See Beck

Simulation and Evaluation of Two Different Skin Thermocouples
A Comparison made with Respect to Measured Temperature
Joel Lundh
Applied Thermodynamics and Fluid Mechanics
Degree Project
Department of Management and Engineering
LIU-IEI-TEK-A--07/0076--SE
Copyright
The publishers will keep this document online on the Internet – or its possible replace-
ment – for a period of 25 years starting from the date of publication barring exceptional
circumstances. The online availability of the document implies permanent permission
for anyone to read, to download, or to print out single copies for his/her own use and to
use it unchanged for non-commercial research and educational purposes. Subsequent
transfers of copyright cannot revoke this permission. All other uses of the document are
conditional upon the consent of the copyright owner. The publisher has taken technical
and administrative measures to assure authenticity, security and accessibility. Accord-
ing to intellectual property law, the author has the right to be mentioned when his/her
work is accessed as described above and to be protected against infringement. For
additional information about Linköping University Electronic Press and its procedures
for publication and for assurance of document integrity, please refer to its www home
page: http://www.ep.liu.se/.
c 2007 Joel Lundh

Abstract
The demand for more accurate measurements is increasing in today’s industry. One
reason for this is to optimize production and thus maximize profits. Another reason is
that in some cases government regulations dictate that supervision of certain parameters
must be followed. At Preemraff Lysekil there are basically four reasons for measuring
skin temperatures inside fired process heaters, namely; because of government regula-
tions, in order to estimate the load of the fired process heater, to estimate the lifetime of
the tubes inside the fired process heater and finally, to determine the need of decoking.
However, only the first three of these reasons are applied to H2301/2/3. The current
skin thermocouple design has been in use for many years and now the question of how
well it measures surface temperature has risen. Furthermore a new weld-free design
is under consideration to replace the old skin thermocouple design. Another question
is therefore how well the new design can measure the surface temperature under the
same operating conditions as the old one. In order to evaluate this, three–dimensional
computer simulations were made of the different designs. As this thesis will show, the
differences in calculated skin thermocouple temperature and calculated surface temper-
ature is about the same for the two designs. However, the current design will show
a lower temperature than the surface temperature, while the new design will show a
higher temperature. Regarding the core of the skin thermocouple designs, namely the
thermocouple, no hard conclusions can be drawn, although the industry appears to
favor type ’N’ over type ’K’.
Sammanfattning
Efterfrågan på bättre mätningar ökar alltjämt i dagens industri. En anledning till

detta är för att optimera produktionen och därigenom maximera vinsten. En annan
anledning är att i vissa fall finns det lagkrav som dikterar att övervakning av vissa
parametrar måste göras. Vid Preemraff Lysekil finns det i praktiken fyra skäl till att
mäta yttemperatur, s.k. skintemperatur, inuti processugnar. Dessa skäl är: det finns
myndighetskrav, för att uppskatta ugnens last, för att göra en livslängdsanalys på ugn-
stuberna samt för att avgöra när avkoksning skall ske. I H2301/2/3 är det dock bara
de första tre anledningarna som är aktuella. I många år så den nuvarande skinelement-
designen använts och nu så har frågan om hur pass rätt den mäter dykt upp. Utöver
detta så har en svetsfri design fångat Preemraff Lysekils intresse då ett eventuellt byte
av design kan vara aktuellt. En annan fråga som har dykt upp är hur den nya desi-
gen står sig mot den gamla gällande avvikelse i uppmätt kontra önskad temperatur.
För att kunna utvärdera den nya designen mot den gamla utfördes tre–dimensionella
datorsimuleringar och som det här examensarbetet visar kommer avvikelsen mellan
den beräknade uppmätta temperaturen och den beräknade tubtemperaturen att vara
ungefär lika för de två olika skinelementen. Den gamla designen kommer dock att visa
en lägre temperatur än tubtemperaturen medan den nya designen kommer att visa en
högre. Angående själva kärnan i skinelementet, nämligen termoelementet, kan inga
bestämda slutsatser dras. Dock verkar industrin i allmänhet ha en tendens till att
favorisera typ ’N’ över typ ’K’.
Acknowledgments
I would like to give my thanks to the following persons which have, in their own way,
contributed to this master thesis.
• Elisabet Blom
• Nils Bjørdal
• Per Carlsson
• Roland Gårdhagen
• Stefan Karlsson
• Saeid Kharazmi
• Nils Larsson
• Faisal Mohamed Ali
• Hans Wernergård
• Joakim Wren
Furthermore, I would like to give my sincere thanks to Jan-Gunnar Alexandersson

for making this master thesis possible and for all guidance.
Last but certainly not least I would like to express my sincere gratitude to professor
Dan Loyd. I am truly amazed at his vast knowledge, rich experience and his ability to
always find time to sit down and talk problems through. His guidance has truly been
invaluable during this master thesis.
Linköping Mars 2007

Joel Lundh
Contents
1 Introduction 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2 Heat Transfer - A Short Repetition 5

2.1 Conduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Convection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3 Skin Thermocouples 9
3.1 The Current Skin Thermocouple . . . . . . . . . . . . . . . . . . . . . . 10
3.2 The New Skin Thermocouple . . . . . . . . . . . . . . . . . . . . . . . 10
4 Thermocouples 13
4.1 The Principle Behind the Thermocouple . . . . . . . . . . . . . . . . . 13
4.2 The Physics Behind the Thermocouple . . . . . . . . . . . . . . . . . . 15
4.2.1 Charge Carrier Diffusion . . . . . . . . . . . . . . . . . . . . . . 15
4.2.2 Phonon Drag . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
4.3 The Eight IEC Standardized Thermocouples . . . . . . . . . . . . . . . 16
4.4 The Type ’K’ and Type ’N’ Thermocouples . . . . . . . . . . . . . . . 18
4.4.1 Corrosion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.4.2 Hysteresis and In Situ Drift . . . . . . . . . . . . . . . . . . . . 19
4.5 Extension Wire and Compensation Cable . . . . . . . . . . . . . . . . . 20
4.6 Considerations at a Thermocouple Installation . . . . . . . . . . . . . . 21
5 Method 23
5.1 Three–Dimension Simulation . . . . . . . . . . . . . . . . . . . . . . . . 23
5.1.1 Geometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
5.1.2 Assumptions and Boundary Conditions . . . . . . . . . . . . . . 26
5.1.3 Mesh . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
6 Results 31
6.1 Three–Dimension Simulations . . . . . . . . . . . . . . . . . . . . . . . 33
6.1.1 Reference Simulation – The Tube . . . . . . . . . . . . . . . . . 33
6.1.2 Current Skin Thermocouple . . . . . . . . . . . . . . . . . . . . 34
6.1.3 New Skin Thermocouple . . . . . . . . . . . . . . . . . . . . . . 36
6.1.4 New Skin Thermocouple – Larger Spray Area . . . . . . . . . . 38
i
7 Discussion 41
7.1 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7.1.1 Simplifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
7.1.2 Mesh Independence . . . . . . . . . . . . . . . . . . . . . . . . . 42
7.1.3 Material Parameters . . . . . . . . . . . . . . . . . . . . . . . . 42
7.2 Thermocouple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.3 Skin Thermocouples . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.4 Long Time Operation of the Skin Thermocouple . . . . . . . . . . . . . 45
8 Conclusions and Recommendations 47
9 Nomenclature 49
A Appendix 53
ii
List of Figures
1.1 This picture shows how burners inside H2301/2/3 heat the vertically
placed tubes. [This picture has been published with the permission of
Preemraff Lysekil] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 This picture shows how a skin thermocouple exits the fired process heater
through the floor. [This picture has been published with the permission
of Preemraff Lysekil] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3.1 Schematic figure of the current skin thermocouple. The tube outer di-
ameter is 114.3 mm and the thickness 6.02 mm. The diameter of the
sheathing is just over 60 mm, the thickness is close to 4 mm and the
length is 150 mm. The dimensions for the plate on which the thermo-
couple is placed upon is approximately 24 x 24 x 3 mm. . . . . . . . . . 10
3.2 Schematic figure of the new skin thermocouple. The height of the thicker
protective part of the skin thermocouple is 13 mm, the length is 38 mm
and the maximum breadth is 16 mm. The contact surface towards the
tube has the dimension 8 x 38 mm. . . . . . . . . . . . . . . . . . . . . 11
3.3 Picture of the new skin thermocouple when fasten to a tube. [This
picture has been published with the permission of Nils Bjørdal] . . . . . 11
4.1 Schematic figure of a thermocouple. DMM is an abbreviation for digital-

