- Email: [email protected]
Improvement in pipe chilldown process using low thermal conductive layer
International Journal of Heat and Mass Transfer 111 (2017) 115–122
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Improvement in pipe chilldown process using low thermal conductive layer Daisuke Takeda a, Katsuyoshi Fukiba a,⇑, Hiroaki Kobayashi b a b
Graduate School of Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, Japan Japan Aerospace Exploration Agency, Chohu 182-8522, Japan
a r t i c l e
i n f o
Article history: Received 16 November 2016 Received in revised form 30 March 2017 Accepted 30 March 2017 Available online 5 April 2017 Keywords: Chilldown Cryogenic Flow boiling Coating
a b s t r a c t A method for reducing the time and total mass of cryogenic fluid required for a chilldown process in piping was experimentally investigated in this study. The inner wall of a pipe with an outer diameter of 1/400 (=6.35 mm) was coated with Polytetrafluoroethylene, which has a low thermal conductivity. Liquid nitrogen (LN2) was supplied to the pipe at a constant tank pressure of 120–170 kPa. The fluctuations of the two-phase flow, which were composed of LN2 and gas phase nitrogen, were observed. A pipe without an insulating layer and three other pipes with insulating layers of thicknesses 23 lm, 63 lm, and 91 lm, respectively, were used in the experiment. The results indicated that the temperature of the minimum heat flux point (MHF) was higher for the pipe with the insulating layer. This increased temperature caused earlier transition to nucleate boiling. Furthermore, the total mass of LN2 consumed in the chilldown process could be retrenched up to a maximum of 64%. The heat flux decreased after reaching the MHF point; however, heat flux after MHF point is not dominant to overall chilldown time. The effect of the layer to increase the temperature of MHF point is dominant to overall chilldown time, which results in the decrease in the chilldown time and the total mass of LN2 consumed in the chilldown process. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction 1.1. Background Cryogenic fluids have recently begun to be increasingly used mainly in the fields of aerospace and superconductive materials. In the aerospace field, liquid hydrogen and oxygen have long been used as fuels for rocket engines. The biggest advantage of using liquid hydrogen as a fuel is that its calorific value per volume is higher than that of hydrocarbon fuels, and a number of reports have reported its use as a fuel for aircrafts in the recent years [1]. In the field of superconductivity, liquid nitrogen and helium are used as coolants in commercial MRI (magnetic resonance imaging) and the Large Hadron Collider (LHC) [2,3]. For all the applications described above, a ‘‘chilldown” process is required before the use of cryogenic fluids in the systems. At initial stage these fluids flow into devices through piping systems that are in the ambient temperature state, which is significantly higher than the temperature of cryogenic fluids. Therefore, when these fluids are introduced into such piping systems, they intensively ⇑ Corresponding author. E-mail address: [email protected] (K. Fukiba). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2017.03.114 0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
boil and evaporate. The gasification of the cryogenic fluids causes intense increase in the volume flow rate, resulting in increase of pressure loss. This makes the chilldown process time consuming and also leads to the discharge of the gases generated into the atmosphere during the process. According to Shaeffer et al. [4], only 8% of the calorific value of the cryogenic fluids is used for cooling the piping system during the chilldown process, which is extremely inefficient. Therefore, reducing the time required for chilldown along with retrenching the mass of the cryogenic fluids used is important. In the earlier studies, mechanisms regarding boiling heat transfer and effects of gravity on this heat transfer in the chilldown process have been reported. Hu et al. [2] described the heat transfer phenomena in the chilldown of vertical pipes using liquid nitrogen. They also experimentally investigated the effects of the direction of the flow on the chilldown process. Shaeffer et al. [4] investigated the mechanism governing the heat transfer in the flow of liquid nitrogen with fluctuating flow rates during the process. For practical applications, chilldown is frequently processed at a relatively low flow rate. Yuan et al. [5] studied the heat transfer phenomena under such conditions. Studies have also reported for cryogenic fluids other than liquid nitrogen. Shirai et al. [6] studied the boiling heat transfer characteristics of liquid hydrogen, whereas Hartwig
116
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
Nomenclature c q r t T Greek
a
specific heat (J kg1 K1) heat flux (W m2) radius (m) time (s) temperature (K)
k
q
thermal conductivity (W m1 K1) density (kg m3)
Subscripts i inner wall o outer wall
thermal diffusivity (m2 s1)
et al. [7] conducted a series of experiments using liquid hydrogen. However, only a few experiments have been conducted using liquid hydrogen because it is a fairly costly and dangerous substance.
1.2. Improvement in the boiling heat transfer There have been a large number of studies on methods to use for improving the heat transfer rate during a boiling heat transfer. One of the main methods proposed involves coating the surface of an object: Vakarelski et al. [8] succeeded in drastically increasing the heat flux by using a hydrophilic coating, which promoted the transition to nuclear boiling. In their experiment, they immersed a heated steel sphere with this coating in water. A number of studies have also investigated improving the boiling heat transfer with coatings and not using cryogenic fluids. Shojaeian et al. [9] conducted a review of the results of such studies. Nevertheless, few studies have been reported where cryogenic fluids have been used. One of these studies was an experiment by Hu et al. [10]; they conducted their experiment under pool boiling using a rod that had a nanoporous surface. They increased the critical heat flux (CHF) by 160%. Unfortunately, the methods used seemed to be complex and expensive, which made them unsuitable for practical applications. However, aside from the methods for improving the heat transfer mentioned above, another method for coating a surface of an object has been proposed, although this one uses thin layers with low thermal conductivity. This seems incongruous, as one would expect such layers to prevent heat transfer, and yet this method can reduce the time needed for the chilldown process. As a result of its contradictory nature, this method has been called the ‘‘Paradox of the Insulating Layer”. This technique has ancient origins. The technique has been used in the process of ‘‘Yakiire” (quenching) to manufacture katanas (i.e., a Japanese sword). In the process heated iron is quenched in order to improve the strength of the sword [11]. As academic thesis, Cowely et al. [12] reported the method for improving boiling heat transfer using thin layers with low thermal conductivity in 1962. After a few years of this study, Maddox [13] conducted pool boiling experiments with coated tubes. Allen [14] applied this method for increasing heat transfer to space chamber cryopanels. Subsequently an experiment was conducted with various coatings using liquid helium [15], and another used water and liquid hydrogen [16]. A detailed mechanism using liquid nitrogen was studied by Nishio et al. [17,18], and an experiment using liquid helium was conducted by Chandratilleke et al. [19]. Kikuchi et al. [20], meanwhile, studied the heat transfer modeling in the case where a low thermal conductive layer was used. Recently, Tsoi et al. [21] studied the effects of the thickness of the insulating layers on the heat transfer by applying grease. All studies above were conducted under pool boiling, where cryogenic fluids cannot flow. Dreitser [22] investigated methods of heat transfer enhancement in channels. This study treated forced convection flow with coating. How-
ever, the main topic of this study is introductions of various methods for enhancing heat transfer. Therefore, the details of the experiments in this study were not explained well. The biggest advantage of this method is its durability. When using hydrophilic coatings, surface contamination causes the performance of the heat transfer to deteriorate. However, by using thin layers with low thermal conductivity, it is possible that the surfaces can withstand longer use because the heat transfer performance is not subject to any contaminations. With all of the above in mind, this study looks to discuss the applicability of the ‘‘Paradox of the Insulating Layer” to the chilldown process. We coat the inside of a pipe with a layer of low thermal conductivity, and we evaluate the chilldown time and the amount of cryogenic liquid consumed. Some previous experiments that chilled down piping systems were conducted at a constant flow rate [2,5]. However, rockets being launched have cryogenic fluids pumped into them at a constant tank pressure. In this study, we conduct an experiment at a constant tank pressure to evaluate the effects that a layer of low thermal conductivity can have under more practical conditions. Furthermore, we observed pulsation of the flow during the start of the chilldown process in spite of using a constant tank pressure. In this paper, report on this phenomenon is also presented.
1.3. Principle behind the paradox of the insulation layer This section provides an explanation about the typical boiling regimes used during chilldown as well as the mechanism in the paradox of the insulating layer that reduces the chilldown time. The temperature profile of a pipe during a typical chilldown process is shown in Fig. 1. During the chilldown process, the initial value of the pipe’s temperature at point A is usually that of the
Fig. 1. Typical temperature profile of a pipe during the chilldown process.
