You are on page 1of 191

Methods for the Characterization of Deposition and

Transport of Magnetite Particles in Supercritical Water

by

Keigo Karakama

B.A.Sc., The University of British Columbia, 2009

A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF

THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF APPLIED SCIENCE

in

THE FACULTY OF GRADUATE STUDIES

(Mechanical Engineering)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

December 2011

© Keigo Karakama, 2011


Abstract

Generation IV CANDU Supercritical Water Reactor (SCWR) is being developed to use a

light water coolant at high temperature and pressure beyond the critical point of water (374⁰C

and 22.1 MPa). The dramatic decrease in the solubility of magnetite in supercritical water

suggests that the precipitation of magnetite particles will occur in the reactor core which can

deposit on the fuel cladding or be transported to the steam turbine.

A once-through flow system was modified to develop experimental techniques for studying

the deposition and transport of magnetite particles in supercritical water onto stainless steel

316L. Experiments were run with temperatures ranging from 200°C to 400°C and a pressure

of 23.7 MPa. A hydrothermal method for synthesizing magnetite particles was adapted for

producing simulated corrosion products in which a typical run had an iron concentration of

0.005 mol/L and lasted for 40 minutes.

An online monitoring technique using thermal resistance to infer deposit loadings showed

deposition and removal cycles of the corrosion product on the tube wall. Scanning electron

microscope images of particles on the tube inner wall and those collected by the high

temperature, high pressure filters revealed magnetite particles which were several hundred

nanometers to several microns in diameter depending on the precursor and condition of the

system. Ultrasound and acid wash cleaning methods were used to remove deposits from the

test section for determining deposit thickness and adhesive strength. The strength of deposit

adhesion was observed to increase along the tube, particularly under supercritical conditions

suggesting precipitation of dissolved species may enhance the strength of the deposit. By

ii
comparing the results, a comprehensive approach was developed to study magnetite fouling

in supercritical water conditions.

Finally, comparison between a simulation model based on mass transport equations and

experimental deposition suggests that mass transport alone can overestimate the deposition

thickness when surface attachment and removal are significant as they were for many

experiments in this study. The simulation predicted as an upper limit scenario that fouling in

a CANDU SCWR could increase the fuel cladding temperature at certain locations by up to

23.9⁰C after one year of operation.

iii
Preface

Chapter 5 is based on experimental work conducted at the mechanical engineering

laboratories by the author, Keigo Karakama. Thermocouple temperature measurements and

its analysis were done by the author. All Atomic absorption spectroscopy analysis, scanning

electron microscopy images and energy dispersive x-ray analysis was completed by the

author at the Materials Engineering Laboratories. X-ray diffraction data and analysis in

Chapter 5 was produced by Anita Lam in the Chemistry Department.

iv
Table of Contents

Abstract ..................................................................................................................................... ii
Preface...................................................................................................................................... iv
Table of Contents ...................................................................................................................... v
List of Tables ......................................................................................................................... viii
List of Figures .......................................................................................................................... ix
List of Symbols and Abbreviations......................................................................................... xv
Acknowledgements ............................................................................................................... xvii
1 Introduction ....................................................................................................................... 1
1.1 Supercritical Water ..................................................................................................... 1
1.2 Systems using Supercritical Water Medium .............................................................. 3
1.3 Scope and Objectives ................................................................................................. 3
2 Literature Review.............................................................................................................. 6
2.1 Applications of Supercritical Water ........................................................................... 6
2.1.1 Supercritical Fossil-fueled Power Plants ............................................................ 6
2.1.2 Supercritical Water Oxidation ............................................................................ 8
2.1.3 Nuclear Energy and the Generation IV Nuclear Reactors .................................. 8
2.1.4 CANDU Supercritical Water Reactors ............................................................. 10
2.2 Materials for SCWR ................................................................................................. 11
2.3 Corrosion and Fouling in Subcritical and Supercritical Water ................................ 13
2.3.1 Properties of Magnetite ..................................................................................... 13
2.3.2 Magnetite and its Solubility in High Temperature and Pressure Water ........... 15
2.3.3 Corrosion........................................................................................................... 17
2.3.4 Fouling .............................................................................................................. 18
2.3.5 Corrosion Product Removal and Transport ...................................................... 22
3 Heat Transfer and Deposition Modeling......................................................................... 25
3.1 Overview .................................................................................................................. 25
3.2 Heat Transfer Calculations of Water........................................................................ 25
3.2.1 Enthalpy of Fluid in the Test Section ............................................................... 25
3.2.2 Heat Transfer Coefficient in Supercritical Water ............................................. 26

v
3.2.3 Temperature of the Oxide Layer ....................................................................... 28
3.2.4 Temperature of the Outside Wall Temperature ................................................ 28
3.2.5 Buoyancy Calculations ..................................................................................... 29
3.3 Deposition Calculations ........................................................................................... 31
3.4 CANDU Supercritical Water Reactor Simulation ................................................... 38
4 Experiments .................................................................................................................... 41
4.1 Supercritical Water Once-through Flow Apparatus ................................................. 41
4.2 Flow and Temperature Control ................................................................................ 44
4.2.1 Safety ................................................................................................................ 44
4.2.2 Pressure Measurements ..................................................................................... 45
4.2.3 Flow Rate Measurements .................................................................................. 45
4.2.4 Fluid Conductivity Measurements .................................................................... 46
4.2.5 Dissolved Oxygen Concentration Measurements ............................................. 46
4.3 Particle Synthesis for Modeling Power Plant Fouling ............................................. 46
4.4 Experimental Procedure ........................................................................................... 48
4.4.1 Experiments with Precursor .............................................................................. 48
4.4.2 Experiments with Magnetite Particles .............................................................. 50
4.4.3 Post Experiment Cleaning................................................................................. 51
4.5 Analytical Methods .................................................................................................. 51
4.5.1 Online Thermal Resistance Monitoring ............................................................ 51
4.5.2 SEM Imaging for Filters and Tube Deposits .................................................... 59
4.5.3 Particulate Filtering System .............................................................................. 60
4.5.4 Deposit Thickness and Strength of Oxide Adhesion ........................................ 67
5 Results and Discussion of Experiments .......................................................................... 69
5.1 Reference Condition ................................................................................................. 70
5.1.1 Temperature Analysis ....................................................................................... 70
5.1.2 Filter Analysis ................................................................................................... 72
5.1.3 Deposit Analysis ............................................................................................... 73
5.2 Effect of Heat Flux ................................................................................................... 80
5.2.1 Temperature Analysis ....................................................................................... 80
5.2.2 Filter Analysis ................................................................................................... 80

vi
5.2.3 Deposit Analysis ............................................................................................... 84
5.3 High pH Conditions ................................................................................................. 87
5.3.1 Temperature Analysis ....................................................................................... 89
5.3.2 Filter Analysis ................................................................................................... 91
5.3.3 Deposit Analysis ............................................................................................... 93
5.4 Low Concentration and Blank Experiment .............................................................. 95
5.5 Overall Summary and Mass Balance ....................................................................... 95
6 Results and Discussion of Simulations ......................................................................... 100
6.1 Comparison between Simulation and Experimental .............................................. 100
6.2 Simulation for CANDU SCWR ............................................................................. 103
6.3 Limitations of the Simulation................................................................................. 107
7 Conclusion .................................................................................................................... 109
8 Recommendations ......................................................................................................... 112
References ............................................................................................................................. 113
Appendix A: MATLAB Code .............................................................................................. 120
Appendix B: Filter Parts Drawings ....................................................................................... 131
Appendix C: Results of Experiments.................................................................................... 133
C. 1. Reference Condition, Ferrous Chloride .............................................................. 133
C. 2. Reference Condition, Ferrous Sulfate ................................................................ 138
C. 3. Low Heat Flux, Ferrous Chloride ....................................................................... 143
C. 4. Low Heat Flux, Ferrous Sulfate ......................................................................... 148
C. 5. No Heat Flux, Ferrous Sulfate ............................................................................ 153
C. 6. High pH, Ferrous Sulfate .................................................................................... 158
C. 7. High pH & Subcritical, Ferrous Sulfate ............................................................. 163
C. 8. Low Concentration, Ferrous Chloride ................................................................ 166
C. 9. Blank Run ........................................................................................................... 169
Appendix D: Filter Flow Rates ............................................................................................. 172
Appendix E: Atomic Absorption Spectroscopy.................................................................... 173

vii
List of Tables

Table 2-1: Generation IV systems and expected deployment year........................................... 9


Table 2-2: Stainless steel AISI 316L composition (American Society for Metals, 1985) ..... 12
Table 3-1: SCWR and fuel assembly design parameters ........................................................ 39
Table 4-1: Input power comparison ........................................................................................ 55
Table 4-2: Glass and silver membrane filter comparison ....................................................... 63
Table 5-1: Summary of experiments....................................................................................... 69
Table 5-2: Results summary table - Part I .............................................................................. 96
Table 5-3: Results summary table - Part II ............................................................................. 97
Table 5-4: Mass balance of experiments #1 - 8 ...................................................................... 99
Table C-1: Summary table of experiment #1 ........................................................................ 133
Table C-2: Summary table of experiment #2 ........................................................................ 138
Table C-3: Summary table of experiment #3 ........................................................................ 143
Table C-4: Summary table of experiment #4 ........................................................................ 148
Table C-5: Summary table of experiment #5 ........................................................................ 153
Table C-6: Summary table of experiment #6 ........................................................................ 158
Table C-7: Summary table of experiment #7 ........................................................................ 163
Table C-8: Summary table of experiment #8 ........................................................................ 166
Table C-9: Summary table of blank run ............................................................................... 169
Table D-1: Flow rate through filter ....................................................................................... 172

viii
List of Figures

Figure 1-1: Pressure - temperature diagram for water, grey region indicating supercritical
conditions............................................................................................................. 1
Figure 1-2: Density of water vs. temperature at various pressures, showing the effects of
pressure on the pseudocritical temperature ......................................................... 2
Figure 2-1: Simplified drawing of a direct cycle coolant for Generation IV CANDU SCWR
........................................................................................................................... 11
Figure 2-2: Pourbaix diagram (a) Fe – H2O at 25ºC and 1 atm, (b) Fe – H2O at 374°C and
220 atm (c) Fe/Cr – H2O at 25°C and 1 atm, (d) Fe/Cr – H2O at 374ºC and 220
atm. Based on SUPCRT92 (Propp et al., 1996) ............................................... 14
Figure 2-3: Magnetite solubility adapted from Cook and Fatoux (2009) for neutral pH water
at 25 MPa, Guzonas et al. (2009) for pH = 9.3 at 30 MPa, Burrill (2000) for
neutral pH at 25 MPa, and Sweeton and Baes Jr. (1970) experimental results. 16
Figure 3-1: Calculation of Grq/Grth for determination of the effects of buoyancy along the
test section using Bazargan et al. (2005) ........................................................... 31
Figure 3-2: Effect of particle diameter and fluid temperature on deposition velocity for
experimental apparatus ...................................................................................... 38
Figure 3-3: Cross-sectional view of CANFLEX 43 rod fuel bundle and drawing of
equivalent tube for simulation calculations (not to scale) ................................. 40
Figure 4-1: SCW system used in all experiments located in the Mechanical Engineering
laboratories. ....................................................................................................... 41
Figure 4-2: Schematic of supercritical once-through flow apparatus................................... 42
Figure 4-3: Mounting block and test section inlet ................................................................ 54
Figure 4-4: Electrical wiring, heater controls and location of surface welded thermocouples
and bulk fluid thermocouples ............................................................................ 57
Figure 4-5: Saturation temperature of 231.51ºC at 417psi, calibration for Experiment #1 . 58
Figure 4-6: Temperature of test section, calculated vs. calibrated thermocouples............... 58
Figure 4-7: Test section sample for SEM analysis ............................................................... 60
Figure 4-8: SS316L LTHP filter installed after the heat exchanger ..................................... 61
Figure 4-9: HTHP filter with silver membrane and SS304 back support ............................ 62
Figure 4-10: Glass fiber membrane and Swin-Lok LTLP filter holder .................................. 63
Figure 4-11: SEM photograph of clean silver membrane, 0.2µm pore size, using a Hitachi
S3000N SEM ..................................................................................................... 65
Figure 4-12: EDX of silver membrane to determine elemental composition of background 65
Figure 4-13: SEM photograph of clean glass fiber filter, 0.7µm pore size, a Hitachi S2300
SEM ................................................................................................................... 66
Figure 4-14: EDX of glass fiber filter to determine elemental composition of background .. 66
Figure 5-1: Temperature measurements of test section vs. time for reference condition with
ferrous chloride precursor (Experiment #1) ...................................................... 71

ix
Figure 5-2: SEM photograph of magnetite deposit on test section tube surface, 0.15 m
location. Reference condition with ferrous chloride precursor (Experiment #1)
........................................................................................................................... 73
Figure 5-3: SEM photograph of magnetite deposit on test section tube surface, 0.60 m
location. Reference condition with ferrous chloride precursor (Experiment #1)
........................................................................................................................... 74
Figure 5-4: SEM photograph of magnetite deposit on test section tube surface, 1.20 m
location. Reference condition with ferrous chloride precursor (Experiment #1)
........................................................................................................................... 74
Figure 5-5: SEM photograph of magnetite deposit on test section tube surface, 1.65 m
location. Reference condition with ferrous chloride precursor (Experiment #1)
........................................................................................................................... 75
Figure 5-6: XRD of tube deposits of test section, reference condition ferrous chloride
precursor (Experment #1) .................................................................................. 76
Figure 5-7: XRD of tube deposit of test section, reference condition ferrous sulfate
precursor (Experiment #2) ................................................................................. 76
Figure 5-8: Deposit thickness on test section for reference condition with ferrous chloride
precursor (Experiment #1) ................................................................................. 79
Figure 5-9: Deposit thickness on test section for reference condition with ferrous sulfate
precursor (Experiment #2) ................................................................................. 79
Figure 5-10: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for
low heat flux with ferrous chloride precursor (Experiment #3) ........................ 81
Figure 5-11: EDX of HTHP filter for low heat flux with ferrous chloride precursor ............ 81
Figure 5-12: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for
low heat flux with ferrous sulfate precursor (Experiment #4)........................... 82
Figure 5-13: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for
no heat flux with ferrous chloride (Experiment #5) .......................................... 82
Figure 5-14: XRD of deposit on HTHP filter for low heat flux with ferrous chloride
precursor (Experiment #3) ................................................................................. 83
Figure 5-15: Deposit thickness on test section, low heat flux with ferrous sulfate precursor
(Experiment #4) ................................................................................................. 84
Figure 5-16: Calibration of the conductivity meter for determining NaOH concentration in
primary tank ....................................................................................................... 88
Figure 5-17: Predicted pH of the system using data from Arden (1950).............................. 89
Figure 5-18: Temperature measurements of test section vs. time for high pH with ferrous
sulfate precursor (Experiment #6). Power turned off briefly at t = 48 minutes. 90
Figure 5-19: SEM photograph of particles collected on HTHP filter, 10.0k magnification, for
high pH, ferrous sulfate (Experiment #6) .......................................................... 91
Figure 5-20: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
high pH, ferrous sulfate (Experiment #6) .......................................................... 92

