- Email: [email protected]
Double condenser pulsating heat pipe cooler
Applied Thermal Engineering xxx (2017) xxx–xxx
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research Paper
Double condenser pulsating heat pipe cooler Daniele Torresin ⇑, Francesco Agostini, Adrian Mularczyk, Bruno Agostini, Mathieu Habert ABB Switzerland Ltd., Corporate Research, Segelhofstrasse 1K, 5405 Baden-Dättwil, Switzerland
h i g h l i g h t s A newly designed pulsating heat pipe cooler (PHP) based on automotive technology is investigated. The PHP has two condenser areas, one above and one below the evaporator. Measurements were performed with the refrigerant fluid R245fa in three different orientations. The results show that the device is orientation-free with identical performance in all tested orientations.
a r t i c l e
i n f o
Article history: Received 9 September 2016 Revised 2 February 2017 Accepted 15 February 2017 Available online xxxx Keywords: Pulsating heat pipe Double condenser Heat transfer Power electronic
a b s t r a c t A novel pulsating heat pipe cooler (PHP) based on automotive technology is presented. This technology uses numerous aluminum Multiport Extruded (MPE) tubes disposed in parallel to achieve the desired compactness. The sub-channels of the MPEs are connected in a serpentine manner by means of fluid distribution elements integrated in the two condenser manifolds. This configuration enables the oscillation of liquid slugs and elongated bubbles between the evaporator and the condenser areas. This power electronics cooling system with a double condenser area was experimentally characterized and the thermal performances measured with R245fa are presented in this paper. The influence of different parameters such as the heat load, fluid filling ratio and orientation are investigated. The results show that the device is orientation-free: the performance was identical in all orientations, vertical, horizontal and anti-vertical. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction Power electronics and microelectronic cooling requirements are becoming every year more demanding due to the continuous increase of power density. Pulsating heat pipes (PHP) have emerged in the last years as suitable cooling devices for dissipating the high heat loads generated by electronic devices since they allow for the extension of air cooling’s applicability into areas nowadays covered by water-cooling [1]. These devices are entirely passive and they comply with long-term operation without maintenance. Furthermore, the possibility to work independently from the orientation allows higher flexibility on their mechanical integration compared to two-phase thermosyphon coolers. The overall functionality of PHPs is still not completely understood [2]. The complex interplay of local hydrodynamics, thermodynamics, and phase change phenomena make a comprehensive model extremely difficult to develop. Reviews of past research as well as unresolved issues relating to PHPs can be found in works by Zhang and Faghri
⇑ Corresponding author.
[3] and Khandekar et al. [4]. Research efforts continue to move toward a more comprehensive understanding of this technology. Experimental investigations test the performance limits of PHPs with different design parameters [5,3,7]. Among these, Lin et al. [6] tested experimentally a water-cooled open-loop PHP with two symmetrical condensers using fluorocarbons as the working fluids. The experimental configuration consisted of a copper tube with an inner diameter of 1.75 mm which was bent in 40 turns. The PHP dissipated a maximum heat load of 2.04 kW, representing a heat flux of 6.4 W/cm2 and a thermal resistance of 48 K/kW. Experiments reported similar thermal conductivities for horizontal and vertical orientations, showing that the heat pipe was orientation independent. Numerical models have also been developed in an attempt to predict the thermal and hydrodynamic behavior of these devices [8–11]. Nikolayev [12] has produced a recent work that numerically and theoretically predicts oscillation development and start-up criterion for a single branch PHP. All current models include essential simplifications in order to keep the problem tractable. Future work must continue to capture effects that have thus far been neglected such as boiling, contact
E-mail address: [email protected] (D. Torresin). http://dx.doi.org/10.1016/j.applthermaleng.2017.02.066 1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
2
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
Nomenclature Cond Cp,air FR _ air m dp Pmax Q Qair
condenser air specific heat capacity (J/kg K) filling ratio (%) air mass flow rate (kg/s) differential pressure (bar) maximum operative pressure input power (W) power extracted by air stream (W)
angle, etc. Continued experimental work is important to provide reference and verification for the models. 2. Experiments Fig. 1 reports a schematic of the novel pulsating heat pipe heat exchanger [13]. The cooler is made of 11 parallel aluminum multi-port extruded (MPE) tubes having capillary dimensions and connected by fluid distribution elements in the two condenser manifolds. Each MPE tube has 7 sub-channels with rectangular cross section and 1.54 mm hydraulic diameter. The heat load is transferred to the internal channels by an aluminum baseplate (evaporator) in which several grooves are machined and the MPEs are embedded. The heat is then transferred to the two opposite ends where the MPEs are thermally connected using louvered fins (condenser) and finally transferred to the external air by forced air convection. The pulsating heat pipe configuration is achieved by connecting in series the sub-channels of the MPEs at the manifolds [14] enabling the pulsation as shown in Fig. 2. The manifold design used results in a total of 27 turns and an open loop configuration. The newly designed cooler has two condensing areas, one at the top and one at the bottom of the evaporator baseplate, where the heat source device is mounted. In such a way, the fluid has the possibility to pulsate/oscillate in both directions symmetrically. Furthermore, this adds the benefit of having the fluid turns in the manifolds with the fluid being most likely in the liquid state thus reducing the local pressure losses. The PHP has a width of 155 mm and a height of 134 mm for both condensing areas and 125 mm for the evaporator baseplate, respectively. A schematic of the test facility is reported in Fig. 3. Tests have been performed by using one centrifugal fan and a T-conduit to split the air flow between the two condensers. To ensure equal mass flow rates, the pressure drop across each condenser was measured (dp1 and dp2) and kept equal by opening and closing the respective valves (V1 and V2). The overall mass flow of air was
