Modeling and Simulation
of Chemical Reactors Cooled by
Thermosyphons
by
Marcus Hothar
Department of Chemical Engineering
Lund University
June 2017
Supervisor: Professor Hans T. Karlsson
Examiner: Associate professor Christian Hulteberg
Postal address
P.O. Box 124
SE-221 00 Lund, Sweden
Web address
www.chemeng.lth.se
Visiting address
Getingevägen 60
Telephone
+46 46-222 82 85
+46 46-222 00 00
Telefax
+46 46-222 45 26
Acknowledgement
This master thesis has been performed at the Department of Chemical Engineering at Lund
University.
I would like to extend my thanks to my supervisor, Professor Hans T. Karlsson for all the help
and support that he has given me as I’ve been working on this (and on previous projects as
well). Secondly, I would also like to thank Professor Bernt Nilsson and Associate Professor
Christian Hulteberg for the advices they have provided when I’ve visited them with questions.
I would also like to thank my family and friends for always supporting me and for giving me
all the time I need to dedicate to my studies.
Abstract
In a theoretical study the use of thermosyphons to improve the safety and efficiency of chemical reactors have been evaluated. Thermosyphons is a simple form of a “heat pipe”, a hollow
and closed tube that’s partially filled with a fluid. When the pipe is heated up to the boiling
temperature of the fluid within a phase change is initiated. The phase-change is used to improve the thermal conductivity of the pipe. Three simulation models have been used in the
study and two of the models is taken from previous studies regarding the thermal safety of
chemical reactors. The results have been analyzed and presents an interesting indication of
positive possibilities. The outstanding heat transfer of the thermosyphons gives according to
the simulation good chance of a firmer control of the reactor temperature and thereby also the
reaction process. Besides the improved temperature control the thermosyphons proved capable of preventing thermal runaways during scenarios when the cooling jacket was out of order.
Sammanfattning
I en teoretisk studie har användbarheten av thermosyphoner för att förbättra säkerheten och
effektiviseringen av kemiska reaktorer. Thermosyphoner är en simpel form av en ”heat pipe”,
ett ihåligt och slutet rör som är delvis fyllt av en fluid. Om röret värms upp till fluidens kokpunkt
inleds en inre fasomvandling vilket används för att förbättra rörets värmeledningsförmåga. Tre
simuleringsmodeller har använts i studien, varav två kommer från tidigare studier rörande
reaktorsäkerhet. Resultaten har analyserats och ger en intressant inblick på potentiella
möjligheter. Thermosyphonernas utomordentliga värmeledningsförmåga ger enligt studien
väldigt goda möjligheter att kontrollera reaktorstemperaturen och därmed även
reaktionsprocessen. Utöver den förbättrade temperaturkontrollen lyckades även
thermosyphoner förhindra skenande reaktioner i scenarion där mantelvärmeväxlaren slutat
fungera.
Table of Contents
1
Background ........................................................................................................................ 1
1.1
Thermal runaways ....................................................................................................... 1
1.2
Heat Pipes .................................................................................................................... 3
2
Aim..................................................................................................................................... 6
3
Method: .............................................................................................................................. 7
4
3.1
Heat Pipes – Theoretical Design ................................................................................. 8
3.2
The Overall Heat Transfer Coefficient ...................................................................... 10
3.3
The Pipes Operating Limit: ....................................................................................... 13
3.4
Simulations Models: .................................................................................................. 18
3.5
A Worst-Case Scenario ............................................................................................. 27
Results & Discussions:..................................................................................................... 33
4.1
The Steinbach Case ................................................................................................... 33
4.2
Hydrolysis of Acetic Anhydride................................................................................ 42
4.3
Fine Chemicals .......................................................................................................... 46
5
Conclusion ....................................................................................................................... 50
6
Further work ..................................................................................................................... 51
7
References ........................................................................................................................ 52
Table of Figures:
Figure 1. A visual display of the process behind a thermal runaway. ..................................................... 1
Figure 2. A simple display of a heat pipe and the three separate regions that its length is divided into.
................................................................................................................................................................. 4
Figure 3. A cross-sectional picture that displays the wick structure within the wicked heat pipe and
the regions for liquid and gas. ................................................................................................................. 5
Figure 4. A figure displaying a cooling jacket surrounding a batch reactor along with four installed
heat pipes. ............................................................................................................................................... 7
Figure 5. A graph displaying the velocity differences of the vapor inside the heat pipe. ....................... 9
Figure 6. A graph displaying the changes of the fanning friction due to temperature of a pipe with a
smooth surface...................................................................................................................................... 11
Figure 7. A graph displaying the heat transfer coefficient of pipes (with a smooth surface) of various
radiuses at different temperature. ....................................................................................................... 12
Figure 8. A graph displaying the heat transfer coefficient at different temperatures, roughness and
pipe radiuses. ........................................................................................................................................ 12
Figure 9. A graph displaying the difference that the roughness of 1 mm brings to the heat coefficient
compared with a pipe without a rough surface. ................................................................................... 12
Figure 10. A fault tree that displays the base events that can trigger a thermal runaway for an
ordinary semi-batch reactor. ................................................................................................................ 29
Figure 11. A fault tree presenting the base events that can cause a thermal runaway to occur in a
semi-batch reactor that’s cooled with both a cooling jacket and thermosyphons. ............................. 31
Figure 12. A graph showing the reactor temperature during the cycle when three, four or five
thermosyphons is used. [rinner= 5 cm] ................................................................................................... 33
Figure 13. A graph displaying the temperature within the reactor during the cycle time. A different
number of thermosyphons was tested yet they all had the same inner radius of 5 cm. ..................... 33
Figure 14. The concentration of reactant A within the reactor during the cycle time ......................... 34
Figure 15. The concentration of the reactant that’s being fed into the semi batch reactor, reactant B,
under the reaction cycle. ...................................................................................................................... 34
Figure 16. The difference of conversion of reactant A during the process cycle when using a different
number of thermosyphons. .................................................................................................................. 35
Figure 17. The energy that’s being released due to the exothermic reaction. ..................................... 36
Figure 18. A graph displaying the reaction rate during the process cycle. ........................................... 36
Figure 19. The changes of the heat coefficient during the process. ..................................................... 37
Figure 20. The heat transfer that each thermosyphon performs during the cycle time. ..................... 37
Figure 21. The reactor temperature during the process cycle when various numbers of
thermosyphons is installed to provide additional cooling .................................................................... 38
Figure 22. The conversion rate of the loaded reactant, component A, during the simulation. ........... 39
Figure 23.The concentration of the loaded reactant, component A, during the dosing time.............. 39
Figure 24. The concentration of the dosed reactant, component B, during the reaction. ................... 39
Figure 25 The overall heat transfer coefficient that determines the amount of heat that’s transfered
from the reactor to the heat pipes. ...................................................................................................... 40
Figure 26. The heat that’s transfered from the reactor by each thermosyphon in comparison with
their dominating limit, the boiling limit. ............................................................................................... 40
Figure 27. A display of the different reaction rates between the two cooling systems. ...................... 40
Figure 28. The energy amount that’s released during the process. ..................................................... 40
Figure 29. The temperature within the reactor in case the cooling jacket isn’t functionable during the
cycle. ...................................................................................................................................................... 41
Figure 30. The effect that three, four, five and six thermosyphon has on the reaction temperature
during a worst case scenario. ................................................................................................................ 41
Figure 31. The heat transfer that each thermosyphon provides in comparison with their dominating
limit. ....................................................................................................................................................... 42
Figure 32. The reaction temperature within the semi-batch reactor when various numbers of
thermosyphons is used.......................................................................................................................... 43
Figure 33. The rate of the hydrolysis reaction when different numbers of thermosyphons was used.44
Figure 34. The concentration of the fed reactant, acetic anhydride. ................................................... 44
Figure 35. The heat transfer that a thermosyphon performs during the process in comparison with its
operating limit. ...................................................................................................................................... 44
Figure 36. The energy that’s released from the hydrolysis reaction during the process cycle............. 44
Figure 37. The reaction temperature within the semi-batch reactor when various numbers of
thermosyphons is used and the cooling jacket is not providing any cooling of the process. ............... 45
Figure 38. The heat transfer of each thermosyphon during the malfunction cycle in comparison with
the pipe’s dominating limit. .................................................................................................................. 45
Figure 39. A display of the reaction temperature when the two cooling systems is compared with
various numbers of thermosyphons installed. ...................................................................................... 46
Figure 40. The temperature within the reactor when two to five thermosyphons is installed in
comparison with the original cooling system........................................................................................ 46
Figure 41. The concentration of reactant A during the cycle time of the batch reactor when different
numbers of thermosyphons was used. ................................................................................................. 47
Figure 42. The concentration of the undesired product, S, during the cycle time of the batch reactor
when different numbers of thermosyphons was used. ........................................................................ 48
Figure 43. The concentration of the desired product, P, during the cycle time of the batch reactor
when different numbers of thermosyphons was used. ........................................................................ 48
Figure 44. A comparison between the two cooling systems effect on the conversion rate of the
reactant, component A, during the process.......................................................................................... 48
Figure 45. A display of the selectivity changes that the thermosyphons provides. .............................. 48
Figure 46. The overall heat transfer coefficient during the reaction cycle. .......................................... 49
Figure 47. A display of the heat transfer that each thermosyphons displays in comparison with the
boiling limit. ........................................................................................................................................... 49
1 Background
1.1 Thermal runaways
“The Chemical industry, more than any other industry, is perceived as a threat to humans, society and the environment.” (Stoessel, Chapter 1 – Introduction to Risk Analysis of Fine
Chemical Processes, 2008)
The reason for the statement above is not only due to the risks that comes with handling hazardous components such as acids and toxins. Chemical reactions can release large quantities of
energy as well as gaseous by-products if not under sufficient control. One of the major threats
that involves the chemical industry is chemical reactions on a runaway. In an investigation that
was performed by the U.S Chemical Safety and Hazard Investigation Board (CSB) a total
amount of 167 serious incidents that occurred in the United States from 1980 to 2001. Of all
the 167 incidents thirty-five percent were the result of a runaway reaction and the most common
location was the reactor. (Hazard Investigation - Improving Reactive Hazard Management,
2002)
Most reactions that occurs in main branches of the chemical industry such as the fine chemicals
and the polymer industries involves exothermic reactions where thermal energy is released as
the reaction proceeds. Since almost all chemical reactions have a positive activation energy the
reaction rate increases with higher temperature, this can lead to what’s best described as a bad
cycle and that’s called a runaway reaction or a thermal runaway. The reaction releases energy
that leads to an increase of temperature, which in turn results in an exponential growth of both
temperature and reaction rate (see Figure 1). All chemical reactors are “usually” installed with
a heating/cooling system that is either to keep the up temperature if the main reaction is endothermic or keep it down if it’s exothermic. For batch reactors, semi-batch reactors and CSTRs
the cooling is usually performed using a cooling jacket, a heat exchanger that surrounds the
shell of the reactor. For reasons such as faults in the process design, scale ups, fouling, human
errors and malfunctions the cooling processes might however be unable to keep the reaction
temperature under control. This leads to the bad cycle that was mentioned previously and what
is called a thermal runway. (Stoessel, Chapter 2 – Fundamentals of Thermal Process Safety,
2008), (Karlsson, 2012)
Exothermic reaction
Heat is released
Thermal
Runaway
Increased reaction rate
Figure 1. A visual display of the process behind a thermal runaway.
1
During a thermal runaway, the reaction rate of the exothermic reaction grows too high due to
the temperature increases and the cooling system isn’t capable of transferring of enough energy
which results in a continuous temperature increase. The high temperature can thereby initiate
undesired side reactions that can release even more energy and result in vaporization and growing pressure. The consequence of the pressure that builds up can be everything from nothing to
severe, it depends on the number of reactants and how much energy that the reaction releases.
In the more severe cases an explosion or possibly a leak of dangerous components from within
the reactor can occur. In August 1998, an explosion and a fire took place in New Jersey, USA,
causing nine people being injured, material damage and the release of hazardous material. The
reason for this was a thermal runaway the occurred within a batch reactor, the temperature
released initiated a decomposition reaction which in turn caused the explosion. (Guinand,
2016), (MAHBulletin, 2016)
There are two things that’s required for a thermal runaway to occur:
At least one exothermic reaction
An improved reaction rate of the exothermic reaction due to the increased temperature
within the reactor
As can be seen in Table 1 several reactions that plays an important part of the chemical industry
involves exothermic reactions. The second part that is required for a thermal runaway is an
improved reaction rate during a temperature increase. All reactions that possess a positive activation energy falls under this category and that’s almost as good as all chemical reactions.
Thereby a thermal runaway is very often a potential risk that needs consideration through risk
analyzations and consequence evaluations. (Karlsson, 2012)
Table 1. A couple of classical reactions and their approximate (experimental) reaction enthalpies. (Grewer, 1999)
Reaction Enthalpies (ΔHreaction [kJ/mol])
Reaction
Neutralization
-55
Polymerization (double bond)
-80
Sulfonation
-60
Nitration
-150
Amination
-220
Hydrogenation
-560
One of the many problems when it comes to the process of designing a batch reactor is the
scale-up step. During a scale-up the volume of the reactor is increased to improve the production yet the surface area of the reactor doesn’t grow at the same pace. Therefore, the area where
the heat transfer occurs per unit of volume is lowered when the reactor volume grows. The
reactor is now larger and will thereby release more energy due to the exothermic reaction yet
2
at the same time the cooling area hasn’t increased as much itself resulting in a higher temperature rise. This can of course be overcome with the aid of a lower temperature of the cooling
medium or improvements of the overall heat transfer coefficient yet both can come at a good
price. (Moulijn, 2008)
A process with absolute safety (no risks) isn’t possible due to possibility that all the protective
measurements that has been taken will fail simultaneously and secondly that there is always the
potential of a human error. However, the more control that can be gained over the temperature
of a chemical reaction the lower will the risk be for a thermal runaway.
1.2 Heat Pipes
A heat pipe is a very efficient tool to achieve thermal control since they have very good heat
transfer capabilities. The idea of what’s today called a heat pipe originated from Gauger in 1944
yet it wasn’t until 1963 that G. M. Grover created the first patent of a heat pipe. During the last
decades, quite many studies have been performed regarding this device and today they are used
in engineering fields such as electronics, aerospace, food and energy for purposes such as cooling and heat recovery. In later years, the studies regarding their uses for environmental issues
as well as safety problems have also been growing. There are patents displaying chemical reactors where the temperature is being controlled through heat pipes yet the tool hasn’t been
used as much in the reaction industry as in several others so far.
All types of heat pipe consist of a hollow, sealed metal pipe which contains a liquid. The pipe
can then be used as an effective way of transferring energy from one end of the other, Figure 2.
