Experimental study of capillary pumped loop for integrated power in gravity field

Applied Thermal Engineering 35 (2012) 166e176

Contents lists available at SciVerse ScienceDirect

Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng

Experimental study of capillary pumped loop for integrated power in gravity field Laurent Lachassagne*, Vincent Ayel, Cyril Romestant, Yves Bertin Institut PPRIME (UPR CNRS 3346), Département Fluides-Thermique-Combustion, ENSMA, 1 av. Clément Ader e BP40109, 86961 Futuroscope-Chasseneuil, France

a r t i c l e i n f o

a b s t r a c t

Article history: Received 18 April 2011 Accepted 10 October 2011 Available online 20 October 2011

Year after year, thermal dissipation due, for instance, to power electronics, is increasing. The efficiency demand is consequently growing for highly efficient cooling systems as classical solutions are becoming outdated. In this context, Capillary Pumped Loops (CPLs) appear as innovative and efficient heat transfer devices but there is still a lack of data concerning their operating characteristics in gravity field for terrestrial applications. Thus, in this work, a particular design of CPL (called CPLIP) with flat evaporator, designed by the Euro Heat Pipes society in Belgium, has been tested at steady state and transient regime in order to provide data and new insights into thermal and hydraulics couplings of these systems. Ó 2011 Elsevier Ltd. All rights reserved.

Keywords: Capillary pumped loop Gravity field Flat evaporator Power electronics cooling

1. Introduction For some years, terrestrial transport applications, such as the French high speed train (“TGV”), have shown a great increase of heat dissipated by power electronics, for instance. Conventional cooling devices like single-phase loops and even nucleate boiling have reached their limitations. Fortunately, two-phase heat transfer devices appear as the possible next step for power electronics cooling in such applications [1,2]. For almost fifty years, Capillary Pumped Loops (CPLs) and Loop Heat Pipes (LHPs) have been developed and tested for space applications such as NASA’s GLAS [3], CNES’s STENTOR [4] and ESA’s COM2PLEX [5], among many others [6e10]. The CPL, designed by Stenger in 1966 [11], has been mostly developed in United States and Europe whereas LHP appeared in Russia almost at the same time [12] with the work of Gerasimov et al. [13]. However, conceptually, CPLs and LHPs remain very similar. Both can be divided into three major components: evaporator, condenser and reservoir. These parts are connected with bendable pipes forming a loop between dissipative and cooling areas. It allows to set up condenser at a specified position regardless of dissipated heat location. As pointed out by Nikitkin and Cullimore [14], the main difference between LHPs and CPLs is the reservoir position in regards to the evaporator. A CPL’s reservoir is located on the liquid line and is thermally separated from the evaporator whereas a LHP’s reservoir

* Corresponding author. E-mail addresses: [email protected] (L. Lachassagne), vincent.ayel@ ensma.fr (V. Ayel), [email protected] (C. Romestant), [email protected] (Y. Bertin). 1359-4311/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.applthermaleng.2011.10.019

is side by side with the evaporator. A secondary wick between the evaporator and the reservoir makes this thermal link even stronger for LHPs [14]. Thus, Startup is easier for LHPs but their behaviour remains less secure thereafter. CPLs need reservoir preconditioning for startup and their operating temperature can be controlled during operation which is an interesting functionality for power electronics cooling. According to Nikitkin and Cullimore, any middle design between CPL and LHP can be proposed. Depending on the wick efficiency, LHPs and CPLs are able to transfer heat over long distances and even produce a capillary pressure rise exceeding the gravitational losses in the loop. That is why it is challenging to adapt these devices to applications involving gravity field. Thus, Euro Heat Pipe society (EHP, based in Belgium) has designed a “Capillary Pumped Loop for Integrated Power” (CPLIP) in order to respond to the increasing demand for efficient heat transfer devices in terrestrial applications. This design is illustrated in Fig. 1(a). This work will focus on the experimental analysis of operating characteristics and thermohydraulics couplings of the CPLIP during steady state and transient regimes. One major originality of this study is the presence of flowmeters on the loop lines allowing to perform a detailed hydraulic analysis of the CPLIP operation. Joung et al. [15] have also recently studied a flat evaporator CPL with reservoir located above the evaporator. Their “FECPL” design remains the most alike to the CPLIP among other CPL configurations that can be found in literature [16e19]. Yet the CPLIP dimensions are far greater as will be detailed later. Then, contrary to Joung et al. CPL, the CPLIP reservoir is divided into two parts with liquid flowing through the lower part (Fig. 1(a)). Finally, evaporator design is particular (with inner design as property of EHP) and its height is sufficient to consider gravity effects inside evaporator during