x
Figure 5-21: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for
high pH, subcritical, ferrous sulfate (Experiment #7) ....................................... 92
Figure 5-22: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
high pH, subcritical, ferrous sulfate (Experiment #7) ....................................... 93
Figure 5-23: Deposit thickness on test section (Experiment #6) ............................................ 94
Figure 6-1: Comparison between simulation and experimental for reference conditions (a)
Experiment #1, ferrous chloride with average particle size = 4 µm, magnetite
concentration = 200 mg/L (b) Experiment #2, ferrous sulfate with average
particle size = 1 µm, magnetite concentration = 299 mg/L. ............................ 100
Figure 6-2: Comparison between simulation and experimental for ferrous sulfate
experiments with (a) Experiment #4, low heat flux with average particle size =
1 µm, magnetite concentration = 108 mg/L and (b) Experiment #5, no heat flux
with average particle size = 2 µm, magnetite concentration = 89 mg/L. ........ 101
Figure 6-3: Comparison between simulation and experimental for Experiment #6, high pH,
ferrous sulfate. Simulation was run with an average particle size = 0.15µm,
magnetite concentration = 321 mg/L. .............................................................. 102
Figure 6-4: Bulk fluid and fuel cladding-surface temperature without particle deposition 104
Figure 6-5: Bulk fluid and fuel cladding-surface temperature after 1 year of operation with
magnetite particles of 150 nm diameter .......................................................... 104
Figure 6-6: Magnetite deposition thickness on fuel cladding after 1 year of operation for
150 nm and 1 µm average particle diameters .................................................. 105
Figure C-1: Temperature measurements of test section vs. time for reference condition with
ferrous chloride precursor. (Experiment #1) ................................................... 134
Figure C-2: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
reference condition with ferrous chloride precursor. Low flow rate in secondary
line resulting in very few particles collected. (Experiment #1) ....................... 134
Figure C-3: SEM photograph of particles collected on LTHP filter, 1.8k magnification, for
reference condition with ferrous chloride precursor. (Experiment #1) ........... 135
Figure C-4: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
reference condition with ferrous chloride precursor. Low flow rate in secondary
line resulting in very few particles collected. (Experiment #1) ....................... 135
Figure C-5: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m,
(b) 0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #1) .................. 136
Figure C-6: Deposit thickness on test section for reference condition with ferrous chloride
precursor. (Experiment #1) .............................................................................. 137
Figure C-7: Comparison between simulation and experimental for reference condition
ferrous chloride. Simulation was run with an average particle size = 4 µm,
magnetite concentration = 200 mg/L. .............................................................. 137

xi
Figure C-8: Temperature measurements of test section vs. time, temperature fluctuations
were found in both the inlet and outlet bulk temperatures which may be
attributed to the inconsistent flow. (Experiment #2) ....................................... 139
Figure C-9: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2) .............................. 139
Figure C-10: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2) .............................. 140
Figure C-11: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2) .............................. 140
Figure C-12: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m,
(b) 0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #2) .................. 141
Figure C-13: Deposit thickness on test section for reference condition with ferrous sulfate
precursor. (Experiment #2) .............................................................................. 142
Figure C-14: Comparison between simulation and experimental for reference condition
ferrous sulfate. Simulation was run with an average particle size = 1 µm,
magnetite concentration = 190 mg/L. .............................................................. 142
Figure C-15: Temperature measurements of test section vs. time for low heat flux ferrous
chloride precursor. (Experiment #3) ................................................................ 144
Figure C-16: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
low heat flux with ferrous chloride precursor. (Experiment #3) ..................... 144
Figure C-17: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
low heat flux with ferrous chloride precursor. (Experiment #3) ..................... 145
Figure C-18: SEM photograph of particles collected on LTLP filter, - magnification, for low
heat flux with ferrous chloride precursor. (Experiment #3) ............................ 145
Figure C-19: SEM photograph of magnetite deposit on test section tube surface, (a) 0.10 m,
(b) 0.30 m, and (c) 0.90 m location. (Experiment #3) ..................................... 146
Figure C-20: Deposit thickness on test section for low heat flux ferrous chloride precursor.
(Experiment #3) ............................................................................................... 147
Figure C-21: Temperature measurements of test section vs. time for low heat flux with
ferrous sulfate precursor. (Experiment #4) ...................................................... 149
Figure C-22: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
low heat flux ferrous sulfate precursor (Experiment #4) ................................. 149
Figure C-23: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
low heat flux ferrous sulfate precursor. (Experiment #4) ................................ 150
Figure C-24: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
low heat flux ferrous sulfate precursor. (Experiment #4) ................................ 150
Figure C-25: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m,
(b) 0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #4) .................. 151
Figure C-26: Deposit thickness on test section for low heat flux ferrous sulfate precursor.
(Experiment #4) ............................................................................................... 152

xii
Figure C-27: Comparison between simulation and experimental for low heat flux, ferrous
sulfate. Simulation was run with an average particle size = 1 µm, magnetite
concentration = 108 mg/L................................................................................ 152
Figure C-28: Temperature measurements of test section vs. time for no heat flux with ferrous
sulfate precursor. (Experiment #5) .................................................................. 154
Figure C-29: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
no heat flux with ferrous sulfate precursor. (Experiment #5).......................... 154
Figure C-30: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
no heat flux with ferrous sulfate precursor. (Experiment #5).......................... 155
Figure C-31: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
no heat flux with ferrous sulfate precursor. (Experiment #5).......................... 155
Figure C-32: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m,
(b) 0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #5) .................. 156
Figure C-33: Deposit thickness on test section for no heat flux with ferrous sulfate precursor.
(Experiment #5) ............................................................................................... 157
Figure C-34: Comparison between simulation and experimental for no heat flux with ferrous
sulfate precursor. Simulation was run with an average particle size = 2 µm,
magnetite concentration = 89 mg/L. ................................................................ 157
Figure C-35: Temperature measurements of test section vs. time for high pH with ferrous
sulfate precursor. (Experiment #6) .................................................................. 159
Figure C-36: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
high pH with ferrous sulfate precursor. (Experiment #6) ................................ 159
Figure C-37: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
high pH with ferrous sulfate precursor. (Experiment #6) ................................ 160
Figure C-38: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
high pH with ferrous sulfate precursor. (Experiment #6) ................................ 160
Figure C-39: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m,
(b) 0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #6) .................. 161
Figure C-40:Deposit thickness on test section for high pH with ferrous sulfate precursor.
(Experiment #6) ............................................................................................... 162
Figure C-41: Comparison between simulation and experimental for high pH, ferrous sulfate.
Simulation was run with an average particle size 0.15 µm, magnetite
concentration = 321 mg/L................................................................................ 162
Figure C-42: Temperature measurements of test section vs. time for high pH & subcritical
with ferrous sulfate precursor. (Experiment #7) .............................................. 164
Figure C-43: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
high pH & subcritical with ferrous sulfate precursor. (Experiment #7) .......... 164
Figure C-44: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
high pH & subcritical with ferrous sulfate precursor. (Experiment #7) .......... 165

xiii
Figure C-45: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
high pH & subcritical with ferrous sulfate precursor. (Experiment #7) .......... 165
Figure C-46: Temperature measurements of test section vs. time for low concentration with
ferrous chloride precursor. (Experiment #8) ................................................... 167
Figure C-47: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
low concentration with ferrous chloride precursor. (Experiment #8).............. 167
Figure C-48: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
low concentration with ferrous chloride precursor. (Experiment #8).............. 168
Figure C-49: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
low concentration with ferrous chloride precursor. (Experiment #8).............. 168
Figure C-50: Temperature measurements of test section vs. time for high pH & subcritical
with ferrous sulfate precursor. (Experiment #9) .............................................. 170
Figure C-51: SEM photograph of particles collected on HTHP filter, 2.0k magnification,
blank run. (Experiment #9) .............................................................................. 170
Figure C-52: SEM photograph of particles collected on LTHP filter, 5.0k magnification,
blank run. (Experiment #9) .............................................................................. 171
Figure C-53: SEM photograph of particles collected on LTLP filter, 2.0k magnification,
blank run. (Experiment #9) .............................................................................. 171
Figure E-1: Calibration curve for iron concentration using AAS, nitric acid matrix ......... 173
Figure E-2: Calibration curve for iron concentration using AAS, nitric acid and copper
sulfate matrix ................................................................................................... 174

xiv
List of Symbols and Abbreviations

AAS Atomic Absorption Spectroscopy

AECL Atomic Energy of Canada Limited

CANDU Canadian Deuterium Uranium

Cb Concentration of magnetite in bulk fluid

cp Heat capacity [kJ/kg·K]

Csat Saturation concentration of magnetite

dp Particle diameter [m]

f Friction factor

Gr Grashoff number

H1 Enthalpy [kJ/kg]

H2 Enthalpy [kJ/kg]

HKF Helgeson-Kirkham-Flowers

HTHP High temperature high pressure

Ka Surface attachment deposition velocity [m/s]

Kd Effective deposition velocity [m/s]

Kt Mass transport deposition velocity [m/s]

LTHP Low temperature high pressure

LTLP Low temperature low pressure

Nu Nusselt number

̇ Mass flow rate [kg/s]

Pr Prandtl number

q Heat input per meter [kW/m]

Re Reynolds number

Sc Schmidt number

xv
SCWO Supercritical Water Oxidation

SCWR Supercritical Water Reactor

SEM Scanning Electron Microscope

SPE Solid Particle Erosion

Tb Bulk fluid temperature [ºC]

Te Temperature of outer wall [ºC]

tp + Dimensionless particle relaxation time

Tw Wall temperature [ºC]

U* Frictional velocity

µ Dynamic viscosity [Pa·s]

XRD X-ray Diffraction

Δx Interval length [m]

ε Tube roughness

ρf Density of fluid [kg/m3]

ρp Density of particle [kg/m3]

τw Wall shear stress

d Overall deposition velocity [m/s]

xvi
Acknowledgements

First and foremost I would like to thank my supervisors, Dr. Steven Rogak and Dr. Akram

Alfantazi for their invaluable insight and guidance into making this thesis possible. I would

also like to express my gratitude to Dr. Edouard Asselin whose advice gave further direction

to this thesis. Furthermore, I would like to thank NSERC and AECL for their financial

support in this research.

I would like to thank Andrej Boskovics for his help on the construction of the electrical

heaters and flow loop. Additionally, I would like to thank my lab colleagues, Tirdad Nickchi,

Hamid Zebardast, Hugo Tjong, Arka Soewono, Mohammad Taghavi, and Kamran Alba for

their help, advice and support.

Also, many thanks to the technicians who were incredibly helpful and patient in teaching me

the operation and analysis of the equipment: Dr. Berend Wassink on atomic absorption

spectroscopy, Jacob Kabel on the scanning electron microscope, Anita Lam for X-ray

diffraction analysis, the mechanical machine shop for their assistance in machining

components, the electronics shop for their advice on the electrical heater and controls, and

Perry Yabuno for always making sure that parts were ordered and received as soon as

possible.

Finally, I would like to thank all my friends and family who provided moral support

throughout the entire journey.

xvii
1 Introduction

1.1 Supercritical Water

Supercritical water (SCW) refers to water above the critical point of 374⁰C and 22.1 MPa.

Above this temperature and pressure, water exists only as a single phase fluid with very

unique thermodynamic and fluid properties (Weingartner & Franck, 2005).

Supercritical
water
Liquid
Pressure

374ºC
22.1MPa

Solid
Vapor

Temperature

Figure 1-1: Pressure - temperature diagram for water, grey region indicating supercritical
conditions.

Viscosity, density, and dissolution of non-polar and polar species can change dramatically

from subcritical to supercritical conditions. At subcritical temperatures, water is highly polar

and is an excellent solvent for other polar substances such as inorganic salts. On the other

hand, organic species are generally non-polar and therefore are usually not soluble in

subcritical water. In supercritical conditions, the characteristics described above are reversed

1
and non-polar organic species and gases become significantly soluble, while ionic species

become almost completely insoluble (Kritzer, 2004; Zhang et al., 2009).

While the critical temperature gives the threshold temperature for achieving supercritical

conditions, when describing the temperature at which thermodynamic changes occur, it is

often times more useful to describe the transition using the “pseudocritical” temperature. The

pseudocritical temperature is the temperature at which fluid thermal expansion is at its

maximum for any given supercritical pressure. This temperature increases as the pressure

increases as shown in Figure 1-2. Other parameters such as heat capacity and ionic product of

water are also affected by pressure and its transition also shifts to higher temperature as

pressure is increased (Kritzer, 2004).

Figure 1-2: Density of water vs. temperature at various pressures, showing the effects of
pressure on the pseudocritical temperature

2
1.2 Systems using Supercritical Water Medium

Supercritical water has been a crucial medium in numerous industrial and research

applications for its interesting properties. Large variations in fluid and thermal properties

near the critical point offers an opportunity to fine tune fast chemical reactions with only

small adjustments in pressure and temperature (Kritzer, 2004). The special properties of

supercritical water as an excellent solvent for organic material have led to the technology of

supercritical water oxidation systems (Shaw et al., 1991) while the excellent heat transport

characteristics have led to power plants operating at higher temperature and pressure.

Material degradation has been a major limiting factor, however, new technology in materials

in the last few decades have allowed coal-fired power plants to operate in supercritical

conditions (Viswanathan et al., 2005). Finally, the need for cleaner and more efficient energy

has recently prompted research into supercritical water reactors (SCWR) for nuclear power

plants. Despite the operational experience with supercritical water in various applications,

corrosion and fouling in this medium remains a significant challenge.

1.3 Scope and Objectives

Both the UBC Mechanical and Materials Engineering departments are collaborating with

AECL in the development of monitoring techniques for corrosion and fouling in supercritical

water. The primary application for the technology is for the CANDU supercritical water

reactors, however, the scope of this research can be applied for many similar systems such as

supercritical fossil-fueled and supercritical light water reactors. The purpose of this particular

study is to explore the deposition and transport characteristics of magnetite formed in

supercritical water with several experimental techniques and simulation models.

3
Experimental investigations were conducted to examine four techniques for measuring and

monitoring fouling and transport of magnetite in supercritical water. These experiments were

focused on determining the effects of heat flux, temperature and pH on the particle size,

deposition rate, and adhesive strength of particle to the wall. A once-through flow loop was

modified to conduct experiments of magnetite deposition in conditions which attempted to

replicate SCWR conditions. A hydrothermal method for synthesizing iron oxide particles in

supercritical water was adapted for producing simulated magnetite corrosion particles.

Experimental procedures and description of the apparatus can be found in Chapter 4.