Rth Tb,max Tai Tao u Vair DT
thermal resistance (K/W) maximum base plate temperature (°C) inlet air temperature (°C) outlet air temperature (°C) uncertainty volumetric air flow rate (m3/h) temperature difference (K)
controlled by adjusting the fan speed and monitored using a flow meter (FM). Furthermore the air inlet temperature was regulated by the heater (H2) and set by a PID controller. The heat load was supplied to the system by means of six ceramic heaters (H1) each having a size of 50 50 mm and monitored using a Hameg (HM8115-2) power meter. All signals have been acquired through a National Instruments SCXI box connected to a laptop running Labview. Thermocouples have been calibrated using an Omega DP97 precision thermometer with platinum probes to measure the reference temperature and a Lauda R207 chiller to control the temperature. 3. Data reduction From the experimental campaign, it is possible to determine the thermal resistances of the cooling device. The thermal resistance is calculated as follows:
Rth ¼
T b;max T ai Q
ð1Þ
where Tai is the inlet air temperature, Q is the heat load to be dissipated by the cooler and Tb,max is the maximum base plate temperature measured with the 9 temperature measurements Tb. The thermocouples are placed inside 1.2 mm diameter holes drilled from the backside of the baseplate and with a depth reaching 1 mm to the top surface. The thermocouple tip was fixed to the baseplate holes by means of a highly thermally conductive epoxy (supplier omega, type 101), thermal conductivity 1.04 W/m K. Several thermocouples have been installed to monitor the air temperature conditions: one at each condenser inlet cone and three on each outlet. From the measured air volumetric flow rate, the heat exchanged on the air side can be calculated with a thermal balance (assuming uniform air flow rate):
_ air cp;air ðT ao T ai Þ Q air ¼ m
ð2Þ
The above equation can be evaluated for each condenser: the term Tao represents the mean outlet temperature of the six measured val-
Fig. 1. Schematic of test section and thermocouple positioning.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
3
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
“cross over” connection
“in line” connection
Fig. 2. Manifolds design and schematic of MPEs’ sub-channels connection.
Fig. 3. Test setup schematic.
_ air .is the air mass flow rate. The heat balance, repreues, while m senting the difference between the electrical power applied to the ceramic heaters and the sensible heat received by the air stream can be easily evaluated by comparing Eq. (2) and the power measured with the power meter. As previously mentioned, the operative conditions of a PHP device are determined by a combination of many parameters. One of the most important among them is the filling ratio. Particular care has been taken in the filling ratio procedure: a precision scale (Mettler Toledo Model MS32001L/01) with a repeatability and readability of 0.1 g has been used to estimate the amount of fluid inside the cooler before running each experiment. Fluid temperature and pressure were also monitored with thermocouples inserted in one channel of the MPE tubes and by using an Omega PX409-500A pressure transducer.
The uncertainties of the measured signals are reported in Table 1. The error of the thermal resistances calculated applying the standard Kline and McClintock uncertainty analysis [15] has been found to be below 0.2 K/kW at 2400 W and below 1 K/kW at 500 W. The maximum uncertainty for the air side heat balance
Table 1 Uncertainties of different measured signals. Description
u
Temperature (°C) Pressure (bar) Power (W) Volumetric air flow rate (m3/h)
±0.1 ±0.02 ±16.2 ±24.6
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
4
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
has been calculated to be below 85 W with a volumetric flow rate of 750 m3/h and 2.4 kW losses. 4. Results The next paragraphs summarize the experimental results obtained with the refrigerant fluid R245fa. Focus is given on the analysis of the effect of several parameters on the thermal performance. Three different orientations have been tested as shown in Fig. 4: vertical (0°), horizontal (90°), and anti-vertical (180°). 4.1. Effect of fluid filling ratio One of the most important parameters to investigate of a PHP is the amount of fluid filled compared to its total internal volume. This parameter is defined as filling ratio. An optimal value is characteristic of each device and depends on the boundary working conditions. Fig. 5 shows the thermal resistance as a function of the filling ratio in vertical orientation with a volumetric air flow rate of 750 m3/h and inlet temperature of 23 °C. The different curves refer to heat loads set from 750 W to 2.4 kW. The filling ratio varies from 40 to 75% and the trend of the thermal resistance is the same for all the heat loads: the minimum of thermal resistance has been found to be 27 K/kW at the optimal filling ratio of 60%. At heat loads higher than 1 kW, no data points are reported below 50% FR. The reason for that is because either the maximum criteria for the operative temperature (Tb,max = 120 °C) or pressure (Pmax = 10 bar) for running the test was reached. The optimization procedure for the filling ratio has also been performed horizontally. As shown in Fig. 6 the trend of thermal resistance is the same as for the vertical orientation. The optimum filling ratio in this orientation was also found to be 60% for all the investigated heat loads. At high filling ratio (75%) in horizontal orientation, the cooler was working up to 2 kW losses while in vertical orientation the upper limit at this high filling ratio was found at 1.75 kW. Compared to the vertical case, in horizontal orientation, also for fillings ratio below 50% the maximum criteria for the operative temperature (Tb,max = 120 °C) or pressure (Pmax = 10 bar) for running the test were not reached even at heat loads of 1.25 kW.
Fig. 5. Thermal resistance as a function of filling ratio in vertical orientation and heat load from 0.75 to 2.4 kW.
Fig. 6. Thermal resistance as a function of filling ratio in horizontal orientation and heat load from 0.75 to 2.4 kW.