multi-meter. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.2 This figure shows the output signal from the different thermocouples
with respect to the temperature in the measuring point. The reference
point is assumed to have a temperature of 0 ◦ C.[20, 19] . . . . . . . . . 17
4.3 This figure shows how the Seebeck coefficient varies with temperature
for the eight IEC standardized thermocouples.[19] . . . . . . . . . . . . 18
4.4 Schematic figure of a MIMS thermocouple. The conventional MIMS
thermocouple has a sheath of Inconel or stainless steel and incorporates
Ni-based thermoelements, i.e. type ’K’ or type ’N’.[7] The insulation used
between the sheathing and the thermoelements is exclusively MgO.[11] 19
4.5 Schematic figure of a oven. The temperature on the inside is more or less
the same and therefore will the total temperature gradient be placed in
the wall. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
iii
5.1 This figure shows the geometry of the current skin thermocouple on which
the simulations were made. The actual tube has an outer diameter of
114 mm, a thickness of 6 mm and a length of 277 mm. The radiation
shield has an outer diameter of 60 mm, a thickness of 4 mm and a length
of 150 mm. The plate has the dimensions 24 x 24 x 3 mm and is placed
on the tube 10 mm from the end of the radiation shield. . . . . . . . . 24
5.2 (a) shows how the new skin thermocouple with the three spray layers
were modeled in COMSOL. The actual tube has an outer diameter of 114
mm, a thickness of 6 mm and a length of 200 mm. The skin thermocouple
is 13 mm at its widest and has a contact area of 8 x 38 mm towards the
tube. The three spray layers are equally thick with a thickness of 1
mm. (b) is just a variant of (a). The difference is that in (b) the spray
area is larger. The total spray area, excluding the contact area of the
thermocouple, is about 0.003 m2 . (c) illustrates the reference geometry,
namely the tube it self. The outer diameter is 114 mm, it has a thickness
of 6 mm and a length of 277 mm. . . . . . . . . . . . . . . . . . . . . . 25
5.3 This figure illustrates a simple 2D mesh. . . . . . . . . . . . . . . . . . 28
5.4 This figure shows two different meshes used in the mesh independence
test. As seen, the mesh in (a) is quite more dense than the one in (b) . 29
6.1 Temperature field on the topside of the metal plate on which the ther-
mocouple is placed upon. The z axis is parallel to the length of the tube
and the tube height is increased with decreasing z. The measuring point
is placed at x = 0 and z = 0.1. . . . . . . . . . . . . . . . . . . . . . . . 34
6.2 Temperature field of the surface between the tube and the skin thermo-
couple. The z axis is parallel to the length of the tube and the tube’s
height is increased with decreasing z. The measuring point is placed at
x = 0 and z = 0.09. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
6.3 Temperature field of the surface between the tube and the skin thermo-
couple. The z axis is parallel to the length of the tube and the tube’s
height is increased with decreasing z. The measuring point is placed at
x = 0 and z = 0.09. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
7.1 The absolute deviance between calculated skin thermocouple temper-

ature and calculated skin temperature for the various simulations. It
should be noted that the current skin thermocouple shows a lower tem-
perature than the actual skin temperature. . . . . . . . . . . . . . . . . 44
iv
List of Tables
4.1 This table shows the IEC-code, which metal/alloy the positive and nega-
tive wire is made of, the field of work and which atmosphere the specific
thermocouple is designed for.[20] . . . . . . . . . . . . . . . . . . . . . 16
4.2 This table shows the letter designation of extension wire and compensa-
tion cable according to IEC standard.[20] . . . . . . . . . . . . . . . . . 20
6.1 Values of the parameters used in the first simulation’s two cases. . . . . 31
6.2 Values of the parameters used in the second simulation’s three cases. . 32
6.3 Calculated skin temperature at an ambient temperature of 800 and 850 ◦ C. 33
6.4 Calculated skin temperature for three different internal heat transfer
coefficients, at an ambient temperature of 800 ◦ C. . . . . . . . . . . . 33
6.5 Deviation between calculated skin temperature and calculated skin ther-
mocouple temperature for the current installation. . . . . . . . . . . . . 35
mocouple temperature for the current installation in an ambient temper-
ature of 800 ◦ C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
mocouple temperature for the new skin thermocouple. . . . . . . . . . . 37
mocouple temperature for the new skin thermocouple in an ambient tem-
perature of 800 ◦ C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
mocouple temperature for the new skin thermocouple with a larger spray
area. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
mocouple temperature for the new skin thermocouple with a larger spray
area, in an ambient temperature of 800 ◦ C. . . . . . . . . . . . . . . . 39
7.1 Results from the mesh independence check. . . . . . . . . . . . . . . . . 42
A.1 This table gives an explanation of the abbreviations regarding metals

and alloys that are common in thermocouples.[12] . . . . . . . . . . . . 53
A.2 Parameters used in the boundary heat flux calculation. . . . . . . . . . 53
A.3 This table presents an overview of the material parameters used in the
simulations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
v
Chapter 1
Introduction
Today the demand for better and more accurate measurements of various parameters,
e.g. temperature, increases. Some of the main reasons for this is that the industry
wants to optimize its processes in order to maximize their profits or to more closely
check certain parameters. Another reason is that in some cases government regulations
dictate that supervision of certain parameters must be done.
1.1 Background
In the oil refinery business unexpected shutdowns of machinery and or processes equal
big monetary losses, especially when it comes to critical equipment. This is why all
equipment, if possible, must be chosen so that they can be easily repaired, replaced or
so that they can last over a long period of time.
Preemraff Lysekil use thermocouples in their fired process heaters to measure the skin
temperature1 . Depending on which type of thermocouple used and how these are fasten
to the tubes, the error in the measured temperature varies.
In this thesis the skin thermocouples2 in the fired process heater designated H2301/2/3
has been considered. In H2301/2/3 naphtha3 and hydrogen are being heated in tubes
placed vertically as figure 1.1 shows. The skin thermocouples are placed both at the
ceiling and the floor of the fired process heater. Figure 1.2 gives a more detailed view of
the tubes and it also shows how one of these skin thermocouples exits the fired process
heater through the floor.
1
Skin temperature is another word for surface temperature.
2
Skin thermocouples are used for measuring surface temperature, see chapter 3.
3
Naphtha is one of the products from the distillation of crude oil. It usually contains hydrocarbons
in the range of C7 – C11.
1
Introduction
Figure 1.1: This picture shows how burners inside H2301/2/3 heat the vertically placed
tubes. [This picture has been published with the permission of Preemraff Lysekil]
Figure 1.2: This picture shows how a skin thermocouple exits the fired process heater
through the floor. [This picture has been published with the permission of Preemraff
Lysekil]
2
1.2 Objective
The reason why skin thermocouples are used in this fired process heater is threefold.
The first reason is due to government regulations. If a fired process heater exceeds a
certain output effect than government regulations dictate that supervision of the tem-
perature must be done. The second reason is to estimate the load of the H2301/2/3 and
the third and final is to estimate the lifetime of the tubes inside the fired process heater.
Every fifth year Preemraff Lysekil stops all production for inspection and maintenance
during a four week period. During this time the skin thermocouples in H2301/2/3 are
being replaced with new ones. The next stop is due in October 2007 and the question of
how reliable the current skin thermocouple design really is has risen. Furthermore, an-
other type of skin thermocouple, one that requires no welding to the tube, has captured
Preemraff Lysekils attention, see section 3.2.
1.2 Objective
Based on the information given in 1.1 this thesis has the following objective:
• To give a short repetition of the fundamentals of heat transfer.
• To give an orientation regarding the function and physics behind the

thermocouple.
• To give a short survey of the eight IEC4 standardized thermocouples that are
used in the industry today and a more detailed one about the type ’K’ and type
’N’ thermocouples.
• To give an analysis, of the present skin thermocouple as well as of the possible re-
placement skin thermocouple, regarding deviations in measured skin temperature
versus actual skin temperature.
Finally the ultimate goal of this master thesis is to:
• Give a recommendation on which type of thermocouple and which skin thermo-

couple that is most suited to use in H2301/2/3.
4
International Electrotechnical Commission.
3
Introduction
4
Chapter 2
Heat Transfer - A Short Repetition
There are basically three fundamental modes that heat can be transferred from one
medium to another, namely by conduction, convection or radiation. In all of these
modes, heat is always transferred from the higher temperature medium to the lower
one.[4]
In real applications however, it is seldom just one mode of heat transfer that is involved,
but rather a combination. This chapter will present a brief overview of each of these
modes.
2.1 Conduction
Conduction can take place in solids as well as in gases or liquids. In gases and liquids
the conduction is due to collisions and diffusion of the molecules during their random
motion. In solids, conduction is a combination of the vibrations the molecules in the
lattice and the energy transport by the free electrons.[4]
The rate at which heat is transferred by conduction through a homogeneous medium

