Development of Calculation Model for

Heat Exchangers in Subsea Systems

Hkon Eriksen

Master of Science in Energy and Environment

Submission date: June 2010
Supervisor:
Arne Olav Fredheim, EPT

Norwegian University of Science and Technology

Department of Energy and Process Engineering

Problem Description
1.
Description of possible concepts for cooling of process streams in subsea systems
2.
Review and description of methods for calculation of heat transfer and pressure drop for
the fluid being cooled on the tubeside and for the seawater being heated on the outside of the
tubes. Implementation of both natural and forced convection must be possible.
3.
Establish a program system in Excel or Matlab for calculation of a complete seawater
cooler. The program must allow for specification of necessary geometric parameters and stream
data, and for implementation of thermo-physical data in table format by the user.
4.
Carrying out simulation of two defined test cases (well stream cooling and compressor
after-cooling)

Assignment given: 25. January 2010

Supervisor: Arne Olav Fredheim, EPT

Preface
This master thesis is written during the tenth and final semester of the
MasterofSciencestudiesinEnergyandenvironmentalengineering,atthe
departmentofenergyandprocessengineering,NTNU.

Thepurposeofthisworkistodevelopacalculationmodelforasubseaheat
exchanger.

I would like to thank my supervisor Arne Olav Fredheim for the guidance
through the period of this study. In addition I would like to thank Erling
Nssfordiscussionsaroundheattransfer,OleJrgenNydalfordiscussions
around twophase flow and Knut Arild Mark for helping me with various
topics.

Trondheim,June2010

HkonEriksen

Abstract
Subseaprocessingcanmakeproductionfromotherwiseunprofitablefields
profitable.Insubseaprocessingcontrolledcoolingoftheprocessfluidwill
often be required. Robust and simple solutions are desirable in subsea
processing. Coolers that rely on natural convection from the surrounding
seawater are therefore interesting, but control of the process fluid outlet
temperatureishardtoobtaininsuchcoolers.

Inthisstudyacalculationmodelforsubseacoolershasbeendeveloped.The
commercial software MATLAB has been used for developing a program.
Heat transfer and frictional pressure drop correlations have been studied
andrecommendationsaremadeforthemodel.Themodelisbasedontubes
in parallel, and the tubes can be oriented vertically or horizontally. The
program allows for open, semiopen and closed arrangements on the
waterside,andbothnaturalandforcedconvectionisimplemented.

The program has been tested through simulations of two test cases and
foundtobeperformingasdesired.

ii

TableofContents
Preface............................................................................................................ i
Abstract ......................................................................................................... ii
TableofContents .......................................................................................... iii
Nomenclature ............................................................................................... vi
Listoffigures............................................................................................... viii
Listoftables ................................................................................................... x
1 Introduction ............................................................................................. 1
2 Heatexchangersinsubseaprocessing ...................................................... 2
2.1 Antisurge/gasrecycle ...............................................................................................2
2.2 Wellstreamcooler ....................................................................................................... 2
2.3 Importantaspectsinsubseacooling .................................................................... 3
3 Conceptsforsubseacooling ..................................................................... 5
3.1 Passiveparalleltubecooler......................................................................................5
3.2 Subseashellandtube.................................................................................................. 6
3.3 Subseaheatexchanger ...............................................................................................7
3.4 Subseacoolerwithpropeller...................................................................................8
3.5 Ongoingprojectsonsubseacoolers ..................................................................... 9
4 Theoreticalbackground.......................................................................... 10
4.1 Introduction ................................................................................................................. 10
4.2 Flowpattern................................................................................................................. 10
4.3 Singlephasefrictionalpressuredrop ............................................................... 11
4.3.1 DefinitionoffrictionfactorandReynoldsnumber .................................11
4.3.2 Effectofroughwalls .............................................................................................12
4.4 Tubesidesinglephaseheattransfercoefficient .......................................... 13
4.4.1 DefinitionofheattransfercoefficientandNusseltnumber ................13
4.4.2 Singlephaseheattransfercoefficient ..........................................................13
4.4.3 Effectofroughsurfaces.......................................................................................14
4.5 Purecomponentcondensationheattransfer ................................................ 14
4.5.1 Introduction .............................................................................................................14
4.5.2 Twophaseheattransfercorrelations ..........................................................15
4.5.3 Comparisonofcorrelationsforheattransfer............................................15
4.6 Multicomponentcondensationheattransfer ................................................ 17
4.6.1 Introduction .............................................................................................................17
4.6.2 Silversmethod ........................................................................................................17
4.6.3 ModificationstoSilversmethod .....................................................................18
4.7 ComparisonofHTCcorrelationsinmulticomponentcondensation ... 19

iii

4.8 Tubewallroughnesseffectsintwophaseflow ........................................... 21

4.9 Twophasepressuredrop ...................................................................................... 24
4.9.1 Staticandmomentumpressuredrop............................................................24
4.9.2 Frictionalpressuredrop .....................................................................................25
4.10 Naturalconvectionaroundimmersedbodies ............................................ 29
4.11 Externalforcedconvection ................................................................................. 32
4.11.1 Flowacrosstubes ................................................................................................32
4.11.2 Flowoverahorizontalplate(tube) ............................................................32
4.12 Combinedforcedandnaturalconvection .................................................... 33
4.13 SeawatervelocitysimpactonoverallHTC.................................................. 35
5 Heatexchangerdesigntheory ................................................................ 36
5.1 Introduction ................................................................................................................. 36
5.2 Overallheattransfercoefficient.......................................................................... 36
5.3 Calculationoftemperatureprofiletubeside................................................. 37
5.4 Watersideconfigurationsandheattransfer ................................................. 38
5.4.1 Naturalconvection................................................................................................38
5.4.2 Cocurrentflow .......................................................................................................39
5.4.3 Countercurrent .......................................................................................................39
5.4.4 Crossflow...................................................................................................................40
5.5 Tubesidemaldistribution...................................................................................... 41
5.6 Foulinginheatexchangers .................................................................................... 41
6 Modeling................................................................................................ 43
6.1 Introduction ................................................................................................................. 43
6.2 Discretizationandnumericalintegration ....................................................... 43
6.3 Temperatureprofiles ............................................................................................... 44
6.4 Pressureprofiles ........................................................................................................ 45
6.5 Evaluatingfluidproperties.................................................................................... 47
6.6 Wallinlettemperature ............................................................................................ 47
6.7 Pressuredropiteration ........................................................................................... 48
6.8 Outletconditions........................................................................................................ 50
6.9 ImplementationoftubesideHTC ...................................................................... 50
6.10 ImplementationofwatersideHTC .................................................................. 51
6.11 Transitionfromsinglephasetotwophase................................................. 51
6.12 Obtainingwatertemperatureprofileincountercurrentflow ........... 52
6.13 Runningtheprogram ............................................................................................ 54
7 Descriptionoftestcases......................................................................... 55
7.1 Introduction ................................................................................................................. 55
7.2 Wellstreamcoolercase .......................................................................................... 55
7.3 Aftercoolercase ........................................................................................................ 57
8 Resultsanddiscussion............................................................................ 60
8.1 Wellstreamcoolercase1....................................................................................... 60

iv

8.2 Wellstreamcoolercase2....................................................................................... 62
8.3 Sensitivityonwellstreamcase2 ........................................................................ 65
8.3.1 Seawatervelocityanddirection......................................................................65
8.3.2 Feedflowrate ..........................................................................................................65
8.3.3 Tubeorientation.....................................................................................................67
8.3.4 Worstcase.................................................................................................................67
8.3.5 Summaryofsensitivityonwellstreamcoolercase2.............................67
8.4 Aftercooler .................................................................................................................. 68
8.5 Sensitivityaftercooler ............................................................................................ 69
8.5.1 Seawatervelocityanddirection......................................................................69
8.5.2 Feedflowrate ..........................................................................................................70
8.5.3 Tubeorientation.....................................................................................................71
8.5.4 Summaryofsensitivityforaftercooler .......................................................71
8.6 Summaryoftestcases ............................................................................................. 71
9 Experiencesandpossibleimprovementofcalculationmodel ................. 73
10 Conclusion............................................................................................ 75
11 Bibliography ......................................................................................... 76
Appendices .................................................................................................. 78
A. Heattransfercorrelations ........................................................................................ 79
B. Pressuredropcorrelations ...................................................................................... 83
C. Testcaseusedincomparisons ............................................................................... 89
D. MATLABfunctionsandscripts............................................................................... 90

Nomenclature
A
area(m2)
c p
isobaricspecificheatcapacity(J/kgK)
C
empiricalconstantinLockhartMartinellicorrelation
C f
Twophaseenhancementfactor
D
diameter(m)
e
equivalentsandcornroughness(m)
E
parameterinFriedelscorrelation
f
frictionfactor,twophaseenhancementfactor
F
parameterinFriedelscorrelation
Fr
Froudenumber
G
massflux(kg/m2s)
g
gravityconstant(m/s2)
Gr
Grashofnumber
h
enthalpy(J/kg)
H
parameterinFriedelscorrelation
k
CorrectionfactorforBoykoandKruzhilinscorrelation
l
length(m)
m
massflowrate(kg/s)
Nu
Nusseltnumber
P
pressure(bar,Pascal),wettedperimeter(m)
Pr
Prandtlnumber
Q
duty(W)
q
heatflux(W/m2)
R
thermalresistance(K/W)
Ra
Rayleighnumber
Re
Reynoldsnumber
S
slipratio
T
temperature(K)
U
overallheattransfercoefficient(W/m2K)
v
velocity(m/s)
We
Webernumber
x
vapormassfraction,distance(m)
X
parameterinLockhartMartinellicorrelation
Y
parameterinChisholmscorrelation
z
longitudinalcoordinate(m)
Zg
ratiobetweensensibleandtotalheatflux

Greeksymbols

heattransfercoefficient(W/m2K)

volumeexpansioncoefficient(K1)

filmthickness(m)

vi


voidfraction

thermalconductivity(W/mK)

dynamicviscosity(Ns/m2)

density(kg/m3)

surfacetension(N/m)

shearstress(N/m2)

kinematicviscosity(m2/s)

twophaseenhancementfactorinFuchscorrelation

twophasefrictionmultiplier

Subscripts
acc
acceleration
c
characteristic
o
outer,external
i
internal,interfacial
f
fouling
film liquidfilm
fric
frictional
g
gas
go
gasonly
gs
gassuperficial
h
homogenous
max maximum
min minimum
l
liquid
lo
liquidonly
sp
singlephase
tp
twophase
v
vapor
w
wall

vii

Listoffigures
Figure2.11Compressionwithgasrecycleandcooler[1]
2
Figure2.21Conventionalgascompressionsystem[1]
3
Figure3.11Openparalleltubecooler
5
Figure3.12Finnedtube
6
Figure3.21Shellandtubeheatexchanger[2]
6
Figure3.31Convectionheatexchanger(AkerSubseaAS)[3]
7
Figure3.41Principalsketchofsubseacoolerwithpropeller(FMC
KongsbergSubseaAS)[4]
8
Figure3.42Powergenerationfromprocessfluid[4]
9
Figure4.21Flowpatternsduringcondensationinsidehorizontaltubes[5]

10
Figure4.22Mapillustratingtransitionboundariesbetweentheflow
patterns[5]
11
Figure4.51Liquidfilmheattransfercoefficient
16
Figure4.61Surfaceroughnesscorrectionfactors
19
Figure4.71EffectivemixtureHTCcomparison
20
Figure4.81Tubewallroughnesscorrectionfactor
22
Figure4.82TwophaseHTCwithwallroughnesscorrectionfactor
23
Figure4.83ComparisonofHTCmodelswithlowwallroughness
24
Figure4.91Frictionalpressuredrop
26
Figure4.92ComparisonoforiginalandcorrectedFuchs
27
Figure4.93FuchscorrectedandFriedelcorrelation
28
Figure4.94HTCwithcorrectedFuchs
29
Figure4.101Curvedboundarylayeraroundtube
31
Figure4.111Boundarylayerseparation[24]
32
Figure4.121OuterHTCinparallelflow
34
Figure4.122OuterHTCincrossflow
35
Figure5.21Heattransfer
36
Figure5.31Heattransferinsegment
38
Figure5.41Cocurrentflow
39
Figure5.42Countercurrentflow
39
Figure5.43Crossflowunmixed
40
Figure5.44Crossflowmixed
40
Figure6.21Discretizationandnodeplacement
43
Figure6.51Propertiesinputtable
47
Figure6.71Iterationprocessinordertogetequalpressuredropineach
tube
49
Figure6.121Iterationprocessforwatertemperatureincountercurrent
flow
53
Figure6.122Illustrationofdistancesinparallelcountercurrentflow
54
Figure7.11Hypotheticalsystemfortestcases
55

viii

Figure7.21Wellstreamphaseenvelopewithinlet(1)andoutlet
conditions(case1(2),case2(2))
57
Figure7.31Phaseenvelopeaftercoolerfluidwithinlet(1)andoutlet(2)
conditions
59
Figure8.11Temperatureprofilewellstreamcase1,G=369.83kg/m2s 60
Figure8.12HTCwellstreamcase1,G=369.83kg/m2s
61
Figure8.21Temperatureprofilewellstreamcase22inchtubes,
G=577.87kg/m2s
62
Figure8.22HTCwellstreamcase22inchtubes,G=577.87kg/m2s
63
Figure8.23Temperatureprofilewellstreamcase21inchtubes,
G=609.99kg/m2s
64
2
Figure8.24HTCwellstreamcase21inchtubes,G=609.99kg/m s
64
Figure8.31Outlettemperaturevsseawatervelocityincrossflowand
parallelflowoneinchtubedesign
65
Figure8.32Outlettemperatureasafunctionofdecreaseintotalmassflow
rate,oneinchtube
66
Figure8.33Pressuredropasafunctionofdecreaseintotalmassflowrate,
oneinchtube
66
Figure8.41Temperatureprofileaftercooler,G=462.30kg/m2s
68
Figure8.42HTCaftercooler,G=462.30kg/m2s
69
Figure8.51Outlettemperatureinparallelandcrossflowforaftercooler70
Figure8.52Outlettemperatureasafunctionofreductioninmassflowrate
inaftercooler
70
Figure8.53Pressuredropatreductioninmassflowrateinaftercooler 71

ix

Listoftables
Table4.91Recommendationstofrictionalpressuredropcorrelations...... 26
Table7.21Compositionofwellstream...................................................................... 56
Table7.22Inletconditionswellstream..................................................................... 56
Table7.23Outletconditionsforwellstream .......................................................... 56
Table7.24Geometrictubespecifications[27] ....................................................... 57
Table7.25Tubewallroughnessandthermalconductivity[7][9]................. 57
Table7.31Compositionaftercoolercase................................................................. 58
Table7.32Inletconditionsaftercooler ..................................................................... 58
Table8.11Designwellstreamcase1 ......................................................................... 61
Table8.21Designwellstreamcase22inchtubes ............................................ 62
Table8.22Designwellstreamcase21inchtubes ............................................. 63
Table8.31Oneinchtubedesignwithverticaltubes ........................................... 67
Table8.41Designaftercooler ....................................................................................... 68

1 Introduction
Subsea processing can make oil and gas production from remote or small
fields more profitable. Subsea processing is currently under development
andnewsolutionsarebeingdevelopedcontinuously.Anexampleofsubsea
processing is compression of the gas in order to maintain a production
level. Another example is cooling of the well stream for separation of
heavier hydrocarbons and separation and injection of water. In order to
achievethedesiredresultsinthementionedexamples,controlledcoolingof
thefluidmaybenecessary.Ifthesurrounding seawater is usedas coolant
thecontrolcanbehardtoachieve.

The aim of this study is to develop a calculation model for a subsea heat
exchanger for cooling. The calculation model is relevant for a cooler with
tubes in parallel, and includes both passive cooling, which utilizes natural
convectionorthenaturalmotionoftheseawater,andactivecooling,which
uses forced convection from circulating water. The commercial software
MATLABisusedforimplementingthemodel.

Thereportstartswithareviewofpossibleconceptsforcoolingofprocess
streams in a subsea processing facility. It continues with theory for
calculationofheattransferontheoutsideandinsideoftubes,inadditionto
calculation of pressure drop on the inside. Heat exchanger theory is then
presented, before the implementation in MATLAB is described. The
calculationmodelisusedononecasestudyonwellstreamcooling,andon
one case study on aftercooling of gas from a compressor. Results and
discussions around the result are presented. Conclusions and suggestions
forfurtherworkonthecalculationmodelarediscussedfinally.

2 Heatexchangersinsubseaprocessing

a process
design philosophy
2.1 Antisurge/gasrecycle

In order to maintain a production level from a field with decreasing

eam compressor
wellhead pressure,technology
subsea compression can be an option. In any
compression solution a recycle bypass is desirable. The recycle loop is
important in order to keep a minimum flow through the compressor for
avoiding a surge situation. Surge is an unstable flow situation that occurs
whenthepressureattheoutletofthecompressorbecomestohighsothat
some gas flows backwards through the compressor. Figure 2.11 shows a
principalsketchofacompressorwithagasrecycleloop.

Cooler

FT

TT

PT

TT

PT

Well stream

stream
Compressor
Figure2.11Compressionwithgasrecycleandcooler[1]

Because of the energy input to the gas through the compressor the
temperatureofthegasincreases.Therecycledgasmustbecooledinorder
to maintain the desired inlet temperature to the compressor. The
compressor performance is better at lower temperatures because the
densityofthegasdecreaseswithincreasingtemperature.

2.2 Wellstreamcooler
Figure2.21showsaconventionalsubseaboostingsolution.

Subsea process design philosophy

Conventional compressor technology (centrifugal)

Anti-surge Valve
SC

FT

TT

PT

TT

PT

Gas
Well stream
Compressor
Scrubber
Well stream

Inlet/antisurge cooler

Liquid
Pump

Figure2.21Conventionalgascompressionsystem[1]

Both the recycled gas and the well stream are cooled in a combined anti
surge and inlet cooler. In general this can also be done in two separate
coolers. The liquid formed in the cooler is separated in the scrubber and
sent to the pump, while the gas goes to the compressor before the two
streamsaremixedtogetherfortransportationinamultiphasepipelinetoa
platformoranonshoreinstallation.

2.3 Importantaspectsinsubseacooling
Outlet temperature on the cooler, whether it is a well stream cooler or an
antisurgecooler,iscrucialtocontrol.Iftheoutlettemperatureofthewell
stream cooler is too high liquid may form in the compressor because of
unsatisfyingseparationandtheperformanceofthecompressorisreduced
and it might also break down. And if the outlet temperature of the anti
surgecoolerbecomestoohighthecompressorperformanceisalsoreduced
because the inlet temperature rises. On the other hand if the temperature
becomes too low hydrates can form. Hydrates are solids of hydrocarbons
andwaterthatcanformwhenfreewaterispresentinthewellstream.At
what conditions hydrates form is dependant on well stream composition,
temperature, pressure and the existence of free water. The formation of
hydratesiscrucialtoavoid.

It is also common to place an aftercooler after the compressor. The after

coolercoolsthegasinordertoincreasethedensityofthemixturesothat
more fluid can be transported through the pipeline. The specifications for
anaftercoolerareforpipelinetransport.

Goodcalculationmodelsforsubseaheatexchangersplaysaveryimportant
rolewhendesigningasubseaprocessingfacility.

3 Conceptsforsubseacooling
Several companies design and produce subsea solutions and components,
butpublicinformationabouttheheatexchangersolutionsusedforsubsea
coolersisdifficulttofind.AthoroughsearchofthewebandScienceDirect
has been carried out. Possible solutions will be presented in this chapter.
Two patents exist at the World Intellectual Property Organization (WIPO).
Thesetwowillbepresentedherealongwithsomebasicsolutions.