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
ambient temperature. At this time the difference in temperature between the pipe and the fluid is still so large that it causes film boiling to occur in the pipe. During film boiling, a film of vapor forms on the surface of the pipe. The resultant heat transfer is low, and the temperature of the pipe only gently decreases. Once the pipe has cooled somewhat (point B), film boiling stops occurring and instead becomes nucleate boiling through transition boiling regime. During nucleate boiling, the amount of heat transferred is much greater than that transferred during film boiling. As a result, the temperature of the pipe drastically decreases. Finally, the chilldown process terminates when the temperature of the pipe reaches that of the cryogenic fluid. With this in mind, we now consider a pipe with a layer of low thermal conductivity. The temperature distribution inside the pipe during the chilldown process is shown in Fig. 2. Because of this layer, the temperature of the surface that comes into contact with the cryogenic fluid is lower than that in the case where there is no layer. As the onset of the transition to nucleate boiling is determined by the temperature at the contact surface, the layer enables the boiling regime to shift more quickly from film to nucleate boiling. As a result, the time taken for the chilldown process is drastically reduced. However, this is theoretical and the actual phenomena are multidimensional and more complicated. In fact, the increase of the temperature at the minimum heat flux (MHF) point cannot be explained quantitatively by just the thermal resistance of the layer. Kikuchi et al. [20] assumed that points exist locally where the surface comes into contact with the liquid even during the film boiling regime. By using this assumption, they developed a model that considers the generation of bubbles from the points. This model can evaluate the increase in the temperature at the MHF point. Meanwhile, Moreaux et al. [16] and Nishio et al. [17] conducted an experiment under pool boiling and added a thin metal layer on top of layer of low thermal conductivity. This resulted in the chilldown process time increasing dramatically. Based on this result, it would seem that the promotion of the transition to nucleate boiling for a surface with a layer of low thermal conductivity is generated locally at the points where the surface contacts the liquid. If an additional metal layer exists on top of this insulating layer, the nucleate boiling cannot be retained because of the thermal conduction through the metal layer and be the cause for the time for the chilldown process increasing.
117
2. Experimental setup and measurements 2.1. Experimental setup and procedure The effects of a layer of low thermal conductivity on the time taken for the chilldown process and the reduction of the mass of the cryogenic fluid were evaluated using an experimental setup that composed of a tank, test section, and measurement system for the flow rate. Liquid nitrogen (LN2) was selected as the working fluid. A schematic diagram of the experimental apparatus is shown in Fig. 3. The system was composed of pipes whose outer diameters were 6.35 mm. LN2 was stored in the tank, and was pumped downstream in the system by compressed air. A three-way valve was located upstream of the test section. At the start of the experiment, the test section was separated from the three-way valve. LN2 flowed into a catch tank underneath the three-way valve, and the upstream part of the test section cooled down. Once the upstream part had cooled sufficiently, the test section was connected to the three-way valve with a joint (Swagelok 1/400 tube fitting), and LN2 was introduced into the test section by switching the three-way valve. The test section was composed of a pipe that had a length of 110 mm and an inner diameter of 4.35 mm. The material of the pipe was stainless steel (SUS 304). The outer surface of the pipe was insulated by polystyrene foam whose thickness was 25 mm. The heat evacuation to the atmosphere through the polystyrene foam was relatively large near the MHF point, where the heat flux into the system reached its minimum value. The maximum heat evacuation at this point was determined to be up to 16%. It is, nevertheless, ignored in the analysis below. A thermocouple was soldered onto the center of the test section, and this was where the temperature of the pipe was measured. This thermocouple was located on the top of the tube. In a preliminary experiment, the temperatures on the top, middle and bottom of the pipe were measured using three thermocouples. The result showed that there were no differences. The distance from the three-way valve to the thermocouple is 55 mm. A coilshaped copper pipe with a length of 7.5 m and a diameter of 6.35 mm was set downstream of the test section so as to act as a heat exchanger, as the LN2 was expected to completely evaporate here. The mass flow rate of the gaseous nitrogen (GN2) was subsequently measured.
2.2. Layer of low thermal conductivity
Fig. 2. Temperature distribution with a layer of low conductivity.
In this experiment, the pipe was coated with a layer that had a thickness of less than 100 lm. The layer was made of Polytetrafluoroethylene (PTFE, Teflon 959G-200) and had a thermal conductivity of 0.25 W m1 K1, according to a catalog. Teflon 959G-200 is so-called one coat paint, which does not require a primer coat. The process of the coating was simple. First, the pipe was cleansed with an organic solvent. Generally, objectives of the coating is prebaked in a kiln in the next process. However, we skipped this process in this time. We also skipped the process of abrasive blasting, which was also conducted in general coating processes. The reason why we skipped abrasive blasting is that the inner diameter of the pipe is so small that we could not blast. Then, the pipe was fixed vertically and the PTFE was supplied as a liquid paint, which was introduced into the pipe. Next, the pipe sintered in a kiln at about 600 K. The thickness of the layer was varied by pouring the paint into the pipe repeatedly. Four pipes were processed using this procedure for each thickness. The three pipes that were not used in the experiment were cut off, and the thickness of the layer of PTFE in them was measured using magnified images of their cross sections.
118
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
Fig. 3. Schematic diagram of the experimental setup.
The enlarged view is shown in Fig. 4. A pipe was cut off at three cross sections. The results of the measurements of the thickness are shown in Table 1. In this table the symbol L means the cross section at 20 mm from the left edge, the symbol C means the center of the pipe, and the symbol R means the cross section at 20 mm from the right edge, respectively. Although the paint was poured into the pipe, no effect of gravity on the thicknesses was observed. Eventually, the layer was coated with three different thickness. If all the thicknesses are averaged, the thicknesses are 25.8, 56.8 and 93.7 lm, respectively. However, in this study the thermocouple is located on the center of the pipe. Therefore, we evaluated the thickness of the layer using the values at the center. In this case, the thicknesses are 22.5, 63.2 and 91.4 lm, respectively. We use these values in this paper. 2.3. Measurements and data reduction In this experiment, the main measurement items were those of the temperature of the pipe, the pressures in the tank and at the outlet of the test section, and the mass flow rate. The temperature was measured by a T-type thermocouple, and the measurement was used to calculate the inner wall temperature as shown below [4]:
2 !! r 2o ri ri dT o Ti ¼ T0 þ 1 2 ln 4a ro ro dt 2 r2o r 2i r2o r 2i d T o 1 ri r 4o ri 4 4 þ ðr 5r Þ ln ln þ o 64a2 i 8a2 r o 16a2 ro 16a2 dt2
The heat flux on the inner wall can be calculated as shown below:
2 ! 2 r i r2o dT o ðqcÞ2 r 3i r 4o r 2o ri r i d To qi ¼ qc þ ln 2 2r i dt k 16 16r i 4 ro dt 3 r2 r3 ri r4 ri ri d T o ðqcÞ3 r 5i 3r 4 r i 3r2 r 3 r6 þ 2 o þ o i o o i ln o ln 384 128 128 384r i 32 ro 32 r o dt3 k
The pressures were converted into electrical signals by a pressure transducer (KYOWA PG-2KU). The mass flow rate was measured by a mass flow meter (Azvil CMS0200). All the measurement items used an analogue-to-digital converter, and the results were
Fig. 4. Layer of low thermal conductivity on the inner surface.