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

167

a

b

Fig. 1. CPLIP design by EHP (a) and pressure/temperature diagram of its operation (b).

operation. Despite the differences between CPLIP and Joung et al. FECPL, these systems are similar enough to allow an interesting comparison between both studies. 2. Operating principles The CPLIP operation is described in the Pressure/Temperature diagram of Fig. 1(b) with respect to CPLIP design of Fig. 1(a). Lines are supposed completely insulated in this operating case. Heat power is applied at the evaporator where the saturation interface is located at point 1. The produced vapour flows through vapour grooves to the evaporator outlet (2), inducing pressures losses DP1e2 and superheat DT1e2 of Fig. 1(b). The vapour then flows to the condenser inlet (3) inducing pressure losses DP2e3. After condensation (4) and liquid subcooling (4e5), the liquid flows through liquid line to reservoir inlet (6). The pressure drop DP5e6 is a combination of frictional pressure losses in the line and gravity pressure drop due to height difference between the condenser outlet (5) and the reservoir inlet (6). Liquid flowing through the lower part of reservoir (6e7) is then heated, due to heat transfer between the two reservoir parts (DT6e7). The pressure difference DP7e8 is a combination of gravity pressure drop and frictional pressure losses in the line between reservoir outlet (7) and

evaporator inlet (8). The liquid heating DT7e8 is due to heat conduction from the evaporator body through the pipe. The liquid in the evaporator artery can then enter the wick between the top (9) and the bottom (90 ) of the evaporator. The pressure difference DP9e90 is almost equal to gravity pressure drop due to height between 9 and 90 . Finally, the liquid flows through the wick to the evaporation interface (1) with pressure losses DPwick shown in Fig. 1(b). It is interesting to note that the CPLIP keeps one major operating characteristic of CPL: the operating temperature controllability by the reservoir. As shown by Eq. (1):



Tsat;1 ¼ f Psat;1 ¼ f Psat;10  DP1e10  
¼ f f 1 Tsat;10  DP1e10

(1)

and Fig. 1, the reservoir high part saturation state (10) is the reference point of CPLIP working cycle. In Eq. (1), f is the saturation function of the working fluid and DP1e10 the amount of pressure losses and drops between the vapourisation interface and the higher part of reservoir, which is set by CPLIP design. Therefore, it appears clearly that controlling the reservoir saturation temperature Tsat,10 directly impacts the saturation temperature in the evaporator (Tsat,1). This functionality is particularly interesting in

168

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

order to maintain the temperature level of dissipative equipments placed side by side with the evaporator. However, more like LHP, CPLIP design allows heat and mass transfer between evaporator and reservoir. Indeed, the reservoir is located above the evaporator, thus, vapour bubbles due to percolation in the wick can rise from the evaporator to the reservoir. We will see thereafter that, to some extent, this singularity does not disturb the CPLIP operation. That is why CPLIP operation appears halfway between CPL and LHP operations as described by Nikitkin and Cullimore [14]. Consequently, as it is well known for both LHP and CPL [20], the capillary pressure drop DPcap at the evaporation interface (point 1 on Fig. 1) balances the amount of pressure losses and pressure drops in the whole loop. For example, at the bottom of the evaporator, the capillary pressure drop can be written as follow:

DPcap ðbottomÞxDPf;v;1e2 þ DPf;v;2e3 þ DPcond þ DPf;l;5e6

Table 1 Dimensions and features of CPLIP prototype. Component

Feature

Dimension

Value

Reservoir

Cylindrical

Diameter (mm) Length (mm) Height (mm) Volume (cm3)

99.6e101.6 308 6 58.7

Size (mm3) Thickness (mm) Size (mm3) Pores (mm) e Porosity (%) Permeability (m2) Trapezoid section (mm) Pitch (mm)

320  81  20 1 283  68  16 6.8 e 73 6.53  1013 0.6e0.6e1.2 1.8

Length (m) Primary diameter (mm) Secondary diameter (mm)

5.1 12e14 17.3e21.3

Vapour line

Length (m) Diameter (mm)

1.94 12e14

Liquid line

Length (m) Diameter (mm)