3.1 Passiveparalleltubecooler
Since the cooling capacity is basically infinite at the sea bottom it seems
convenient to expose the fluid carrying tubes to seawater and rely on
natural and forced convection from the seawater. In order to increase the
heat transfer surface and decrease the pressure drop the flow should be
divided into several parallel tubes. This is known as a passive cooler
becauseitusesnoexternalmeansforcirculatingtheseawater.

Figure3.11Openparalleltubecooler

Figure 3.11 shows a birds view of a principal sketch of a parallel tube

cooler. In reality these tubes will have bends in order to make the design
morecompact.Thisisthesimplestkindofheatexchanger.Whendesigning
a subsea processing facility it is desirable to make the design robust and
simple.Inthatrespectthepassiveparalleltubecoolerisadvantageous.On
the other hand, the disadvantage is that it is difficult to control the outlet
temperature.Iftherearecurrentsintheseawatertheheattransferratecan
increasesignificantlyandthetemperaturemaybecometoolow.Iftherest
of the subsea processing facility is very reliant on a specific outlet
temperaturefromthecoolerthisdesigncancauseoperationaldifficulties.

Ifthereisnocurrentinthesurroundingseawaterpoorheattransfercanbe
experienced. This leads to increased area requirements in the heat
exchanger.Inordertomaketheheatexchangermorecompact,finnedtubes
canbeusedforincreasingtheheattransferarea.Afinnedtubeisillustrated
inFigure3.12

Figure3.12Finnedtube

3.2 Subseashellandtube
Awellknowntypeofheatexchangeristheshellandtubeheatexchanger.It
consistsofmanyparalleltubeswithashellwrappedaroundthetubes.The
flowontheshellsidecanbeorganizedinmanyways.Figure3.21showsan
illustration of one kind of shell and tube heat exchanger.This type of heat
exchangeristhemostcommonlyusedheatexchangerinonshoreprocesses.
Itisrobust,andalotofresearchhasbeendoneonit.Basedonthisitseems
promisingtousedthiskindofheatexchangersubseaaswell.

Figure3.21Shellandtubeheatexchanger[2]

This is an active cooler, which needs an external pump for circulating the
coolant on the shellside. The coolant can be seawater, but also other
coolants can be used if the heat exchanger is connected to a platform for
example. Good calculation models and software exist for this kind of heat
exchanger, which is an advantage. An additional disadvantage to the
external pump is the material costs. The shell needs to withstand the high

static pressure of the surrounding seawater if other coolants than the

seawaterareused.

3.3 Subseaheatexchanger
AtWIPOapatentisregisteredforasubseacoolerthatusesconvectionfrom
the surrounding water to cool the fluid [3]. The purpose of this heat
exchanger is to control the velocity of the seawater in order to control
outletconditions.Figure3.31showsanillustrationoftheheatexchanger.

Figure3.31Convectionheatexchanger(AkerSubseaAS)[3]

The hot fluid inlet is at (3) and outlet at (5). The seawater inlet is at (11)
and seawater outlet at (13). The fluid carrying tubes (7) are in direct
contact with the seawater. The tubes are confined in an enclosure (9). A
pump (15) is placed at the seawater outlet for generating motion in the
seawater.

The pump enables the operator to control the flow of the seawater and
hence the outlet temperature of the tubeside fluid. The heat exchanger is
hydrostaticallybalancedwiththesurroundingseawater.Theonlypressure
differenceisthatcausedbythepump.Thisfactsgivestheadvantagesthat
the walls of the heat exchanger does not have to sustain a big pressure
differenceandthemotorforthepumpdoesnothavetobebig.Inthecase
that no cooling is desirable the pump can be turned off and the seawater
inside the enclosure will reach the process fluid temperature, and act as
insulation.

The obvious disadvantage here is the need for a motor and a pump. The
motor and pump solution must be designed so that no leakages can
penetrateintothemotor. Theinventor alsodescribes a solutionwithouta
pump. If the arrangement of the heat exchanger is in such a way that the
seawater flow is vertical the temperature rise of the seawater can be
utilized. As the seawater flows through the convections section it gets
heated.Thisleadstoadensitydifferencefromseawateroutlettoinlet.This
density difference gives a pressure difference. The flow of seawater can
thenbecontrolledwithavalveattheseawaterinletoroutlet.

3.4 Subseacoolerwithpropeller
AtWIPOanotherpatentisregisteredforasubseacooler.Thispatent[4]is
for a heat exchanger that uses a propeller for generating fluid motion
aroundthefluidcarryingtubes.Figure3.41showsaprincipalsketchofthe
heatexchanger.

Figure3.41Principalsketchofsubseacoolerwithpropeller(FMCKongsbergSubsea
AS)[4]

Thehotfluidinletisat(18)andoutletat(20).Thetubearrangement(10)
canconsistofmanytubesinparallel.Adistributionunit(22)isplacedatthe
inlet,andagatheringunit(24)attheoutlet.

The purpose of the invention is to increase the flow of seawater past the
tubes by the means of a propeller. The propeller (14) is rotated by power
from the motor (30). The controller (34) regulates the motor, which
generates the desired velocity of the seawater, which in turn controls the
outlet temperature of the process fluid. In order to enhance the cooling
effectanopenendedduct(12)isplacedaroundthetubes.Theseawateris
thenforcedtoflowpastthetubes.


This invention also needs external power to drive the motor for the
propeller. The inventor suggests several solutions for generating some of
the required power from the energy of the process fluid. One of these
solutionsisillustratedinFigure3.42.

Figure3.42Powergenerationfromprocessfluid[4]

Theprocessfluidgoesthroughapropeller(64)aftertheoutlet,causingthis
propeller to rotate. The propeller is connected to a shaft (66), which is
connectedtoagenerator(68).Theenergytakenoutfromthefluidmustbe
ofcoursebeaddedinthecompressoragain,soiftheheatexchangerispart
of a compressor solution this solution has no energy savings. But it could
haveareasavingsbecauseoftheincreasedheattransfer.

3.5 Ongoingprojectsonsubseacoolers
TherearetwomasterthesesonsubseacoolersattheNorwegianUniversity
ofScienceandTechnology(NTNU),inadditiontothisstudy.Oneofthemis
a study on natural convection around tubes by computational fluid
dynamics (CFD) modeling. The other one is an experimental study for
investigating heat transfer coefficients in natural convection. At the
University in Oslo (UiO) there is a research project with a doctor degree.
The project is both an experimental and numerical investigation of heat
transferinnaturalconvectionforsubseaheatexchangers.Theresultsofthe
threementionedprojectswillbeinterestingtofollow.

rnal of

using

4 Theoreticalbackground
4.1 Introduction
Goodestimationofheattransferandpressuredroparecrucialinanyheat
exchangerdesign.Inthischapter,methodsforestimatingheattransferand
Heat pressure
and Mass
Transfer
46 (2003)
drop
inside tubes
will be33493363
discussed. Both single phase and
condensation heat transfer will be reviewed. In addition, methods for
estimating heat transfer on the outside of tubes will be discussed. Both
naturalandforcedconvectionheattransferwillbeexamined.

4.2 Flowpattern
Somemethodsuseflowpatternforbasisofcalculationofpressuredropand
heat transfer. In order to decide pressure drop and heat transfer in such
methodsknowledgeofflowpatternisrequired.Thedifferentflowpatterns
that can be experienced during condensation inside horizontal tubes are
illustratedinFigure4.21fromElHajaletal[5].

Figure4.21Flowpatternsduringcondensationinsidehorizontaltubes[5]

Numerous flow pattern maps have been proposed for predicting flow
pattern in twophase flow. El Hajal et al [5] proposed a new flow pattern
mapforcondensationbasedonaflowpatternmapdevelopedbyKattanet
al[6]forflowboiling.Figure4.22showstheflowpatternmapforR134at
40 C in 8 mm tubes. A denotes annular flow, I intermittent, SW stratified
wavy,SstratifiedandMFmistflow.

Fig. 5. Two-phase flow patterns in horizontal tubes from

Collier and Thome [22]: (a) evaporation, (b) condensation with
high liquid loading, (c) condensation with low liquid loading.

tent
flow refers to both the10plug and slug flow regimes (it
is essentially a stratified-wavy flow pattern with large

J. El Hajal et al. / International Journal of Heat and Mass Transfer 46 (2003) 33493363

and AV are the corresponding cross-sectional a

cupied by the liquid and vapor. Four of these
sions are normalized using the tube internal dia
to obtain four dimensionless variables:
hLd

hL
;
d

Pid

Pi
;
d

ALd

AL
;
d2

AVd

AV
d2

Rather than using the RouhaniAxelsson void

equation to obtain e as in Thome and El Hajal
new LMe void fraction equation is used for e in
extend application of the flow pattern map to
duced pressures. Then, from the cross-sectional
the tube A, the values of AL , AV , ALd and AVd are
determinable as
AL A1 $ e

AV Ae
Figure4.22Mapillustratingtransitionboundariesbetweentheflowpatterns[5]
Fig. 6. KattanThomeFavrat flow pattern map illustrating
The area AL here ignores any liquid formed
Gwavy transition boundaries for evaporation and condensation.
condensation on the upper perimeter of the tu
The map in Figure 4.22 is only valid for the mentioned
the
given
stratifiedfluid
angleat
hstrat
in Fig.
7 remains the only pa
conditions and is only meant as an illustration. The
transition
boundaries
which
must
be
solved
for
in
an iterative manner f
tified-wavy flow. Hence, dryout does not occur for
must
be calculated
for
actualcurve
fluidlabeled
and conditions.
methoddefined
for equation:
followingThe
geometrically
condensation.
Therefore,
thethe
transition
calculatingtheseboundariesispresentedin[5].
Gwavy can be supposed to reach its minimum value and
then continue on horizontally to the vapor quality of

ALd 182p $ hstrat $ sin2p $ hstrat &

1.0, asSinglephasefrictionalpressuredrop
shown in Fig. 6. This means that a saturated
4.3

(Note: A non-iterative expression equivalent to

is given as Eq. (19) in Part 2.) The dimensionles
vapor
enters
at
x

1:0
and
goes
directly
into
either
the
Total pressure drop consists of three components,
namely
height can
then befrictional
determined from the geome
annular flow regime or the stratified-wavy flow regime,
pressuredrop,accelerationormomentumpressuredrop,andgravitational
pression:
depending on whether G is greater or less that Gwavy . The
!
""
!
pressuredrop.Thissectiondealswithsinglephasefrictionalpressuredrop.
2p $ hstrat
other boundaries remain the same, assuming the gravityhLd 0:5 1 $ cos
controlled condensing film around the upper perimeter
2
4.3.1
does notDefinitionoffrictionfactorandReynoldsnumber
affect them. The bubbly flow regime occurs at
The geometric expression for Pid in terms of hstr
mass velocities higher than those shown on the present
Thefrictionfactorisadimensionlessparameterusedfordecidingfrictional
!
"
2p $ hstrat
map and this regime is also beyond the range of our
pressuredrop.Itisdefinedas
Pid sin
2
current database. In a mist flow, it can be envisioned
8 w
that
the
layer
of
condensate
will
be
sheared
from
the
f = 2
The transition curve from stratified-wavy flow t
wall
and
v that a new condensate layer will begin to grow
mittent and annular flow for evaporation is det
again in its place.
Equation4.31
using the updated expression of Z
urcher et al.
Fig. 7 defines the geometrical dimensions of a stratGwavy , where Gwavy is in kg/(m2 s):
ified flow, where PL is the stratified perimeter around the
(
"
or
bottom of the tube, PV is the non-stratified perimeter
16A3Vd gdqL qV
p2
dptop of the tube, h is the height of the
1 $ x$
Gwavy
2
0:5
around
the
L
25h2Ld
(
)
x2 p2 1 $ 2hLd $ 1
stratified dz
liquid,
fric Pi is the length of the interface, and AL
#)0:5
! "$F2 q

f = 2
We
2
v / (2D)
'
1
50 $ 75e$x $0:9
Fr L
Equation4.32

The friction factor is a function of geometry, surface roughness and

The non-dimensional
empirical
Reynolds number. The Reynolds number is the ratio
of inertia forces
to exponents acc
for
the
effect
of
heat
flux
on
dryout during evap
viscousforcesandisdefinedas:
are F1 q and F2 q:

Fig.
tube.

11
7. Geometrical parameters for two-phase flow
in a circular

!
"2
!
"
q
q
64:8
F1 q 646:0
qcrit
qcrit
!
"
q
F2 q 18:8
1:023
qcrit

Re =

vlc

Equation4.33

Where lc is the characteristic length and defined as the tube diameter for
singlephase flow in tubes. For laminar flow in tubes, the friction factor is
definedas

f =

64

Re

Equation4.34

The transition to turbulent flow is at Reynolds number at 2300. The

acceptedformulaforthefrictionfactorforturbulentflowinsmoothtubesis
[7]:

1
f

1/2

= 2.0 log(Re f 1/2 ) 0.8

Equation4.35

Equation 4.35 is implicit but several alternative explicit approximations

exist.ThatofBlasiusiswidelyused,butisonlyvalidforReynoldsnumbers
upto100000.

4.3.2 Effectofroughwalls
Rough walls increase the frictional pressure drop. Colebrook suggested a
formula that covers both smooth and rough tubes. The formula is an
interpolationbetweenEquation4.35andanexpressionforfullyroughflow
andcanbeusedforthewholerange.Theformulaisgivenas[7].

1
f

1/2

= 2.0 log(

e/D
2.51
+
)
3.7 Re f 1/2

Equation4.36

Where e is the equivalent sand corn roughness. Equation 4.36 is implicit,

but Haaland [8] proposed an alternative explicit formula that deviates
maximum only 2 percent from Equation 4.36 [7]. The explicit formula is
givenas

1
f

1/2

= 1.8 log((

e / D 1.11 6.9
) +
)
3.7
Re

Equation4.37

12

4.4 Tubesidesinglephaseheattransfercoefficient
4.4.1 DefinitionofheattransfercoefficientandNusseltnumber
The heat transfer coefficient (HTC) is the ratio between heat flux and the
temperature difference between the bulk fluid and the tube wall, and is
definedbyEquation4.41.

(Tb Tw )

Equation4.41

It is common to group variables into dimensionless groups in order to

reduce the number of variables. The heat transfer coefficient
nondimensionalized is known as the Nusselt number. Equation 4.42
definestheNusseltnumber.

Nu =

lc

Equation4.42

The Nusselt number can be interpreted as the ratio between convective

heat transfer to conductive heat transfer. The characteristic length for
singlephaseheattransferinsidetubesisthetubediameter.
4.4.2 Singlephaseheattransfercoefficient
In heat exchanger design it is important with good predictions of heat
transfer coefficients for singlephase flow. Also for prediction of
condensationheattransfer,becausesinglephaseHTCsareused.

SinglephaseHTCsareusuallyexpressedasafunctionofReynoldsnumber
andPrandtlnumber.ThePrandtlnumberaccountforphysicalpropertiesof
the fluid and can be interpreted as the ratio of momentum diffusivity to
thermaldiffusivityandisdefinedas:

Pr =

c p

Equation4.43

Asimpleandoftenusedcorrelation forsinglephase HTC is that of Dittus

Boelter. Equation 4.44 gives the expression for cooling. It is valid for
Prandtlnumbersbetween0.7and60andReynoldsnumbersabove10000
(fullydevelopedturbulentflow).

13

Nu = 0.023Re 0.8 Pr 0.3

Equation4.44

Equation 4.44 may give errors as large as 25 % [9]. A more accurate

correlation is proposed by Gnielinski [10]. He used the work of Pethukov
[11]andexpandedthecorrelationintothetransitionareabetweenlaminar
andturbulentflow.Thiscorrelationisthemostusedandissuggestedbythe
HeatExchangerDesignHandbook[12].Itisexpressedas:

Nu =

( fsmooth / 8)(Re 1000)Pr

1 + 12.7( fsmooth / 8)0.5 (Pr 2 / 3 1)

Equation4.45

The friction factor should be determined from an appropriate correlation

forsmoothtubes(i.e.Equation4.35).ThecorrelationisvalidforReynolds
numbersfrom2000to5000000,andPrandtlnumbersfrom0.5to2000.

4.4.3 Effectofroughsurfaces
Surface roughness disturbs the laminar sublayer where most of the
resistanceagainstheattransferis(forliquidswithPrandtlnumbers>0.6).
Hence, rough surfaces improve heat transfer. Norris [13] proposes the
followingcorrectionfortheaffectofsurfaceroughnessonheattransfer:

Nu
f n
=(
)
Nu smooth
fsmooth
Equation4.46

where n=0.68Pr0.215. For f/fsmooth > 4 the HTC no longer increases. f is

calculatedfromEquation4.37andfsmoothfromEquation4.35.

4.5 Purecomponentcondensationheattransfer
4.5.1 Introduction
Condensation heat transfer is closely linked with flow pattern. In
condensation of pure component one assumes that the vapor is saturated,
sothatalltheresistanceagainstheattransferisintheliquidthatisformed.
IftheliquidfilmformedislaminarNusseltstheory[14]yields,andtheheat
transfer is governed by the conductivity of the liquid film and the film
thickness.Butathighmassvelocitiesasubstantialinterfacialshearwillbe
present. The interfacial shear induces turbulence in the liquid film so that
convectiveeffectswillgoverntheheattransfer.

Different models exist for prediction of convective condensation for pure

components. Some use a twophase multiplier and a singlephase liquid

14

heattransfercorrelation,sometakeuseoftheinterfacialshearforcewhich
inturnrequiresknowledgeofafrictionalpressuregradient,andsomeusea
flowpatternbasedapproach.Threedifferentmodels,representingeachof
thementionedalternativeswillbepresented.

Annular flow (see Figure 4.21) will be the most commonly experienced
flowpatterninthepresentstudybecauseofhighmassvelocitiesandhigh
vapormassfractions,sothefocuswillbeonheattransferinannularflow.

4.5.2 Twophaseheattransfercorrelations
Threecorrelationshavebeenlookedat:
TheHeatExchangerdesignhandbook(HEDH)correlation[12]
ThemodifiedBoykoandKruzhilincorrelation[15]
Thomescorrelation[16]
ThethreecorrelationsandthebasicdataaredescribedinappendixA.The
HEDH correlation is chosen because it is a renowned source and the
handbookiscommonlyusedinheatexchangerdesign.ThemodifiedBoyko
and Kruzhilin correlation is interesting because the modification is made
based on experiments with highpressure hydrocarbons. Thomes
correlation is based on a database of results from experiments with
hydrocarbons and refrigerants, and has a minimum of empirically
determinedconstants.

4.5.3 Comparisonofcorrelationsforheattransfer
Figure4.51showsacomparisonbetweenthethreementionedcorrelations
forthefluidandflowsituationdescribedinappendixC.Thefluidusedhere
isnotasinglecomponentfluid,butonlytheliquidfilmHTCisshownhere.
The correlation from Thome et al deviates significantly from the others
betweenvaporfractionsof0.3andup.Forthisparticularfluidthethermal
conductivity of the liquid varies a lot as function of mass fraction. This is
clearlyseenintheHTCpredictedbytheThomecorrelation,whichismore
dependent on the liquid thermal conductivity than the other. The
correlation from Boyko and Kruzhilin is more dependent on the mass
fraction and the density ratio. While the HEDH correlation is very
dependantontheinterfacialshearforce.