4.2. Effect of orientation Fig. 7 reports the hotspot thermal resistance as a function of the heat load at the best filling ratio (60%) for three different orientations: vertical (0°), horizontal (90°) and anti-vertical (180°). The thermal resistance is decreasing with the heat load in all the
orientations. In particular, in vertical and anti-vertical orientation the performance are the same up to 1.5 kW. For values higher than 1.5 kW the thermal resistance in anti-vertical orientation becomes
Fig. 4. Schematic of tested configurations.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
5
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
Fig. 7. Thermal resistance as a function of the heat load in three different orientations.
independent from the heat load (29 K/kW) while for the vertical and horizontal cases it continues to decrease with the heat load. An explanation of the different behavior between vertical and anti-vertical orientation at heat loads higher than 1.5 kW might be the different design of top and bottom manifolds (see Fig. 2). For heat loads lower than 1 kW, the thermal resistance in horizontal orientation is lower by 10–15% compared to the vertical and anti-vertical cases. A plausible explanation of this behavior is that in horizontal orientation the non-presence of the gravity effect allows a balanced fluid flow in both condensers. 4.3. Heat balance The graph in Fig. 8 compares the heat removed by the air stream calculated with Eq. (2) and the input heat measured with the power meter for all three orientations. The graph reports also the heat balance of the two different air streams going through the PHP. The dotted line borders represent a ±5% margin of error compared to the measured input power and the dash-dotted line rep-
resents the case of an even split of power into the two airstreams. Note that Cond 1 refers to the condenser that is in the bottom position and Cond 2 to the condenser on the top when referring to the 0° orientation. Furthermore, in this paragraph ‘‘top” refers to the condenser being physically in the top position and ‘‘bottom” to the condenser in the physical bottom position respectively. In the investigated range of heat loads, at the best filling ratio, the heat balance is with a ±5% error band. It can be observed that at input powers higher than 1.25 kW (named as Mode 2 area in the graph) condenser 1 extracts more heat from the system than condenser 2 regardless of its physical orientation. In vertical and anti-vertical orientations for heat loads lower than 750 W (named as Mode 1 area in the graph), the condenser that is in the top position is removing exclusively the heat from the system. This behavior suggests that the bottom condenser is full of liquid and only the top part is active and the PHP behaves as a thermosyphon with rising vapor and liquid returning through some of the channels. With increasing heat loads this thermosyphon state is overshadowed by the pulsating behavior of the PHP developing. These two states are connected by a transition area between 750 W and 1250 W. Contrary to the vertical orientations, below 750 W the horizontally oriented system shows almost equal power sharing between both condensers, showing that there are capillary induced pulsations at low heat load as well. With increasing heat load the vertical performance approaches the horizontal system’s behavior. However also in horizontal orientation condenser 2 extracts more heat than condenser 1. 4.4. Pressure signal analysis In order to gain a better understanding of the working mechanisms of the PHP, the pressure signal for different heat loads and , orientations was analyzed. Table 2 lists the average pressure p ^ and the frequency f at which the average pressure amplitude p the pressure changes in presence of pressure pulsations. The values reported in brackets represent the ratio between the pressure amplitude signal compared to the vertical case at 750 W. Values are given for 750 W, 1250 W and 2250 W as a representation of the low heat load, the transition and the high heat load region respectively. The frequency was obtained from an FFT analysis of the pressure signal. The system pressure appears to be largely independent of the system’s orientation and changes by 2.2 bars (which corresponds to 25 °C change in the saturation temperature) with an increase of heat load from 750 W to 2250 W. Note that for 750 W in vertical orientation, the amplitude of the pressure signal does not exceed the measuring uncertainty and no underlying frequency was
Table 2 Pressure signal analysis.
Fig. 8. Total heat balance and split of heat load between two condensers in three different orientations.
[bar] p
^ [bar] p
f [Hz]
Power [W] Vertical Horizontal Anti-vertical
750 W 2.28 2.04 2.18
0.017 0.059 (3.5) 0.033 (1.9)
– 0.6 –
Power [W] Vertical Horizontal Anti-vertical
1250 W 2.65 2.67 2.63
0.08 (4.7) 0.116 (6.8) 0.113 (6.6)
0.5 1 1
Power [W] Vertical Horizontal Anti-vertical
2250 W 4.50 4.33 4.32
0.252 (14.8) 0.307 (18.1) 0.275 (16.2)
1.4 1.6 1.4
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
6
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
resistance. This is in agreement with other studies available in the literature [16–18]. 4.5. Baseplate temperature distribution In Fig. 9 the effect of the temperature uniformity on the evaporator baseplate (defined as the difference between the maximum and minimum baseplate temperatures) as a function of the heat load is presented for the three tested orientations. The temperature inhomogeneity is increasing with the heat load independently on the orientation. In the horizontal case, the temperature is more uniform in the investigated heat load range compared to the vertical and anti-vertical orientations: at 2250 W the temperature difference across the base plate in horizontal orientation is 10 K while is 11 K in vertical orientation and 15 K in anti-vertical orientation. 4.6. Air temperature and flow rate effect
found. Similarly, for the anti-vertical orientation, despite an amplitude larger than the measuring uncertainty, the FFT did not yield an underlying frequency. In horizontal orientation, an amplitude of 0.059 bar was measured and the frequency found to be 0.6 Hz. Increasing the heat load to 1250 W results in the horizontal and anti-vertical orientation having almost equal behavior. An amplitude of 0.11 bar was observed and the frequency was found to be 1 Hz for both orientations. At 2250 W the vertical and anti-vertical frequencies are equal at 1.4 Hz. The average pressure is however 0.2 bar higher in the vertical case while the amplitude is 0.02 bar lower. In horizontal orientation the system shows the highest values for frequency (1.6 Hz) and amplitude (0.3 bar). Interestingly, at 2250 kW, in vertical and anti-vertical orientations the main frequency is the same (1.4 Hz) but the amplitude is higher in the anti-vertical case. Since it has been shown in Fig. 7 that in anti-vertical orientation the thermal resistance is higher than in the vertical case, it can be concluded that higher frequencies and lower amplitude pressure signals yield lower thermal
The effect of air volumetric flow rate and temperature have also been investigated. During these tests the PHP has been oriented horizontally and filled at the best filling ratio (60%). As shown in Fig. 10, at 23 °C inlet air temperature, increasing the air velocity results in a decrease of the thermal resistance on the air side, thus improving the thermal performance. The same trend is also expected in vertical orientation since the increase of the air flow rate results in a decrease of the condenser thermal resistance due to the higher air velocity. In Fig. 11 the effect of the inlet air temperature in the range from ambient (23 °C) to 40 °C as a function of the heat load at 750 m3/h air flow in horizontal orientation is reported. The effect of higher air temperatures is beneficial in the entire range of tested heat loads. The increase of air temperature from 23 °C to 40 °C allows a reduction of the stack thermal resistance by 10% at 750 W and 6% at 2 kW. This is due to the increase of the air heat transfer coefficient thanks to the decrease of the air viscosity but also to the increase of the boiling heat transfer coefficient: in fact, an increase of the air temperature results in higher fluid temperature and pressure which is beneficial for the evaporation process. For all the curves the trend is the same with the thermal resistance decreasing with the heat load.
