THERMAL DIFFUSION OF HEAT PULSEIN SUBCOOLED LIQUID NITROGEN

H. M. Chang1, J. J. Byun1, J. H. Choi1, C. J. Ha1, M. J. Kirn1, H. M. Kirn2, andT. K. Ko3

!Hong Ik UniversitySeoul, 121-791, Korea2LS Industrial Systems Inc.Cheongju, 361-720, Korea

3Yonsei UniversitySeoul, 120-749, Korea

ABSTRACT

Transient heat transfer caused by a heat pulse in subcooled liquid nitrogen is investigatedexperimentally. This study is part of our ongoing efforts to develop a stable cryogenic coolingsystem for superconducting fault current limiters (SFCL) in Korea. A thin heater attached byepoxy on one surface of a GFRP plate is immersed in a liquid-nitrogen bath at temperaturesbetween 77 K and 65 K. A strong heat flux up to 150 W/cm2 is generated for 100 ms, and thetemperature of the heater surface is measured as a function of time. The behavior of bubbleson the heating surface can be indirectly explained by comparing the measured temperaturehistory for vertical and two different horizontal (up and down) orientations. It is concludedthat subcooling liquid nitrogen below 70 K is a very effective method to suppress bubbles andresult in better thermal protection and faster recovery from a heat pulse.

KEYWORDS: Heat Transfer, Liquid Nitrogen, Heat Pulse, Fault Current LimiterPACS: 44.90,+c

INTRODUCTION

One of the most promising HTS (high temperature superconductor) power devices undercurrent development is the superconducting fault current limiter (SFCL). A specific cryogenic

CP823, Advances in Ciyogenie Engineering: Transactions of theCryogenic Engineering Conference - CEC, Vol. 51, edited by J. G. Weisend II

© 2006 American Institute of Physics 0-7354-0317-1/06/$23.00

488

requirement of an SFCL is to safely protect superconducting elements from a heat pulsecaused by a fault current and to promptly recover them to normal operation [1-5]. Thisrequirement is especially important in resistive-types of SFCL where the energy of anelectrical fault is immediately dissipated into heat. This paper is part of our ongoing researchtowards an efficient and stable cryogenic cooling system for SFCL under the 21C FrontierR&D Program funded by the Korean Ministry of Science and Technology.

Cooling superconducting elements in subcooled liquid-nitrogen at around 65 K andatmospheric pressure is an excellent option for SFCL in many respects. The critical currentdensity of HTS is considerably greater at 65 K than at 77 K, thus the size of superconductingelements can be reduced [6]. A more important benefit of subcooling the liquid is thesuppression of bubbles that may be generated from an internal or external heat load. Sincebubbles play a critical role in the deterioration of electrical insulation in liquid nitrogen,subcooling is essential for high-voltage devices such as SFLC or transformers [7].

A number of efforts have been directed at subcooled liquid-nitrogen systems for HTSpower applications, and it is now a common practice in SFCL to employ a cryocooler forcontinuous operation and to take advantage of the active natural convection of liquid nitrogenfor temperature uniformity [8]. Several cryogenic systems had been successfully developed forsmall or full-scale SFCL prototypes in subcooled liquid-nitrogen [2-4]. On the other hand, thetransient thermal phenomena caused by a heat pulse have been rarely reported. Even though anumber of pool-boiling correlations in subcooled liquid nitrogen are available includingnucleation, peak nucleate boiling, and film boiling [9], we think the steady-state information isnot directly applicable to this specific situation, because the transient inertial factor of the fluidis dominant in the short fault period (typically, 100 ms). This study investigates the thermaldiffusion of a heat pulse in subcooled liquid nitrogen. As a first step, we intend in this paper todirectly measure the temperature history in similar experimental conditions for a betterunderstanding of the thermal phenomena. These results should be useful in predicting the peaktemperature and the subsequent recovery of the superconducting elements in the SFCL.

EXPERIMENTAL APPARATUS

FIGURE 1 is a schematic overview of our heat pulse diffusion experiment. A single-stageGM cryocooler (Cryomech AL60) is mounted on the top plate of a cryostat, and a 35 L vesselof liquid nitrogen is placed at the upper position of a cryostat for close access to the coldheadof the cryocooler. A circular copper plate (5 mm thick and 25 cm in diameter) is horizontallyattached to the coldhead as an extended cooling surface. The cooling capacity of thecryocooler is regulated by a Thermofoil™ heater (Minco Model HK5562) attached on thecoldhead and a DC power supply. FIGURE 2 is a cool-down history to a subcooled liquidstate at 65 K and 101 kPa. During the initial cool-down (about 13 hours), liquid nitrogen is ina saturated state so that the cryostat pressure decreases in accordance with the vapor pressureof nitrogen. When the liquid reaches the intended temperature, the coldhead heater is turnedon at the preset value and helium gas is supplied to pressurize the cryostat to 101 kPa. Theliquid temperature under the cooling plate is spatially very uniform by natural convection, andthe test modules are suspended at a distance of approximately 5 cm below the plate.

