Applied Thermal Engineering 130 (2018) 343–353

Contents lists available at ScienceDirect

Applied Thermal Engineering

journal homepage: www.elsevier.com/locate/apthermeng

Research Paper

Experimental heat transfer analysis of a cryogenic nitrogen pulsating

heat Pipe at various liquid fill ratios
Luis Diego Fonseca ⇑, Franklin Miller, John Pfotenhauer
Department of Mechanical Engineering, University of Wisconsin-Madison, 1500 Engineering Drive, Madison, WI 53706, USA

h i g h l i g h t s

A nitrogen based PHP has been operated at temperatures between 77 K and 80 K.

Various fill ratios below 40% and heat loads below 4.5 W have been applied.

A minimum startup heat load of 2 W is required to operate the PHP successfully.

Results show a maximum thermal conductivity of 70 kW/m/K at a fill ratio of 20%.

a r t i c l e i n f o a b s t r a c t

Article history: The experimental results of extensive testing of a cryogenic Pulsating Heat Pipe (PHP) using nitrogen as
Received 23 November 2016 the working fluid operating between 77 K and 80 K is presented. Fifty-two different test runs were ana-
Revised 24 August 2017 lyzed over a range of different fill ratios, heater input powers, and orientations in order to map thermal
Accepted 5 November 2017
performance as a function of the system operating parameters. The temperature difference between the
Available online 8 November 2017
evaporator and condenser section were obtained using platinum resistance thermometers (PRTs), and a
pressure transducer was used to record the fluid’s pressure oscillations. As opposed to other N2 pulsating
heat pipes which use less than 16 parallel tubes, this experimental PHP consists of 40 parallel tubes. The
PHP was operated at different liquid fill ratios ranging between 10% and 40% and heat loads between 1 W
to 4.5 W. Results show that the PHP’s temperature difference between section are small and resulted in
high effective thermal conductivities up to 70,000 W/m/K at fill ratios around 20%. Also, it was shown the
PHP stopped oscillating at heat loads below 1.5 W but operated successfully above 2 W. Results from
experiments conducted in both horizontal and vertical orientations show that gravity has a significant
effect on the thermal performance of the PHP even with a high number of turns.
Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction an entire SCM system or any extensive surface area since its cool-
ing region is localized.
1.1. Motivation For these reasons companies are focusing on effective alterna-
tive cooling mechanisms such as pulsating heat pipes (PHPs) that
Devices such as MRIs use liquid cryogens to cool down their may be connected to a single cryocooler and distrubute its cooling
superconducting magnets (SCM). Cryocoolers are used in these power, maintain a uniform temperature distribution, mitigate hot
systems to minimize any boil-off of the cryogens, but still the cost spots and thermal gradients.
per year of refilling the MRIs are an important factor to consider. Space agencies such as NASA are also interested in absorbing
For example, as shown in Fig. 1, 22% of helium was consumed heat loads from multiple locations and rejecting heat to one cry-
solely for MRI magnets in 2011 [1]. Currently, liquid helium costs ocooler. Potential space applications are:
around $9.5 per liter but can reach as high as $22 per liter [1]. Gen-
eral Electric (GE) allocated 6 million liters per year to service its
Cooling heat shields of large cryo-propellant tanks at multiple
MRI magnets at hospitals and other sites in 2012 [2]. Generally, locations which would replace current systems which utilize
another disadvantage is that a cryocooler cannot alone cool down magnetic stir-bars to distribute heat loads throughout the cryo-
gen. For proposed manned missions to Mars, large amounts of
liquid cryogens will be necessary in order to travel the 67.8 mil-
⇑ Corresponding author. lion mile round-trip to the red planet. Current systems which
E-mail address: [email protected] (L.D. Fonseca). utilize extreme insulation and stir-bars would be vastly ineffec-

https://doi.org/10.1016/j.applthermaleng.2017.11.029
1359-4311/Ó 2017 Elsevier Ltd. All rights reserved.
344 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353

Nomenclature

Ac total cross-sectional area (m2) Subscripts

Dcrit critical diameter (m) amb ambient condition
fliq fill ratio (–) cond condenser section
g acceleration due to gravity (m/s2) evap evaporator section
k thermal conductivity (W/m-K) eff effective
L length (m) final final value of a variable
N number of parallel tubes (–) PHP part of PHP
Ntotal total mass in moles P pressure
P pressure (Pa) Q heat load
Q_ heat load or power applied (W) T temperature
R gas constant (J/mole-K) tank part of supply tank
T temperature (K) l liquid phase
T average temperature (K) v saturation phase
V volume (m3) gas-lines part of gas lines
r surface tension (N/m2) sat saturation
q density (kg/m3 or mole/m3)
q average density (kg/m3 or mole/m3)
m uncertainty