Fig. 10. Thermal resistance as a function of the heat load in horizontal orientation. Effect of volumetric air flow rate.
Fig. 11. Thermal resistance as a function of the heat load in horizontal orientation. Effect of inlet air temperature.
Fig. 9. Maximum baseplate temperature difference as a function of heat load at the best filling ratio at three orientation.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
5. Conclusions The experimental results obtained on a newly designed pulsating heat pipe cooler based on automotive technology having two condenser areas, above and below the evaporator, have been presented. Measurements have been performed with the refrigerant fluid R245fa. In vertical orientation (0°) a thermal resistance of 27 K/kW and baseplate temperature uniformity of 11 K were measured at 2.4 kW heat load, 750 m3/h air flow rate at 23 °C and 60% filling ratio. In horizontal orientation (90°) a maximal thermal resistance of 27 K/kW and baseplate temperature uniformity of 10 K at 2.4 kW heat load, 750 m3/h air flow rate at 23 °C and 60% filling ratio. In anti-vertical orientation (180°) the thermal resistance is the same as in vertical orientation up to 1.5 kW losses and then remain constant at 29 kW up to 2.25 kW. On the contrary the temperature inhomogeneity is the highest compared to the other two cases. A fast Fourier analysis of the pressure signal is performed showing that higher frequencies and lower amplitude pressure signals yield lower thermal resistance. The effects of higher air flow rate and air inlet temperature have shown beneficial effects in the entire range of tested heat loads.
References [1] G. Karimi, JR. Culham, Review and assessment of pulsating heat pipe mechanism for high heat flux electronic cooling, in: Inter-Society Conference on Thermal Phenomena, 2004. [2] Vadim S. Nikolayev, A dynamic film model of the pulsating heat pipe, J. Heat Transf. 133 (8) (2011) 081504.
7
[3] Y. Zhang, A. Faghri, Advances and unsolved issues in pulsating heat pipes, Heat Transf. Eng. 29 (2008) 20–44. [4] S. Khandekar, K.P. Pradipta, F. Lefevre, J. Bonjour, Local hydrodynamics of flow in a pulsating heat pipe: a review, Front. Heat Pipes 1 (1) (2010) 20. [5] Vladimir G. Pastukhov, Yuri F. Maydanik, Combined LHP and PHP based heat transfer system, Int. J. Therm. Sci. 74 (2013) 81–85. [6] Lanchao Lin, Rengasamy Ponnappan, John Leland, Experimental investigation of oscillating heat pipes, J. Thermophys. Heat Transf. 15 (4) (2001) 395–400. [7] Honghai Yang, S. Khandekar, M. Groll, Operational limit of closed loop pulsating heat pipes, Appl. Therm. Eng. 28 (1) (2008) 49–59. [8] Manoj Rao et al., Heat and mass transfer mechanisms of a self-sustained thermally driven oscillating liquid–vapour meniscus, Int. J. Heat Mass Transf. 86 (2015) 519–530. [9] S.M. Pouryoussefi, Yuwen Zhang, Numerical investigation of chaotic flow in a 2D closed-loop pulsating heat pipe, Appl. Therm. Eng. (2016). [10] Balkrishna Mehta, Sameer Khandekar, Taylor bubble-train flows and heat transfer in the context of pulsating heat pipes, Int. J. Heat Mass Transf. 79 (2014) 279–290. [11] Bhawna Verma, Vijay Lakshmi Yadav, Kaushal Kumar Srivastava, Nanofluid – an alternative fluid in pulsating heat pipe/oscillating heat pipe, IOSR J. Electron. Commun. Eng. (2012) 28–37. [12] Vadim S. Nikolayev, Oscillatory instability of the gas–liquid meniscus in a capillary under the imposed temperature difference, Int. J. Heat Mass Transf. 64 (2013) 313–321. [13] B. Agostini, M. Fabbri, Heat exchanger based on pulsating heat pipe principle, U.S. patent, Patent number 97369, 2012. [14] D. Torresin, M. Habert, F. Agostini, B. Agostini, V. Mounier, Experimental characterization of an open loop pulsating heat pipe cooler, in: 12th International Conference on Nanochannels, Microchannels and Minichannels, August 3–7, Chicago, Illinois, USA, 2014. [15] S.J. Kline, F.A. McClintock, Describing uncertainties in single-sample experiments, Mech. Eng. (January) (1953) 3. [16] Jong-Soo Kim et al., The study on pressure oscillation and heat transfer characteristics of oscillating capillary tube heat pipe, KSME Int. J. 17 (10) (2003) 1533–1542. [17] Yong-ho Park et al., Characterizing pressure fluctuation into single-loop oscillating heat pipe, J. Central South Univ. 19 (9) (2012) 2578–2583. [18] Y. Miyazaki, H. Akachi, Heat transfer characteristics of looped capillary heat pipe, in: Proceedings of the 5th International Heat Pipe Symposium, Melbourne, Australia, 1996.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
Contents lists available at ScienceDirect
Applied Thermal Engineering journal homepage: www.elsevier.com/locate/apthermeng
Research Paper
Double condenser pulsating heat pipe cooler Daniele Torresin ⇑, Francesco Agostini, Adrian Mularczyk, Bruno Agostini, Mathieu Habert ABB Switzerland Ltd., Corporate Research, Segelhofstrasse 1K, 5405 Baden-Dättwil, Switzerland
h i g h l i g h t s A newly designed pulsating heat pipe cooler (PHP) based on automotive technology is investigated. The PHP has two condenser areas, one above and one below the evaporator. Measurements were performed with the refrigerant fluid R245fa in three different orientations. The results show that the device is orientation-free with identical performance in all tested orientations.
