including the 4 rows is 1145 mm At the top o each row, the pressure is measured with a WIKA type RB manometer, at a requency o Hz, with an accuracy o 1% ater calibration The range o the manometers is MPa The saturation temperature o R134a is given by the Antoine relation (2) obtained rom data rom NIST [4] with temperatures in degrees Celsius and pressure in kpa α tot Q = Aχ T lm (4) with the corresponding Q rom Eq (3), A the heat transerring area o whether the hot or the cold side, χ a geometrical correction actor, here 99 [11], i each side is seen as a cross low heat exchanger with 4 passes and Tmax T Tlm = T max ln Tmin min (5) with T max and T min the maximum and minimum temperature dierences between the airlow and heat pipes o the irst and last row The heat transer rom the air stream to the ins and rom the ins to the tube is described with a in eiciency according to [12] The Nusselt number or the airlow between the ins is given by Hewitt [13] 2 18 14 a S h Nu = 19 Re Pr b d d = 1124Re 65 65 33 (6) with a the tube distance in a row, b the distance between the tube in two successive rows, S the in distance, d the tube diameter and h the in length in gas low direction R The heat resistance o the heat pipe is given by ( r r ) ln 1 = o i w 2πλw L = cond α w A (7) w B T = C A ln ( p ) v (2) with r o and r i the outer and inner radius o the pipe, λ w the thermal conductivity o the copper pipe
and L cond the length o the evaporator or the condenser section respectively The heat transer rom the air stream to the ins and rom the ins to the tube can be described with a in eiciency according to [12] ( ) tanh ml in η in = (8) mlin with δ m = 2α 1+ ( λ δ ) in in in in lin 1 4 (9) with l in the length rom in tip to tube wall, α in the heat transer coeicient to the in, δ in the in thickness, here 2 mm, λ in the thermal conductivity o the in material, here aluminium, 236 W/mK Every tube in the tube bank is observed as having its own segment o ins This leads to a local in length l in o hal the distance between two tubes The heat transer coeicient α in is determined with the Nusselt relation (6) The total heat transerring area equals the heat transerring area o the ins, so rom this assumption ollows [12]: R in = 1 η α A () in in in 23 Phase change heat transer The heat resistance o the condensate in the thermosyphon is deined as: lows down to the evaporator conserving the ilm thickness in the part between the two measuring sections o the wind tunnel At the evaporator, the ilm starts thinning under inluence o the added heat Under stationary operating conditions o the thermosyphon, the overall ilm thickness over the condenser can be described by Eq (13) 4µ λ Tcond L cond δ L = 2 ρ g h g 1 4 (13) with L cond the location at the condenser The Reynolds number o the ilm is given by [12]: Re 4m cond = ɺ (14) µ with mɺ cond the mass low rate o liquid per unit o periphery The condensate mass low rate is equal to Eq (15), see also [5], with the heat low rate obtained rom Eq (3) Q mɺ cond = (15) h g Assuming laminar low, similar to the condenser section, the Nusselt theory can be applied again, now or overall ilm thickness o downward lowing evaporating ilms: 1 1 3 2 3 mɺ cond 2 2 g 4ρ g 3µ 3µ δ = = ρ Re 1/ 3 (16) R δ ( y) 1 α A = = (11) λ A The heat transer coeicient or a vertical tube is given by: with δ the thickness o the condensate layer and λ the thermal conductivity o the luid The local thickness o the condensate is [4], assuming laminar low ollowing Nusselt theory: δ 4µ λ Ty x = 2 ρ g h g 1 4 (12) with ρ the luid mass density, g the gravitational constant, δ x the thickness o the condensate ilm layer, µ the dynamic viscosity o the condensate, T the temperature dierence between saturation temperature o the vapour and wall temperature and h g the latent heat o the luid By draining heat at the condenser rom the vapour, a ilm o condensate orms, which becomes thicker in the direction o the evaporator The ilm 1 2 3 3 4 g 2 3 λ ρ λ α = = Re δ µ 1/ 3 (17) and all luid properties evaluated at the average ilm temperature, see Eq (18), at the inlet: T T T gas,in in,average ilm = (18) 2 Another model or ilmwise condensation is obtained rom Butterworth [15]: Re α = λ µ ρ ρ 18Re 52 122 ( ( ) g ) 1 2 3 v (19) with ρ v the mass density o the vapour In the evaporator the condensed ilm evaporates due to pool boiling The models o Cooper, see Eq