Fig. 1. Helium Consumption 2011.. Source [1]

tive for missions of such magnitude. Thus, NASA is interested in

pulsating heat pipes as the core component of a revolutionary Fig. 2. Simple PHP schematic of a pulsating heat pipe.
thermal management system. Data indicate that the PHPs have
operated successfully, however further investigation is required
to understand how these systems behave under various operat- pressure would equilibrate with the supply tank. This valve open
ing conditions. case resulted in fill ratios between 27% and 46% and achieved a

In addition, PHPs could be used to cool infrared (IR) optics maximum effective thermal conductivity of 35,000 W/m-K at a
which are currently used in detectors. IR detectors need to be heat load of 3.5 W and fill ratio of 27%. The data presented and ana-
at cryogenic temperatures and instead of using copper straps lyzed in this paper is for the case when the valve is closed. An
to connect these detectors to a cryocooler PHP’s could transfer extensive set of 52 test runs were performed. In contrast to only
the heat with lower system mass because the work at much 4 test runs with the valve open. For cases where the valve is closed,
higher effective thermal conductivities. the PHP had higher effective thermal conductivities. The pulsating
heat pipe described in this work consists of 40 parallel tubes with
Results from this PHP apparatus were published in a previous an inner diameter of 0.5 mm and an adiabatic length of 80 mm
paper [10], however results of further testing are reported in this when valve is closed. Most of the experiments were conducted at
paper. In the previous published paper [10] the data was collected lower fill ratios in order to achieve higher effective thermal con-
when the PHP valve was open, as shown in Fig. 2; therefore the ductivities since high fill ratios resulted in lower thermal
PHP’s capillary tubing, gas-lines and supply tank were directly con- performance.
nected to each other. With the PHP valve open the experiments In order to achieve a slug/vapor flow it is necessary that the sur-
were able to verify that cryogenic PHPs can be successfully oper- face tension forces be larger than the gravity forces. This criterion
ated in this form, however the paper did not show if the PHP oper- leads to, a critical diameter of the tube, Dcrit [1]
ates in an optimal behavior by having the valve closed. A main rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
r
advantage of operating the PHP with the valve open is if the cry- Dcrit 6 2 ð1Þ
ocooler and relief valve malfunctions. In such a case, the PHP’s gðql  qv Þ
 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353 345

where g (m/s2) is the acceleration due to gravity and r (N/m), ql and rpm and 4000 rpm. Results show that when the fan operated at
qv (kg/m3) are the surface tension, liquid density and vapor density 2000 rpm the CEOHP was able to dissipate a maximum heat load
properties of the fluid respectively. Fig. 3 shows the critical diame- 34.4 W which maintained the chip at a working temperature of
ters as a function of saturation pressure for different cryogens such 54 °C. While at 4000 rpm, the maximum heat load dissipated was
as Helium, Hydrogen, Nitrogen, Neon and Argon. Nitrogen shows a 44.5 W and maintained a chip temperature of 53 °C. This successful
maximum Dcrit around 2 mm. For this experiment a diameter of 0.5 testing stated that this CEOHP module had a better performance
mm was chosen to ensure that Eq. (1) was satisfied and also allow than conventional heat sink
the use of other cryogens in the future. In addition, investigation using other fluids at different fill
ratios have been a priority for researchers to characterize and pin
1.2. Background point the maximum heat that can be transferred. For example,
Bhawna Verma et al. [6] tested a PHP using methanol and de-
Since PHPs exhibit high heat transfer rates they can be useful ionized water as the working fluids. They reported that for the DI
for many applications involving room temperature cooling sys- water case, the minimum thermal resistance achieved was 0.492
tems. For example, PHPs could be used to cool down photon detec- K/W (2.03 W/K) at a filling ratio of 50% and at a heat load of 100
tors since they can provide a bridge from the detectors to the cold W. While using methanol as the working fluid, the PHP reached a
sink instead of using a copper thermal buss. The main advantage in minimum thermal resistance of 0.51 K/W at a fill ratios of 40%
replacing the thermal copper busses with PHPs would be to mini- and heat load of 100 W. Patel et al. [7] tested more working fluids
mize mass constraints and lower temperature differences between such as dionized water, ethanol, methanol, acetone and including
the evaporator and condenser section. Recently, The Japanese water based mixtures with ethanol, methanol and acetone. A heat
Aerospace Exploration Agency (JAXA) has tested two room temper- load between 10 W and 110 W was applied at a 50% fill ratio.
ature PHP prototypes [4] in order to cool down their photon detec- Results showed a thermal resistance down to 0.95 k/W for water
tors. These detectors will be used in the balloon-borne General and 0.64 k/W for acetone at an average heat load of 57.3 W. Faghri
Anti-Particle Spectrometer (GAPS). These two prototypes have a et al. [8] and Cui et al. [9] summarized work done on room temper-
U-shaped and O-shaped configuration and have a total length of ature PHPs/OHPs experiments.
6 and 8 m, respectively. They are longest room temperature PHPs For the last ten years, few cryogenic pulsating heat pipes have
reported yet. For the U-shaped configuration, the condenser, evap- been reported in the literature [10–15,17,18] compared to the
orator and adiabatic section were 2 m long each and with a total of room temperature studies mentioned in [7,9]. Results from four
16 loop turns, while the O-shaped configuration has two parallel experiments using nitrogen as a working fluid have been published
adiabatic section. Another characteristic is that these PHPs have to date [3,10–12,18] and are compared extensively to this experi-
check valves installed at every other turn in order to guarantee mental apparatus in the Results section. This experiment has a
fluid circulation throughout the capillary tubing. This PHP used smaller diameter capillary tubing and these tubes are attached to
R410A as the working fluid and was built with copper capillary a cylindrical evaporator and condenser section, which provides
tubing with an inner and outer diameter of 1 mm and 1.6 mm, the following advantages:
respectively. The O-shaped PHP performed best and retrieved a
total of 250 W. GAPS requires 800 W of cooling capacity therefore
All cryogens can be tested using a capillary tubing with inner
four O-shaped PHPs will be necessary. diameter of 0.5 mm.
One of the most important application for room temperature
A total of 40 parallel tubes were installed. Other nitrogen PHP
PHPs are to cool down electronics components such as CPU Chips. experiments listed above have around 10–16 parallel tubes.
For this reason, Rittedech et al. [5] tested a Closed-end Oscillating More tubes provide more pressure perturbations throughout
Heat Pipe (CEOHP) which was designed to cool a Pentium 4 CPU, the PHP. Hence a start-up oscillation behavior can be easily
model SL 6 PB 2.26 GHz, with an electric power of 58 W. The achieved.
CEOHP used R134a as the working fluid and operated at a fill ratio
Cylindrical shape allows a more simplified installation inside a
of 50%. The condenser section was air cooled using a fan set 2000 vacuum dewar and easy to attach it to a cryocooler.