The reason they are of such interest is that they can transfer impressive amounts of heats with
almost no heat loss. The heat transfer rate that they are capable of achieving is thousands of
times greater than that which a solid heat conductor of equal size could perform and this is the
reason they are sometimes referred to as the superconductors. (Dincer, Chapter 7 - Heat Pipes)
The heat transfer is achieved through a phase-changes within the pipe. One part (most often the
lower) of the pipe is heated at a temperature above the boiling limit of the working fluid within
the pipe which leads to an evaporation. The vapor that’s created causes a forced convection on
the inside of the heat pipe which increases the heat transfer coefficient. The vapor then rises to
the opposite end of the pipe where the heat pipe is cooled and a condensation is performed as
the vapor meets the inner wall of the pipe. The condensate then slides down along the pipes
surface to the end where the evaporation takes place creating a passive loop. The great effect
of the heat pipe is mainly because of the high heat transfer coefficient that the tube can reach
and the high latent heat of the fluids that’s being used.
3
Evaporation region
Adiabatic Region
Heat
Condensation Region
Heat
Figure 2. A simple display of a heat pipe and the three separate
regions that its length is divided into.
There are many kinds of working fluids that can be used such as water, acetone, ethanol, sodium
and potassium depending on which temperature the boiling point is desired. Alkali metals such
as sodium and lithium are for example formidable for the purpose due to their massive latent
heat yet they can’t be used at low temperatures due to their high boiling temperatures. In temperature ranges of between twenty to two-hundred degrees Celsius water is very common since
it has high latent heat and a low cost.
The simple construction, impressive heat transfer capabilities and the reliability of the heat pipe
makes it a promising tool to be used. They themselves don’t require regulations since the main
motions within is based almost purely on phase changes. For this study two main kinds of heat
pipes will be analyzed and be used for theoretical studies regarding their capabilities to improve
the temperature control as well as the efficiency of a chemical reactor.
1.2.1 Thermosyphons
The thermosyphon (or thermosiphon, heat pipes assisted by gravity, Two-phase closed thermosyphon) is the simplest kind of a heat pipe, it’s a simple hollow metal pipe that’s partly filled
with a working fluid and sealed under a suitable pressure to ensure the desired boiling point.
They can perform good heat transfer and are very cheap due to their simple construction. They
are being used for many purposes for example transferring solar energy and cooling electronics.
Thermosyphons is often used and in reboilers as well since they generally provide higher heat
fluxes and an improved heat transfer coefficient. As one of the device many names states however it has one weakness compared to other heat pipes, it requires the assistance of gravity. The
evaporator end of the thermosyphon must be beneath the condenser end or the device won’t
function properly. Considering however that chemical reactors tends to be stationary this isn’t
much of a problem in this case. (Abdollahi, 2015)
1.2.2 Wicked Heat Pipes
A wicked heat pipe has a wick at the inside of the hollow tube, the wick is a structure that allows
the vapor to pass through it yet can’t be penetrated by the liquid. This causes the wicked heat
pipe to have two cross sectional regions, the liquid region that’s in contact with the inner surface
4
of the pipe and the vapor region that’s within the wick structure. By adding a wick to a heat
pipe the device now functions even if it’s not gravity assisted yet its angle and position still
matters when the full capacity of the heat pipe is to be determined as will be discussed later.
The wick structure
The vapor region
The liquid region
The pipe
Figure 3. A cross-sectional picture that displays the wick structure within the wicked heat pipe
and the regions for liquid and gas.
5
2 Aim
If heat pipes were installed in a reactor a large amount of heat could be transferred from away
from the reaction to keep the reaction rate from reaching dangerous levels. This could either be
as a safety precaution or as a standard cooling mechanism. For instance, in a batch reactor that’s
cooled by a cooling jacket heat pipes could be installed with a boiling point that is above the
desired temperature within the reactor. If the temperature would grow above the desired level
the boiling point will be reached and the heat pipes could begin to transfer heat away from the
reactor.
The heat pipes possess excellent thermal conductivity and therefor a big heat transfer can be
reached without high temperatures rises within the reactor. This as well as the flexibility of the
number of pipes that can be installed could be of useful aid for controlling the temperature
within the chemical reactor.
The aim of this work is to perform an initial theoretical study and perform simulations to analyze the value of the cooling devices known as wicked heat pipes and thermosyphons. The
simulations will be regarding the additional cooling can grant beneficial results without escalating into something of unreasonable scale. The emphasis of the discussions regarding the results will then be placed on process safety and reactor efficiency.
6
3 Method:
The test is performed using several different simulations of reaction processes that occurs
within a batch rector and a semi batch reactor. The reason for this choice is to the initial nonsteady state in both reactors and that semi-batch reactors is most often used to reduce the risk
of a thermal runaways. A comparison of the results is then performed in the following combinations:
A reactor cooled only by a cooling jacket
A reactor that’s cooled by both a cooling jacket and heat pipes
A reactor where the efficiency of the cooling jacket is completely removed yet where the
heat pipes is still functional
Every process that’s simulated will be gone over separately and the mathematical model that’s
being used will be explained. However, for each simulation, the following assumptions are
made:
Ideal heat pipes that’s both adiabatic and isothermal. It’s assumed that there is no superheating of the working fluid and total condensation occurs of the vapor in the condensation
region.
o For thermosyphons the temperature difference is dependent on the length of the
pipe. Tests have been performed that examined the wall temperature of a wickless
heat pipe (at an inclination angle of 90 degrees and diameter of 20 mm) which contained water. These proved that the temperature difference decreased with length
up to the longest pipe that was examined (950 mm). (Khalid, 2000)
Adiabatic reactors and isothermal.
The volume of the heat pipes is neglected and not added to re-calculate the size of the reactor to uphold the decided volume.
Figure 4. A figure displaying a cooling jacket
surrounding a batch reactor along with four
installed heat pipes.
7
3.1 Heat Pipes – Theoretical Design
3.1.1 Geometry of the pipes:
Since heat pipes of various sizes will be used for the simulations certain simplifications will be
made. First off is an assumption of the correlation between the inner radius of the pipe and the
outer radius of the pipe. The inner radius covers the distance from the inner surface of the pipe
to the center while the outer radius includes the thickness of the surrounding metal. The assumption is that the outer radius is always 10% of the inner radius.
𝑇ℎ𝑒 𝑃𝑖𝑝𝑒𝑠 𝑖𝑛𝑛𝑒𝑟 𝑟𝑎𝑑𝑖𝑢𝑠: 𝑟𝑖𝑛𝑛𝑒𝑟
𝑇ℎ𝑒 𝑃𝑖𝑝𝑒𝑠 𝑂𝑢𝑡𝑒𝑟 𝑟𝑎𝑑𝑖𝑢𝑠: 𝑟𝑜𝑢𝑡𝑒𝑟 = 1.1𝑟𝑖𝑛𝑛𝑒𝑟
The length of a heat pipe can be divided up into three sections dependent on the what kind of
heat transfer that occurs in the region, see Figure 2.
The evaporation region
The adiabatic region
The condensation region
In the evaporator region the heat transfer into the heat pipe takes place while the heat transferred
out of the pipe takes place in the condensation region. Within the adiabatic region no heat transfer occurs whether into or out of the pipe and the fluid simply flows though it without any
changes.
To determine the length of the different heat pipe regions the geometry of the reactor is required.
For each reactor, an assumption is made that the height and radius of the reactor is the same.
Thereby the radius and height of the reactor can be calculated according to:
𝑅𝑒𝑎𝑐𝑡𝑜𝑟 ℎ𝑒𝑖𝑔ℎ𝑡 & 𝑟𝑎𝑑𝑖𝑢𝑠: ℎ𝑟𝑒𝑎𝑐𝑡𝑜𝑟 = 𝑟𝑟𝑒𝑎𝑐𝑡𝑜𝑟 = (
𝑉𝑇𝑜𝑡𝑎𝑙 1⁄3
𝜋
)
[𝑚]
The height of the fluid within the reactor can be considered as the evaporator region while the
reminding height of the reactor is the adiabatic region. This results in a difference between the
calculations for a semi-batch reactor and a batch reactor. In a semi-batch reactor, the filled
volume within the reactor increases during the dosing period while in a batch reactor the height
is constant.
It’s assumed that the heat pipe reaches all the way down to the bottom of the reactor which
allows the length to be calculated as the height of the reactants within the reactor:
𝑙𝑒𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑜𝑟 = (𝜋∗𝑟
𝑉(𝑡)
𝑟𝑒𝑎𝑐𝑡𝑜𝑟
2)
[𝑚]
Where the volume functions calculate the fluid volume within the reactor, which means that for
a batch reactor V is constant yet for a semi-batch reactor the volume is:
𝑉𝐹𝑖𝑛𝑎𝑙 = 𝑉(𝑡𝑑𝑜𝑠𝑖𝑛𝑔 )
𝑉0 = 𝑉(0)
8
The length of the condenser is assumed to be twice the length of the evaporator region, this is
due to the lower heat transfer coefficient in the condensation region:
𝑙𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑒𝑟 = 2 ∗ 𝑙𝑒𝑣𝑎𝑝,𝑚𝑎𝑥 [𝑚]
The total length of the heat pipe is therefor to be the same as the sum of the assumed condenser
region along with the maximal height of the reactor:
𝑇𝑜𝑡𝑎𝑙 𝑙𝑒𝑛𝑔𝑡ℎ 𝑜𝑓 𝑡ℎ𝑒 𝐻𝑃: 𝑙𝐻𝑃 = 𝐻𝑒𝑖𝑔ℎ𝑡𝑅𝑒𝑎𝑐𝑡𝑜𝑟 + 𝑙𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑒𝑟 [𝑚]
𝑙𝑎𝑑𝑖𝑎𝑏𝑎𝑡𝑖𝑐 = 𝑙𝐻𝑃 − 𝑙𝑒𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑜𝑟 − 𝑙𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑒𝑟 [𝑚]
Effective length:
To determine the maximal heat transfer capacity of the heat pipe the so-called effective length
is required to be determined. It is needed for the equations that involves the vapor and liquid
pressure drops along within the pipe. The calculation bases on simple one-dimensional equations and the “effective length” makes up for the different velocities in the evaporator and condenser regions of the heat pipe. The velocity is at its highest in the adiabatic section and then
linearly lowered within the evaporator and condenser sections as can be seen in Figure 5.
Figure 5. A graph displaying the velocity differences of the vapor inside the heat pipe.
Thereby only half of the length of the evaporator and condenser region is considered for determining the effective length:
𝑙𝑒𝑓𝑓 = 𝑙𝑎𝑑𝑖𝑎𝑏𝑎𝑡𝑖𝑐 +
𝑙𝑒𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑜𝑟 +𝑙𝑐𝑜𝑛𝑑𝑒𝑛𝑠𝑜𝑟
2
[𝑚]
3.1.2 Suitable working fluids:
To reach an optimal heat transfer the working fluid within the heat pipe is of great importance.
A suitable working fluid comes with:
9
-
Good thermal stability
High latent heat
High thermal conductivity
Low liquid and vapor viscosities
High surface tension
Compatibility with the wick and the container
Two fluids that is often used for lower or medium temperatures is ammonia and water due to
properties such as their high latent heat and reasonably high densities.
To simplify the simulation the properties of working fluid have been assumed to be constant.
Table 2. A table displaying the physical parameters of the working fluids that’s used in this
work. (Reay, Appendix 1, Working fluid properties, 2013)
Working Fluid Parameter:
Ammonia (273 K)
Water (313 K )
Vapor pressure (Pvapor)
4.24 bar
0.07 bar
Vapor density (ρvapor)
3.48 kg/m3
0.05 kg/m3
Liquid density (ρliquid)
638.6 kg/m3
992.3 kg/m3
Surface tension (σliquid)
2.48*10-2 N/m
6.96*10-2 N/m
Liquid viscosity (μliquid)
0.25*10-3 Pa*s
0.65*10-3 Pa*s
Vapor viscosity (μvapor)
0.92*10-5 Pa*s
1.04*10-5 Pa*s
Latent heat (L)
1263 kJ/kg
2402 kJ/kg
3.2 The Overall Heat Transfer Coefficient
“For a well-designed heat pipe, effective thermal conductivity can range from 10-10 000
times the effective conductivity of copper depending on the length of the pipe.” (Dincer,
Chapter 7 - Heat Pipes)
Considering the high thermal conductivity of copper as well as the heat coefficient that the
boiling within the pipe will create the dominant process that will determine the speed of the
heat transfer is the heat coefficient of the liquids inside the reactor. To determine this the correlation between the dimensionless numbers along with the Chilton-Colburn Analogy will be
used. Beside this a rough surface is sometimes used to intensify the turbulence at the pipe’s
surface (and therefor also the heat coefficient).
Since it’s a matter of a flow across tubes (pipes) the characteristic length that’s used to determine the Reynold’s number is the diameter of the tube.
𝑅𝑒 =
𝜌𝑣𝑑
𝑃𝑟 =
𝜇
10
𝜇𝐶𝑝
𝜆
Fanning Friction (f):
𝜀
𝑓 = (−1.737 ln (0.269 −
𝑑
2.185
𝑅𝑒
𝜀
ln (0.269 +
𝑑
14.5
𝑅𝑒
)))
−2
Where d is the charismatic length of the pipe (the outer diameter), and the ε is the roughness of
the outer surface of the pipe. The Chilton-Colburn Analogy then uses the fanning friction as a
factor to determine the heat transfer coefficient of the fluid within the reactor.
𝑓
𝐶ℎ𝑖𝑙𝑡𝑜𝑛 − 𝐶𝑜𝑙𝑏𝑢𝑟𝑛 𝐴𝑛𝑎𝑙𝑜𝑔𝑦: 𝑗ℎ = =
2
ℎ
𝜌𝐶𝑝 𝑣
𝑃𝑟 2⁄3 → ℎ =
𝑗ℎ 𝜌𝐶𝑝 𝑣
𝑃𝑟 2⁄3
Considering that the heat conductivity of copper and the nucleate boiling that takes place within
the heat pipe it can be assumed that the heat transfer coefficient can be roughly estimated as:
𝑘𝑃𝑖𝑝𝑒 ≈ ℎ
This heat coefficient along with the surface area of the evaporator region and the boiling temperature of the working liquid is then used to determine the heat transfer that the heat pipe is
capable of accomplishing. The equation presented is dependent on the assumption that the temperature within the heat pipe remains constant after the working fluid reaches its saturation
temperature.
𝑄𝑃𝑖𝑝𝑒 = 𝑘𝑃𝑖𝑝𝑒 𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 (𝑇 − 𝑇𝐵𝑜𝑖𝑙 )
𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 = 𝜋2𝑟𝑜𝑢𝑡𝑒𝑟 𝑙𝑒𝑣𝑎𝑝
It’s assumed that the stirring within the reactor gives a constant velocity to the fluid of three
meters per second. The physical parameters are then assumed to be the same as water and the
heat coefficient is calculated according to different temperatures, roughness of the metal and
sizes of the pipes.