(), and Gorenlo, see Eq (21) describes this phenomena [15] 12 4343ln ( R ) α = 55p p r ( 4343ln ( r )) 55 5 67 p M q () with p r the reduced pressure, R p surace roughness in µm, M molecular weight o the condensate in kg/kmol and q the heat lux ( Rp 4) ( q ) 133 3 9 3 pr α = 4F with PF ( ) 27 r r r r (21) F = 12 p + 25p + p 1 p (22) PF The total heat transer coeicient, see Eqn (23) and (24), is ound rom the summation o the partial heat resistances, which are given by Eqn (7), () and the desired heat transer coeicient o condensation or pool boiling α = 1 tot R A (23) R tot tot in 1 1 1 = + + (24) α A α A α A 3 RESULTS 3 2 w w w in in = 4 kg/s, F e = 19% T 3-76 C histories during a measurement This igure shows that the variation is less than 1 C The heat low rate is measured rom the temperature dierence over the heat exchanger at the evaporator and condenser part o the heat pipe At steady state both heat low rates should be equal Figure 4 shows the comparison o the heat low rates at the evaporator side and condenser part o the experiments This igure shows that the heat low rate o evaporator is about 4% larger than the heat low rate o the condenser, or which we have no explanation Q evap [kw] 25 15 5 5 15 25 Q cond [kw] Figure 4 Comparison o measured heat low rates at evaporator and condenser side o the heat pipe T 4 [ C] 7 65 6 = 4 kg/s, F e = 19% = 12 kg/s, F e = 19% = 4 kg/s, F e = 59% = 12 kg/s, F e = 59% Air temperature [ C] 2 1 1 T 4-59 C T 1-24 C T 2-39 C 55 T 3 = 7 C, T 1 = 4 C 3 4 6 Location rom top evaporator [mm] 6 1 18 24 3 Time [s] Figure 3 Typical histories o air temperatures, see also Fig 2, up- and downstream o the heat exchanger: T 3 =7821 ± 3 C, T 4 =684 ± 2 C, T 1 =2476 ± 3 C and T 2 =413 ± 2 C The measurements are perormed at steady state and each condition lasts 5 minutes Figure 3 shows a typical example o the airlow temperature Figure 5 The eect o illing degree and o mass low rate on temperature distribution downstream o the evaporator In some cases the heat low rate is that high that the heat pipe can dry out Ten Pt s were mounted downstream the evaporator to measure the temperature distribution along the evaporator Figure 5 shows our distributions at two process conditions or two illing degrees o the heat pipe A local, nongradual increase in temperature along
the evaporator indicates a dry out The inner wall i the thermosyphon is in this situation not ully covered with liquid This occurs at low illing degree and high heat low rate (Fig 5) I dry-out occurs, the measurement is skipped rom the analysis Figures 6 and 7 show the perormance o the heat pipe at the evaporator side or various Reynolds numbers and illing degrees In Fig 6 the total heat transer coeicient at F e o 19% is shown, whereas Fig 7 shows results at the higher illing degree The igures show that the perormance increase with increasing heat low rate An increase o the Reynolds number o the airlow leads also to a better perormance Some process conditions have been repeated with a higher illing degree The results are given in Fig 7 A higher illing degree gives a higher overall heat transer coeicient at otherwise identical process conditions 6 4 3 Re = 2 Re = Re = 14 Re = 2 Re = 14 Re = 5 15 25 Figure 6 Measured heat transer coeicient evaporator side or various Reynolds numbers at F e =19% 6 4 3 Re = 2 Re = Re = 14 Re = 2 Re = Re = 14 5 15 25 Figure 7 Measured heat transer coeicient evaporator side or various Reynolds numbers at F e =59% Figures 8 and 9 show the perormance o the heat pipe at the condenser side or various Reynolds numbers and illing degrees Fig 8 presents the total heat transer coeicient at F e o 19% and that o the illing degree o 59% is shown in Fig 9 The igures show that the perormance improves with increasing heat low rate An increase o the Reynolds number o the airlow leads also to a better perormance Some process conditions have been repeated with a higher illing degree The results are given in Fig 9 A higher illing degree gives a higher overall heat transer coeicient at some process conditions The Figs 6-9 show that the perormance o the condenser is better than that o the evaporator at the same heat low rate, i perormance is measured in terms o net heat transer coeicient 6 4 3 Re = 4 Re = 4 5 15 25 Figure 8 Measured heat transer coeicient condenser side or various Reynolds numbers at F e =19% 6 4 3 Re = 4 Re = 4 5 15 25 Figure 9 Measured heat transer coeicient condenser side or various Reynolds numbers at F e =59% 4 ANALYSIS Figures and 11 show a comparison o the measured total heat transer coeicient and predictions based on models o pool boiling o Gorenlo and Cooper [15] Fig shows the comparison at airlow Reynolds number o 2 whereas Fig 11 presents the comparison at Re = 8 In both cases the Gorenlo correlation predicts a higher transer coeicient than Cooper Both
