Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Article history: In this study, the ability of a boiling process to thermally condition (homogenisation and cooling) batter-
Received 5 April 2013 ies is investigated. Thereto, a series of experiments are performed and discussed. Subjects that are treated
Accepted 5 December 2013 are the dielectric property of the proposed cooling fluid, its cooling capability compared to that of air, the
ability of the boiling fluid to thermally homogenise a battery and the influence of pressure variation on
the boiling process. It turns out that the proposed cooling fluid conducts no electricity, has good cooling
Keywords: characteristics compared to those of air and, when boiling, is able to thermally homogenise the battery.
Boiling heat-transfer
Furthermore, pressure variation seems to offer a good method to regulate the boiling process.
Battery
Pool boiling
Ó 2013 Elsevier Ltd. All rights reserved.
Thermal homogenisation
Experiments
Numerical simulation
0196-8904/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.enconman.2013.12.006
10 R.W. van Gils et al. / Energy Conversion and Management 79 (2014) 9–17
(1-methoxyheptafluoropropane)), since its boiling point at atmo- determination of homogeneity and/or heterogeneity. Orientation
spheric pressure equals 34 °C, which is in the battery safety limits. and dimensions of Setup 1 and the sensor configurations are sche-
Two setups are utilised for the experiments. The first setup (Set- matically depicted in Fig. 2.
up 1) is used to:
3.2. Setup 2
Test the dielectric properties of the fluid.
Perform discharge experiments at atmospheric pressure. These The second setup is used to investigate the influence of the
experiments are done with the battery submerged in the liquid pressure on the boiling process. Thereto, a much smaller container
and in open-air. Results of both experiments give insight into is used, so that the liquid starts boiling due to the heat produced by
the cooling capacity of the liquid compared to that of air. the battery only. Hence, pre-heating as in Setup 1 is unnecessary.
Perform pulse-charge–discharge (PCD) experiments at atmo- The small boiling chamber is connected to an injection syringe
spheric pressure, with the battery submerged in the liquid. and a water manometer for pressure measurement. The syringe
With these experiments the heat generation by the battery is enables variation of the boiling chamber volume, which thus indi-
maximal. These experiments are used to investigate the ability rectly regulates the pressure. Setup 1 and the sensor configurations
of the boiling process to thermally homogenise batteries. are schematically depicted in Fig. 2. Configuration and dimensions
of Setup 2 are schematically depicted in Fig. 3.
The second setup (Setup 2) is used to investigate the influence
of pressure variation on the boiling process. The primary goal is
4. Experimental analysis
determining to what extent the pressure is suitable for regulation
of the system. Although the influence of pressure on a boiling pro-
In this section the experiments performed are discussed. Sub-
cess has been discussed in e.g. [10,22], this experiment is per-
jects that are treated are the dielectric property of the fluid, its
formed to specifically investigate whether this principle is
cooling capability compared to that of air, its ability for thermal
applicable for controlling the thermal conditioning. It is important
homogenisation and the influence of pressure variation on the
that the boiling process can be actively regulated by variation of
boiling process.
the pressure in the boiling chamber, since the boiling process must
be maintained in the nucleate boiling regime (see Fig. 1). Control of
the boiling process must be done rapidly as it has to cope with the 4.1. Electric interaction between battery and cooling liquid
highly dynamic heat production by the battery pack. The (future)
challenge is to regulate the boiling process actively and rapidly First the dielectric property of the liquid is investigated in order
via the pressure under dynamical conditions. This is essential for to rule out possible influences due to electric conduction. The evo-
the application of pool-boiling considered here. The experiments lution of the voltage and current of the battery, fully submerged in
discussed here lay the groundwork for more detailed (future) the liquid, are given in Fig. 4. As can be seen, first the battery is
investigations. charged using a constant-current constant-voltage (CCCV) scheme
In both setups, the current and voltage are supplied, regulated (Fig. 4(a)) and then it is maintained at rest (open circuit voltage
and measured by a battery-testing system (MACCOR 2300, USA). Mea- (OCV)). The close-up of the rest period in Fig. 4(b) exposes hardly
surements are done with a sampling rate of 1 Hz or when the volt- any decay of the voltage over time, indicating that the battery does
age has changed more than 1 mV. The accuracy of the Maccor not appreciably discharge during open circuit conditions. Hence,
system is <1 mV in voltage and <1 mA in current. The battery can the electric conduction by the fluid will not play a role in the fol-
be (dis)charged at a maximum of 5 A. This system also supplies lowing experiments.