4.5
Helium
4 Hydrogen
Nitrogen
3.5 Neon
Argon
3
D crit (mm)

2.5

1.5

0.5

0
100 150 200 250 300
P sat (kPa)

a) b)
Fig. 3. (a) Critical diameter versus (a) saturation pressure (left) and (b) saturation temperature (right).
346 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353

2. Experimental apparatus and PHP design

Fig. 4 shows the main components of the test apparatus. The

PHP is mounted inside a 20 cm diameter by 33 cm long vacuum
chamber.
As shown in Fig. 5, the PHP is divided into three major sections:
the evaporator, adiabatic and condenser sections. The evaporator
section was constructed of 110 copper machined into a wall cylin-
der that has an inner diameter of 50 mm, an outer diameter of 65
mm and a length of Le = 70 mm [3]. In addition, the evaporator sec-
tion includes a 50-Ohm electric heater that acts as the thermal
load. The condenser section was also constructed from this wall
cylinder made of 110 grade copper with an inner diameter of
25.4 mm, an outer diameter of 65 mm and a length of Lc = 70
mm. As shown in Fig. 6 we can calculate the DT, from one end to
the other of the evaporator/condenser section, using Fourier’s Fig. 5. PHP Core Dimensions.
law while assuming a constant thermal conductivity of copper k
= 553 W/m-K at 77 K. Hence all temperature differences for the
condenser and evaporator are shown in Table 2.1 at different heat
loads. The importance of this analysis is to show that the thick-
nesses chosen for these sections result in very small temperature
differences allowing to attach only one temperature sensor on each
section and assume a uniform temperature difference and to fur-
ther calculate the effective thermal conductivities.
The condenser section is thermally anchored to a Sunpower CTÒ
cryocooler that can maintain the fluid’s saturation temperature
from 64 K to 126 K. The capillary tubing was soldered to both the
evaporator and condenser sections; this tubing has an inner diam-
eter of 0.5 mm and an outer diameter of 0.8 mm. These capillary
tubing were bent using 4 mm shoulder screws and are equally
spaced around the perimeter, resulting in a total of 20 loops at each
end. More details of the final construction are detailed in [3]. Fig. 6. Change in Temperature for evaporator/condenser sections.

Fig. 4. Test Apparatus.