can be described as:
dT
Q̇ = −kA (2.1)
dx
dT
where Q̇ is the heat flux, k is the thermal conductivity, A is the area and dx
is the
temperature gradient.
This equation, (2.1), is called Fourier’s law of heat conduction. The negative sign on
the right-hand-side is a consequence of the second law of thermodynamics which states
that heat must flow from a high temperature to a low temperature.[15]
In electrical circuits the flow of electric current can be calculated as the voltage potential
divided by the electrical resistance. This method is also applicable for heat flow in a
thermal circuit. This states that the heat flux, Q̇, is equal to the temperature difference,
∆T , divided by the thermal resistance, RT , as equation (2.2) shows.[15]
5
∆T
Q̇ = (2.2)
RConduction
where the thermal resistance for a plate or wall can be written as:
t
RConduction = (2.3)
Ak
or for a tube or pipe it will be as follows:
ln( rroi )
RConduction = (2.4)
2πkL
2.2 Convection
Convection is a way of energy transfer between a solid and a gas or a liquid. It can
be divided into two subcategories, free convection and forced convection. Convection is
called forced convection when the fluid is forced over the solid e.g. by a fan or a pump.
Convection is called free convection or natural convection when the fluid’s motion is
caused by buoyancy forces which are created by density differences due to temperature
variations in the fluid.[4]
Convection can be described by Newtons’s law of cooling, equation (2.5):
Q̇ = hA(Ts − T∞ ) (2.5)
where Q̇ is the heat flux, A is the surface area, Ts is the surface temperature, T∞ is
the temperature of the fluid far from the surface and h is the average convection heat
transfer coefficient.
The value of the convection heat transfer coefficient, h, is dependent on many parame-
ters such as the geometry of the surface, the physical properties of the fluid, the fluids
velocity and in some cases even the temperature difference, Ts − T∞ .[15, 4] It is quite
clear that all of these properties are not constant over a surface and that h may vary
from point to point. However, in most engineering applications an average value of the
convection heat transfer coefficient is satisfactory.
In the same manner as with conduction a thermal resistance can be calculated, as shown
in equation (2.6).
1
RConvection = (2.6)
hA
6
2.3 Radiation
and thus the heat flux can be calculated as:

Ts − T∞
Q̇ = (2.7)
RConvection
2.3 Radiation
Radiation is energy emitted from a body by electromagnetic waves. All solids, liquids
and gases emit, absorb and/or transmit radiation of varying degrees. However, radia-
tion is mostly considered a surface phenomena for opaque solids.[4]
The maximum heat flux a body can emit by radiation is given by the Stefan-Boltzmann
law, equation (2.8):
Q̇emit,max = σATs4 (2.8)

where Q̇ is the heat flux, σ is the Stefan-Boltzmann constant, A is the area and T is
the temperature in Kelvin.
It should be noted that this equation, equation (2.8), is only valid for an idealized
surface, called a blackbody. In real life no body can emit that much radiation.[4] Because
of this, the property emissivity is introduced and equation (2.8) can now be written as:
Q̇emit = σǫATs4 (2.9)

It should also be noted that the emissivity is in the interval of 0 <ǫ <1 and is dependent
on the form and texture of the body.
However, a body will not only emit but also absorb radiation. In other words it is the
net rate of radiation heat transfer that is interesting. Between two surfaces this can be
described as:
Q̇ = σǫA(Ts4 − T∞
4
) (2.10)
In this case, equation (2.10), the emissivity factor ǫ is based not only on the body, for
which the calculations are made, but also on the surrounding surfaces.
As for conduction and convection a thermal resistance can also be described for radia-
tion, as equation (2.11) shows.
Ts − T∞
RRadiation = (2.11)
Aǫσ(Ts4 − T∞
4)
The heat flux can then be calculated as:

Ts − T∞
Q̇ = (2.12)
RRadiation
7
8
Chapter 3
Skin Thermocouples
Thermocouples can be used to measure temperature at many various locations, e.g. the
temperature of the air in a duct, the temperature inside a wall or the surface temper-
ature of a tube. Another expression for surface temperature is skin temperature hence
the term skin thermocouple.
It is however not uncommon to mount the thermocouple to the surface in quite different
ways, according to the situation. In some cases it might be more practical, from an
installation point of view, to have the thermocouple welded to a plate which later on
is placed on the objects surface. In this case the whole design with thermocouple and
plate is referred to as the skin thermocouple.
In this thesis two types of skin thermocouples were studied. First the current installation
which has been in use for many years and secondly a new solution which has not been
tried out at Preemraff Lysekil yet.
9
Skin Thermocouples
3.1 The Current Skin Thermocouple

Figure 3.1 illustrates the current installation. Here the whole design including metal
plate, thermocouple, insulation and radiation shield is referred to as the skin thermo-
couple.
(a) Cross-sectional view from above (b) Cross-sectional view from side
Figure 3.1: Schematic figure of the current skin thermocouple. The tube outer diameter
is 114.3 mm and the thickness 6.02 mm. The diameter of the sheathing is just over 60
mm, the thickness is close to 4 mm and the length is 150 mm. The dimensions for the
plate on which the thermocouple is placed upon is approximately 24 x 24 x 3 mm.
In this current installation, see figure 3.1, the metal plate is welded to the tube on three
sides allowing for expansion in one direction. This is also the case with the thermocouple
which is welded on three sides to the metal plate. The protective shield placed over the
thermocouple is welded to the tube and the space between the shield and the tube is
filled with verilight, which is a kind of cement. The thermocouple used in this design
is of type ’K’. More information of this specific type of thermocouple can be found in
section 4.4.
3.2 The New Skin Thermocouple

This type of skin thermocouple is, compared to the current installation, very simple. In
fact it is just a thermocouple which has been fasten to the tube by two layers of metal
and finally an additional layer of aluminiumoxide. In order to protect the measuring
section of the thermocouple its’ sheathing is a bit thicker in that area, otherwise it is
a normal thermocouple. In order to fasten the thermocouple to the tube, it is first
smoothen by blasting. This is done in order to get a good contact surface between the
thermocouple and the tube. The thermocouple is then fasten to the tube with metal ties
and a metal spray of hastalloy fixates it. In order to make the installation mechanically
robust a second layer of Ni-200 is added upon the thermocouple and finally to enclose
and protect the thermocouple from oxidation and reduction a layer for aluminiumoxide
is added. A big advantage with this design is that no welding has to take place. As
stated in the beginning of this chapter this type of skin thermocouple has not been
10
3.2 The New Skin Thermocouple
tested at Preemraff Lysekil yet. Figure 3.2 gives a schematic picture of the new skin
thermocouple and figure 3.3 shows how it will look when fasten to a tube.
(a) Cross-sectional view from above (b) Cross-sectional view from side
Figure 3.2: Schematic figure of the new skin thermocouple. The height of the thicker
protective part of the skin thermocouple is 13 mm, the length is 38 mm and the max-
imum breadth is 16 mm. The contact surface towards the tube has the dimension
8 x 38 mm.
Figure 3.3: Picture of the new skin thermocouple when fasten to a tube. [This picture
has been published with the permission of Nils Bjørdal]
11
Skin Thermocouples
12
Chapter 4
Thermocouples
The first and foremost thing to know about thermocouples is that they do not measure
an absolute temperature, but rather a temperature difference.[20] This chapter will
give a brief description of the principle and the physics behind the thermocouple. This
chapter also includes a short presentation of different thermocouples and a concluding
section in which a more detailed study on type ’K’ and type ’N’ thermocouples is
presented.
4.1 The Principle Behind the Thermocouple

In 1821 Thomas Johann Seebeck discovered that when a conductor, e.g. a metal bar,
is exposed to a temperature gradient, it will generate a voltage. When two different
metals, which are subjected to a temperature gradient, are put together to form a closed
circuit, a continuous current will flow in the conductors. This is due to the voltage, the
thermoelectric emf1 , between the two metals. This effect is called the Seebeck effect or
the thermoelectric effect.[21]
Consider the the circuit shown i figure (4.1).