the current and measures the voltage drop over the thermocouples
in order to measure the temperature on various locations. Thermo- 4.2. Battery cooling: air versus single-phase liquid
couples used are Pt100/1516, Grade A thermocouples (METATEMP)
([23]) for both setups their accuracy/stability is increased by taking The 1 Ah battery is discharged at 5 A. Two series of experiments
a moving average of 50 measurement points. are done: (i) discharging in open air; (ii) discharging while the bat-
tery is submerged in Novec7000, in order to compare the cooling
3.1. Setup 1 capability of the fluid with that of air. These experiments are per-
formed using Setup 1 with Sensor configuration 1. The voltage
In the first setup, the battery is fully submerged in a pool of li- curves of the experiment are depicted in Fig. 5(a). Curve (a) shows
quid inside a double-walled container. The enclosure between the the voltage upon discharge while submerged in Novec7000, while
double walls can be filled with air, which thermally insulates the Curve (b) shows that in open-air. The difference in discharge pro-
battery and liquid from the ambient, since the air volume is sealed files is due to the difference in temperature for both experiments,
in this case and motion of air is restricted. They can be also filled the corresponding temperature evolutions are depicted in Fig. 5(b).
with water at a set temperature (the water temperature is regu- Here Curve (a) corresponds to the experiment where the battery is
lated via a ‘Thermo Scientific HAAKE Bath’), in order to pre-heat submerged in Novec7000, while Curve (b) corresponds to that in
the liquid prior to an experiment. The temperature control is pro- open-air. Curve (c) gives the constant temperature of the ambient
vided with an accuracy of ±0.1 °C), temperature loss along the tube during the experiments. Due to the elevated temperatures during
is not accounted for and is considered small. Temperature mea- the discharge in open-air, the battery becomes more efficient.
surements are performed at two and three locations for Sensor However, undesired side-reactions that deteriorate the battery will
configuration 1 and 2, respectively. For Sensor configuration 1, increase as well due to these elevated temperatures.
the temperature of the battery wall is measured externally be- These results reveal that when discharged in air, the battery
tween the top and bottom of the battery and the ambient temper- temperature (Curve (b)) rises to temperature levels above the boil-
ature is measured. For Sensor configuration 2, the external ing point of Novec7000 (34 °C). During the experiments in No-
temperature of the positive pole and the negative pole and the vec7000, the battery is fully submerged in a pool of liquid,
temperature of the liquid are measured. These locations are chosen however. Due to the vertical position of the (main part of the)
since with two battery measurements available, all first order fluid-heater interface, the temperature differences along this
heterogeneity can be observed and thus provide a viable first interface result in fluid flows as a consequence of natural convec-
12 R.W. van Gils et al. / Energy Conversion and Management 79 (2014) 9–17
Fig. 4. Evolution of the current and voltage upon battery charging (panel a) and while at rest (panel b). During both experiments the battery is submerged in Novec7000.
Fig. 5. Voltage curves of the discharge experiment (panel a). Temperature evolution of the battery and the ambient for a 5 A discharge in air and Novec7000 (panel b).
tion. As a result, the fluid is drained away from the battery before it
can start boiling. This has been observed clearly during these
experiments. Consequently, the temperature of the battery (Curve
(a)) does not rise more than 5 °C. Hence, the favourable cooling
characteristics compared to that of air are solely the result of nat-
ural convection. In the following section, boiling is induced by
pulse-charge–discharge (PCD) of the battery, as it produces more
heat and over a larger period of time.
Experiments with the walls fixed at 20 °C, 25 °C and 30 °C dem- perature difference T bat T bulk can be considered a measure for the
onstrate that as long as the liquid does not start boiling (at 34 °C), intensity of the boiling process, where lower temperature differ-
the evolution of the temperature in all three measurement points ences correspond to a more violent mode of boiling. This means that
of Sensor configuration 2 (Fig. 2) is, relative to the wall tempera- the liquid starts to boil more intensively during the experiments
ture, approximately the same. This evolution is represented with with the walls fixed at 33 °C compared to 32 °C, which is also ob-
the walls fixed at 20 °C in Fig. 7(a). An offset between the temper- served during these experiments.