 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353 347

Table 2.1 sors and pressure sensors were used to calculate the liquid volume
Condenser/Evaporator DeltaTs. fraction as described in Section 4.
Q (W) DT Condenser (K) DT Evaporator (K)
0.05 2.25E-03 4.67E-03
0.1 4.50E-03 9.34E-03
4. Experimental procedure
0.5 2.25E-02 4.67E-02
1 4.50E-02 9.34E-02 4.1. Purging process
1.5 6.75E-02 1.40E-01
2 9.00E-02 1.87E-01
As shown in Fig. 8-b, the 3.8 L gas supply tank (blue1) contains
2.5 1.13E-01 2.34E-01
3 1.35E-01 2.80E-01 the necessary pressure to fill the PHP’s capillary tubing. If necessary,
3.5 1.58E-01 3.27E-01 the gas supply tank can be refilled by opening the reservoir valve
4 1.80E-01 3.74E-01 which is connected directly to the gas reservoir. Before turning on
4.5 2.03E-01 4.20E-01 the cryocooler it is very important to clean the capillary tubing to
any impurities that could freeze and obstruct the system. Therefore
a purging process is initiated, ultra-high purity nitrogen gas is
As shown in Fig. 4 for component (5), a 0.5 mm thick copper injected by opening the supply valve and filling all the gas-lines
thermal jacket with outer diameter 141 mm and a length 305 leading to the PHP. Afterwards the supply valve is closed and the
mm encloses the PHP core in order to reduce the parasitic radiation purge valve is slowly opened in order to evacuate all the gas using
from the vacuum chamber walls. This thermal jacket was also ther- the pumping system which consists of a turbo-molecular pump
mally anchored to the CT cryocooler in order to provide an envi- and a mechanical pump in series. The turbo pump should be turned
ronmental surrounding temperature of approximately 80 K for off initially until the mechanical pump reaches a pressure of 102
the experiment. A 14 layer MLI blanket covers the thermal jacket torr. In addition, the vacuum valve should always be closed during
and an additional 7 layer MLI blankets were placed at the inner this process to avoid pressuring the vacuum chamber (dewar).
surface of the vacuum chamber. A parasitic radiation heat load Finally, the purge valve is closed and the purging process is repeated
from the dewar to the thermal jacket was estimated to be 174 five times.
mW by creating a simple thermal resistance network model based
on residual gas, number of layers and boundary temperatures. This
MLI model was based on equations presented in [16] where the 4.2. PHP operation
vacuum pressure, number of MLI Layers, and boundary tempera-
tures were assumed to be 106 torr, 21 layers, and 300 K and 80 The following tasks were performed in order to reach a stable
K, respectively. PHP operation after the purging process:

1. Fill the gas supply to the necessary pressure and afterwards

3. Measurements close supply valve.
2. Evacuate all remaining gas inside the capillary tubing and gas-
A total of 5 lakeshore platinum RTDs, model PT-103, were used lines using the pumping system: PHP valve open, purge valve
to measure the temperatures within the experimental apparatus. A open and vacuum valve closed.
copper casing was built to hold them in place. As shown in Fig. 7, 3. Once the PHP pressure sensor measures 106 torr, the purge
the thermometers were placed at the (1) cold end of the cry- valve was closed and vacuum valve opened.
ocooler, (2) the thermal bus, (3) center of the condenser section, 4. Open the supply valve and turn on cryocooler, PHP valve should
(4) center of the evaporator section and (5) the top endplate of still remain open.
thermal jacket. As shown in Table 2.1, the temperature throughout 5. Close the PHP valve once a desirable pressure it met as gas is
the evaporator and condenser section can be assumed to be uni- condensed inside the PHP. It will take about 10 h for the cry-
form, therefore no other temperature sensors were necessary. ocooler to reach a steady state temperature of 80 K.
Two pressure transducers, Endevco model 8510B-500, were 6. Apply a heat load to the evaporator section and run experiment.
used to measure the pressure of the nitrogen supply tank and
the oscillatory pressure of the evaporator section. As shown in Fig. 9 shows the cooldown time of the experiment. For this case
Fig. 8-a and -b, the first pressure sensor is located at the top of the gas tank was initially filled at a pressure of 170 kPa and
the 3.6 L nitrogen supply tank and the second sensor is located decreased to 140 kPa once the condenser temperature reached
between the PHP valve and evaporator section. Temperature sen- 80 K. The Sunpower CT Cryocooler took around 10 h to reach a
steady state temperature from 300 K to 80 K while the evaporator
temperature finally reached an equilibrium state condition in 12 h.