If T2 = T1 + ∆T and ∆V is the voltage observed at b-c, then the thermopower, i.e.
Seebeck coefficient, is defined by:
∆V
SAB = lim (4.1)
∆T →0 ∆T
As stated before, the thermoelectric effect occurs only when two dissimilar conductors
are used. This means that the thermoelectric effects are determined by the properties
of the individual conductors and thus:
SAB = SB − SA (4.2)
Where SA and SB are the Seebeck coefficients of the metals A and B. Both SA and SB
depend on the respective conductors’ material, molecular structure and usually also the
1
Electromotive force.
13
Thermocouples
temperature.[10] Furthermore it is important to point out that the resulting Seebeck

coefficient SAB is a non-linear parameter.
T2
a
A B
T1 DMM T1
b c
Figure 4.1: Schematic figure of a thermocouple. DMM is an abbreviation for digital-
multi-meter.
If there is a finite difference between the temperature T1 and T2 then the voltage gen-
erated in the circuit shown in figure (4.1) is:
Z T2
Vc − Vb = SB (T ) − SA (T ) dT (4.3)
T1
or rewritten with equation 4.2:
Z T2
∆V = SAB (T )dT (4.4)
T1
In some cases equation 4.4 can be simplified into:
∆V = SAB (T2 − T1 ) (4.5)
The simplification from equation (4.4) to equation (4.5) is only valid when the Seebeck
coefficient is constant enough. The variation of the Seebeck coefficient, for the eight IEC
standardized thermocouples, with respect to the temperature can be found in figure 4.3
in section 4.3.
14
4.2 The Physics Behind the Thermocouple
4.2 The Physics Behind the Thermocouple
As stated in section 4.1 it is the Seebeck effect which allows a conversion of temperature,
or rather temperature difference, into electricity. This is due to two effects, namely
charge carrier diffusion and phonon drag.[10]
4.2.1 Charge Carrier Diffusion
If there is a thermal gradient in a conductor, the charge carriers2 will diffuse from one
end to the other. This means that the hot carriers will diffuse to the cold end of the
conductor due to the lower density of hot carriers there. By the same token cold charge
carriers will diffuse to the hot end of the conductor. The conductors however are not
perfect. There are imperfections and impurities which scatter the diffusing charges and
if the scattering is energy dependent, e.g. if the hot electrons scatter more then the
cold ones, the hot and cold carriers will diffuse at a different rate. This will create a
density difference between the two ends of the conductor and thus create a potential
difference, i.e. a voltage.[10]
4.2.2 Phonon Drag
In the simplest model of charge carrier diffusion, it’s assumed that the phonons3 always
are in thermal equilibrium. This of course is not completely true and phonons will move
along the temperature gradient.[10]
The phonons will interact with crystal imperfections and with electrons and thus lose
some of their momentum. If the greater part of the interaction is between phonons and
electrons then the phonons will lose their momentum to the electrons and push them to
one side of the conductor. This will contribute to the thermoelectric field.The phonon
drag contribution will be at its’ most when the phonon-electron interaction is predomi-
nant. This occurs for temperatures T ≈ 15 ΘD where ΘD is the Debye temperature4 .[10]
2
In metals the charge carriers are electrons and in semiconductors they are electrons and holes.[10]
3
”The quantum of acoustic or vibrational energy, considered a discrete particle and used especially
in mathematical models to calculate thermal and vibrational properties of solids.”.[2, 3]
4
”Debye temperature: In the Debye model of the heat capacity of a crystalline solid, ΘD = hvkD ,
where h is Planck’s constant, k is the Boltzmann constant, and vD is the maximum vibrational fre-
quency the crystal can support.”.[16]
15
Thermocouples
4.3 The Eight IEC Standardized Thermocouples
The IEC has standardized eight types of thermocouples out of the very many that exist
in the world today.[20] Out of these eight there are five that are made of non-precious
metal alloys and three that are. These thermocouples have the designations E, J, K,
N, T and S, R, B respectively.[20] Table 4.1 gives a quick overview over the eight IEC
standardized thermocouples. In table A.1 in Appendix a more detailed explanation of
the metal and alloy abbreviations can be found.
Type IEC-code +Wire/-Wire Field of Work Atmosphere

Good in oxidizing
E Violet Chromel/Constantan -200 - 900 ◦ C
atmosphere
Not for use in
J Black Fe/Constantan -200 - 760 ◦ C oxidizing atmosphere
or in acids
Good in oxidizing
K Green Chromel/Alumel -200 - 1200 ◦ C
atmosphere
Good in oxidizing
N Pink Nicorsil/Nisil 0 - 1300 ◦ C
atmosphere
Not for use in
T Brown Cu/Constantan -200 - 370 ◦ C
oxidizing atmosphere
Ceramic
S Orange Pt-10%Rh/Pt 0 - 1480 ◦ C
protective pipe
Ceramic
R Orange Pt-13%Rh/Pt 0 - 1480 ◦ C
protective pipe
Ceramic
B Gray Pt-30%Rh/Pt-6%Rh 0 - 1700 ◦ C
protective pipe
Table 4.1: This table shows the IEC-code, which metal/alloy the positive and negative
wire is made of, the field of work and which atmosphere the specific thermocouple is
designed for.[20]
16
4.3 The Eight IEC Standardized Thermocouples
As figure 4.2 shows, the difference in output voltage between the thermocouples of type
’R’ and type ’S’ is extremely small and as table 4.1 indicates it is but a small difference
in composition in the positive thermoelement that separates them. The reason that
they are both included in the standard is that Europeans and Americans could not
agree upon which of the thermocouples, type ’S’ from Germany or type ’R’ from the
US, to use.[20]
80
E
J
T
K
70 N
R
S
B
60
50
40
emf [mV]
30
20
10
−10
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Temperature [°C]
Figure 4.2: This figure shows the output signal from the different thermocouples with
respect to the temperature in the measuring point. The reference point is assumed to
have a temperature of 0 ◦ C.[20, 19]
As can be seen in figure 4.2 the eight different thermocouples differs from each other
regarding the output in voltage with change in temperature. Since all real systems have
a certain amount of background noise, it is favorable to have a large difference in output
voltage per degree Celsius since changes in output signal will be easier to detect.
The Seebeck coefficient for a thermocouple is not linear and depends on the physical
parameters of the included thermoelements as well as the temperature, i.e. the Seebeck
coefficient is a material parameter dependent on the temperature. Figure 4.3 shows how
the Seebeck coefficient varies with increased temperature for the eight IEC standardized
thermocouples.
17
Thermocouples
90
E
J
80 T
K
N
R
70 S
B
Seebeck coefficient [ µV/°C]
60
50
40
30
20
10
−10
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Temperature [°C]
Figure 4.3: This figure shows how the Seebeck coefficient varies with temperature for
the eight IEC standardized thermocouples.[19]
4.4 The Type ’K’ and Type ’N’ Thermocouples
The type ’N’ thermocouple was developed to overcome the instabilities of the conven-
tional type ’K’ thermocouple.[6] The main difference between the two is the fact that
type ’N’ does not experience the aging process of type ’K’. The aging process will re-
sult in a drift in output signal. If this drift is large enough the output signal will be
unreliable. However, type ’N’ will not give these unreliable output signals, to the same
extent, since it will most likely fail before that occurs.
In the industry, when measuring higher temperatures, it is necessarily to protect the

thermoelements that makes up the thermocouple because ordinary insulation like PVC5
will melt. The thermocouples used are so called MIMS thermocouples, where MIMS
stands for Mineral-insulated metal-sheathed. A schematic picture of the MIMS ther-
mocouple can be found in figure 4.4.
5
Polyvinyl chloride, a type of plastic.
18
4.4 The Type ’K’ and Type ’N’ Thermocouples
Figure 4.4: Schematic figure of a MIMS thermocouple. The conventional MIMS ther-
mocouple has a sheath of Inconel or stainless steel and incorporates Ni-based thermoele-
ments, i.e. type ’K’ or type ’N’.[7] The insulation used between the sheathing and the
thermoelements is exclusively MgO.[11]
4.4.1 Corrosion
The most known and most severe material change that can occur in the type ’K’ ther-
mocouple is the corrosion of chromium. In unfavorable conditions, i.e. at 850 – 1000 ◦ C
and with a low content of oxygen, some risk of corrosion of the Chromel wire exists
even in MIMS thermocouples.[11, 20] What happens is that the chromium oxidizes at
the surface of the wire. This means that less metallic chromium is at the surface of the
wire and thus there will be a continuous flow of chromium towards the surface until
it’s depleted. The change in material composition will give a lower Seebeck coefficient
which means that the measured output emf will differ from the expected one.[11] The
change in output emf due to this corrosion may give deviations in temperature of around
10 ◦ C/1000h in operation around 1000 ◦ C.[13]
In type ’N’ thermocouples the amount of chromium is increased and the large amount
of silicon gives a stronger oxide layer. There has been no reports of corrosion problems
in type ’N’ thermocouples.[11]
4.4.2 Hysteresis and In Situ Drift
Hysteresis6 in the type ’K’ thermocouple predominates at lower temperatures, i.e.