15

Figure4.51Liquidfilmheattransfercoefficient

FromThomeetal[16]onecanseethatthecorrelationgivesgoodaccuracy
to the values from refrigerants in the database, 85% of the data points
(1850points)lieswithin20%ofthepredictedHTCs.Butforhydrocarbons
(921points)itonlypredicts55%oftheHTCswithin20%.Ifthefactthatit
deviates from the other correlations mentioned here also is taken into
account,thiscorrelationseemsuncertain.TheThomecorrelationisclosest
tothesinglephaseliquidHTCatx=0.

The Boyko and Kruzhilin correlation is a complete theoretical model [15],

which makes it interesting if it accounts for all the effects. The fact that it
uses a liquid only HTC ensures that it will have a smooth transition as x
approaches zero. But with the correction factor proposed by Neeraas it
hits x equal to zero 13% (11/1.15) below the singlephase liquid HTC.
NeeraasfoundthattheBoykoandKruzhilincorrelationwithhiscorrection
gavegoodagreementtohisexperimentswithpropane.Onlyoneofthe38
pointsintheexperimentsdeviateswithmorethan10%fromthemodified
correlation.

The HEDH does not mention what data the correlation is based on. But it
seems to predict relatively similar values to the corrected Boyko and
Kruzhilin.Becausetheshearforceiscalculatedfromthefrictionalpressure
gradient, the correlation for frictional pressure gradient must be accurate.
ThecorrelationfromFuchsisusedforfrictionalpressuregradientinFigure

16

4.51.TheHEDHcorrelationdeviatesmostfromthesinglephaseliquidHTC
atx=0.Thiscouldbecausedbythefactthatitisacorrelationdevelopedfor
highly sheared flow, which might not be the case at very low vapor
fractions.

4.6 Multicomponentcondensationheattransfer
4.6.1 Introduction
In multicomponent condensation the temperature at which the
condensationoccursisgliding.Thelessvolatilecomponentswillcondense
more readily than the more volatile components making the liquid and
vapor equilibrium compositions vary as condensation takes place. As
opposedtocondensationofpurecomponentsensibleheateffectswillalso
bepresent.Inthissectionthemostcommonlyusedmethodforpredicting
twophaseHTCformulticomponentmixturesispresented.
4.6.2 Silversmethod
The method is based on calculating an effective condensing side HTC that
includes the resistance in the liquid film and the resistance in the vapor
phase.TheeffectiveHTCisdefinedas[12]:

eff = (

1 Z g 1
+ )
l g

Equation4.61

whereliscalculatedfrompurecomponentcorrelation(sec.4.5),andgis
calculatedfromasinglephasecorrelation(sec.4.4)withReynoldsnumber
basedonvg (EquationA19)andthetubediameter.TheparameterZgisthe
ratiobetweensensibleheatfluxandtotalheatfluxandisdefinedas:

Z g = xcpg

dT

dh

Equation4.62

The temperature in the gas core is assumed to follow the equilibrium

condensing temperature of the mixture. Here the film thickness has been
ignored so that the heat transfer area for the vapor against the liquid is
assumedequaltotheheattransferareabetweentheliquidandthewall.At
low vapor fractions the heat transfer area between the gas and the liquid
canbemuchsmallerthanbetweentheliquidandthewall.Ifthisisthecase
the gasphase HTC should be multiplied with the ratio between the heat
transferareas.

Twoimportantassumptionshasbeenmadehere[12]:

17

1. Masstransfereffectsisneglected
2. Anytwophaseenhancementmechanismshavebeenneglected.
Correction for twophase enhancement mechanisms is dealt with in the
following section. Neeraas [15] found that mass transfer effects are small
andcanbeneglected.
4.6.3 ModificationstoSilversmethod
Price and Bell [17] suggested a correction factor for interfacial surface
roughness.ThecorrectionfactorismultipliedwiththegasphaseHTC.

dp fric
( )tp
C f = dz

dp
( )gsfric
dz

0.445

Equation4.63

ThenewgasphaseHTCis

gcf = C f g
Equation4.64

Thome et al[18] suggests the same interfacial roughness correction factor

for the gas phase HTC as the liquid phase HTC, namely fi defined by
Equation A 18. The new gas phase HTC with the interfacial roughness
correctionfactoristhen

gfi = fi g
Equation4.65

ThemethodbyThomeetalwastestedagainstmeasuredHTCsfordifferent
refrigerant mixtures. The mixtures are different compositions of
R125/236ea (98 points), R22/124 (77 points) and R290/600
(propane/butane, 659 points), and the ternary refrigerant R407C (38
points), which is a blend of R32, R125 and R124a (23,25,52 % by mass).
Massfluxesintherange100755kg/m2sandpressuresupto17bar.Vapor
fractionsfrom0.1to0.9.

Figure4.61showsfiandCfasafunctionofvaporfractionforthefluidand
flowsituationdescribedinappendixC.

18

Figure4.61Surfaceroughnesscorrectionfactors

A correlation from Fuchs[19] has been used to determine Cf. Cf becomes

verylargeasthevaporfractionapproacheszero,becausethesinglephase
gas pressure drop approaches zero as the vapor fraction approaches zero.
Atlowvaporfractionsmostoftheresistanceagainstheattransferisinthe
liquid so the fact that Cf becomes very large does not affect the effective
twophaseHTCverymuch.

4.7 Comparison of HTC correlations in multicomponent

condensation
Figure 4.71 shows effective HTCs as a function of vapor fraction for the
flowsituationandfluiddescribedinappendixC.Sincethesamecorrelation
is used for gasphase HTC for all the effective mixture HTCs, the only
additional deviation between the correlations compared to Figure 4.51 is
thegasphaseenhancementfactor.TheenhancementfactorfromPriceand
Bell,Cf,isusedonthegasphaseHTCfortheHEDHandmodifiedBoykoand
Kruzhilincorrelation,whilefiisusedontheThomecorrelationasspecified
inthedescriptionofthemodel.

19

Figure4.71EffectivemixtureHTCcomparison

ThemixtureHTCdropssignificantlyatthetransitionfromsinglephasegas
to twophase. In contrast to condensation of pure components, where the
HTCtakesasuddenjumpatthex=1(Figure4.51).Thisisbecausemostof
theresistanceisinthegasphaseathighvaporfractions.Thedropiscaused
by that the liquid that is formed smoothes out the tube walls, so the tube
walls seem smooth from the gasside. In singlephase gas flow the tube
roughness tears up the laminar sub layer where most of the resistance
against heat transfer is. In addition the interfacial roughness correction
factorsare1atthetransition,becausethewavesthatenhanceheattransfer
attheinterfaceareproportionaltothefilmthicknessandthefilmthickness
islowathighvaporfractions.

ThemodelbyThomeetal[18]predicted75%oftheHTCsmeasuredfrom
theR125/236eaandR407Cmixturestowithin10%,butonlypredicts50%
of the measured HTCs (736 points) from the hydrocarbon mixtures to
within10%.ThisimpliesthatthismodeldoesnotpredictaccurateHTCsfor
hydrocarbons.

Neeraas[15]foundthatthemodifiedBoykoandKruzhilinmethodwithCf
for interfacial enhancement gave good agreement with his experiments
with methane/propane and ethane/propane mixtures. Mass fluxes in the
experiments were in the range 150400 kg/m2s, vapor fractions between

20

0.08 and 0.98, and pressures between 16 and 40 bar. All measured HTCs
werewithin10%.

Another thing worth mentioning is that the deviation that the correlation
from Thome showed without the mixture effects is suppressed by the fact
thatitisthegasphaseHTCthatisgoverninginthatmassfractioninterval.
Though it is still lower than the other two correlations at mass fractions
from0.5andupto0.9.

4.8 Tubewallroughnesseffectsintwophaseflow
How to handle the effect of tube wall roughness in twophase flow is not
described in the literature. The tube wall roughness will act on the liquid
film, and tear up the thermal boundary layer in the liquid film. Intuitively,
thesameeffectsasinsinglephaseflowintubeswithroughwallsshouldbe
encounteredbytheliquidfilminannularflow.Atwophasefrictionfactoris
hard to define so the following tubewallroughness correction factor is
proposedbytheauthor:
n

film
film, smooth

dp

dz fric

=
dp

dz

fric, smooth

Equation4.81

wheren=0.68Prl0.275.ThisisbasedonEquation4.46.

Figure4.81showsthecorrectionfactorasafunctionofvaporfractionfor
thefluidandflowsituationinappendixC.

21

Figure4.81Tubewallroughnesscorrectionfactor

ThecorrelationfromFuchsisusedfordeterminingfrictionalpressuredrop.
Itseemslogicthatthetubewallroughnesshasmostimpactwhentheliquid
filmisthin,becausethethermalboundarylayeristhickercomparedtothe
liquidfilmthicknessathighvaporfractions,thanatlowervaporfractions.

Figure4.82showstheHTCswiththeproposedcorrectionfactor.Allofthe
correlationspredictmuchhigherHTCsthanthesinglephaseliquidHTCat
vapor fraction equal to zero. This implies that the proposed correction
factorisnotsuitedfortwophaseflow.

22

Figure4.82TwophaseHTCwithwallroughnesscorrectionfactor

ItmustbepointedoutthatEquation4.81hasnotbeentestedagainstany
datafromexperimentsandissolemnlyapropositionbytheauthor.

InthecorrelationbyBoykoandKruzhilintheeffectoftubewallroughness
isincludedintheliquidonlyHTC,whileintheHEDHcorrelationtheeffects
isindirectlyincludedintheinterfacialshearforcewhichiscalculatedfrom
thefrictionalpressuregradient.

Figure4.83showsacomparisonforthesameflowsituationasdescribedin
appendix C, but with a lower tube wall roughness (0.002e3 m). The
correlation from Thome is unchanged while the two other decreases. This
suggests that the modified Boyko and Kruzhilin or the HEDH correlation
shouldbeusedfortubeswithroughwalls.

23

Figure4.83ComparisonofHTCmodelswithlowwallroughness

Based on the discussion in the previous sections the modified Boyko and
KruzhilincorrelationforliquidHTCandCfasinterfacialenhancementfactor
on gasphase HTC seems most promising for HTC in condensation of
mixtures.

4.9 Twophasepressuredrop
4.9.1 Staticandmomentumpressuredrop
As mentioned previously pressure drop consists of three components:
frictional pressure drop, momentum or acceleration pressure drop and
pressuredropduetochangeinstaticheight,expressedas
dptot = dpstatic + dpmomentum + dp frictional
Equation4.91

Determining momentum and static pressure drop is a question of

determiningvoidfraction.Staticpressuredropisdefinedas
dpstatic = tp gdz
Equation4.92

Wheredzistheelevationchangeandthetwophasedensityisdefinedas
tp = g + (1 )l
Equation4.93

24

Staticpressuredropiszeroforhorizontaltubes.Momentumpressuredrop
is a function of condensation effects and is expressed as the difference
betweeninletandoutletconditionsandcanbecalculatedas:

dpacc

x 2
x2
(1 x)2
(1 x)2
= G
+
+


g (1 )l out g (1 )l in
2

Equation4.94

4.9.2 Frictionalpressuredrop
Frictionalpressuredropisdefinedas[20]

dp fric
dz

wP

Equation4.95

wherewistheshearstressatthetubewall,andPisthewettedperimeter
(tubecircumferenceforflowinsidetubes).

Thereexistmanyempiricalcorrelationsforfrictionalpressuredropintwo
phaseflow.Fourrenownedcorrelationshavebeencomparedinthisstudy:
LockhartandMartinelli[21]
Chisholm[22]
Friedel[23]
Fuchs[19]
Description of the correlations can be found in Appendix B. The Lockhart
and Martinelli correlation is historically the most widely used correlation,
but it does not account for mass flux effects. Therefore more modern
correlations such as Friedels or Chisholms correlations should be
employedwhenpossible[20].ThecorrelationbyFuchsisdevelopedatthe
Norwegian Institute of Technology (now NTNU), department of
refrigeration engineering (now department of energy and process
engineering). The correlation has shown good agreement with various
experimentsattheinstitute[15],andisthereforeinteresting.

4.9.2.1

Comparison between the twophase frictional pressure drop

correlations
Figure 4.91 shows the predictions of pressure drop calculated from the
differentcorrelationsforthefluidandflowsituationdescribedinappendix
C.

25

Figure4.91Frictionalpressuredrop

TheLockhartandMartinellicorrelationpredictshigherpressuredropthan
alltheothercorrelations,whiletheotherthreearefairlysimilar.Handbook
of multiphase systems [20] gives these recommendations to when the
differentcorrelationsshouldbeused:
Friedels correlation should be
l
< 1000
g
used

l
> 1000 andG>100
g

Chisholms correlations should

l
> 1000 andG<100
g

Lockhart

beused
and

Martinellis

correlationshouldbeused

Table4.91Recommendationstofrictionalpressuredropcorrelations

The correlation by Friedel is developed from a database of 25000 data

points and is for vertical upward and horizontal flow. The standard
deviation was about 4050% for twocomponent flows [20]. Chisholms
correlation is derived from the LockhartMartinelli approach, but with
additionalparameterstoaccountforfluidpropertiesandmassflux.

The correlation from Fuchs was not mentioned in the Handbook of

multiphasesystems,butNeeraas[15]foundittooverpredictthepressure
drop for his experiments. He also found that the deviation increased with

26

pressure.Basedonthisheproposedacorrectionfactorasafunctionofthe
liquidgas density ratio and the vapor fraction. The correction is found in
appendixB.

The corrected Fuchs correlation was within 10% of his experiments at

vapor fractions between 0.3 and 0.9. At vapor fractions above 0.9 he
explainedthedeviationwiththedesiretohitthesinglephasevalueatx=1.
Atvaporfractionsbelow0.3hestatedthatthedeviationcouldbecausedby
achangeinflowpattern.

Figure 4.92 shows a comparison between the corrected Fuchs and the
originalFuchs.

Figure4.92ComparisonoforiginalandcorrectedFuchs

The corrected Fuchs predicts generally lower values than the original
Fuchs, which corresponds to Neeraas experiments. For this specific case
the viscosity ratio is between 5 and 10. The handbook of multiphase
systems suggests the Friedel correlation for viscosity ratio below 1000.
Figure4.93showsacomparisonbetweenFriedelandthecorrectedFuchs.
Theresemblanceisstriking.

27

Figure4.93FuchscorrectedandFriedelcorrelation

Since Neeraas experiments were carried out with mixtures of

hydrocarbons and he found good agreement with the corrected Fuchs
correlation,thiscorrelationisusedinthemodel.

IntheHEDHcorrelationforliquidHTCduringcondensationtheinterfacial
shear force is calculated from the frictional pressure gradient. In Figure
4.71 the original Fuchs correlation is used. In addition, the frictional
pressuregradientisusedinthetwophaseenhancementfactorforthegas
phaseHTCintwophaseflow(Equation4.63).Figure4.94showsthesame
comparison, but with the corrected Fuchs for calculation of the interfacial
shearforceandthegasphaseenhancementfactor(Cf).

28

Figure4.94HTCwithcorrectedFuchs

The HEDH correlation generally becomes a little lower because the lower
predictedinterfacialshearforcebythecorrectedFuchs.

4.10 Naturalconvectionaroundimmersedbodies
If tubes are immersed in a stagnant fluid and there is a temperature
differencebetweenthetubewallandthebulkfluid,adensitydifferencewill
occur and this will induce fluid motion. The fluid motion increases heat
transfer compared to pure conduction. This heat transfer mechanism is
knownasnaturalorfreeconvection.Inthissectionmethodsforcalculating
the HTC for natural convection around immersed tubes will be presented.
ThewholesectionisbasedonHEDH[12]andCengel[9].

ForflowinchannelstheReynoldsnumberisameasureofinertiaforcesto
viscous forces, and is an indication of if the flow is turbulent or not. The
correspondingdimensionlessnumberfornaturalconvectionistheGrashof
number. The Grashof number represents the buoyancy forces to viscous
forces,andisdefinedas:

Gr =

g (Ts T )lc3

Equation4.101

Where is volume of expansion, T is the bulk fluid temperature, lc is the

characteristic length, which is the tube diameter for a horizontal tube and
theaxialpositionforaverticaltube,Tsisthesurfacetemperatureand is

29

the kinematic viscosity. For vertical tubes the characteristic length varies
withthedistancefromthetubesleadingedge.Thisisbecausetheboundary
layerincreaseswithposition.ThecriticalGrashofnumberforverticalplates
is 109, that is, the boundary layer becomes turbulent when the Grashof
numberexceeds109.Innaturalconvectionitisusualtospecifytherangeof
where a correlation is valid based on the Rayleigh number, which is the
productoftheGrashofnumberandthePrandtlnumber.
Ra = Gr Pr
Equation4.102

In natural convection around horizontal cylinders the thickness of the

boundarylayeraroundtheperipheryofthecylinderwilldiffer(seeFigure
4.101), but correlations for the average Nusselt number around the
periphery have been proposed. HEDH proposes the following correlation,
whichisvalidforbothlaminarandturbulentboundarylayer:

Nu f =

1/ 4

0.518Ra

1 + (0.559 / Pr)9 /16

4 /9

3.47 10 1 Ra
1 +
16 /9

1 + (0.559 / Pr)9 /16

1/12

Equation4.103

A correction factor for the curved thickness of the boundary layer is also
proposed:

Nu =

ln(1 + 2 / Nu f )

Equation4.104

Thefluidpropertiesshouldbeevaluatedatthefilmtemperature,definedas
the arithmetic mean between the wall temperature and the bulk fluid
temperature.

T film =

Ts + T

Equation4.105

30

Figure4.101Curvedboundarylayeraroundtube

For vertical plates (assumed valid for vertical tubes) HEDH[12] proposes
the following relationship for the laminar regime with constant wall
temperature, Rayleigh numbers (based on the axial position) between 104
and109
1/ 4

16 /9

0.492 9 /16

Nu = 0.503 Rax 1 +
Pr

Equation4.106

For Rayleigh numbers greater than 1012, turbulent boundary layer, the
followingexpressionissuggestedbyHEDH
1/ 3

16 /9

0.492 9 /16

Nu = 0.15 Rax 1 +

Pr

Equation4.107

The subscript x denotes that it is Rayleigh number based on the distance

along the tube. The transition from laminar to turbulent seems hard to
decide. But based on the mentioned ranges for Rayleigh numbers in the
laminar and turbulent regime, the transition starts at 109. Based on this
Equation 4.106 and Equation 4.107 can be combined into the following
expressionsthatyieldsbothforthetransitionandtheturbulentregime[12].
Thatis,Rayleighnumbersgreaterthan109.
9 /16 16 /9

0.492
1/4
Nu = 89.4 Pr 1 +

Pr

1/4

1/3
+ 0.15 Ra1/3
x 1000 Pr

Equation4.108

31

9 /16 16 /9

0.492
1 +

Pr

1/3

4.11 Externalforcedconvection
External forced convection is basically governed by the same mechanisms
as internal convection heat transfer described in sec 4.4, but flow around
bodies introduce additional effects. In this section it is important to
differentiatebetweenflowacrosstubes,knowascrossflow,andflowalong
tubes,whichcanbelookeduponasflowoveraflatplate.

4.11.1 Flowacrosstubes
2984
V.K. Patnana et al. / Chemical Engineering Science 64 (2009) 2978 -- 2999
Flowacrosstubesinvolveflowseparation,thatis,theboundarylayerdoes
not manage to follow the curvature of the tube and separates and forms a
wakeattherearofthecylinderasillustratedinFigure4.111.

Figure4.111Boundarylayerseparation[24]

This fact makes it hard to handle analytically, but correlations have been
formed that correlate the available data well. Cengel[9] suggests the
followingcorrelationfromChurchillandBernstein:

Nu = 0.3 +

0.62 Re1/2 Pr1/ 3

1/ 4

1 + (0.4 / Pr)2 / 3

Re 5 /8
1 +

282000

4 /5

Equation4.111

The Nusselt number is the same as for internal flow and defined by
Equation 4.42 with the tube diameter as the characteristic length. The
characteristiclengthintheReynoldsnumberisalsothetubediameter,and
thepropertiesshouldbeevaluatedatthefilmtemperature.