a r t i c l e
i n f o
Article history: Received 9 September 2016 Revised 2 February 2017 Accepted 15 February 2017 Available online xxxx Keywords: Pulsating heat pipe Double condenser Heat transfer Power electronic
a b s t r a c t A novel pulsating heat pipe cooler (PHP) based on automotive technology is presented. This technology uses numerous aluminum Multiport Extruded (MPE) tubes disposed in parallel to achieve the desired compactness. The sub-channels of the MPEs are connected in a serpentine manner by means of fluid distribution elements integrated in the two condenser manifolds. This configuration enables the oscillation of liquid slugs and elongated bubbles between the evaporator and the condenser areas. This power electronics cooling system with a double condenser area was experimentally characterized and the thermal performances measured with R245fa are presented in this paper. The influence of different parameters such as the heat load, fluid filling ratio and orientation are investigated. The results show that the device is orientation-free: the performance was identical in all orientations, vertical, horizontal and anti-vertical. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction Power electronics and microelectronic cooling requirements are becoming every year more demanding due to the continuous increase of power density. Pulsating heat pipes (PHP) have emerged in the last years as suitable cooling devices for dissipating the high heat loads generated by electronic devices since they allow for the extension of air cooling’s applicability into areas nowadays covered by water-cooling [1]. These devices are entirely passive and they comply with long-term operation without maintenance. Furthermore, the possibility to work independently from the orientation allows higher flexibility on their mechanical integration compared to two-phase thermosyphon coolers. The overall functionality of PHPs is still not completely understood [2]. The complex interplay of local hydrodynamics, thermodynamics, and phase change phenomena make a comprehensive model extremely difficult to develop. Reviews of past research as well as unresolved issues relating to PHPs can be found in works by Zhang and Faghri
⇑ Corresponding author.
[3] and Khandekar et al. [4]. Research efforts continue to move toward a more comprehensive understanding of this technology. Experimental investigations test the performance limits of PHPs with different design parameters [5,3,7]. Among these, Lin et al. [6] tested experimentally a water-cooled open-loop PHP with two symmetrical condensers using fluorocarbons as the working fluids. The experimental configuration consisted of a copper tube with an inner diameter of 1.75 mm which was bent in 40 turns. The PHP dissipated a maximum heat load of 2.04 kW, representing a heat flux of 6.4 W/cm2 and a thermal resistance of 48 K/kW. Experiments reported similar thermal conductivities for horizontal and vertical orientations, showing that the heat pipe was orientation independent. Numerical models have also been developed in an attempt to predict the thermal and hydrodynamic behavior of these devices [8–11]. Nikolayev [12] has produced a recent work that numerically and theoretically predicts oscillation development and start-up criterion for a single branch PHP. All current models include essential simplifications in order to keep the problem tractable. Future work must continue to capture effects that have thus far been neglected such as boiling, contact
E-mail address: [email protected] (D. Torresin). http://dx.doi.org/10.1016/j.applthermaleng.2017.02.066 1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
2
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
Nomenclature Cond Cp,air FR _ air m dp Pmax Q Qair
condenser air specific heat capacity (J/kg K) filling ratio (%) air mass flow rate (kg/s) differential pressure (bar) maximum operative pressure input power (W) power extracted by air stream (W)
angle, etc. Continued experimental work is important to provide reference and verification for the models. 2. Experiments Fig. 1 reports a schematic of the novel pulsating heat pipe heat exchanger [13]. The cooler is made of 11 parallel aluminum multi-port extruded (MPE) tubes having capillary dimensions and connected by fluid distribution elements in the two condenser manifolds. Each MPE tube has 7 sub-channels with rectangular cross section and 1.54 mm hydraulic diameter. The heat load is transferred to the internal channels by an aluminum baseplate (evaporator) in which several grooves are machined and the MPEs are embedded. The heat is then transferred to the two opposite ends where the MPEs are thermally connected using louvered fins (condenser) and finally transferred to the external air by forced air convection. The pulsating heat pipe configuration is achieved by connecting in series the sub-channels of the MPEs at the manifolds [14] enabling the pulsation as shown in Fig. 2. The manifold design used results in a total of 27 turns and an open loop configuration. The newly designed cooler has two condensing areas, one at the top and one at the bottom of the evaporator baseplate, where the heat source device is mounted. In such a way, the fluid has the possibility to pulsate/oscillate in both directions symmetrically. Furthermore, this adds the benefit of having the fluid turns in the manifolds with the fluid being most likely in the liquid state thus reducing the local pressure losses. The PHP has a width of 155 mm and a height of 134 mm for both condensing areas and 125 mm for the evaporator baseplate, respectively. A schematic of the test facility is reported in Fig. 3. Tests have been performed by using one centrifugal fan and a T-conduit to split the air flow between the two condensers. To ensure equal mass flow rates, the pressure drop across each condenser was measured (dp1 and dp2) and kept equal by opening and closing the respective valves (V1 and V2). The overall mass flow of air was
Rth Tb,max Tai Tao u Vair DT
thermal resistance (K/W) maximum base plate temperature (°C) inlet air temperature (°C) outlet air temperature (°C) uncertainty volumetric air flow rate (m3/h) temperature difference (K)
controlled by adjusting the fan speed and monitored using a flow meter (FM). Furthermore the air inlet temperature was regulated by the heater (H2) and set by a PID controller. The heat load was supplied to the system by means of six ceramic heaters (H1) each having a size of 50 50 mm and monitored using a Hameg (HM8115-2) power meter. All signals have been acquired through a National Instruments SCXI box connected to a laptop running Labview. Thermocouples have been calibrated using an Omega DP97 precision thermometer with platinum probes to measure the reference temperature and a Lauda R207 chiller to control the temperature. 3. Data reduction From the experimental campaign, it is possible to determine the thermal resistances of the cooling device. The thermal resistance is calculated as follows:
Rth ¼
T b;max T ai Q
ð1Þ
where Tai is the inlet air temperature, Q is the heat load to be dissipated by the cooler and Tb,max is the maximum base plate temperature measured with the 9 temperature measurements Tb. The thermocouples are placed inside 1.2 mm diameter holes drilled from the backside of the baseplate and with a depth reaching 1 mm to the top surface. The thermocouple tip was fixed to the baseplate holes by means of a highly thermally conductive epoxy (supplier omega, type 101), thermal conductivity 1.04 W/m K. Several thermocouples have been installed to monitor the air temperature conditions: one at each condenser inlet cone and three on each outlet. From the measured air volumetric flow rate, the heat exchanged on the air side can be calculated with a thermal balance (assuming uniform air flow rate):
_ air cp;air ðT ao T ai Þ Q air ¼ m
ð2Þ
The above equation can be evaluated for each condenser: the term Tao represents the mean outlet temperature of the six measured val-
Fig. 1. Schematic of test section and thermocouple positioning.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
3
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
“cross over” connection
“in line” connection
Fig. 2. Manifolds design and schematic of MPEs’ sub-channels connection.