Two different types of test modules are prepared in this experiment, as shown in FIGURE3. In the first, a square shape of Thermofoil™ heater (Minco HK5583) is attached by epoxy onthe surface of a rectangular GFRP plate (40 mm x 37 mm and 3 mm thick). A temperaturesensor (E-type thermocouple as described below) is located in a direct contact with the heater

489

FIGURE 1. Schematic overview of experimental apparatus.

FIGURE 2. Typical cool-down history to 65 K by a GM cryocooler.

at the center of the square and the thermocouple wires are connected through a tiny hole (2.0mm diameter) from the opposite direction of heating surface. Three different orientations(vertical, heating surface up and down) are tested and compared to examine the effect ofbubbles. In the second module, the same heater is sandwiched by two identical GFRP squareplates. Because of its symmetry, the temperature sensor is placed only on one side of theheater, and two orientations (vertical and horizontal) are tested.

In order to simulate the heat pulse in an actual SFCL, a preset voltage is applied to theheater for 100 ms by an AC power supply (Pacific Power Smart Source™ 360-AMX).FIGURE 4 is an example of the measured voltage applied to the heater when Vrms = 90 V and/= 60 Hz. The heat diffusion in the test module is considered basically one-dimensional, sincethe heat flux is spatially uniform over the heating surface. The heating power is considerednearly constant over the heating period (that is, step-up and step-down profile), since the acperiod (1/60 s) is very short in comparison with the thermal relaxation time constant in the

490

CJHIP S«wie

(a) Dimensions of test module (unit: mm) (b) Orientations of test modules

FIGURE 3. Schematic of test modules: GFRP plate attached by Thermofoil heater and temperature sensor.

off

> 100

1 50o> 0

21 -50

^ -100

-ISO

;

: V v

A A :

jv V V :

-50 0 50 100

Time (ms)

FIGURE 4. An example of measured voltage across the heater (Vrms = 90 V and/= 60 Hz)

heater and the electrical resistance of the heater is nearly constant over temperatures up to 600K. The range of the heating power per unit area is 10-150 W/cm2 in this experiment.

Temperatures are measured with E-type (chromel-constantan) thermocouples (Omega5TC-TT-36AWG) at the coldhead of the GM cryocooler, the test modules, and severallocations in the liquid pool. The accuracy of the thermocouples is ± 30 mK for temperatures inthe range of 63-77 K, which is confirmed by a simultaneous measurement with a silicon diode(Lakeshore DT-470-SD) at the coldhead. We prepared a few test modules where both athermocouple and a silicon diode are installed side-by-side. Following a heat pulse, thesensitivity of the two sensors was practically the same, but the measured response time ofthermocouple was clearly shorter. Consequently, we decided to use thermocouples for theremaining transient measurements. The temperature of the test modules is sampled andrecorded every 1 ms with a high-speed data acquisition board (National Instrument DAQPad-6015). In order to verify repeatability, we executed the same experiment ten times for severalselected conditions and compared the temperature history. In every case, the temperaturediscrepancy was very small (the order of 0.1 K or less) during the sharp heating-up period. Onthe other hand, there was a few percent of discrepancy in the early stage of the recovery period,which is mainly due to the somewhat erratic behavior of bubbles as discussed later. Additionalthermocouples are attached at several axial locations of a vertical GFRP rod for the purpose ofchecking the liquid level and confirming the uniformity of the liquid temperature.

491

- 2 - 1 0 1 2 3Distance from center (mm)

(a) Calculated temperature distributions

0.5 1.0 1.5

Time (s)2.0

(b) Calculated and measured temperature history

FIGURE 5. Calculated and measure temperatures of GFRP sandwich for 100 W/cm2 at 65 K.

RESULTS AND DISCUSSION

GFRP Sandwich

We present first the computational results for the GFRP sandwich in subcooled liquid at65 K to compare with the experimental measurement. The temperature is calculated bynumerically solving the one-dimensional heat diffusion equation

0)dt dx dx

where p, C, and k are the density, specific heat, and thermal conductivity of GFRP asfunctions of temperature. FIGURE 5(a) shows the evolution of the calculated temperaturedistribution when 100 W/cm2 is supplied to the center surface (x = 0) for Q<t <0.1 s. TheCrank-Nicolson method [10] is incorporated in the calculation and an adiabatic condition isassumed at the surface at x = ± 3 mm, because the thermal penetration depth [11] is less than 3mm and the effect of the convection cooling on the temperature distribution is negligibly small.A sharp peak formed at* = 0 during initial 0.1 second is gradually spread out symmetrically.