4.11.2 Flowoverahorizontalplate(tube)
Flowoverflatplatesinvolvesalaminarboundarylayeratstartoftheplate,
transition to turbulent, and fully developed turbulent boundary layer. The
HTC differs in the different boundary layer regimes. The transition to
turbulent boundary layer depends on many variables, but the Reynolds
numbercharacterizestheboundarylayerbest.Thecharacteristiclengthin
the Reynolds number and the Nusselt numbers is the distance x along the
platefromtheleadingedge,

32

Re =

Vx

Equation4.112

LocalNusseltnumbersforlaminarandturbulentflowarerespectively[9]
Nu = 0.332 Re 0.5 Pr1/ 3
Equation4.113

Nu = 0.0296 Re 0.8 Pr1/ 3

Equation4.114

The transition regime starts at Reynolds number of 105 and becomes fully
turbulent at about 3*106 [9]. Again, the properties should be evaluated at
thefilmtemperature.

4.12 Combinedforcedandnaturalconvection
Externalforcedconvectionisalwaysaccompaniedwithnaturalconvection,
becausetherewillalwaysbeadensitydifferenceinafluidwhenthereisa
temperaturedifference.BecauseHTCtendstoincreasewhenfluidvelocity
increases we normally ignore the natural convection affect when forced
convection is present. Cengel[9] states that the parameter Gr/Re2
represents the importance of natural convection relative to forced
convection.WhenGr/Re2<0.1naturalconvectionisnegligible,whenGr/Re2
>10 forced convection is negligible. In the range between neither is
negligibleandbothaffectsmustbetakenintoaccount.

The buoyant motion will have different direction depending on if the

surface is hot or cold, in this study the surface will be hotter than the
surrounding fluid so the buoyant motion will be upwards. If the external
flow is opposing to the buoyant motion the forced convection resists the
naturalconvection,leadingtoadecreaseinHTC,similarlyitwillleadtoan
increaseinHTCiftheexternalflowisinthesamedirectionasthebuoyant
motion, known as assisting flow. In cross flow the buoyant motion is
perpendiculartoforcedflow.Thecrossflowleadstoenhancedmixing,and
hence,increasedheattransfer.

Both Cengel[9] and HEDH[12] proposes the following expression for HTC
forcombinednaturalandforcedconvection:
n
Nucombined = (Nu nforced Nu natural
)1/n
Equation4.121

33

Theminussignforopposingflowandplusforassistingandcrossflow.The
exponentnvariesbasedongeometry.Forimmersedbodiesincrossflowor
assistingflowHEDHsuggestnequalto4.

Figure4.121showsouterHTCinparallelflowoverahorizontaltubeasa
functionofseawatervelocityfordifferenttemperaturedifferencesbetween
thetubewallandtheseawater.

Figure4.121OuterHTCinparallelflow

Thenaturalconvectiontermincreaseswithtemperaturedifference,andhas
an influence on the combined HTC up to higher seawater velocities. But
even at the highest temperature difference the forced convection term
dominatesthecombinedHTCatseawatervelocitiesfrom0.2andupward.

Figure4.122showsthesameasFigure4.121,butforcrossflow.Here,the
forced convection term dominates the combined HTC even at very low
seawatervelocities.

34

Figure4.122OuterHTCincrossflow

Aconclusionthatcanbedrawnfromthisisthatknowledgeofthedirection
of the seawater and the seawater velocity is crucial for estimating HTC on
theoutsideoftubes.Inasubseaheatexchangerwiththesurroundingwater
ascoolantthisinformationisveryuncertainandmaychangecontinuously.
In an active cooler where the coolant velocity is controlled, the outlet
temperaturecaneasierbecontrolled.

4.13 SeawatervelocitysimpactonoverallHTC
If the heat transfer mechanism is solemnly natural convection on the
outsideofthetubes,theresistanceagainstheattransferwillnormallybeon
theoutsideofthetubes.ThesmallestHTCwillbethelimitingfactorinthe
overallHTC,(seeEquation5.25).Thismeansthatwhentheresistanceison
theoutside,goodpredictionoftheouterHTCismuchmorecrucialthanthe
inner HTC. But as the velocity increases the limiting HTC can change from
the outside to the inside. If this is the case, then good prediction of inner
HTC is crucial. In the condensation of mixtures different HTC models give
different results, Therefore, choice of inner HTC correlation must be paid
greatattentionifthevelocityoftheseawaterissubstantial.

35

5 Heatexchangerdesigntheory
5.1 Introduction
Therearenumerouskindsofheatexchangers,butthetheoryinthischapter
focuses on tube heat exchangers with various types of flow orientation on
the outside of the tubes. The flow on the outside of the tubes can be
countercurrent,cocurrent,crossfloworstagnant.

Thischapterisbasedonreferences[9],[25]

5.2 Overallheattransfercoefficient
In order to decide how much heat that is transferred from the fluid inside
the tubes to the fluid on the outside of the tubes for a given temperature
differenceitisnecessarytodecidetheoverallheattransfercoefficient.The
heatistransferredbyconvectionfromthebulkfluidtothetubewall,then
by conduction through the tube wall, and finally by convection from the
tubewalltothebulkfluidontheoutsideofthetubes.

Figure5.21Heattransfer

Energyconservationgivesthattheheattransferratemustbeconstantfrom
insidethetube,tooutsidethetube.Therateofheattransferforasegment
ofthetubeisthen

dQ = dAi i (Ti Twi ) =

(Twi Two )
= dAo o (Two To )
Rw

Equation5.21

36

EliminatingthewalltemperaturesfromEquation5.21givesthefollowing
expression

dQ =

Ti To

1
1
+ Rwall +
i dAi
o dAo

Equation5.22

where

Do
Di

=
2w dL
ln

Rwall

Equation5.23

It is desirable to express the rate of heat transfer from the temperature

difference, an overall HTC, and a heat transfer area. But since the heat
transferareaisdifferentontheoutsideofthetubeandontheinsideofthe
tubeitisnecessarytospecifyiftheoverallHTCisbasedoninnerorouter
heat transfer area. The final expressions if the overall HTC based on the
outerareais:
dQ = dAoU o (Ti To )
Equation5.24

where
1

Do ln o
1
D
dAo
1
1
Di
Uo =
+ Rwall dAo + = o +
+
o
2 w
o
dAi i
Di i

Equation5.25

5.3 Calculationoftemperatureprofiletubeside
Theheattransferredfromthefluidinsidethetubesis:

dQ = m dhi
Equation5.31

Forheattransferwithoutcondensationthisexpressionsimplifiesto

dQ = mc p dTi
Equation5.32

37

Combining Equation 5.24 and Equation 5.32 gives the following

differential equation for the temperature inside the tubes in singlephase
flow:

dTi
U (T To )
= o i

dAo
mc
p

Equation5.33

TheoverallHTCandtheheatcapacitymayvaryalongthelengthofthetube,
and must be calculated for each segment. The temperature of the fluid on
the outside may not be known, depending on configuration. This will be
discussedinsection5.4.

If condensation occurs the rate of heat transferred in each segment is

calculatedfromEquation5.24,thenewenthalpyofthetwophasemixture
is calculated by subtracting the heat transferred in the segment from the
enthalpy in the previous segment. The new temperature and properties is
thenobtainedbyinterpolatingintablesatthenewenthalpy.

Figure5.31Heattransferinsegment

5.4 Watersideconfigurationsandheattransfer
The fluid on the outside of the tubes will from now on be referred to as
water, and the outside will be referred to as waterside. In this study
seawaterwillbethecoolantontheoutsideofthetubes,soitisconvenient
tousethesenames.Theequationsfortemperatureprofilesontheseawater
in the following section will only be necessary to solve in a closed
arrangement.Inanopenarrangementthetemperatureoftheseawaterwill
beconstant.

5.4.1 Naturalconvection
Innaturalconvectionthewatertemperatureisassumedconstantandequal
to seawater temperature. The outer HTC is explicitly a function of the
outside wall temperature, and implicitly through the properties, which
shouldbeevaluatedatthefilmtemperature,asmentioned.

38

5.4.2 Cocurrentflow
Inacocurrentarrangementtheexternalflowisparalleltotheinternalflow
asillustratedinFigure5.41.

Figure5.41Cocurrentflow

TherateofheattransferredtothewateriscalculatedfromEquation5.24,
and the temperature profile of the water is calculated by integrating the
followingequationfromtubeinlettotubeoutlet:

dTo U o (Ti To )
=

dAo
m o cp
o

Equation5.41

5.4.3 Countercurrent
In a countercurrent configuration the external flow is flowing in the
oppositedirectionoftheinternalflow,illustratedbelow.

Figure5.42Countercurrentflow

In this case the external and internal temperatures are not known at the
same location. This leads to that the water temperature at the inlet of the
tube has to be guessed. And the following differential equation integrated
fromtubeinlettotubeoutlet.

dTo
U (T To )
= o i

dAo
m o cp
o

Equation5.42

Ifthecalculatedwatertemperatureatthetubeoutletisequaltotheactual
water temperature the problem is solved, if not, a new guess of water
temperature at the tube inlet has to be made. This must be repeated until
theactualwatertemperatureisreachedatthetubeoutlet.

39

5.4.4 Crossflow
In a cross flow configuration the flow is perpendicular to the tube axis.
Crossflowcanbefurtherclassifiedintounmixedormixedflow.Thecross
flow is said to be unmixed if the flow is prohibited from moving in the
transverse direction with physical separation, or if it is assumed not to
move in the transverse direction. Figure 5.43 illustrates this from a birds
view.

Figure5.43Crossflowunmixed

Thetemperatureofthewaterwillvaryalongthelengthofthetube,andcan
becalculatedas:

To (z, ntube + 1) = To (z, ntube) +

dQz

m o cpo
Equation5.43

WherentubedenotesthetubenumberanddQzistheheattransferredata
distancezalongthetube.dQziscalculatedfromEquation5.24.

Inmixedcrossflowthefluidisfreetomoveinthetransversedirection,and
the temperature does not vary along the axis of the tube because of this
mixing.

Figure5.44Crossflowmixed

40

Thewatertemperatureiscalculatedas:

To (ntube + 1) = To (ntube) +

Qntube

m o cpo
Equation5.44

WhereQntubeisthetotalheattransferredfromtheprevioustube.

5.5 Tubesidemaldistribution
Varyingexternalconditionscaninducemaldistribution.Thepressuredrop
through each tube must be equal for the flow to be stable. If the water
temperature varies outside the different tubes the heat transfer will be
affectedandthiswillaffectthepressuredropaswell.Naturewilldistribute
the total mass flow into the tubes so that the pressure drop in each tube
becomesequal.Thisleadstodifferentoutletconditionsineachtube.

Intwophaseflowtherecanalsobemaldistributionattheinlet.Iftheliquid
is distributed unequally through the header nature distributes the gas in
suchawaythatequalpressuredropisexperiencedineachtube.Thisisnot
included in the model because this would lead to different equilibrium
condensationcurvesineachtube.Inordertosolvethisanequationofstate
mustbeimplemented,andthatisnotunderthescopeofthisstudy.

5.6 Foulinginheatexchangers
Foulingisacoatingthatformsinheatexchangersovertime.Theformation
offoulingisanadditionalresistanceagainstheattransfer.Inasubseaheat
exchanger fouling on the outside of tubes is likely to occur. Extra heat
transfer area is necessary in order to reach the specified outlet
temperature.Sinceitisdifficulttoaddextraheattransferareaafteraheat
exchanger has been installed, future fouling should be included in the
design process. The fouling resistance is included in the overall HTC as
follows:

Do ln o
D
1
Di
Uo = o +
+ R f ,o +
2 w
o
Di i

Equation5.61

Where Rf,o is the fouling resistance on the outside of tubes. There are
several types of fouling, and the principal processes that cause the fouling

41

classify them. The types of fouling that is most likely to form in a subsea
heat exchanger are deposition fouling, corrosion fouling, particle fouling
andbiologicalfouling.Thesearedescribedin[25].Thecoatingofthetubes
will also affect the heat transfer and this resistance can be included in the
foulingresistance.

42

6 Modeling
6.1 Introduction
Thischapterdealswithhowthecalculationmodelisbuilt.Thecalculation
model is implemented in the commercial software MATLAB (Matrix
Laboratory). The most important parts of the programming, and special
considerations is mentioned under this chapter. An overview and
description of the routines and the complete mfiles can be found in
appendixD.

The model is based on tubes in parallel with various configuration of the

seawater flow. Both horizontal are vertical tubes are implemented. The
tubesareassumedstraight(nobends).

6.2 Discretizationandnumericalintegration
Inordertousenumericalintegrationthecontinuoustubemustbedivided
into discrete segments where the fluid temperature, pressure and fluid
propertiesareconstant.Thenodescaneitherbeplacedinthemiddleofthe
segmentorattheendofeachsegment.Hereitischosentoplacethematthe
endofeachsegment,asshowninFigure6.21.

Figure6.21Discretizationandnodeplacement

The differential equations Equation 5.33, Equation 5.41 and Equation

5.42mustbediscretizedinordertosolvethemnumerically.Sincethetube
diameter is constant the left hand side of the mentioned equations
simplifiestoconstant*dT/dz,wherezisthedistancealongthetube.Bythe
useaforwarddifferencethisgradientcanbediscretizedasfollows[26]

dTi Ti +1 Ti
=

dz
z

Equation6.21

BysolvingtheequationwithrespecttoTi+1theforwardEulerscheme[26]is
obtained

43

 dT

Ti +1 = Ti + z
dz i
Equation6.22

Eulersmethodisafirstordermethod.Thismeansthattheglobalerrorisof
order z. This requires small values of z in order to achieve accurate
results.

6.3 Temperatureprofiles
InsertingfordTi/dzfromEquation5.33givesthefollowingexpressionfor
thetubesidetemperatureinsinglephaseflow.

Ti +1 = Ti z

U o (Ti To,i ) Do
.

m cp
Equation6.31

Ti denotes the bulk temperature inside the tubes in ith node, and To,i
denotesthebulktemperatureontheoutsideofatubeattheithnode.The
overallHTCandspecificheatcapacityshould(asimpliedbyEquation6.22)
beevaluatedattheconditionsoftheithnode.Similarlytheexpressionsfor
water temperature in cocurrent and countercurrent flow in a closed
arrangementarerespectively

To,i +1 = To,i z

U o (Ti To ) Do

m o cpo
Equation6.32

To,i +1 = To,i + z

U o (Ti To ) Do

m o cpo
Equation6.33

In the case of an open heat exchanger the mass flow of the water will be
infiniteandthewatertemperaturewillbeconstant.

In condensing flow an explicit relation cant be formed without use of an

equation of state. This is not under the scope of this study. Instead the
temperatureisassumedtofollowtheequilibriumcondensingtemperature
slope for the mixture. The temperature in the next node is obtained by
calculatingtheheattransferredintheithsegmentas:
Qi = U o Do (Ti To,i )z
Equation6.34

Thenewenthalpyis

44

hi +1 = hi Qi

Equation6.35

Thenthe temperatureis obtainedbyinterpolating in a table with the new

enthalpy.

The outside wall temperature, TW, of the tube is necessary in order to

calculate the properties of the water. As mentioned earlier, water
propertiesshouldbeevaluatedatthefilmtemperature,0.5(TW+TO).Froma
heatbalancethewalltemperaturecanbecalculatedas:

Tw,i +1 = Ti

Uo
(Ti To,i )
U iw

Equation6.36

WhereUiwisthetotalHTCfromthefluidinsidethetubetothetubeoutside
wall
1

D
D
D
U iw = o + o ln o
Di i 2 w Di
Equation6.37

Allthementionedtemperatureprofilesinthissectionmayvaryfromtube
to tube if the water flow is in cross flow with the tubes so that the water
temperature differs from tube to tube. In that case the temperature terms
should have an additional index. For example Equation 6.31 could be
writtenasfollows

Ti+1 = Ti z
j

U o Ti j To,ij Do
.

m cp

Equation6.38

But in order to keep the expressions as plain as possible this index is not
includedintheexpressions.Thesamegoesforthepressureprofilesinthe
followingchapter.

6.4 Pressureprofiles
The pressure profile in a tube is calculated in the same manner as the
temperatureprofiles.

dp
Pi +1 = Pi + z
dz i
Equation6.41

45

As mentioned earlier the pressure gradient is the sum of three

contributions. For singlephase flow the frictional pressure gradient is
calculated from Equation 4.32 with Equation 4.37 for the friction factor.
The expressions for momentum and static pressuredrop must be
discretized. For the static pressure drop in singlephase flow in vertical
tubesthefollowingexpressionisused

+ i +1
dp
= i
g

dz static,i
2

Equation6.42

Equation4.94discretizedgivesthemomentumpressuredropas:
2
2
2
2
1 xi +1 ) x i
1 xi )
(
(
G 2 x i+1
dp
=
+
+

dz momentum,i
z i +1g,i +1 (1 i +1 )l,i +1 i g,i (1 i )l,i

Equation6.43

It is possible to use the conditions at i+1 because the temperature is

calculated before the pressure. Equation A 15Equation A 17 is used for
calculatingthevoidfraction.

If the tubes are vertical the static pressure drop in twophase flow is
calculatedbythefollowingdiscretization

g,i +1 + g,i i +1 + i l,i +1 + l,i i +1 + i

dp
= g

1 2
2 +
dz static,i
2
2

Equation6.44

For the frictional pressure gradient in twophase flow the corrected

correlation by Fuchs is used. Introducing the discretized syntax and
rearranging,theexpressionbecomes

dp

dp
dp
dp
=

(x
)

i

dz frictional,i
dz fric,go,i dz fric,lo,i dz fric,lo,i

f l,i , xi
g,i

Equation6.45

Thesinglephasepressuregradientsarecalculatedasdescribedforsingle
phaseflow,butwiththerespectiveReynoldsnumbers.

Totalpressuregradientis:

dp dp
dp
dp

=
dz i
dz momentum,i dz static,i dz frictional,i
Equation6.46

46

The momentum term is chosen positive because during cooling and

condensation this term increases the pressure. This sign convention has
beenchoseninordertogetallthequantitiespositive.

6.5 Evaluatingfluidproperties
Anequationofstateisnotincludedinthecalculationmodel.Therefore,the
user must specify thermodynamic and physical properties. This is most
convenientlydoneinatableinexcelformatorsuch.Inordertoextractthe
properties at a specific temperature, interpolation in the table must be
carriedout.MATLABcanreadexcelfilesandextractthedatainthetables.
In order to generalize the program code the table should be in a specified
form.TheformisshowninFigure6.51

Figure6.51Propertiesinputtable

When the table is read by MATLAB and the data is saved in variables in
MATLABthebuiltinfunctioninterp1canbeusedtointerpolateinthetable.
The method of interpolation can be specified in the interp1 function.
Problems can occur at the transition from single phase to twophase flow
because there are no points to interpolate between. In order to avoid this
problem a method that extrapolates if necessary is chosen. The method
spline uses cubic spline interpolation, and extrapolates if necessary. This
methodischosenintheprogramcode.

6.6 Wallinlettemperature
At the inlet, the outside wall temperature is not known. The outer wall
temperatureisnecessaryinordertoobtaintheouterHTC.Sincetheseare
both unknown an iteration loop is necessary for deciding outer HTC and

47

outside wall temperature at the inlet. From a heat balance the following
relationcanbeestablishedatthetubeinlet:
dAoU o (T1 To,1 ) = dAoUiw (T1 Two,1 )
Equation6.61

Uo and Uiw are functions of the wall temperature. An iteration loop where
Two is changed until Equation 6.61 is satisfied within a given tolerance is
implementedinthefunctiontwallstart.