4.3. Steady state condition measurements

Fig. 10 shows the different transient and equilibrium states of

the condenser and evaporator temperatures of the PHP when a
heat load of 2.5 W and 2.8 W were applied successively. It can be
noticed that for these cases an equilibrium state was achieved after
1 h. The evaporator and condenser section have masses of 1.155 kg
and 1.818 kg respectively, clearly decreasing these masses would
decrease drastically the transient state time. The data of the equi-
librium state are retrieved and used to get the average condenser

1
For interpretation of color in Fig. 8, the reader is referred to the web version of
Fig. 7. Temperature Sensor Locations. this article.
348 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353

Fig. 8. (a) Experimental apparatus, (b) Schematic of experimental apparatus.

Fig. 9. Cooldown Time of Nitrogen PHP.

 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353 349

Fig. 10. Transient and Equilibrium State Data.

and evaporator temperatures T cond and T ev ap respectively and a

minimum of 70 min of equilibrium data is recorded.

5. Results

The effective thermal conductivity keff (W/m-K), can be calcu-

lated using the following equations [13]

Q_ Leff
keff ¼ ð2Þ
Ac ðT ev ap  T cond Þ


where Q_ is the heat load applied at the evaporator section, T ev ap and

T cond are the average temperatures (K) of the evaporator and con-
denser sections respectively. Leff is the length of adiabatic section
and finally the Ac is the total cross-sectional area of the fluid inside Fig. 11. Contour plot of Effective thermal conductivity versus fill percentage and
of the capillary tubing. heat load.
For calculating the liquid volume Vl (m3) inside the capillary
tubing, Eq. (2) can be used. Where Ntotal (Pa), is the total number
ture PHPs observed similar behavior at low and high fill ratios and
of moles of gaseous nitrogen in the supply tank before filling the
provided the following explanations for its behavior that may also
PHP. Vtank, VPHP, and Vgas-lines (m3) are the nitrogen tank, PHP and
apply to the data presented in this paper:
gas line volumes respectively. Pfinal is the average final tank pres-
sure (Pa). Tamb and Tgas-lines are the average temperatures (K) of

Low fill ratios: in this case the fluid approaches a vapor single
the tank and gas lines respectively. The densities ql,sat and qv,sat
phase, resulting in longer vapor bubbles and low amounts of
are the molar densities (moles/m3) of the saturated liquid and sat-
liquid to form many distinct plugs. Therefore a tendency to
urated vapor respectively.
dry-out is encountered in the evaporator sections.
Pfinal V tank Pfinal V gaslines
High fill ratios: the main function of vapor bubbles are to pro-
Ntotal ¼ þ þq
 l;sat V l þ q
 v ;sat ðV PHP  V l Þ ð3Þ
RT amb RT gaslines vide the necessary perturbations to drive the fluid and over-
come induced forces such as friction, buoyancy and
Finally, the liquid volume ratio fliq can be calculated using Eq. gravitational forces. At high fill ratios the working fluids
(3). approaches a liquid single phase, therefore less vapor slugs
Vl are present in the system and the probability of having instabil-
f liq ¼ ð4Þ ities is hindered.
V PHP

Optimal fill ratios: distinct vapor plugs are formed throughout
A total of 52 successful tests were performed by varying the the PHP providing the necessary pumping force to drive the
heat load from 1.5 W to 3.5 W and varying the fill ratio from 13% fluid.
to 45%. It should be noted that these tests were conducted in a hor-
izontal orientation. As shown, the fill ratio and heater power are Not all PHPs will encounter the same optimal fill ratio since the
primary parameters that influence the effective thermal conductiv- chosen fluid’s properties will have a great effect. Kandekar [19]
ity keff. Fig. 11 shows a contour plot of all three variables, where the found an optimal fill ratio of 30% with water. Here at the University
power ðQ_ Þ is on the y-axis, and the fill ratio fliq is on the x-axis. The of Wisconsin-Madison, our current helium PHP has shown an opti-
contours and colors represent the keff. It can be concluded that this mal fill ratio of 70% [21]. In addition, Deng et al. [17] determined,
PHP configuration achieves highest values of keff at high heater for their hydrogen based PHP, an optimal effective thermal conduc-
powers but at lower fill ratios (approximately 20%), where the tivity of 18.7 kW/m-K at a fill ratio of 35% and at a heating power of
maximum conductivity achieved is 70,000 W/m-K at 3.5 W. Kan- 5 W. For example, the size of the vapor bubble and liquid slug
dekar [19] who used flow visualization to observe room tempera- depends on the fluid’s density at the saturation conditions. The
350 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353

