See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268370823

A Loop Thermosyphon for Asphalt Tank Heating

Article

CITATIONS READS

2 1,511

2 authors:

Fernando H. Milanez Marcia Mantelli

Federal University of Santa Catarina Federal University of Santa Catarina
39 PUBLICATIONS   177 CITATIONS    149 PUBLICATIONS   562 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development of passive aircraft cooling systems View project

OPTIMUM EFFICIENCY ANALYSIS OF REGENERATIVE CYCLE WITH FEED WATER HEATERS View project

All content following this page was uploaded by Marcia Mantelli on 28 January 2015.

The user has requested enhancement of the downloaded file.

 A Loop Thermosyphon for Asphalt Tank Heating

F. H. Milanez, M. B. H. Mantelli
Heat Pipe Laboratory
Department of Mechanical Engineering
Federal University of Santa Catarina
Phone: +55 44 3331-9379, FAX: +55 48 3331-7615
[email protected], [email protected]

Abstract

The objective of this work is to evaluate the existing steam coils of the asphalt tanks of the Brazilian
Petroleum Company as a condenser of a loop-thermosyphon. The existing asphalt tank heating system uses
steam generated in a 10 bar boiler placed far from the tanks. The basic idea is to replace this system by loop-
thermosyphons using the existing coils. The evaporator would be placed nearby the tank and the heat source
would be gas combustion. Then the fabrication of the new system would not involve modification of the coils
inside the tanks, which means minimum construction time. The main advantage of using loop-thermosyphons
for asphalt tank heating is the independency from the plant boiler. Also, the loop-thermosyphon condenser
temperature could be controlled by burning gas according to the desired heating power. Therefore, it would not
be restricted to the maximum boiler temperature of 180°C. A small scale model of the system was built and
tested in the laboratory. The results show that the system works very well when the condenser heat transfer
coefficient between the condenser and the heat sink is small. However, when the heat transfer coefficient is
larger, the system does not start-up completely: a large portion of the condenser is not reached by the vapor.
This phenomenon could be due to either the presence of non-condensable gases or a failure in the start-up
associated to the vapor flow sonic limit.

Key Words: loop-thermosyphon, asphalt heating, industrial heaters.

1. INTRODUCTION existing coil. The evaporator will be placed nearby

the tank and will be heated by natural gas
Asphalt is the more viscous and dense product combustion.
obtained from petroleum refining. Its main
application is road paving. Because of its high
viscosity, asphalt storage is made inside heated
tanks. The higher the temperature, the lower is the
viscosity and the easier is to transport it from one
tank to the other. Petrobras, the Brazilian
Petroleum Company, employs several large
asphalt storage tanks in their plants with a capacity
of more than a thousand tons. In order to keep the
asphalt at the required temperature level of 140°C,
the tanks are equipped with steam coils placed at
the bottom. Figure 1 shows a schematic of the
steam coil lay out. The steam is provided from a Figure 1. Asphalt tank steam coil lay out
10 bar boiler that is responsible for all the steam
used inside the Plant. However, as the maximum The advantages of the loop-thermosyphon over the
vapor temperature inside the coil is 180°C, the current equipment are basically two:
resulting asphalt temperature in most of the tanks independency from the plant boiler and more
is of 130°C. Therefore, in order to attend the flexibility regarding to the temperature operation.
140°C demand, a new heating system is required. As the actual system is limited to a 180°C coil
temperature, as already mentioned, the asphalt
This work presents the preliminary results of an does not reach the required temperature level of
experimental study on an engineering model of a 140°C because of the heat losses. With the loop-
loop-thermosyphon for application on heating of thermosyphon, the temperature could be
asphalt tanks. The objective of the thermosyphon controlled by burning gas in the evaporator section
is to replace the actual heating system. The according to power input needed. Also, the system
condenser of the loop-thermosyphon will be the

1
would be independent from the Plant central boiler order to evaluate the applicability of the loop-
and more compact, which means less heat losses thermosyphon conception for asphalt tank heating.
to transport steam from the boiler to the tank. As the actual system is too large to be tested in a
laboratory, an engineering model was designed
Loop thermosyphon heat exchangers, also known and built. Figure 3 presents the experimental set-
as separated heat pipes, have been successfully up. The coil diameter was built in a 1:6 scale of
applied in industrial waste heat recovery systems. the actual system, which means 6 mm i.d and 11
Dube et al. [1] present a study on such a system. m long. As one can see, the geometry of the
These researchers were particularly focused on the experimental setup condenser is not identical to
effects of non-condensable gases on the system the coil of Fig. 1. However, the number of U-turns
performance. In a typical loop thermosyphon heat in both cases is the same, and the total length of
exchanger the geometry of the condenser differs the condenser is 1:6 of the actual system. These
significantly from the lay out sketched in Fig. 1. differences between the model and the actual coil
Figure 2 presents a schematic drawing of a typical are not expected to cause inconveniences.
loop thermosyphon heat exchanger. Both the
evaporator and the condenser are geometrically Regarding to the condenser heat sink, two types of
very similar. They consist of two horizontal tests were performed: air under natural convection
headers (upper and lower) connected by several and forced convection of a thermal fluid
vertical tubes in parallel. The vapor line coming (ethylene-glycol). In the forced convection tests,
from the evaporator is connected to the upper the condenser is placed inside a tank filled with
header. As vapor condenses, the liquid flows by the thermal fluid. The temperature of the thermal
gravity to the lower header, which is connected fluid is strictly controlled thought a LAUDA®
back to the evaporator. PR855 controlled temperature thermal bath.