6.7 Pressuredropiteration
Thetemperatureandpressureprofilesaresolvedforeachtubeseparately.
A forloop with a given step size (z) is used for integrating through each
tube. This is done in a tubebytube manner. The first tube is integrated
over the length of the tube first, then the second and so on. Since the
seawater temperature may differ around each tube (in cross flow with a
finitewatermassflowrate)thepressuredropmayalsovary,asmentioned
before.Whenthetemperatureandpressureprofilesineachtubehavebeen
achievedthepressuredropineachtubeiscompared.Thepressuredropin
each tube is saved in a variable, and with help of the builtin MATLAB
functionsmaxandmintheminimumandmaximumpressuredropisfound.
ThebuiltinMATLABfunctionsmaxandminalsoreturninwhichtubethe
maximum and minimum pressuredrops are. If the difference between
maximumandminimumpressuredropisgreaterthanagivetolerancethe
totalmassflowhastoberedistributeduntilthedifferenceinpressuredrop
isunderthetolerancelevel.Figure6.71illustratesthisiterationprocess.

48

Figure6.71Iterationprocessinordertogetequalpressuredropineachtube

TheunderlinedsymbolsinFigure6.71arevectors.
The first guess is of course to distribute the total mass flow rate evenly
throughallthetubes.Whenredistributingthemassflowratethetotalmass
balance must be fulfilled. That is, if the mass flow rate is increased in one
tube,itmustbereducedbythesamequantityinanothertube.Inorderto
gettheprogramtorunreasonablyfasttheredistributionshouldbedonein
a smart way. The following logic is used in the program code for deciding
howtoredistributethemass:
p ~ lG 2
Equation6.71

pmax pmin
~ G 2l
2

Equation6.72

1/2
pmax pmin )
(
m =
D2

2l

Equation6.73

49

missubtractedfromthetubewiththehighestpressuredrop,andadded
to the tube with the least pressuredrop. It seems reasonable that m
should be high when there is a big difference in pressure drop, that m
should be smaller for longer tubes than shorter, and that m should be
biggerfortubeswithbiggerdiameterthanforsmallerdiametertubes.

With this iteration method the mass flow rates in only two tubes are
updated in each iteration. This has the advantage that its easy to respect
the mass balance, and its easy to implement. The disadvantage is that if
therearebigdifferencesinexternalconditionsineachtubetheruntimeof
theprogramincreasessignificantly.

Thetoleranceischosenasarelativetoleranceof103,thatis
tolerance = 10 3 pmax
Equation6.74

6.8 Outletconditions
Theoutlettemperature,Tout,isatypicalspecificationforaheatexchanger.If
the temperature profiles are different in the tubes, the outlet temperature
and mass fraction (if condensation occurs) in each tube will differ. If the
mass flow rate is different in each tube Tout cant be obtained by a simple
averageoftheoutlettemperaturesineachtube.

Toutisobtainedfromthetotaldutyoftheheatexchanger.Thetotalenthalpy
at the outlet is the total inlet enthalpy minus the total duty of the heat
exchanger. Tout is obtained from interpolation at the total enthalpy at the
outlet. The mass fraction at the outlet is then obtained by interpolating at
thecalculatedoutlettemperature.

Thepressureineachtubeshouldobviouslybeequal,butbecauseofafinite
tolerancelevelsmalldeviationsmayoccur.Theoutletpressureiscalculated
fromanaverageofthepressuredrops.

6.9 ImplementationoftubesideHTC
ThevalueZginEquation4.61isimplementedasfollows

Z g,i = xi c pg,i

Ti Ti 1

hi hi 1

Equation6.91

If there is twophase flow at the inlet the following approximation is used

forZgattheinlet

50

Z g,1 = x1c pg,1

T1 (T1 + 0.5)

h1 hT1 + 0.5

Equation6.92

As mentioned before, choice of tubeside HTC in condensing flow must be

done with care. It is possible to choose between the three mentioned
correlationsfortwophaseflow(see6.13).

ThecorrelationbyGnielinski(Equation4.45)isusedforsinglephaseflow
withEquation4.46forcorrectionforroughtubewalls.Thesilvermethod
(Equation4.61)isusedforHTCincondensationofmixtures.FortheHEDH
correlation in twophase flow, Cf from Equation 4.63 is used as
enhancementfactoronthegasphaseHTC,andthecorrectedFuchsisused
forcalculatingtheinterfacialshearforce.

Cf is also implemented as enhancement factor for gasphase HTC with the

Boyko and Kruzhilin correlation for liquid HTC. The modification made by
NeeraasisimplementedsoliquidfilmHTCiscalculatedfromEquationA10.
The Thome correlation is used in its original form. The DittusBoelter
correlation (Equation 4.44) is used for gasphase HTC for both the HEDH
andBoykoandKruzhilincorrelation.TheliquidonlyHTCintheBoykoand
Kruzhilin correlation is calculated from Gnielinski (Equation 4.45) with
correctionfortubewallroughness.

6.10 ImplementationofwatersideHTC
FornaturalconvectionaroundhorizontaltubesEquation4.103combined
withEquation4.104isused.Fornaturalconvectionaroundverticaltubes
Equation 4.106 and Equation 4.108 are used for laminar and
transition/turbulentrespectively.

In forced convection in parallel flow Equation 4.113 and Equation 4.114

areusedforlaminarandturbulentflow,respectively.IncrossflowEquation
4.111 is used. Equation 4.121 is consequently used for determining the
combined HTC from natural and forced convection. The value of n is set
equalto4assuggestedbyHEDH[12]andCengel[9].

6.11 Transitionfromsinglephasetotwophase
Thetransitionfromsinglephasetotwophaseflowisobtainedbychecking
if the mass fraction of the vapor is between 0 and 0.99. At the transition
there is a change in correlation for tubeside HTC. The HTC drops
significantlyforthetestcase(appendixC)(seeFigure4.71).Thisleadstoa

51

discontinuityintheoverallHTCandtheoutsidewalltemperature.Though
it is not physically logic that the walltemperature takes a sudden drop, it
doesnotcauseanyproblemsforrunningtheprograminthecurrentstate.
Butgenerally,discontinuitiescanleadtothataprogrambreaksdownatthe
discontinuity. This can be dealt with by using single phase HTC down to
x=0.95forexample,orusingsomekindofinterpolationfromx=1tox=0.95
in order to get a smooth profile on the tube side HTC and the wall
temperature.

Since the flow pattern at vapor fractions between 0.95 and 1 probably is
liquiddropletsentrainedinthegascore,itseemslogictousesinglephase
expressions for tube side HTC down to x=0.95. The models used are for
annularflow,andannularflowrequiresenoughliquidtoformaliquidfilm
aroundtheperipheryofthetube.

6.12 Obtainingwatertemperatureprofileincountercurrentflow
As mentioned before, the water temperature and tubeside fluid
temperature are not known at the same position in countercurrent flow.
Therefore, the water temperature at the tubeside inlet is guessed. This is
onlynecessaryintheclosedcase.Afirstguessisthatthewatertemperature
has a rise of 5 degrees. The water temperature at the tube outlet is then
checked against the actual water temperature. If its higher the guess is
reduced, if it is lower the guess is heightened. This is repeated until the
actual water temperature is reached at the outlet with a given tolerance.
Figure6.121illustratesthisiterationprocess.

52

Figure6.121Iterationprocessforwatertemperatureincountercurrentflow

Thetoleranceis setto0.5degrees.TheT should be chosenbased on the

difference between the calculated seawater temperature and the actual
seawater temperature, but its not chosen as the value of the difference
because the heat transfer is affected by the temperature difference. It is
chosenashalfthedifferencebetweenTo,outandTsw.

T = abs(

To,out Tsw
)
2

Equation6.121

SincetheHTCinparallelflowisdependantonthelocaldistancealongthe
tube, and the distance outside the tube starts at the opposite end as the
inside of the tube, the distance outside the tube must me implemented
separately. This distance starts at zero at the outlet for the process fluid,
and goes to the tube length at the inlet for the process fluid. This is
illustratedinFigure6.122.

53

Figure6.122Illustrationofdistancesinparallelcountercurrentflow

6.13 Runningtheprogram
The program is run from the mfile heatex. In heatex all necessary input
parameters are set. Geometric parameters such as tube diameter, tube
orientation,tubelength,numberoftubesandtubewallroughnessmustbe
specified.Thethermalconductivityofthetubewallmustalsobespecified.

The file where the thermodynamic and physical properties are must be
specified.TheMATLABfunctionimportfilereadsthefileandsavesthedata
into variables. Total mass flow rate, inlet temperature and inlet pressure
arealsospecified.

Inheatexthefloworientationoftheseawatermustbespecifiedalongwith
theseawatervelocityandtemperature.Ifaclosedsolutionisdesirablethe
watermassflowratemustbespecified.Ifanopensolutionisdesirablethe
mass flow should be specified as Inf (infinity). The water orientation is
specifiedinthevariablewaterorientwithdifferentintegersforthedifferent
orientations. What integer that corresponds to which water orientation is
commented in the mfile. A constant outer HTC can be specified in the
variablehokonst.

It is possible to choose between the three mentioned correlations for HTC

insidetubesduringcondensation.Thisisdoneinthevariablemod.mod=1
for HEDH, 2 for the modified Boyko and Kruzhilin and 3 for Thome. A
constant tube side HTC can also be specified. This is done by setting the
variablemodequalto4andspecifyingthedesiredinnerHTCinthevariable
hikonst. The step size (z) is also chosen in heatex. It is recommended to
choose this value small compared to the tube length. 0.01% of the tube
lengthisanokayvalue.

Thecombinedfoulingandcoatingresistanceontheoutsideofthetubesis
alsosetinheatex.

The outlet conditions, pressuredrop, duty and heat transfer area will be
displayedintheMATLABworkspacewhentheprogramhasfinished.

54

7 Descriptionoftestcases
7.1 Introduction
Inordertoensurethatthecalculationmodelworksasdesired,simulation
of two test cases have been carried out. The test cases are connected
throughahypotheticalsubseacompressionsystem.Onewellstreamcooler,
where the cooler should condense out liquid for separation and one
compressor aftercooler for pipeline transport are the basis for the test
cases.ThesystemisillustratedinFigure7.11.Thecoolingcapacityofthe
seawaterisassumedtobeinfinite,sotheseawatertemperatureisconstant.
ThemodifiedBoykoandKruzhilincorrelationisusedforliquidHTCintwo
phaseflow.

Figure7.11Hypotheticalsystemfortestcases

7.2 Wellstreamcoolercase
The well stream cooler should cool a typical well stream for separation
before compression of the gas and pumping of the liquid. The well stream
composition is shown in Table 7.21. The well stream is mostly methane,
buttherearealsosomeheavierhydrocarbons.

55

Table7.21Compositionofwellstream

TheinletconditionsareshowninTable7.22.

Table7.22Inletconditionswellstream

Two cases have been established for the well stream cooler. The outlet
conditionsforthetwocasesareshowninTable7.23.Case2isforavoiding
theformationofhydrates.

Table7.23Outletconditionsforwellstream

Figure7.21showsthephaseenvelopeforthefluidwiththeinletandoutlet
conditions. 1 denotes inlet, 2 denotes outlet case 2, and 2 denotes outlet
case 1. In both cases the outlet condition is in the twophase region. This
ensuresthatthemodelistestedinthetwophasearea.

56

Figure 7.21 Well stream phase envelope with inlet (1) and outlet conditions (case 1
(2),case2(2))

The water temperature is 5 C and the maximum seawater velocity is

assumedtobe1m/s.Horizontaltwoinchtubesareusedforbothcases,but
in case 2 a design with oneinch tubes is also carried out. Geometric tube
specificationsareshowninTable7.24.

Table7.24Geometrictubespecifications[27]

A conservative design, i.e. only natural convection, is carried out for both
cases.Incase2asensitivitystudyiscarriedout.Thesensitivityparameters
are flow rate of well stream, combined forced and free convection on the
waterside,andtubeorientation,verticalorhorizontal.

Atypicalroughnessforstainlesssteeltubesisusedtogetherwithatypical
thermalconductivity.TheseareshowninTable7.25.

Table7.25Tubewallroughnessandthermalconductivity[7][9]

7.3 Aftercoolercase
The gas flow rate from the well stream case 2 is used for the aftercooler
case.Thisisalighterfluidbecausesomeoftheheavierhydrocarbonshave
beenremoved.ThecompositionisshowninTable7.31

57

Table7.31Compositionaftercoolercase

The inlet conditions are shown in Table 7.32. The pressure and
temperatureisincreasedthroughthecompressor,butareductioninmass
flowratebecauseoftheseparationofliquid.

Table7.32Inletconditionsaftercooler

Adesignwiththesametwoinchtubesasusedinthewellstreamcases(see
Table 7.24 and Table 7.25) is carried out. The outlet temperature and
pressure drop are as in well stream case 1, 9 C and 0.5 bar. The phase
envelope with inlet (1) and outlet (2) conditions for the fluid is shown in
Figure7.31.Attheoutletconditionthereissomeliquid.

58

Figure7.31Phaseenvelopeaftercoolerfluidwithinlet(1)andoutlet(2)conditions

59

8 Resultsanddiscussion
8.1 Wellstreamcoolercase1
Thisdesignhasverylowdrivingforcesattheoutletsincethetemperature
difference is only 4 C. In addition the HTC on the outside decreases with
decreasingtemperaturedifferencebetweentubewallandtheseawater.The
combinationofthesetwoleadstohugeheattransferarearequirementsat
the cold side of the cooler. This is illustrated in Figure 8.11 for one tube.
Thisshowsthata160mlongtubeisneededforachievingthedesiredoutlet
temperatureforthisparticulartubemassflux.

Figure8.11Temperatureprofilewellstreamcase1,G=369.83kg/m2s

There is nothing to gain by increasing the mass flux in each tube, because
most of the resistance is on the outside. This is illustrated in Figure 8.12.
ThetubesideHTCandpressuredropwillincreaseifthemassfluxineach
tube is increased, but the overall HTC wont be affected very much. The
allowedpressuredropcantbeutilizedforincreasingtheheattransfer.

60

Figure8.12HTCwellstreamcase1,G=369.83kg/m2s

Based on this discussion the design is just dependant on what is the most
practical of fewer longer tubes or many shorter tubes. A design has been
madethatisacompromise.ThedesignisasdescribedinTable8.11.

Table8.11Designwellstreamcase1

The outlet temperature is not reached. The area requirements for the last
halfdegreeisverybig.Themaxallowedpressuredropisnotutilized.This
wouldhaveleadtofewerandlongertubes.

Thedesignmadehereisnotasuitabledesign.250tubesof160meterseach
arenotpractical.Constructionandtransportationofsuchacoolerwouldbe
verydifficultandveryexpensive.Aconclusionthatcanbedrawnfromthis
is that passive coolers that rely solemnly on natural convection are not
suitable if a small temperature difference at the outlet is required. The
allowed pressure drop could instead be used through a combined
expander/pumpforgeneratingmotionontheseawateraroundtheoutlet.

61

8.2 Wellstreamcoolercase2
Atthemassfluxchosenforcase1therequiredtubelengthcanbeestimated
fromFigure8.11.Atthesamemassflux250tubesofroughly50metersis
required. But in order to utilize the allowed pressure drop and for
decreasing the number of tubes the mass flow in each tube is increased.
Basedonthetrialanderrorprocedurethedesignwithtwoinchtubesare
asstatedinTable8.21.

Table8.21Designwellstreamcase22inchtubes

ThetemperatureprofilesareshowninFigure8.21.

Figure8.21Temperatureprofilewellstreamcase22inchtubes,G=577.87kg/m2s

Figure 8.22 shows the tubeside, outside and overall HTCs. The tubeside
HTC is higher compared to case 1 due to the higher mass flux, but the
overallHTCisnotaffectedbecausethewatersideHTCismuchsmallerthan
thetubeside.

62

Figure8.22HTCwellstreamcase22inchtubes,G=577.87kg/m2s

Thisdesignismorerealistic.Theareaisdecreasedwith72%comparedto
thecasewith9Cattheoutlet,butthedutyisonlydecreasedwith27%.This
illustrates how much the area requirements increase as the temperature
differencedecreases.

ThedesignwithoneinchtubesisshownTable8.22.Theallowedpressure
dropwasutilizedinthisdesignaswell.Theimportantdifferencebetween
the oneinch and twoinch design is the tube length and the number of
tubes. The total area is pretty similar but it is distributed differently. The
oneinch design allows shorter tubes, but requires 3.7 times more tubes.
Thisisatradeoffthatmustbeconsideredwhendecidingwhichtubesare
bestsuited.

Table8.22Designwellstreamcase21inchtubes

Figure8.23showsthetemperatureprofileswiththeoneinchtubes.

63

Figure8.23Temperatureprofilewellstreamcase21inchtubes,G=609.99kg/m2s

Figure 8.24 shows the HTCs. The tubeside HTC is higher for the oneinch
tubes compared to the twoinch tubes due to the higher mass flux, but as
discussedpreviouslythisdoesnotaffecttheoverallHTCconsiderably.

Figure8.24HTCwellstreamcase21inchtubes,G=609.99kg/m2s

64

8.3 Sensitivityonwellstreamcase2
8.3.1 Seawatervelocityanddirection
Asdiscussedinsection4.12theseawatervelocityinfluencesthewaterside
HTC considerably. Figure 8.31 shows the outlet temperature of the one
inch tube design in parallel and cross flow as a function of seawater
velocity.Incrossflowtheoutlettemperaturedropssignificantlyevenatlow
seawatervelocities.Butataboutvsw=0.3theprofilestraightensout.Thisis
because the constraining HTC shifts from the outside to the inside. So it is
the tube side HTC that is the constraining HTC. In parallel flow the outlet
temperature decreases more steadily. This shows that the waterside HTC
stillistheconstrainingHTCevenatsweatervelocitiesat1m/s.

Figure8.31Outlettemperaturevsseawatervelocityincrossflowandparallelflow
oneinchtubedesign

The pressure drop is not affected very much by the change in seawater
velocity. A small decrease in pressure drop is experienced because of the
factthatmoreliquidisformedwhichleadstoalowertwophasevelocity.A
similar study has been carried out on the twoinch design, and the results
areverysimilar.Therewasnoextradiametereffect.

This discussion shows that the seawater velocity influences the outlet
temperature significantly. In addition the direction of the seawater affects
theoutlettemperature.

8.3.2 Feedflowrate
Duringproductionitwillbenecessarytodecreasetheflowratebecauseof
unforeseen situations or during startup and shutdown. Its interesting to
seehowthecoolerrespondstosuchchangesinflowrate.Thefeedflowrate

65

has been varied between 50 and 100 % of design flow rate. Figure 8.32
showstheoutlettemperaturewithdecreasingmassflowratesfortheone
inch tube design. The effect from the reduction in mass flow rate on the
outlet temperature is not as strong as the case was for cross flow forced
convection. The tubeside HTC becomes lower with decreasing mass flow,
but it is still the outer HTC that is the constraining one. The reduction in
outlettemperatureisbecausethereislessmassineachtubetocooldown.

Figure8.32Outlettemperatureasafunctionofdecreaseintotalmassflowrate,one
inchtube

The reduction in mass flow rate has a very strong effect on the pressure
drop. This is illustrated in Figure 8.33. The reduction in pressure drop is
caused by the reduction in velocity. The low pressuredrops can induce
maldistribution.