correlations have the same trend with respect to dependency on heat lux as the corresponding measurements The dierence between the two models and the measured ones could be caused by an overestimate o the Nusselt number or the airlow to the ins Here it is assumed that the ins have a constant temperature due to high thermal conductivity o the ins I the temperature is not homogenously distributed the Nusselt number is lower The heat transer rom the air to the ins has a large inluence to the heat transer, so any uncertainty in it is directly relected in discrepancies in comparisons like those o Figs -11 6 4 3 Re = 2 Gorenlo measured Cooper 5 15 25 Figure Comparison o measured and predicted total heat transer coeicient o the evaporator at airlow Reynolds number o 2 6 4 3 Cooper Gorenlo measured 5 15 25 Figure 11 Comparison o measured and predicted total heat transer coeicient o the evaporator at airlow Reynolds number o 8 Figures 12 and 13 show a comparison o the measured total heat transer coeicient and predictions based on models o ilmwise condensation o Butterworth and Nusselt [15] The ilm Reynolds number has been taken into account (laminar gas and laminar liquid and condensate layer with waves) Fig 12 shows the comparison at airlow Reynolds number o 4 whereas Fig 13 presents the comparison at Fig 12 shows a airly good agreement between the predictions and the measurements At higher airlow Reynolds number the dierence between prediction and measured heat transer coeicient becomes large (Fig 13) and in this case the models underpredict the heat transer In both cases Butterworth predicts a higher result than Nusselt However, in both cases (Figs 11-12) the predicted heat lux decreases with increasing heat low rate, which is a dierent trend than the one measured 6 4 3 Butterworth measured Nusselt Re = 4 5 15 25 Figure 12 Comparison o measured and predicted total heat transer coeicient o the condenser at airlow Reynolds number o 4 6 4 3 measured Nusselt Butterworth 5 15 25 Figure 13 Comparison o measured and predicted total heat transer coeicient o the condenser at airlow Reynolds number o 8 For the predictions in Figs -14 the Nusselt relation (6) or airlow has been used Apparently the best predictions are obtained with correlations or boiling in the heat pipe (Gorenlo or even better Cooper) and or convective heat transer on the gas side (Nusselt-relation (6)), i the last one is corrected slightly The reason why this correction is necessary is not ully clear, but might have to do with some loss o contact between ins and tubes 5 CONCLUSIONS The perormance o a heat pipe equipped heat exchanger on a laboratory test rig has been measured and analyzed at the most common process
conditions: various mass low rates o ambient air, various temperature dierences between hot and cold sides o the heat pipe and various illing degrees o the heat pipe The overall heat transer o the heat exchanger has been assessed At the evaporator side to 4 W/m 2 K has been measured and at the condenser side o the heat pipe to W/m 2 K The temperature distribution over the evaporator has been ound to be indicative o proper illing degree The results are rewarding, although more research has to be carried out to ind the most suitable working luid, the optimal heat pipe geometry, operating limits and illing degree A model to predict the heat transer and to calculate the perormance o the heat pipe equipped heat exchanger has been set up This model is a irst step towards a ull predictive model to optimize the heat pipe equipped heat exchanger The result o this study is that a heat pipe equipped heat exchanger can replace a water-cooled heat exchanger without loss o perormance The tested process conditions are typical or warmer countries like Bahrain This study thereore demonstrates that it is possible to apply heat-pipebased cooling equipment in practical conditions o warmer countries [9] Lee, HS, Yoon JI, Kim, JD and Pradeep Bansal, 5, Evaporating heat transer and pressure drop o hydrocarbon rerigerants in 952 and 127 mm smooth tube, Int J Heat Mass Transer, Vol 48, pp 2351-2359 [] Kline, SJ, McKlintock, FA, 1953, Describing uncertainties in single-sample experiments, Mech Eng, Vol 75, pp 3-8 [11] VDI-Wärmeatlas, 1991,: Berechnungsblaetter uer den Waermeuebergang, 6 erw Aulage, VDI Verlag GmbH [12] Rohsenow, W M, Hartnett, J Pand Cho, Y I, 1998, Handbook o Heat Transer, 3 rd ed, McGraw-Hill [13] Hewitt, GF, 1998, Heat Exchanger Design Handbook, Begell House [14] Baehr, HD and Stephan, K, 1998, Heat and Mass Transer, Springer [15] Collier JG and Thome, JR, 1994, Convective boiling and condensation, Clarendon Press REFERENCES [1] Dunn, PD and Reay, DA, 1994, Heat Pipes, ourth edition, Pergamon [2] Wadowski, T, Akbarzadeh, A and Johnson, P, 1991, Characteristics o a gravity assisted heat pipe based heat exchanger, Heat Recovery Systems CHP, Vol 11, pp 69-77 [3] Morgan MJ and Shapiro, HN, 1992 Fundamentals o Engineering Thermodynamics, 2nd ed, John Wiley & Sons, Inc [4] NIST Standard Reerence Database 69, June 5 Release: NIST Chemistry WebBook [5] Unk J, 1988, Ein Beitrag zur Theorie des geschlossenen Zweiphasen-Thermosiphons, Dissertation Technische Universität Berlin [6] Fröba, AP, Penedo Pellegrino, L and Leipertz, A, 4, Viscosity and Surace Tension o Saturated n-pentane, Int J Thermophysics, Vol 25, pp 1323-1337 [7] Fröba, A P, Will, S and Leipertz, A,, Saturated liquid viscosity and surace tension o alternative rerigerants, 14th Symposium on Thermophysical Properties, Boulder, Colorado, USA [8] Reay, DA, 1984, Heat exchanger selection part 4: Heat pipe heat exchangers, Int Research & Development Co Ltd