Fig. 3. Test setup schematic.
_ air .is the air mass flow rate. The heat balance, repreues, while m senting the difference between the electrical power applied to the ceramic heaters and the sensible heat received by the air stream can be easily evaluated by comparing Eq. (2) and the power measured with the power meter. As previously mentioned, the operative conditions of a PHP device are determined by a combination of many parameters. One of the most important among them is the filling ratio. Particular care has been taken in the filling ratio procedure: a precision scale (Mettler Toledo Model MS32001L/01) with a repeatability and readability of 0.1 g has been used to estimate the amount of fluid inside the cooler before running each experiment. Fluid temperature and pressure were also monitored with thermocouples inserted in one channel of the MPE tubes and by using an Omega PX409-500A pressure transducer.
The uncertainties of the measured signals are reported in Table 1. The error of the thermal resistances calculated applying the standard Kline and McClintock uncertainty analysis [15] has been found to be below 0.2 K/kW at 2400 W and below 1 K/kW at 500 W. The maximum uncertainty for the air side heat balance
Table 1 Uncertainties of different measured signals. Description
u
Temperature (°C) Pressure (bar) Power (W) Volumetric air flow rate (m3/h)
±0.1 ±0.02 ±16.2 ±24.6
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
4
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
has been calculated to be below 85 W with a volumetric flow rate of 750 m3/h and 2.4 kW losses. 4. Results The next paragraphs summarize the experimental results obtained with the refrigerant fluid R245fa. Focus is given on the analysis of the effect of several parameters on the thermal performance. Three different orientations have been tested as shown in Fig. 4: vertical (0°), horizontal (90°), and anti-vertical (180°). 4.1. Effect of fluid filling ratio One of the most important parameters to investigate of a PHP is the amount of fluid filled compared to its total internal volume. This parameter is defined as filling ratio. An optimal value is characteristic of each device and depends on the boundary working conditions. Fig. 5 shows the thermal resistance as a function of the filling ratio in vertical orientation with a volumetric air flow rate of 750 m3/h and inlet temperature of 23 °C. The different curves refer to heat loads set from 750 W to 2.4 kW. The filling ratio varies from 40 to 75% and the trend of the thermal resistance is the same for all the heat loads: the minimum of thermal resistance has been found to be 27 K/kW at the optimal filling ratio of 60%. At heat loads higher than 1 kW, no data points are reported below 50% FR. The reason for that is because either the maximum criteria for the operative temperature (Tb,max = 120 °C) or pressure (Pmax = 10 bar) for running the test was reached. The optimization procedure for the filling ratio has also been performed horizontally. As shown in Fig. 6 the trend of thermal resistance is the same as for the vertical orientation. The optimum filling ratio in this orientation was also found to be 60% for all the investigated heat loads. At high filling ratio (75%) in horizontal orientation, the cooler was working up to 2 kW losses while in vertical orientation the upper limit at this high filling ratio was found at 1.75 kW. Compared to the vertical case, in horizontal orientation, also for fillings ratio below 50% the maximum criteria for the operative temperature (Tb,max = 120 °C) or pressure (Pmax = 10 bar) for running the test were not reached even at heat loads of 1.25 kW.
Fig. 5. Thermal resistance as a function of filling ratio in vertical orientation and heat load from 0.75 to 2.4 kW.
Fig. 6. Thermal resistance as a function of filling ratio in horizontal orientation and heat load from 0.75 to 2.4 kW.