If the roughness is set as zero then the results indicates clearly that a smaller diameter of the
pipe will have a large influence of the heat coefficient.
𝜀=0
The fanning friction decreases with temperature as well as increased diameter of the pipes
which must be the reason for the clear difference the pipe diameter provides.
Figure 6. A graph displaying the changes
of the fanning friction due to temperature of
a pipe with a smooth surface
11
A roughness that’s independent of the pipe’s diameter is also tested to observe the results. For
the pipes with an inner radius of beneath 5 cm a roughness of 5 mm was used and for the larger
pipes it was set as 1 mm.
The results on the heat transfer coefficient can be seen in Figure 7 and 8.
Figure 7. A graph displaying the heat transfer
coefficient of pipes (with a smooth surface)
of various radiuses at different temperature.
Figure 9. A graph displaying the difference
that the roughness of 1 mm brings to the heat
coefficient compared with a pipe without a
rough surface.
Figure 8. A graph displaying the heat transfer
coefficient at different temperatures, roughness
and pipe radiuses.
The two graphs show that the rough surface of the pipe increases a large increase to the heat
transfer coefficient that determines the heat transfer from the reactor fluid to the pipe. In Figure
8 a comparison of the two coefficients can be viewed. It’s clear that the smaller the radius the
bigger is the effect and that the heat transfer coefficient of the pipe with an inner radius of 1 cm
is at least three times bigger when a rough surface of 1 mm is added to the pipe. Yet it’s still
also very clear that the roughness has a big influence even at the pipes with of larger size and
that their heat coefficients are increased of something between two to three times depending on
the fluid temperature.
12
3.3 The Pipes Operating Limit:
The capability of the heat pipe depends greatly on the operational limitation. The operating
limit gives a perspective of the heat transfer capacity that the heat pipe is capable of accomplishing. These limitations depend greatly on the working fluid, the size of the heat pipe as well
as the wick structure that is been used for a wicked heat pipe.
If the heat transfer would overcome the operating limit, the heat transfer will either decrease or
cease to function for various reasons.
The five major operating limits for a wicked heat pipe is the:
Sonic limit
Entrainment limit
Capillary limit
Vapor pressure limit
Boiling limit
Three of these are linked to the liquid flow (Entrainment, Capillary and Boiling) while the remaining two (Sonic and Vapor pressure) are connected to the vapor flow.
A thermosyphon has only four major operating limits due to the removal of the wick structure:
Sonic limit
Vapor pressure limit
Boiling limit
Flooding limit
For both types:
3.3.1 The Sonic limit:
The vapor between the evaporator and the condenser must not exceed the local speed of sound,
if the vapor velocity is too high the flow will choke. This limits the mass transfer ability and
thereby also the heat transfer capabilities of the heat pipe. The working medium and the crossarea of the vapor section is very important when determining the sonic limitation.
It’s assumed that the vapor flow inside the vapor section is one dimensional.
𝛾∗𝑅0 ∗𝑇𝑚𝑖𝑛𝑖𝑚𝑢𝑚
𝑞𝑠𝑜𝑛𝑖𝑐 = 𝜌𝑣𝑎𝑝𝑜𝑟 𝐿√
2(𝛾+1)
(Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
𝑄𝑠𝑜𝑛𝑖𝑐 = 𝑞𝑠𝑜𝑛𝑖𝑐 ∗ 𝐴𝑟𝑒𝑎𝑣𝑎𝑝𝑜𝑟
For ammonia and water the heat capacity ratio (gamma, γ) can be taken as:
𝐶𝑝
𝛾𝑎𝑚𝑚𝑜𝑛𝑖𝑎 = 𝐶 = 1.4
𝛾𝑤𝑎𝑡𝑒𝑟 = 1.3
𝑉
13
3.3.2 The Vapor Pressure limit (Viscous limit):
The vapor pressure limitation is encountered at low temperatures, when the heat pipe operates
at a temperature below its design. At these low temperatures, the viscous forces are dominant
in the vapor flow and the vapor pressure is very small. (Heat Transfer Limitations of Heat Pipes,
2017)
𝑞𝑣𝑎𝑝𝑜𝑟 =
𝑟𝑖𝑛𝑟𝑒 𝐿∗𝑟ℎ𝑜𝑙𝑖𝑞𝑢𝑖𝑑 𝑃𝑣𝑎𝑝𝑜𝑟
(Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
16𝜇𝑣𝑎𝑝𝑜𝑟 𝑙𝑒𝑓𝑓
𝑃𝑣𝑎𝑝𝑜𝑟 = 4.24 ∗ 105 [𝑃𝑎]
𝑄𝑣𝑎𝑝𝑜𝑟 = 𝑞𝑣𝑎𝑝𝑜𝑟 ∗ 𝐴𝑟𝑒𝑎𝑣𝑎𝑝𝑜𝑟
Specific limitations for Wicked Heat Pipes:
3.3.3 The Entrainment limit:
The vapor and liquid moves in opposite directions which creates a shear force at the liquidvapor interface. In case of very high velocities the liquid particles can be pulled from the structure of the wick and entrain it into the vapor that streams towards the condenser. If too much
liquid is entrained into the vapor flow the evaporator will eventually dry out and the heat pipe
will no longer be functional.
2𝜋𝜌𝑣𝑎𝑝𝑜𝑟 𝐿2 𝜎𝑙
𝑞𝑒𝑛𝑡𝑟𝑎𝑖𝑛𝑚𝑒𝑛𝑡 = √
𝑧
(Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
For water (0.01 – 647°C) the following equations and values can be used to calculate the surface
tension as a function dependent on the temperature:
𝜎𝑙 = 𝐵(1 − 𝑇𝑟 )𝑢 ∗ (1 − 𝑏(1 − 𝑇𝑟 )) (Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
𝑇𝑟 = 𝑇
𝑇𝑚𝑎𝑥
𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙
𝑇𝑐𝑟𝑖𝑡𝑖𝑐𝑎𝑙 = 647.096 [𝐾]
𝐵 = 235.8 ∗ 10−3 [𝑁/𝑚]
𝑏 = 0.625
𝑢 = 1.256
To determine the entrainment limit of the wicked heat pipe the characteristic dimension of the
liquid-water interface must be known. It’s dependent of the mesh that’s used to separate the
two phases from each other inside the pipe. If it’s assumed that a fine mesh is being used the
value of z can be taken as:
𝑧 = 0.036 ∗ 10−3 [𝑚] (Reay, Chapter 4 Design Guide, 2014)
𝑄𝑒𝑛𝑡𝑟𝑎𝑖𝑛𝑚𝑒𝑛𝑡 = 𝑞𝑒𝑛𝑡𝑟𝑎𝑖𝑛𝑚𝑒𝑛𝑡 ∗ 𝐴𝑟𝑒𝑎𝑣𝑎𝑝𝑜𝑟
14
3.3.4 The Boiling Limit:
When a high radial heat flux causes the boiling to occur in the wick structure which results in
that the mass circulation is seriously reduced and the boiling limit is reached. The boiling limit
depends quite a bit on the wick structure that’s being used and for a screen wick the limit is
usually reached at a heat flux of about 5-10 watts per square centimeter. If the wick is made of
powder metal a higher heat flux can be achieved and the boiling limit is only reached at about
20-30 watts per square centimeter. (Dincer, Chapter 7 - Heat Pipes)
Water and non-metallic liquids: 130 kW/m2 (Reay, Chapter 2 Heat transfer and fluid flow
theory, 2014)
𝑄𝐵𝑜𝑖𝑙𝑖𝑛𝑔,𝑤𝑖𝑐𝑘𝑒𝑑 = 𝐴𝑟𝑒𝑎𝐸𝑣𝑎𝑝,𝑠𝑢𝑟𝑓𝑎𝑐𝑒 ∗ 5 ∗ 104 [𝑊]
3.3.5 The Capillary limit (Wick & Fluid limit):
Sometimes called the hydrodynamic limitation, it occurs when the pumping rate within the heat
pipe is insufficient and too little liquid is brought to the evaporation section. It’s the most common limitation for low-temperature heat pipes. It occurs when the capillary pressure doesn’t
meet up to the pressure drops and depends on the working fluid as well as the wick structure
and material. In case the capillary limit is exceeded a dry out will occur in the evaporator. (Heat
Transfer Limitations of Heat Pipes, 2017)
To express the maximum heat flow due to the wick and fluid limitations there is three assumptions that is required. (Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
The liquid properties are constant within the length of the pipe
The wick is uniform along the pipe
The pressure drop that’s caused by the vapor flow is neglected
The maximum capillary pressure drop needs to sum up to the liquid and gravitational pressure
drops:
∆𝑝𝑐 = ∆𝑝𝑙 + ∆𝑝𝑔
The different pressure drops can be calculated and if put together the capillary limit of the
wicked heat pipe can be determined. (Reay, Chapter 2 Heat transfer and fluid flow theory, 2014)
𝐶𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑑𝑟𝑜𝑝: ∆𝑝𝑐 =
𝐿𝑖𝑞𝑢𝑖𝑑 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑑𝑟𝑜𝑝: ∆𝑝𝑙 =
2∗𝜎𝑙 ∗cos 𝜃
𝑟𝑐𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦
𝜇𝑙 𝑄𝑤𝑖𝑐𝑘 𝑙𝑒𝑓𝑓
𝜌𝑙 𝐿𝐴𝑤 𝐾
𝐺𝑟𝑎𝑣𝑖𝑡𝑎𝑡𝑖𝑜𝑛𝑎𝑙 𝑝𝑟𝑒𝑠𝑠𝑢𝑟𝑒 𝑑𝑟𝑜𝑝: ∆𝑝𝑔 = 𝜌𝑙 𝑔ℎ
𝑄𝑐𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦 = (
𝜌𝑙 𝜎𝑙 𝐿
𝑄𝑐𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦 = (
𝜌𝑙 𝜎𝑙 𝐿
𝜇𝑙
)(
𝐾𝐴𝑤𝑖𝑐𝑘
)(
𝐾𝐴𝑤𝑖𝑐𝑘
𝑙
)( −
2
𝜌𝑙 𝑔𝑙
2
𝜌𝑙 𝑔𝑙
𝑟𝑒
𝜎𝑙
sin 𝜙)
Since the evaporator is below the condenser and working at an inclination of 90 degrees to the
horizontal the equation becomes:
𝜇𝑙
𝑙
)( +
𝑟𝑒
𝜎𝑙
)
15
The wick diameter, capillary radius and volume fraction (of the solid phase, ε) are assumed to
be: (Reay, Chapter 4 Design Guide, 2014)
𝑑𝑤𝑖𝑐𝑘 = 0.025 ∗ 10−3 [𝑚]
𝑟𝑐𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦 = 0.029 ∗ 10−3 [𝑚]
𝜀 = 0.314
This wick permeability can then be calculated as:
(1−𝜀)3
2
𝐾 = 𝑑𝑤𝑖𝑐𝑘
66.6∗𝜀 2
Perfect wetting is assumed which means that the contact angle is to be. (Reay, Chapter 2 Heat
transfer and fluid flow theory, 2014)
𝜃 = 0°
The capillary limit can then be calculated as:
𝑚𝑚𝑎𝑥 =
𝜌𝑙𝑖𝑞𝑢𝑖𝑑 𝜎𝑙
𝜇𝑙𝑖𝑞𝑢𝑖𝑑
∗
flow theory, 2014)
𝐾∗𝐴𝑤𝑖𝑐𝑘
𝜇𝑙𝑖𝑞𝑢𝑖𝑑 𝑙𝑒𝑓𝑓
∗(
2𝜎𝑙
𝑟𝑝𝑜𝑟𝑒
− 𝜌𝑙𝑖𝑞𝑢𝑖𝑑 𝑔𝑙𝑒𝑓𝑓 ) (Reay, Chapter 2 Heat transfer and fluid
𝑄𝑤𝑖𝑐𝑘𝑖𝑛𝑔 = 𝑚𝑚𝑎𝑥 ∗ 𝐿
Specific limitations for Thermosyphons:
3.3.6 The Boiling limit:
Just like in a wicked heat pipe the boiling limit can be reached in a thermosyphon even without
the wick. This limitation often occurs in thermosyphons with a high radial heat flux along with
a large liquid fill ratio. It’s based on that the amount of vapor bubbles that’s being created at
the inner surface of the pipe can create a vapor surface that prevents the liquid from coming in
contact with the surface of the evaporator region. (Dr. Ahmad, 2013)
𝑑 = 2𝑟𝑜𝑢𝑡𝑒𝑟
𝐴𝑟𝑒𝑎𝑟𝑎𝑑𝑖𝑎𝑙 = 𝜋𝑑𝑙𝑒𝑣𝑎𝑝
𝐾𝑢𝐵𝐿 = 0.16 (1 − 𝑒
(
−𝑑
𝑙𝑒𝑣𝑎𝑝
𝜌𝑙𝑖𝑞𝑢𝑖𝑑 0.13
)(
𝜌𝑣𝑎𝑝𝑜𝑟
)
)
𝑄𝐵𝑜𝑖𝑙 = 𝐴𝑟𝑒𝑎𝑟𝑎𝑑𝑖𝑎𝑙 𝐿√𝜌𝑣𝑎𝑝𝑜𝑟 𝑔𝜎𝑙 (𝜌𝑙𝑖𝑞𝑢𝑖𝑑 − 𝜌𝑣𝑎𝑝𝑜𝑟 )
0.25
𝐾𝑢𝐵𝐿
3.3.7 The Flooding limit:
The flooding limit (also called the counter current flow) is a very important limitation that’s
often the dominant one for thermosyphons that operates with a high fillers ratio along with a
16
high axial heat flux and a small radial heat flux. When it’s overstepped then the condensate
from can’t return to the evaporator region due to the vapor shear within the pipe. (Dr. Ahmad,
2013)
2
𝐴𝑟𝑒𝑎𝐴𝑥𝑖𝑎𝑙 = 𝜋𝑟𝑜𝑢𝑡𝑒𝑟
𝐵𝑜 = 𝑑√
𝑔(𝜌𝑙𝑖𝑞𝑢𝑖𝑑 −𝜌𝑣𝑎𝑝𝑜𝑟 )
𝜌𝑙𝑖𝑞𝑢𝑖𝑑
𝐾𝑢𝑐𝑐𝑙 = (
𝜌𝑣𝑎𝑝𝑜𝑟
𝜎𝑙
0.14
)
∗ tanh(𝐵𝑜0.25 )2
𝑄𝐹𝑙𝑜𝑜𝑑 = 𝐾𝑢𝑐𝑐𝑙 𝐿𝐴𝑟𝑒𝑎𝐴𝑥𝑖𝑎𝑙 𝑔𝜎𝑙 (𝜌𝑙𝑖𝑞𝑢𝑖𝑑 − 𝜌𝑣𝑎𝑝𝑜𝑟 )
0.25
−0.25
−0.25
(𝜌𝑣𝑎𝑝𝑜𝑟
− 𝜌𝑙𝑖𝑞𝑢𝑖𝑑
)
−2
3.3.8 The maximal heat transfer capacity of the heat pipe and the selection:
The lowest limitation is the dominating one and determines the maximum heat transfer that the
pipe can accomplish. The dominating limitation is the lowest limitation of the ones that effect
either the heat pipe or the thermosyphon: (Reay, Chapter 2 Heat transfer and fluid flow theory,
2014), (Dr. Ahmad, 2013)
The wicked heat pipe:
𝑄𝑙𝑖𝑚𝑖𝑡 = 𝑚𝑖𝑛(𝑄𝑆𝑜𝑛𝑖𝑐 , 𝑄𝐸𝑛𝑡𝑟𝑎𝑖𝑛𝑚𝑒𝑛𝑡 , 𝑄𝐶𝑎𝑝𝑖𝑙𝑙𝑎𝑟𝑦 , 𝑄𝑉𝑎𝑝𝑜𝑟 , 𝑄𝐵𝑜𝑖𝑙𝑖𝑛𝑔 )
The thermosyphon:
𝑄𝑙𝑖𝑚𝑖𝑡 = 𝑚𝑖𝑛(𝑄𝑆𝑜𝑛𝑖𝑐 , 𝑄𝐵𝑜𝑖𝑙𝑖𝑛𝑔 , 𝑄𝐹𝑙𝑜𝑜𝑑𝑖𝑛𝑔 , 𝑄𝑉𝑎𝑝𝑜𝑟 )
In the simulation, this is considered in the following fashion:
𝑖𝑓 𝑄𝑃𝑖𝑝𝑒 > 𝑄𝑙𝑖𝑚𝑖𝑡 → 𝑄𝑃𝑖𝑝𝑒 = 𝑄𝑙𝑖𝑚𝑖𝑡
This prevents the pipes from providing heat transfer that over exceeds their capability. However, the processes will be designed to make sure that the heat transfer does not reach the operating limit. The easiest way to accomplish this is to increase the number of heat pipes since this
will lower the temperature rise as well as the heat transfer coefficient.