recorded at a sampling rate of 20 Hz. The uncertainties in each measurement are shown in Table 2, and were calculated according to literature provided by Moffat [23]. The point in time when the temperature of the pipe reached 290 K is defined as 0 s in our graphs below. 3. Results 3.1. Without the insulating layers First, the experiments were conducted four times under the same conditions (tank pressure of 150 kPa, no insulating layer), and the temperatures variations were obtained; these are shown in Fig. 5. Here, the time needed for the chilldown process is defined as the time until the temperature variation per second becomes less than 0.1 K s1. When the process was conducted at a constant tank pressure, a variation in the time taken for the chilldown process to finish occurs; this is believed to be due to an unstable flow with the oscillation forms. The precision index of the time taken for this process was calculated from the four experiment, and was found to be 16 s, as shown in Fig. 5. In Fig. 6, the temperature variations of the pipes without the layers of low thermal conductivity are shown for tank pressure of 120, 135, 150, and 170 kPa. The experiments were conducted several times at each tank pressure. The temperature histories that were closest to the average of the time taken for the chilldown process at each pressure are shown in Fig. 6. From this, we can see that the time decreased as the tank pressure increased. Fig. 7 presents the time taken for the chilldown process against the four tank pressures. The length of the error bars in this figure indicates the precision index, which means the deviation of the data [23]. As the pressure in the tank increased from 120 kPa to 170 kPa, the time taken for the process decreased from 329 s to 115 s. The precision index of the time taken for the chilldown process decreased as the tank pressure increased. In Fig. 8, the pressure downstream of the test section is shown for a tank pressure of 150 kPa, along with the temperature variation of the pipe. The tank pressure was kept relatively constant through adjustments made by the regulator. It can be seen that the downstream pressure oscillated strongly, and at the beginning exceeded that of the tank pressure. Fig. 8(b) shows the downstream pressure variation from 45 to 55 s. The pressure oscillated with an amplitude of up to 15 kPa at a frequency of 2 Hz. The oscillations converged as the temperature of the pipe decreased. This oscillation is thought to be a densitywave oscillation. A number of studies have looked into these [24–26], and so the mechanism governing this oscillation will not be discussed in this paper. In Fig. 9, the mass flow rate variation for each tank pressure is shown. At higher tank pressures, even if the tank pressure remained constant, the mass flow rate increased as time passed. This is because pressure loss decreased with decrease of void
119
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122 Table 1 Thickness of the layer and its precision index [23] (unit: lm). Number of times of coating
Cross section
Sample 1
Sample 2
Sample 3
1
L C R L C R L C R
49.5 34.3 30.3 45.0 77.3 64.7 68.2 86.7 101.5
19.1 22.0 25.7 54.6 62.3 52.6 103.8 91.8 105.2
18.2 11.4 21.5 51.6 50.1 52.7 104.9 95.7 85.9
2
3
Average
Precision index
22.5
11.5
63.2
13.6
91.4
4.5
Table 2 Uncertainties in the measurements made. Parameter
Uncertainty
Pressure Mass flow rate Temperature
±1.5 kPa ±0.05 g/s 3K
Fig. 7. Time taken for the chilldown process against tank pressure variation.
Fig. 5. Repeatability of the experiments under the same condition at the tank pressure of 150 kPa.
Fig. 10 shows the chilldown curves calculated from the temperature histories. Although it is difficult to distinguish a difference in the graph, the heat flux increased alongside the tank pressure during the film boiling. The difference in the heat flux in this regime resulted in the difference observed in the time taken for the chilldown process. During the regime from the MHF point to the end of the chilldown process via the nucleate boiling regime, the higher the tank pressure became, the larger heat flux became. At lower tank pressures, the heat flux oscillated and the CHF point was unclear. The temperature of the MHF point varied from 124 to 130 K, which indicates that the effect of the tank pressure on the temperature of the MHF point is small.
3.2. With the insulating layer
Fig. 6. Temperature variation of the test section with four different tank pressures.
fraction as the pipe cooled down. The total mass at each tank pressure, which is the integrated mass flow rate, will be shown in Fig. 13 along with the results of the total mass obtained by the layer of low thermal conductivity. Nevertheless, it can be seen from Fig. 9 that the mass flow rate increased as the tank pressure increased, although the time for the chilldown process decreased. As a result, the total mass was least at the tank pressure of 170 kPa.
Fig. 11 shows the temperature variation of the metal surface of the pipe that is covered by layer of low thermal conductivity. Here, the vertical axis does not mean the surface temperature of the layer, but rather the surface temperature of the stainless tube. The tank pressure for this experiment was set to 150 kPa. Note that the temperature histories that were closest to the average of the time taken for the chilldown process are shown in Fig. 11. Although the behavior during the film boiling regime hardly changed, the layer enabled the transition in the boiling regimes to start earlier. After the transition, the temperature steeply decreased. The temperature variation of the pipe with the layer has a gentler slope after the start of the transition boiling than that of the pipe without the layer. In Fig. 12, the effect of the thickness of the layer on the time taken by the chilldown process is shown at each tank pressure. The error bars in this figure indicate the mean dispersion (or precision index) of the experiments, and these were obtained by conducting the experiments several times over under the same conditions. The dispersion got smaller as the tank pressure
120
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
(a) From the start to the end of the test.
Fig. 10. Chilldown curves conducting without an insulating coating for tank pressures of 120, 135, 150 and 170 kPa.
(b) Pressure downstream of the test section from 45 to 55 s.
Fig. 11. Temperature profile for pipes with and without the layer of low thermal conductivity.
Fig. 8. Time variation of the downstream pressure. The tank pressure was 150 kPa.
Fig. 9. Variation of the mass flow rate with tank pressure.
increased. Some error bars overlap with the symbols and can be therefore not visible. It is clear from Fig. 12 that the time needed for the chilldown process was drastically reduced by the insulating layer. In Fig. 12 (b), the time is normalized at each tank pressure by the time taken for the experiments conducted without the layer. Although it seems the variations in tank pressure made a difference in the time taken, overall the time was reduced by 40–60%. There was no significant difference between the pipes that had a layer of thickness
of 63 lm and 91 lm. Previous studies conducted on pool boiling reported that the reduction in the time taken for the chilldown process to occur plateaued at a certain thickness of the layers as the thickness of the layers increased [14]. This was observed because the effect of the insulating layer as heat resistance exceeded the effect to reduce the time taken until the transition boiling began. In Fig. 12(a), the time for the thickness of 91 lm and at 120 kPa is 129 s. The time without a coating (0 lm) and at 170 kPa is 115 s. These times are similar, which means that the increase in the flow rate (or Re number) provides the similar effect on the chilldown time. Generally, the heat flux of the wall increases with the Re number in the film boiling regime. Consequently, the chilldown time seems to decrease. We need more detailed investigation on this point in the future. In Fig. 13, the total mass of LN2 consumed until the termination of the chilldown process, as determined through the integration of the mass flow rate, is shown. It is clear that the total mass of LN2 used was reduced by the layer of low thermal conductivity. As in Fig. 12, there was no large difference between the results obtained by the thicknesses of 63 and 91 lm. The rate of the reduction in chilldown time, as shown in Fig. 13 b), is larger at the lower tank pressures, where the total mass of LN2 consumed is relatively high. The rate of reduction reached its maximum when the thickness was 91 lm and the tank pressure was 135 kPa. The total mass under these conditions was 35% of that obtained without the layer. Fig. 14 shows the chilldown curve when the insulating layer was used. The tank pressure used was 150 kPa. The heat fluxes during the film boiling regime were slightly lower when the layer was
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
(a) Chilldown time
(a) Total mass
(b) Normalized chilldown time
(b) Non-dimensionalized total mass
Fig. 12. Effect of the coating thickness on the time taken for the chilldown process.
used than when it was not, though this difference is smaller than the uncertainty. The temperature at the MHF point increased as the thickness increased: while the temperature at the MHF point was about 135 K without the layer, it was 160 K with a 23 lm layer and 185 K with the layers of 63 and 91 lm. The heat flux at the CHF point with the layer drastically decreased compared to that without the layer. The heat flux decreased as the thickness increased. The heat flux at the CHF point for a thickness of 91 lm was half of the value it had for a thickness of 63 lm. However, the proportion of time for the transition and nucleate boiling as compared to the total time for the whole chilldown process was small. Therefore, the heat flux after the transition have little impact on the overall chilldown time. Consequently, the time needed for the chilldown process was slightly longer for the scenario where the layer thickness was 91 lm as compared to the case where the thickness was 63 lm. As regards the decrease in the heat flux at the CHF point with the layer, Chandratilleke et al. [27] pointed out that there are two factors to decrease the heat flux. One is the low thermal conductivity of the layer. Generally, low thermal conductivity layers decrease heat flux into a wall. The other is the difference of nucleation sites of the surface. At transition and nucleate boiling regime, the increase in the number of the nucleation sites causes the increase in the heat flux. The surface of PTFE is smoother than that of the metal, which means less nucleation sites. These two factors decrease the heat flux. In previous studies on the paradox of the insulating layer, although some researchers indicated only temperature profiles [13,16,21], the gradients of the temperature profile at the transition and nucleate boiling regime decreased with the low
121
Fig. 13. Total mass of LN2 as calculated by integrating the mass flow rate.