A high-current, electrically-heated test section with temperature control and tube-wall

temperature measurements was constructed by the author for this study to measure fouling

and removal rates. High and low temperature filters were also fabricated and installed after

the test section to collect particles in each experiment. These filters were analyzed using

Scanning Electron Microscopy (SEM) to provide insight into the effects of pH on the

magnetite particle size and structure produced in supercritical water. In addition, deposition

morphology and thickness was determined using SEM and Atomic Absorption Spectroscopy

(AAS). A novel, two-step cleaning procedure was developed for qualitatively determining

the bond strength of the magnetite deposit.

A heat and mass transfer model was created to compare simulations with experimental

deposition thicknesses to determine if a mass transport limited model can accurately predict

deposition of magnetite particles in supercritical water. Due to the lack of experimental data

on the deposition mechanism of magnetite particles in supercritical water, deposition was

4
assumed to be mass transport limited. Simulations were also run to quantify the deposition of

magnetite onto a CANDU nuclear reactor for one year. The results and its implications on a

SCWR are presented in Chapter 6. Finally the conclusions and recommendations for future

work are summarized in Chapters 7 and 8 respectively.

5
2 Literature Review

2.1 Applications of Supercritical Water

2.1.1 Supercritical Fossil-fueled Power Plants

The thermal efficiency of fossil-fueled power plants has increased over the past century by

operating steam at higher pressure and temperature, and thereby increasing the

thermodynamic efficiency (Masuyama, 2001). In the 1920’s, the pressure and temperature

was limited to 4 MPa and 370°C due to the mechanical properties of carbon steel. However,

technological improvements in materials such as the development of nickel-based alloys and

high temperature resistant steels have allowed coal-fired power plants to operate as high as

24 MPa and 600°C. With nearly two dozen power plants worldwide currently operating at

580 – 600°C and 24 – 25 MPa, there is considerable operational experience with supercritical

water in power generation (Viswanathan et al., 2005). These experiences with supercritical

fossil-fueled power plants can provide invaluable insight into possible operational issues for

a nuclear SCWR.

Studies on the scaling of austenitic steels in supercritical fossil-fueled power plants show a

duplex oxide structure in which the outer layer is likely to exfoliate compared to an inner

layer of chromium and nickel oxide which almost never exfoliates. Once the scale thickness

reaches approximately 100 – 200 µm, oxides on austenitic steel begin to exfoliate due to the

differences in thermal expansion between the tube surface and the outer oxide layer,

especially during startup and stop time of the boiler (Masuyama, 2001). It was also found

that fine-grained steels and high-Cr steels had very slow oxide growth with very little

6
exfoliation. In general, chromium content above 20% was effective at preventing steam oxide

buildup. Tube plugging due to exfoliated material buildup can cause overheating in the

portion of the tube, resulting in creep rupture of the piping when the pressure due to the

fouling buildup exceeds the creep strength of the tubing (Masuyama, 2001). In addition,

exfoliated material has also been found to erode turbine components in a supercritical fossil-

fueled power plant if solid particles are found in the fluid stream. Therefore, scale build-up

and its resistance to exfoliation is a major concern for any system running SCW.

There are several key differences between a supercritical fossil-fueled power plant and

nuclear SCWR design requirements which make it particularly challenging for SCWR (Was

et al., 2007). First, compared to supercritical fossil-fueled power plants which usually can

have fire tubes with wall thickness of up to 10 mm, the fuel cladding in a nuclear reactor core

is restricted to 0.5 - 0.6 mm due to thermal and neutron efficiency (Zhang et al., 2009).

Larger amount of corrosion product deposits are often found on superheater tubes operating

above the critical point in a supercritical fossil-fueled power plant and it is not unusual to

find oxide films of several hundred micrometer thickness on these tubes (Cook & Fatoux,

2009). Such high fouling rates and deposit thickness would be unacceptable for a fuel

cladding in a SCWR. Second, the cooling water in the boiler tubes sees a geometrically

smooth surface along the boiler while the core of a SCWR will have a fuel assembly of

intricate geometry (Was et al., 2007).

7
2.1.2 Supercritical Water Oxidation

The high solubility of organic species in supercritical water has led to the research in

“supercritical water oxidation” (SCWO), technology which was developed for destroying

hazardous organic waste (Helling & Tester, 1988). Despite its initial success, problems with

fouling and corrosion resulted in high operating costs which has forced the majority of

commercial SCWO plants constructed and operated in the late 90’s and early 2000’s to

permanently shut down (Marrone et al., 2005). There was extensive experimental work

conducted on fouling and corrosion in supercritical conditions for SCWO applications,

however, the environment is often highly oxidizing and in some cases highly acidic which

are significantly different from the water chemistry which will likely be used for a CANDU

SCWR (Kritzer, 2004). It has been well observed from SCWO research that acidic water

chemistry promotes rapid corrosion, especially slightly below the pseudocritical temperature.

With inorganic salts in supercritical conditions, very high fouling rates are observed and the

deposit can completely plug flow of a system resulting in increased system pressure and the

need for immediate shut down and cleaning (Teshima, 1997).

2.1.3 Nuclear Energy and the Generation IV Nuclear Reactors

As energy demand continues to increase and the effects of energy production on the global

environment escalates, the need for cleaner and more efficient technology grows ever more

important. Nuclear energy provides approximately 13.8% of the world’s energy demand and

its fuel provides energy with low greenhouse gas emissions making it an ideal candidate for

future energy production (International Atomic Energy Agency, 2010). These facilities

require some of the most advanced technology available and the Generation III++ is the most

8
recent design incorporating the technology developed for the industry over the last half

century. Generation IV nuclear technology is to succeed the current Generation III++ nuclear

reactors with state-of-the-art technology to improve on safety, lower byproduct waste, and

higher efficiency (Cook & Fatoux, 2009; Khartabil, 2009). There are currently six emerging

technologies for Generation IV nuclear reactors, one of these systems being SCWR in which

the coolant will operate above the critical point of water. The first of the Generation IV

nuclear reactors utilizing SCWR technology are expected to go online in 2025 as presented in

Table 2-1 (Generation IV International forum SCWR Committee, 2002) .

Table 2-1: Generation IV systems and expected deployment year

Generation IV R&D Systems Year


Gas-Cooled Fast Reactor System 2025
Lead-Cooled Fast Reactor System 2025
Molten-Salt Reactor System 2025
Sodium-Cooled Fast Reactor System 2015
Supercritical-Water-Cooled Reactor System 2025
Very-High Temperature Reactor System 2020

The higher operating temperature and pressure of a SCWR is expected to increase the

thermodynamic efficiency of nuclear power plants to 45 - 50% in comparison to the 33% -

35% for conventional nuclear power plants (Naidin et al., 2009). This is due to the single

phase coolant which improves heat transfer and simplifies the overall design by reducing the

number of equipment and flow loops. In addition, the higher enthalpy of the coolant at higher

temperatures especially near the critical point offers an opportunity to decrease the mass flow

of the coolant for the same thermal output, resulting in smaller pumps, piping and housing

(Generation IV International forum SCWR Committee, 2002).

9
2.1.4 CANDU Supercritical Water Reactors

The CANDU (Canadian Deuterium Uranium) nuclear reactor is a Canadian-designed nuclear

reactor developed by Atomic Energy of Canada Limited (AECL). The CANDU nuclear

design is unique in that it uses heavy water as a moderator versus light water, which allows

the reactors to use non-enriched uranium as fuel. In addition, the CANDU reactors use

pressure tubes rather than a pressure vessel in the reactor core, simplifying high pressure

designs and making the movement towards a supercritical coolant the next logical step

(Torgerson et al., 2006).

The coolant would enter the reactor core at 350⁰C and leave the core at above 600⁰C and

25MPa, finally leading to a high pressure turbine in the case of a direct cycle system (Naidin

et al., 2009). A simplified drawing of the proposed CANDU SCWR is provided in Figure

2-1.

The low solubility of inorganic salts in supercritical water suggest that there is greater

possibility for higher fouling rates of metal oxides in a SCWR core than previously seen in a

subcritical CANDU reactor. Metal oxides may nucleate from solution to form nano to

micrometer sized particles which may foul onto various heat transfer surfaces, leading to

higher than expected temperatures and accelerating material degradation. Particles which do

not deposit may be carried in the bulk fluid to the steam turbines which can lead to solid

particle erosion and out-of-core radiation (Guzonas et al., 2009).

10
Turbine Generator

Reactor Core

Condenser

Preheater

Pump

Figure 2-1: Simplified drawing of a direct cycle coolant for Generation IV CANDU SCWR

2.2 Materials for SCWR

There are currently five materials that are under evaluation as potential materials to be used

in SCWR for various components such as the fuel cladding, pressure tubes, etc. (Zhang et al.,

2009).

 Nickel Alloys: 690, 625, 718

 Austenitic Stainless Steels: 304NG, 316L, D9, AL-6XN

 Ferretic / Martensitic Steels: P92, P112

 Oxide Dispersed Steels (ODS): 9Cr-ODS, 12Cr-ODS, MA956

 Modified zircaloys

11
Experiments presented in this study are constructed from mainly SS316L material. Austenitic

Stainless Steels such as AISI 304 and AISI 316 are part of a family of stainless steels with

nickel content which preserve the austenite crystal structure during cooling. Compared to

martenitic and ferritic stainless steels, austenitic stainless steel enjoys the advantages of being

easily formable, weldable, and ductile.

The ‘L’ of 316L indicates that the particular stainless steel has lower carbon content than that

of unstabilized austenitic steels such as 316 and 304. When types 316 and 304, are subject to

high temperatures in the range of 550°C to 880ºC, chromium carbide tend to precipitate

along the grain boundaries, forming areas depleted of chromium below the minimum 12%.

These areas become susceptible to corrosion in and near these boundaries where the material

has become sensitized (Sinha, 1989). In operational power plants, these sensitized austenitic

steels have been found to fail by intergranular stress corrosion cracking (IGSCC) in the

tubing of boiler water reactors (R. L. Jones et al., 1993). By lowering the carbon content to

below 0.03%, the amount of M23C6 that forms is minimized. Furthermore, the addition of

molybdenum (Mo) in 316L SS as opposed to 304L SS which does not have Mo, improves

corrosion resistance of stainless steel in chloride aggressive environments by forming a

uniform passive layer (Bastidas et al., 2002).

Table 2-2: Stainless steel AISI 316L composition (American Society for Metals, 1985)

C Mn Si Cr Ni P S Mo N
16.0 - 10.0 -
0.03 2.00 1.00 0.045 0.03 2.0 - 3.0 0.10
18.0 14.0

12
2.3 Corrosion and Fouling in Subcritical and Supercritical Water

2.3.1 Properties of Magnetite

Magnetite is an iron oxide mineral with the chemical composition of Fe3O4. It is also known

as a ferric-ferrous oxide as it is one part wüstite (FeO) and one part hematite (Fe2O3). It is

part of the spinel group which is a group of minerals that have a chemical composition of

A2+B23+O42-. It typically forms as an octahedral, less commonly as dodecahedral and rarely as

a cubic (Anthony et al., 1997). Magnetite, which is the stable form of iron oxide in SCWR

conditions, is expected to be the dominant metal oxide formed in the core due to the presence

of iron in many high temperature alloys. It is a common corrosion product found on the tubes

of fossil-fueled and nuclear power plants (Propp et al., 1996).

Pourbaix diagrams, also known as Eh-pH diagrams indicate the stable regions for metal

species. Solid lines indicate the border between two species while the diagonal dashed lines

indicate the stable region for water. Above the upper dashed line, water is oxidized to oxygen

while below the lower line, water is reduced to hydrogen (D. A. Jones, 1996). The

equilibrium equations of water in neutral or alkaline solutions are shown below:

2-1

2-2

Figure 2-2 is a Pourbaix diagram for iron and iron-chromium created from SUPCRT92.

Because of its dependence on temperature and pH, the borders for stable species will vary

13
from one temperature to another and change dramatically in supercritical conditions (Propp

et al., 1996).

2.20

Fe3+ FeO42-
1.45
FeO42-

0.70
Fe2O3 Fe2O3
2+
-0.05 Fe 2+
Fe
Fe3O4
Fe3O4
-0.88

Fe
Fe
-1.55 HFeO2
Eh(V)

2.20

Cr2O72- Cr2O72-
1.45 FeO42-
FeO42-
CrO42-
CrO2
0.70 CrO2
3+
CrO42-
Cr
Fe2O3
-0.05 Cr3+ Fe2O3

Fe2+ Cr2O3
-0.88 Cr2+ Cr2O3
CrO
Fe CrO33-
-1.55 Fe Cr
Cr

-2 2 6 10 14 -2 2 6 10 14
pH

Figure 2-2: Pourbaix diagram (a) Fe – H2O at 25ºC and 1 atm, (b) Fe – H2O at 374°C and
220 atm (c) Fe/Cr – H2O at 25°C and 1 atm, (d) Fe/Cr – H2O at 374ºC and 220
atm. Based on SUPCRT92 (Propp et al., 1996)

14
2.3.2 Magnetite and its Solubility in High Temperature and Pressure Water

The solubility of corrosion products play a primary role in corrosion and fouling in aqueous

systems (Kritzer, 2004). In particular, the solubility of magnetite, an iron oxide of the form

Fe3O4, in high temperature and pressure water is a crucial information to operators in the

power industry due to the prevalence of the mineral found on heated tube surfaces.

Solubility of magnetite in alkaline and acidic water conditions has been measured up to

300⁰C by several groups. Sweeton and Baes Jr. (1970) tested magnetite solubility with

varying pH using potassium hydroxide and hydrochloric acid while Tremaine and LeBlanc

(1980) adjusted pH with sodium hydroxide and hydrochloric acid. The dissolution of

magnetite requires the reduction of FeIII to FeII. The equilibrium equation is provided in

equation 2-3 (Tremaine & LeBlanc, 1980).

( ) ( ) ( ) 2-3

In order to predict the magnitude of super-saturation and potential fouling rates in

supercritical water, the solubility of magnetite under these conditions is required.

Unfortunately, the experimental solubility data for magnetite in supercritical water is

currently unavailable. There are several major problems which make measuring solubility of

magnetite extremely difficult. First, the low solubility of iron means that measurements must

be done in parts per billion and parts per trillion ranges. Preventing iron contamination of the

solution from tubing, pumps, valves, etc. becomes increasingly difficult as materials that

withstand supercritical conditions are limited to high temperature alloys which often have

iron content eg. 316L, Inconel 625, Alloy 800.

15
To overcome the deficiency in experimental solubility data for magnetite, thermodynamic

studies utilizing the Helgeson-Kirkham-Flowers (HKF) extrapolation for the Gibbs free

energy at high temperature and pressure have been used to predict the solubility of magnetite

in supercritical conditions (Cook & Fatoux, 2009; Guzonas et al., 2009). In another study by

Burrill (2000), magnetite solubility data from Bohnsack was extrapolated from subcritical to

supercritical conditions, assuming a direct dependence of magnetite solubility with water

density. The simulation results are provided for comparison in Figure 2-3.