1.92 6e8

Lower part Evaporator

Housing Wick

Vapour grooves Condenser

Coaxial

þ rl gðh6  h5 Þ þ DPf;l;6e7 þ rl gðh8  h7 Þ þ rl gðh90  h9 Þ þ DPwick

(2)

where DPf stands for frictional pressure losses due to vapour or liquid fluid flow, DPcond is the pressure difference between condenser ends and DPwick stands for pressure losses due to fluid flow through the wick. Notice here that, compared with the liquid ones, the gravitational pressure losses in vapour phase have been neglected for the tested working fluid. 3. Experimental features 3.1. CPLIP prototype The whole CPLIP prototype is described in Fig. 2 and Table 1 which gathers the dimensions of each CPLIP component. The lines, reservoir, condenser and evaporator housing sides are made of stainless steel whereas the evaporator wick and heated faces are

made of nickel. Notice that the reservoir/evaporator assembly has been manufactured by Euro Heat Pipe society. The working fluid used for these experiments is ethanol and the fluid charge for these experiments was fixed at 60% of the total CPLIP volume. As mentioned above, the CPLIP reservoir is divided into two parts. This singularity makes the reservoir regulation easier by enhancing the reservoir higher part temperature stability, especially for highly transient regime. Indeed, during operation, liquid flows through the lower part of the reservoir while diphasic higher part is maintained thanks to a PID regulator linked to a temperature probe, a dedicated power supply and a heating cartridge, as shown in Fig. 2. The reservoir and the evaporator are linked by a stainless steel line of 25 cm length. The evaporator is vertically set up and the

Fig. 2. CPLIP prototype design and instrumentation.

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

cross-section in Fig. 2 shows the liquid inlet dividing into three arteries through the sintered nickel wick. Transverse vapour grooves are machined on the whole wick surface area in order to collect vapour in the side grooves appearing in Fig. 2. Notice that this evaporator shape is symmetrical and can thus be heated on one or both faces. CPLIP evaporator and reservoir are linked to the condenser by stainless steel lines whose dimensions are detailed in Table 1. The condenser is a coaxial counter-current heat exchanger with seven return bends. The secondary fluid is a mixture of water and ethylene-glycol at 40%. 3.2. Test set-up The test bench set-up detailed in Fig. 3 allows us to control three boundary conditions during the CPLIP operation:  heat power applied on one evaporator face called “active face”,  secondary fluid temperature at the condenser inlet,  temperature of the reservoir higher part. Consequently, the test sequence is always beginning with reservoir preheating until temperature regulation. Thus the loop is completely filled with liquid before startup. The secondary fluid temperature is then fixed on the chiller and finally a power cycle is applied at the evaporator. Concerning other test-bench features, the non-heated face of the evaporator is supported by an insulator block machined in order to ensure the assembly planarity. The active face is covered with a thermal film of thickness 74 mm, giving a thermal contact resistance of 0.08 kW1 m2. Then, the whole heating surface between the active face and the 8 copper heating blocks is equal to 148 cm2. The copper blocks are linked to four heat power supplies of 1500 W, allowing the total applied heat power to reach 6000 W. Notice finally that the CPLIP lines are not insulated because of a particular application specification.

169

3.3. Measurements The whole CPLIP is fitted with type T thermocouples (0.5  C). On lines and condenser these thermocouples of 0.1 mm wire diameter are located (stuck on the pipe and insulated) as shown in Fig. 2. On the evaporator, the thermocouples of 0.5 mm external diameter are sheathed and inserted in grooves machined on both evaporator faces. There are 16 thermocouples on the active face and 16 other thermocouples on the non-heated face in vis-à-vis ones to the others. Notice that temperatures of the evaporator faces given in the following are mean temperature of the 16 thermocouples. The positioning uncertainty in the grooves is 0.3 mm. Therefore, on the active face, knowing that heat flux can reach 10 W cm2, the whole uncertainty of thermocouples measuring chain reaches 1.5  C. Concerning the interior of the reservoir, the thermal sensor appearing in Fig. 2 is a platinum probe PT100 of 1.6 mm diameter and temperature range [75, 350  C], with an accuracy of 0.1  C at 0  C and 0.26  C at 100  C. The accuracy of these probes allows to identify precisely the saturation state in the reservoir which is also the reference of the CPLIP working cycle (point 10 in Fig. 1). Concerning hydraulic measurements, pressure sensors pictured as squares in Fig. 2 are six absolute pressure transmitters GE Druck PMP 4070 of range [0e3.5 bars]. Finally, two Coriolis flowmeters (MicroMotion Elite manufactured by Emerson society) are inserted on the test bench on the vapour and liquid lines (triangles on Fig. 2). In comparison with other flowmeters, these Coriolis flowmeters induce less pressure losses in the loop with higher measurement accuracy. Indeed, the uncertainty on mass flow rate measurement is low enough not to appear on graphics scale. However, another kind of error appeared during experiments, concerning vapour mass flow rate. Fluid inertia is a key parameter to ensure the accuracy of these equipments, and yet the fluid density measured in the flow at the evaporator outlet showed that vapour condensation starts in the vapour line before entering the flowmeter. In this case, measures given by the vapour flowmeter will appear on the graphics but cannot be analysed.