ature at the bottom of the battery (negative pole) (Curve (b)) and The heat-transfer coefficient is determined using a thermal bat-
that of the top of the battery (positive pole) (Curve (a)) of approx- tery model. Such models can form the basis for designing control
imately 0:7 C can be observed. When the battery orientation is laws as discussed in the introduction of this study. The tempera-
standard, the positive pole is its upper side. In this case, the top ture evolution of the battery for the PCD experiments is deter-
is warmer than the bottom of the battery. With the battery turned mined via simulations and compared to the experimental results
upside down the positive pole (bottom) is approximately 0:2 C of the previous section. The mass, dimensions and specific heat
warmer than the negative pole (top). Since for this orientation of the battery are determined and are given by
the bottom of the battery is warmer than its top, an unstable strat- m ¼ 32:5 g; ðl; dÞ ¼ ð49; 18Þ mm and cp ¼ 0:87 J=ðg KÞ, respectively.
ification of the liquid surrounding the battery occurs, which results The temperature evolution of the battery can be described by
in vertical convection. As a result, the temperature difference be-
dT bat 1
tween the battery poles diminishes. The temperature difference ¼ ðQ hAðT bat T bulk ÞÞ: ð2Þ
observed in Fig. 7(a) thus is not solely the result of natural convec- dt cp m in
tion (for both experiments vertical streamlines can be observed) If h and Q in are constant, the exact solution of this differential
but is probably the combined result of convection and a nonuni- equation is given by
form temperature distribution inside the battery.
For the experiments with the container walls fixed at a temper- T bat ðtÞ ¼ ðT bat ð0Þ T 1 Þet=s þ T 1 ; ð3Þ
ature larger than about 31 °C the liquid starts to boil at the battery
where s ¼ mcp =ðhAÞ is the characteristic time scale of the system
poles. Fig. 7(b) represents the temperature evolution in the sensor
evolution and T 1 ¼ Q in =ðhAÞ is the equilibrium temperature. Note
points for the walls fixed at 33 °C and reveals that the top and bot-
that this simplified model assumes a uniform internal battery tem-
tom temperature of the battery (Curve (a) and (b), respectively) be-
perature. As can be seen, the characteristic time of the system evo-
come nearly indistinguishable, demonstrating that the boiling
lution decreases with increased heat-transfer coefficients (h). This
process indeed thermally homogenises the battery.
means that boiling, besides high heat fluxes and thermal homogeni-
The eventual temperature difference of the battery relative to
sation, offers a third advantage over conventional conditioning
the bulk temperature, i.e. the liquid temperature (Curve (c) in
method: the ability of rapid regulation of the temperature.
Fig. 7, equals approximately 1.1 °C for the walls fixed at 32 °C
For the experiments Q in is determined from the Maccor system,
and approximately 0.8 °C for the walls fixed at 33 °C. In the non-
i.e. the energy difference between that charged in the battery and
boiling regime this amounts to 1.3 °C for the bottom and 2.0 °C
that discharged from the battery divided by the time of one cycle
for the top of the battery, for all experiments done. Since, the
and
PCD procedure is equal for all experiments, it can be assumed that
the heat flux from battery to liquid Q is constant for all experi- Q in
ments. Assuming convective heat-transfer, i.e. h¼ ; ð4Þ
AðT eq eq
bat T bulk Þ
Q ¼ hðT bat T bulk Þ; ð1Þ where superscript ‘eq’ denotes that the battery and liquid are in
thermal equilibrium. The heat-transfer coefficient (h) can then be
this means that a smaller temperature difference implies a relative determined from the experiments discussed above. The obtained
larger heat-transfer coefficient h, which signifies a higher degree of heat-transfer coefficients are given in Fig. 8 as function of wall
boiling. The heat-transfer coefficient is given by the slope of the temperature for the experiments done. It can be seen that the
boiling curve, see Fig. 1 (also see, for example [13]). That is, the tem- heat-transfer coefficients in the non-boiling regime are approxi-
Fig. 7. Temperature evolution of the battery and the liquid (close to the container wall) for a 5 A PCD in Novec7000. The temperature of the walls of the container are fixed at
20 °C (panel a) and 33 °C (panel b).