Fig. 14. The chilldown curves for scenarios where the layer of low thermal conductivity was, and was not, used. The tank pressure was 150 kPa.
thermal conductive layer, which means the decrease in the heat flux. However, for example, the heat flux did not decrease in the experiment using liquid helium by Chandratilleke et al. [19]. Further studies are needed to explain these phenomena. Fig. 15 shows the time variation of the heat flux and the amount of heat transferred from the pipe. There is no difference between the heat fluxes at the film boiling regime. The differences exist only after the transition and nucleate boiling start. Fig. 15(b) shows the amount of heat transferred from the pipe. These values are calculated by integrating the heat fluxes. In this figure, the dotted line
122
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
a small proportion of the overall time taken for the chilldown process. In fact, because the layer increased the temperature of the MHF point, we saw a decrease in the overall time needed for the chilldown process. Acknowledgements This study has been conducted in the collaborative research with Japan Aerospace Exploration Agency titled ‘‘Research on the technology for providing hydrogen fuel to aircraft”. We are grateful to Prof. Hidetoshi Ohkubo for providing useful comments. References
(a) Time variation of the heat flux with and without the layer.
(b) Time variation of the amount of heat transferred from the pipe. The dotted line is the total amount of heat needed to cool the pipe. Fig. 15. Time variation of the heat flux and the amount of heat transferred from the pipe.
indicates the total amount of heat needed to cool the pipe, which is obtained with the mass of the pipe, the specific heat and the temperature difference from the start to the end of the experiment. These figures show that the total amounts of heat transferred are almost the same. We can understand that the timing where the transition boiling starts determines the time needed for the chilldown. 4. Conclusions In this study, a method for reducing the time and total mass of cryogenic fluid needed for the chilldown process was experimentally investigated. Our setup involved an arrangement of pipes, and these were coated with Polytetrafluoroethylene; this was because the resin has low thermal conductivity. Our experiment used LN2, and the pressure of the two phase flow of this substance oscillated even though it was being supplied at a constant pressure by a storage tank. Experiments were conducted using stainless pipes, and three different thicknesses of the Polytetrafluoroethylene layers were used. Our results show that, the greater the thickness of this layer, the higher the temperature at the minimum heat flux (MHF) point becomes. These experiments also revealed that the layer enables transition boiling to start earlier. Consequently, the total mass of LN2 consumed was reduced by a maximum of 64%. The heat flux decreased after reaching the MHF point, though this was, overall,
[1] B. Khandelwal, A. Karakurt, P.R. Sekaran, V. Sethi, R. Singh, Hydrogen powered aircraft: the future of air transport, Prog. Aerosp. Sci. 60 (2013) 45–59. [2] H. Hu, J.N. Chung, S.H. Amber, An experimental study on flow patterns and heat transfer characteristics during cryogenic chilldown in a vertical pipe, Cryogenics 52 (2012) 268–277. [3] L. Evans, The large hadron collider, Annu. Rev. Nucl. Part Sci. 61 (2011) 435– 466. [4] R. Shaeffer, H. Hu, J.N. Chung, An experimental study on liquid nitrogen pipe chilldown and heat transfer with pulse flows, Int. J. Heat Mass Transfer 67 (2013) 955–966. [5] K. Yuan, Y. Ji, J.N. Chung, Cryogenic chilldown process under low flow rates, Int. J. Heat Mass Transfer 50 (2007) 4011–4022. [6] Y. Shirai, H. Tatsumoto, M. Shiotsu, K. Hata, H. Kobayashi, Y. Naruo, Y. Inatani, Boiling heat transfer from a horizontal flat plate in a pool of liquid hydrogen, Cryogenics 50 (2010) 410–416. [7] J. Hartwig, H. Hu, J. Styborski, J.N. Chung, Comparison of cryogenic flow boiling in liquid nitrogen and liquid hydrogen chilldown experiments, Int. J. Heat Mass Transfer 88 (2015) 662–673. [8] I.U. Vakarelski, N.A. Patankar, J.O. Marston, D.Y.C. Chan, S.T. Thoroddsen, Stabilization of Leidenfrost vapour layer by textured superhydrophobic surfaces, Nature 489 (2012) 274–277. [9] M. Shojaeian, A. Kosar, Pool boiling and flow boiling on micro- and nanostructured surfaces, Exp. Thermal Fluid Sci. 63 (2015) 45–73. [10] H. Hu, C. Xu, Y. Zhao, R. Shaeffer, K.J. Ziegler, J.N. Chung, Modification and enhancement of cryogenic quenching heat transfer by a nanoporous surface, Int. J. Heat Mass Transfer 80 (2015) 636–643. [11] T. Uehara, T. Inoue, Quenching process simulation of Japanese sword covered with clay, J. Soc. Mat. Sci., Japan, 44 (498) (1995) 309–315. (in Japanese). [12] C.W. Cowley, W.J. Timson, J.A. Sawdye, A method for improving heat transfer to a cryogenic fluid, Adv. Cryog. Eng. 7 (1962) 385–390. [13] J.P. Maddox, T.H.K. Frederking, Cooldown of insulated metal tubes to cryogenic temperatures, Adv. Cryog. Eng. 11 (1966) 536–545. [14] L.D. Allen, A method of increasing heat transfer to space chamber cryopanels, Adv. Cryog. Eng. 11 (1966) 547–553. [15] A.P. Butler, G.B. James, B.J. Maddock, W.T. Norris, Improved pool boiling heat transfer to helium from treated surfaces and its application to superconducting magnets, Int. J. Heat Mass Transfer 13 (1) (1970) 105–115. [16] F. Moreaux, J.C. Chevrier, G. Beck, Destabilization of film boiling by means of a thermal resistance, Int. J. Multiph. Flow 2 (2) (1975) 183–190. [17] S. Nishio, Study on minimum heat-flux point during boiling heat transfer on horizontal plates: effects of transients and thermal conductance of wall, Jap. Soc. Mech. Eng., Ser. B 51 (462) (1985) 582–590 (in Japanese). [18] S. Nisio, Y. Serizawa, Control of minimum heat flux point temperature by the thermal conductance of an additional surface layer, Jap. Soc. Mech. Eng., Ser. B 53 (487) (1987) 1061–1064 (in Japanese). [19] G.R. Chandratilleke, S. Nishio, Pool boiling heat transfer to saturated liquid helium on coated surfaces, Cryogen. Superconduct. Soc. Japan, 23(3) (1988) 128–133. [20] Y. Kikuchi, T. Hori, I. Michiyoshi, Minimum film boiling temperature for cooldown of insulated metals in saturated liquid, Int. J. Heat Mass Transfer 28 (6) (1985) 1105–1114. [21] A.N. Tsoi, A.N. Pavlenko, Enhancement of transient heat transfer at boiling on a plate surface with low thermoconductive coatings, Thermophys. Aeromech. 22 (6) (2015). [22] G.A. Dreitser, Heat transfer enhancement in channels for film boiling of cryogenic liquids, Appl. Therm. Eng. 25 (2005) 2512–2521. [23] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Thermal Fluid Sci. 1 (1988) 3–17. [24] J.A. Boure, A.E. Bergles, L.S. Tong, Review of two-phase flow instability, Nucl. Eng. Des. 25 (1973) 165–192. [25] S. Kakac, B. Bon, A Review of two-phase flow dynamic instabilities in tube boiling systems, Int. J. Heat Mass Transfer 51 (3–4) (2008) 399–433. [26] L.C. Ruspini, C.P. Marcel, A. Clausse, Two-phase flow instabilities: a review, J. Heat Mass Transfer 71 (2014) 521–548. [27] G.R. Chandratilleke, A. Sato, Heat transfer characteristics of cooling, Low Temp. Technol. 28 (5) (1993) 244–252 (in Japanese).