Fig. 3. Test bench set-up.

170

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

4. Results and discussion 4.1. CPLIP specification test Figs. 4 and 5 show the CPLIP response to a specification test performed on the test bench. During the first part of these tests, rising power steps of 200 W are applied on the evaporator active face, waiting for steady-state regime between each step. Then, after reaching steady-state operation at the maximum applied heat power (fixed by user), decreasing power steps of 200 W are applied until power shutdown. This kind of test has been performed for several cold source temperatures. Notice on Fig. 4 that the maximum temperature variation of the evaporator active face (baseplate of power electronics) that was observed is only 9  C for a heat power applied variation of 1600 W. Moreover, assuming that for some kind of application, a non dissipative power electronics module is stacked on evaporator back face, it would also be maintained at the regulation temperature level during the CPLIP operation. Concerning the others temperatures of the CPLIP, it appears that mean liquid line temperature remains stable at 20  C, meaning that the counter-current heat exchanger of this test bench is oversized enough to ensure that the outlet temperature of working fluid remains the same as the inlet temperature of the cooling fluid. The gap between the evaporator/ reservoir line temperature and the evaporator inlet temperature can be explained by heat conduction inside the stainless steel pipe. Concerning the thermal response of the CPLIP, the mean temperature of vapour line remains stable at 72  C. Thus, vapour condensation probably starts inside vapour line. Indeed, lines are not insulated and consequently sensitive to convective heat transfer inside the test room, leading to the little lines temperature instabilities appearing in Fig. 4. It appeared during tests that this feature has no influence on CPLIP behaviour. Fig. 5(a) gathers the two corresponding mass flow rates given by the liquid and vapour Coriolis flowmeters. An issue for vapour mass flow rate measurement appears immediately. For 1600 W of heat power applied, the mean value of vapour mass flow rate is decreasing and a gap of 0.54  1013 kg s1 appears between the vapour and liquid mass flow rate measures. At 200 W heat power level, the vapour mass flow rate value appears also wrong. The fluid density, also given by the flowmeter, can lead to an explanation for

Fig. 5. CPLIP response to a specification test (reservoir 73  C and condenser 20  C): (a) Mass flow rate and (b) absolute pressure.

this behaviour. Whatever the heat power step is, fluid density value given by the vapour flowmeter shows that some liquid is always present inside the pipe. The vapour condensation starts in the vapour line on this test bench, as expected due to non insulation of lines. One can then assume reasonably that the liquid and vapour phases distribution inside the line impacts directly the mass flow rate measurement. Unfortunately, this assumption cannot be verified on the present test bench. However, all along the tests, fluid density given by the liquid flowmeter remains constant and equal to the NIST value for liquid density of ethanol [21]. That is why in this work, mass flow rate analysis during CPLIP operation has been based on liquid flowmeter measures only. Finally, for each heat power step, liquid mass flow rate average value is related, at first order, to the CPLIP energy equation:

_ ¼ m

Fig. 4. CPLIP response to a specification test (reservoir 73  C and condenser 20  C): Temperature.

Qevap Hlv

(3)

Fig. 5(b) shows the absolute pressure variations during this test measured by the six sensors of the CPLIP test bench. First, the absolute pressures given by the six sensors can be gathered into two groups separated by pressure level of around 5000 Pa, which corresponds to gravity pressure drop in the liquid line. In the same