The evaporator of the model is made of a SS 316

tube, of 100 mm i.d. by 400 mm long, where the
longitudinal axis is in the horizontal position. The
vapor
working fluid is distillated water and the filling
Condenser ratio is 80% of the evaporator volume. The vapor
line is connected to the top of the horizontal
cylinder, while the liquid return line is connected
to the side of the cylinder, below the liquid pool
level. Heat is provided by eight 20 mm o.d.
liquid cartridge type electrical heaters immersed in the
liquid pool.

Evaporator The evaporator of the model does not correspond

to the actual system. The evaporator of the actual
system should be more like the evaporator of the
Figure 2. Loop thermosyphon heat exchanger loop heat pipe of Fig. 2, and the heat source
should be the hot exhaustion gases from a gas
As for the system under investigation, the main burner. However, the study is focused on the
concern is about the geometry of the condenser, condenser only, as already mentioned. Electrical
which is the existing steam coil shown in Fig. 1. heaters were employed as the heat source, since
The active length of the condenser is mostly in the they provide a better heat power control. As the
horizontal orientation, apart from the “U”-turns, entire system is well insulated with glass wool, by
which are slightly tilted towards the condensate measuring the electrical resistance and the current,
flow direction. The main objective of this work is the heat power input could be accurately measured.
to verify if the steam coil of Fig. 1 can be used as
the condenser of a loop thermosyphon. The main The condenser can also be tilted with respect to
concerns are the start-up and the effect of non- the horizontal position. During the tests, the angle
condensable gases on the performance of the “θ” of the condenser measured from the horizontal
system. orientation (see Fig. 3) assumed values of 0°, 7°,
13°, 45° and 90°.

2. EXPERIMENTAL STUDY The condenser was instrumented with seven K-

type thermocouples distributed evenly over its
2.1. Experimental Set-up length. The first thermocouple was placed 50 mm
from the start of the condenser. The seventh
An experimental set-up was built and tested in thermocouple was placed 50 mm from the end of

2
the condenser.
2.3. Experimental Uncertainty

vapor
After calibration of the system, the uncertainty in
temperature measurement is ±0.60°C in the range
of temperatures of interest (ambient to 350°C).
condenser The uncertainty in voltage measurement is ±0.1%,
and the uncertainty of the heater electrical
θ resistance value was found to be ±4%.

liquid
Following the procedure described by Holman [2],
the uncertainty in the thermal resistance
evaporator measurement during the tests had a maximum
value of ±8 %.
ed
trol e
con eratur
p
tembath
3. RESULTS AND DISCUSSION

Previous experience with conventional stainless

Figure 3. Experimental set-up steel-water thermosyphons showed that under
certain circumstances the vapor does not reach the
The vapor line was instrumented with two K-type end of the condenser, especially for large values of
thermocouples, one at the beginning and one at the the condenser external heat transfer coefficient
end. The liquid return line was instrumented with (Refs. [3] to [5]). When the heat transfer
eight K-type thermocouples evenly spread over its coefficient is small, such as air in natural
length. convection, the vapor generally gets very close to
the condenser end. On the other hand, when the
The evaporator inside was instrumented with two heat transfer coefficient is large, such as cooling
K-type thermocouples: one immersed in the liquid by a liquid under forced convection, the vapor
pool and one immersed in the vapor space above only wets a small part of the condenser. In this
the liquid pool. case, the temperature of the region of the
condenser that is not reached by the vapor
The temperature, voltage and current were becomes equal to the heat sink temperature, i. e.,
measured with a HP 3947-A® Data Acquisition there is no heat transfer between that portion of
System connected to a personal computed, which the condenser and the heat sink, which means that
stored the data for further treatment. part of the condenser is inactive. This behavior can
be attributed to non-condensable gases as
described by Dube et al. [1], Peterson [6], Busse et
2.2. Test Procedure al. [7], among others, or by the fact that the start
up is not complete, as described by Mantelli et al.
For each set of tests, i.e., air natural convection or [3] and by Silverstein [8]. Non-condensable gases
thermal fluid forced convection, and for each of could be present because of either the hydrogen
the condenser tilt angles, the heater was turned on, generated by chemical reaction between the water
starting from the thermal equilibrium with the and the tube walls (stainless-steel) [1], [6], [7]
ambient. The power input was first set to 42 W. and/or by undetected leakage of air into the system.
After the system reached steady state, the power The incomplete start-up of the system could
input was increased by 6 steps up to 2000 W. In happen when the vapor flow reaches the sonic
each power step, the system was left to reach limit [8].
steady state. In this condition, the thermal
resistance R [°C/W] of the condenser is computed With the knowledge of this behavior, as already
as: mentioned, the condenser were cooled by both
T −T natural convection and forced convection. Based
R= e c , (1) on the success of past laboratory experiments, the
Q&
natural convection tests were performed first, in
where Te [°C] is the temperature of the vapor order to check the system proper operation,
above the evaporator liquid pool, Tc [°C] is the especially when the setup is subjected to the
average of the seven thermocouples attached to several different condenser tilt angle “θ” (see Fig.
the condenser and Q& [W] is the heater power 3). However, air in natural convection is not a
good representation of the heat transfer
input. coefficients between the loop and the asphalt. For