FIGURE 5(b) is a reproduction of FIGURE 5(a) to show the temperature history at x = 0,0.1, 0.2, and 0.3 mm in comparison with the experimental measurement. The experimentaltemperature history is also the result of a heat pulse of 100 W/cm2 for 100 ms, and the sameresults are obtained regardless of its orientation (vertical or horizontal). It is noted that themeasured temperature reaches a peak value of 173 K at around t = 0.15 s and then drops downgradually by heat diffusion. The main reason for the delay of the peak temperature is thedistance between the heater and the temperature sensor.

A fairly good agreement is observed between the calculated temperature at 0.2 mm andthe experiment. It is a coherent reasoning that in the experiment we have actually measuredthe temperature approximately at 0.2 mm away from the center, because the heater is 0.2 mmthick (or 0.1 mm from the center) and the thermocouple junction is also about 0.2 mm indiameter. Therefore, the experimental data presented below does not quantitatively representthe maximum temperature, but does provide significant information about the diffusion of theapplied heat pulse to the GFRP plate, depending upon the various cooling conditions of liquid.

492

150£

18.S 100

50-0.5 0.0 0.5 1.0

Time(s)1.5 2.0

200

150

100

heater7 down

-0.5 0.5 1.0

Time(s)1.5 2.0

(a) 11 K (b) 65 K

FIGURE 6. Measured temperature history of GFRP plates at different orientations for 100 W/cm2

Orientation of Heating Surface

FIGURE 6(a) is the temperature history for the GFRP plate at three different orientationswhen 100 W/cm2 is supplied at 77 K. It is clearly noticed that the temperature-increasing rateis independent of the orientation during the heating period, but the recovery is faster for theheating surface up than the vertical surface, and slower for the heating surface down. Thesephenomena are closely related with the behavior of bubbles. Recall that the thermalpenetration depth at t = 0.1 s is less than 0.06 mm for GFRP as demonstrated in FIGURE 5(a)[11]. Thus the bubbles, if any, may be generated only on the heating surface of the plate.Obviously, the bubbles can escape more easily from the heating surface when it is facing upthan when it is facing down, which results in the difference of the recovery speed. On theother hand, during the short heating period (100 ms) the buoyant effect depending upon thesurface orientations does not seriously affect the dynamic behavior of bubbles near the surface.We think for this reason that any existing (steady-state) pool-boiling correlations [9] could notbe directly applicable to this extremely transient heat transfer.

FIGURE 6(b) is the temperature history when liquid nitrogen is in a subcooled state at 65K and all the other conditions are the same as in FIGURE 6(a). A distinctive feature of thesecurves is that the temperature change is much less dependent upon the orientation not only forthe short heating period, but also for most of the recovery period. This means that even thoughthe heat pulse may have caused temporary vaporization of liquid on the surface, the vapormust be re-condensed immediately by the subcooled liquid, before escaping as bubbles. Weclaim through these experimental results that the subcooling of liquid nitrogen at 65 K indeedplays a significant role in suppressing the bubbles for the short fault period in an SFCL.

Temperature of Subcooled Liquid

We repeated the same heating procedure for a vertical GFRP plate at various liquid-temperatures, and the temperature history is plotted in FIGURE 7. It is right in general that thelower the liquid temperature is below 77 K, the lower is the peak temperature and the faster isthe recovery. On the other hand, the curves can be sorted in two groups, which means that theeffect of subcooling becomes significant only if the degree of subcooling exceeds a certain

493

limit (3 to 6 degrees in this specific case). We think this behavior is also closely related withthe probability of bubbles on the surface in contact with the subcooled liquid. In FIGURE 7,for example, no noticeable advantage in peak temperature and recovery is expected if liquidnitrogen is subcooled at 74 K, because 3 K of subcooling may not be enough to suppressbubbles. In order to take advantage of subcooling during the fault period of SFCL, we stronglyrecommend that the liquid temperature should be 70 K or lower at atmospheric pressure.

Intensity of Heating Power

The burnout temperature of HTS elements is another crucial factor in the design of anSFCL. It is very difficult in this experiment to provide a quantitative correlation between theheating power and the exact peak temperature. However, it is meaningful to examine thecorrelation between the heating power and the "measured" peak temperature in the experiment,as we may obtain somewhat qualitative and useful information.