Contents lists available at ScienceDirect
International Journal of Heat and Mass Transfer journal homepage: www.elsevier.com/locate/ijhmt
Improvement in pipe chilldown process using low thermal conductive layer Daisuke Takeda a, Katsuyoshi Fukiba a,⇑, Hiroaki Kobayashi b a b
Graduate School of Engineering, Shizuoka University, 3-5-1 Johoku, Naka-ku, Hamamatsu, Japan Japan Aerospace Exploration Agency, Chohu 182-8522, Japan
a r t i c l e
i n f o
Article history: Received 16 November 2016 Received in revised form 30 March 2017 Accepted 30 March 2017 Available online 5 April 2017 Keywords: Chilldown Cryogenic Flow boiling Coating
a b s t r a c t A method for reducing the time and total mass of cryogenic fluid required for a chilldown process in piping was experimentally investigated in this study. The inner wall of a pipe with an outer diameter of 1/400 (=6.35 mm) was coated with Polytetrafluoroethylene, which has a low thermal conductivity. Liquid nitrogen (LN2) was supplied to the pipe at a constant tank pressure of 120–170 kPa. The fluctuations of the two-phase flow, which were composed of LN2 and gas phase nitrogen, were observed. A pipe without an insulating layer and three other pipes with insulating layers of thicknesses 23 lm, 63 lm, and 91 lm, respectively, were used in the experiment. The results indicated that the temperature of the minimum heat flux point (MHF) was higher for the pipe with the insulating layer. This increased temperature caused earlier transition to nucleate boiling. Furthermore, the total mass of LN2 consumed in the chilldown process could be retrenched up to a maximum of 64%. The heat flux decreased after reaching the MHF point; however, heat flux after MHF point is not dominant to overall chilldown time. The effect of the layer to increase the temperature of MHF point is dominant to overall chilldown time, which results in the decrease in the chilldown time and the total mass of LN2 consumed in the chilldown process. Ó 2017 Elsevier Ltd. All rights reserved.
1. Introduction 1.1. Background Cryogenic fluids have recently begun to be increasingly used mainly in the fields of aerospace and superconductive materials. In the aerospace field, liquid hydrogen and oxygen have long been used as fuels for rocket engines. The biggest advantage of using liquid hydrogen as a fuel is that its calorific value per volume is higher than that of hydrocarbon fuels, and a number of reports have reported its use as a fuel for aircrafts in the recent years [1]. In the field of superconductivity, liquid nitrogen and helium are used as coolants in commercial MRI (magnetic resonance imaging) and the Large Hadron Collider (LHC) [2,3]. For all the applications described above, a ‘‘chilldown” process is required before the use of cryogenic fluids in the systems. At initial stage these fluids flow into devices through piping systems that are in the ambient temperature state, which is significantly higher than the temperature of cryogenic fluids. Therefore, when these fluids are introduced into such piping systems, they intensively ⇑ Corresponding author. E-mail address: [email protected] (K. Fukiba). http://dx.doi.org/10.1016/j.ijheatmasstransfer.2017.03.114 0017-9310/Ó 2017 Elsevier Ltd. All rights reserved.
boil and evaporate. The gasification of the cryogenic fluids causes intense increase in the volume flow rate, resulting in increase of pressure loss. This makes the chilldown process time consuming and also leads to the discharge of the gases generated into the atmosphere during the process. According to Shaeffer et al. [4], only 8% of the calorific value of the cryogenic fluids is used for cooling the piping system during the chilldown process, which is extremely inefficient. Therefore, reducing the time required for chilldown along with retrenching the mass of the cryogenic fluids used is important. In the earlier studies, mechanisms regarding boiling heat transfer and effects of gravity on this heat transfer in the chilldown process have been reported. Hu et al. [2] described the heat transfer phenomena in the chilldown of vertical pipes using liquid nitrogen. They also experimentally investigated the effects of the direction of the flow on the chilldown process. Shaeffer et al. [4] investigated the mechanism governing the heat transfer in the flow of liquid nitrogen with fluctuating flow rates during the process. For practical applications, chilldown is frequently processed at a relatively low flow rate. Yuan et al. [5] studied the heat transfer phenomena under such conditions. Studies have also reported for cryogenic fluids other than liquid nitrogen. Shirai et al. [6] studied the boiling heat transfer characteristics of liquid hydrogen, whereas Hartwig
116
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
Nomenclature c q r t T Greek
a
specific heat (J kg1 K1) heat flux (W m2) radius (m) time (s) temperature (K)
k
q
thermal conductivity (W m1 K1) density (kg m3)
Subscripts i inner wall o outer wall
thermal diffusivity (m2 s1)
et al. [7] conducted a series of experiments using liquid hydrogen. However, only a few experiments have been conducted using liquid hydrogen because it is a fairly costly and dangerous substance.
1.2. Improvement in the boiling heat transfer There have been a large number of studies on methods to use for improving the heat transfer rate during a boiling heat transfer. One of the main methods proposed involves coating the surface of an object: Vakarelski et al. [8] succeeded in drastically increasing the heat flux by using a hydrophilic coating, which promoted the transition to nuclear boiling. In their experiment, they immersed a heated steel sphere with this coating in water. A number of studies have also investigated improving the boiling heat transfer with coatings and not using cryogenic fluids. Shojaeian et al. [9] conducted a review of the results of such studies. Nevertheless, few studies have been reported where cryogenic fluids have been used. One of these studies was an experiment by Hu et al. [10]; they conducted their experiment under pool boiling using a rod that had a nanoporous surface. They increased the critical heat flux (CHF) by 160%. Unfortunately, the methods used seemed to be complex and expensive, which made them unsuitable for practical applications. However, aside from the methods for improving the heat transfer mentioned above, another method for coating a surface of an object has been proposed, although this one uses thin layers with low thermal conductivity. This seems incongruous, as one would expect such layers to prevent heat transfer, and yet this method can reduce the time needed for the chilldown process. As a result of its contradictory nature, this method has been called the ‘‘Paradox of the Insulating Layer”. This technique has ancient origins. The technique has been used in the process of ‘‘Yakiire” (quenching) to manufacture katanas (i.e., a Japanese sword). In the process heated iron is quenched in order to improve the strength of the sword [11]. As academic thesis, Cowely et al. [12] reported the method for improving boiling heat transfer using thin layers with low thermal conductivity in 1962. After a few years of this study, Maddox [13] conducted pool boiling experiments with coated tubes. Allen [14] applied this method for increasing heat transfer to space chamber cryopanels. Subsequently an experiment was conducted with various coatings using liquid helium [15], and another used water and liquid hydrogen [16]. A detailed mechanism using liquid nitrogen was studied by Nishio et al. [17,18], and an experiment using liquid helium was conducted by Chandratilleke et al. [19]. Kikuchi et al. [20], meanwhile, studied the heat transfer modeling in the case where a low thermal conductive layer was used. Recently, Tsoi et al. [21] studied the effects of the thickness of the insulating layers on the heat transfer by applying grease. All studies above were conducted under pool boiling, where cryogenic fluids cannot flow. Dreitser [22] investigated methods of heat transfer enhancement in channels. This study treated forced convection flow with coating. How-
ever, the main topic of this study is introductions of various methods for enhancing heat transfer. Therefore, the details of the experiments in this study were not explained well. The biggest advantage of this method is its durability. When using hydrophilic coatings, surface contamination causes the performance of the heat transfer to deteriorate. However, by using thin layers with low thermal conductivity, it is possible that the surfaces can withstand longer use because the heat transfer performance is not subject to any contaminations. With all of the above in mind, this study looks to discuss the applicability of the ‘‘Paradox of the Insulating Layer” to the chilldown process. We coat the inside of a pipe with a layer of low thermal conductivity, and we evaluate the chilldown time and the amount of cryogenic liquid consumed. Some previous experiments that chilled down piping systems were conducted at a constant flow rate [2,5]. However, rockets being launched have cryogenic fluids pumped into them at a constant tank pressure. In this study, we conduct an experiment at a constant tank pressure to evaluate the effects that a layer of low thermal conductivity can have under more practical conditions. Furthermore, we observed pulsation of the flow during the start of the chilldown process in spite of using a constant tank pressure. In this paper, report on this phenomenon is also presented.
1.3. Principle behind the paradox of the insulation layer This section provides an explanation about the typical boiling regimes used during chilldown as well as the mechanism in the paradox of the insulating layer that reduces the chilldown time. The temperature profile of a pipe during a typical chilldown process is shown in Fig. 1. During the chilldown process, the initial value of the pipe’s temperature at point A is usually that of the
Fig. 1. Typical temperature profile of a pipe during the chilldown process.