Figure8.33Pressuredropasafunctionofdecreaseintotalmassflowrate,oneinch
tube

66

8.3.3 Tubeorientation
The designs so far have been with horizontal tubes. An advantage with
verticaltubesisthattheseawatervelocitywillalwaysbeperpendicularto
thetubeorientation,sothetubeswillalwaysbeincrossflowwhenthereis
motionintheseawater.Thismakesiteasiertopredicttheheattransfer.On
theotherhand,thepressuredropwillincreaseduetothechangeinstatic
height. Table 8.31 shows the outlet conditions and design if the oneinch
coolerdesignisplacedverticallyinsteadofhorizontally.

Table8.31Oneinchtubedesignwithverticaltubes

The outlet temperature is not changed significantly; this shows that the
outer HTC on vertical tubes are quite similar to horizontal tubes. The
difference is that the boundary layer increases along the tubes. The
pressuredropincreaseswith52%.Theincreaseisduetothegravitational
pressure drop. The pressure drop increases more with the twoinch tubes
becauseofthelongertubes.

8.3.4 Worstcase
The worst case in this sensitivity study is 50% flow rate and seawater in
cross flow at 1 m/s. Simulations of both the oneinch and twoinch design
showsthattheoutlettemperatureis6degreesifworstcaseisencountered.
Six degrees is only one degree above the seawater temperature. This
illustrates how the effects of varying operating conditions influence the
performanceofthecooler.

8.3.5 Summaryofsensitivityonwellstreamcoolercase2
Thedesignedheatexchangersperformancesareverysensitivetoseawater
velocity.Theoutlettemperaturecanbecomesignificantlylower,especially
in seawater cross flow. The designs are not very sensitive to tube
orientations,butifplacedverticallyonlycrossflowisexperienced,andthis
makes it easier to predict. The mass flow rate sensitivity is actually
relativelysmall,excepttheobvioussensitivitybecauseoflessmasstocool.
This is because the overall HTC is not very sensitive to mass flow rate,
becausetheresistanceisontheoutsideofthetubes.

67

8.4 Aftercooler
A similar design procedure as in the previous cases was carried out. A
designthatmeetstherequirementsisshowninTable8.41.

Table8.41Designaftercooler

Therequiredheattransferareaisverylarge.172tubesof220metersisnot
easily manufactured and transported. The reason for the huge area
requirementsis,asthecasewasforthewellstreamcoolercase1,thesmall
temperature difference at the outlet. Small driving forces and a low HTC
leads to big area requirements. The temperature profiles are shown in
Figure8.41.Atapproximatelyhalfthetubelength92%ofthetemperature
drop has been achieved. This means that 50% of the heat transfer area is
used for the last 8% of the temperature drop. At the outlet the outer wall
temperatureandthefluidtemperaturearealmostequal.

Figure8.41Temperatureprofileaftercooler,G=462.30kg/m2s

ThevariousHTCsareshowninFigure8.42.ThetubesideHTCdropsfrom
approximately1950W/m2Ktoapproximately1450W/m2Katz75m.The
dropiscausedbytransitionfromsinglephasetotwophase.Thisisa25%

68

drop, but the overall HTC only experiences a small drop. At the outlet the
overallHTCandthewatersideHTCarealmostequal.

Figure8.42HTCaftercooler,G=462.30kg/m2s

8.5 Sensitivityaftercooler
Asensitivitystudyiscarriedoutontheaftercooler.Thesameparameters
that were changed in the sensitivitystudy of well stream case 2 are
changed.
8.5.1 Seawatervelocityanddirection
In cross flow the outlet temperature reaches the seawater temperature
when there is any motion at all in the seawater. In parallel seawater flow
theoutlettemperaturedecreasesmoresteadily.ThisisillustratedinFigure
8.51.Onlysmalldeviationsinpressuredropwereexperienced.

69

Figure8.51Outlettemperatureinparallelandcrossflowforaftercooler

8.5.2 Feedflowrate
Figure8.52showstheoutlettemperatureatreducedmassflowrates.The
outlettemperatureis,asexpected,reducedasthemassflowrateisreduced.
Thelinearityoftheprofilesuggeststhatitisonlythereductioninmassthat
causes the drop in outlet temperature, and not any significant changes in
overallHTC.

Figure8.52Outlettemperatureasafunctionofreductioninmassflowrateinafter
cooler

Figure 8.53 shows the pressure drop at reduced mass flow rates. As
expectedthepressuredropdecreasesasthemassflowratesdecreasesdue
tothelowervelocityinthetubes.

70

Figure8.53Pressuredropatreductioninmassflowrateinaftercooler

8.5.3 Tubeorientation
Theoutlettemperatureincreaseswithhalfadegreeifthetubeorientation
is vertical instead of horizontal. The pressure drop increases substantially
becauseofthelongtubesandthebigheightdifference.Thepressuredrop
increases to 3.32 bar. This is a 564% increase in pressuredrop, but in
reality tubes of this length will have bends. The bends would make some
partsofthetubesverticallyorientedandsomepartshorizontallyoriented.

8.5.4 Summaryofsensitivityforaftercooler
The outlet temperature is not very sensitive to changes in operational
conditions.Thisisbecausetheoutlettemperatureisonly4degreeshigher
thantheseawatertemperatureinthedesigncase.Thepressuredropinthe
designed cooler is very sensitive to tube orientation because of the long
tubesthatcauseabigdropinstaticpressureifplacedvertically.

8.6 Summaryoftestcases
Four coolers have been designed, three designs as well stream cooler and
one design as compressor aftercooler. The area requirements for subsea
heatexchangersthatrelyonnaturalconvectionareverylarge.Especiallyin
the cases where there is a small temperature difference at the outlet. This
small temperature difference leads to low driving forced and low overall
HTC.Sensitivitystudieswerecarriedoutonthewellsteamcoolerwith25C
outlet temperature, and on the aftercooler. The well stream cooler outlet
temperature is very sensitive to external conditions, but overall HTC only
experience a small sensitivity to reduction in feed flow rate. The after
cooleroutlettemperatureisnotverysensitivetoeitherreductioninmass

71

flow rate or external conditions. Because of the long tubes the pressure
dropisverysensitivetotubeorientationduetothedropinstaticpressure.

Thedesignsmadeforthecoolerswith9Cattheoutletrequireagreatheat
transfer area. This introduces great challenges in production and
transportation.Ontheotherhand,therearebasicallynoconstraintsonthe
size with regards to space on the seabed. The coolers with the low
temperature difference at the outlet are very robust to changes in
operationalconditions,whichisanadvantage.

An important factor for the designs is the seawater velocity. The designs
madehereareconservativedesigns.Inreality,therecouldbemotioninthe
seawater continuously. If that is the case, the conservative design
overestimatestherequiredheattransferareagreatly.Infuturedesignsan
averageseawatervelocitycouldbeused.Ameasureforgeneratingmotion
intheseawateraroundthetubescouldbeusedwhennecessary.Thiswould
leadtoamuchmorecompactheatexchanger,butthetradeoffistheadded
complexitybecauseofthemeasureforgeneratingmotion.

72

9 Experiencesandpossibleimprovementofcalculation
model
Thecalculationmodelinitscurrentstateisafirstversionandthatmustbe
kept in mind. Even though the calculation model is usable now,
improvements can be made for a more efficient program and it can be
expandedtoaccountforadditionalaspects.

In general the program works as desired. Heat exchangers for subsea

applicationscanbedesignedwiththisprogram.Itseasytosetandchange
theparameters,whichmakesthefunctionalityoftheprogramgood.Studies
oninfluenceofthedifferentparametersareeasilycarriedout.

The runtime of the program can be improved. As shown in the test cases
verylongtubesmayberequired,andthisleadstolongruntime.Apossible
improvement is to change the stepsize (z) based on the temperature
gradient.Atlowtemperaturegradientsalongerstepsizecouldbeusedfor
decreasingtheruntime.

The only problem encountered when running the program is that the
function twallstart does not always find the wall temperature at the inlet.
This is dependant on the chosen step size and the tolerance level. The
problemiseasilysolvedthroughdecreasingthestepsizeorincreasingthe
tolerancelevel.

The model developed in this study is for straight parallel tubes. As

mentioned earlier, bends will make the design more compact. By
introducing bends the pressure drop increases on the tube side, and this
mustbeaccountedforbypressurelosscoefficients.Byintroducingbendsin
aclosedheatexchangerwherethewatertemperaturevariesfromlayerto
layer, the calculation procedure should be divided into segments. But the
modelinitscurrentstatecanbeusedasbasisforeachsegment,exceptthe
segmentwherethebendsare.Inaclosedarrangementbafflesmayalsobe
usedonthewatersideforleadingtheflowandinduceturbulence.Butother
commercial software exists for calculating traditional shell and tube heat
exchangers.

A natural expansion of the model is to include an equation of state.

Currently,theusermustspecifythermodynamicandphysicalpropertiesin
tables.Afutureversionoftheprogramshouldincludeanequationofstate
so that the user only specifies composition, and the properties are

73

calculatedfromtheequationofstate.Thiswouldmakehemodeleasierto
useinthesensethatpropertytablesdoesnottohavetobeproduced.

Another measure for making the program userfriendlier is a graphical

interface for the user. An interface where the user can specify geometric
specificationsandotherinletconditionsisdesirable.An.exefileshouldbe
made for running of the program. This would give the program a more
commerciallook.

In commercial software for designing heat exchangers the program

automatically iterates until the specifications are reached. For example,
allowed pressure drop, size and outlet temperature are specified and the
program iterates until the specifications are reached. Since this is the first
versionoftheprogramallproblems/flawsmaynothavebeenencountered
orfoundyet.Therefore,suchaniterationloophasnotbeenincluded.When
it is established that the program functions as desired this should be
implemented.

74

10 Conclusion
A calculation model for a subsea heat exchanger has been developed. The
model is implemented in the commercial software MATLAB. The model is
based on straight tubes that can be horizontally or vertically oriented.
Differentorientationsofthesurroundingseawaterareimplementedinthe
model.

Heat transfer and frictional pressure drop correlations have been studied,
and based on this study the modified Boyko and Kruzhilin correlation is
recommended for heat transfer coefficient in condensing twophase flow
inside tubes and the corrected Fuchs correlation is recommended for
calculationoffrictionalpressuredropintwophaseflowinsidetubes.

Themodelhasbeentestedontwotestcases:onetestcaseforsimulatinga
well stream cooler, and one for simulation of an aftercooler in a subsea
compression system. Experiences from simulation of two test cases show
that the model is suited for design, and parameter sensitivity studies, of
subseaheatexchangers.

75

11 Bibliography
[1] T.BjrgeandL.Brenne,Subseagascompression;Availabletechnology
andfutureneeds,February2008.
[2] (2006,March)WikimediaCommons.[Online].
http://commons.wikimedia.org/wiki/File:Straight
tube_heat_exchanger_1pass.PNG
[3] B.VanDerRest,"SubseaHeatExchanger,"PCT/NO2009/000248,July
3,2008.
[4] V.StenHalvorsen,E.Baggerud,andT.Hollingster,"SubseaCooler,"
PCT/NO2008/000196,Juni01,2007.
[5] J.ElHajal,J.R.Thome,andA.Cavallini,"Condensationinhorizontal
tubes,part1:twophaseflowpatternmap,"Int.J.HeatandMass
transfer,vol.46,pp.33493363,2003.
[6] N.Kattan,J.R.Thome,andD.Favrat,"Flowboilinginhorizontaltubes:
Part1Developmentofadiabatictwophaseflowpatternmap,"Trans.
ASME,vol.120,pp.140147,1998.
[7] F.M.White,FluidMechanics,6thed.:McGrawHill,2008.
[8] S.E.Haaland,"Simpleandexplicitformulasforthefrictionfactorin
turbulentpipeflow,"J.Fluids.Eng,pp.8990,1983.
[9] Y.A.Cengel,HeatandmasstransferApracticalapproach,3rded.,Y.A.
Cengel,Ed.:McGrawHill,2006.
[10] V.Gnielinski,"Newequationsforheatandmasstransferinturbulent
pipeandchannelflow,"Int.Chem.Eng.,vol.16,pp.359368,1976.
[11] B.S.Pethukov,"Advancesinheattransfer,"vol.6,1970.
[12] G.F.Hewitt,HeatExchangerDesignHandbook.:Hemispherepublishing
corporation,1983.
[13] R.H.Norris,"Augmentationofconvectionheatandmasstransfer,"
Americansocietyofmechanicalengineers,1971.
[14] W.Nusselt,"DieOberachenkondensationdesWasserdampfes,"
Zeitschr.Ver.Deutch.Ing,vol.60,pp.541569,1916.
[15] B.O.Neeraas,CondensationofhydrocarbonmixturesincoilwoundLNG
heatexchangers.:Departmentofrefrigerationengineering(NTNU),
1993,Doctorthesis.
[16] J.R.Thome,J.E.Hajal,andA.Cavallini,"Condensationinhorizontal
tubes,part2:newheattransfermodelsbasedonflowregimes,"Int.J.

76

heatandmasstransfer,vol.46,pp.33653387,2003.
[17] B.C.PriceandJ.K.Bell,"Designofbinaryvaporcondensersusingthe
ColburnDrewanalogy,"AIChEsymp.series,vol.70,no.138,pp.163
171,1974.
[18] D.DelCol,A.Cavallini,andJ.R.Thome,"Condensationofzeotropic
mixturesinhorizontaltubes:NewSimplifiedHeatTransferModel
BasedonFlowRegimes,"J.HeatTransfer,vol.127,pp.221230,2005.
[19] P.H.Fuchs,"Pressuredropandheattransferduringflowof
evaporatingliquidinhorizontaltubesandbends,"PhDthesis,1975.
[20] G.Hetsroni,G.F.Hewitt,J.M.Delhaye,andA.Lieberman,
"Measurementtechniques,"inHandbookofmultiphasesystems,Gad
Hetsroni,Ed.:McGrawHillBookCompany,1982,pp.10110180.
[21] R.W.LockhartandR.C.Martinelli,"Proposedcorrelationofdatafor
isothermaltwophase,twocomponentflowinpipes,"Chem.Eng.Prog,
vol.45,pp.3948,1949.
[22] D.Chisholm,"Pressuregradientsduetofrictionduringtheflowof
evaporatingtwophasemixturesinsmoothtubesandchannels,"Int.J.
Heatmasstransfer,vol.16,pp.347348,1973.
[23] L.Friedel,"Improvedfrictionpressuredropcorrelationforhorizontal
andverticaltwophasepipeflow,"Europeamtwophaseflowgroup
meet.,Ispra,Italy,p.paperE2,1979.
[24] V.K.Patnana,R.P.Bharti,andR.P.Chhabra,"Twodimensinal
unsteadyflowofpowerlawfluidsoveracylinder,"Chem.Engin.
Science,vol.64,pp.29782999,2009.
[25] Various,CompendiuminTEP07Industrialheatengineering.:
Departmentofenergiandprocesstechnology(NTNU),2009.
[26] J.B.Aarseth,Numericalcalculationmethods(Innorwegian).:
DepartmentofconstructionNTNU,2008.
[27] NorwegianSteelAssociation.Norsok.[Online].
http://www.standard.no/norsok/L001/pclass/FD30.htm
[28] D.Butterworth,"Simplifiedmethodsforcondensationonavertical
surfacewithvaporshear,"UKAEA,1981.
[29] E.P.Anandiev,L.D.Boyko,andG.N.Kruzhilin,"Heattransferinthe
presenceofsteamcondensationinahorizontaltube,"Int.Heattransfer
conf.,vol.2,pp.290295,1961.
[30] B.O.Neeraas,Personalcommunication,2010.

77

Appendices

78

A. Heattransfercorrelations
TheHeatExchangerdesignhandbookcorrelation:
This correlation is based on annular flow, or shear dominated flow. HEDH
[12]suggestacorrelationdevelopedby Butterworth[28]. A critical liquid
Reynoldsnumberisdefinedbasedontheinterfacialshearforce.Theliquid
Reynoldsnumberisdefinedas:

Rel =

GD(1 x)

EquationA1

Thedimensionlessshearforceisdefinedas:

I+ =

l I

( l g)2 / 3
2
l

EquationA2

For+<9.04thecriticalReynoldsnumberisgivenas:
Re c = 1600 226 I+ + 0.667( I+ )3
EquationA3

andfor+>9.04,thecriticalReynoldsnumberis50.IftheReynoldsnumber
isbelowthecriticalReynoldsnumbertheHTCisgivenas:
+ = 1.41Rel1/2 ( I+ )1/2
EquationA4

andforReynoldsnumbersabovethecriticalReynoldsnumbertheHTCis:
1/ m
+
= (1.41Rel1/2 )m + (0.071Pr1/2 Re 1/24 )m
+ 1/2
( I )

EquationA5

Wherem=0.5(Pr+3).ThedimensionlessHTCisdefinedasfollows

+ =

l2 1/ 3
( )
l g

EquationA6

Thismodelneedscalculationoftheinterfacialshearforce.Thesimplestway
ofcalculatingitistotake[12]:

I =

D dp f
(
)
4
dz

EquationA7

79

ThemodifiedBoykoandKruzhilincorrelation:
Neeraas [15] made a modification to the original Boyko and Kruzhilin
correlation [29]. Boyko and Kruzhilins correlation has no empirically
determinedconstantsandisthereforeinteresting.Theoriginalcorrelation
isgivenas:

= lo (1 + x(

l
1))0.5
g

EquationA8

Neeraasmodifieditbasedonhisexperimentswithpropane.Heproposeda
correctionfactor:
k = (1.15 0.275x)1
EquationA9

SothatthenewexpressionforHTCis

= k lo (1 + x(

l
1))0.5
g

EquationA10

loiscalculatedasiftheliquidoccupiesallofthetubeandisreferredtoas
liquidonly.TheliquidonlyReynoldsnumberisdefinedas:

Relo =

GD

EquationA11

Equation 4.44 or Equation 4.45 can then be used for calculating lo. The
mass flux in the experiments with propane ranges from 150350 kg/m2s,
pressuresbetween12and20bar,andvaporfractionsbetween0.150.88.

Thomescorrelation:
Thomeetal[16]developedanewcorrelationbasedonflowregimes.Their
objective was to develop a new correlation with a minimum of
experimentally determined constants. The correlation was compared to
data points from 15 fluids, including nbutane, isobutane, propylene and
propane, which are hydrocarbons. The other fluids are the single
component refrigerants R11, R12, R22, R32, R113, R125, R134a,
R236ea, and the binary azeotropic mixtures R32/R125 (60%/40%),
R404A and R410A. The range of mass fluxes is 241022 kg/m2s, vapor
fractions in the range from 0.03 to 0.97, reduced pressures in the range
from 0.02 to 0.80. 921 of 2771 data points in total are hydrocarbons data
points.

80

Since it is based on flow regimes the expression for HTC differs from flow
regimetoflowregime.Hereonlythemodelfortheannularflowregimewill
bepresented.TheexpressionforHTCforannularflowis

= 0.003Rel0.74 Prl0.5
EquationA12

l
fi

Where is the film thickness and fi is an interfacial roughness correction

factor.TheresemblancetotheDittusBoelterequationisobviousbutwith
thediameterreplacedwiththefilmthicknessandacorrectionfactorforthe
interfacial roughness. The constants in Equation A 12 are determined
empiricallyfromtheHTCdatabase.