In general, the thermosyphons is capable to transferring more energy than the wicked heat pipe
due to the capillary limit and therefore thermosyphons have been determined to be used for the
simulations. A study that performed a comparison on the overall heat transfer coefficient of a
wicked heat pipe and a thermosiphon evaluated however that the wicked heat pipe possessed
the better coefficient. A smooth surfaced wicked heat pipe and a thermosiphon of equal dimensions and during the same heat transfer (700 W) obtained an overall heat transfer coefficient of
9950 respectively 4950 W/m2*K. The difference between the two values is significant considering that the heat transfer coefficient of the wicked heat pipe is more than twice that of the
thermosiphon. Since this study doesn’t evaluate different heat transfer coefficients for the two
kinds of heat pipes this is however neglected though it’s noted as something that needs to be
examined later. Another thing that’s worth to take into consideration is that the wicked heat
pipe required a much larger amount of working fluid to gain its optimal heat transfer coefficient
than the thermosiphon. This could have an important effect on the time that it takes for the
17
working fluid to reach its saturation temperature which is something that will be discussed further on. (Dr. Ahmad, 2013)
3.4 Simulations Models:
3.4.1 The Steinbach Case
It is now time to make an overview of the different processes that is simulated and examined.
The first one is an exothermic reaction that has been used for studies regarding thermal runaways before.
The model consists of a reaction that occurs within the semi-batch reactor. The reaction is irreversible, of second order, homogeneous, carried out in liquid phase and does not involve any
phase changes.
The reaction formula is:
𝐶𝐴 + 𝐶𝐵 → 𝐶𝐶
Table 3. A table that displays the initial conditions of the reactor model. (Mas, 2006)
Parameter
Initial Condition
CA,0
3400 [mol/m3]
CB,0
0 [mol/m3]
CC,0
0 [mol/m3]
V0
10.43 m3
T0
263 K
A continuous and constant feed streams into the reactor during the dosing period and the reactor
reaches a final volume after a long dosing period of 18 000 seconds.
𝑉𝑓𝑖𝑛𝑎𝑙 = 17.73 [𝑚3 ]
𝑡𝑑𝑜𝑠𝑖𝑛𝑔 = 18 000 [𝑠]
The volumetric flow is calculated using the volume changes during the process cycle and the
dosing time.
𝑣𝐹𝑒𝑒𝑑 =
𝑉𝑓𝑖𝑛𝑎𝑙 −𝑉0
𝑡𝑑𝑜𝑠𝑖𝑛𝑔
[𝑚3 /𝑠]
Something that is interesting with this is that the final reactor volume (20 m 3) is not reached
during the entire cycle. This means that there will always be an adiabatic region of the thermosyphons that’s used to cool the process.
18
The feed stream consists purely of component B and is dosed into the reactor with a constant
concentration during the entire dosing period.
The reaction kinetics:
𝑟 = 𝑘𝐶𝐴 𝐶𝐵
−𝐸
𝑘 = 𝑘0 𝑒 (𝑅∗𝑇)
Table 4. The reaction kinetics and volumetric heat capacity of the reactor fluid. (Mas, 2006)
Reaction Parameter
Value
Frequency Factor (k0)
4.43*109 m3/mol*s
Activation Energy (E)
86 881.3 J/mol
Reaction enthalpy (ΔHreaction)
85 000 J/mol
Volumetric heat capacity (ρcp)
1700 kJ/m3*K
The concentration of the feed (CB,feed)
4860 mol/m3
Differential equations for concentrations and volume:
𝑑𝐶𝐴
𝑑𝑡
𝑑𝐶𝐵
𝑑𝑡
𝑑𝐶𝐶
𝑑𝑡
= −𝑟 −
= −𝑟 −
=𝑟−
𝐶𝐴 𝑣𝐹𝑒𝑒𝑑
𝑉
𝑣𝐹𝑒𝑒𝑑 𝐶𝐵
𝑉
𝑣𝐹𝑒𝑒𝑑 ∗𝐶𝐶
𝑉
0 < 𝑡 < 𝑡𝑑𝑜𝑠𝑖𝑛𝑔 →
𝑡 > 𝑡𝑑𝑜𝑠𝑖𝑛𝑔 →
𝑑𝑉
𝑑𝑡
+
𝑑𝑉
𝑑𝑡
=0
𝐶𝐵,𝑓𝑒𝑒𝑑 𝑣𝐹𝑒𝑒𝑑
𝑉
=𝑣
3.4.1.1 The cooling systems:
3.4.1.1.1 Cooling jacket:
The reactor is assumed to be both adiabatic and isothermal. The cooling capacity is also assumed to have a constant heat transfer factor (that doesn’t change as the surface area increases)
and that’s set as: (Mas, 2006)
𝑈𝐴0 = 6000 𝑊/𝐾
𝑇𝑓𝑒𝑒𝑑 = 298 °𝐾
19
By summarizing the energy that the exothermic reaction releases along with the heat transfer
of the cooling jacket and the energy that’s required to warm up the cold feed the differential
equation for the reactor temperature can be written as:
𝑑𝑇
𝑑𝑡
=
𝑟∗∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
𝑐𝑝 𝜌
−
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
−
𝑈𝐴0 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
𝑉𝑐𝑝 𝜌
3.4.1.1.2 A Cooling jacket along with thermosyphons
A simplified simulation where the thermosyphons is assumed to reach the boiling temperature
at the same moment as the reactor enables the following model to be used for thermosyphons
operating in combination with a cooling jacket.
𝐼𝑓: 𝑇𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 ≥ 𝑇𝑏𝑜𝑖𝑙𝑖𝑛𝑔
𝑑𝑇
𝑑𝑡
=
𝑟∗𝑑𝐻𝑟𝑒𝑎𝑘𝑡𝑖𝑜𝑛
𝑐𝑝 𝜌
−
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
−
𝑈𝐴0 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
−
𝑈𝐴0 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
𝑉𝑐𝑝 𝜌
𝑄𝑃𝑖𝑝𝑒 = 𝑘𝑃𝑖𝑝𝑒 𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 (𝑇 − 𝑇𝐵𝑜𝑖𝑙𝑖𝑛𝑔 )
− 𝑁𝑢𝑚𝑏𝑒𝑟 ∗
𝑄𝑃𝑖𝑝𝑒
𝑉𝜌𝐶𝑝
𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 = 𝜋2𝑟𝑜𝑢𝑡𝑒𝑟 𝑙𝑒𝑣𝑎𝑝
𝑙𝑒𝑣𝑎𝑝 =
𝑉
2
𝜋𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟
𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟 = (
𝑉𝑇𝑜𝑡𝑎𝑙 1⁄3
𝜋
𝐼𝑓: 𝑇𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 < 𝑇𝑏𝑜𝑖𝑙𝑖𝑛𝑔
𝑑𝑇
𝑑𝑡
=
𝑟∗𝑑𝐻𝑟𝑒𝑎𝑘𝑡𝑖𝑜𝑛
𝑐𝑝 𝜌
−
)
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
𝑉𝑐𝑝 𝜌
3.4.2 Hydrolysis of Acetic Anhydride:
The second process that’s to be simulated and examined is a hydrolysis of acetic anhydride
which produces acetic acid. Acetic anhydride is a common component that’s used in the production of products such as aspirin, organic synthesis, explosives and of course acetic acid. The
reaction is exothermic with a reaction enthalpy more than twice as high as the one that was used
in the previous model. Moreover, acetic anhydride is a dangerous component as it’s a highly
flammable liquid, can release toxic vapor when in gas phase. Besides this mixtures of acetic
anhydride and air can become explosive at temperatures above 322 degrees Kelvin. (García,
2016)1
The process that will be simulated is a hydrolysis reaction of acetic anhydride in a semi-batch
reactor. Just like in the previous model the reaction that occurs within the semi-batch reactor in
the simulation is irreversible, homogeneous, liquid phase, doesn’t involve any phase changes
and of second order. The reactor size and the reaction kinetics is based on a previous study
1
García, M. Thermal stability and dynamic analysis of the acetic anhydride hydrolysis reaction, Elsevier, Chemical Engineering Science, Vol. 142, 2016.
20
regarding dimensioning and simulations of batch reactors. (Bjerle, I , Berggren, J-C och
Karlson, H., 1977)
The reaction formula is:
(𝐶𝐻3 𝐶𝑂)2 𝑂 + 𝐻2 𝑂 → 2𝐶𝐻3 𝐶𝑂𝑂𝐻
The reactor that’s simulated in the process has a full capacity of 235 liters and the dosing time
is determined to be 20 minutes. During the dosing, a full amount of 235 mol of acetic anhydride
is to be fed into the reactor.
Table 5. The density and molar weight of the two reactants along with the final volume of the
semi-batch reactor. (Bjerle, I , Berggren, J-C och Karlson, H., 1977)
Parameter
Value
ρAcetic Anhydride
1082 kg/m3
ρWater
1000 kg/m3
MAcetic Anhydride
102.09 kg/kmol
MWater
18 kg/kmol
VFinal
235 dm3
nAcAN
235 mol
𝜌
𝐶𝐴𝑐𝐴𝑛,𝑓𝑒𝑒𝑑 = 𝑀𝐴𝑐 𝐴𝑛 = 10.599 ≈ 10.6
𝐴𝑐 𝐴𝑛
𝑡𝑑𝑜𝑠𝑖𝑛𝑔 = 20 [𝑚𝑖𝑛]
𝑛𝐴𝑐𝐴𝑛
𝑣𝐹𝑒𝑒𝑑 = 𝐶
𝐴𝑐𝐴𝑛,𝑓𝑒𝑒𝑑 ∗𝑡𝐷𝑜𝑠𝑖𝑛𝑔
𝑘𝑚𝑜𝑙
𝑚3
= 1.109 ∗ 10−3
𝑚3
𝑚𝑖𝑛
Before the dosing cycle begins the reactor is filled with 213 liters of pure water. This enables
the last of the initial condition to be determined as:
𝜌
𝐶𝑊𝑎𝑡𝑒𝑟,0 = 𝑀𝑊𝑎 = 55.56
𝑊𝑎
𝑘𝑚𝑜𝑙
𝑚3
Table 6. The initial conditions of the reaction process.
Parameter
Initial Condition
CAcetic Anhydride,0
0 mol/m3
CWater,0
55.56 kmol/m3
21
CAcetic Acid,0
0 mol/m3
V0
213 dm3
T0
303 K
The volumetric flow:
0 < 𝑡 < 𝑡𝑑𝑜𝑠𝑖𝑛𝑔 → 𝑣𝐹𝑒𝑒𝑑 =
𝑡 > 𝑡𝑑𝑜𝑠𝑖𝑛𝑔 → 𝑣𝐹𝑒𝑒𝑑 = 0
𝑑𝑉
𝑑𝑡
𝑉𝐹𝑖𝑛𝑎𝑙 −𝑉0
𝑡𝑑𝑜𝑠𝑖𝑛𝑔
= 𝑣𝐹𝑒𝑒𝑑
The reaction kinetics:
𝑟 = 𝑘𝐶𝐴
−𝐸𝐴
𝑘 = 𝑘0 𝑒 ( 𝑅∗𝑇
)
Table 7. The reaction kinetics that’s used to simulate the reaction. (Bjerle, I , Berggren, J-C
och Karlson, H., 1977)
Reaction Parameter
Value
Frequency Factor (k0)
7.94*107 [m3/mol*min]
(EA/R)
5949 [K]
Reaction enthalpy (ΔHreaction)
2.1*105 [J/mol]
In the semi batch reactor, the differential equations for component concentrations can be written
as:
𝑑𝐶𝐴𝑐𝐴𝑛
𝑑𝑡
𝑑𝐶𝑊𝑎
𝑑𝑡
(𝐶𝐴𝑐𝐴𝑛,𝑓𝑒𝑒𝑑 −𝐶𝐴𝑐𝐴𝑛 )
= −𝑟 −
𝑑𝐶𝐴𝑐𝐴𝑐
𝑑𝑡
=
=𝑟−
𝑉
𝐶𝑊𝑎
𝑉
𝐶𝐴𝑐𝐴𝑐
𝑉
𝑣𝐹𝑒𝑒𝑑
𝑣𝐹𝑒𝑒𝑑 − 𝑟
𝑣𝐹𝑒𝑒𝑑
22
3.4.2.1 The cooling system:
3.4.2.1.1 Cooling jacket:
The coolant temperature and the overall heat coefficient that determines the heat transfer from
the reactor to the heat exchanger is assumed to have constant values that’s set as:
𝑈𝑉𝑉𝑋 = 900 [𝑊 ⁄𝑚2 𝐾 ]
𝑇𝑓𝑒𝑒𝑑 = 303 °𝐾
𝑇𝑐𝑜𝑜𝑙𝑒𝑟 = 303 °𝐾
Unlike the previous simulation, the surface area of the cooling jacket is not assumed to be constant. The surface area that’s in contact with the liquid inside the reactor is assumed to grow
with the volume. The surface area of the cooling jacket is calculated in accordance to the fluid
volume under the assumption that the radius and the height of the final reactor volume remains
equal.
𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟 = (
𝑉𝑇𝑜𝑡𝑎𝑙 1⁄3
𝜋
)
𝐻𝑒𝑖𝑔ℎ𝑡𝑅𝑒𝑎𝑐𝑡𝑜𝑟 = 𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟
𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 = 𝐷𝑖𝑎𝑚𝑒𝑡𝑒𝑟𝑅𝑒𝑎𝑐𝑡𝑜𝑟 𝐻𝑒𝑖𝑔ℎ𝑡𝑅𝑒𝑎𝑐𝑡𝑜𝑟 𝜋
Similarly, to before the differential equation for the reaction temperature can be written as:
𝑑𝑇
𝑑𝑡
=
𝑟𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
𝑚𝑐𝑝
−
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
−
𝑈𝑉𝑉𝑋 𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
𝑚𝑐𝑝
3.4.2.1.2 Cooling jacket and Thermosyphons:
The same assumptions as was used previously is made. The thermosyphons doesn’t provide
any heat transfer up until they reach the boiling temperature which they are assumed to do at
the same time as the reactor.
𝐼𝑓: 𝑇𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 ≥ 𝑇𝑏𝑜𝑖𝑙𝑖𝑛𝑔
𝑑𝑇
𝑑𝑡
=
𝑟𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
𝑚𝑐𝑝
−
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
−
𝑈𝑉𝑉𝑋 𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
𝑄𝑃𝑖𝑝𝑒 = 𝑘𝑃𝑖𝑝𝑒 𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 (𝑇 − 𝑇𝑏𝑜𝑖𝑙𝑖𝑛𝑔 )
𝑚𝑐𝑝
𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 = 𝜋2𝑟𝑜𝑢𝑡𝑒𝑟 𝑙𝑒𝑣𝑎𝑝
𝑙𝑒𝑣𝑎𝑝 =
𝑉
2
𝜋𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟
𝐼𝑓: 𝑇𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 < 𝑇𝑏𝑜𝑖𝑙𝑖𝑛𝑔
23
− 𝑁𝑢𝑚𝑏𝑒𝑟 ∗
𝑄𝑃𝑖𝑝𝑒
𝑚𝑐𝑝
𝑟𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛
𝑚𝑐𝑝
−
𝑣(𝑇−𝑇𝑓𝑒𝑒𝑑 )
𝑉
−
𝑈𝑉𝑉𝑋 𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 (𝑇−𝑇𝑐𝑜𝑜𝑙𝑒𝑟 )
𝑚𝑐𝑝
3.4.3 Fine Chemicals:
In processes that produces bulk chemicals and fine chemicals there is several common differences. Fine chemicals rely not as heavily on catalysis as bulk chemicals do and the processes
usually involves several side reactions that can result in the formation of large amounts of byproducts such as inorganic salts. Besides this the purity of the fine chemicals is also of great
interest which means that the by-products need to be removed through several separations steps
following the reactor. The reaction rate of the side reactions increases often in correlation with
the temperature which means that a large portion of the by-products is produced during a hotspot period if the process occurs in a batch or semi-batch reactor. The by-products can also
bring a potential risk to the process in case they are strongly exothermic, which is quite often
the case. By using the additional cooling that’s supplied by the pipes the reaction rate of undesired side reactions can be halted which would result that the reactor produces a product of
higher purity which could potentially decrease the amount of separation steps that’s necessary
afterwards. Besides this the decreased reaction rate of the side reactions could also limit the
energy that’s being released within the reactor and increase the loss prevention of the process.
A simulation of a relatively simple theoretical process is to be used as a test to see the result of
the additional cooling of the pipes. The process is set up as a desired reaction of first order
followed by an undesired side reaction of the product, for instance a decomposition reaction.
The component named A, P and S is the reactant, the desired product, and the undesired product.
Both the reactions are highly exothermic and the first reaction is relatively slow while the second is faster.
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 1: 𝐴 → 𝑃
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 2: 𝑃 → 𝑆
Table 8. A table displaying the reaction kinetics, the physical parameters and the as the
parameters for determining the heat transfer of the cooling jacket.
Parameter
Value
k0,1
0.5 s-1
E1
20 000 J/mol
ΔHr,1
-300 000 J/mol
k0,2
1011 s-1
E2
100 000 J/mol
ΔHr,1
-250 000 J/mol
ρ
1000 kg/m3
cp
4000 J/kg*K
24
V
6.3 m3
AreaJacket
16.38 m2
UCooling Jacket
500 W/m2*K
TCool
305 K
The reaction kinetics of the two reactions is taken into the Arrhenius equation to determine the
different reaction rates.
−𝐸𝑖
𝑘𝑖 = 𝑘0,𝑖 𝑒 ( 𝑅𝑇 )
𝑟1 = 𝑘1 𝐶𝐴
, 𝑟2 = 𝑘2 𝐶𝑃
Since the reaction is performed in a batch reactor the differential equations are only effected by
the reactions and not by feeding or volume changes.
𝑑𝐶𝐴
𝑑𝑡
𝑑𝐶𝑃
𝑑𝑡
𝑑𝐶𝑆
𝑑𝑡
= −𝑟1
= 𝑟1 − 𝑟2
= 𝑟2
Considering that there is more than one reaction in this process the selectivity of the reaction
can be examined. The selectivity functions as a measurement of how many reactants or products
that was wasted by the undesired side reactions.
𝑆𝑡𝑜𝑖𝑐ℎ𝑖𝑜𝑚𝑒𝑡𝑟𝑖𝑐 𝑓𝑎𝑐𝑡𝑜𝑟 = 1 → 𝑆𝑒𝑙𝑒𝑐𝑡𝑖𝑣𝑖𝑡𝑦 =
𝐷𝑒𝑠𝑖𝑟𝑒𝑑 𝑃𝑟𝑜𝑑𝑢𝑐𝑡 𝑃𝑟𝑜𝑑𝑢𝑐𝑒𝑑
𝑅𝑒𝑎𝑐𝑡𝑎𝑛𝑡𝑠 𝐶𝑜𝑛𝑠𝑢𝑚𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑅𝑒𝑎𝑐𝑡𝑜𝑟
Another difference with this simulation from the previous ones is that the radius and height of
the reactor isn’t the same anymore. The reactor will have a much larger height instead of radius
which will aid both the cooling of the surface jacket as well as the pipes.
Table 9. The initial conditions of the simulated model.
Parameter
Initial Condition
CA,0
1000 mol/m3
CP,0
0 kmol/m3
CS,0
0 mol/m3
25
295 K
T0
3.4.3.1 The cooling System:
3.4.3.1.1 Cooling Jacket:
The reactor is assumed to be adiabatic and isothermal. This allows the following equation to be
used to determine the temperature for a reactor that’s cooled by a cooling jacket:
𝑑𝑇
𝑑𝑡
=
𝑟2 𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛,1
𝑚𝑐𝑝
−
𝑟2 𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛,2
𝑚𝑐𝑝
−
𝑈𝑉𝑉𝑋 𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 (𝑇−𝑇𝑐𝑜𝑜𝑙 )
𝑚𝑐𝑝
Since the reaction is performed in a batch reactor instead of a semi-batch reactor the energy lost
to heat the cold feed is no longer in the equation. There are also now two exothermic reactions
that can release energy within the reactor.
3.4.3.1.2 The Cooling Jacket + Thermosyphons
When the additional cooling of the thermosyphons is added to the process the differential equation can be written as:
𝑑𝑇
𝑑𝑡
=
𝑟2 𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛,1
𝑚𝑐𝑝
−
𝑟2 𝑉∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛,2
𝑚𝑐𝑝
−
𝑈𝑉𝑉𝑋 𝐴𝑟𝑒𝑎𝐽𝑎𝑐𝑘𝑒𝑡 (𝑇−𝑇𝑐𝑜𝑜𝑙 )
𝑚𝑐𝑝
− 𝑁𝑢𝑚𝑏𝑒𝑟 ∗
𝑄𝑃𝑖𝑝𝑒
𝑚𝑐𝑝
The working fluid within the thermosyphon is water with a vapor pressure at 70 mbar, which
gives the water a boiling temperature of 313 degrees Kelvin. This allows the heat transfer that
each pipe performs to be calculated as:
𝑄𝑃𝑖𝑝𝑒 = 𝑘𝑃𝑖𝑝𝑒 𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 (𝑇 − 𝑇𝐵𝑜𝑖𝑙 )
𝐴𝑟𝑒𝑎𝑃𝑖𝑝𝑒 = 𝜋2𝑟𝑜𝑢𝑡𝑒𝑟 𝑙𝑒𝑣𝑎𝑝
𝑙𝑒𝑣𝑎𝑝 =
𝑉
2
𝜋𝑅𝑎𝑑𝑖𝑢𝑠𝑅𝑒𝑎𝑐𝑡𝑜𝑟
One difference between this model and the previous two is however that the temperature within
the pipe is going to be examined as well. This is mainly to observe the time that’s required to
bring the working fluid up to its saturation temperature. The starting temperature of the working
fluid is assumed to be the same as the temperature of the feed. Before the working fluid has
reached its boiling temperature, the heat transfer will be much lower since there is no forced
convection inside the pipe without the evaporation. This means that the convective heat transfer
within the pipe is first in the form of a free convection before it’s turned into a forced convection. To simulate this an adjustment to the calculation of the overall heat transfer coefficient is
required since before the boiling initiates the heat transfer coefficient within the thermosyphon
is going to be dominant.
The overall heat transfer coefficient is calculated using the following equation. (Alveteg, 2013)
26
1
𝑘𝑡𝑜𝑡𝑎𝑙
=
𝑑
( 𝑖𝑛𝑛𝑒𝑟 )
𝑑𝑜𝑢𝑡𝑒𝑟
𝛼𝑤𝑖𝑡ℎ𝑖𝑛
+
𝑑
𝑑𝑜𝑢𝑡𝑒𝑟 ln( 𝑜𝑢𝑡𝑒𝑟 )
𝑑𝑖𝑛𝑛𝑒𝑟
2𝜆𝑐𝑜𝑝𝑝𝑒𝑟
+𝛼
1
𝑜𝑢𝑡𝑠𝑖𝑑𝑒
The inner diameter is determined using the inner and outer radius of the thermosyphon that’s
used in the simulation. The heat transfer coefficient from outside the thermosyphon is calculated using the Chilton-Colburn Analogy and the thermal conductivity of the pipe material (copper) is assumed as a constant. The heat transfer coefficient of the working fluid within the pipe
is given two constant values. The first constant is used during the boil up period and when the
saturation temperature is reached the second constant is taken in as the new heat transfer coefficient. The values of the constants that has been chosen can be seen in Table 10.
Table 10. The constant heat transfer coefficients of the working fluid along with the thermal
conductivity of the pipe metal.
Heat transfer coefficient
αBoil up
300 W/m2*K
αBoiling
15 000 W/m2*K
Thermal Conductivity
λcopper
394 W/m*K
𝑇𝑤𝑖𝑡ℎ𝑖𝑛 < 𝑇𝐵𝑜𝑖𝑙𝑖𝑛𝑔 𝑝𝑜𝑖𝑛𝑡
𝑑𝑇𝑤𝑖𝑡ℎ𝑖𝑛
𝑑𝑡
=
𝑘𝑡𝑜𝑡 𝐴𝑟𝑒𝑎𝑒𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑜𝑟 (𝑇−𝑇𝑤𝑖𝑡ℎ𝑖𝑛 )
𝑚𝑡ℎ𝑒𝑟𝑚𝑜𝑠𝑦𝑝ℎ𝑜𝑛 𝐶𝑝,𝑤𝑎𝑡𝑒𝑟
𝑇𝑤𝑖𝑡ℎ𝑖𝑛 ≥ 𝑇𝐵𝑜𝑖𝑙𝑖𝑛𝑔 𝑝𝑜𝑖𝑛𝑡
𝑑𝑇𝑤𝑖𝑡ℎ𝑖𝑛
𝑑𝑡
= 0
The amount of water that’s suitable to be within the thermosyphon is assumed in accordance to
a study that indicated that lower filler ratios of the evaporator region resulted in an improved
heat transfer. Though since not very many different filler ratios were tried out a slightly larger
ratio has been selected for this simulation. The reason for this is that the larger amount of water
should slow the boiling up process and that if the boiling up period is acceptable at high filler
ratios it should be also be reasonable at lower ones. (Dr. Ahmad, 2013)
𝑚𝑡ℎ𝑒𝑟𝑚𝑜𝑠𝑦𝑝ℎ𝑜𝑛 = 0.45𝑉𝑃𝑖𝑝𝑒 𝜌𝑤𝑎𝑡𝑒𝑟
2
2
= ℎ𝑟𝑒𝑎𝑐𝑡𝑜𝑟 𝜋𝑟𝑜𝑢𝑡𝑒𝑟
𝑉𝑃𝑖𝑝𝑒 = 𝑙𝑒𝑣𝑎𝑝,𝑓𝑖𝑛𝑎𝑙 𝜋𝑟𝑜𝑢𝑡𝑒𝑟
3.5 A Worst-Case Scenario
As was mentioned previously in chapter 3 a simulation test would be performed where the
cooling jacket is no longer operational yet the heat pipes continues to function. This will be
27
referred to as a worst-case scenario and will be simulated for the models involving semi-batch
reactors. The entirety of the scenario is that it’s based on a situation where the cooling of the
thermal jacket is no longer functional yet no emergency actions (such as stopping the feed flow)
is taken. The thermal stability of the process will then be analyzed and it shall be evaluated
whether thermosyphons can prevent the thermal runaway.
To compare the probability of a thermal runaway between an ordinary semi-batch reactor and
one that’s additionally cooled by thermosyphons a fault tree analysis has been made. Considering that no statistics have been gathered and that the scenario that has been set up is purely
theoretical the exact frequency cannot be evaluated yet a comparison is still possible.
The ordinary semi-batch reactor is estimated to come with several safety precautions such as
emergency cooling as well as both an automatic and a manual possibility of stopping the dosing.