density ratio ql,sat/qv,sat, at saturation conditions of 77 K for nitro- after the first 50 min the PHP stops operating. This is evident
gen, is 182; while for helium at 4.2 K is 7.6 and for water at 373 because the temperature of the evaporator section temperature
K is 1604. Therefore, the optimal fill ratio will be dependent on starts increasing while the temperature of the condenser section
the fluid’s density. In addition, as the heat load is applied, more decreases. The slow temperature decrease after 150 min was when
vapor bubbles will be generated causing the necessary pressure the heater was turned-off. Fig. 14 shows the temperature oscilla-
perturbations to move the fluid much faster until a dry-out condi- tions with a period of approximately 5 min for both evaporator
tion is encountered. In addition, the whole PHP system is quite and condenser sections at a heat load of 2 W. Fig. 15 shows the
dependent on the cryocooler’s cooling power since the volumetric temperature change as the heat load is increased to 2.8 W. This
contraction of the PHP occurs in the condenser section. Fig. 11 graph shows that at heat loads above 2 W the oscillation periods
compiles a large quantity of experimental data into one graph to decreased drastically to 18 s and 42 s. It should be noted that at
show the dependence of thermal performance on heater power heat loads below 2 W the PHP does not consistently operate and
and fill ratio for this specific PHP configuration. oscillations can be reduced or completely stop as shown in
Fig. 11 would be different if a more powerful cryocooler were Fig. 13. Therefore, the critical heat load qcrit of 2 W is the minimum
used or if the condenser section were immersed in liquid nitrogen heat load needed to operate this PHP continuously. Kandekar [19]
which would act as an infinite heat sink; therefore, higher heat observed similar behavior in his water/ethanol PHPs and stated
loads could be applied. Table 5.1 shows a comparison of this PHP two important conclusions:
to other nitrogen ones and important characteristics are shown
such as the effective length Leff, inner diameter ID, number of par- (a) In general, start-up by a step power level is only possible
allel tubes N, maximum heat load applied Q_ , maximum effective beyond a minimum heat flux. Beyond the critical heat flux,
thermal conductivity keff and the fill ratio applied during the test- no ‘‘stop-over” was ever detected and continuous operation
ing. It should be noticed that experiments [10,11,18] were per- was always achieved.
formed when the condenser section was submerged into a (b) The thermal resistance continuously decreases with increas-
nitrogen pool, for this reason high heat loads were achieved, since ing heat input until the heat transfer gets limited by external
the nitrogen pool can be considered as an infinite heat sink. How- air-side heat transfer coefficient.
ever, the data present in [13] and this paper are experiments con- In statement b, the so called ‘‘external air-side heat transfer”
ducted with the same PHP configuration but at different fill ratios refers to the condenser section which is cooled down by an exter-
and cases when the PHP valve is opened/closed. In addition, Mito nal forced airflow. In our case our limiting cooling factor is the CT
et al. [12] used a cryocooler but achieved a maximum effective cryocooler. In addition, the sensible and latent heat are the two
thermal conductivity of 15,800 W/m-K. Another major difference main fluid properties that affect the heat transfer of the PHP. Zhang
between configurations are the number of parallel tubes used, et al. [20] stated the following:
which ranged between 6 and 16 in a rectangular evaporator/con-
denser geometry. The PHP tested in this work had a total of N =
A low latent heat will cause the liquid to evaporate more
40 tubes were connected to a cylindrical evaporator and condenser quickly at a given temperature and a higher vapor pres-
section and achieved the highest effective thermal conductivity to sure; the liquid slug oscillating velocities will be increased
date. Zhang et al. [20] stated that the heat flux of a PHP decreases and the heat transfer performance of the PHP will be
as the number of turns increases and therefore an optimal number improved.
of turns might exist that would result in a maximum heat flux. In
Because the majority of the total heat transfer in a PHP is due to
contrast, by reducing the inner diameter to 0.5 mm would cause sensible heat, a high specific heat is desirable.
the heat flux to increase and generate vapor bubbles more easily.
It can be concluded through this experiment, that the highest ther- Therefore, more vapor bubbles are generated as the heat load is
mal performance was achieved by decreasing the ID and increasing increased which will cause an increase in pressure perturbations
N with a cylindrical configuration. and flow velocities; hence higher heat transfer rates will be
Fig. 12, shows a comparison of the effective thermal conductiv- observed until dry-out conditions are reached.
ity in the horizontal and vertical orientations. The vertical orienta- As observed in Fig. 16 the pressure oscillations periods of the
tion is the case when condenser is above the evaporator. Both cases PHP are much shorter than the temperature oscillations; notice
start with the same initial charge pressures before turning on the that the oscillations are about 2 s in period and have a peak-to-
heater. These results show that the vertical orientation experiences peak pressure amplitude of 2 kPa. In addition, Fig. 17 shows a fre-
greater effective conductivities due to the presence of gravitational quency spectrum of the pressure oscillations at heat loads of 2.2 W
forces that help the liquid slugs move to the evaporator section. and 3 W, and demonstrates that most frequency peaks are
Other PHP experiments [12,14,15] have demonstrated that the between 0.2 Hz and 0.5 Hz. The frequency spectrum for the 2.2
thermal performance of cryogenic PHPs change at different inclina- W case has various visible peaks below 0.5 Hz, while for the 3 W
tion angles. case the most noticeable peaks are close to 0.5 Hz corresponding
The conclusion that higher heat loads tend to increase the fluid to the 2 s period as observed in Fig. 16. The difference between
velocity is supported by the data shown in Figs. 13–15. Fig. 13 the temperature and pressure oscillations arises from the fact that
shows the PHP operating at a low heat load of 1.5 W, it shows that