around 400 ◦ C. For the type ’N’ alloys it is in the vicinity of 700 ◦ C that hystere-
sis is greatest. For both types of thermocouples these peaks cause changes in the net
Seebeck coefficient of about 1-1.5 %. This means that hyseteresis contributes with
about 2 ◦ C to long time drift.[7, 8]
6
Phenomenon in which the response of a physical system to an external influence depends not only
on the present magnitude of that influence but also on the previous history of the system.[1]
19
Thermocouples
In situ drifts7 in the emf, due to changes in the Seebeck coefficient, are smaller for
wires in Inconel8 sheathing than for those in stainless steel. This change is even less if
a thermoelement of type ’N’ is used rather than one of type ’K’. The change in emf is
linked to the presence of the element manganese in stainless steel, Alumel and to some
extent in Inconel.[9]
If a type ’N’ thermocouple is sheathed with stainless steel and is used at high tem-
peratures, i.e. temperatures in the vicinity of 1100 ◦ C, then an in situ drift in emf
corresponding to 25 – 30 ◦ C is to be expected when used for 1000 h. If however a
sheathing of Inconel is used instead, the variation in emf would be equivalent to 3 – 5
◦
C.[9]
4.5 Extension Wire and Compensation Cable

The difference between the extension wire and the compensation cable is that the com-
pensation cable is made out of a different material than the thermocouple but has the
same thermoelectric properties, at least in a narrow temperature range. The extension
wire on the other hand is made of the same material as the thermocouple although the
tolerance demands have a limited temperature range compared to the thermocouple.[20]
Regarding thermocouples of type ’K’ and type ’N’ there is virtually no reason for using
anything other than thermocouple cable. However, regarding thermocouples of type
’S’, ’R’ and ’B’, i.e. noble thermocouples, there is much money to be saved using a
compensation cable.
Extension wire and compensation cable for different thermocouples have been stan-
dardized. Table 4.2 shows the different designations according to the IEC.
Designation Explanation
K Type of thermocouple, according to standard.
KX Extension wire of the same material as the thermocouple.
Compensation cable of a different material than the thermocouple.
KCA
The last letter, i.e. the A, stands for the type of alloy used.
Table 4.2: This table shows the letter designation of extension wire and compensation
cable according to IEC standard.[20]
7
In situ drift refers to the changes in emf while the emf is monitored at a fixed T, i.e. with a
constant temperature profile along the MIMS thermocouple.[9]
8
Inconel is a registered trademark referring to a family of austenitic nickel-based high-performance
alloys.
20
4.6 Considerations at a Thermocouple Installation
4.6 Considerations at a Thermocouple Installation

As described in section 4.1 a thermocouple will generate a voltage, i.e. an emf, when
subjected to a temperature gradient. This means that every part of the thermocouple,
that is in a temperature gradient, will contribute to the output emf.
Consider figure 4.5. The temperature is, more or less, the same inside the oven. This
means that there will be no output emf from the part of the thermocouple that is inside
the oven. However, the part of the thermocouple that is in the wall is subjected to a
large temperature gradient. It is here that all of the output emf is generated. This
behavior is quite an important thing to have in mind when installing a thermocouple.
T6
-
x
Figure 4.5: Schematic figure of a oven. The temperature on the inside is more or less
the same and therefore will the total temperature gradient be placed in the wall.
Consider figure 4.5 again. If the extension is made inside the oven all of the emf will be
generated in the extension cable. This might be very undesirable if the extension cable
and the thermocouple cable differs in thermoelectric properties. If this in fact is the
case, then the output signal from the thermocouple will correspond to a temperature
that differs from the real temperature.
21
Thermocouples
22
Chapter 5
Method
The method used to evaluate the skin thermocouples is based on two cornerstones,
namely simulation and the experience of competent people in and around the temper-
ature measuring business. Regarding the simulations made, it is important to realize
that these are based on a model which is just an approximation of the real world and
not an exact replica.
5.1 Three–Dimension Simulation

In the three–dimension simulation process it was the program COMSOL, formerly
FEMLAB, that was used. In fact all of the simulation process, the implementation
of the geometry, the meshing, the simulation and the after study were done in COM-
SOL.
The simulations were divided into two sets with two and three cases respectively. All
geometries, see 5.1.1, were used in each case. The fundamental difference between the
simulation sets is the boundary condition on the inside of the tube, see 5.1.2. In the first
simulation set a constant heat flux is used and in the second simulation set a convective
boundary condition is utilized.
23
Method
5.1.1 Geometries
In all, four different geometries has been considered. The first geometry was that of
the current skin thermocouple. As figure 5.1 shows some simplifications were made.
For one, the actual thermocouple has been left out. Another is that, due to symmetry,
the tube has been cut in half. The reason for these simplifications is non other than to
reduce the number of cells in the mesh and thus minimize the number of calculations.
Figure 5.1: This figure shows the geometry of the current skin thermocouple on which
the simulations were made. The actual tube has an outer diameter of 114 mm, a
thickness of 6 mm and a length of 277 mm. The radiation shield has an outer diameter
of 60 mm, a thickness of 4 mm and a length of 150 mm. The plate has the dimensions
24 x 24 x 3 mm and is placed on the tube 10 mm from the end of the radiation shield.
24
The second geometry studied was the new skin thermocouple, as shown in figure 5.2 (a).
Even here some simplifications to the geometry has been made. Just as for the current
skin thermocouple, the actual thermocouple were left out and the tube were cut in two.
Furthermore the spray were idealized to cover just the thermocouple and only a small
area of the tube. The third geometry considered, figure 5.2 (b), were in fact a variant
of the second geometry. In this case however the spray were applied on a larger area of
the tube, as can bee seen in 5.2. The fourth and last geometry considered is used as a
reference in the oncoming simulations, figure 5.2 (c).
(a) Geometry 2: New Skin Thermocouple (b) Geometry 3: New Skin Thermocouple –
Larger Spray Area
(c) Geometry 4: Reference geometry
Figure 5.2: (a) shows how the new skin thermocouple with the three spray layers were
modeled in COMSOL. The actual tube has an outer diameter of 114 mm, a thickness
of 6 mm and a length of 200 mm. The skin thermocouple is 13 mm at its widest and
has a contact area of 8 x 38 mm towards the tube. The three spray layers are equally
thick with a thickness of 1 mm. (b) is just a variant of (a). The difference is that in
(b) the spray area is larger. The total spray area, excluding the contact area of the
thermocouple, is about 0.003 m2 . (c) illustrates the reference geometry, namely the
tube it self. The outer diameter is 114 mm, it has a thickness of 6 mm and a length of
277 mm.
25
Method
5.1.2 Assumptions and Boundary Conditions

The validity of a model is, as in all models, based on the boundary conditions and
assumptions made for the model. In this master thesis two simulation sets were utilized
and the assumptions and boundary conditions for these are as follows.
Simulation Set 1
For the two simulation cases in the first simulation set the following boundary conditions
were set:
• BC 1: The inside of the tube has a heat flux boundary condition.
• BC 2: The outer side of the tube and the skin thermocouple has a radiative and
convective boundary condition.
• BC 3: The cut surfaces of the tube has symmetric boundary conditions.
The boundary condition on the inside of the tube can be justified, at least from an
engineers point of view, as follows. The temperature of the naphtha is known in both
ends of the pipe and thus can a mean value of the heat transfered be calculated. The
values of the parameters used in the calculation of the heat flux can be found in table
A.2.
Simulation Set 2
For the three simulation cases in the second simulation set the following boundary
conditions were set:
• BC 1: A convective boundary condition is used on the inside of the tube.
• BC 2: The outer side of the tube and the skin thermocouple has a radiative and
convective boundary condition.
• BC 3: The cut surfaces of the tube has symmetric boundary conditions.
26
The assumptions made in the two simulation sets were as follows:
• A steady state situation.
• A constant ambient temperature.
• The heat flux from the inside of the tube to the naphtha is uniformed over the
whole pipe. (First simulation set)
• The heat transfer coefficient is the same along the inside of the tube.
(Second simulation set)
• All sides of the skin thermocouple is subjected to radiation.
• The measured temperature is the same as the temperature in the point where the
measuring point of the thermocouple is placed.
• No heat flux due to convection on the outer side of the skin thermocouple.
Regarding the last assumption, it has been made due to two facts. The first one is that
in this high temperature environment the heat flux due to radiation is much greater
than the heat flux due to convection. Secondly, it is very hard to calculate/approximate
the convective heat transfer coefficient from the flue gas to the skin thermocouple.
Since some parameters of certain materials could not be found, they were substituted
with those of materials with similar characteristics. Table A.3 shows the values of all
material parameters used in the calculations.
27
Method
5.1.3 Mesh
As stated early in chapter 5.1, COMSOL were used to generate a mesh. The mesh
generation in COMSOL is not complex per se. However if a mesh is made to fine, a
great deal of computational power is required since the equations are to be solved in all
nodes. To illustrate this problem consider figure 5.3. This figure illustrates a simple two
dimensional mesh where ∆x and ∆y are the distance between the nodes. Given a fix
area the number of nodes will increase with a decreasing distance in x and y direction
and hence a increased number of calculations for a finer mesh.
∆x
z }| {