In order to use Equation A 12 the film thickness needs to be calculated,

whichagainrequirescalculationofvoidfraction.Voidfractionisdefinedas

Ag
A

A Al

EquationA13

FilmthicknessiscalculatedfromgeometryfromEquationA14

D
D2

2
2

EquationA14

Thome et al [16] used a void fraction model that is a logarithmic mean

between the homogenous model (phases move with the same velocity,
EquationA15)andthemodelfromRouhaniandAxelsson,whichisanon
homogenous model [16] (Equation A 16). The Rouhani and Axelsson void
fractionequationisbasedonthedriftfluxmodel.
1

1 x g
h = 1 + (
)( )
x
l

EquationA15
1

0.25

x 1 x 1.18(1 x) g ( l g
x

ra = [1 + 0.12(1 x)] +
+
0.5
g

v
l
l

EquationA16

Thelogarithmicmeangivesthefollowingexpressionforthevoidfraction:

81

h ra

h
ln
ra

EquationA17

The interfacial roughness correction factor proposed by Thome et al is

calculatedfromEquationA18.

vg
fi = 1 +
v
l

0.5

1/ 4

( l g )g 2

EquationA18

Thevaporandliquidvelocitiesarecalculatedas:

vl =

G(1 x)
Gx
, vg =

l (1 )
v

EquationA19

TheliquidReynoldsnumberinEquationA12iscalculatedfromEquationA
20.

Rel =

4G(1 x)

(1 )l

EquationA20

82

B. Pressuredropcorrelations
All the correlations use singlephase expressions for frictional pressure
drop,thereforesomenecessaryquantitieswillbedefinedfirst.Thesingle
phase friction factors can be calculated from equations in section 4.3 with
the respective Reynolds number. Rel is calculated from Equation A 1, Reg
fromEquationB1,RelofromEquationA11,andRegofromEquationB2.

Re g =

GxD

EquationB1

Re go =

GD

EquationB2

The singlephase frictional pressure gradients are calculated through the

followingequations[20].Forsinglephasegas:

fg G 2 x 2
dp fric
dz = 2D
g
g
EquationB3

Forsinglephaseliquid:

dp fric
fl G 2 (1 x)2

dz =
2Dl
l
EquationB4

Forliquidonly:

dp fric
floG 2
dz = 2D
l
lo
EquationB5

Forgasonly:

fgoG 2
dp fric
dz = 2D
g
go
EquationB6

LockhartandMartinelliscorrelation:
This is the most used correlation historically [20], but its accuracy is
limited.Itisbasedonatwophasepressuredropmultiplier.Thetwophase
pressure drop can either be expressed by the singlephase liquid pressure
gradientorthesinglephasegaspressuregradient:

83

dp fric
dp fric
= g2
= l2

dz g
dz l
dz

dp fric

EquationB7

ThepressuredropmultipliersisafunctionoftheparameterX2definedas:

dp fric
dz
2
l
X =

dp fric
dz
g
EquationB8

Thetwophasemultiplierscanthenbefoundbythefollowingequations:

l2 = 1 +

C
1
+ 2
X X

EquationB9

g2 = 1 + CX + X 2
EquationB10

WhereCisanempiricalconstantthatdependsoniftheflowislaminaror
turbulent. TableB 1showsthevaluesofC dependingon if thegasand/or
liquidareturbulentorlaminar[20].

TableB1ValuesofCinLockhartandMartinellifrictionalpressuredropcorrelation

Chisholmscorrelation:
Chisholms correlation is based on a liquid only twophase friction
multiplier.

dp fric
= lo2

dz lo
dz

dp fric

EquationB11

Where2loisgivenbythefollowingexpression[20]:

84

2n
(
)
( 2n )

lo2 = 1 + (Y 2 1) Bx 2 (1 x) 2 + x 2 n

EquationB12

nisthepowerofthesinglephasefrictionalfactorrelationship(see4.3).B
isgiveninTableB2

TableB2ValuesofBinChisholmfrictionalpressuredropcorrelation

WhereYis

Y2 =

dp fric
dz
go
dp fric
dz
lo

EquationB13

Friedelscorrelation:
Friedels correlation is also based on the liquid only twophase friction
multiplier,butthefrictionmultiplieriscorrelatedasfollows[20]:

lo2 = E +

3.24FH

Fr 0.045We0.035

EquationB14

Where

E = (1 x)2 +

x 2 l fgo

g flo

EquationB15

F = x 0.78 (1 x)0.24
EquationB16

85


H = l
g

0.91

0.19

1
l

0.7

EquationB17

TheFroudenumberisdefinedas:

G2
Fr =

gDtp2
EquationB18

TheWebernumberisdefinedas:

G2D
We =

tp
EquationB19

Wherethetwophasedensityisdefinedhereas:
1

x 1 x

tp = +
g
g
EquationB20

Fuchscorrelation:
Fuchscorrelation[19]isbasedonexperimentswithR12andconsistsofa
twophaseenhancementfactor[15].Thetwophaseenhancementfactoris
definedas:

dp dp

dz fric dz fric
lo
dp
dp
dz dz
fric go fric lo

EquationB21

Thetwophaseenhancementfactorisgenerallyafunctionofvaporfraction,
Froudenumber and density ratio between the liquid and the gas, but for
Froudenumbersabove1.45itisonlyafunctionofvaporfraction.Theflow
is then expected to be annular [15]. The equation for the twophase
enhancementfactoris:

86

1
= 6740.331720 * x11
36759.087741* x10 + 85275.119778 * x 9

110168.145383 * x 8 + 87170.939162 * x 7 43797.819250 * x 6

+14021.596088 * x 5 2790.120307 * x 4 + 324.432076 * x 3

18.611125 * x 2 + 2.414768 * x 0.000141

EquationB22

The equation is not retrieved directly from Fuchs [19] because it was not
possibletoobtain,butitisretrievedthroughpersonalcommunicationwith
BengtO.Neeraas[30].

Thedatafortheexperimentsare[15]:
Pressure:1.22.6bar
Massflux:65775kg/m2s
Vaporfraction:01

CorrectedFuchs:
Neeraas[15]correctedthecorrelationbyFuchsfromhisexperimentswith
propane, propane/methane and propane/ethane. The mass fluxes were in
therangeof150400kg/m2s,pressuresbetween12and40bar,andvapor
fractionsbetween0and1.
fuchs

dp
dp
=
dz fric
dz fric

f l , x
g

EquationB23

Wherethecorrectionfunctionfis

f l , x = 1.0 1.0 0.3g l h(x)
g
g

EquationB24

Thedensitycorrectionisdefinedthroughthethreefollowingequations:

1.0 l
g l =
, > 20
g 0.3 g
EquationB25

g l = l
g g

0.4

, 6.5

l
20
g

EquationB26

87

g l = 6.5 0.4 , l < 6.5

g
g
EquationB27

Andthevaporfractioncorrectiondefinedbythefollowingtwoequations:

h(x) =

sin( x) + sin( x 3 )
, x 0.725
1.69

EquationB28

x 0.725 5 x
h(x) = 1.0 sin
, x > 0.725

2 0.275

7/2

EquationB29

88

C. Testcaseusedincomparisons
A typical well stream is used for comparing HTC and pressure drop
correlations.TableC1showstubeandflowspecifications.

TableC1Tubeandflowspecificationsfortestcase

ThecompositionofthefluidisshowninTableC2andthephaseenvelope
inFigureC1.

TableC2Testfluidcomposition

FigureC1Phaseenvelopetestfluid

89

D. MATLABfunctionsandscripts
Table D 1 shows an overview with description of the different scripts and
functions in the calculation model. The scripts for the builtin MATLAB
functionsarenotshownhere,butthescriptsfortheimplementedroutines
areincludedafterthetable.

Function
Description
Built-in MATLAB functions
Returns minimum and position of minimum
min(vector)
in vector
Returns maximum and position of maximum
max(vector)
in vector
interp1
Interpolates in specified tables
Implemented routines
twallstart
Returns wall temperature at inlet
Returns tubeisde HTC in single-phase flow
(Gnielinski)
hindre
Return the implicit function for friction
factor in smooth tubes (Equation4.35)
fricsmooth
Returns HTC in natural convection around
hytre
horizontal cylinder
Returns HTC in forced cross flow over
cylinder
hytref
Returns combined natural and forced
hytrekomb
convection HTC
Returns
single-phase
friction
factor
trykkfall
(Haaland)
Reads to specified file with properties.
Importfile is generated by MATLAB
importfile
Return single-phase HTC for smooth tubes
hindreglatt
(Dittus-Boelter)
hliquid
Returns liquid HTC from HEDH correlation
Returns liquid HTC from modified Boyko and
Kruzhilin correlation
hliquidboy
Returns HTC in natural convection around
hytrevert
vertical tubes
Return HTC in parallel flow over horizontal
hytrecocurr
tubes
Returns void fraction from (Equation A 15EquationA17)
lmevoid
Returns frictional pressure gradient in twofuchscorrected
phase flow (corrected Fuchs correlation)
fuchs
Returns frictional pressure gradient in two-

90

euler1naturlig
euler1cocurrent
euler1crossmixed
euler1crossunmixed
euler1countercurrent

heatex

phase flow from original Fuchs correlation

Solves temperature and pressure profiles in
pure natural convection
Solves temperature and pressure profiles in
cocurrent parallell flow
Solves temperature and pressure profiles in
mixed cross flow
Solves temperature and pressure profiles in
unmixed cross flow
Solves temperature and pressure profiles in
countercurrent parallell flow
m-file where all inlet parameters, tube
geometry and water- and tube orientations
is chosen. File where properties are located
is specified in heatex

TableD1MATLABroutines

twallstart
functiontw1=twallstart(Ti,hi,vsw,teta,l,Tsw,waterorient)
globalDoDikdlhokonstRf
a=TiTsw;
ifhokonst==0
ho=hytrekomb(a,vsw,teta,l,Tsw,waterorient);
else
ho=hokonst;
end
htot=1/((1/ho)+Rf+(Do/(hi*Di))+(log(Do/Di)*Do/(2*k)));
hiw=1/((Do/(hi*Di))+Rf+(log(Do/Di)*Do/(2*k)));
sjekk=pi*Do*dl*(htot*(TiTsw)hiw*(Tia));
dt=0.1;
eps=.5;
while((abs(sjekk)>eps)&&(a>Tsw))
a=adt;
ifhokonst==0
ho=hytrekomb(a,vsw,teta,l,Tsw,waterorient);
else
ho=hokonst;
end
htot=1/((1/ho)+Rf+(Do/(hi*Di))+(log(Do/Di)*Do/(2*k)));
hiw=1/((Do/(hi*Di))+Rf+(log(Do/Di)*Do/(2*k)));
sjekk=Do*pi*dl*(htot*(TiTsw)hiw*(Tia));
end
tw1=a;

91

hindre
functionhi=hindre(Re,Pr,Di,l,lambda,fg)
ifRe>3000
f=fzero(@(f)fricsmooth(f,Re),0.04);
else
f=64/Re;
end
if(Re>2300)&&(Re<5e4)
Nusmooth=(f/8)*Re*Pr/(1+12.7*(f/8)^(1/2)*(Pr^(2/3)1));

elseifRe>5e4
Nusmooth=(f/8)*Re*Pr/(1+12.7*(f/8)^(1/2)*(Pr^(2/3)1));
else
Nusmooth=3.66;
end
n=0.68*Pr^0.215;
surfcorr=(fg/f)^n;
Nu=Nusmooth*surfcorr;
hi=Nu*lambda/Di;

fricsmooth
functiong=fricsmooth(f,Re)
g=2.0*log10(Re*f^(1/2))0.8(1/f^(1/2));

hytre
functionho=hytre(Tw,Tsw)
globalDog
T=5:5:60;
beta=1e3*[0.0150.07330.1380.1950.2470.2940.3370.3770.4150.451
0.4840.517];
Prandtl=[11.29.458.097.016.145.424.834.323.913.553.252.99];
my=1e3*[1.519 1.307 1.138 1.002 0.891 0.798 0.720 0.653 0.596 0.547
0.5040.467];
ro=[999.9999.9999.1998997996994992.1990.1988.1985.1983.3];
kond=[0.5710.5800.5890.5980.6070.6150.6230.6310.6370.6440.649
0.654];
Tf=(Tw+Tsw)/2;
b=interp1(T,beta,Tf);
Pr=interp1(T,Prandtl,Tf);
u=interp1(T,my,Tf);
p=interp1(T,ro,Tf);
v=u/p;
k=interp1(T,kond,Tf);

92

Gr=g*b*(TwTsw)*Do^3/v^2;
Ra=Pr*g*b*(TwTsw)*Do^3/v^2;
Nuf1=(0.518*Ra^(1/4)/(1+(0.559/Pr)^(9/16))^(4/9));
Nuf2=(1+((3.47*1e7*Ra)/(1+(0.559/Pr)^(9/16))^(16/9)))^(1/12);
Nuf=Nuf1*Nuf2;
temp=1+2/Nuf;
Nu=2/log(temp);
ho=Nu*k/Do;

hytref
functionhof=hytref(vsw,Tw,Tsw)
globalDo
T=5:5:60;
Prandtl=[11.29.458.097.016.145.424.834.323.913.553.252.99];
my=1e3*[1.519 1.307 1.138 1.002 0.891 0.798 0.720 0.653 0.596 0.547
0.5040.467];
ro=[999.9999.9999.1998997996994992.1990.1988.1985.1983.3];
kond=[0.5710.5800.5890.5980.6070.6150.6230.6310.6370.6440.649
0.654];
Tf=(Tw+Tsw)/2;
Pr=interp1(T,Prandtl,Tf);
u=interp1(T,my,Tf);
p=interp1(T,ro,Tf);
k=interp1(T,kond,Tf);
Re=p*vsw*Do/u;
Nu=
0.3
+
(0.62*Re^(1/2)*Pr^(1/3)*(1+(Re/282000)^(5/8))^(4/5))/(1+(0.4/(Pr^(2
/3))))^(4/5);
hof=Nu*k/Do;

hytrekomb
functionhokomb=hytrekomb(Tw,vsw,teta,l,Twater,waterorient)
ifvsw==0&&teta==0
hokomb=hytre(Tw,Twater);
return
elseifvsw==0&&teta==90
hokomb=hytrevert(Tw,l);
return
elseifvsw~=0&&(waterorient==2||waterorient==3)
hf=hytref(vsw,Tw,Twater);
ifteta==0
hn=hytre(Tw,Twater);

93

else
hn=hytrevert(Tw,l);
end
hokomb=(hn^4+hf^4)^(1/4);
elseifvsw~=0&&(waterorient==4||waterorient==5)
hn=hytre(Tw,Twater);
hf=hytrecocurr(vsw,Tw,Twater,l);
hokomb=(hn^4+hf^4)^(1/4);
end

trykkfall
functionf=trykkfall(Re,k,Di)
ifRe<3000
f=64/Re;
return
else
f=(1.8*log10((6.9/Re)+(k/(3.7*Di))^1.11))^(2);
end

importfile(GeneratedbyMATLAB)
functionimportfile(fileToRead1)
newData1=importdata(fileToRead1);
vars=fieldnames(newData1);
fori=1:length(vars)
assignin('base',vars{i},newData1.(vars{i}));
end

hindreglatt
functionhi=hindreglatt(Re,Pr,lambda,l)
Nu=0.023*Re^(0.8)*Pr^(0.4);
hi=Nu*lambda/l;

hliquid
functionhl=hliquid(Re,rol,ul,lambdal,cpl,taul)
globalg
taupluss=rol*taul/((rol^2*ul*g)^(2/3));
Pr=ul*cpl/lambdal;
m=0.5*(Pr+3);
ny=ul/rol;
iftaupluss<=9.04
Rec=1600226*taupluss+0.667*(taupluss^3);
else
Rec=50;

94

end
ifRe>Rec
apluss=((1.41/(Re^0.5))^m
(0.071*(Pr^0.5)/(Re^(1/24)))^m)^(1/m)*taupluss^(1/2);
else
apluss=1.41*Re^(0.5)*taupluss^(0.5);
end
hl=apluss*lambdal*g^(1/3)/(ny^(2/3));

hliquidboy
functionhl=hliquidboy(massfrac,Relo,lambdal,Pr,rol,rog)
globalDiks
f=trykkfall(Relo,ks,Di);
hlo=hindre(Relo,Pr,Di,0,lambdal,f);
hl=(1.150.275*massfrac)^(1)*hlo*(1+massfrac*((rol/rog)1))^(0.5);

hytrevert
functionho=hytrevert(Tw,l)
globalTswg
T=5:5:60;
beta=1e3*[0.0150.07330.1380.1950.2470.2940.3370.3770.4150.451
0.4840.517];
Prandtl=[11.29.458.097.016.145.424.834.323.913.553.252.99];
my=1e3*[1.519 1.307 1.138 1.002 0.891 0.798 0.720 0.653 0.596 0.547
0.5040.467];
ro=[999.9999.9999.1998997996994992.1990.1988.1985.1983.3];
kond=[0.5710.5800.5890.5980.6070.6150.6230.6310.6370.6440.649
0.654];
Tf=(Tw+Tsw)/2;
b=interp1(T,beta,Tf);
Pr=interp1(T,Prandtl,Tf);
u=interp1(T,my,Tf);
p=interp1(T,ro,Tf);
v=u/p;
k=interp1(T,kond,Tf);
Ra=Pr*g*b*(TwTsw)*l^3/v^2;
w=(1+(0.492/Pr)^(9/16))^(16/9);
ifRa<1e9
Nu=0.503*(Ra*w)^(1/4);
else
Nu=89.4*Pr^(1/4)*w^(1/4)+0.15*(Ra^(1/3)1000*Pr^(1/3))*w^(1/3);
end
ho=Nu*k/l;

95


hytrecocurr
functionho=hytrecocurr(vsw,Tw,Tsw,x)
T=5:5:60;
Prandtl=[11.29.458.097.016.145.424.834.323.913.553.252.99];
my=1e3*[1.519 1.307 1.138 1.002 0.891 0.798 0.720 0.653 0.596 0.547
0.5040.467];
ro=[999.9999.9999.1998997996994992.1990.1988.1985.1983.3];
kond=[0.5710.5800.5890.5980.6070.6150.6230.6310.6370.6440.649
0.654];
Tf=(Tw+Tsw)/2;
Pr=interp1(T,Prandtl,Tf);
u=interp1(T,my,Tf);
p=interp1(T,ro,Tf);
k=interp1(T,kond,Tf);
ifx==0
x=0.02;
end
Re=p*vsw*x/u;
Nu=0.0296*Re^(0.8)*Pr^(1/3);
ho=Nu*k/x;

lmevoid
functione=lmevoid(x,rog,rol,G,sigma)
globalg
era=(x/rog)*((1+0.12*(1x))*((x/rog)+((1x)/rol))+(1.18*(1
x)*(g*sigma*(rolrog)^0.25/(G*rol^0.5))))^(1);
eh=(1+((1x)/x)*(rog/rol))^(1);
e=(ehera)/(log(eh/era));

fuchscorrected
functiondpdz=fuchscorrected(x,G,Relo,Rego,rol,rog)
dpfuchs=fuchs(x,G,Relo,Rego,rol,rog);
densrat=rol/rog;
ifdensrat>20
g=1.0/0.3;
elseif(densrat>=6.5)&&(densrat<20)
g=densrat^(0.4);
else
g=6.5^(0.4);
end
ifx<=0.725
h=(sin(pi*x)+sin(pi*x^3))/1.69;