3
the tube employed (8 mm o.d.) and for air under 250
natural convection at an ambient temperature of 307 W 871 W
42 W
25°C, the global heat transfer coefficient between 200

the thermosyphon working fluid and the heat sink

Temperature [°C]
150
(air) was estimated according to Incropera and De
Witt [9] to be approximately 8 W/m²⋅°C. For the 100 evaporator 1960 W
actual system of Fig. 1 with asphalt, the global condenser
heat transfer coefficient is approximately 80 50

W/m²⋅°C. On the other hand, the heat transfer 0

coefficient for the forced convection test with 0 5000 10000 15000 20000 25000

ethylene-glycol, is approximately 170 W/m²⋅°C, Time [s]

which is a better representation of the actual data. Figure 4. Evaporator and condenser temperature
Therefore, if the condenser worked correctly readings as a function of time
under the two extreme values of 8 and 170
W/m²⋅°C, it would also work for the asphalt. 0.1
90°

Thermal Resistance [°C/W]

45°
3.1. Natural Convection Tests
0°
13°
During these tests, the tank of the condenser was 0.01 7°
open and the condenser was cooled by natural
convection of air. Figure 4 presents the collected
temperature data as a function of time for a typical
test. Only the thermocouple placed inside the
0.001
evaporator to measure the vapor temperature and 0 500 1000 1500 2000 2500
the seven thermocouples spread on the condenser Power input [W]
external surface are displayed here. As the system
was turned on with a power input of 42 W, one can Figure 5. Condenser thermal resistance as a function
see that it did not start up since only the of the heat power input
evaporator thermocouple was above the room
temperature (approximately 25 °C). As the power
input was increased to 307 W, the first five 3.2. Forced Convection Tests
condenser thermocouples started to feel the
presence of the vapor, one after the other. The In order to study the effect of the condenser
readings of these five first thermocouples were external heat transfer coefficient on the system
practically identical and approximately 5 °C below performance, the thermosyphon was tested under
the evaporator temperature. As the power input forced convection cooling, as already mentioned.
was increased to 871 W the vapor reached quickly Only tests for tilt angles of 0 and 7° were
the entire length of the condenser and all the seven performed. The controlled thermal bath
condenser thermocouples read practically the temperature was tested for 20, 80 and 120°C. In
same temperature. all cases, the vapor front never went beyond the
second thermocouple of the condenser, which
Figure 5 presents the values of the thermal means at least 70% of the condenser length was
resistance of the condenser, defined according to inactive. Only a small portion of the condenser
Eq. (1), as a function of the heat power input, for was enough for the vapor to condense. This
all the condenser tilt angles tested during the observation agrees with previous works on
natural convection tests. As one can see, the conventional stainless steel-water thermosyphons
thermal resistance decreases as the power input [3] to [5].
increases. Also, the differences among the thermal
resistance values for different tilt angles are large According to references [1] and [6], conventional
for small values of power input (less than 1700 thermosyphons can experience similar behavior in
W). In this case, the smaller values of thermal the presence of non-condensable gases. According
resistance were found for tilt angles of 7 to 13°. to Busse et al. [7], stainless steel-water
Interestingly, the maximum tilt angle, i. e. 90°, in thermosyphons generate hydrogen, which obstruct
general yielded larger thermal resistances. the condenser, leading to the appearance of a cold
However, for power input of 1700 W and more, zone. Those authors studied several cleaning and
the effect of the tilt angle is negligible on the fabrication procedures, including passivation by
condenser thermal resistance. It also can be noted addition of oxygen in order to avoid the
that a further increase in the power input would appearance of non-condensable gases. They
not lead to a significant decrease in the condenser concluded that none of these procedures were
thermal resistance. helpful in avoiding the presence of non-