∆y

Figure 5.3: This figure illustrates a simple 2D mesh.
In order to validate that the mesh did not have any fundamental impact on the solution,
several simulations were made with the same assumptions and boundary conditions but
with different meshes. Figure 5.4 illustrates two of the different meshes used in the three
dimensional simulations.
28
(a) Mesh with about 112 000 elements
(b) Mesh with about 8 100 elements
Figure 5.4: This figure shows two different meshes used in the mesh independence test.
As seen, the mesh in (a) is quite more dense than the one in (b)
29
Method
30
Chapter 6
Results
The results presented in section 6.1 were calculated with the values given in table 6.1
and in table 6.2. As pointed out in section 5.1.2 all geometries were used in all of the
simulation cases. As stated in section 5.1 the main difference between the two simula-
tion sets is the inside boundary condition, namely constant heat flux and a convective
boundary condition.
Simulation set 1
Parameter Value Description
Based on measurements of
Tamb , ambient temperature 800 ◦ C
the flue gas temperature
Estimated value on the
ǫ, thermal emissivity 0.9 outside of the tube
Case 1
and the radiation sheathing
Calculated from process data
q̇, heat flux 43500 W/m2
given by Preemraff Lysekil
No convective heat flux added
ho , heat transfer coefficient 0 W/m2 ◦ C
on the outside of the tube
Case 2
Calculated from process data
q̇, heat flux 43500 W/m2
Table 6.1: Values of the parameters used in the first simulation’s two cases.
31
Results
Simulation set 2
Case 1
Based on process data
hi , heat transfer coefficient 50 W/m2 ◦ C
Case 2
ǫ, thermal Emissivity 0.9 outside of the tube
Case 3
Table 6.2: Values of the parameters used in the second simulation’s three cases.
32
6.1 Three–Dimension Simulations

Just as in the beginning of chapter 5 it is important to stress that these results are
based upon a model which is a simplification of the real world. With this in mind one
should realize that there will be deviations between the model and the real world.
6.1.1 Reference Simulation – The Tube

Simulation 1
Ambient Calculated
Temperature Skin Temperature
800 ◦ C 593 ◦ C
850 ◦ C 681 ◦ C
Table 6.3: Calculated skin temperature at an ambient temperature of 800 and 850 ◦ C.
Simulation 2
Heat Transfer Calculated

Coefficient Skin Temperature
50 W/m2 ◦ C 748 ◦ C
100 W/m2 ◦ C 708 ◦ C
800 W/m2 ◦ C 548 ◦ C
Table 6.4: Calculated skin temperature for three different internal heat transfer coeffi-
cients, at an ambient temperature of 800 ◦ C.
33
Results
6.1.2 Current Skin Thermocouple
Simulation 1
(a) Ambient temperature of 800 ◦ C
(b) Ambient temperature of 850 ◦ C
Figure 6.1: Temperature field on the topside of the metal plate on which the thermo-
couple is placed upon. The z axis is parallel to the length of the tube and the tube
height is increased with decreasing z. The measuring point is placed at x = 0 and z =
0.1.
34
Calculated
Ambient Calculated
Skin Thermocouple ∆T
Temperature
800 ◦ C 593 ◦ C 574 ◦ C - 19 ◦ C
850 ◦ C 681 ◦ C 646 ◦ C - 35 ◦ C
Table 6.5: Deviation between calculated skin temperature and calculated skin thermo-
couple temperature for the current installation.
Simulation 2
Calculated
Temperature
50 W/m2 ◦ C 748 ◦ C 739 ◦ C - 9 ◦C
100 W/m2 ◦ C 708 ◦ C 695 ◦ C - 13 ◦ C
800 W/m2 ◦ C 548 ◦ C 540 ◦ C - 8 ◦C
Table 6.6: Deviation between calculated skin temperature and calculated skin ther-
mocouple temperature for the current installation in an ambient temperature of 800
◦
C.
35
Results
6.1.3 New Skin Thermocouple
Simulation 1
Figure 6.2: Temperature field of the surface between the tube and the skin thermocou-
ple. The z axis is parallel to the length of the tube and the tube’s height is increased
with decreasing z. The measuring point is placed at x = 0 and z = 0.09.
36
Calculated
Ambient Calculated
Temperature
800 ◦ C 593 ◦ C 648 ◦ C + 55 ◦ C
850 ◦ C 681 ◦ C 725 ◦ C + 44 ◦ C
couple temperature for the new skin thermocouple.
Simulation 2
Calculated
Temperature
50 W/m2 ◦ C 748 ◦ C 760 ◦ C + 12 ◦ C
100 W/m2 ◦ C 708 ◦ C 728 ◦ C + 20 ◦ C
800 W/m2 ◦ C 548 ◦ C 583 ◦ C + 35 ◦ C
couple temperature for the new skin thermocouple in an ambient temperature of 800
◦
C.
37
Results
6.1.4 New Skin Thermocouple – Larger Spray Area
Simulation 1
Figure 6.3: Temperature field of the surface between the tube and the skin thermocou-
ple. The z axis is parallel to the length of the tube and the tube’s height is increased
with decreasing z. The measuring point is placed at x = 0 and z = 0.09.
38
Calculated
Ambient Calculated
Temperature
800 ◦ C 593 ◦ C 629 ◦ C + 36 ◦ C
850 ◦ C 681 ◦ C 709 ◦ C + 28 ◦ C
couple temperature for the new skin thermocouple with a larger spray area.
Simulation 2
Calculated
Temperature
50 W/m2 ◦ C 748 ◦ C 755 ◦ C + 7 ◦C
100 W/m2 ◦ C 708 ◦ C 720 ◦ C + 12 ◦ C
800 W/m2 ◦ C 548 ◦ C 567 ◦ C + 19 ◦ C
Table 6.10: Deviation between calculated skin temperature and calculated skin ther-
mocouple temperature for the new skin thermocouple with a larger spray area, in an
ambient temperature of 800 ◦ C.
39
Results
40
Chapter 7
Discussion
7.1 Simulations
7.1.1 Simplifications
In all simulations a steady state situation has been considered. This is of course a
simplification since there will be fluctuations both in the temperature on the outside of
the tube due to variations of the fuel gas feed rate and quality of the fuel gas but also
on the inside of the tube. The fluctuations on the inside of the tube will mainly depend
on the product feed rate and the quality of the naphtha. However these variations are
fairly small over time and so the approximation to a steady state is valid.
Another simplification that has been made regards the thermal conductivity. In real life
the thermal conductivity is dependent on the temperature, in the simulations however
it has been approximated as a constant.
In the models, the thermocouple has been excluded. This has been done for two main
reasons. The first reason is simply to reduce the number of calculations that has to be
done. In order to include the thermocouple in the simulations a very fine grid must
be used and thus more calculations must be done. The second reason is to avoid mesh
related problems. In early models, which included a thermocouple, problems with the
grid occurred. In order to get a working mesh of good quality a very large number of
cells were required. Unfortunately the computational power available was not enough.
In short, the exclusion of the thermocouple in the models, will have an impact on the
results; the extent of which has not been investigated in this thesis.
41
Discussion
7.1.2 Mesh Independence
The mesh independence was controlled in the following manner. Four meshes with
different number of elements were made for the current skin thermocouple. The mean
temperature over outside boundary were then compared amongst the meshes. Table
7.1 presents the results.
Mean Value of Temperature Deviation From