Figure 2-3: Magnetite solubility adapted from Cook and Fatoux (2009) for neutral pH water
at 25 MPa, Guzonas et al. (2009) for pH = 9.3 at 30 MPa, Burrill (2000) for
neutral pH at 25 MPa, and Sweeton and Baes Jr. (1970) experimental results.

16
All three solubility simulations show a drop in solubility between 350ºC to 400ºC as the

density of water drops and the water transitions from polar to non-polar solvent. For models

based on HKF, the solubility increases once again as the temperature increases even higher

after the initial drop. The HKF model has been successful in correlating equilibrium

constants for hundreds of inorganic aqueous species in high temperature and high pressure

water including supercritical water (Sue et al., 2002). Cook and Fatoux model was adopted

for the simulation in this study for this reason as well as because it used pressure closer to

those expected in a CANDU SCWR of 25 MPa. Under this model, minimum solubility is

reached at 400°C and 25 MPa under neutral pH conditions with an iron concentration of

approximately 0.2µg/L. At supercritical conditions, neutral species such as Fe(OH)2 and

Fe(OH)3 are expected to be the dominant dissolved species in the solution (Cook & Fatoux,

2009).

2.3.3 Corrosion

Due to significant changes in the thermal and chemical properties of water from subcritical to

supercritical conditions, corrosion rates and mechanisms can be substantially different in

supercritical water environment (Was et al., 2006). In supercritical water systems, general

corrosion and stress corrosion cracking are found to be one of the dominant corrosion

mechanisms (Was et al., 2007; Zhang et al., 2009). For general corrosion, the surface of the

alloy undergoes oxidation and forms metal oxide corrosion product. This can be of particular

concern since SCW can be an aggressively oxidizing environment if certain species or

dissolved oxygen are present, resulting in very high corrosion rates (Lister, 1980). It has been

17
found experimentally that at high temperatures, oxides are preferentially formed over

hydroxides as the primary form of corrosion product (Kritzer, 2004).

However, not all corrosion products are undesirable and the corrosion products can produce a

protective film that reduce the rate of corrosion of the underlying metal (Lister, 1980).

Corrosion of austenitic stainless steel and nickel alloys subject to supercritical water is

observed to form a multilayer oxide on the surface. These are found to be tens to hundreds of

nanometers thick after several hundred hours in a deaerated supercritical condition (Was et

al., 2007). In high temperature aqueous systems, iron ions diffuse to the outer surface while

oxygen from the solution diffuses into the metal alloy. This produces a duplex oxide

structure: a dense inner chromium-iron rich layer and a porous magnetite outer layer. The

inner chromium rich layer has a spinel structure and provides protection to the alloy while

the porous outer layer provides no further protection (Gao et al., 2007; Was et al., 2006).

Therefore, the corrosion rate is dependent on the oxygen diffusion through the inner layer

following an Arrhenius behavior in which the high temperature provides the activation

energy for diffusion to occur (Zhang et al., 2009)

2.3.4 Fouling

Another method for corrosion products to form on the surface is the deposition of corrosion

products from the bulk solution through “ex-situ” fouling. Ex-situ fouling can be further

divided into two subcategories of crystallization and particulate fouling. In crystallization

fouling, the bulk fluid is super-saturated with, for example, iron species which may have

come from other parts of the power plant where corrosion had occurred and been transported

18
to the surface of interest. Once these iron species reach the wall, super-saturation causes

particles to nucleate and deposit onto the wall. In particulate fouling, nucleation happens in

the bulk fluid and the corrosion products are carried to the wall as solid particles. Lister

(1980) hypothesized that the deposits of these ex-situ corrosion products may be the same

form as the outer layer oxide formed from the alloy. As the iron species diffuse from the

alloy to the surface, the iron first dissolves into the solution. Then, due to local super-

saturation, it re-deposits as porous outer layer resembling ex-situ fouling.

Ex-situ fouling of oxides onto heat transfer surfaces can be detrimental to power systems as

they can decrease thermal efficiency and increase the temperature of the heat transfer

material promoting earlier degradation (Lister, 1980). To track thermal efficiency issues due

to deposition, heat transfer performance is regularly monitored on steam generators in

pressurized water reactors (PWR) (Schwarz, 2001). The steam generator of a PWR is a heat

exchanger where the reactor coolant transfers the heat to the feed water / steam cycle. The

accumulation of deposits, mainly iron hydroxides and oxides was found to impair heat

transfer between the two systems. Schwarz (2001) compiled data from twelve Siemens steam

generators, many spanning over a decade and one for 27 years. Fouling was determined using

energy balance and heat exchanger equation shown below.

2-4

where Q is the thermal energy into the system, ΔTlog is the mean logarithmic temperature and

U is the overall heat transfer coefficient for the heat exchanger, calculated by equation 2.3.

19
2-5

where Rf is the fouling factor representing the additional thermal resistance due to the iron

oxide deposit. The thermal power Q was determined from the change in enthalpy of the

feedwater, derived by the inlet and outlet temperatures at constant pressure. Accuracy of the

fouling measurements was found to be highly sensitive on the accuracy of the steam

generator outlet temperature and pressure measurements. Using the heat exchanger and the

calculations described above, it was possible to observe the fouling rates on the steam

generator over its operational time. It was found that switching from phosphate-treatment to

an all-volatile treatment (AVT) with hydrazine to ensure high pH, significantly reduced the

fouling rate on the SG. The fouling factor increased gradually over time under phosphate-

treatment which occasionally would encounter a sharp drop after undergoing chemical

cleaning (Schwarz, 2001).

Experimental, simulations and theoretical studies of magnetite deposition in aqueous systems

has been the subject of many previous studies. A classic study of particle deposition was

conducted by Thomas and Grigull (1974) using magnetite powder in high temperature and

pressure system. Their experiment used an electrical heater in a flow loop system and is one

of the few studies which reached supercritical conditions of 24.6 MPa and 471°C. They

observed that the deposition of magnetite in both subcritical and supercritical single phase

flow was approximately linear to Reynolds number and concentration which is an indication

that the deposition was mass transport limited (Thomas & Grigull, 1974). More recent

studies by Turner and Klimas (2000) and Basset et al. (2000) used sol-gel methods to form

20
narrow size-distribution magnetite particles for deposition experiments. Deposition

experiments have shown the effects and differences of deposition of single phase, boiling,

and non-boiling two phase flow. Under single-phase flow, removal was not found to be a

significant factor and only significant in sub-cooled boiling (Basset et al., 2000).

Since the proposal of the CANDU SCWR reactor design, deposition of corrosion products in

the reactor core has been identified as a problem by a number of researchers prompting

simulations for deposition to be developed. Using the solubility mentioned in section 2.2,

Burrill (2000) created a simulation with a CANDU 6 fuel channel design and the temperature

went from 350°C to 384°C at 25 MPa. Ferrous ions were assumed to diffuse towards the wall

and crystallize as magnetite deposits. Deposition was assumed to be mass transport limited

and the diffusivity coefficient of ferrous species was calculated using empirical formulations

by Miller (1982). Burrill (2000) determined that after one year of operation the peak deposit

was 108mg Fe/cm2 occurring 4m from the channel inlet. The effect of dissolved iron

concentration on deposition was also investigated by setting the concentration at the inlet of

the reactor core to 18.22 µg/L and 5 µg/L. The higher inlet concentration led to deposition

nearly six times higher than the lower concentration, suggesting that the uncertainty in

magnetite solubility can have a significant impact on the simulation results (Burrill, 2000).

In a similar study, Cook and Fatoux (2009) proposed the use of steam generator such as the

one used in a PWR for the CANDU SCWR to prevent corrosion products from travelling to

the steam turbine. A simulation using the HKF model for saturation of iron was used and the

kinetics of mass transport was arbitrarily assumed to be described by constant deposition

21
velocities of 0.01 cm/s and 0.001 cm/s. The peak deposition thickness in the steam generator

was calculated to be up to 300µm or 50µm depending on the kinetic constant assumed. The

deposit decreased the efficiency of the steam generator by 4% after five years of operation

(Cook & Fatoux, 2009).

2.3.5 Corrosion Product Removal and Transport

Once attached to the tube surface, adhesive forces prevent the particle from flowing back into

the fluid stream. Adhesive forces are generally classified into three categories; The first class

of adhesive forces is intermolecular forces which can include van der Waals, the second are

chemical bonds including hydrogen bonds, and the third is sintering effects such as diffusion.

(Cooper et al., 2001).

As particles deposit, they are located in the viscous sub-layer, a very thin layer near the wall

where viscous effects are dominant and the flow is laminar (Cengel & Cimbala, 2006).

Detailed investigations of particulate removal by Cleaver and Yates (1973) have shown that

the viscous sub-layer is unsteady in turbulent flows, and that there are turbulent “bursts” that

can cause particle removal when the shear flow generate lift forces exceeding adhesive forces

of the particle and surface. In contrast, if the viscous sub-layer thickness is much greater than

the particle deposited, the viscous forces can rapidly dissipate the turbulent bursts (Cengel &

Cimbala, 2006; Turner et al., 1990). Previous studies for calculating removal rates have used

an exponential relationship (equation 2-6) and incorporated the frictional velocity U*

(equation 2-7).

22
̇ ̇ 2-6

The value ̇ is found experimentally, t is time [s] and λ is removal rate constant [1/s]

calculated using equation 2-7.

2-7

where f is the friction factor and ν is the kinematic viscosity.

While fouling of corrosion product on heated surfaces is of concern, the transport of

particulates outside the reactor core is also of great importance to the reliability and safe

operation on a power system.

Solid particle erosion (SPE) can be problematic in power generating systems where the

protective oxide layer of piping and other components are eroded by particles suspended in

the fluid stream (Crockett & Horowitz, 2010). SPE can contribute to increased maintenance

costs and the decrease in the thermal efficiency of a turbine which can be expensive (Dai et

al., 2007). For a SCWR, a direct cycle from the reactor core to the steam turbine is expected

and the problem of SPE may be enhanced due to lower solubility of oxides in supercritical

water and the formation of particulates. Another issue with corrosion products being

transported in a nuclear reactor is the transport of radioactive material. Corrosion products

can form in the reactor core, attaching to fuel cladding and pressure tubes which are then

irradiated (Burrill, 1977). Once the particle exfoliates, it can travel to other parts of the

23
system outside the core and become a source of hazardous radiation to its operators and

maintainers (Lister, 1980).

The need for in-situ particulate sampling systems can be significant for power plant

applications. Information about the properties of the particle such as size, shape, hardness

and strength can lead to information about potential damage that particulates can have on

system components surfaces (McCabe et al., 1985). The composition of the particles give an

indication as to where they may be coming from (eg. fuel cladding, piping, etc.) as well as a

predicted hardness and strength of the particles.

Furthermore, sampling at high temperature has been shown to provide better representation

of the particulates in the system. Turner and Klimas (2000) made measurements of

particulates before and after a filter and found that by placing the filter before the heat

exchanger, the concentration of iron was twice as large as when the filter was sampled cold.

Turner argued that such difference is likely due to thermophoresis where the particles diffuse

down a temperature gradient, therefore transported to the colder surface and attaching to the

heat exchanger wall. In addition, sampling ports in real power-generating systems are often

far away from the sampling line which can cause losses in the particulates found at the ports.

In PWR, it was found that 50-75% of particles were lost or dissolved in the solution due to

long sampling lines (Srisukvatananan et al., 2007)

24
3 Heat Transfer and Deposition Modeling

3.1 Overview

A heat transfer and deposition model was created in MATLAB to aid in the design of the

experiments and to predict the effects of magnetite deposition on a CANDU SCWR fuel

cladding. Only particle deposition was considered in this model due to the lack of data on the

deposition rate constant for ferrous species in supercritical water. Simulation models were

also used to compare expected deposition thicknesses with those found in the experiments.

The heat transfer and deposition model for the CANDU SCWR and the experimental

simulation are based from the same code. However, due to different geometry and flow

characteristics, adjustments to the code were made where necessary. The MATLAB program

source code can be found in Appendix A.

3.2 Heat Transfer Calculations of Water

3.2.1 Enthalpy of Fluid in the Test Section

The test section in the once-through flow loop is electrically heated and the heat flux is

assumed to be constant throughout the test section. As a result, the enthalpy of the bulk fluid

is a linear function of the distance along the tube, and at any point, the enthalpy can be

determined from equation 3-1:

3-1
̇

25
where H1 and H2 is the enthalpy [kJ/kg] at a distance Δx [m] away, ̇ is the mass flow rate

[kg/s], and q is the heat input per meter [kW/m]. The enthalpy of the inlet and outlet of the

test section is determined by using the inlet and outlet bulk fluid temperatures [⁰C] at the

specified pressure P [bar]. The bulk fluid temperatures were measured experimentally using

in-situ thermocouples before and after the test section, as described in detail in Chapter 4.

3.2.2 Heat Transfer Coefficient in Supercritical Water

Heat transfer correlations developed for subcritical conditions are not applicable for

supercritical conditions due to the dramatic changes in the thermodynamic and transport

properties near the critical region. Therefore, a number of correlations have been developed

for supercritical conditions in previous studies. Calculating non-dimensional numbers for

heat transfer generally require the use of either bulk or wall temperatures, however, there can

be dramatic changes in the heat transfer properties if bulk and wall temperatures are near the

pseudocritical region. Correlations such as those developed by Swenson et al. (1965) and

Yamagata et al. (1971) attempt to reflect these changes in the properties by using both wall

and bulk temperature to determine a suitable average. The equations provided by Swenson

are a relatively straightforward method for calculating the Nusselt number and has been

shown to predict heat transfer well, demonstrated by Teshima (1997).

( ) 3-2

26
By using equation 3-2, one can determine the Nusselt number for finding the heat transfer of

the tube to fluid. This requires the Reynolds number as well as the Prandtl number found in

equation 3-3.

̅
3-3

where ̅ is the integrated average heat capacity.

̅ ∫ 3-4

A recent study by Bazargan and Fraser (2009) on the heat transfer correlation in supercritical

water in a horizontal pipe showed that many previous correlation used were insufficient in

predicting the Nusselt number accurately. Bazargan concluded that large variations in fluid

properties near the pseudocritical zone dramatically enhanced heat transfer. A comprehensive

explanation and discussion of the approach is described by Bazargan and Fraser (2009).

Both the correlation of (Swenson et al., 1965) and (Bazargan & Fraser, 2009) were

programmed in MATLAB for comparison between the empirical models and experimental

results. These results are presented in section 4.5.1. The Nusselt number calculated from the

correlations is then used to calculate the wall temperature using equation 3-5.

3-5

To determine the fluid properties at the wall, the wall temperature is required in the previous

calculation. Since this is the parameter which is currently unknown, a wall temperature must

27
first be predicted by the software. The program runs through an iterative loop, first

calculating the wall temperature with the initial guess, calculating the new wall temperature

and modifying the initial guess until the two temperature converges to within 0.1ºC.

3.2.3 Temperature of the Oxide Layer

As the particles begin to foul the inner wall of the tube, the deposits create a heat transfer

resistance between the fluid and tubing wall interface. The temperature at the wall calculated

from section 3.2.2 is therefore the temperature of the oxide/fluid layer as deposition occurs.