R.W. van Gils et al. / Energy Conversion and Management 79 (2014) 9–17 15
mately the same. When the liquid starts boiling, the heat-transfer tery running a PCD procedure at 5 A. This experiment is carried out
coefficient increases, which indicates a more aggressive mode of at atmospheric pressure, for an underpressure of 0:05 atm and for
boiling. A doubling of the heat-transfer coefficient can be observed an overpressure of 0:05 atm in the boiling chamber. The tempera-
when going from the non-boiling to the boiling regime, meaning the ture evolutions of these experiments show that the battery tem-
reaction-time (s) increases with a factor two as well. perature in thermal equilibrium relates to the pressure in the
The evolution of the measurements can be compared with those boiling chamber as shown in Fig. 10. The temperature increases
obtained via simulation of the model. This is shown in Fig. 9(a) and with pressure, as a result of the dependency of the boiling point
(b) for the experiment with the walls fixed at 20 °C and at 33 °C, of the liquid on the pressure. This demonstrates that the tempera-
respectively. The evolution of the experiments with the walls fixed ture can indeed be regulated via the pressure.
at 25 °C, 30 °C and 32 °C are given in the Appendix. The model and In order to attain further prove of the principle of manipulating
experiment transient nearly overlap, meaning the obtained heat- the boiling process via adaptation of the pressure, a video is made
transfer coefficients can be assumed to be accurate. Note that in during an experiment. The primary goal is determining to what ex-
Fig. 9 the average battery temperature is simulated, while the tent the pressure is suitable for regulation of the system. Although
experimental signals correspond to the top and bottom of the the influence of pressure on the boiling process has been discussed
battery. in e.g. [10,22], this experiment is performed to investigate whether
this principle is also applicable for thermal conditioning. The (fu-
ture) challenge is to regulate the boiling process actively and rap-
4.4. Regulation of the conditioning process
idly via the pressure under dynamical conditions. Here the
pressure is momentarily decreased and increased, while the liquid
Finally, it is investigated whether boiling can occur when the li-
is boiling and the battery is running a PCD procedure at 4.5 A. This
quid is not heated externally and how the pressure influences the
video can be found at http://youtu.be/8oBIF6lpL_s. It can be seen
boiling process. A different experimental setup is used, where the
quite well that increasing the pressure, instantaneously suppresses
volume of liquid surrounding the battery is much smaller. There-
the boiling process, while decreasing the pressure makes the liquid
fore, the liquid starts boiling due to the heat produced by the bat-
36
35.5
35
ylabel
34.5
34
33.5
33
0.9 0.95 1 1.05 1.1
xlabel
Fig. 8. Heat-transfer coefficient at thermal equilibrium as function of wall Fig. 10. Temperature at thermal equilibrium ( liquid boiling point) as function of
temperature. pressure.
Fig. 9. Temperature evolution of the battery and the liquid (close to the container wall) for a 5 A PCD in Novec7000. The temperature of the walls of the container are fixed at
20 °C (panel a) and 33 °C (panel b).
16 R.W. van Gils et al. / Energy Conversion and Management 79 (2014) 9–17
Fig. 11. Visualisation of the boiling experiments at atmospheric pressure (panel a), at sub-atmospheric pressure (panel b) and at super-atmospheric pressure (panel c).
Fig. 12. Temperature evolution of the battery cell and the liquid (close to the container wall) for a 4.9A PCD in Novec7000. The temperature of the walls of the container are
fixed at 25 °C (panel a), 30 °C (panel b) and 32 °C (panel c). The experimental signals are smoothed out by taking a moving average over 50 data points.
If the battery is able to get the liquid boiling, the boiling process [5] Pesaran A, Vlahinos A, Burch S. Thermal performance of EV and HEV battery
modules and packs. In: Proceedings of the 14th international electric vehicle
is able to thermally homogenise the battery. Temperature dif-
symposium, Orlando, Florida; December 15–17, 1997.
ferences between positive and negative pole of the battery of [6] Heckenberg T. Li-ion battery cooling: more than just another cooling task.
approximately 0.7 °C are observed in the non-boiling regime, Internal report, BEHR; 2009.
while these temperature differences are eliminated when the [7] Chu R, Simons R, Ellsworth M, Schmidt R, Cozzolino V. Review of cooling
technologies for computer products. IEEE Trans Device Mater Reliab
liquid around the battery starts boiling. 2004;4(4):568–85.
The boiling process can be influenced by the pressure in the [8] Simons R. The evolution of ibm high performance cooling technology. IEEE
boiling chamber. Decreasing the pressure, instantaneously Trans Comp Packag Manuf Technol Part A 1995;18(4):805–11.