Table 5.1
Benchmark with other PHPs.

Year Reference Fluid Condenser Type Leff (mm) ID (mm) N Max Q (W) Max keff (W/m-K) Fill Ratio (%)
2008 [18] N2 Nitrogen Pool 120 1 6 120 n/a n/a
2009 [10] N2 Nitrogen Pool 75 1.65 16 380.1 40,757 50
2011 [12] N2 Cryocooler 100 2.1 10 22 15,800 17–70
2014 [11] N2 Nitrogen Pool 100 0.9 10 7 18,000 50
2015 [3] N2 Cryocooler 80 0.5 40 3.5 35,000 27–46
2016 This Paper N2 Cryocooler 80 0.5 40 3.5 70,000 10–40
 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353 351

Fig. 12. Orientation Performance, cases with valve closed.

Fig. 13. PHP operating at 1.5 W, oscillations stopped.

PHP is oscillating in a chaotic behavior at different frequencies and smaller than the value of keff. 98.6% of the uncertainty of mkeff comes
that the evaporator and condenser sections have a masses of 1.155 from the temperature uncertainty mT, and 1.4% comes from the
kg and 1.818 kg, respectively. These masses act as a thermal capac- heat load uncertainty mQ. However, the lowest and higher fill ratio
itance, attenuating the amplitude of higher frequencies. uncertainties are very similar and are mostly affected by the pres-
sure uncertainty mP; hence any temperature sensitivity is
negligible.
6. Uncertainty analysis

As shown in Table 6.1, the uncertainty of the measured values 7. Conclusion

of pressure P, temperature T and heat load applied Q_ are 2 kPa,
23 mK and 1% of Q_ , respectively. As shown in Eq. (1), the effective A nitrogen based pulsating heat pipe with a total of 40 parallel
thermal conductivity keff is dependent on temperature and heat tubes and with a cylindrical geometry has been successfully oper-
load. While the fill ratio is dependent on temperature and pressure. ated at different fill ratios when the PHP valve was closed. The
The lowest and highest uncertainties of keff and fliq (mkeff and mfliq results show that a maximum effective thermal conductivity of
respectively) are shown in Table 6.2. This values were obtained for 70,000 W/m/K can be achieved at fill ratios close to 20% and heat
those cases with an initial fill ratio fliq of 20%. The highest uncer- loads above 2 W. It was shown the PHP stopped oscillating at
tainty of keff was found to be ±5523 W/m-K which is one order 1.5 W but operated successfully above 2 W. This thermal perfor-
352 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353

Fig. 14. Temperature Oscillations at 2 W.

Fig. 15. Temperature Oscillations at 2.8 W.

Fig. 16. Pressure Oscillations at stable operation.

 L.D. Fonseca et al. / Applied Thermal Engineering 130 (2018) 343–353 353

Fig. 17. Pressure Power Spectrum: Vertical Orientation.