The temperature of the oxide/tube layer can be calculated as follows:

3-6

where koxide is the thermal conductivity of the fouling oxide (Holman, 2002). Magnetite

thermal conductivity was calculated using the following empirical formula

3-7

where T is the temperature of magnetite (Electric Power Research Institute Inc., 2003).

3.2.4 Temperature of the Outside Wall Temperature

Calculating the outside wall temperature of the tube was necessary because the

thermocouples on the test section are measuring outside wall temperature. The tube has

internally generated heat from the low AC voltage and high current applied to the tube. The

28
equation for determining the external wall temperature with internally generated heat

assuming a perfectly insulated outside wall is shown in equation 3-8.

( ) 3-8

where A is the ratio of the inner and outer radius found by equation 3-9.

3-9

3.2.5 Buoyancy Calculations

Buoyancy effects were calculated for the experimental conditions to determine if it would

have an effect on the temperature profile of the tubing wall. Buoyancy effects can be

significant for fluids near the pseudocritical point due to large variations in the fluid density

at the bulk and wall temperatures (Bazargan et al., 2005). This can cause the low density

supercritical water to flow at the top of the tube while the denser subcritical water flows at

the bottom of the tube. Since the heat transfer coefficient varies dramatically from

supercritical to subcritical, this means that a difference in top and bottom surface

temperatures can have significant implications for a horizontal flow in tube.

Buoyancy effects were calculated according to Petukhov et al. by a threshold value for the

Grashof number, which below the threshold the buoyancy can be neglected. The threshold is

defined as follows:

29
̅̅̅ ̅̅̅ 3-10

where

̅̅̅ 3-11

The Grashof number is calculated using a heat-flux-related definition of equation 3-12.

̅
3-12

where

̅ 3-13

Evaluation is done by determining if Grq < Grth or that the Grq/Grth is less than one.

30
Figure 3-1: Calculation of Grq/Grth for determination of the effects of buoyancy along the
test section using Bazargan et al. (2005)

Since Grq<Grth for all experiments run, buoyancy should not affect the temperature of the

outer wall.

3.3 Deposition Calculations

Fouling is a complex phenomenon in which many of the parameters are unknown or specific

to the condition of deposition. In a real SCWR core, formation of magnetite may be a

combination of particles formed in the bulk solution and crystallization of magnetite on the

tube wall. In order to develop a practical simulation with limited data available, certain

assumptions for the formation of magnetite and deposition characteristics are made and

presented. For the simulation under experimental conditions, particles are assumed to be of

uniform size and the formation is assumed instantaneous at the inlet of the test section. For

31
the CANDU SCWR simulation, the particles are assumed to form instantaneously after

super-saturation.

In addition, corrosion products which foul onto the heat transfer surface can come from the

tube surface itself or from the environment. For the case of the experimental simulation, only

ex-situ fouling will be considered since the experiments were run for a short period (40

minutes) making corrosion negligible. For CANDU SCWR simulation, corrosion is again

neglected because it will be highly dependent on the material used for the fuel cladding

which is currently undecided.

There are five sequential steps in fouling; induction, transport, attachment, aging and finally

removal (Turner, 1993). For particulate fouling, the induction period is not important and is

primarily limited by transport and/or attachment. Early deposition models tended to use a

single mass flux term which depended on the concentration. A sticking probability ‘S’ was

used to take into consideration the probability of particles reaching the wall without sticking,

depending on the conditions of the particle of interest. In later models, the sticking

probability was replaced with equations of surface attachment. This is described in the Kern

and Seaton model which is most commonly used in particulate fouling (Epstein, 1987). The

deposition is said to be a two-step process; mass transport where the particle moves from the

bulk to near the surface, and an attachment process where the particle attaches to the wall.

Therefore, the deposition is the net resistance of the two processes and can be described by

the equation below.

32
3-14

where Kd, Kt and Ka is the effective, transport and attachment deposition velocity [m/s]

respectively (Basset et al., 2000).

The term is the attachment deposition velocity and is believed to be caused by van der

Waals interactions as well as the electrical double layer when the particle is approximately

10nm from the surface (Turner, 1993). Van der Waals forces arise from dipole-dipole and

dipole-induced-dipole interactions between particle and wall (Turner & Klimas, 2000). These

forces are always attractive between a particle and wall in a water medium (Hong, 2007).

The electric double layer forms at the surface of an object that is immersed in a liquid. Solid

objects generally attain a surface charge when immersed in a liquid which can be formed

through the absorption of ions present in the liquid or the dissociation of molecules on the

surface of the object (Bott, 1995). This creates a redistribution of ions in the medium in near

vicinity to the immersed particle with similar charged ions being repelled and opposite

charged ions being attracted. Therefore, a second layer in the medium called the diffuse layer

is formed by coulomb forces with oppositely charged ions. As a particle approaches the wall,

the diffuse layers of the particle and wall in the solution overlap and create a potential

energy. If the charges are oppositely charged, the forces are attractive and the limiting rate is

the mass-transport velocity. If the charges are the same, the force is repulsive and depending

on the magnitude of the repulsive force, the overall deposition may be surface attachment

limited (Basset et al., 2000). In such case, the particle that approaches the wall will have to

33
overcome this potential energy barrier to attach to the wall. This follows an Arrhenius

behavior and can be described as:


3-15

where E is the activation energy, R is the gas constant, A is an experimentally determined

factor and Ts is the temperature of the surface (Turner, 1993).

When deposition is mass transfer controlled, Ka is much larger than Kt and the mass transport

velocity is approximately equal to the effective deposition velocity.

3-16

For deposition onto the inside of a tube, the mass deposition rate along any section of the

tube is then determined by:

̇ 3-17

where is the density of the particle fouling the surface. The deposition velocity is

calculated from equation 3-18.

3-18

where U* = frictional velocity and determined by:

34

( ) 3-19

where is the fluid density and is the wall shear stress and calculated by:

3-20

The friction factor f in a fully developed turbulent pipe flow can be calculated using the well-

known Colebrook equation. In many cases an approximate but explicit relation can be

substituted for the Colebrook friction equation using Haaland’s equation shown in the

equation below.

( ( ) ) 3-21

where is the roughness factor which is was taken to be 0.002 mm for a clean stainless steel

tube. The uncertainty in such a roughness factor can be as much as ±60% (Cengel &

Cimbala, 2006). As the tube begins to foul onto the surface, it is assumed that the roughness

of the tube was then the sum of the deposit and initial roughness:

3-22

The non-dimensional deposition velocity Kd+ was predicted from empirical equations fit by

Papavergos and Hedley (1984) for aerosol particles landing onto a flat plane. Experimental

work done by Teshima and Khan showed that these equations provided agreement for salt

deposition in supercritical water (Khan, 2005; Teshima, 1997).

35
( ) 3-23

Kd+ depends on both the dimensionless Schmidt number as well as heavily on the particle

relaxation time. The Schmidt number is given in equation 3-24.

3-24

where Sc = dimensionless Schmidt number

µ = dynamic viscosity of the fluid [Pa∙s]

D = diffusion coefficient [m2/s]

The diffusion coefficient was determined from the Stokes-Einstein equation:

3-25

where bk = Boltzman’s constant [J/K]

µw = dynamic viscosity of fluid [Pa∙s]

Tw = temperature of fluid [⁰C]

Finally, the non-dimensional parameter relaxation time is calculated using the following

formula (Turner, 1993):

36
( ) 3-26

The regions of Kd+ can be categorized into three regions depending on the relaxation particle

time. For , the particle motion is dominated by Brownian diffusion and deposition is

inversely proportional to the size of the diffusing particle (Beal, 1970). As the particle size

increases, the relaxation particle time increases and momentum effects gain importance. The

deposition velocity reaches a minimum when is 0.2 and inertial coasting becomes the

dominant mechanism for transport which increases exponentially with diameter size. At even

higher relaxation particle times above 20, the particle inertia becomes very large and the

stopping distance exceeds the viscous and buffer layers. The stopping distance marks the

distance at which the particle will coast to the wall simply by their momentum. In this region,

the particles are no longer affected by the fluid forces, therefore reaching a constant value

(Chen & Ahmadi, 1997).

Figure 3-2 was produced using the equations described above and shows the effects of

particle diameter and temperature on deposition velocity under the experimental conditions.

37
Figure 3-2: Effect of particle diameter and fluid temperature on deposition velocity for
experimental apparatus

3.4 CANDU Supercritical Water Reactor Simulation

To predict the rate of deposition that might occur in a CANDU SCWR, the design of a

CANDU ACR-700 was used as a baseline since the design for a SCWR pressure tube has yet

to be finalized. The CANFLEX 43-rod fuel bundle configuration which is used in the ACR-

700 provided geometric parameters for the heat transfer surfaces and is listed Table 3-1.

38
Table 3-1: SCWR and fuel assembly design parameters

Coolant pressure (MPa) (Naidin et al., 2009) 25


Inlet temperature (⁰C) (Naidin et al., 2009) 350
Outlet temperature (⁰C) (Naidin et al., 2009) 625
Maximum cladding temperature (⁰C) (Naidin et al., 2009) 850
Coolant total flow rate (kg/s) (Naidin et al., 2009) 1320
Number of fuel channels (Naidin et al., 2009) 300
Heat flux (kW/m2) (Li et al., 2009) 1000
Fuel elements diameter (mm) (Li et al., 2009) 12
Fuel elements length (m) (Li et al., 2009) 6.0
Pressure tube diameter (m) (Li et al., 2009) 0.10

As an approximation of a real fuel bundle which has complex geometry, the simulation

developed in this study was conducted by looking at a single fuel element and approximating

the fluid and thermal properties around the element without the influence of neighboring

elements. Temperature rise of the coolant was predicted by setting the fuel cladding wall as

the single source of heat to increase the enthalpy of the coolant linearly along the 6 m

distance.

For the hypothetical single fuel element developed in this study, the fuel bundle was

converted into an equivalent heated tube with the coolant flowing on the inside, and therefore

the MATLAB code used for this simulation is nearly identical to the simulation used for the

experiment with the exception of the parameters. The mass flow rate for a single fuel element

was calculated by the total flow rate in the reactor divided by the 43 fuel elements. An

effective diameter for a single fuel element was calculated by keeping the velocity of the

fluid the same as the velocity in a full scale fuel bundle.

39
Pressure Tube Cladding
Coolant
Tbulk
Toxide
Tcladding

CANFLEX fuel bundle Equivalent tube

Figure 3-3: Cross-sectional view of CANFLEX 43 rod fuel bundle and drawing of
equivalent tube for simulation calculations (not to scale)

The surface most interesting for these deposition studies is on the fuel cladding which covers

the fuel pellets and prevents the uranium from coming into contact and contaminating the

water. The fuel cladding is often made zirconium alloys due to its low neutron absorption

property compared to other alloys. However, zirconium alloys are limited by their reduced

mechanical properties at high temperature and therefore may require other alloys as a

substitute for the fuel cladding for future SCWR. Results from the simulations are reported in

Chapter 6 and discussed considering the measurements described in the next few chapters.

40
4 Experiments

4.1 Supercritical Water Once-through Flow Apparatus

A once-through flow apparatus shown in the Figure 4-1 was operated for all high temperature

experiments reported in this study. The apparatus was originally constructed for running

experiments for supercritical water oxidation and therefore, system components were added

and/or modified to run the system for corrosion product transport experiments and simulate

supercritical water reactor conditions. A schematic of the entire system can be found in

Figure 4-2.

Back pressure
regulator

Control Panel

Effluent tank

Injection tank Injection pump

Figure 4-1: SCW system used in all experiments located in the Mechanical Engineering
laboratories.

41
Nitrogen Nitrogen

Primary
Tank

Relief
Valve

P-1 T1 T2
Preheater 1 Preheater 2
Nitrogen

Pump 1

Back P-2 T-C Tb1


Tb2 Injection
Pressure
Primary Heat Exchanger Test Section Preheater 3 Tank
Regulator

LTHP
Effluent Tank Filter
Secondary Line Heat
Exchanger
Pump 2

LTLP HTHP
Effluent Tank Filter Filter

Figure 4-2: Schematic of supercritical once-through flow apparatus

42
The majority of the system tubing is constructed from 1/4” seamless AISI stainless steel 316L.

Tube fittings for the system use SS316 Swagelok fittings where contact with high temperature

and pressure water is present. A 20L plastic tank and a glass flask were used for the primary

feedwater tank and the injection solution tank respectively. It was important to keep oxygen

content low in the system favoring reducing conditions, and therefore the tanks were deaerated

by vigorously bubbling nitrogen into the water.

A high pressure Bran Leubre positive displacement pump was connected to the primary tank for

delivering the majority of the fluid. Flow rates are manually controlled and can attain up to 8L/hr

at a pressure of 25.0 MPa. To suppress the fluctuations in the pressure due to the positive

displacement pump, a pulsation damper (Flowguard DS-10-NBR-A) pre-charged with nitrogen

was installed near the Bran Leubre pump.

Three ceramic preheaters with a combined total wattage of 8.6kW are located downstream of the

main pump which externally heat the fluid to the desired temperature before entering the test

section. The ceramic heaters are split into top and bottom halves allowing for easy access to the

tube during cooling and setup.

An Eldex Laboratories high pressure pump (BB-4-VS) injects the ferrous precursor solution into

a flow-tee which joins the injection stream to the hot fluid stream. The pump has two heads, both

taking fluid from the injection tank and reconnecting through a tee at the outlet. The pump has a

manually adjustable flow rate using a Bodine Electric Company, RPM controller. To ensure

complete thermal and chemical mixing of the two fluids, a perforated stainless steel 304 was

43
placed inside the tube immediately prior to the bulk thermocouple. A detailed description of the

test section and the temperature measurements are given in section 4.7.

After the test section, a final tee splits the stream towards a secondary line consisting of a high

temperature, high pressure filter and the primary line connecting directly into a heat exchanger.

The secondary line has its own heat exchanger and a needle valve to cool and depressurize the

fluid to room temperature and ambient pressure. A low temperature, low pressure filter is located

after the needle valve in the secondary line.

On the primary line, a 2.0 m long tube and shell heat exchanger cools the hot supercritical fluid

flowing in the tube side with cold water flowing in the shell side. After the primary heat

exchanger, a low temperature high pressure filter and a back pressure regulator (TESCOM 26-

1762-24) are located immediately downstream to adjust the pressure of the system

4.2 Flow and Temperature Control

4.2.1 Safety

The safety of the operator, as well as other laboratory personnel was considered of utmost

importance. Hence, online temperature and pressure monitoring with automatic shutdown control

was put into place. Temperature was monitored at five locations throughout the system to keep

temperatures below tube temperature ratings. If temperatures of the heated section were to

exceed 500⁰C, the controllers are programmed to trip the breaker causing all heaters to turn off

automatically while the primary pump remains online to remove residual heat from the system.