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
ambient temperature. At this time the difference in temperature between the pipe and the fluid is still so large that it causes film boiling to occur in the pipe. During film boiling, a film of vapor forms on the surface of the pipe. The resultant heat transfer is low, and the temperature of the pipe only gently decreases. Once the pipe has cooled somewhat (point B), film boiling stops occurring and instead becomes nucleate boiling through transition boiling regime. During nucleate boiling, the amount of heat transferred is much greater than that transferred during film boiling. As a result, the temperature of the pipe drastically decreases. Finally, the chilldown process terminates when the temperature of the pipe reaches that of the cryogenic fluid. With this in mind, we now consider a pipe with a layer of low thermal conductivity. The temperature distribution inside the pipe during the chilldown process is shown in Fig. 2. Because of this layer, the temperature of the surface that comes into contact with the cryogenic fluid is lower than that in the case where there is no layer. As the onset of the transition to nucleate boiling is determined by the temperature at the contact surface, the layer enables the boiling regime to shift more quickly from film to nucleate boiling. As a result, the time taken for the chilldown process is drastically reduced. However, this is theoretical and the actual phenomena are multidimensional and more complicated. In fact, the increase of the temperature at the minimum heat flux (MHF) point cannot be explained quantitatively by just the thermal resistance of the layer. Kikuchi et al. [20] assumed that points exist locally where the surface comes into contact with the liquid even during the film boiling regime. By using this assumption, they developed a model that considers the generation of bubbles from the points. This model can evaluate the increase in the temperature at the MHF point. Meanwhile, Moreaux et al. [16] and Nishio et al. [17] conducted an experiment under pool boiling and added a thin metal layer on top of layer of low thermal conductivity. This resulted in the chilldown process time increasing dramatically. Based on this result, it would seem that the promotion of the transition to nucleate boiling for a surface with a layer of low thermal conductivity is generated locally at the points where the surface contacts the liquid. If an additional metal layer exists on top of this insulating layer, the nucleate boiling cannot be retained because of the thermal conduction through the metal layer and be the cause for the time for the chilldown process increasing.
117
2. Experimental setup and measurements 2.1. Experimental setup and procedure The effects of a layer of low thermal conductivity on the time taken for the chilldown process and the reduction of the mass of the cryogenic fluid were evaluated using an experimental setup that composed of a tank, test section, and measurement system for the flow rate. Liquid nitrogen (LN2) was selected as the working fluid. A schematic diagram of the experimental apparatus is shown in Fig. 3. The system was composed of pipes whose outer diameters were 6.35 mm. LN2 was stored in the tank, and was pumped downstream in the system by compressed air. A three-way valve was located upstream of the test section. At the start of the experiment, the test section was separated from the three-way valve. LN2 flowed into a catch tank underneath the three-way valve, and the upstream part of the test section cooled down. Once the upstream part had cooled sufficiently, the test section was connected to the three-way valve with a joint (Swagelok 1/400 tube fitting), and LN2 was introduced into the test section by switching the three-way valve. The test section was composed of a pipe that had a length of 110 mm and an inner diameter of 4.35 mm. The material of the pipe was stainless steel (SUS 304). The outer surface of the pipe was insulated by polystyrene foam whose thickness was 25 mm. The heat evacuation to the atmosphere through the polystyrene foam was relatively large near the MHF point, where the heat flux into the system reached its minimum value. The maximum heat evacuation at this point was determined to be up to 16%. It is, nevertheless, ignored in the analysis below. A thermocouple was soldered onto the center of the test section, and this was where the temperature of the pipe was measured. This thermocouple was located on the top of the tube. In a preliminary experiment, the temperatures on the top, middle and bottom of the pipe were measured using three thermocouples. The result showed that there were no differences. The distance from the three-way valve to the thermocouple is 55 mm. A coilshaped copper pipe with a length of 7.5 m and a diameter of 6.35 mm was set downstream of the test section so as to act as a heat exchanger, as the LN2 was expected to completely evaporate here. The mass flow rate of the gaseous nitrogen (GN2) was subsequently measured.
2.2. Layer of low thermal conductivity
Fig. 2. Temperature distribution with a layer of low conductivity.
In this experiment, the pipe was coated with a layer that had a thickness of less than 100 lm. The layer was made of Polytetrafluoroethylene (PTFE, Teflon 959G-200) and had a thermal conductivity of 0.25 W m1 K1, according to a catalog. Teflon 959G-200 is so-called one coat paint, which does not require a primer coat. The process of the coating was simple. First, the pipe was cleansed with an organic solvent. Generally, objectives of the coating is prebaked in a kiln in the next process. However, we skipped this process in this time. We also skipped the process of abrasive blasting, which was also conducted in general coating processes. The reason why we skipped abrasive blasting is that the inner diameter of the pipe is so small that we could not blast. Then, the pipe was fixed vertically and the PTFE was supplied as a liquid paint, which was introduced into the pipe. Next, the pipe sintered in a kiln at about 600 K. The thickness of the layer was varied by pouring the paint into the pipe repeatedly. Four pipes were processed using this procedure for each thickness. The three pipes that were not used in the experiment were cut off, and the thickness of the layer of PTFE in them was measured using magnified images of their cross sections.
118
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
Fig. 3. Schematic diagram of the experimental setup.
The enlarged view is shown in Fig. 4. A pipe was cut off at three cross sections. The results of the measurements of the thickness are shown in Table 1. In this table the symbol L means the cross section at 20 mm from the left edge, the symbol C means the center of the pipe, and the symbol R means the cross section at 20 mm from the right edge, respectively. Although the paint was poured into the pipe, no effect of gravity on the thicknesses was observed. Eventually, the layer was coated with three different thickness. If all the thicknesses are averaged, the thicknesses are 25.8, 56.8 and 93.7 lm, respectively. However, in this study the thermocouple is located on the center of the pipe. Therefore, we evaluated the thickness of the layer using the values at the center. In this case, the thicknesses are 22.5, 63.2 and 91.4 lm, respectively. We use these values in this paper. 2.3. Measurements and data reduction In this experiment, the main measurement items were those of the temperature of the pipe, the pressures in the tank and at the outlet of the test section, and the mass flow rate. The temperature was measured by a T-type thermocouple, and the measurement was used to calculate the inner wall temperature as shown below [4]:
2 !! r 2o ri ri dT o Ti ¼ T0 þ 1 2 ln 4a ro ro dt 2 r2o r 2i r2o r 2i d T o 1 ri r 4o ri 4 4 þ ðr 5r Þ ln ln þ o 64a2 i 8a2 r o 16a2 ro 16a2 dt2
The heat flux on the inner wall can be calculated as shown below:
2 ! 2 r i r2o dT o ðqcÞ2 r 3i r 4o r 2o ri r i d To qi ¼ qc þ ln 2 2r i dt k 16 16r i 4 ro dt 3 r2 r3 ri r4 ri ri d T o ðqcÞ3 r 5i 3r 4 r i 3r2 r 3 r6 þ 2 o þ o i o o i ln o ln 384 128 128 384r i 32 ro 32 r o dt3 k
The pressures were converted into electrical signals by a pressure transducer (KYOWA PG-2KU). The mass flow rate was measured by a mass flow meter (Azvil CMS0200). All the measurement items used an analogue-to-digital converter, and the results were
Fig. 4. Layer of low thermal conductivity on the inner surface.