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

way, mean pressure differences between reservoir higher and lower parts as well as the evaporator inlet are for most part due to gravity. This figure also points out two kinds of instabilities inside the CPLIP. The first kind could be called “low amplitude instabilities” which are detected by all the sensors but appear clear on the measures located inside the reservoir and at the evaporator ends. These variations amplitude never exceeds 1000 Pa. These low amplitude instabilities are directly linked to the reservoir thermal regulation. In fact, in order to maintain the temperature in the reservoir higher part, the PID regulator permanently adjusts the power supply of the reservoir heater. Then, the temperature in the reservoir two-phase part varies with respect to time with an amplitude of 0.3  C around the regulation temperature of 73  C. Now, for ethanol at saturation state, this temperature variation leads to a pressure variation of around 1000 Pa, which corresponds to the amplitude of this first kind of instabilities. The second kind of pressure variations will be called “high amplitude instabilities” and appears only at the condenser ends, where they stack on the low amplitude variations already mentioned. These instabilities amplitude reaches 10,000 Pa for the sensors located in the vapour flow at the condenser inlet and 35,000 Pa for the sensor located in the liquid flow at the condenser outlet. Thus, the particular flow inside the condenser leads to strong instabilities which fortunately do not reflect at the evaporator. In that case, the vapourisation interface would certainly be destroyed because the maximum value of capillary pressure drop is around 10,000 Pa. Notice that loud noise snaps located at the condenser can be heard during tests, which are probably linked to these pressure instabilities. A more detailed and instrumented test bench dedicated to the condenser could allow deeper investigation of this phenomenon. In the following, the CPLIP response during startup is presented, then the transient response to heat power variations and finally, some studies concerning the CPLIP steady-state operation conclude this experimental work.

171

increases (blue curve of Fig. 7(a)). This liquid withdrawal from the reservoir should lead to a pressure decrease with liquid evaporation inside the two-phase part of the reservoir. That’s why one could expect a decrease of the reservoir saturation temperature, which does not appear on the figure. It seems that the reservoir capacity is sufficient to balance this phenomenon and avoid this issue. CPLIP operation quickly returns to steady state with a lower amount of subcooling. Concerning absolute pressure measurements (not included in this paper), tests have shown that neither increasing, nor decreasing heat power steps have any influence on the amplitude and frequency of the pressure instabilities observed at both ends of the condenser. 4.3. CPLIP response to heat power cycling The power electronics cooling application will lead the CPLIP to be subjected to heat power cycling on the evaporator active face. This part presents CPLIP response to two different power cycles appearing in Fig. 8. The first one is a harsh power cycle similar to heat power dissipation of power modules chips. The second one is built by taking running average of first power cycle. The result is

4.2. CPLIP response to heat power steps A detailed study of a CPL response to heat power steps was provided by Pouzet et al. in 2004 [22]. Their comparison between experiments and numerical simulations has pointed out some issues in their loop response to decreasing heat power steps due to condenser/reservoir couplings. Here, the following study of CPLIP response to heat power steps will provide experimental data for the case where gravity has a major influence on the CPL operation. As shown in Fig. 6(a), an increasing heat power step of 700 W (from 300 W to 1000 W) leads to an evaporator active face temperature increase of 1.5  C. The little peak at the evaporator inlet temperature indicates that percolation has occurred through the wick leading to vapour rise in the evaporator/reservoir line, but this peak does not appear for thermocouple located above on the reservoir/evaporator line. It means that this vapour has certainly condensed before entering the reservoir. Indeed, the mass flow rate peak (seven times the average value of mass flow rate before the heat power step) observed in Fig. 6(b) shows that a great amount of cold liquid flows from condenser to reservoir during this rising power step. Then, this additional amount of subcooled liquid is sufficient to stop percolation and fill the wick again. By this way, as other kinds of loops, CPLIP response to rising heat power steps seems more stable than to startup. Fig. 7 shows a decreasing heat power step of 700 W. There is no sign here of percolation through the wick. However, as shown by the red curve of Fig. 7(b), liquid mass flow rate in the CPLIP decreases quickly, until reversing and finally going back to a steadystate value. Consequently, some liquid withdraws from the reservoir to the condenser and the liquid line temperature slightly

Fig. 6. CPLIP response to an increasing heat power step of 700 W (reservoir 73  C and condenser 20  C): (a) Temperature and (b) mass flow rate.

172

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

Fig. 7. CPLIP response to a decreasing heat power step of 700 W (reservoir 73  C and condenser 20  C): (a) Temperature and (b) mass flow rate.