44
Pressure is also monitored by the controller and can turn off both the heaters and the pump if the

pressure exceeds 34.5 MPa, signaling a possible plug in the system due to fouling. In addition, a

pressure relief valve just downstream of the pump was set to open at 27 MPa.

If temperatures at the exit of the heat exchanger begin to rise, this may indicate the cooling water

may be off or have insufficient flow rates which may result in high temperature water/steam

exiting the system and potentially damaging downstream equipment. The controller was set to

automatically turn off the heaters if the temperature of the tube after the heat exchanger exceeded

50⁰C.

4.2.2 Pressure Measurements

Pressure was monitored with a GP-50 pressure transducer (P-1 in Figure 4-2), calibrated by the

manufacturer with an accuracy of ±2% full scale output (FSO) of 34.5 MPa. These were read and

recorded manually from the control panel board. A Bourdon pressure gauge (P-2 in Figure 4-2)

downstream of the primary heat exchanger was also frequently checked as a safety measure.

4.2.3 Flow Rate Measurements

The primary pump was first turned on and pressurized to the desired pressure. Flow rates of the

effluent were measured three times and averaged using a graduated cylinder and a stop watch.

Then the injection pump was turned on and after several minutes the flow rate was measured

once again. The calculated difference between the two averaged flow rates was the injection

pump flow rate.

45
4.2.4 Fluid Conductivity Measurements

An Omega CDH-287 micro-conductivity meter was calibrated using a CDE3052 conductivity

standard (1413µS/cm). The conductivity meter has an accuracy of ±0.3% FSO. Conductivity of

the inlet of the primary tank was measured to check that the deionized water had low

conductivity indicating ions are not present. The conductivity meter was also used for high pH

experiments for measuring sodium hydroxide content in the tank. A detailed description of this

calibration and use for high pH experiments is explained in section 5.5.3.

4.2.5 Dissolved Oxygen Concentration Measurements

An Omega DOB21 measured the dissolved oxygen (DO) content in the two feedwater tanks. The

DO has a relative accuracy of 0.02% FSO, with a minimum resolution of 0.01ppm.

4.3 Particle Synthesis for Modeling Power Plant Fouling

Due to the low solubility of magnetite, deposition of magnetite from a dissolved state was

expected to be slow, taking several years to form measurable deposit thicknesses. To create a

feasible experiment, it was necessary to introduce ferrous species artificially. To simulate ferrous

corrosion products, ferrous salts were mixed into deionized water and injected into the system.

The ferrous salts then form iron oxide particles after the process described here.

Formation of magnetite particles has been of interest in colloidal science for its industrial

applications, as well as through its presence in the power generating industry as corrosion

products from iron and steel (Zhao et al., 2009). Work of Sugimoto and Matijevic (1980)

46
introduced a procedure of producing spherical magnetite particles using ferrous hydroxide gels.

Later work by Adschiri et al. (1992) reported a rapid hydrothermal process using supercritical

water to form sub-micron metal oxide particles in a continuous production. Hydrothermal

processes for producing oxide particles became a popular method as it offered control over

morphology and composition depending on the precursor, temperature and density of the fluid.

Under supercritical conditions, reaction kinetics is very high and the formation of metal oxide

particles occurs rapidly due to its low solubility. In previous studies, Fe(NO3)3, Fe2(SO4)3, FeCl2

and Fe(NH4)∙2H∙(C6H5O7)2 were used as precursors in forming various iron oxides in

supercritical water. Fe(NH4)2H(C6H5O7)2 formed spherical magnetite (Fe3O), while all others

formed spherical hematite (α-Fe2O3).

Both in the sol-gel method and rapid hydrothermal process, the metal oxide goes through two

reactions; first the metal salt undergoes hydrolysis in which the salt will react with water to form

a hydroxide and second, a dehydration process where the hydrogen is separated to form an oxide.

Dehydration occurs from the outer surface of the hydroxide particle and therefore smaller

particles may favor faster dehydration rates (Adschiri et al., 1992). These reactions are given in

equations 4-1and 4-2 respectively.

4-1

4-2

In equation 4-1, B is the anion of the ferrous salt which is either Cl- or SO4- in this study. Due to

excess hydrogen ions, the solution is slightly acidic and the salt is said to be a lewis acid.

47
At higher temperatures, dehydration occurs in which the iron is oxidized from Fe2+ to Fe3+. This

forms magnetite if the environment is anaerobic while forming hematite or other hydroxides in

an aerobic environment. This is known as Schikorr reaction and is a common reaction which

takes place in boiler tubes forming magnetite (Shreir et al., 1994).

4-3

A residence time of approximately 2 minutes yielded 50 nm diameter hematite particles from

Fe(NO3)3 in supercritical water. Adshiri (1992) also tested reaction rates and found that it formed

hematite nano-particles even when the reactor residence time was cut down to less than 1 second

although particles were smaller (<20 nm) suggesting that reaction times are very high (Adschiri

et al., 1992). Synthesis in supercritical conditions occur faster partly due to the lower dielectric

constant of water and the resulting enhancement of the reaction rate according to Born’s theory

(Adschiri et al., 2000).

Although Aschiri had used ferrous chloride in his experiment, it is believed that the system still

contained oxygen, making it a highly oxidizing environment under supercritical conditions. This

would explain why hematite was formed in Aschiri’s experiment with ferrous chloride.

4.4 Experimental Procedure

4.4.1 Experiments with Precursor

Before each experiment, the primary pump was turned on and the system was flushed with

deionized water for about one hour. Both the primary and injection tanks were continually

48
sparged with nitrogen throughout the experiment to deaerate the system. The dissolved oxygen

sensor was placed in the injection tank to deaerate the solution until it reached 0.00ppm which

usually took one to two hours. Once the dissolved oxygen content reached the desired level in

the injection tank, the DO sensor was moved to monitor dissolved oxygen levels in the primary

tank for the remaining part of the experiment.

The back-pressure regulator was adjusted until the pressure at the transducer was reading the

required pressure for the particular experiment. Although the pressure transducer is located

before the heaters and several feet away from the test section, it is assumed to that the readings

are representative of the pressure in the test section as the pressure drop between the two points

is small and negligible compared to the system pressure of 24 MPa. Flow rates of the primary

pump were measured and adjusted according to the required 3.0L/hr. The injection pump

connected to a tank of deionized water was turned on and the new flow rate was adjusted to

3.5L/hr using the injection pump. The pressure remained constant during this time and the

injection pump flow rate was calculated as the difference between the two flow rates. Once the

desired flow rate and pressure were achieved, the heaters were turned on and allowed to heat the

system to supercritical conditions. Temperatures of the bulk fluid inlet and outlet of the test

section were checked until the system reached steady state temperature. The injection pump was

momentarily turned off in order to switch the injection tank from deionized water to precursor

solution. This point would be observed in the temperature measurements by a momentary

increase in the temperature profile of the test section due to reduced flow rates. Once the tank

with precursor solution was connected, the injection pump was turned on once again and the

temperature of the test section dropped to its original temperature. The flow rate at the exit was

49
measured once more to check if they were consistent from before. Each of the experiments ran

for approximately 40 minutes, with the exception of the low concentration ferrous chloride

experiment which lasted for 400 minutes at one tenth the concentration. Therefore, the total mass

of the salt injected was kept constant in all experiments.

For low heat flux, ferrous chloride experiment, a 1.0m, 1/8” x 0.028” outer diameter stainless

steel 316L tube at 6V was used rather than the 1.8m at 12V used for the rest of the experiments.

It was realized during this experiment that there was significant electrical resistance in the wiring

and heat loss through the mounting blocks such that the temperature rise in the test section was

very small. Therefore, for all subsequent experiments, the test section was modified to run at

1.8m and 12V yet maintain a maximum current under design ratings of 200A.

4.4.2 Experiments with Magnetite Particles

Experiments with magnetite particle injection was also initially tested using commercially

available magnetite powder by Sigma Aldrich (Iron (II,III) oxide, > 98% Fe3O4). The use of

prepared magnetite powder for the experiments offered several advantages over using a

precursor; the amount of magnetite particles in the system could be accurately known without the

need to determine yield rates and the size of the particles could also be externally controlled. The

magnetite powder had a diameter of 50 nm or less and was suspended in deaerated solution in

the injection tank. The solution was continually mixed with an electrically powered mixer to

prevent particle settlement. The magnetite particles were observed to agglomerate and stick to

the glass flask and tubing, lowering the overall concentration of magnetite powder in the

solution. To mitigate this problem, the pH of the solution was increased to 10 using sodium

50
hydroxide where the zeta potential of the solution/magnetite favored dispersion. However, the

HPLC injection pump continued to experience difficulty pumping at consistent concentration of

magnetite nano-particles, therefore further experiment with magnetite slurry was not pursued.

4.4.3 Post Experiment Cleaning

After each experiment with the exception of experiments 7 and 8, the test section was removed

and replaced with a temporary tube and the system was chemically cleaned. For experiments 7

and 8, both the test section tube and the rest of the system underwent chemical cleaning. The

system was cleaned using 1N hydrochloric acid at 80⁰C for one hour before each test to ensure

that magnetite deposits were removed from the system. After cleaning with hydrochloric acid,

the entire system was flushed out with de-ionized water for several hours. When required,

mechanical cleaning of tubes and fittings was also conducted by disassembling the sections and

cleaning the section with an ultrasonic cleaner and/or a mechanical brush.

4.5 Analytical Methods

4.5.1 Online Thermal Resistance Monitoring

The electrical heater was designed and constructed to serve several purposes. First, the heaters

increase the temperature from subcritical to supercritical temperatures, allowing for a wide range

of temperatures and heat flux conditions which can be tested. Second, it offered an online

monitoring of fouling in the system. Deposits on the heat transfer surface would increase thermal

resistance which would be detected by an increase in temperature of the tube outer wall. By

modeling magnetite deposits and heat transfer characteristics, the deposition and removal rates

51
can be calculated. Calculated values can be compared to the deposition found in SEM images

and cleaning methods.

The test section is electrically wired with current running directly through the tube causing the

electrical resistance of the 1/8” tube to dissipate power in the form of heat. It is assumed that the

electrical resistivity remains constant along the tube and therefore the heat flux into the water

will also remain constant along the tube. This method for heating the tube was chosen over

heating tapes and ceramic heaters for its uniform heat flux demonstrated by Teshima (1997) and

Khan (2005). The test section tube is connected to a 2400VA transformer with its primary coil

wired in series and secondary coil wired in parallel using 6V center taps. Power input was

provided by two of three legs of a three-phase delta-configured power supply, which supply a

voltage of 208V RMS to the transformer. Fuzzy Pro logic controllers monitored the temperatures

for both the preheaters and test section. These have PID control which read temperatures from

the thermocouples and send a signal for optimal power level to the solid-state rectifiers (SSR)

which act as a gate for the electrical current.

Metal oxide varistors (MOVs) were installed both at the inlet before and after the SSR to prevent

voltage spikes from damaging the SSR and transformer. The transformer connects to the tube

with one connection to the center and two ground wires to each of the ends. This is to prevent

any ground loops from forming within the system, which may create safety hazards. The

resistance of the 1/8” SS316L tube was measured to be on average ~0.13 Ω per meter, however

the effective resistance is reduced to a quarter of its value due to the parallel connection.

52
Thermal expansion of the tubing was also considered which can be as high as a few millimeters

from room temperature to experimental temperatures. The heat exchanger is free to move axially

a short distance, allowing thermal expansion to occur without introducing significant stress on

the test section.

For both safety and protection of the equipment from voltage on the tube, secondary grounding

wires were attached at the inlet of the injection line and after the heat exchanger. Three tube

mounting blocks were constructed from SS303 for clamping the tube and the electrical conductor

together. These mounting blocks were specially designed and optimized to offer high thermal

resistance to decrease heat loss from the test section yet maintain high electrical conductivity for

the electrical path. Aluminum heat sinks were attached to the mounting blocks to dissipate any

excess thermal heat. The configuration of the test section and mounting blocks are provided in

Figure 4-3.

53
From main pump

Test section
From
injection
pump

Wall temperature
thermocouple Mounting
blocks
Tee with inlet
bulk
thermocouple

Heat exchanger

Figure 4-3: Mounting block and test section inlet

Temperature along the tube was measured using K-type Chromel/Alumel thermocouples with

twisted shield for protection against electrical noise. Although not as accurate as RTD, they are

very robust even at high temperature and can be calibrated to be accurate within less than 1⁰C.

Six thermocouples were connected to the control panel for input power control and safety

reasons. Eleven thermocouples were spot welded to the 1/8” test section tubing on the outside

wall. These thermocouples were placed on the top of the tube, however, since buoyancy is

negligible in these conditions, the location around the circumference of the tubing should not

have a significant impact on the temperature reading. Thermocouples were clamped onto the

54
outside wall of the outlet of heaters 1 and 2 and the tube downstream of the heat exchanger of the

main line. Bulk fluid temperatures were measured using Inconel sheathed, ungrounded probes

which were inserted at the Tees with the probe end approximately at the center of fluid flow. The

outlet bulk temperatures were fed into the control panel and were used to control the temperature

of the electrically heated section. A wiring and thermocouple placement diagram is shown in

Figure 4-4.

AC voltage and current into the heated tube sections were measured for calculating power input.

Although these measurements are not used for determining the enthalpy of the water, it provided

a secondary confirmation for ensuring temperature rise corresponding to the electrical power

provided by the system. A fluke clamp meter was used to measure AC current with resolution of

±0.1A. Measurements of the voltage across each half-section were performed with a Mastercraft

multimeter with a resolution of ±0.01V.

Input power calculations derived from current/voltage measurements closely matched input

power calculations using bulk temperature measurements, shown in the table below as the

electrical and thermal power respectively. The small discrepancy between the two values is

likely due to thermal loss to the mounting blocks and surrounding insulation, hence the electrical

power is slightly higher than the actual heat gained by the fluid.

Table 4-1: Input power comparison

First half of tube Second half of tube Power [W] Power [W]
Experiment
Current [A] Voltage [V] Current [A] Voltage [V] (electrical) (thermal)
1 54.3 10.00 51.5 9.93 1054 1025

55
Calibration of the thermocouples was conducted by measuring the outside wall temperatures

with the bulk inlet and outlet temperature of the test section set at a saturation temperature and

pressure, and the section was thermally insulated with no input power. This process was

completed before the experiment each time a new test section was installed. Temperature was

recorded for thermocouples and any significant temperature differences were noted. For the

thermocouples, the temperature measured usually fell within one or two degrees of expected wall

temperature. These were used to correct the temperature readings for any future measurements

for each experiment.

Subsequently, the accuracy of the temperature profile calculated by the MATLAB simulation

fluidproperties.m was checked by comparing the simulation to experimental temperature profile

with inlet and outlet temperatures at 350⁰C and 400⁰C respectively. The measured and calculated

temperatures using correlations of Swenson and Bazargan are shown in Figure 4-6.