4
 condensable gases. Dube et al. [1] also coils inside the tanks, which means minimum
acknowledge the presence of non-condensable construction time. The main advantage of using
gases in stainless steel-water thermosyphons. As loop-thermosyphons for asphalt tank heating is the
there is no way to avoid their presence, they independency from the plant boiler (10 bar). With
studied experimentally the best position to place a the loop-thermosyphon, the temperature could be
gas reservoir to accommodate the non- controlled by burning gas according to the desired
condensable gases during operation of a loop- heating power.
thermosyphon similar to the one shown in Fig. 2.
Acknowledgements

4. SUMMARY AND CONCLUSIONS The authors would like to acknowledge the

support of PETROBRAS and CNPq during this
A coil type condenser geometry is analyzed in this project.
work for application as a condenser of a loop
thermosyphon for asphalt tank heating. This type References
of geometry differs significantly from the
geometry generally employed for loop 1. Dube, V., Akbarzadeh, A., Andrews, J., “The
thermosyphon heat exchangers. Effects of Non-condensable Gases on the
Performance of Loop Thermosyphon Heat
A small scale prototype of the condenser was built Exchangers”, Applied Thermal Engineering, Vol.
and tested in order to check the system operation 24, 2004, pp. 2439-2451.
conditions. Natural convection of air and forced
2. Holman J. P., “Experimental Methods for
convection of ethylene-glycol were used to
Engineers”, 6th Edition, McGraw-Hill, Singapore,
remove heat from the condenser, in order to test a
1994, pp. 49-103.
large range of possible heat transfer coefficients
between the condenser and the heat sink. 3. Mantelli, M. B. H., Martins, G. J., Zimmermann,
R., Reis, F., Rocha, G. K. L., Landa, H. G.,
The results showed that under a poor external heat “Experimental Study of Vertical Thermosyphons
transfer coefficient (natural convection of air), the for Industrial Heat Exchangers Applications”,
vapor spreads through the entire condenser length. Proceedings of the 13th International Heat Pipe
On the other hand, when the external heat transfer Conference, Shanghai, China Astronautic
coefficient is larger (forced convection of Publishing House, 2004, pp. 531-536.
ethylene-glycol), only a small portion of the
4. Molz, M., Mantelli, M. B. H., Milanez, F. H.,
condenser is active. According to the literature,
Landa, H. G., “Transient Modeling of a Closed
this behavior could be either due to the presence
Two-Phase Thermosyphon for Heat Exchanger
of non-condensable gases or by a failure in the
Applications”, Proceedings of the 13th
start-up of the system.
International Heat Pipe Conference, Shanghai,
China Astronautic Publishing House, 2004, pp.
The presence of an inactive portion of the
89-95.
condenser is undesired for the application under
study. It is necessary that the entire condenser 5. Mantelli, M. B. H., Milanez, F. H., Nielitz, G.,
length be wetted by the vapor, so the asphalt be “Tree Configuration Thermosyphon Study”, 38th
uniformly heated inside the tank. This problem AIAA Thermophysics Conference, Toronto, Paper
could be solved by installing a reservoir to trap the No. AIAA-2005-4953, 2005.
non-condensable gases, similarly to the work of
6. Peterson, G. P., “An Introduction to Heat Pipes”,
Dube et al. [6], provided the source of the problem
John Wiley & Sons, New York, 1994, pp. 247-248.
is indeed the presence of non-condensable gases.
Further work is needed to investigate this 7. Busse, C.A., Loens, J., Campanile, A.,
phenomenon. A loop thermosyphon made of glass, “Hydrogen Generation in Water Heat Pipes at
similar to the one used in this study, is already 250°C”, 1st International Heat Pipe Conference,
being built to better understand this issue. Stuttgart, Paper No. 4-2, 1973.
8. Silverstein, C. C., “Design and Technology of
Could this problem of limited active condenser Heat Pipes for Cooling and Heat Exchange”,
length be solved, the existing steam coils could be Taylor & Francis, 1992, pp. 38-45.
used as condensers of loop thermosyphons to heat
the asphalt tanks. The optimum condenser tilt 9. Incropera, F. P., De Witt, D. P., “Fundamentos
angle appears to be approximately 7°, according to de Transferência de Calor e de Massa”, 3ª. Edição,
experimental results, which is the tilt angle of the Ed. Guanabara Koogan S.A., Rio de Janeiro, 1992,
actual system. Therefore, the fabrication of the pp. 176-255.
new system would not involve modification of the

5
View publication stats