Table 6.1 [5] S. Rittidech, A. Boonyaem, P. Tipnet, CPU cooling of desktop PC by closed-end
Uncertainty of Pressure, Temperature and Heat Load. oscillating heat-pipe (CEOHP), Am. J. Appl. Sci. 2 (12) (2005) 1574–1577.
[6] B. Verma, V.L. Yadav, K.K. Srivastava, Experimental studies on thermal
mP (kPa) Pressure mT (mK) Temperature mQ (mK) Temperature of performance of a pulsating heat pipe with methanol/DI water, J. Electron.
uncertainty Uncertainty Heat Load Cool. Thermal Control 3 (1) (2013) 27–34.
[7] V.M. Patel, H.B. Mehta, Influence of working fluids on startup mechanism and
±2 ±23 ±1% of Q_ applied thermal performance of a closed loop pulsating heat pipe, Appl. Therm. Eng.
110 (2017) 1568–1577.
[8] A. Faghri, Y. Zhang, Thermal modeling of unlooped and looped pulsating heat
pipes, J. Heat Transfer 123 (6) (2001) 1159.
Table 6.2 [9] X. Cui, Z. Qiu, J. Weng, Z. Li, Heat transfer performance of closed loop pulsating
Uncertainty of keff and fliq. heat pipes with methanol-based binary mixtures, Exp. Thermal Fluid Sci. 76
(2016) 253–263.
Uncertainty variable Lowest Uncertainty Highest Uncertainty [10] A.J. Jiao, H.B. Ma, J.K. Critser, Experimental investigation of cryogenic
mkeff (W/m-K) ±1766 ±5523 oscillating heat pipes, Int. J. Heat Mass Transfer 52 (15) (2009) 3504–
mfliq (%) ±5.088 ±5.139 3509.
[11] Y. Li, Q. Wang, S. Chen, B. Zhao, Y. Dai, Experimental investigation of the
characteristics of cryogenic oscillating heat pipe, Int. J. Heat Mass Transfer 79
(2014) 713–719.
mance is much higher than the results presented in [3] when the [12] T. Mito, K. Natsume, N. Yanagi, H. Tamura, T. Tamada, K. Shikimachi, S. Nagaya,
Achievement of high heat removal characteristics of superconducting magnets
PHP valve was opened. In addition, a comparison between other with imbedded oscillating heat pipes, IEEE Trans. Appl. Supercond. 21 (3)
PHPs indicate that this PHP configuration achieved the highest (2011) 2470–2473.
thermal conductivity where the number of turns and inner diame- [13] L.D. Fonseca, F. Miller, J. Pfotenhauer, A helium based pulsating heat pipe for
superconducting magnets, in: AIP Conference Proceedings (Vol. 1573, No. 1,
ter were the two major differences observed. Experiments nitrogen pp. 28-35). AIP, 2014.
PHPs showed smaller effective thermal conductivities even with [14] F. Bonnet, P. Gully, V. Nikolayev, Development and test of a cryogenic
having the condenser section submerged in liquid nitrogen. The pulsating heat pipe and a pre-cooling system, in: AIP Conference Proceedings
(Vol. 1434, No. 1, pp. 607-614). AIP, 2012.
data validated that higher heat loads tend to increase the fluid [15] D. Xu, L. Li, H. Liu, Experimental investigation on the thermal performance of
velocity since the decrease in oscillation periods were noticeably helium based cryogenic pulsating heat pipe, Exp. Thermal Fluid Sci. 70 (2016)
observed since more vapor bubbles are generated as the heat load 61–68.
[16] M. Chorowski, P. Grzegory, C. Parente, G. Riddone, Experimental and
is increased which will cause an increase in pressure perturbations. mathematical analysis of multilayer insulation below 80 K (No. LHC-Project-
The performance between the vertical and horizontal orientations Report-385), 2000.
is noticeable and shows that even with many loops the gravita- [17] H.R. Deng, Y.M. Liu, R.F. Ma, D.Y. Han, Z.H. Gan, J.M. Pfotenhauer, Experimental
investigation on a pulsating heat pipe with hydrogen, in: IOP Conference
tional effects on PHP performance are still measurable.
Series: Materials Science and Engineering (Vol. 101, No. 1, p. 012065). IOP
Publishing, 2015.
References [18] X. Han, H. Ma, A. Jiao, J.K. Critser, Investigations on the heat transport
capability of a cryogenic oscillating heat pipe and its application in achieving
[1] M.S. Reisch, Coping with the helium shortage, Chem. Eng. News 91 (5) (2013) ultra-fast cooling rates for cell vitrification cryopreservation, Cryobiology 56
18–19. (3) (2008) 195–203.
[2] W. Forest, Helium shortage lingers for hospitals and imaging centers, 2012. [19] Sameer Khandekar, Thermo-hydrodynamics of closed loop pulsating heat
Retrieved from <http://www.auntminnie.com>. pipes, 2004.
[3] L.D. Fonseca, F. Miller, J. Pfotenhauer, Design and operation of a cryogenic [20] Y. Zhang, A. Faghri, Advances and unsolved issues in pulsating heat pipes, Heat
nitrogen pulsating heat pipe, in: IOP Conference Series: Materials Science and Transfer Eng. 29 (1) (2008) 20–44.
Engineering (Vol. 101, No. 1, p. 012064). IOP Publishing, 2015. [21] L.D. Fonseca, F. Miller, J. Pfotenhauer, Thermal Performance of a Helium
[4] S. Okazaki, H. Fuke, Y. Miyazaki, H. Ogawa, Development of meter-scale U- Pulsating Heat Pipe, in: 14th IIR International Conference Proceedings, 2017.
Shaped and O-Shaped Oscillating Heat Pipes for GAPS, J. Astronom. Instrum. 3
(02) (2014) 1440004.