recorded at a sampling rate of 20 Hz. The uncertainties in each measurement are shown in Table 2, and were calculated according to literature provided by Moffat [23]. The point in time when the temperature of the pipe reached 290 K is defined as 0 s in our graphs below. 3. Results 3.1. Without the insulating layers First, the experiments were conducted four times under the same conditions (tank pressure of 150 kPa, no insulating layer), and the temperatures variations were obtained; these are shown in Fig. 5. Here, the time needed for the chilldown process is defined as the time until the temperature variation per second becomes less than 0.1 K s1. When the process was conducted at a constant tank pressure, a variation in the time taken for the chilldown process to finish occurs; this is believed to be due to an unstable flow with the oscillation forms. The precision index of the time taken for this process was calculated from the four experiment, and was found to be 16 s, as shown in Fig. 5. In Fig. 6, the temperature variations of the pipes without the layers of low thermal conductivity are shown for tank pressure of 120, 135, 150, and 170 kPa. The experiments were conducted several times at each tank pressure. The temperature histories that were closest to the average of the time taken for the chilldown process at each pressure are shown in Fig. 6. From this, we can see that the time decreased as the tank pressure increased. Fig. 7 presents the time taken for the chilldown process against the four tank pressures. The length of the error bars in this figure indicates the precision index, which means the deviation of the data [23]. As the pressure in the tank increased from 120 kPa to 170 kPa, the time taken for the process decreased from 329 s to 115 s. The precision index of the time taken for the chilldown process decreased as the tank pressure increased. In Fig. 8, the pressure downstream of the test section is shown for a tank pressure of 150 kPa, along with the temperature variation of the pipe. The tank pressure was kept relatively constant through adjustments made by the regulator. It can be seen that the downstream pressure oscillated strongly, and at the beginning exceeded that of the tank pressure. Fig. 8(b) shows the downstream pressure variation from 45 to 55 s. The pressure oscillated with an amplitude of up to 15 kPa at a frequency of 2 Hz. The oscillations converged as the temperature of the pipe decreased. This oscillation is thought to be a densitywave oscillation. A number of studies have looked into these [24–26], and so the mechanism governing this oscillation will not be discussed in this paper. In Fig. 9, the mass flow rate variation for each tank pressure is shown. At higher tank pressures, even if the tank pressure remained constant, the mass flow rate increased as time passed. This is because pressure loss decreased with decrease of void
119
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122 Table 1 Thickness of the layer and its precision index [23] (unit: lm). Number of times of coating
Cross section
Sample 1
Sample 2
Sample 3
1
L C R L C R L C R
49.5 34.3 30.3 45.0 77.3 64.7 68.2 86.7 101.5
19.1 22.0 25.7 54.6 62.3 52.6 103.8 91.8 105.2
18.2 11.4 21.5 51.6 50.1 52.7 104.9 95.7 85.9
2
3
Average
Precision index
22.5
11.5
63.2
13.6
91.4
4.5
Table 2 Uncertainties in the measurements made. Parameter
Uncertainty
Pressure Mass flow rate Temperature
±1.5 kPa ±0.05 g/s 3K
Fig. 7. Time taken for the chilldown process against tank pressure variation.
Fig. 5. Repeatability of the experiments under the same condition at the tank pressure of 150 kPa.
Fig. 10 shows the chilldown curves calculated from the temperature histories. Although it is difficult to distinguish a difference in the graph, the heat flux increased alongside the tank pressure during the film boiling. The difference in the heat flux in this regime resulted in the difference observed in the time taken for the chilldown process. During the regime from the MHF point to the end of the chilldown process via the nucleate boiling regime, the higher the tank pressure became, the larger heat flux became. At lower tank pressures, the heat flux oscillated and the CHF point was unclear. The temperature of the MHF point varied from 124 to 130 K, which indicates that the effect of the tank pressure on the temperature of the MHF point is small.
3.2. With the insulating layer
Fig. 6. Temperature variation of the test section with four different tank pressures.
fraction as the pipe cooled down. The total mass at each tank pressure, which is the integrated mass flow rate, will be shown in Fig. 13 along with the results of the total mass obtained by the layer of low thermal conductivity. Nevertheless, it can be seen from Fig. 9 that the mass flow rate increased as the tank pressure increased, although the time for the chilldown process decreased. As a result, the total mass was least at the tank pressure of 170 kPa.
Fig. 11 shows the temperature variation of the metal surface of the pipe that is covered by layer of low thermal conductivity. Here, the vertical axis does not mean the surface temperature of the layer, but rather the surface temperature of the stainless tube. The tank pressure for this experiment was set to 150 kPa. Note that the temperature histories that were closest to the average of the time taken for the chilldown process are shown in Fig. 11. Although the behavior during the film boiling regime hardly changed, the layer enabled the transition in the boiling regimes to start earlier. After the transition, the temperature steeply decreased. The temperature variation of the pipe with the layer has a gentler slope after the start of the transition boiling than that of the pipe without the layer. In Fig. 12, the effect of the thickness of the layer on the time taken by the chilldown process is shown at each tank pressure. The error bars in this figure indicate the mean dispersion (or precision index) of the experiments, and these were obtained by conducting the experiments several times over under the same conditions. The dispersion got smaller as the tank pressure
120
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
(a) From the start to the end of the test.
Fig. 10. Chilldown curves conducting without an insulating coating for tank pressures of 120, 135, 150 and 170 kPa.
(b) Pressure downstream of the test section from 45 to 55 s.
Fig. 11. Temperature profile for pipes with and without the layer of low thermal conductivity.
Fig. 8. Time variation of the downstream pressure. The tank pressure was 150 kPa.
Fig. 9. Variation of the mass flow rate with tank pressure.
increased. Some error bars overlap with the symbols and can be therefore not visible. It is clear from Fig. 12 that the time needed for the chilldown process was drastically reduced by the insulating layer. In Fig. 12 (b), the time is normalized at each tank pressure by the time taken for the experiments conducted without the layer. Although it seems the variations in tank pressure made a difference in the time taken, overall the time was reduced by 40–60%. There was no significant difference between the pipes that had a layer of thickness
of 63 lm and 91 lm. Previous studies conducted on pool boiling reported that the reduction in the time taken for the chilldown process to occur plateaued at a certain thickness of the layers as the thickness of the layers increased [14]. This was observed because the effect of the insulating layer as heat resistance exceeded the effect to reduce the time taken until the transition boiling began. In Fig. 12(a), the time for the thickness of 91 lm and at 120 kPa is 129 s. The time without a coating (0 lm) and at 170 kPa is 115 s. These times are similar, which means that the increase in the flow rate (or Re number) provides the similar effect on the chilldown time. Generally, the heat flux of the wall increases with the Re number in the film boiling regime. Consequently, the chilldown time seems to decrease. We need more detailed investigation on this point in the future. In Fig. 13, the total mass of LN2 consumed until the termination of the chilldown process, as determined through the integration of the mass flow rate, is shown. It is clear that the total mass of LN2 used was reduced by the layer of low thermal conductivity. As in Fig. 12, there was no large difference between the results obtained by the thicknesses of 63 and 91 lm. The rate of the reduction in chilldown time, as shown in Fig. 13 b), is larger at the lower tank pressures, where the total mass of LN2 consumed is relatively high. The rate of reduction reached its maximum when the thickness was 91 lm and the tank pressure was 135 kPa. The total mass under these conditions was 35% of that obtained without the layer. Fig. 14 shows the chilldown curve when the insulating layer was used. The tank pressure used was 150 kPa. The heat fluxes during the film boiling regime were slightly lower when the layer was
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
(a) Chilldown time
(a) Total mass
(b) Normalized chilldown time
(b) Non-dimensionalized total mass
Fig. 12. Effect of the coating thickness on the time taken for the chilldown process.
used than when it was not, though this difference is smaller than the uncertainty. The temperature at the MHF point increased as the thickness increased: while the temperature at the MHF point was about 135 K without the layer, it was 160 K with a 23 lm layer and 185 K with the layers of 63 and 91 lm. The heat flux at the CHF point with the layer drastically decreased compared to that without the layer. The heat flux decreased as the thickness increased. The heat flux at the CHF point for a thickness of 91 lm was half of the value it had for a thickness of 63 lm. However, the proportion of time for the transition and nucleate boiling as compared to the total time for the whole chilldown process was small. Therefore, the heat flux after the transition have little impact on the overall chilldown time. Consequently, the time needed for the chilldown process was slightly longer for the scenario where the layer thickness was 91 lm as compared to the case where the thickness was 63 lm. As regards the decrease in the heat flux at the CHF point with the layer, Chandratilleke et al. [27] pointed out that there are two factors to decrease the heat flux. One is the low thermal conductivity of the layer. Generally, low thermal conductivity layers decrease heat flux into a wall. The other is the difference of nucleation sites of the surface. At transition and nucleate boiling regime, the increase in the number of the nucleation sites causes the increase in the heat flux. The surface of PTFE is smoother than that of the metal, which means less nucleation sites. These two factors decrease the heat flux. In previous studies on the paradox of the insulating layer, although some researchers indicated only temperature profiles [13,16,21], the gradients of the temperature profile at the transition and nucleate boiling regime decreased with the low
121
Fig. 13. Total mass of LN2 as calculated by integrating the mass flow rate.