A fault tree has then been made to estimate what kinds of events that is required to trigger a
thermal runaway, see Figure 10.
28
A Runaway Reaction
A
High temperature within the
reactor
No Emergency actions
B
C
Cooling Malfunctions
Fouling of the Cooling
Jacket
No emergency cooling
Continued Dosing
2
1
D
Malfunction:
Pump
E
Malfunction:
Control system
3
Automatic regulations fail
Manual regulations fail
F
G
4
Malfunction:
Automatic Control
Malfunction:
temperature sensor
Malfunction:
Manual Control
6
5
Operator doesn t act
7
H
Malfunction:
Alarm
Sleeping operator
8
9
Figure 10. A fault tree that displays the base events that can trigger a thermal runaway for an
ordinary semi-batch reactor.
A Minimal Cut Set (MCS):
To evaluate the most frequent and thereby the most dangerous set of events that can trigger a
thermal runaway all the cut sets from the fault tree above needs to be determined. A cut set is
the determination of the number of basic events that needs to occur to reach the final event, in
this case a runaway reaction. One example of a cut set would be in case the following base
events would happen:
29
The fouling in the cooling jacket causes the temperature to rise to a dangerous level (Base
event 1)
There is no emergency cooling that can be applied (base event 2)
The automatic and manual regulations fail because of two separate malfunctions (base event
5 and 7)
Together these four base events create a situation where a runaway reaction could be possible and the possibility is named CS1. All the cut sets that can trigger the final event has
been calculated and named in Table 10.
Table 11. A display of all the cut scenes that can lead to a thermal runaway.
CS1=1,2,5,7
CS2=1,2,5,8
CS3=1,2,5,9
CS4=1,2,6,7
CS5=1,2,6,8
CS6=1,2,6,9
CS7=3,2,5,7
CS8=3,2,5,8
CS9=3,2,5,9
CS10=3,2,6,7
CS11=3,2,6,8
CS12=3,2,6,9
CS13=4,2,5,7
CS14=4,2,5,8
CS15=4,2,5,9
CS16=4,2,6,7
CS17=4,2,6,8
CS18=4,2,6,9
The cut sets are then ranked in accordance to the number of base events that’s required and the
possibility of each base event. In the table above all the cut sets consist of four base events and
thereby it’s the possibility of the separate base events that’ll determine the MCS.
Without any direct experiments or statistics, the base events will be ranked in possibility in
accordance to the kind of error it is. Human errors are most common and besides this the frequency of errors in active components (such as pumps) is also more common than errors among
passive components (such as tanks). (Karlsson, 2012)
Therefor the frequency is ranked as:
Frequency: Human errors > Error in an active component > Error in a passive component
Through this CS3, CS9, CS15 along with CS6, CS12 and CS18 can be evaluated as the minimal
cut set and therefor the events that has the highest risk of triggering the runaway reaction. The
reason for this is simply that they all involves the possibility of a human error and the only
difference between them is their errors in an active component.
If now a new scenario is visualized. It’s a semi-batch reactor just like in the previous scenario
yet beside the cooling jacket it’s also cooled by installed thermosyphons. The condensation
region of the thermosyphons is cooled by a heat exchanger that is driven by a separate pump
and a different regulation system than the cooling jacket. In case the thermosyphons would be
able to provide the cooling that’s required to prevent a runaway reaction a new fault tree could
be written, see Figure 11. The difference with the previous one is that this fault tree now involves two actions that’s required for the reactor temperature to reach a dangerous level. Both
the cooling jacket and the thermosyphons needs to be out of order or there will be enough
cooling capacity of the process will be enough to prevent the runaway.
30
A Runaway Reaction
A
High temperature within
the reactor
No Emergency actions
X
C
Malfunction:
Thermosyphon
Malfunction:
Cooling Jacket
No emergency
cooling
Continued Dosing
2
Y
B
E
Overstepped
Operating limit
Insufficient condensation
Cooling Malfunctions
Fouling of the
Cooling Jacket
Automatic regulations
fail
Manual regulations fail
F
G
10
1
Z
D
Malfunction (ex. pump or
control system)
Fouling in
Condensor HEX
Malfunction:
Automatic Control
Malfunction:
Pump
11
Malfunction:
Pump 2
5
3
V
Malfunction:
Temperature sensor
Operator doesn t act
7
6
4
H
Malfunction:
Control system 2
12
Malfunction:
Manual Control
Malfunction:
Control system
Malfunction:
Alarm
Sleeping operator
13
8
9
Figure 11. A fault tree presenting the base events that can cause a thermal runaway to occur
in a semi-batch reactor that’s cooled with both a cooling jacket and thermosyphons.
Table 12. All the separate cut sets that can trigger a runaway reaction in the system that’s
presented. The minimal cut scenes are marked out as red.
CS1=
1,2,5,7,10
CS2=
1,2,5,8,10
CS3=
1,2,5,9,10
CS4=
1,2,6,7,10
CS5=
1,2,6,8,10
CS6=
1,2,6,9,10
CS7=
3,2,5,7,10
CS8=
3,2,5,8,10
CS9=
3,2,5,9,10
CS10=
3,2,6,7,10
CS11=
3,2,6,8,10
CS12=
3,2,6,9,10
CS13=
4,2,5,7,10
CS14=
4,2,5,8,10
CS15=
4,2,5,9,10
CS16=
4,2,6,7,10
CS17=
4,2,6,8,10
CS18=
4,2,6,9,10
CS19=
1,2,5,7,11
CS20=
1,2,5,8,11
CS21=
1,2,5,9,11
CS22=
1,2,6,7,11
CS23=
1,2,6,8,11
CS24=
1,2,6,9,11
CS25=
3,2,5,7,11
CS26=
3,2,5,8,11
CS27=
3,2,5,9,11
CS28=
3,2,6,7,11
CS29=
3,2,6,8,11
CS30=
3,2,6,9,11
31
CS33=
4,2,5,7,11
CS34=
4,2,5,8,11
CS34=
4,2,5,9,11
CS135=
4,2,6,7,11
CS36=
4,2,6,8,11
CS37=
4,2,6,9,11
CS38=
1,2,5,7,12
CS39=
1,2,5,8,12
CS40=
1,2,5,9,12
CS41=
1,2,6,7,12
CS42=
1,2,6,8,12
CS43=
1,2,6,9,12
CS44=
3,2,5,7,12
CS45=
3,2,5,8,12
CS46=
3,2,5,9,12
CS47=
3,2,6,7,12
CS48=
3,2,6,8,12
CS49=
3,2,6,9,12
CS50=
4,2,5,7,12
CS51=
4,2,5,8,12
CS52=
4,2,5,9,12
CS53=
4,2,6,7,12
CS54=
4,2,6,8,12
CS55=
4,2,6,9,12
CS56=
1,2,5,7,13
CS57=
1,2,5,8,13
CS58=
1,2,5,9,13
CS59=
1,2,6,7,13
CS60=
1,2,6,8,13
CS61=
1,2,6,9,13
CS62=
3,2,5,7,13
CS63=
3,2,5,8,13
CS64=
3,2,5,9,13
CS65=
3,2,6,7,13
CS66=
3,2,6,8,13
CS67=
3,2,6,9,13
CS68=
4,2,5,7,13
CS69=
4,2,5,8,13
CS70=
4,2,5,9,13
CS71=
4,2,6,7,13
CS72=
4,2,6,8,13
CS73=
4,2,6,9,13
Just like before all the cut sets share the same number which means that they can only be ranked
according to the kind of base event. Among the four new base events that’s been presented in
the fault tree the following three can are errors among active components:
Malfunction: Pump 2
Malfunction: Control System 2
Fouling in the condenser heat exchanger
The last one that refers to the possibility that the heat transfer limit of the thermosyphon have
been crossed can be seen from two different perspectives. Firstly, the calculation of determining
the limit of the thermosyphon can be a human error. This is in case the designer made any errors
when designing the process for this kind of scenario. Besides this the thermosyphon is a passive
device and thereby the error could be classified as an error of a passive component. It is though
assumed that the process has been designed properly and that the base event can be seen as an
error of a passive component. With this information, the minimal cut scenes can be determined
and the events where the process provides insufficient condensation becomes part of the MCS.
All the minimal cut scenes have been marked out in Table 11.
The minimal cut scenes from the two different scenarios can now be compared with each
other. The minimal cut scenes from both the fault trees contains one human error while the remaining ones are an error of an active component. Therefore, the only real difference is the
number of base events that the two scenarios require to trigger a thermal runaway. The second
scenario when a semi-batch reactor is cooled by both a cooling jacket and a thermosyphon
contains one more active error than the first scenario. This proves that as long the thermosyphons is designed to be capable of handling worst-case scenarios use of them can lower the
risk of thermal runaways.
32
4 Results & Discussions:
4.1 The Steinbach Case
The primary reason for this study is to observe the pipes effect on the reactor temperature and
determine if a thermal runaway can be avoided as well as if a hot spot can be lowered. The
process consists of a semi-batch reactor cooled with a cooling jacket and a different amount of
thermosyphons. The dosing time is 18000 seconds and reactant B is fed into the reactor during
the entire dosing period. The simulations that was performed using the equations mentioned in
the chapter 3.4 presented the following results. Thermosyphons was selected to perform the
additional cooling and between one to eight pipes was added to the reactor, each with the same
inner radius, 5 cm.
Figure 13. A graph displaying the temperature
within the reactor during the cycle time. A different number of thermosyphons was tested yet
they all had the same inner radius of 5 cm.
Figure 12. A graph showing the reactor temperature during the cycle when three, four or
five thermosyphons is used. [rinner= 5 cm]
Table 13. The highest temperature that’s reached during the process cycle.
Number of
Thermosyphons
Tmax
0
3
4
5
298.72 ºK
287.57 ºK
284.70 ºK
282.68 ºK
Figure 13 displays that thermosyphons can clearly lower the reactions hot spot and bring a much
lower temperature difference during the cycle. The effect of the cooling pipes starts to appear
when the reactor temperature reaches about 275 degrees Kelvin. It can also be seen that the
effect is lowered when too many pipes is used for cooling. One of the reason for this is that the
hot spot doesn’t reach a significant temperature difference from the boiling point of the working
fluid. This means that there is only a small temperature difference that will create the driving
force of the heat transfer. Secondly it was also seen in Figure 7 that the heat coefficient of the
pipe increased with higher temperatures. This means that the maximal heat coefficient will decrease if the process reaches a lower hot spot for both of these reasons.
33
Three to five thermosyphons was determined to be used to study the effect of the pipes on the
reaction process. The number was chosen since the pipes lowers the hot spot considerably and
that there is a clear difference between each of them, as can be seen in Figure 12.
To begin with the concentration of the reactants within the reactor is examined. The reaction
rate is lower due to the temperature decrease. In Figure 14 it can be noted that there is a slope
decrease of reactant A at the same time as the hot spot reaches its peak. When the pipes is
installed and the hot spot is lowered the reaction rate becomes more constant and the concentration of reactant A is lowered more stably.
Figure 14. The concentration of reactant A
within the reactor during the cycle time
Figure 15. The concentration of the reactant
that’s being fed into the semi batch reactor, reactant B, under the reaction cycle.
In Figure 14 and 15 the concentrations of the reactants can be seen and the decrease in temperature brings with it a growing accumulation of the reactants, especially component B. Until
about one third of the dosing time has ended the concentration of reactant B increases since the
reaction rate is lower than the incoming flow. This accumulation of the reactant results in an
increased reaction rate. The new reaction rate causes a large energy release and the hot spot
begins to grow up until the accumulated concentration of component B has decreased. The
difference between the two kinds of cooling system that was examined can first be seen after
the accumulation of the reactant has reached its peak. This is because the temperature within
the reactor hasn’t reached the boiling point of the working fluid and therefor the heat transfer
hasn´t initiated yet. When the hot spot begins to grow the effect of the thermosyphons can be
seen. Firstly, the accumulated concentration of component B doesn´t decrease to the same level
as when only a cooling jacket is used to cool the reactor. It can also be noted that concentration
decreases less when more thermosyphons is used. The reason for this is the correlation of the
temperature and the reaction rate. The more thermosyphons that’s installed the lower the temperature increases and therefor the reaction rate and the feeding rate becomes equivalent at
different concentration levels.
The second thing to note is that the final accumulation of reactant B increases when thermosyphons is added to the cooling system. The number of thermosyphons that was installed doesn’t
seem to affect the final concentration of component B very much yet clear differences is easy
to during the rise of the hot spot. The reason for this is most likely that while the temperature is
34
being decreased the reaction rate recovers due to an increased accumulation of reactants. The
high accumulation of the dosed component, reactant B, at the end of the reaction cycle comes
from that the reaction rate decreases due to a low concentration of the loaded reactant. To prevent the accumulation from growing so large the reaction rate could be increased either with a
temperature adjustment (such as turning off the cooling jacket) or with a temporary dosing of
the reactant A.
The growing accumulation of unused reactants that has been observed could be noted as a disadvantage and potentially a safety risk. The reasoning for this is simply that either the reaction
cycle or reactor size needs to be increased to uphold the previous production per cycle and that
the accumulation could. To observe the progress of the process the conversion rate of the loaded
reactant is calculated according to the equation below.
𝐶𝑜𝑛𝑣𝑒𝑟𝑠𝑖𝑜𝑛 𝑜𝑓 𝑟𝑒𝑎𝑐𝑡𝑎𝑛𝑡 𝐴 (%):
𝑋 = 100
𝐶𝐴,0 𝑉0 − 𝐶𝐴 𝑉
𝐶𝐴,0 𝑉0
The conversion is calculated when various numbers of thermosyphons is installed and the results is displayed in Figure 16. The clearest difference can be seen when the reactor temperature
reaches the boiling temperature of the working fluid. The difference between using three to five
thermosyphons is very low. However, when examining the difference between the two cooling
systems the difference is obvious, which is displayed in Table 14. The lower reaction rate has
resulted in a conversion decrease of a least 10 %. This means that the additional cooling requires
a longer cycle time to reach the same conversion as the original process. To counter the new
cycle time the reactor size could be altered to increase the production during the cycle time.
Table 14. The final conversion rate of the loaded reactant, component A.
Cooling system
Conversion of
reactant A
Cooling
Jacket
79.03 %
Cooling jacket +
3 Thermosyphon
67.85 %
Cooling jacket + 4 Cooling jacket + 5
Thermosyphon
Thermosyphon
67.00 %
66.47 %
Figure 16. The difference of conversion of reactant A during the process cycle when using a
different number of thermosyphons.
To observe the process more closely and to evaluate potential risks the changes of four factors
is analyzed:
35
The reaction rate
The released energy
The heat transfer coefficient
The heat transfer of a thermosyphon
Three of these factors is calculated out of the results from the differential equation according to
the equations beneath and the overall heat transfer coefficient is determined according to the
same equations that was presented in chapter 3.2.