Number of Elements
on the Outside Boundary [◦ C] Densest Mesh [%]
112136 596.683 –
64816 596.701 0.003
62992 596.701 0.003
8102 596.810 0.02
Table 7.1: Results from the mesh independence check.
As table 7.1 shows, there is little difference between a mesh with about 112 000 ele-
ments and one with about 8 000 elements. It is therefore quite safe to assume mesh
independence. However when a mesh with few elements is used, every element cover
a larger piece of the geometry and in certain regions, e.g. areas with small details,
problems with the resolution might arise.
7.1.3 Material Parameters
In the simulation process it is important to try to approximate the thermal conductivity

of the different materials as well as possible since it might have a large impact on the
solution. To give an example consider the current skin thermocouple and in particular
the verilight. In this simulation verilight has been approximated with ordinary cement
and with a thermal conductivity of 1.77 W/m K. If however a higher thermal conduc-
tivity is used say 35 W/m K, as the one for aluminiumoxide, the plate on which the
thermocouple is fasten to will experience a 140 ◦ C change upward in temperature in
an ambient surrounding of 800 ◦ C. This will mean that from measuring a lower tem-
perature than the real tube temperature, it will now measure a higher one. However,
it should be realized that this example was a sort of worst case scenario, where the
thermal conductivity changed with about 2000 %. For smaller uncertainties, e.g. the
one with aluminumoxide which according to [5] is in the range of 18 – 35 W/m K, the
impact on the solution will hardly be noticable. In the case with the aluminumoxide
the change from 35 W/m K, which was used in the simulations, to 18 W/m K gave an
over all change in temperature of less than 2 ◦ C in an ambient surrounding of 800 ◦ C.
42
7.2 Thermocouple
7.2 Thermocouple
The life expectancy of the thermocouples are very difficult to anticipate. Although it
can be said that the life expectancy will decrease with increased temperature. The life
expectancy will also decrease if the thermocouple is located in an unfavorable atmo-
sphere. Since the thermocouple is embedded in either cement or in two metal layers and
a aluminiumoxide layer, the atmosphere is assumed to have little impact on the thermo-
couple and thus the lifetime of the thermocouple is more dependent on the temperature
than on the atmosphere in the fired process heater. However at high temperatures some
materials will be no more dense than a sieve.
Regarding the differences between the type ’K’ and type ’N’ thermocouple, it is hard
to draw any good conclusions. Information obtained in articles, e.g. [11, 20, 12] points
to an slight advantage for type ’N’. One big advantage with the type ’N’ is that it does
not age in the same manner as type ’K’ and thus the output signal from a type ’N’
will be more reliable. Furthermore, in todays industry more and more companies are
replacing their old type ’K’ thermocouples with new type ’N’.
Thermocouples can be designed in various ways. One example of this is the placement
of the measuring point, which can be connected to ground through the sheathing or
not. However, in this thesis the effects on the thermocouple due to the placement of
the measuring point has not been investigated.
7.3 Skin Thermocouples

As can be viewed in chapter 6 the current skin thermocouple presents a 20 – 40 ◦ C
lower temperature with respect to the actual tube temperature, given the set conditions
in the first and second case in the first simulation set. The boundary condition on the
inside of the tube is however, relatively harsh. If this condition is replaced with a more
gentle one, as that in cases 1 – 3 in the second simulation set, the skin thermocouple
temperature will differ downwards with about 10 – 15 ◦ C from the actual tube tem-
perature. If this information is used in a lifetime analysis of the tubes it will result
in an overestimation of the lifetime of the tubes. Having the thermocouple in a lower
temperature however, is favorable for the thermocouple itself since it will last longer.
Regarding the new skin thermocouple it is important to realize that it will work as the
opposite of a cooling flange. If only a small area is utilized when fixating the ther-
mocouple to the tube, it will mean that a large outside area will be connected to a
small area on to the tube and thus a hot-spot will be formed. This however is heavily
dependent on which type of material used in the fixating-, i.e. spray-, process.
In a comparison between the new skin thermocouple and the current one, the simula-
tions shows that the deviation from the tube skin temperature is about the same for
the two. However the new skin thermocouple will, according to the simulations, show
a too high temperature meanwhile the current one will show a too low temperature
43
Discussion
with respect to the actual tube temperature. Figure 7.1 shows the deviation between
the calculated skin thermocouple temperature and the calculated skin temperature. It
should be noted that the heat transfer coefficient will most likely be in the range of
100 – 800 W/m2 ◦ C in the real case.
Difference in temperature − Calculated Measured Temperature Vs Calculated Tube Temperature

(Absolute Values)
60
Current Skin Thermocouple
New Skin Thermocouple
New Skin Thermocouple − Larger Spray Area
50
40
[°C]
30
20
10
0
Tamb = 800 °C Tamb = 850 °C h = 50 W/m K h = 100 W/m K h = 800 W/m K
Figure 7.1: The absolute deviance between calculated skin thermocouple temperature
and calculated skin temperature for the various simulations. It should be noted that the
current skin thermocouple shows a lower temperature than the actual skin temperature.
44
7.4 Long Time Operation of the Skin Thermocouple
7.4 Long Time Operation of the Skin Thermocouple

One important thing to have in mind is that once the skin thermocouple is in place it
will run continuously for, at least, five years and without any means of calibration. In
order to validate the thermocouple, a reference system should be implemented. It is not
important that the reference method used gives a correct reading of the temperature,
what is important is that the error in the reference system is constant.
As stated before, there is no means of replacing or calibrating the skin thermocouples

today. Once they are installed they will be in use at least five years. If however a
solution with a sheathing through the ceiling or the floor of the fired process heater was
designed, it would mean that the thermocouple could be replaced and or calibrated.
It is important to realize that the thermocouple is just a measuring device and like all
measuring devices it is in need of regular calibration in order to be reliable.
45
Discussion
46
Chapter 8
Conclusions and Recommendations
It is hard to draw any conclusions regarding which type of thermocouple to use. The
type ’N’ thermocouple has the advantage that it will not drift in output signal in the
same manner as type ’K’. Furthermore, many industries are making the switch from
type ’K’ to type ’N’. An educated guess would then be that, since they are not changing
back, the type ’N’ thermocouple is better or at least just as good as the type ’K’. In
order to validate this, it is recommended that a survey is carried out on companies that
have gone from type ’K’ to type ’N’.
According to the simulations made, there is little difference between the two different
skin thermocouples, in a measured temperature point of view. The current skin ther-
mocouple will show a lower temperature than the skin temperature on the tube and the
new skin thermocouple will show a higher one. The difference between measured and
wanted temperature is however almost the same. It is therefore recommended that the
decision on which skin thermocouple to use should be based more on other parameters,
e.g. the usage of welding or not.
The biggest drawback with the skin thermocouples is that it is not possible to replace
or remove them for calibration during operation. It is therefore highly recommended
that, if not a removable solution is implemented, a reference system is utilized.
47
Conclusions and Recommendations
48
Chapter 9
Nomenclature
Symbol Description Unit

σ Stefan-Boltzmann constant W/m2 K 4
ǫ Emissivity –
Θ Debye temperature ◦
C
A Area m2
h Convection heat transfer coefficient W/m2 K
k Thermal conductivity W/m K
L Length m
Q Heat flux W
R Thermal resistance K/W
ro Outer radius m
ri Inner radius m
SAB Resulting Seebeck coefficient V/◦ C
SA Seebeck coefficient for material A V/◦ C
SB Seebeck coefficient for material B V/◦ C
t Thickness m
T1 Lower temperature ◦
C
T2 Higher temperature ◦
C
Ts Surface temperature ◦
C
T∞ Ambient temperature ◦
C
V Voltage V
49
Nomenclature
50
Bibliography
[1] The Columbia Electronic Encyclopedia. Columbia University Press, sixth edition,
2003.
[2] The American Heritage R

Dictionary of the English Language. Houghton Mifflin
Company, fourth edition, 2004.
[3] Encyclopedia of Science and Technology. McGraw-Hill Companies, Inc., 2005.
[4] Yunus A.Çengel and Robert H. Turner. Fundamentals of Thermal-Fluid Sciences.

McGraw-Hill, 2001. ISBN 0-07-239054-9.
[5] Accuratus. Aluminum oxide. http://www.accuratus.com/alumox.html.
[6] G. Bailleul. Type n (nicrosil-nisil) thermocouple - comparison with type k and

type r. Technical report, Philips electronic instruments: Norcross, Georgia.
[7] R E Bentley. Thermoelectric hysteresis in nicrosil and nisil. Physics E: Scientific

Instruments 20, 1987.
[8] R E Bentley. Thermoelectric hysteresis in nickel-based thermocouple alloys.