56
Omega 1200 High Speed
Isolated Measurement System

Flow

15 cm

Mounting
blocks

Parallel

Transformer

Series

SSR
Fuzzy Pro logic Fuzzy Pro logic
controller controller

Figure 4-4: Electrical wiring, heater controls and location of surface welded thermocouples and bulk fluid thermocouples

57
Figure 4-5: Saturation temperature of 231.51ºC at 417psi, calibration for Experiment #1

Figure 4-6: Temperature of test section, calculated vs. calibrated thermocouples

58
It was observed from the comparison that Swenson’s correlation predicted temperatures well

both in subcritical and supercritical conditions. Bazargan’s correlation over-predicted heat

transfer enhancement at bulk fluid temperatures above 380°C. Therefore, all subsequent

simulations presented in this study use Swenson’s correlation for determining the heat transfer

coefficient. The temperatures read were quite stable (std = 0.7⁰C for experiment #1) and

therefore, any temperature differences due to fouling can be measured with an accuracy to 1⁰C.

Temperature measurements of the test section thermocouples were recorded using a data

acquisition hardware and computer. The Omega Multiscan 1200 has 24 electrically isolated

channels and two internal channels which are used for cold junction compensation. The 16-bit

temperature measurements offer 3.12 uV or 0.1⁰C resolution. Electrically isolated channels were

needed for the temperature measurements since the thermocouples were welded and electrically

connected to parts of the test section which are at various voltage potentials. AC line rejection

and averaging helped to reduce noise caused by the AC voltage applied to the test section in the

measurement. Temperature data is sampled at 20kHz frequency and the data was sent to

“TempView” software, displaying real-time temperature measurements. The measurements were

automatically averaged over a 5 second period and then outputted to the text file which was later

analyzed in MATLAB.

4.5.2 SEM Imaging for Filters and Tube Deposits

A Hitachi S-3000N in the Frank Forward Materials Engineering building at UBC was used for

SEM imaging, with the exception of those photographs which are labeled as Hitachi 2300 SEM,

also located in the Frank Forward building.

59
To observe the morphology of the deposit on the tube inner-wall surface, the test section was cut

into 1.5cm long samples using a 1/8” tube cutter. The sample was then carefully cut in the

middle of the tube by grinding halfway through the tube until there was exposure to the other

inner wall. Then the tube was bent slightly to open the section and the sample was placed into a

mounting clamp for SEM imaging.

Figure 4-7: Test section sample for SEM analysis

4.5.3 Particulate Filtering System

Three filters were constructed and installed at various locations in the system to collect particles

formed in supercritical water. After each experiment, the filters were removed and analyzed with

SEM/EDX with the objective of determining particle size, shape, and composition.

The main body for the low temperature, high pressure (LTHP) filter holder shown in Figure 4-8

was constructed from stainless steel 316L. The filter was designed and constructed with the

60
intention of allowing future filter holders to be constructed from any material and allow for easy

assembly and reassembly after each experiment (see Appendix B for details). The metal spiral

wound sealant allows the filter to reach high temperature and pressure applications and

compensates for uneven or scratched surfaces of the mating parts. Glass fiber filters were

inserted into the LTHP filter to collect particles in the primary line after the heat exchanger as

well as to protect the back pressure regulator which was sensitive to particulate fouling. After

each experiment, the glass filter was carefully removed, left to dry overnight, and was gold

coated using a gold sputtering machine for SEM/EDX analysis.

Figure 4-8: SS316L LTHP filter installed after the heat exchanger

61
A high temperature, high pressure (HTHP) holder was made from Swagelok 316L fittings shown

in Figure 4-9. A stainless steel 304 perforated metal with a porosity of 22% and a hole diameter

of 0.02” was fitted into the holder to provide back support for the filter. Glass fiber filters were

initially tested in supercritical water, however, the filters were destroyed when exposed to

supercritical water. It is thought that at high temperature and pressure, the glass fiber mechanical

properties degrade significantly and the pressure drop across the filter may have caused it to

break down. It was decided that the glass fiber filters would be used to collect particles only at

low temperatures. In replacement of glass fiber filters, sintered silver membranes from SPI

supplies with absolute retention pore size of 0.2 µm and thickness of 50 µm were used for high

temperature applications. The thin filter makes it ideal for higher flow rates and the conductive

properties of the silver membrane make it excellent for SEM/EDX and XRD analysis. The filter

holder and its components are shown in Figure 4-9.

Figure 4-9: HTHP filter with silver membrane and SS304 back support

62
Finally, a polypropylene 25mm Swin-Lok filter holder was used for the low temperature, low

pressure (LTLP) filter shown in Figure 4-10. The same glass fiber filters used in the LTHP filters

were also used for the LTLP filter. A summary of the properties of the glass and sintered silver

filters are described in Table 4-2.

Figure 4-10: Glass fiber membrane and Swin-Lok LTLP filter holder

Table 4-2: Glass and silver membrane filter comparison

Commercial Effective Pore Thickness Water Flow


Supplier size [µm] [µm] [mL/min/cm2] Temp Max
Sintered Silver SPI Supplies 0.2 50 17 550
Glass Fiber Millipore 0.7 380 1.4 500

SEM and EDX analysis were taken of clean filters to determine its composition, and therefore

providing information about the background of the filters. This helps to identify which elements

63
found on the filter are coming from the particle and which are the background of the filter. The

silver membrane was found to be composed of only silver and oxygen. The glass fiber filter had

Si, O, Ca, K, Ti and Zn.

64
Figure 4-11: SEM photograph of clean silver membrane, 0.2µm pore size, using a Hitachi
S3000N SEM

Figure 4-12: EDX of silver membrane to determine elemental composition of background

65
Figure 4-13: SEM photograph of clean glass fiber filter, 0.7µm pore size, a Hitachi S2300 SEM

Figure 4-14: EDX of glass fiber filter to determine elemental composition of background

66
4.5.4 Deposit Thickness and Strength of Oxide Adhesion

To quantify the mass of magnetite deposited on the tube, two techniques for magnetite removal

was used; mechanically using an ultrasonic cleaner and chemically using acid wash procedures.

5.0 cm sample tubes were cut at 30.0cm intervals along the test section for magnetite deposition

quantification. For ultrasonic cleaning, the sample tube was immersed in 25.0mL of 1N

hydrochloric acid at room temperature and cleaned in a Cole-Parmer ultrasonic cleaner for 20

minutes. The sample tube was then removed and the solution was heated on a bunsen burner

with a watch plate and evaporated in a fume hood. Once the beaker was completely dry, it was

washed with 25.0mL of 1% nitric acid solution. 2.0mL was taken from the sample and diluted in

another sample container of 25.0mL, 1% nitric acid. This was repeated once more to achieve

three sets of sample solutions which have of 1×, 13.5× and 182.25× dilution ratios.

For acid wash, the sample tube was immersed in 15.0mL of 1N preheated hydrochloric solution

(≈75°C) for 2 minutes. The sample tube is then removed from the solution and the solution is

further heated to evaporate it completely. The beaker was then washed using the same procedure

as the ultrasonic cleaning method.

A Varian AA240 Atomic Absorption Spectroscopy (AAS) was used to analyze iron

concentrations in the samples. AAS uses measured absorbance to determine analyte

concentrations according to the Beer-Lambert law. Calibration of the AAS was conducted by

first measuring the absorbance using standards of known concentration. For these experiments,

the calibration was done with 0.0, 1.0, 2.0, 5.0, and 7.5mg/L of iron standards in 1% nitric acid

67
matrix. All sample solutions were also digested in a 1% nitric acid matrix. The detection limit of

atomic absorption spectroscopy for iron is approximately 0.1 mg/L (Willard et al., 1969).

Since stainless steel 316L also contains iron, there would be dissolution of iron from the tubing

itself and not just the deposited magnetite. To account for this, clean tubes without deposits were

exposed to the same conditions as those of the sample tubes and its iron content was measured

using AAS. This was done for both the ultrasonic cleaning and acid wash procedures and the

resulting iron concentration from the clean samples were subtracted from the raw sample tube

concentration.

A mass balance of the entire system was pursued by measuring the concentration of ferrous

species in the effluent, filters and test section. Effluent concentrations were determined by taking

20.0mL samples and analyzing them using Atomic Absorption Spectroscopy in a similar manner

to the test section; first by dissolving any possible magnetite that might have passed through the

filter, evaporating the solution, and re-dissolving in 1% nitric acid matrix. For experiments with

ferrous chloride, the effluent samples were taken during the experiment directly from the system

effluent. It was determined that due to changing deposition/exfoliation rates over time, an

averaged effluent concentration would be more representative for mass balance purposes.

Therefore for all ferrous sulfate experiments, the effluent was stored in the high-density-

polyethylene (HDPE) tank until the experiment was completed and a sample was taken of the

total effluent. In addition, for all ferrous sulfate experiments, sulfate was known to interfere with

the AAS in measuring iron, therefore as a way of mitigating this, a copper sulfate of 0.1 mol/L

and 1% nitric acid matrix was used for these standards and sample solutions.

68
5 Results and Discussion of Experiments

A total of eight experiments and one blank were conducted using the SCW once-through flow

apparatus. After each experiment, filters, effluent, and tube samples were analyzed using SEM

and AAS to determine significant findings. A brief summary of the results is provided in Table

5-1 while a more detailed overview and collection of all results can be found in Appendix C.

The two precursors used were ferrous chloride tetrahydrate (FeCl2·4H2O) from Fisher Scientific

and ferrous sulfate heptahydrate (FeSO4·7H2O) from BDH. The experiments are broken down

into the following categories: reference condition, effect of heat flux, high pH, low concentration

and blank run with no precursor.

Table 5-1: Summary of experiments

Concentration
Tube Heat [mmol Fe2+/L]
Pressure Bulk Temperature
Run Category length Flux pH
[MPa]
[m] [kW/m2] Dissolved
Inlet [⁰C] Outlet [⁰C] Precursor
Fe2+
1 Reference 23.7 1.8 350 400 102 3 5.48 FeCl2·4H2O
2 Reference 23.9 1.8 350 397 97 3.5 4.97 FeSO4·7H2O
3 Heat Flux 22.4 1.0 372 376 24 3 5.31 FeCl2·4H2O
4 Heat Flux 23.0 1.8 371 376 13 3.5 5.04 FeSO4·7H2O
5 Heat Flux 23.3 1.8 384 383 - 3.5 5.32 FeSO4·7H2O
6 High pH 23.8 1.8 350 396 97 9 5.11 FeSO4·7H2O
7 High pH 23.7 1.8 200 370 92 9 5.17 FeSO4·7H2O
8 Concentration 23.7 1.8 350 395 96 3 0.56 FeCl2·4H2O
- Blank 23.7 1.8 351 399 99 Neutral 0.00 -

69
5.1 Reference Condition

Experiments were run with ferrous chloride and ferrous sulfate precursors with the test section at

maximum heat flux of approximately 100 kW/m2. These were considered to be reference

experiments that all other experiments could be compared to. Temperatures of the bulk fluid inlet

and outlet were 350°C and 400°C respectively at 23.7MPa. The injection tank consisted of 1L of

precursor in deaerated, deionized water while the primary tank had deaerated, deionized water,

producing an acidic solution (pH≈3) once both fluids were mixed. The injection method is

similar to those described by Adschiri et al. (1992) in hydrothermal synthesis.

5.1.1 Temperature Analysis

The transient temperature of the test section outer wall for reference condition with ferrous

chloride precursor is shown in Figure 5-1. As mentioned in Chapter 4, the large spike and

subsequent fall in the temperature at 15-20 minutes is due to the momentary shutoff/on of the

injection pump for switching the supply line from deionized water to ferrous chloride solution.

At subcritical temperatures, the temperature remain constant throughout the experiment. At

supercritical temperature (1.65m from the inlet), the wall temperatures increase rapidly at the

start and reaches an asymptotic temperature after 10 minutes of injection.

70
Injection

Figure 5-1: Temperature measurements of test section vs. time for reference condition with
ferrous chloride precursor (Experiment #1)

Fouling was observed by the sharp rise in temperature as well as an increase in system pressure

to 26.2 MPa during the injection period as a result of restricted flow. The temperature profile at

the 1.65m location has an asymptotic fouling behavior similar to those found in several studies of

iron oxide fouling (Müller-Steinhagen et al., 1988; Newson et al., 1983; Thomas & Grigull,

1974). It was speculated that the removal rate increases as the deposition thickness increases,

since the bond at the top layer of the surface weakens as the deposit grows higher and is more

likely to be removed by the shear force of the fluid (Newson et al., 1983). Another explanation

presented by Thomas and Grigull (1974) suggested that the initial high deposition rate was due

to the particles adhering to the troughs in an initially rough tube. As deposition continued, the

71
troughs would be filled and the tube surface would become effectively smooth and more difficult

for the particles to adhere to.

High fouling rates observed at the 1.65m location from the inlet was converted to an equivalent

magnetite deposition thickness at the asymptotic temperature. This was found to be 0.47 mm or

1.5g of magnetite in a 15 cm long section of the tube. Such high fouling rates are unlikely with

only magnetite fouling, and therefore the high increase in temperature may be a product of mixed

recrystallization of ferrous chloride as well as magnetite particles. Such a scenario should be

considered as the solubility of any salt, including ferrous chloride would be very low in

supercritical conditions.

Temperature measurements for ferrous sulfate precursor showed variations in the test section

temperature but also in the bulk fluid temperatures (Appendix C. 2). This may be due to

inconsistent flow rate of the injection pump for this experiment. Once this was taken into

consideration, it was concluded that no significant fouling was observed from the temperature

measurements.

5.1.2 Filter Analysis

In the experiment with a ferrous chloride precursor, it was determined after the experiment that

the secondary line was at a significantly lower flow rate than initially set. The HTHP and LTLP

filter was visibly unchanged and showed very few particles collected through SEM images. On

the other hand, the HTHP filter for the ferrous sulfate precursor had a very fine but visible black

film on the silver membrane after the experiment. SEM analysis of the HTHP filter revealed

72
significant amount of particles collected, with particle diameters ranging from less than a micron

to several microns in diameter. These particles are much larger than those seen by Adschiri et al.

(2000) in hydrothermal synthesis of iron oxide particles which observed the formation of ~50nm

diameter particles. However, there is a possibility that the larger particles were captured by their

system filter which had not been analyzed.

5.1.3 Deposit Analysis

SEM images of the test section revealed that the ferrous chloride precursor produced magnetite

particles early in the test section. The particles which deposited near the inlet were smaller with

few larger micron-sized particles. Further down the test section at higher temperatures, the

particles deposited grew in size to several microns in which deposition was dense and compact.