𝑅𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑒: 𝑟 = 𝑘𝐶𝐴 𝐶𝐵
𝑅𝑒𝑙𝑒𝑎𝑠𝑒𝑑 𝐸𝑛𝑒𝑟𝑔𝑦: 𝐸 = 𝑟∆𝐻𝑟𝑒𝑎𝑐𝑡𝑖𝑜𝑛 𝑉
𝐻𝑒𝑎𝑡 𝑡𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝑜𝑓 𝑡ℎ𝑒 𝑡ℎ𝑒𝑟𝑚𝑜𝑠𝑦𝑝ℎ𝑜𝑛: 𝑄𝑃𝑖𝑝𝑒 = 𝑘𝑝𝑖𝑝𝑒 𝐴𝑟𝑒𝑎𝐸𝑣𝑎𝑝𝑜𝑟𝑎𝑡𝑜𝑟 (𝑇 − 𝑇𝐵𝑜𝑖𝑙 )
As was previously mentioned when the conversion was analyzed the additional cooling that the
thermosiphons apply decreases the reaction temperature and thereby also the reaction rate. In
Figure 19 the reaction rate can be compared and it’s clear that the high peak that the hot spot
provides is lowered by at least 40 % when three or more thermosiphons is used.
Figure 18. A graph displaying the reaction rate
during the process cycle.
Figure 17. The energy that’s being released due
to the exothermic reaction.
Since the reaction rate is one of the factors that determines the heat generation that the exothermic process provides a clear correlation can be seen in Figure 20 where the released amount of
energy that’s released every second is compared under the same circumstances. Since the only
difference between the two parameters is the adding of a constant reaction enthalpy the curves
shape is identical. When it comes to the differences between the two cooling systems the biggest
change can be seen during the hot spot period. The reaction rate of the process that’s only cooled
by a cooling jacket grows dramatically due to the temperature peak before it begins to decline
thanks to the low concentration of the dosed reactant. When the second cooling system is examined the reaction rate reaches its maximal level during the hot spot period yet the growth is
much lower in response to the decreased temperature rise. When five thermosyphons is installed
in the reactor the highest reaction rate and energy release is roughly half when compared to the
original cooling system. Another interesting difference that the second cooling system provides
is that the reaction rate is almost constant during the final part of the reaction. The thermosyphons can thereby help with decreasing the changes of the reaction rate during the process. That
36
the hot spot can be controlled and adjusted like this can prove to be very useful considering it’s
then that the risk of a thermal runaway is at its highest level and that most undesired products
is usually produced. Since the process that’s being observed in this simulation doesn’t involve
any unwanted reactions the second statement can’t be examined more closely yet it will be
discussed later.
Since a thermosyphons has heat transfer limitations and that the overall heat transfer coefficient
is crucial to determine the number of pipes that’s required to use these parameters needs to be
examined. As displayed previously (Figure 7) the heat transfer coefficient increases with temperature and thereby it’s clear to see that the peak value is reached as the cycle is at its hot spot
(see Figure 21). Since the temperature rise of the hot spot decreases when more thermosiphons
are used for cooling the heat transfer coefficient decreases as well yet the changes is not of
bigger scale.
The theoretic heat transfer that the thermosiphon is supposed to perform during different moments of the process cycle is easy to compare with the dominating limit of the pipe. The dominating limit of the thermosiphon that’s been used for the simulation is the boiling limit. When
the limitation is compared to the energy that’s being transferred through the pipe when three
pipes (and less of course) is being used the heat transfer oversteps the boiling limit. To overcome this at least four pipes needs to be used. Depending on potential risks for increased reaction temperature such as jacket fouling or malfunctions of the regulation system it can be recommendable to use even more thermosyphons to minimize the risk that the boiling limit is
reached.
Figure 20. The heat transfer that each thermosyphon performs during the cycle time.
Figure 19. The changes of the heat coefficient
during the process.
4.1.1 A pipe with a rough outer surface:
As was discussed in chapter 3.2 an efficient way to improve the heat transfer coefficient is to
use pipes with a rough surface. Figure 8 displayed that a roughness of 1 mm could at least
double the heat transfer coefficient of a pipe that was of the same size as those that was used
for the simulations above. Since this can increase the amount of energy that the thermosiphons
can transfer away from the reactor such an increase would be most desirable yet it comes at the
cost that the boiling limit will be reached at even smaller temperature differences. Since the
boiling limit is directly connected to the radial area of the pipe, thermosyphons of a slightly
larger scale was determined to be put to the test. The effect of various number of thermosiphons
37
with an inner radius of six centimeters and a rough surface of 1 mm was tested for the process
simulation.
𝜀 = 1 𝑚𝑚
𝑟𝑖𝑛𝑛𝑒𝑟 = 6 𝑐𝑚
As was expected, due to the increased heat transfer coefficient the pipes cooling effect was
improved. In Figure 17 it can be seen that the hot spot of the process was greatly diminished
when two or more thermosiphons was used and that when four thermosiphons are used the hot
spot didn’t go even five degrees above the boiling temperature of the working fluid within the
pipe.
Table 15. The highest temperature that’s reached during the process cycle.
Number of
Thermosyphons
Tmax
0
2
3
4
298.72 ºK
282.89 ºK
279.11 ºK
277.56 ºK
Figure 21. The reactor temperature during the
process cycle when various numbers of
thermosyphons is installed to provide
additional cooling
Figure 24 and 23 displays the concentration and the conversion of the loaded reactant, component A. Just like before the additional cooling of the thermosyphons have a negative effect of
the conversion. The low temperature decreases the reaction rate and the production is lowered.
The conversion is also displayed in table 15 where the exact amount of decrease can be seen.
Considering an even more impressive temperature control with the rough-surfaced thermosyphons the disadvantages can be assumed to point in the same direction. When observing the
conversion rate the difference is still very clear between the two cooling systems. Something
that’s worth to note is however that the different number of pipes still doesn’t affect the end
results very much. The conversion rate is lowered by almost 15 % when four surface treated
thermosyphons is installed and the production is thereby lowered as well. It can also be observed that the conversion is lower when using four treated thermosyphons than when five ones
with a smooth surface was used.
38
Figure 23.The concentration of the loaded
reactant, component A, during the dosing time.
Figure 22. The conversion rate of the loaded
reactant, component A, during the simulation.
Figure 24. The concentration of the dosed
reactant, component B, during the reaction.
Table 16. The final conversion rate that’s reached during various simulations of the model.
Cooling system
Conversion of
reactant A
Cooling
Jacket
79.03 %
Cooling jacket +
2 Thermosyphon
66.41 %
Cooling jacket + 3 Cooling jacket + 4
Thermosyphon
Thermosyphon
65.44 %
64.76 %
In Figure 20 the concentration of component B within the reactor is displayed. Just as in the
previous simulation it’s clear that the accumulation of the component grows when the additional
cooling is used yet that the difference between the numbers of pipes isn’t overwhelming. The
reasons behind the results should still be the same as before since the only thing that has been
altered between the two simulations is the heat coefficient.
Just like previously the process is analyzed more closely by examining the following parameters:
The reaction rate
39
The released energy
The heat transfer coefficient
The heat transfer of a thermosyphon
There is a considerable change in the heat transfer coefficient when a roughed surface pipe is
used instead of a smooth. The heat transfer coefficient that determines the heat transfer into the
pipe is displayed in Figure 25.
In Figure 26 the heat transfer of a thermosyphon is compared with the dominating limit, the
boiling limit. When two pipes are being used the heat transfer oversteps the limitation due to
the growing heat transfer coefficient and the temperature difference. If four thermosyphons is
being used the heat transfer reaches its peak at a level of about 50 % of the limit. It’s also worth
to mention that the high heat transfer coefficient enables this to occur even though the reaction
temperature is less than 5 K above the boiling temperature of the working fluid.
Figure 25 The overall heat transfer coefficient Figure 26. The heat that’s transfered from the
that determines the amount of heat that’s reactor by each thermosyphon in comparison
transfered from the reactor to the heat pipes.
with their dominating limit, the boiling limit.
When it comes to the reaction rate and the energy released into the reactor the curves is very
similar to before with only a difference in the number of pipes that’s required.
Figure 27. A display of the different reaction
rates between the two cooling systems.
40
Figure 28. The energy amount that’s released
during the process.
4.1.2 In case of fouling or a malfunction:
The simulations have given a taste that a thermosiphon can be of good use for lowering the hot
spot that’s generated by an exothermic reaction in a semi-batch reactor yet what would happen
in case of an accident? As was mentioned before a thermal runway is often caused due to either
fouling or a malfunction. What effect would the additional cooling have if the worst accident
would happen and that a malfunction in the regulation system or an accident would cause the
flow within the cooling jacket to be stopped completely and yet the dosing wasn’t ceased?
The simulation is performed under the same circumstances as before except that the cooling of
the cooling jacket is completely removed. Thermosyphons with a rough surface was chose to
be tried due to their promising results in the previous simulation. This means that the pipes have
an inner radius of six centimeters and a roughness of 1 mm.
𝜀 = 1 𝑚𝑚
𝑟𝑖𝑛𝑛𝑒𝑟 = 6 𝑐𝑚
As can be seen in Figure 29, the reaction temperature increased with more than 100 degrees if
no cooling process was installed which could have become a major accident if for example a
decomposition reaction would have been initiated. Without the cooling jacket, at least three
thermosyphons is required to keep the temperature under control. In Figure 30 a very clear
difference between using either three or four thermosyphons can be seen. If less than three
thermosyphons is installed a thermal runaway is initiated and the temperature continues to rise
during the entire dosing time. When using three the temperature reaches up to 300 K and remains relatively constant for the remainder of the cycle. If four thermosyphons is applied the
temperature within the reactor doesn’t even go above the previous hot spot of the original process that was only cooled by a cooling jacket. Thereby, even if the cooling jacket is completely
useless three or more thermosyphons can provide the required cooling to control the reaction.
Figure 30. The effect that three, four, five and
Figure 29. The temperature within the reactor
six thermosyphon has on the reaction
in case the cooling jacket isn’t functionable
temperature during a worst case scenario.
during the cycle.
41
Table 17. The highest temperature that’s reached in the reactor if the cooling jacket is out of
order.
Number of
Thermosyphons
Tmax
0
3
4
5
6
365.06 ºK
297.14 ºK
283.82 ºK
279.89 ºK
278.53 ºK
From the previous simulation, it was noted that a very small temperature difference was enough
to bring the heat transfer of the thermosyphon up to the boiling limitation. In Figure 31 a graph
comparing the heat transfer and the limitation is displayed and even though three or four thermosyphons is only required to cool the reaction it takes at least six to bring it underneath the
boiling limitation. This is important to note and it gives a good indication that for safety
measures it’s best to install at least six thermosyphons into the reactor to keep their functionality
safe. It can also be observed that even when using six thermosyphons the boiling limitation
remains fairly close to the maximal heat transfer and thereby it might be wise to install even
more.
Figure 31. The heat transfer that each thermosyphon provides in comparison with their
dominating limit.
4.2 Hydrolysis of Acetic Anhydride
The process was simulated following the model of the hydrolysis process that was presented in
chapter 3. It’s an adiabatic semi-batch reactor that’s loaded with a high volume of water and
dosed with a continuous flow of acetic anhydride during a dosing time of twenty minutes. The
results that’s presented is the outcomes when a cooling jacket performs all the cooling as well
as a reactor with both a cooling jacket and thermosyphons. Due to the improved heat transfer
coefficient, thermosyphons with rough surfaces was determined to be used for the first simulation test of the process. Pipes with an inner radius of five centimeters and a roughness of 1 mm
was simulated as an additional cooling of the process. The saturation temperature of the working fluid within the thermosyphons is 313ºK which results that no bigger differences in the
results can be seen until the reaction temperature has reached this point.
In Figure 32 the reaction temperature during the cycle time is displayed. When a cooling jacket
is installed there is a clear temperature increase during the entire dosing time. The reason for
the quick temperature decrease that follows the end of the dosing period can be seen in Figure
34 where the concentration of acetic anhydride is displayed. The concentration doesn’t build
up to any large amounts of the reactant due to the relatively fast reaction rate and the reaction
42
ends after 25 minutes due to reactant shortage. One thing that’s worth pointing out is that when
two or more thermosyphons is added then the hot spot is drawn out. Table 18 displays that the
thermosyphons provides a very decent control of the reaction temperature since the hot spot
doesn’t even go above 3 degrees over the boiling point of the working medium.
Table 18. A display of the maximal reaction temperatures that’s reached when different numbers of thermosyphons is used to cool the process.
Number of
Thermosyphons
Tmax
0
1
2
3
4
331.4 ºK
319.9 ºK
317.3 ºK
316.1 ºK
315.5 ºK
Figure 32. The reaction temperature within the
semi-batch reactor when various numbers of
thermosyphons is used.
When the reaction temperature is compared with the reaction rate it can be seen that the reaction
rate is relatively similar when using one to four thermosyphons. That the reaction rate isn’t
more effected by the temperature differences is however reasonable when the concentration of
acetic anhydride is considered. The accumulated amount of the reactant increases with the number of thermosyphons and the concentration increase makes up for the temperature loss, keeping
the reaction rate at a reasonably fast and constant speed.
43
Figure 34. The concentration of the fed reac- Figure 33. The rate of the hydrolysis reaction
tant, acetic anhydride.
when different numbers of thermosyphons was
used.
The energy that’s released during the reaction is more similar to the reaction rate than it was in
the previous simulations. This is due to that there is less volume changes in this reactor than it
was in the previous semi-batch reactor. The heat coefficient has already been noted as being
strongly effected of the reaction temperature and as can be seen the heat transfer of the thermosyphons decreases when larger numbers of pipes is used due to the temperature decrease.
The differences isn’t however of any larger scale. Due to the small temperature rises the heat
transfer that each pipe is able to perform is also beneath the dominating limit, the boiling limit,
when two or more thermosyphons is put to use.
Figure 36. The energy that’s released from the
hydrolysis reaction during the process cycle.
Figure 35. The heat transfer that a thermosyphon performs during the process in comparison with its operating limit.
4.2.1 In case of fouling or a malfunction:
Just like in the previous malfunction simulation (Chapter 4.1.2), the worst case of a scenario
would be if the cooling of the cooling jacket would stop while the dosing couldn’t be ceased.
What would the result be in case a malfunction in the equipment or the automation would stop
the cooling stream yet retain the cooling of the thermosyphons?