4.2. Effect of orientation Fig. 7 reports the hotspot thermal resistance as a function of the heat load at the best filling ratio (60%) for three different orientations: vertical (0°), horizontal (90°) and anti-vertical (180°). The thermal resistance is decreasing with the heat load in all the
orientations. In particular, in vertical and anti-vertical orientation the performance are the same up to 1.5 kW. For values higher than 1.5 kW the thermal resistance in anti-vertical orientation becomes
Fig. 4. Schematic of tested configurations.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
5
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
Fig. 7. Thermal resistance as a function of the heat load in three different orientations.
independent from the heat load (29 K/kW) while for the vertical and horizontal cases it continues to decrease with the heat load. An explanation of the different behavior between vertical and anti-vertical orientation at heat loads higher than 1.5 kW might be the different design of top and bottom manifolds (see Fig. 2). For heat loads lower than 1 kW, the thermal resistance in horizontal orientation is lower by 10–15% compared to the vertical and anti-vertical cases. A plausible explanation of this behavior is that in horizontal orientation the non-presence of the gravity effect allows a balanced fluid flow in both condensers. 4.3. Heat balance The graph in Fig. 8 compares the heat removed by the air stream calculated with Eq. (2) and the input heat measured with the power meter for all three orientations. The graph reports also the heat balance of the two different air streams going through the PHP. The dotted line borders represent a ±5% margin of error compared to the measured input power and the dash-dotted line rep-
resents the case of an even split of power into the two airstreams. Note that Cond 1 refers to the condenser that is in the bottom position and Cond 2 to the condenser on the top when referring to the 0° orientation. Furthermore, in this paragraph ‘‘top” refers to the condenser being physically in the top position and ‘‘bottom” to the condenser in the physical bottom position respectively. In the investigated range of heat loads, at the best filling ratio, the heat balance is with a ±5% error band. It can be observed that at input powers higher than 1.25 kW (named as Mode 2 area in the graph) condenser 1 extracts more heat from the system than condenser 2 regardless of its physical orientation. In vertical and anti-vertical orientations for heat loads lower than 750 W (named as Mode 1 area in the graph), the condenser that is in the top position is removing exclusively the heat from the system. This behavior suggests that the bottom condenser is full of liquid and only the top part is active and the PHP behaves as a thermosyphon with rising vapor and liquid returning through some of the channels. With increasing heat loads this thermosyphon state is overshadowed by the pulsating behavior of the PHP developing. These two states are connected by a transition area between 750 W and 1250 W. Contrary to the vertical orientations, below 750 W the horizontally oriented system shows almost equal power sharing between both condensers, showing that there are capillary induced pulsations at low heat load as well. With increasing heat load the vertical performance approaches the horizontal system’s behavior. However also in horizontal orientation condenser 2 extracts more heat than condenser 1. 4.4. Pressure signal analysis In order to gain a better understanding of the working mechanisms of the PHP, the pressure signal for different heat loads and , orientations was analyzed. Table 2 lists the average pressure p ^ and the frequency f at which the average pressure amplitude p the pressure changes in presence of pressure pulsations. The values reported in brackets represent the ratio between the pressure amplitude signal compared to the vertical case at 750 W. Values are given for 750 W, 1250 W and 2250 W as a representation of the low heat load, the transition and the high heat load region respectively. The frequency was obtained from an FFT analysis of the pressure signal. The system pressure appears to be largely independent of the system’s orientation and changes by 2.2 bars (which corresponds to 25 °C change in the saturation temperature) with an increase of heat load from 750 W to 2250 W. Note that for 750 W in vertical orientation, the amplitude of the pressure signal does not exceed the measuring uncertainty and no underlying frequency was
Table 2 Pressure signal analysis.
Fig. 8. Total heat balance and split of heat load between two condensers in three different orientations.
[bar] p
^ [bar] p
f [Hz]
Power [W] Vertical Horizontal Anti-vertical
750 W 2.28 2.04 2.18
0.017 0.059 (3.5) 0.033 (1.9)
– 0.6 –
Power [W] Vertical Horizontal Anti-vertical
1250 W 2.65 2.67 2.63
0.08 (4.7) 0.116 (6.8) 0.113 (6.6)
0.5 1 1
Power [W] Vertical Horizontal Anti-vertical
2250 W 4.50 4.33 4.32
0.252 (14.8) 0.307 (18.1) 0.275 (16.2)
1.4 1.6 1.4
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
6
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
resistance. This is in agreement with other studies available in the literature [16–18]. 4.5. Baseplate temperature distribution In Fig. 9 the effect of the temperature uniformity on the evaporator baseplate (defined as the difference between the maximum and minimum baseplate temperatures) as a function of the heat load is presented for the three tested orientations. The temperature inhomogeneity is increasing with the heat load independently on the orientation. In the horizontal case, the temperature is more uniform in the investigated heat load range compared to the vertical and anti-vertical orientations: at 2250 W the temperature difference across the base plate in horizontal orientation is 10 K while is 11 K in vertical orientation and 15 K in anti-vertical orientation. 4.6. Air temperature and flow rate effect
found. Similarly, for the anti-vertical orientation, despite an amplitude larger than the measuring uncertainty, the FFT did not yield an underlying frequency. In horizontal orientation, an amplitude of 0.059 bar was measured and the frequency found to be 0.6 Hz. Increasing the heat load to 1250 W results in the horizontal and anti-vertical orientation having almost equal behavior. An amplitude of 0.11 bar was observed and the frequency was found to be 1 Hz for both orientations. At 2250 W the vertical and anti-vertical frequencies are equal at 1.4 Hz. The average pressure is however 0.2 bar higher in the vertical case while the amplitude is 0.02 bar lower. In horizontal orientation the system shows the highest values for frequency (1.6 Hz) and amplitude (0.3 bar). Interestingly, at 2250 kW, in vertical and anti-vertical orientations the main frequency is the same (1.4 Hz) but the amplitude is higher in the anti-vertical case. Since it has been shown in Fig. 7 that in anti-vertical orientation the thermal resistance is higher than in the vertical case, it can be concluded that higher frequencies and lower amplitude pressure signals yield lower thermal
The effect of air volumetric flow rate and temperature have also been investigated. During these tests the PHP has been oriented horizontally and filled at the best filling ratio (60%). As shown in Fig. 10, at 23 °C inlet air temperature, increasing the air velocity results in a decrease of the thermal resistance on the air side, thus improving the thermal performance. The same trend is also expected in vertical orientation since the increase of the air flow rate results in a decrease of the condenser thermal resistance due to the higher air velocity. In Fig. 11 the effect of the inlet air temperature in the range from ambient (23 °C) to 40 °C as a function of the heat load at 750 m3/h air flow in horizontal orientation is reported. The effect of higher air temperatures is beneficial in the entire range of tested heat loads. The increase of air temperature from 23 °C to 40 °C allows a reduction of the stack thermal resistance by 10% at 750 W and 6% at 2 kW. This is due to the increase of the air heat transfer coefficient thanks to the decrease of the air viscosity but also to the increase of the boiling heat transfer coefficient: in fact, an increase of the air temperature results in higher fluid temperature and pressure which is beneficial for the evaporation process. For all the curves the trend is the same with the thermal resistance decreasing with the heat load.