Physics D: Applied Physics 22, 1989.
[9] R E Bentley and TL Morgan. Ni-based thermocouples in the mineral-insulated

metal-sheathed format:thermoelectric instabilities to 1100 ◦ c. Physics E: Scientific
Instruments 19, 1985.
[10] Robert M. Besançon, editor. The Encyclopedia of Physics. Van Nostrand, third
edition, 1985. ISBN 0-442-25778-3.
[11] G. Bruce. Termoelement typ n och k, en jämförelse. Technical report, Pentronic

AB, 1997.
[12] Noel A. Burley. Nicrosil/nisil type n thermocouples.
[13] P B Coates. The replacement of type k by nicrosil-nisil thermocouples. Physics E:

Scientific Instruments 14, 1981.
[14] Goodfellow. Fe90/cr 9/mo 1. http://www.goodfellow.com/csp/active/

gfHome.csp.
Brooks/Cole
[15] Frank Kreith and Mark S. Bohn. Principles of Heat Transfer. T homsonLearning
, sixth
edition, 2001. ISBN 0-534-37596-0.
51
BIBLIOGRAPHY
[16] David R. Lide, editor. CRC Handbook of Chemistry and Physics. Taylor and
Francis, 2007. Internet Version 2007.
[17] MatWeb. Hastelloy, sanicro 31 ht. http://www.mymatweb.com.
[18] High Temp Metals. Ni-200. http://www.hightempmetals.com/techdata/

hitempNi200data.php.
[19] Pentronic. http://www.pentronic.se/svensk/katalog/pdf/kap10.pdf.
[20] Hans Wenegård. Felkällor vid temperaturmätning. Technical report, Pentronic

AB.
[21] Wikipedia. Thermoelectric effect. http://en.wikipedia.org/wiki/

Peltier-Seebeck_effect.
52
Appendix A
Metal/Alloy Description/Composition (Weight-%)

Chromel (90.1) Nickel, (9.5) Chromium and (0.4) Silicon
Constantan (54) Copper, (44) Nickel, (1) Silicon, (0.5) Cobalt and 0.5 Iron
Fe Iron
(93.585) Nickel, (3) Manganese, (2) Aluminum,
Alumel
(1) Silicon and (0.015) Magnesium
Nicrosil (84.4) Nickel, (14.2) Chromium and (1.4) Silicon
Nisil (95.5) Nickel, (4.4) Silicon and (0.1) Magnesium
Cu Copper
Pt Platinum
Rh Rhodium
Table A.1: This table gives an explanation of the abbreviations regarding metals and
alloys that are common in thermocouples.[12]

T1 408 ◦ C Temperature on the incoming product
T2 543 ◦ C Temperature on the outgoing product
Cp 3266 J/kgK Mean value of the specific heat of the product
ρ 744 kg/m3 Density of the naphtha
x 60 Number of tubes
di 120.26 mm Inner diameter of the tubes
L 2·12.81 m Length of tubes from roof to ceiling
Table A.2: Parameters used in the boundary heat flux calculation.
53
Appendix
Material Parameter Value Usage Description

Estimation based on the
Thermal 26 Material used
A335 P9 thermal conductivity
conductivity W/m K in the tube
of a 9Cr 1Mo Steel.[14]
Thermal 12 Used in plate and
Hastelloy thermal conductivity of
conductivity W/m K in first spray layer
Hastelloy at 26 ◦ C.[17]
Thermal 44 Used in second
Ni-200 thermal conductivity
conductivity W/m K spray layer
of Ni-200 at 20 ◦ C.[18]
Thermal 35 Used in third Estimation based on
Al2 O3
conductivity W/m K spray layer 99.5 % pure Al2 O3 .[5]
744
Density The fluid
kg/m3 Values given by
Naphtha that flows
3266 Preemraff Lysekil.
Specific heat in the pipe
J/kg K
Used in the Estimation based on the
Sanicro Thermal 19 protective sheathing thermal conductivity of
31HT conductivity W/m K of the current Sanicro 31HT at
skin thermocouple 400 ◦ C.[17]
Used as filling Estimation based on the
Thermal 1.8
Verilight in the current thermal conductivity
conductivity W/m K
skin thermocouple of cement.[15]
Table A.3: This table presents an overview of the material parameters used in the
simulations.
54
Presentationsdatum Institution och avdelning
2007-03-01 Institutionen för ekonomisk och

Publiceringsdatum (elektronisk version) industriell utveckling, avdelningen för
mekanisk värmeteori och strömningslära
2007-03-14
Språk Typ av publikation

ISBN
Svenska Licentiatavhandling
ISRN LIU-IEI-TEK-A–07/0076–SE
x Annat (ange nedan) x Examensarbete
C-uppsats
D-uppsats Serietitel
Engelska Rapport
Antal sidor Annat (ange nedan)
54 Serienummer/ISSN
URL för elektronisk version

http://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-8519
Publikationens title
Simulation and Evaluation of Two Different Skin Thermocouples

A Comparison made with Respect to Measured Temperature
Författare
Joel Lundh
Sammanfattning
The demand for more accurate measurements is increasing in today’s industry. One reason for this is to optimize production
and thus maximize profits. Another reason is that in some cases government regulations dictate that supervision of certain
parameters must be followed. At Preemraff Lysekil there are basically four reasons for measuring skin temperatures inside
fired process heaters, namely; because of government regulations, in order to estimate the load of the fired process heater, to
estimate the lifetime of the tubes inside the fired process heater and finally, to determine the need of decoking. However,
only the first three of these reasons are applied to H2301/2/3. The current skin thermocouple design has been in use for many
years and now the question of how well it measures surface temperature has risen. Furthermore a new weld-free design is
under consideration to replace the old skin thermocouple design. Another question is therefore how well the new design can
measure the surface temperature under the same operating conditions as the old one. In order to evaluate this, three–
dimensional computer simulations were made of the different designs. As this thesis will show, the differences in calculated
skin thermocouple temperature and calculated surface temperature is about the same for the two designs. However, the
current design will show a lower temperature than the surface temperature, while the new design will show a higher
temperature. Regarding the core of the skin thermocouple designs, namely the thermocouple, no hard conclusions can be
drawn, although the industry appears to favor type ’N’ over type ’K’.
Nyckelord
Skin thermocouple, Heat Transfer, Type K, Type N, Seebeck coefficient

See Beck

Enviado por

Dados do documento

Descrição original:

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

See Beck

Enviado por

Direitos autorais:

Formatos disponíveis

Simulation and Evaluation of Two Different Skin Thermocouples

A Comparison made with Respect to Measured Temperature

Applied Thermodynamics and Fluid Mechanics

c 2007 Joel Lundh

Efterfrågan på bättre mätningar ökar alltjämt i dagens industri. En anledning till

• Faisal Mohamed Ali

Furthermore, I would like to give my sincere thanks to Jan-Gunnar Alexandersson

Linköping Mars 2007

2 Heat Transfer - A Short Repetition 5

8 Conclusions and Recommendations 47

4.1 Schematic figure of a thermocouple. DMM is an abbreviation for digital-

7.1 The absolute deviance between calculated skin thermocouple temper-

7.1 Results from the mesh independence check. . . . . . . . . . . . . . . . . 42

A.1 This table gives an explanation of the abbreviations regarding metals

• To give a short repetition of the fundamentals of heat transfer.

• To give an orientation regarding the function and physics behind the

Finally the ultimate goal of this master thesis is to:

• Give a recommendation on which type of thermocouple and which skin thermo-

Heat Transfer - A Short Repetition

The rate at which heat is transferred by conduction through a homogeneous medium

or for a tube or pipe it will be as follows:

Convection can be described by Newtons’s law of cooling, equation (2.5):

and thus the heat flux can be calculated as:

Q̇emit,max = σATs4 (2.8)

Q̇emit = σǫATs4 (2.9)

The heat flux can then be calculated as:

3.1 The Current Skin Thermocouple

3.2 The New Skin Thermocouple

4.1 The Principle Behind the Thermocouple

Consider the the circuit shown i figure (4.1).

temperature.[10] Furthermore it is important to point out that the resulting Seebeck

or rewritten with equation 4.2:

In some cases equation 4.4 can be simplified into:

∆V = SAB (T2 − T1 ) (4.5)

4.2 The Physics Behind the Thermocouple

4.2.1 Charge Carrier Diffusion

4.2.2 Phonon Drag

4.3 The Eight IEC Standardized Thermocouples

Type IEC-code +Wire/-Wire Field of Work Atmosphere

4.4 The Type ’K’ and Type ’N’ Thermocouples

In the industry, when measuring higher temperatures, it is necessarily to protect the

4.4.2 Hysteresis and In Situ Drift

Hysteresis6 in the type ’K’ thermocouple predominates at lower temperatures, i.e.

4.5 Extension Wire and Compensation Cable

4.6 Considerations at a Thermocouple Installation

5.1 Three–Dimension Simulation

(c) Geometry 4: Reference geometry

5.1.2 Assumptions and Boundary Conditions

• BC 1: The inside of the tube has a heat flux boundary condition.

• BC 3: The cut surfaces of the tube has symmetric boundary conditions.

• BC 1: A convective boundary condition is used on the inside of the tube.

• BC 3: The cut surfaces of the tube has symmetric boundary conditions.

The assumptions made in the two simulation sets were as follows:

• A steady state situation.

• A constant ambient temperature.

• All sides of the skin thermocouple is subjected to radiation.

Figure 5.3: This figure illustrates a simple 2D mesh.

(a) Mesh with about 112 000 elements

(b) Mesh with about 8 100 elements

6.1 Three–Dimension Simulations

6.1.1 Reference Simulation – The Tube

Heat Transfer Calculated

6.1.2 Current Skin Thermocouple