Fig. 14. The chilldown curves for scenarios where the layer of low thermal conductivity was, and was not, used. The tank pressure was 150 kPa.
thermal conductive layer, which means the decrease in the heat flux. However, for example, the heat flux did not decrease in the experiment using liquid helium by Chandratilleke et al. [19]. Further studies are needed to explain these phenomena. Fig. 15 shows the time variation of the heat flux and the amount of heat transferred from the pipe. There is no difference between the heat fluxes at the film boiling regime. The differences exist only after the transition and nucleate boiling start. Fig. 15(b) shows the amount of heat transferred from the pipe. These values are calculated by integrating the heat fluxes. In this figure, the dotted line
122
D. Takeda et al. / International Journal of Heat and Mass Transfer 111 (2017) 115–122
a small proportion of the overall time taken for the chilldown process. In fact, because the layer increased the temperature of the MHF point, we saw a decrease in the overall time needed for the chilldown process. Acknowledgements This study has been conducted in the collaborative research with Japan Aerospace Exploration Agency titled ‘‘Research on the technology for providing hydrogen fuel to aircraft”. We are grateful to Prof. Hidetoshi Ohkubo for providing useful comments. References
(a) Time variation of the heat flux with and without the layer.
(b) Time variation of the amount of heat transferred from the pipe. The dotted line is the total amount of heat needed to cool the pipe. Fig. 15. Time variation of the heat flux and the amount of heat transferred from the pipe.
indicates the total amount of heat needed to cool the pipe, which is obtained with the mass of the pipe, the specific heat and the temperature difference from the start to the end of the experiment. These figures show that the total amounts of heat transferred are almost the same. We can understand that the timing where the transition boiling starts determines the time needed for the chilldown. 4. Conclusions In this study, a method for reducing the time and total mass of cryogenic fluid needed for the chilldown process was experimentally investigated. Our setup involved an arrangement of pipes, and these were coated with Polytetrafluoroethylene; this was because the resin has low thermal conductivity. Our experiment used LN2, and the pressure of the two phase flow of this substance oscillated even though it was being supplied at a constant pressure by a storage tank. Experiments were conducted using stainless pipes, and three different thicknesses of the Polytetrafluoroethylene layers were used. Our results show that, the greater the thickness of this layer, the higher the temperature at the minimum heat flux (MHF) point becomes. These experiments also revealed that the layer enables transition boiling to start earlier. Consequently, the total mass of LN2 consumed was reduced by a maximum of 64%. The heat flux decreased after reaching the MHF point, though this was, overall,
[1] B. Khandelwal, A. Karakurt, P.R. Sekaran, V. Sethi, R. Singh, Hydrogen powered aircraft: the future of air transport, Prog. Aerosp. Sci. 60 (2013) 45–59. [2] H. Hu, J.N. Chung, S.H. Amber, An experimental study on flow patterns and heat transfer characteristics during cryogenic chilldown in a vertical pipe, Cryogenics 52 (2012) 268–277. [3] L. Evans, The large hadron collider, Annu. Rev. Nucl. Part Sci. 61 (2011) 435– 466. [4] R. Shaeffer, H. Hu, J.N. Chung, An experimental study on liquid nitrogen pipe chilldown and heat transfer with pulse flows, Int. J. Heat Mass Transfer 67 (2013) 955–966. [5] K. Yuan, Y. Ji, J.N. Chung, Cryogenic chilldown process under low flow rates, Int. J. Heat Mass Transfer 50 (2007) 4011–4022. [6] Y. Shirai, H. Tatsumoto, M. Shiotsu, K. Hata, H. Kobayashi, Y. Naruo, Y. Inatani, Boiling heat transfer from a horizontal flat plate in a pool of liquid hydrogen, Cryogenics 50 (2010) 410–416. [7] J. Hartwig, H. Hu, J. Styborski, J.N. Chung, Comparison of cryogenic flow boiling in liquid nitrogen and liquid hydrogen chilldown experiments, Int. J. Heat Mass Transfer 88 (2015) 662–673. [8] I.U. Vakarelski, N.A. Patankar, J.O. Marston, D.Y.C. Chan, S.T. Thoroddsen, Stabilization of Leidenfrost vapour layer by textured superhydrophobic surfaces, Nature 489 (2012) 274–277. [9] M. Shojaeian, A. Kosar, Pool boiling and flow boiling on micro- and nanostructured surfaces, Exp. Thermal Fluid Sci. 63 (2015) 45–73. [10] H. Hu, C. Xu, Y. Zhao, R. Shaeffer, K.J. Ziegler, J.N. Chung, Modification and enhancement of cryogenic quenching heat transfer by a nanoporous surface, Int. J. Heat Mass Transfer 80 (2015) 636–643. [11] T. Uehara, T. Inoue, Quenching process simulation of Japanese sword covered with clay, J. Soc. Mat. Sci., Japan, 44 (498) (1995) 309–315. (in Japanese). [12] C.W. Cowley, W.J. Timson, J.A. Sawdye, A method for improving heat transfer to a cryogenic fluid, Adv. Cryog. Eng. 7 (1962) 385–390. [13] J.P. Maddox, T.H.K. Frederking, Cooldown of insulated metal tubes to cryogenic temperatures, Adv. Cryog. Eng. 11 (1966) 536–545. [14] L.D. Allen, A method of increasing heat transfer to space chamber cryopanels, Adv. Cryog. Eng. 11 (1966) 547–553. [15] A.P. Butler, G.B. James, B.J. Maddock, W.T. Norris, Improved pool boiling heat transfer to helium from treated surfaces and its application to superconducting magnets, Int. J. Heat Mass Transfer 13 (1) (1970) 105–115. [16] F. Moreaux, J.C. Chevrier, G. Beck, Destabilization of film boiling by means of a thermal resistance, Int. J. Multiph. Flow 2 (2) (1975) 183–190. [17] S. Nishio, Study on minimum heat-flux point during boiling heat transfer on horizontal plates: effects of transients and thermal conductance of wall, Jap. Soc. Mech. Eng., Ser. B 51 (462) (1985) 582–590 (in Japanese). [18] S. Nisio, Y. Serizawa, Control of minimum heat flux point temperature by the thermal conductance of an additional surface layer, Jap. Soc. Mech. Eng., Ser. B 53 (487) (1987) 1061–1064 (in Japanese). [19] G.R. Chandratilleke, S. Nishio, Pool boiling heat transfer to saturated liquid helium on coated surfaces, Cryogen. Superconduct. Soc. Japan, 23(3) (1988) 128–133. [20] Y. Kikuchi, T. Hori, I. Michiyoshi, Minimum film boiling temperature for cooldown of insulated metals in saturated liquid, Int. J. Heat Mass Transfer 28 (6) (1985) 1105–1114. [21] A.N. Tsoi, A.N. Pavlenko, Enhancement of transient heat transfer at boiling on a plate surface with low thermoconductive coatings, Thermophys. Aeromech. 22 (6) (2015). [22] G.A. Dreitser, Heat transfer enhancement in channels for film boiling of cryogenic liquids, Appl. Therm. Eng. 25 (2005) 2512–2521. [23] R.J. Moffat, Describing the uncertainties in experimental results, Exp. Thermal Fluid Sci. 1 (1988) 3–17. [24] J.A. Boure, A.E. Bergles, L.S. Tong, Review of two-phase flow instability, Nucl. Eng. Des. 25 (1973) 165–192. [25] S. Kakac, B. Bon, A Review of two-phase flow dynamic instabilities in tube boiling systems, Int. J. Heat Mass Transfer 51 (3–4) (2008) 399–433. [26] L.C. Ruspini, C.P. Marcel, A. Clausse, Two-phase flow instabilities: a review, J. Heat Mass Transfer 71 (2014) 521–548. [27] G.R. Chandratilleke, A. Sato, Heat transfer characteristics of cooling, Low Temp. Technol. 28 (5) (1993) 244–252 (in Japanese).