Fig. 10. Thermal resistance as a function of the heat load in horizontal orientation. Effect of volumetric air flow rate.
Fig. 11. Thermal resistance as a function of the heat load in horizontal orientation. Effect of inlet air temperature.
Fig. 9. Maximum baseplate temperature difference as a function of heat load at the best filling ratio at three orientation.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066
D. Torresin et al. / Applied Thermal Engineering xxx (2017) xxx–xxx
5. Conclusions The experimental results obtained on a newly designed pulsating heat pipe cooler based on automotive technology having two condenser areas, above and below the evaporator, have been presented. Measurements have been performed with the refrigerant fluid R245fa. In vertical orientation (0°) a thermal resistance of 27 K/kW and baseplate temperature uniformity of 11 K were measured at 2.4 kW heat load, 750 m3/h air flow rate at 23 °C and 60% filling ratio. In horizontal orientation (90°) a maximal thermal resistance of 27 K/kW and baseplate temperature uniformity of 10 K at 2.4 kW heat load, 750 m3/h air flow rate at 23 °C and 60% filling ratio. In anti-vertical orientation (180°) the thermal resistance is the same as in vertical orientation up to 1.5 kW losses and then remain constant at 29 kW up to 2.25 kW. On the contrary the temperature inhomogeneity is the highest compared to the other two cases. A fast Fourier analysis of the pressure signal is performed showing that higher frequencies and lower amplitude pressure signals yield lower thermal resistance. The effects of higher air flow rate and air inlet temperature have shown beneficial effects in the entire range of tested heat loads.
References [1] G. Karimi, JR. Culham, Review and assessment of pulsating heat pipe mechanism for high heat flux electronic cooling, in: Inter-Society Conference on Thermal Phenomena, 2004. [2] Vadim S. Nikolayev, A dynamic film model of the pulsating heat pipe, J. Heat Transf. 133 (8) (2011) 081504.
7
[3] Y. Zhang, A. Faghri, Advances and unsolved issues in pulsating heat pipes, Heat Transf. Eng. 29 (2008) 20–44. [4] S. Khandekar, K.P. Pradipta, F. Lefevre, J. Bonjour, Local hydrodynamics of flow in a pulsating heat pipe: a review, Front. Heat Pipes 1 (1) (2010) 20. [5] Vladimir G. Pastukhov, Yuri F. Maydanik, Combined LHP and PHP based heat transfer system, Int. J. Therm. Sci. 74 (2013) 81–85. [6] Lanchao Lin, Rengasamy Ponnappan, John Leland, Experimental investigation of oscillating heat pipes, J. Thermophys. Heat Transf. 15 (4) (2001) 395–400. [7] Honghai Yang, S. Khandekar, M. Groll, Operational limit of closed loop pulsating heat pipes, Appl. Therm. Eng. 28 (1) (2008) 49–59. [8] Manoj Rao et al., Heat and mass transfer mechanisms of a self-sustained thermally driven oscillating liquid–vapour meniscus, Int. J. Heat Mass Transf. 86 (2015) 519–530. [9] S.M. Pouryoussefi, Yuwen Zhang, Numerical investigation of chaotic flow in a 2D closed-loop pulsating heat pipe, Appl. Therm. Eng. (2016). [10] Balkrishna Mehta, Sameer Khandekar, Taylor bubble-train flows and heat transfer in the context of pulsating heat pipes, Int. J. Heat Mass Transf. 79 (2014) 279–290. [11] Bhawna Verma, Vijay Lakshmi Yadav, Kaushal Kumar Srivastava, Nanofluid – an alternative fluid in pulsating heat pipe/oscillating heat pipe, IOSR J. Electron. Commun. Eng. (2012) 28–37. [12] Vadim S. Nikolayev, Oscillatory instability of the gas–liquid meniscus in a capillary under the imposed temperature difference, Int. J. Heat Mass Transf. 64 (2013) 313–321. [13] B. Agostini, M. Fabbri, Heat exchanger based on pulsating heat pipe principle, U.S. patent, Patent number 97369, 2012. [14] D. Torresin, M. Habert, F. Agostini, B. Agostini, V. Mounier, Experimental characterization of an open loop pulsating heat pipe cooler, in: 12th International Conference on Nanochannels, Microchannels and Minichannels, August 3–7, Chicago, Illinois, USA, 2014. [15] S.J. Kline, F.A. McClintock, Describing uncertainties in single-sample experiments, Mech. Eng. (January) (1953) 3. [16] Jong-Soo Kim et al., The study on pressure oscillation and heat transfer characteristics of oscillating capillary tube heat pipe, KSME Int. J. 17 (10) (2003) 1533–1542. [17] Yong-ho Park et al., Characterizing pressure fluctuation into single-loop oscillating heat pipe, J. Central South Univ. 19 (9) (2012) 2578–2583. [18] Y. Miyazaki, H. Akachi, Heat transfer characteristics of looped capillary heat pipe, in: Proceedings of the 5th International Heat Pipe Symposium, Melbourne, Australia, 1996.
Please cite this article in press as: D. Torresin et al., Double condenser pulsating heat pipe cooler, Appl. Therm. Eng. (2017), http://dx.doi.org/10.1016/j. applthermaleng.2017.02.066