44
In Figure 37 the reaction temperature is displayed in a semi-batch reactor where the cooling
jacket doesn’t provide any heat transfer at all. When there is no cooling whatsoever the temperature increases during the entire dosing time up to about 350ºK. The boiling points of each
component within the reactor is above 373 degrees Kelvin yet considering the small size of the
reactor the temperature rise is quite noticeable. When two or more thermosyphons (rinner = 5
cm, ε= 1 mm) is used the reaction temperature remains much more stable. The maximal temperature that is reached when using two or more thermosyphons is almost the same as when the
cooling jacket was installed.
Figure 37. The reaction temperature within
the semi-batch reactor when various numbers
of thermosyphons is used and the cooling
jacket is not providing any cooling of the process.
Table 19. The maximal temperature that is reached during the cycle time when the cooling
jacket is out of operation.
Number of
Thermosyphons
Tmax
0
1
2
3
4
350.33 ºK
322.84 ºK
310.06 ºK
307.82 ºK
306.65 ºK
The reason for not using less than two thermosyphons is that when one is used the temperature
within the reactor rises too high and the heat transfer exceeds the boiling limit. As can be seen
in Figure 39 the heat transfer is far from the boiling limit when two or more are installed. This
is very good for the safety of the reaction since it indicates that the thermosyphons will be
efficient during the entire cycle in case such a malfunction would occur.
Figure 38. The heat transfer of each thermosyphon during the malfunction cycle in comparison with the pipe’s dominating limit.
45
4.3 Fine Chemicals
The results from the previous two simulations has indicated that the thermosyphons can bring
a benefit to the temperature control and therefor to the thermal safety. Considering that both the
simulations didn’t involve any side reactions they can’t be used to examine the selectivity improvement that the temperature control enables.
When using the model that was presented in chapter 3.4.3, a batch reactor that involves a desired
reaction along with an undesired side reaction the effect of the thermosyphons on the reaction
efficiency can be observed. The batch reactor has a volume of 6.3 cubic meters and is surrounded by a cooling jacket with a constant cooling temperature of 305°K. The thermosyphons
that’s being tested is partially filled with water with a boiling temperature of 313 degrees Kelvin. For the following results thermosyphons with a rough surface (1 mm) and an inner radius
of six centimeters were simulated as an additional cooling system.
In Figure 39 and 40, the thermosyphons effect on the reaction temperature can be witnessed.
The different shape of the curves when compared with the previous simulations models is due
to the reactor type. As this is a batch reactor there is no slow reaction rates in the start as the
concentration of the dosed reactant builds up, instead a more linear reaction rate is achieved.
The reason is most likely that the temperature rise makes up for the decrease of reactants. The
differences of adding additional pipes becomes relatively small when five or more is used and
thereby the results of using two to five thermosyphons will be focused.
Figure 39. A display of the reaction tempera- Figure 40. The temperature within the reactor
ture when the two cooling systems is compared when two to five thermosyphons is installed in
with various numbers of thermosyphons in- comparison with the original cooling system
stalled.
The temperature within the reactor is strongly affected by the thermosyphons due to lower reaction rates and heat transfer. When two or more thermosyphons is used to cool the process the
reaction temperature becomes much more linear after the evaporation within the pipes have
initiated. By adding five thermosyphons the maximal temperature that’s reached after about
one third of the cycle time is hardly changed at all during the remaining process. The temperature within the reactor can most likely be lowered even more if the boiling temperature of the
working fluid is adjusted. As was mentioned previously the temperature within the pipes were
examined during this process. By comparing the reaction temperature and the temperature
within the thermosyphon it can be noted that it requires 120 seconds longer for the working
46
fluid within the thermosyphons to reach the boiling point. Considering that 120 is only 2 % of
the cycle time this shouldn’t be of any concern.
Table 20. A display of the maximal temperature that’s reached during the cycle and the time
required for the reactor and the thermosyphons to reach the boiling point of the working fluid.
Number of
thermosyphons
Tmax
TimeReactor
0
2
3
4
5
328.00°K
320.67°K
319.02°K
317.61°K
316.91°K
-
1459 s
1464 s
1468 s
1473 s
-
1576 s
1578 s
1584 s
1590 s
-
117 s
114 s
116 s
116 s
(Treaktor=313°K)
TimeThermosyphon
(TPipe=313°K)
ΔTime
Another of the main reasons for performing this simulation was to examine the efficiency of
the reactor when the two cooling systems was used. To begin with the concentration levels of
each component is presented in the following graphs. In Figure 41 the concentration of the
reactant, component A, is displayed. Similarly, to the previous models that has been simulated
the additional cooling provides a decrease in the reaction rate and therefor a larger amount of
the reactant remains at the end of the reaction cycle. The differences that the number of thermosyphons that’s installed isn’t of any larger scale yet a small increase in remaining reactants
can be seen when more pipes were used.
Figure 41. The concentration of reactant A during the cycle time of the batch reactor when different numbers of thermosyphons was used.
When examining Figure 41 which presents the concentration of the desired product, component
P, a similar result can be seen. The lower reaction rates have led to a production decrease. As
was mentioned previously this can be countered by either prolonging the cycle time or increasing the reactor volume.
47
Figure 43. The concentration of the desired
product, P, during the cycle time of the batch
reactor when different numbers of thermosyphons was used.
Figure 42. The concentration of the undesired
product, S, during the cycle time of the batch
reactor when different numbers of thermosyphons was used.
The main difference between this simulation model and the ones that’s been tested previously
can be seen in Figure 42. The undesired reaction consumes the desired product and creates an
unwanted product, component S. The concentration of the unwanted product doesn’t reach any
high amounts during this simulation though it can be observed that the temperature control
decreases the production.
To examine the effect that the thermosyphons brings to the process the conversion rate and
selectivity is analyzed. The selectivity displays a measurement of how well controlled the reaction is by examining the number of used reactants and the production of the desired product.
Considering that the stoichiometric factor is one, the selectivity of the process can be calculated
with the following equation:
𝑆𝑒𝑙𝑒𝑐𝑡𝑖𝑣𝑖𝑡𝑦: 𝑌 = 𝐶
𝐶𝑃
𝐴,0 −𝐶𝐴
Figure 44. A comparison between the two cooling systems effect on the conversion rate of the
reactant, component A, during the process.
48
Figure 45. A display of the selectivity changes
that the thermosyphons provides.
When observing the results in Figure 44 and 45, it can be concluded that the additional cooling
can be seen as both an advantage and a disadvantage. The lower temperature results in a lower
production yet as well with a more selective process. By using five thermosyphons which
caused the temperature within the reactor to remain almost constant after the low hot spot the
selectivity is above 99 percent. However, at the cost of the selective process comes a decrease
of production of almost 6 percent. For industries that produces bulk chemicals the conversion
decrease could be of more importance yet the thermal safety could perhaps make up for the loss
since. The high flexibility of the thermosyphons makes it most likely possible to create an extremely effective process with a very low production of undesired by-products. The selectivity
and control of the multiple reactions that occur is often of very high importance in industries
producing fine chemicals and therefor this cooling system could be most suitable for such operations.
Table 21. An evaluation of the process efficiency through conversion and selectivity when two
to five thermosyphons is used in comparison with the original design.
Number of
thermosyphons
0
2
3
4
5
Conversionfinal
80.77 %
77.05 %
76.11 %
75.35 %
74.95 %
-
-3.72 %
-4.66 %
-5.42 %
-5.82 %
96.95 %
98.61 %
98.83 %
98.99 %
99.06 %
-
1.66 %
1.98 %
2.04 %
2.11 %
ΔConversionfinal
Selectivityfinal
Δ Selectivityfinal
Figure 47 shows a graph that displays the heat transfer that is performed by each thermosyphons
during the different simulations. The heat transfer is initially not very impressive during the
boil up yet after the boiling temperature is reached the heat transfer increases which corresponds
with the overall heat transfer coefficient that’s presented in figure 46. Two thermosyphons is
enough to bring the heat transfer underneath the limit yet it’s very close and more pipes would
be recommendable.
Figure 47. A display of the heat transfer that
each thermosyphons displays in comparison
with the boiling limit.
49
Figure 46. The overall heat transfer coefficient
during the reaction cycle.
5 Conclusion
According to the theoretical simulations that has been performed in this work the thermosyphons can be of good use to ensure the safety and efficiency of a chemical reactor. By adding
thermosyphons to cool the temperature within a batch or semi-batch reactor hosting an exothermic reaction the temperature control has increased significantly. Hot spots that’s classically
displayed during a batch reaction can be considerably lowered with an acceptable number of
medium sized thermosyphons. By lowering the hot spot the safety of the process is increased
and a more efficient control of the reaction rates can be achieved. Thermosyphons might also
be a better alternative to wicked heat pipes considering that they are capable of transferring
more heat and comes at a cheaper price. Since heat pipes have proven to have a superior overall
heat transfer coefficient in previous studies this might be debatable and something that requires
experiments and consideration.
During the simulations it was also noted that less thermosyphons were required to be used if
the surface wall of the evaporator region were rough instead of smooth. The rough surface gave
a significant improvement to the heat transfer coefficient of the reactor fluid which improved
the pipes the heat transfer capabilities. Something that speaks against the surface treatment
could be that they come at a higher economical cost and that it might be more economical to
install a few more smooth pipes than a couple of less rough one.
Thermosyphons is generally considered to be relatively cheap yet the golden rule of the lower
the number the lower the cost still stands. The simulations have however indicated that thermosyphons is very sensitive to temperature differences and that the impressive overall heat
transfer coefficient enables the limitation to be reached at low temperatures differences. To
counter this and ensure that the thermosyphons stays below their dominating limit a higher
number of pipes can be installed. This will make sure that during fouling or malfunctions their
heat transfer capabilities will not be lost or decreased dramatically. Another way to lower the
temperature sensitivity is to not give the evaporator regions a rough surface. The rough surface
gave a significant improvement to the heat transfer coefficient of the reactor fluid and by lowering this a higher temperature difference is required to reach the heat transfer limitation.
To summarize the conclusion the theoretical simulations have indicated that a cooling system
that contains heat pipes in the form of thermosyphons is capable to improve the reaction selectivity and the thermal stability of exothermic reactions. The simulations have even gone so
far as to show that it’s theoretically possible to avoid thermal runaways during worst case scenarios such as a complete regulation malfunctions. The effect that the cooling have had on the
production capabilities of the reactor is not desirable yet logical considering that the majority
of exothermic reactions speeds up by increased temperatures. This makes thermosyphons a
notable tool to be used in processes that requires high selectivity due to factors such as expensive reactants. Another beneficial situation to use the additional cooling system is when the
process contains undesired exothermic reactions such as decompositions since they can lower
the runaway risk.
50
6 Further work
To further evaluate how precious thermosyphons can be to the thermal safety and efficiency of
chemical process more studies, experiments and simulations is required. The overall heat transfer coefficient that’s being used in this study is simplified and a more advanced model should
be used where the process isn’t cooled by thermosyphons with ideal temperature and pressure.
This is especially important since the heat transfer coefficient during a phase change is very
dependent on the temperature of the heating medium. Since only a small temperature rise above
the boiling temperature of the working fluid was obtained during most of the simulations the
heat transfer coefficient within the pipes might not be high enough to be neglected.
Something that’s also required to be examined is the cooling of the condensation region of the
pipes. In this study, the cooling has been assumed to be ideal and thereby there is not much that
indicates the cooling requirements that’s necessary to achieve this.
Besides this it’s also worth to note that thermal runaways also tend to occur in storage facilities
where the stored components is heated up to dangerous levels. Heat pipes could maybe prove
useful in minimizing the risks of such events. Pipes could be in contact with the stored component while having their condensation regions within a container of a safe fluid.
51
7 References
Abdollahi, M. A. (2015). Water Boiling Heat Transfer in Vertical Jacketed Pipe: A CFD Model.
International Conference on Chemical, Civil and Enviromental Engineering, p. 28-31.
Alveteg, M. (2013). Handbook, Physical properties, correlationa and equations in Chemical
Engineering. Lund: MediaTryck.
Bjerle, I , Berggren, J-C och Karlson, H. (1977). Kemiska reaktorer - del 1: Dimensionering
och Simulering av Satsreaktorn. Kemiska Tidsskrifter.
Dincer, I. (u.d.). Chapter 7 - Heat Pipes. i Refrigeration Systems and Applications (s. 405). John
Wiley & Sons.
Dincer, I. (u.d.). Chapter 7 - Heat Pipes. i Refrigeration System and Applications. John Wiley
& Sons.
Dr. Ahmad, H. (2013). Comparison between a Heat Pipe and a Thermosyphon Performance
with variable evaporator length. Al-Rafidain Engineering, Vol. 21 p. 1-12.
García, M. (2016). Thermal stability and dynamic analysis of the acetic anhydride hydrolysis
reaction. Chemical Engineering Science, p. 269-276.
Grewer, T. (1999). Prediction of thermal hazards of chemical reactions. Journal of Loss
Prevention in the Process Industries, Vol. 12, p. 391-398.
Guinand, C. (2016). Thermal Process Safety Based on Reaction Kinetics and Reactor
Dynamics. AIDIC, Vol. 48 2016, Vol 48. p. 19-24.
(2002). Hazard Investigation - Improving Reactive Hazard Management. U.S. CHEMICAL
SAFETY AND HAZARD INVESTIGATION BOARD.
Heat Transfer Limitations of Heat Pipes. (den 14 February 2017). Hämtat från Thermal-Fluids
Central:
http://www.thermalfluidscentral.org/encyclopedia/index.php/Heat_Transfer_Limitatio
ns_of_Heat_Pipes
Karlsson, H. (2012). Processriskanalys. Lund: Mediatryck.
Khalid, J. (2000). Improvements of gravity asssited wickless heat pipes. Energy Conversion &
Management, p. 2041-2061.
MAHBulletin. (2016). Chemical Accidents Prevention and Preparedness. European
Commission.
Mas, E. (2006). Safe Operations of Stirred-Tank Semibatch Reactors Subject to Thermal
Hazard. American Institute of Chemical Engineers Journal, 3570-3582.
Moulijn, J. (2008). Chapter 10. Fine Chemicals. i Chemical Process technology (p. 307-330).
John WIley & Sons.
52
Reay, D. (2013). Appendix 1, Working fluid properties. i Heat Pipes: Theory, Design and
Applications (p.228, 232). London: Elsevier.
Reay, D. (2014). Chapter 2 Heat transfer and fluid flow theory. i Heat Pipes Theory, Design
and Application. Elsevier.
Reay, D. (2014). Chapter 4 Design Guide. i Heat Pipes Theory, Design and Applications.
Elsevier.
Stoessel, F. (2008). Chapter 1 – Introduction to Risk Analysis of Fine Chemical Processes. i
Thermal Safety of Chemical Processes (p. 3). Wiley-WCH.
Stoessel, F. (2008). Chapter 2 – Fundamentals of Thermal Process Safety. i Thermal Safety of
Chemical Processes (p. 34-35.). Wiley-WCH.
53