a power cycle with smoothed heat power transitions compared with the first one. Fig. 9 gathers some CPLIP temperature variations and liquid line mass flow rate for a better understanding of CPLIP response to each power cycle. First observation: as shown by Fig. 9, the CPLIP temperature regulation is well operating in the second case (smoothed power cycle), whereas, for harsh heat power cycle, CPLIP does not completely operate as could be expected, due to the highly transient nature of heat power applied. This is why some particular investigations are ongoing about this particular test set-up in order to gain a better understanding of CPLIP behaviour in this case. Concerning startup, for both cases the first “temperature rising stage” lasts around 15 min. For the first heat power cycle (Fig. 9(a)), evaporator temperature does not stabilize. During 15 min after the vapourisation onset, the loop operates with enough subcooling. But around 30 min after the test beginning, some temperature peaks occur at the evaporator active face. It indicates the wick’s dry-out with percolation which spreads step by step at the whole evaporator top. The highly transient nature of increasing or decreasing heat power variations allows the liquid to fill again the wick after each dry-out. But heat flux entering the reservoir, due to percolation in the evaporator added to liquid withdrawals from reservoir to condenser (negative mass flow rates and liquid line temperature peaks), leads to regulation shutdown and an increase of the reservoir saturation temperature for more than 10  C. Consequently, reference pressure in the CPLIP (not shown in this paper) is also increasing leading to evaporator temperature rise, according to Eq. (1). Only lower amplitude and frequency of power variations at the end of the cycle (after 15 min) allow the CPLIP to operate again as expected. Concerning the smoothed power cycle, Fig. 9(b) shows some reservoir/evaporator temperature rises meaning percolation through the wick, or negative mass flow rate indicating liquid withdrawal from the reservoir to the condenser. However, none of these phenomena has a sufficient amplitude to annihilate balance between heat transfers inside reservoir, thus leading to regulation failure. This is the expected CPLIP operation in response to heat load: with always enough subcooling to ensure no wick dry-out or wick dry-out recovery if percolation occurs through the wick. A compromise certainly needs to be found with thermal inertia and thermal resistance between evaporator face and power electronic

Fig. 8. Power cycling on the evaporator for a 10 min sample.

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

173

evaporator temperature. Notice that only working point for 1400 W heat power applied and around 10  C secondary fluid inlet temperature seems to contradict this conclusion. In this case, the temperature at the evaporator wall is around 2  C higher than the mean level for 1400 W applied heat power. This may be due to ethanol liquid viscosity rising at low temperatures, leading to sufficient additional pressure losses with respect to the gravitational ones for the liquid flow inside the CPLIP. Yet, this sensitivity remains minor (around 2  C) and does not invalidate the regulation ability of the CPLIP evaporator temperature whatever the heat power applied and the cold source temperature in tested range. This very interesting behaviour for thermal regulation application is allowed by the adaptability of the subcooling length in the testbench condenser. Fig. 10(b) shows the mass flow rate measured at condenser outlet by liquid flowmeter for same working points as Fig. 10(a). For analysis, this value will stand for mass flow rate in the whole loop at steady state because of uncertainty already pointed out on vapour line flowmeter measurements. Thus, mass flow rate in the CPLIP slightly increases with increasing cold source temperature but has no real sensitivity to this parameter. When cold source temperature is rising, the temperature of the fluid entering the evaporator also becomes higher leading to a heat leak decrease. Then, higher part of applied power is involved in liquid vapourisation (Qevap in Eq. (3)). This leads to a higher resulting mass flow rate as shown by the following equations, resulting from Eq. (3):

_ ¼ m

_ Qtot  Qloss  mCp Tevap  Tart Hlv


(4)

where Qtot is the total heat power applied, Qloss stands for heat losses to the ambient, Tevap is the vapourisation temperature and Tart is the liquid temperature at the evaporator inlet. Then,

_ ¼ m

Fig. 9. Temperature and mass flow rate variation in the CPLIP during power cycling (reservoir 73  C and condenser 20  C).

baseplate in order to “smooth” heat power variations applied at the evaporator. 4.4. Steady-state operation 4.4.1. Sensitivity to cold source temperature We present here the sensitivity analysis of the CPLIP steadystate operation to cold source temperature. Steady-state working points presented in Figs. 10 and 11 stand tests with 73  C fixed at the reservoir and a secondary fluid inlet temperature range of [10  C; 40  C]. Fig. 10(a) presents the average temperature of active and back evaporator faces as functions of secondary fluid inlet temperature for applied powers of 600, 1000 and 1400 W. This figure clearly indicates that cold source temperature has no influence on the

Qtot  Qloss
Hlv þ Cp Tevap  Tart

(5)