[9] Mudawar I. Assessment of high-heat-flux thermal management schemes. IEEE
increases the intensity of the boiling process. This means that Trans Comp Packag Technol 2001;24(2):122–41.
in principle the boiling process can be actively and rapidly con- [10] Kwatny H, Fink L. Acoustics, stability, and compensation in boiling water
trolled by regulation of the pressure in the boiling chamber. reactor pressure control systems. IEEE Trans Automat Control
1975;20(6):727–39.
[11] van Gils RW, Speetjens MFM, Nijmeijer H. Boiling heat transfer in battery
Boiling heat-transfer has several significant advantages in that electric vehicles. In: EEVC 2011, Brussels, Belgium; October 26–28, 2011.
it allows for higher heat fluxes, better thermal homogenisation of [12] Peter Cohan 2013. <http://www.forbes.com/sites/petercohan/2013/02/06/
thermal-runaway-in-787-dreamliner-batteries-must-be-stopped/> [visited
the battery and a more rapid response compared to conventional Febraury 2013].
methods. [13] Dhir V. Boiling heat transfer. Annu Rev Fluid Mech 1998;30(1):365–401.
[14] Pioro IL, Rohsenow W, Doerffer SS. Nucleate pool-boiling heat transfer. i:
Review of parametric effects of boiling surface. Int J Heat Mass Transfer
Acknowledgment
2004;47(23):5033–44.
[15] Jungho Kim. Review of nucleate pool boiling bubble heat transfer mechanisms.
This research is funded by the High Tech Automotive Systems, Int J Multiphase Flow 2009;35(12):1067–76.
Range Extender Innovations Project (HTAS-REI). [16] Blum J, Lüttich T, Marquardt W. Temperature wave propagation as a route
from nucleate to film boiling? In: Celata G, DiMarco P, Shah R, editors.
Proceedings of the second international symposium on two-phase flow
Appendix A modelling and experimentation, vol. 1, Rome. Pisa: Edizioni ETS; 1999.
[17] Speetjens M, Reusken A, Marquardt W. Steady-state solutions in a nonlinear
pool boiling model. Commun Nonlinear Sci Numer Simul 2008;13(8):1475–94.
In Fig. 12 the temperature evolution of the battery from the [18] van Gils RW, Speetjens MFM, Nijmeijer H. Feedback stabilisation of a one-
model-simulation and experiments for container walls fixed at dimensional nonlinear pool-boiling system. Int J Heat Mass Transfer
2010;53(11-12):2393–403.
25 °C, 30 °C and 32 °C are given.
[19] van Gils RW, Speetjens MFM, Zwart HJ, Nijmeijer H. Feedback stabilisation of a
two-dimensional pool-boiling system by modal control. Int J Therm Sci
References 2012;61:38–49.
[20] Tang Z, Zhu Q, Lu J, Wu M. Study on various types of cooling techniques
[1] Pesaran A, Vlahinos A, Stuart T. Cooling and preheating of batteries in hybrid applied to power battery thermal management systems. In: 2nd International
electric vehicles. In: 6th ASMR-JSME thermal engineering conference, Hawaii, conference on energy, environment and sustainable development, EESD 2012,
USA; March 16–20, 2003. Jilin; 12–14 October 2012. Code94737.
[2] Pesaran A. Current and future needs in electric drive vehicle batteries. In: [21] Rao Z, Wang S, Wu M, Lin Z, Li F. Experimental investigation on thermal
Proceedings of the 14th international heat transfer conference, IHTC14, management of electric vehicle battery with heat pipe. Energy Convers
Washington, DC, USA; August 8–13, 2010. Manage 2013;65:92–7.
[3] Ramadass P, Haran B, White R, Popov B. Capacity fade of sony 18650 cells [22] Dhir V. Nucleate and transition boiling heat transfer under pool and external
cycled at elevated temperatures: Part ii. Capacity fade analysis. J Power flow conditions. Int J Heat Fluid Flow 1991;12(4):290–314.
Sources 2002;112(2):614–20. [23] Metatemp 2012. <http://www.metatemp.nl/page1.htm> [visited July 2012].
[4] Besselink IJM, Hereijgers JAJ, van Oorschot PF, Nijmeijer H. Evaluation of
20000 km driven with a battery electric vehicle. In: Proceedings of the EEVC
2011, Brussels, Belgium; October 2011.