FIGURE 8(a) shows the temperature history of the GFRP plate at 65 K for various valuesof heating power, and FIGURE 8(b) plots the measured peak temperature and the temperatureat the end of heating period (t = 0.1 s) as functions of the intensity of heating power. It is noted

200

150

100

•„........„..„..,...........:

-0.5 0.0 0.5 1.0 1.5 2.0

Time (s)

FIGURE 7. Measured temperature history of vertical GFRP plate at different temperatures for 100 W/cm2

150

100

0 50 100 1:

Heating Power (W/cm2)

(a) Temperature history (b) Temperatures at peak and at 0.1 s

FIGURE 8. Measured temperature of vertical GFRP plate at 65 K for different heating powers

494

that the peak temperature increases almost linearly with intensity, but the difference betweenthe two temperatures (and the delay time of the peak temperature from the end of heating)becomes longer as the heating power increases. This means that the actual maximumtemperature in the heater may increase at a higher rate than the measured peak, because a veryintensive heating will vaporize adjacent liquid immediately, but the vapor layer near thesurface may make a thermal barrier of heat diffusion to liquid for the short period.

CONCLUSIONS

Thermal diffusion of a heat pulse in subcooled liquid-nitrogen is experimentallyinvestigated as part of our ongoing SFCL projects. From the measured temperature history forvarious conditions, a few valuable conclusions can be drawn for the thermal phenomena. First,the typical duration of a fault current (-100 ms) is too short for vaporized nitrogen to form anymovable bubbles on the heating surface, thus the rate of temperature increase during the shortperiod is independent of the surface inclination in liquid. Second, the recovery speed from theheat pulse is sensitive to the behavior of bubbles. Third, subcooling liquid nitrogen below 70K is an effective method to suppress bubbles, which will lower the peak temperature andshorten the recovery time. Fourth, the peak temperature caused by a heat pulse may increasevery sharply as the intensity of the heating power increases.

ACKNOWLEDGMENTS

This study is supported by the research funds of the CAST (Center for AppliedSuperconductivity Technology) under the 21C Frontier R&D Program in Korea. The authorsthank Prof. Jae Sang Ro at Hong Ik University for sharing the ac power supply and Dr. HongBeom Jin at Duksung Inc. for supplying components of the experiment.

REFERENCES

1. Wolsky, A. M., Cooling for future power sector equipment incorporating ceramic superconductors. ArgonneNational Laboratory Report, April 2002. (available fVom:http://www.iea.org/tech/scond/scond.)

2. Yazawa, T., et al., "66kV-Class Superconducting Fault Current Limiter Magnet-Subcooled NitrogenCryostat," in Advances in International Cryogenic Engineering 19, edited by G. G. Baguer and P. Seyfert,Narosa, New Delhi, 2002, pp.261-264.

3. Ohtani, Y., et al., "Subcooled Nitrogen Cryostat for 66kV/750A Superconducting Fault Current LimiterMagnet" in Advances in Cryogenic Engineering 49A, edited by J.Waynert et al., American Institute ofPhysics, New York, 2004, pp.867-874.

4. Kang, H., Kirn, H. J., Bae, D. K., Ahn, M. C. and Ko, T. K., Cryogenics 45, pp.65-69 (2005).5. Kang, H., Ahn, M. C., Kim, H.J., Chang, H. M. and Ko, T. K., IEEE Transactions on Applied

Superconductivity 14, pp.883-886 (2004).6. Chang, H. M., Choi, Y.S., Van Sciver, S. W. and Choi, K. D., Cryogenics 43, pp.589-596 (2003).7. Choi, Y. S., Van Sciver, S. W. and Chang, H. M., "Natural Convection of Subcooled Liquid Nitrogen in a

Vertical Cavity/' in Advances in Cryogenic Engineering 49B, edited by J.Waynert et al., American Instituteof Physics, New York, 2002, pp.473-480.

8. Yoshida, S., et al., "1 atm Subcooled Nitrogen Cryogenic system with GM-Refrigerator for a HTS PowerTransformer" in Advances in Cryogenic Engineering 44A, edited by S.Breon et al., American Institute ofPhysics, New York, 2004, pp.867-874.

9. Barren, R. F., Cryogenic Heat Transfer, Taylor & Francis, Philadelphia, 1999, pp.160-172.10. Chapra, S. C. and Canale, R. P., Numerical Methods for Engineers, McGraw-Hill, Boston, 2002, pp.849-852.11. Arpaci, V. S., Conduction Heat Transfer, Addison-Wesley, London, 1966, pp.76-83.

495