This mass flow rate theoretical value also appears in Fig. 10(b). The heat losses to the ambient on the CPLIP test bench has been estimated around 80 W by experiments performed with an empty loop. The values taken for calculation of Tevap and Tart were respectively the value of regulation temperature inside the reservoir (72  C) and the temperature measured on the reservoir/evaporator line. Fig. 10(b) shows good agreement between this theoretical values and mass flow rate given by the liquid flowmeter whatever the working point of the CPLIP is. Finally, according to Fig. 10(c), for 1000 W applied heat power, cold source temperature has no influence on pressure drops in the loop. One can note that no trend clearly appears concerning the condenser and liquid line pressure drops evolutions. These symmetrical variations are due to high instability of absolute pressure measurements at the condenser inlet as shown in Fig. 5 for the specification test of the CPLIP. 4.4.2. Hysteresis phenomenon To conclude this steady-state operation analysis, an hysteresis phenomenon observed on the evaporator active face temperature is presented. This phenomenon was also observed by Lossouarn [1], but Joung et al. [15] found it negligible for their FECPL. Fig. 11 points out the fact that the mean temperature of the evaporator active face is higher for decreasing power steps than for increasing ones. This hysteresis amplitude can reach 2  C. Lossouarn [1] links this phenomenon to the vapourisation interface penetration through the wick. Indeed, in a porous media saturated by two fluids, drainage operation is easier than imbibition. Drainage stands for a non-wetting invading fluid displacing a wetting fluid whereas

174

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

Fig. 10. CPLIP working points as function of cold source temperature (reservoir 73  C): (a) evaporator temperature wall, (b) mass flow rate and (c) pressure drops measurements.

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176

175

Fig. 11. Temperature hysteresis of evaporator active face as function of heat power applied (reservoir 73  C and condenser 20  C).

imbibition occurs when a wetting fluid displaces a non-wetting fluid. When power is increasing, the vapourisation interface penetrates more and more through the porous wick (evaporator porous media saturated with liquid and vapour is in drainage situation), leading to the evaporator wall temperature increase. Whereas when power is decreasing, the harder imbibition of the wick by liquid must lead to the observed heat transfer deterioration. This phenomenon seems close to “capillary hysteresis” pointed out by Vershinin and Maydanik concerning Loop Heat Pipes operation [23]. In addition, the wick geometry, which is quite different between CPLIP and Joung’s FECPL (CPLIP wick volume is around 35 times greater than FECPL one), can have a major influence on hysteresis amplitude. It seems that the thicker the wick is, the stronger the hysteresis will be. Notice finally that the average temperature of the evaporator back face is not affected by this particular behaviour.

rate of a vapour flow with these instruments. Indeed, the reliability of these measurements depends totally on the presence of condensation inside the flowmeter pipe. 4. Transient operation tests with power cycling at the evaporator have pointed out the compromise to find with thermal inertia and thermal resistance between evaporator face and power electronic baseplate, in order to “smooth” heat power variations without lowering too much the CPLIP temperature regulation ability. 5. An hysteresis phenomenon has been revealed on the evaporator active face. This hysteresis may be due to difference between drainage and imbibition in the wick. It seems that the thicker the wick is, the higher the hysteresis amplitude will be. 6. Finally, in the tested range, no sensitivity of the CPLIP operation to cold source temperature has been observed. This is allowed by the adaptability of the subcooling length in the test-bench condenser.

5. Conclusions Acknowledgements The Capillary Pumped Loop for Integrated Power developed by EHP society has proven its efficiency as a power electronics thermal regulation device. The tests performed at PPRIME Institute provided numerous and useful thermohydraulic data and allowed analysis of both steady state and transient regimes of the CPLIP. The major conclusions drawn out by this study are: 1. The reservoir regulation has confirmed its usefulness for controlling the loop temperature level. Then, the CPLIP temperature regulation ability essentially depends on thermal resistance between vapourisation interface and power electronics baseplate which can lead to variation of only 9  C for a power applied variation of 1600 W on this test bench. 2. The key of CPLIP operation is the balance between heat fluxes inside the reservoir, especially due to liquid subcooling from the condenser and vapour rising from the evaporator. Tests have shown that CPLIP can operate well with percolation through the wick as long as liquid subcooling remains sufficient to balance the vapour amount to the reservoir. 3. Mass flow rate measurements were performed on this test bench with accurate Coriolis flowmeters. This work has pointed out the difficulty of measuring correctly the mass flow