96

else
h=1.0sin((pi/2)*((x0.725)/0.275)^(5*x^(7/2)));
end
corr=1.0(1.00.3*g)*h;
dpdz=dpfuchs*corr;

fuchs
functiondpfdl=fuchs(x,G,Relo,Rego,rol,rog)
globalDiks
flo=trykkfall(Relo,ks,Di);
fgo=trykkfall(Rego,ks,Di);
dpldl=flo*G^2/(2*Di*rol);
dpgdl=fgo*G^2/(2*Di*rog);
fuch=6740.331720*x^1136759.087741*x^10+85275.119778*x^9
110168.145383*x^8+87170.939162*x^7
43797.819250*x^6+14021.596088*x^5
2790.120307*x^4+324.432076*x^318.611125*x^2+2.414768*x
0.000141;
dpfdl=fuch*(dpgdldpldl)+dpldl;

euler1naturlig
global Do Di k ks vsw mtot Tstart Pinn data l mod textdata xtubes
waterorientmswTswshikonsthokonstcpswRfgdl
fori=1:length(textdata)
ifstrcmp(textdata(i),'Enth(kJ/kg)')==1
entrow=i;
elseifstrcmp(textdata(i),'Gasmolfrac(mol%)')==1
molfracrow=i;
elseifstrcmp(textdata(i),'Gasmassfrac(mass%)')==1
massfracrow=i;
elseifstrcmp(textdata(i),'Totaldensity(kg/m3)')==1
totdensrow=i;
elseifstrcmp(textdata(i),'Gasdensity(kg/m3)')==1
gasdensrow=i;
elseifstrcmp(textdata(i),'Liquiddensity(kg/m3)')==1
liqdensrow=i;
elseifstrcmp(textdata(i),'Gasviscosity(cP)')==1
gasviscrow=i;
elseifstrcmp(textdata(i),'Liquidviscosity(cP)')==1
liqviscrow=i;
elseifstrcmp(textdata(i),'GasCP(kj/kgK)')==1
gascprow=i;
elseifstrcmp(textdata(i),'LiquidCP(kj/kgK)')==1

97

liqcprow=i;
elseifstrcmp(textdata(i),'Gasthermalconduc(mW/mC)')==1
gascondrow=i;
elseifstrcmp(textdata(i),'Liquidthermalconduc(mW/mC)')==1
liqcondrow=i;
elseifstrcmp(textdata(i),'Oil/gasinterfacialtension(mN/m)')==1
surftensrow=i;
end
end
g=9.81;
n=round(l/dl);
dA=pi*Do*l/n;
T=zeros(xtubes,n);
Tw=zeros(xtubes,n);
Twater=zeros(1,xtubes);
foryt=1:xtubes
Twater(yt)=Tsw;
end
m=zeros(1,xtubes);
dampfrac=zeros(1,xtubes);
P=T;
htottab=T;
hitab=T;
hwtab=T;
hiwtab=T;
hutside=T;
friksjon=T;
Reynold=T;
x=T;
Prandtl=T;
trykktap=m;
dp=0;
dm=0;
forror=1:xtubes
m(ror)=mtot/xtubes;
end
[dpmin,jmin]=min(trykktap);
[dpmax,jmax]=max(trykktap);
while(deltadp>1e2*dpmax)||(trykktap(1)==0)
if(trykktap(1))~=0
dm=(deltadp/2)^(1/2)*Di^2/l;
m(jmax)=m(jmax)dm;
m(jmin)=m(jmin)+dm;

98

end
Qtot=0;
forj=1:xtubes
G=m(j)*4/(pi*Di^2);
T(j,1)=Tstart;
ro=interp1(data(1,:),data(gasdensrow,:),T(j,1),'spline');
u=(interp1(data(1,:),data(gasviscrow,:),T(j,1),'spline'))/1000;
cp=(interp1(data(1,:),data(gascprow,:),T(j,1),'spline'))*1000;
lambda=(interp1(data(1,:),data(gascondrow,:),T(j,1),'spline'))/1000;
Pr=u*cp/lambda;
molfrac=(interp1(data(1,:),data(molfracrow,:),T(j,1)))/100;
massfraci=(interp1(data(1,:),data(massfracrow,:),T(j,1)))/100;
ul=(interp1(data(1,:),data(liqviscrow,:),T(j,1),'spline'))/1000;
rol=interp1(data(1,:),data(liqdensrow,:),T(j,1),'spline');
lambdal=(interp1(data(1,:),data(liqcondrow,:),T(j,1),'spline'))/1000;
cpl=(interp1(data(1,:),data(liqcprow,:),T(j,1),'spline'))*1000;
sigma=(interp1(data(1,:),data(surftensrow,:),T(j,1),'spline'))/1000;
Prl=ul*cpl/lambdal;
ifisnan(massfraci)==1
massfraci=1;
elseifmassfraci>0.99&&massfraci<1
massfraci=.99;
end
ifmassfraci==1
v=m(j)/(ro*(pi*Di^2/4));
Re=ro*v*Di/u;
f=trykkfall(Re,ks,Di);
ifmod~=4
hi=hindre(Re,Pr,Di,l,lambda,f);
elseifmod==4
hi=hikonst;
end
elseif(massfraci<=1)&&(massfraci>=0)
entalp1=(interp1(data(1,:),data(entrow,:),(T(j,1)+.5),'spline'))*1000;
entalp2=(interp1(data(1,:),data(entrow,:),T(j,1),'spline'))*1000;
dtdh=((T(j,1)+.5)Tstart)/(entalp1entalp2);
zg=massfraci*cp*dtdh;
mliq=abs(m(j)*(1massfraci));
e1=lmevoid(massfraci,ro,rol,G,sigma);
delta=Di/2(e1*Di^2)^(1/2)/2;
Rel=4*G*(1massfraci)*delta/((1e1)*ul);
Relo=G*Di/ul;
Rego=G*Di/u;

99

dpfdl=fuchscorrected(massfraci,G,Relo,Rego,rol,ro);
vgs=G*massfraci/ro;
Regs=ro*vgs*Di/u;
vg=G*massfraci/(ro*e1);
Re=ro*vg*Di/u;
vl=G*(1massfraci)/(rol*(1e1));
fgs=trykkfall(Regs,ks,Di);
ifmod~=4
ifmod==2
hliq=hliquidboy(massfraci,Relo,lambdal,Prl,rol,ro);
dpfgsdl=trykkfall(Regs,ks,Di)*ro*vgs^2/(2*Di);
cf=(dpfdl/dpfgsdl)^0.445;
hg=cf*hindreglatt(Re,Pr,lambda,Di);
elseifmod==1
taui=dpfdl*Di/4;
hliq=hliquid(Rel,rol,ul,lambdal,cpl,taui);
dpfgsdl=trykkfall(Regs,ks,Di)*ro*vgs^2/(2*Di);
cf=(dpfdl/dpfgsdl)^0.445;
hg=cf*hindreglatt(Re,Pr,lambda,Di);
else
fi=1+(vg/vl)^(0.5)*((rolro)*g*delta^2/sigma)^(1/4);
hliq=0.003*Rel^0.74*Prl^0.5*lambdal*fi/delta;
hg=fi*hindreglatt(Re,Pr,lambda,Di);
end
hi=1/((1/hliq)+(zg/hg));
elseifmod==4
hi=hikonst;
end
elseifmassfraci==0
v=m(j)/(rol*(pi*Di^2/4));
Rel=rol*v*Di/ul;
f=trykkfall(Rel,ks,Di);
ifmod~=4
hi=hindre(Rel,Prl,Di,l,lambdal,f);
elseifmod==4
hi=hikonst;
end
end
P(j,1)=Pinn;
Q=0;
dpftp=0;
dpfsp=0;
dpa=0;

100

dph=0;
dphdl=0;
x(1)=0;
Tw(j,1)=twallstart(T(j,1),
hi,
vsw,teta,
(x(1)+dl),
Twater(j),waterorient);
dp=0;
fori=1:(n)
x(i+1)=x(i)+dl;
ifhokonst==0
ho=hytrekomb(Tw(j,i),vsw,teta,x(i+1),Twater(j),waterorient);
else
ho=hokonst;
end
hutside(j,i)=ho;
ro=interp1(data(1,:),data(gasdensrow,:),T(j,i),'spline');
u=(interp1(data(1,:),data(gasviscrow,:),T(j,i),'spline'))/1000;
cp=(interp1(data(1,:),data(gascprow,:),T(j,i),'spline'))*1000;
lambda=(interp1(data(1,:),data(gascondrow,:),T(j,i),'spline'))/1000;
ul=(interp1(data(1,:),data(liqviscrow,:),T(j,i),'spline'))/1000;
rol=interp1(data(1,:),data(liqdensrow,:),T(j,i),'spline');
lambdal=(interp1(data(1,:),data(liqcondrow,:),T(j,i),'spline'))/1000;
cpl=(interp1(data(1,:),data(liqcprow,:),T(j,i),'spline'))*1000;
sigma=(interp1(data(1,:),data(surftensrow,:),T(j,i),'spline'))/1000;
Pr=u*cp/lambda;
Prl=ul*cpl/lambdal;
massfrac=(interp1(data(1,:),data(massfracrow,:),T(j,i)))/100;
ifisnan(massfrac)==1
massfrac=1;
end
hwtab(j,i)=2*k/(Do*log(Do/Di));
ifmassfrac==1
v=m(j)/(ro*(pi*Di^2/4));
Re=ro*v*Di/u;
f=trykkfall(Re,ks,Di);
ifmod~=4
hi=hindre(Re,Pr,Di,l,lambda,f);
elseifmod==4
hi=hikonst;
end
h=1/((1/ho)+Rf+(Do/(hi*Di))+(log(Do/Di)*Do/(2*k)));
hiw=1/((Do/(hi*Di))+Rf+(log(Do/Di)*Do/(2*k)));
T(j,i+1)=T(j,i)+dA*h*(Twater(j)T(j,i))/(m(j)*cp);
Tw(j,i+1)=T(j,i)(h/hiw)*(T(j,i)Twater(j));

101

dQ=m(j)*cp*(T(j,i)T(j,i+1));
dpfdl=f*v^2*ro/(2*Di);
ro2=interp1(data(1,:),data(gasdensrow,:),T(j,i+1),'spline');
dpadl=G^2*((1/ro2)(1/ro));
ifteta==90
dphdl=(ro+ro2)*g/2;
dph=dph+dphdl*dl;
end
dpa=dpa+dpadl;
dpfsp=dpfsp+dpfdl*dl;
P(j,i+1)=P(j,i)dpfdl*dl/1e5+dpadl/1e5dphdl*dl/1e5;
hitab(j,i)=hi;
hiwtab(j,i)=hiw;
htottab(j,i)=h;
elseifmassfrac<1&&massfrac>0
entalp1=(interp1(data(1,:),data(entrow,:),T(j,i),'spline'))*1000;
ifi>1
entalp2=(interp1(data(1,:),data(entrow,:),
T(j,i
1),'spline'))*1000;
dtdh=(T(j,i)T(j,i1))/(entalp1entalp2);
zg=massfrac*cp*dtdh;
end
mliq=abs(m(j)*(1massfrac));
e1=lmevoid(massfrac,ro,rol,G,sigma);
delta=Di/2(e1*Di^2)^(1/2)/2;
vg=G*massfrac/(ro*e1);
vl=G*(1massfrac)/(rol*(1e1));
vgs=G*massfrac/ro;
Re=ro*vg*Di/u;
Rel=4*G*(1massfrac)*delta/(ul*(1e1));
Relo=G*Di/ul;
Rego=G*Di/u;
Regs=ro*vgs*Di/u;
dpfdl=fuchscorrected(massfrac,G,Relo,Rego,rol,ro);
fgs=trykkfall(Regs,ks,Di);
ifmod~=4
ifmassfrac<0.95
ifmod==1
dpfgsdl=trykkfall(Regs,ks,Di)*ro*vgs^2/(2*Di);
cf=(dpfdl/dpfgsdl)^0.445;
taui=dpfdl*Di/4;
hliq=hliquid(Rel,rol,ul,lambdal,cpl,taui);
hg=cf*hindreglatt(Re,Pr,lambda,Di);

102

elseifmod==2
hliq=hliquidboy(massfrac,Relo,lambdal,Prl,rol,ro);
dpfgsdl=fgs*ro*vgs^2/(2*Di);
cf=(dpfdl/dpfgsdl)^0.445;
hg=cf*hindreglatt(Re,Pr,lambda,Di);
elseifmod==3
fi=1+(vg/vl)^(0.5)*((rolro)*g*delta^2/sigma)^(1/4);
hliq=0.003*Rel^0.74*Prl^0.5*lambdal*fi/delta;
hg=fi*hindreglatt(Re,Pr,lambda,Di);
end
heff=1/((1/hliq)+(zg/hg));
elseifmassfrac>0.95
heff=hindre(Regs,Pr,Di,l,lambda,fgs);
end
elseifmod==4
heff=hikonst;
end
h=1/((1/ho)+Rf+(Do/(heff*Di))+(log(Do/Di)*Do/(2*k)));
dQ=h*dA*(T(j,i)Twater(j));
entalp=entalp1*m(j)dQ;
hiw=1/((Do/(heff*Di))+Rf+(log(Do/Di)*Do/(2*k)));
T(j,i+1)=interp1(data(entrow,:),
data(1,:),
entalp/(m(j)*1000),'spline');
Tw(j,i+1)=T(j,i)(h/hiw)*(T(j,i)Twater(j));
ro2=interp1(data(1,:),data(gasdensrow,:),T(j,i+1),'spline');
rol2=interp1(data(1,:),data(liqdensrow,:),T(j,i+1),'spline');
massfrac2=(interp1(data(1,:),data(massfracrow,:),T(j,i+1)))/100;
e2=lmevoid(massfrac2,ro2,rol2,G,sigma);
ifteta==90
dphdl=((ro+ro2)*(e1+e2)/4+((rol+rol2)/2)*(1(e1+e2)/2))*g;
dph=dph+dphdl*dl;
end
dpadl=G^2*((massfrac2^2/(e2*ro2))+
((1massfrac2)^2/((1
e2)*rol2))((massfrac^2/(e1*ro))+((1massfrac)^2/((1e1)*rol))));
dpa=dpa+dpadl;
dpftp=dpftp+dpfdl*dl;
hitab(j,i)=heff;
htottab(j,i)=h;
P(j,i+1)=P(j,i)dpfdl*dl/1e5+dpadl/1e5dphdl*dl/1e5;
elseifmassfrac==0
v=m(j)/(rol*(pi*Di^2/4));
Re=rol*v*Di/ul;
f=trykkfall(Rel,ks,Di);

103

ifmod~=4
hi=hindre(Rel,Pr,Di,l,lambdal,f);
elseifmod==4
hi=hikonst;
end
h=1/((1/ho)+(Do/(hi*Di))+(log(Do/Di)*Do/(2*k)));
hiw=1/((Do/(hi*Di))+(log(Do/Di)*Do/(2*k)));
T(j,i+1)=T(j,i)+dA*h*(Twater(j)T(j,i))/(m(j)*cpl);
Tw(j,i+1)=T(j,i)(h/hiw)*(T(j,i)Twater(j));
dQ=m(j)*cpl*(T(j,i)T(j,i+1));
dpfdl=f*v^2*rol/(2*Di);
ro2l=interp1(data(1,:),data(liqdensrow,:),T(j,i+1),'spline');
dpadl=G^2*((1/ro2l)(1/rol));
dpa=dpa+dpadl;
dpfsp=dpfsp+dpfdl*dl;
P(j,i+1)=P(j,i)dpfdl*dl/1e5dpadl/1e5;
hitab(j,i)=hi;
hiwtab(j,i)=hiw;
htottab(j,i)=h;
end
dp=dp+dpfdl*dldpadl+dphdl*dl;
Q=Q+dQ;
end
trykktap(j)=dp;
dampfrac(j)=massfrac;
Qtot=Qtot+Q;
end
[dpmin,jmin]=min(trykktap);
[dpmax,jmax]=max(trykktap);
end
deltaH=interp1(data(1,:),data(entrow,:),Tstart,'spline')*mtot*1000Qtot;
Tout=interp1(data(entrow,:),data(1,:),deltaH/(mtot*1000));
massfracout=interp1(data(1,:),data(massfracrow,:),Tout)/100;
deltaP=mean(trykktap)/1e5;
output.Duty_kW=Qtot/1e3;
output.T_out=Tout;
output.Pressure_loss_Bar=deltaP;
output.Outer_Area_m2=pi*Do*l*xtubes;
output.Mass_fraction=massfracout;
disp(output)
x=0:dl:l;
figure(1)
ji=plot(x,T(1,:),x,Tw(1,:));

104

gridon
xlabel('z[m]')
ylabel('T[C]')
legend('Tube','Wall');
set(ji,'LineWidth',2.5)
figure(2)
x2=0:dl:(ldl);
gi=plot(x2,hitab(1,:),x2,hutside(1,:),x2,htottab(1,:));
gridon
xlabel('z[m]')
ylabel('HTC[W/m2K]')
legend('TubesideHTC','WatersideHTC','OverallHTC');
set(gi,'LineWidth',2.5)

euler1cocurrent
Similar to euler1naturlig, but with the following additional lines for
updatingthewatertemperature.Smallfontillustratesexistingcode.

.
htottab(j,i)=h;
end

Twater(j,i+1)=Twater(j,i)+dQ/(msw*(1/xtubes)*cpsw);

euler1crossmixed
Similar to euler1naturlig, but with the following additional lines for
updatingthewatertemperature:

dampfrac(j)=massfrac;

ifj<xtubes
Twater(j+1)=Q/(msw*cpsw)+Twater(j);
end

euler1crossunmixed
Similar to euler1naturlig, but with the following additional lines for
updatingthewatertemperature:

htottab(j,i)=h;
end

ifj<xtubes
Twater(j+1,i)=Twater(j,i)+dQ/(msw*(dl/l)*cpsw);
end

euler1countercurrent
Similar to euler1cocurrent, but with an additional whileloop for the
determiningtheseawatertemperatureattheinlet:

105

for j=1:xtubes

Tswguess=Tsw+5;
Tswend=0;
while(abs(TswendTsw))>0.5
..
Qtot=Qtot+Q;
Tswend=Twater(j,i);
tempdiff=abs(TswendTsw);
dT=tempdiff*0.5;
ifTswend<Tsw
Tswguess=Tswguess+dT;
elseifTswend>Tsw
Tswguess=TswguessdT;
end
end

heatex
clearall
global Do Di k ks vsw mtot Tstart Pinn data l mod textdata xtubes
waterorientmswTswshikonsthokonstcpswRf
s=warning('off','all');
%Tubespesifications
Do=0.0603;
veggtykkelse=0.00391;
Di=Do2*veggtykkelse;
l=220;
k=15;%Tubethermalconductivity
ks=0.002e3;%Tubewallroughness
xtubes=1;%NUmberoftubes
teta=0;
%Inletconditions
Tstart=80;
mtot=1;%totalmassflow
Pinn=100;%inletpressure

%Stepsize
dl=0.04;
%Watersideflow
%1fornaturalconvection,2forcrossflowmixed,3forcrossflowunmixed,
4forcocurrrent,5for
%countercurrent
waterorient=1;
Tsw=5;

106


%Foulingresistance
Rf=0;

Atot=Do*pi*l*xtubes;
rosw=1000;
cpsw=4200;
ifwaterorient==1
vsw=0;
elseifwaterorient==2||waterorient==3
msw=Inf;
vsw=0.6;
elseifwaterorient==4||waterorient==5
msw=Inf;
vsw=1;
end
%filewithproperties
importfile('gas100.csv')
data;
textdata;
%CorrelationforHTCincondesing:1forHEDH,2forBoyko&
%Kruzshuilin,3forthome4forkonst
mod=2;

ifmod==4
hikonst=1000;%ConstanttubesideHTC
end

%Hokonstequaltozeroifnotused.ifdesirablewithconstantouterHTC
%sethokonsttodesiredvalue
hokonst=0;

ifwaterorient==1
euler1naturlig
elseifwaterorient==2
euler1crossmixed
elseifwaterorient==3
euler1crossunmixed
elseifwaterorient==4
euler1cocurrent
elseifwaterorient==5
euler1countercurrent
end

107

108