This work was performed thanks to the support of PSA PeugeotCitroën, a partner of the European project HI-CEPS (“Highly Integrated Combustion Electric Propulsion System”). See http://www. hi-ceps.eu/ for more informations. References [1] D. Lossouarn, Etude théorique et expérimentale du refroidissement diphasique à pompage capillaire de convertisseurs de puissance à haute densité de flux de chaleur pour la traction ferroviaire, Thèse de doctorat, Université de Poitiers, ENSMA, Futuroscope, France, 2008. [2] V. Dupont, S.V. Oost, L. Barremaecker, S. Nicolau, Experimental investigations on a methanol capillary pumped loop equipped with four flat evaporators, in: 15th International Heat Pipe Conference (15th IHPC), Clemson, USA. [3] D. Douglas, J. Ku, T. Kaya, Testing of the Geoscience Laser Altimeter System (GLAS) Prototype Loop Heat Pipe, Technical Report, AIAA Paper 99-0473, 1999. [4] F. Mena, W. Supper, C. Puillet, Design and Development of Loop Heat Pipes Technical Report 2000-01-2315. Society of Automotive Engineers, Warrendale, PA, 2000. [5] R. Schlitt, M. Dubois, L. Ounougha, W. Supper, COM2PLEX e A Combined European LHP Experiment on CPACEHB/QUEST Technical Report 2000-012457. Society of Automotive Engineers, Warrendale, PA, 2000. [6] K.A. Goncharov, M.N. Nikitkin, O.A. Golovin, Y.G. Fershtater, Y.F. Maidanik, S.A. Piukov, Loop Heat Pipes in Thermal Control Systems for OBZOR Spacecraft

176

[7]

[8]

[9]

[10]

[11] [12] [13]

[14]

L. Lachassagne et al. / Applied Thermal Engineering 35 (2012) 166e176 Technical Report SAE951555. Society of Automotive Engineers, Warrendale, PA, 1995. D. Kozmine, K. Goncharov, M. Nikitkin, Y.F. Maidanik, Y.G. Fershtater, S. Fiodor, Loop Heat Pipes for Space Mission Mars 96 Technical Report SAE961602. Society of Automotive Engineers, Warrendale, PA, 1996. C.L. Baker, W.B. Bienert, A.S. Ducao, Loop Heat Pipe Flight Experiment Technical Report SAE981580. Society of Automotive Engineers, Warrendale, PA, 1998. Q. Mo, J. Liang, A novel design and experimental study of a cryogenic loop heat pipe with high heat transfer capability, International Journal of Heat and Mass Transfer 49 (2006) 770e776. G. Wang, D. Mishkinis, D. Nikanpour, Capillary heat loop technology: space applications and recent Canadian activities, Applied Thermal Engineering (2008) 284e303. F.J. Stenger, Experimental Feasibility Study of Water-filled Capillary-pumped Heat-transfer Loop, NASA TM X-1310, Lewis Research Center, Cleveland, OH. Heat pipe, USSR Inventors Certificate 449213, 1974. Y.F. Gerasimov, Y.F. Maidanik, et al., Low-temperature heat pipes with separate channels for vapor and liquid Technical Report, Engineering Physics Journal 28 (6) (1975) 957e960 (in Russian). M. Nikitkin, B. Cullimore, CPL and LHP technologies: what are the differences, what are the similarities?, in: International Conference on Environmental Systems, number 981587 in SAE Technical Paper.

[15] W. Joung, H. Hwang, J. Lee, Experimental study on the operating characteristics of a capillary pumped loop with a flat evaporator, International Journal of Heat and mass Transfer 53 (2010) 268e275. [16] Y. Cao, A. Faghri, Conjugate analysis of a flat-plate type evaporator for capillary pumped loops with three-dimensional vapor flow in the groove, International Journal of Heat and Mass Transfer 37 (1994) 401e409. [17] J. Yu, H. Chen, H. Zhao, Y. Li, An experimental investigation on capillary pumped loop with the meshes wick, International Journal of Heat and Mass Transfer 50 (2007) 4503e4507. [18] H. Lin, W. Lin, An axial heat transfer analytical model for capillary-pumped loop vapor line temperature distributions, Applied Thermal Engineering (2007) 2086e2094. [19] E. Bazzo, R. Riehl, Operation characteristics of a small-scale capillary pumped loop, Applied Thermal Engineering 23 (2003) 687e705. [20] Y.F. Maydanik, Loop heat pipes, Applied Thermal Engineering 25 (2004) 635e657. [21] NIST Scientific and Technical Database, National Institute of Standards and Technology, http://webbook.nist.gov/chemistry/fluid/. [22] E. Pouzet, J. Joly, V. Platel, J. Grandpeix, C. Butto, Dynamic response of a capillary pumped loop subjected to various heat load transients, International Journal of Heat and Mass Transfer 47 (2004) 2293e2316. [23] S.V. Vershinin, Y.F. Maydanik, Hysteresis phenomena in loop heat pipes, Applied Thermal Engineering 27 (2007) 962e968.