Figure 5-2: SEM photograph of magnetite deposit on test section tube surface, 0.15 m location.
Reference condition with ferrous chloride precursor (Experiment #1)

73
Figure 5-3: SEM photograph of magnetite deposit on test section tube surface, 0.60 m location.
Reference condition with ferrous chloride precursor (Experiment #1)

Figure 5-4: SEM photograph of magnetite deposit on test section tube surface, 1.20 m location.
Reference condition with ferrous chloride precursor (Experiment #1)

74
Figure 5-5: SEM photograph of magnetite deposit on test section tube surface, 1.65 m location.
Reference condition with ferrous chloride precursor (Experiment #1)

X-ray diffraction (XRD) was conducted on the oxide deposits found in the test section tube by

mechanically removing the deposit. Figure 5-6 and Figure 5-7 show XRD results for ferrous

chloride and ferrous sulfate precursors respectively. For ferrous chloride, a General Area

Detector Diffraction System (GADDS) using a copper radiation operated at 40kV and 40mA

scanned the deposit from 21.4⁰ to 71.8⁰ at 0.050⁰ increments. A diffracted beam graphite

monochromator and Hi-Start detector was used for producing and detecting the signal. Signals

were relatively weak and the peaks were broad due to the small quantity of deposit that was

available from the tube (see Figure 5-6).

75
Figure 5-6: XRD of tube deposits of test section, reference condition ferrous chloride precursor
(Experment #1)

Figure 5-7: XRD of tube deposit of test section, reference condition ferrous sulfate precursor
(Experiment #2)

76
For ferrous sulfate reference experiment, a D-8 Advance X-ray Diffractometer using a copper

anode tube operated at 40kV and 40mA, scanned the filter from 5.0⁰ to 70.0⁰ at 0.040⁰

increments. A diffracted beam graphite monochromator and NaI scintillation detector was used

for producing and detecting the signal. XRD of the deposit indicated a mixture of magnetite and

hematite which may have formed due to oxygen contamination during the experiment (see

Figure 5-7).

To determine the deposition thickness along the test section, cleaning methods described in

section 4.5.4 were applied to the sample tubes. This is similar to chemical analysis methods for

deposition determination used in previous studies by dissolving magnetite layers from

experimental tubes using hydrochloric acid (Newson et al., 1983).

Between ferrous chloride and ferrous sulfate precursors, the amount and location of the deposit

was relatively similar. At lower temperatures near the inlet, the deposit was found to be less and

the deposit thickness grew as the temperature increased towards the outlet of the test section.

Ultrasonic cleaning was able to remove the majority of the deposit, however, some of the deposit

remained intact to the tube and was only removed after the acid wash procedure. It was observed

that for tube sections which were subject to supercritical temperatures, depositions remained

intact and were much more difficult to remove than those exposed to subcritical temperatures.

This suggests an increase in the oxide adhesive strength to the surface at supercritical conditions,

shown by the increased amount of magnetite which was removed only after an “aggressive”

cleaning method, regardless of the precursor injected.

77
In a study conducted by Yeon et al. (2006), the deposition characteristic of hematite particles

onto a heated zircaloy tubing was found to change by adding ferrous species to the solution.

Under single phase flow conditions, there was almost no deposition of hematite particles onto a

zircaloy surface unless the particles were mixed with ferrous ions. The deposit was found to form

two layers on the surface; a top layer which was easily removed with ultrasonic cleaning using

distilled water and an inner layer which was only removed after ultrasonic cleaning with

concentrated acid. Yeon et al. (2006) concluded that the top layer was formed from particle

deposition while the inner layer was formed from the precipitation of ferrous ions. Therefore, the

crystallization of dissolved ferrous species formed a stronger bond to the surface which then

allowed the loosely attached particles to stick to the wall. In the case of the ferrous precursors in

supercritical water, the higher strength of the oxide in supercritical conditions is believed to be

due to small amounts of crystallization occurring at the interface between particles and the tube

surface due to the solubility drop of magnetite in supercritical water.

78
Figure 5-8: Deposit thickness on test section for reference condition with ferrous chloride
precursor (Experiment #1)

Figure 5-9: Deposit thickness on test section for reference condition with ferrous sulfate
precursor (Experiment #2)

79
5.2 Effect of Heat Flux

Experiments were run with low heat flux and no heat flux conditions and are compared with the

reference condition. Low heat flux of approximately 20 kW/m2 in the test section was conducted

for both ferrous chloride and ferrous sulfate precursor, with temperature ranging from 371°C to

376°C. A third experiment under no heat flux condition had ferrous sulfate precursor solution

and the temperature was at 384°C.

5.2.1 Temperature Analysis

Temperature measurements were taken for the low heat flux conditions, however, no significant

differences in temperature were found in these cases.

5.2.2 Filter Analysis

SEM analysis showed sub-micron to several micron-sized particles scattered along the

membrane similar to those seen in the reference conditions for ferrous sulfate (see Figure 5-10).

The filter shows particles of various shapes and sizes, some which appear to be crystalline.

20.0keV area EDX was applied on the filter which confirmed the presence of iron oxide along

with silver from the background as shown in Figure 5-11. The absence of other metal oxides

such as nickel, chromium, or molybdenum oxides support the assumption that corrosion of the

stainless steel 316L test section under experimental conditions is negligible. The HTHP filter for

ferrous sulfate at low heat flux was fairly uniform magnetite particles of 1 µm size with

agglomerates of smaller particles on HTHP filter suggesting that the type of precursor has some

influence on the particles formed under the same conditions.

80
Figure 5-10: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for low
heat flux with ferrous chloride precursor (Experiment #3)

Figure 5-11: EDX of HTHP filter for low heat flux with ferrous chloride precursor

81
Figure 5-12: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for low
heat flux with ferrous sulfate precursor (Experiment #4)

Figure 5-13: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for no
heat flux with ferrous chloride (Experiment #5)

82
For the experiment run with ferrous sulfate with no heat flux and supercritical temperature, the

particles on the HTHP filter were of larger diameter than in low heat flux experiments and some

of these larger particles also appear to have an octahedral structure as seen in Figure 5-13, which

is a common crystal structure for magnetite (Anthony et al., 1997). XRD was carried out on the

HTHP silver membrane filter of Experiment #3 to determine the crystal structure of the

deposited particles. Results show that the peaks match very well with those of magnetite and

comparison of these peaks to other iron oxide such as hematite indicated that these were not

present in the sample. Carbon graphite is also present in the sample and is believed to have been

detected due to the carbon tape used on the filter for SEM.

Figure 5-14: XRD of deposit on HTHP filter for low heat flux with ferrous chloride precursor
(Experiment #3)

83
5.2.3 Deposit Analysis

Deposition thickness of the test section for ferrous sulfate precursor under low and no heat flux

conditions were examined and the results are found in Appendix C. SEM images of the tubes

show very little deposition and these findings are confirmed using cleaning methods.

Figure 5-15: Deposit thickness on test section, low heat flux with ferrous sulfate precursor
(Experiment #4)

The effect of heat flux compared to reference condition suggests that lower heat flux results in

lower deposition. This compares to previous models which showed that deposition had a direct

influence from heat flux (Electric Power Research Institute Inc., 2003). In previous studies,

experiments showed that deposition of magnetite particles in bulk concentration of 0.16mg to

5.04mg of Fe kg/L was 30% lower on an unheated surface than one with heat flux of 400kW/m2

(Newson et al., 1983).

84
Although the original Kern and Seaton model does not account for the influence of heat flux on

particle deposition, there have been several attempts in later studies to include heat flux by

incorporating thermophoresis and an Arrhenius term with an activation temperature (Müller-

Steinhagen et al., 1988). Thermophoresis has been shown to create a measurable change in the

deposition of a particle onto a heated surface for particles as large as 11 µm (Epstein, 1997),

although it primarily observed to influence smaller submicron particles. The thermophoretic

velocity is dependent on the fluid properties and temperature gradient and can be calculated

using the equation below:

5-1

where is a coefficient determined by:

5-2

The constant k is 1.8 for gases and 0.26 for liquids, λp and λl represent the conductivity of the

particle and liquid respectively (Epstein, 1997). The overall deposition velocity accounting for

thermophoresis is described in equation 5-3:

( ) 5-3

The equation resulted in values for the thermophoretic velocity of Kth in the range of 10-10 m/s,

much smaller than the mass transport velocity of 10-4 m/s for these experiments, suggesting that

thermophoresis should have an insignificant impact on deposition.

85
In addition to thermophoresis, high heat flux conditions usually result in higher surface

temperature given constant flow rates, which provide the activation energy required for the

particle to stick to the wall. In an experiment with alumina particles depositing on a heated

surface, Müller-Steinhagen et al. (1988) observed that an asymptotic fouling resistance had a

maximum at any given heat flux for a fixed flow rate, concentration and bulk temperature. To

model the data, Müller-Steinhagen included both thermophoresis and activation energy into the

Kern and Seaton model. The model showed that the increase in heat flux initially resulted in the

increase of surface temperature resulting in higher deposition rates. However, as Tw increasingly

exceeds Tb, deposition is counteracted and eventually overtaken by thermophoresis thereby

decreasing deposition with increasing heat flux. In general, the correlation over-predicted

deposition velocities and a correction factor had to be introduced to give a reasonable fit to the

experimental data. Such a procedure demonstrates the difficulty in accurately modeling the

complex nature of particles in heat transfer flow.

Although the Arrhenius term predicts lower deposition rates for lower heat flux conditions, it

fails to adequately explain the extent to which deposition rate decreased in these experiments

compared to reference conditions. Müller-Steinhagen et al. (1988) postulated that surface/fluid

temperature may also affect the structure of the deposit, making it more porous and loosely

packed in certain conditions. It is not known if this is affected by absolute wall temperature or by

the wall-liquid temperature difference. If it is assumed that the lower heat flux results in loosely

packed deposits which can easily be subject to removal, this may explain the lower deposition of

the low heat flux condition observed in this study. However, further research would be required

86
in order to make any conclusions regarding if heat flux has a direct influence on the deposit

structure.

5.3 High pH Conditions

Current CANDU reactors normally operate with a pH of 9 ~ 10 using a simple water chemistry

of lithium hydroxide for pH control (Burrill, 2000). Two experiments were conducted with

ferrous sulfate as a precursor but at a higher pH of 9 (at room temperature). The experiments

were run with temperatures running at 350°C to 400°C and 200°C to 370ºC.

The pH of the system was increased by adding sodium hydroxide salt to the feedwater. At first,

ferrous sulfate and sodium hydroxide was mixed in the injection tank and the pH was measured

directly. However, the introduction of sodium hydroxide to a ferrous sulfate solution

immediately produced particles in the system which would cause problems with the injection

pump. As an alternative approach, sodium hydroxide solution was introduced in the primary tank

and allowed to mix with the ferrous sulfate in-situ of the system. Due to size limitations of the

main tank, sodium hydroxide was added to the first tank after heating the entire system. The

sodium hydroxide was weighed and mixed into a deaerated tank and a sample was drawn from

the tank. The conductivity was measured and compared to the amount of sodium hydroxide

required using a calibration curve shown in Figure 5-16. A linear regression gave the equation 5-

3 for the concentration as a function of conductivity.

5-4

where C is the concentration [mmol/L] and σ is the conductivity [mS/cm].

87
Figure 5-16: Calibration of the conductivity meter for determining NaOH concentration in
primary tank

The pH could not be measured directly in the system due to the high temperature and pressure of

the solution at the mixing tee. Instead, the pH of the ferrous sulfate solution was determined by

using data from Arden (1950) for ferrous sulfate and sodium hydroxide solution. Conditions in

Arden’s experiment are similar to the conditions in this study. The solution was deaerated to

eliminate the presence of oxygen and ferric ions which have been found to create large errors in

the pH measurements. Oxygen in the system promotes the formation of ferrosic hydroxide, a

ferric ion compound instead of the desired ferrous hydroxide (Arden, 1950).

The conductivity of the primary tank after dissolving sodium hydroxide was 2.02 and 2.10

mS/cm which correspond to a concentration of 10.49 and 10.91 mmol/L for experiments #6 and

88
#7 respectively. The ratio of Na to Fe was calculated with equation 5-4 taking into consideration

the dilution of the primary stream with the injection stream.

̇ ̇ 5-5
̇ ̇

The molar ratio of Na to Fe was compared to data with Arden (1950) and the pH was determined

to be approximately 9 in both experiments, as shown below.

Figure 5-17: Predicted pH of the system using data from Arden (1950)

5.3.1 Temperature Analysis

For the experiment with the test section heated from subcritical to supercritical, large

temperature changes were found at locations 0.15 m and 1.20 m while a steady temperature rise

was found near the outlet at 1.65m. The temperature at the 1.20 m location approached 500°C

89
after 30 minutes which was considered too high and the power to the test section was

momentarily turned off to cool the section. Once the temperature dropped to below 400°C, the

test section was turned on once more to continue the injection at the original temperatures until

the end of 40 minutes.

Injection

Figure 5-18: Temperature measurements of test section vs. time for high pH with ferrous sulfate
precursor (Experiment #6). Power turned off briefly at t = 48 minutes.

Temperature measurements at 0.15 m from the inlet suggest that deposits exfoliated from the

tube surface relatively easily and quickly. These deposition-removal cycles found in the

temperature was also observed by Khan (2005) with the injection of sodium carbonate in

supercritical water. These occurred with combined particle and crystallization deposition and

signaled a weaker adhesion to the tube wall than pure crystalline deposit.

90
5.3.2 Filter Analysis

Observing the particles captured on the HTHP and LTHP filters under an SEM, revealed that the

particle sizes are consistently much smaller than in the reference case, falling in the 100-200nm

diameter range. Formation of smaller particles in higher pH solution are confirmed by literature

studies using sol-gels. Sugimoto and Matijevic (1980) displayed that under excess OH- ions, the

magnetite particles formed were small, tens of nanometer in diameter and had a cubic

morphology. On the other hand, large magnetite particles of 1.1 µm formed under 0.03M excess

Fe2+ conditions.

The LTHP filter for both experiments had a significant amount of small particles deposited on

the glass fiber. In experiment #7, it can be seen that the deposit was enough to completely cover

all signs of the glass fiber matrix.

Figure 5-19: SEM photograph of particles collected on HTHP filter, 10.0k magnification, for
high pH, ferrous sulfate (Experiment #6)

91
Figure 5-20: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for high
pH, ferrous sulfate (Experiment #6)

Figure 5-21: SEM photograph of particles collected on HTHP filter, 5.0k magnification, for high
pH, subcritical, ferrous sulfate (Experiment #7)

92
Figure 5-22: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for high
pH, subcritical, ferrous sulfate (Experiment #7)

5.3.3 Deposit Analysis

SEM images of the tube surface after the test section was disassembled confirm that there was

very few particles attached to the surface. Furthermore, deposit thickness determination using

cleaning methods indicate that the deposit is less than 1 µm shown by Figure 5-23.

93
Figure 5-23: Deposit thickness on test section (Experiment #6)

As mentioned in Section 2.3.3, the particles that deposit on the tube are submerged in the viscous

sub-layer which dissipates turbulent bursts responsible for particle removal. The viscous sub-

layer thickness was calculated for the high pH at supercritical experiment using equation 5.5

below (Cengel & Cimbala, 2006).

5-6

The calculations showed that this distance was approximately 25 µm at 200°C, 14 µm at 350°C,

and 6 µm at 400ºC under the experimental conditions. In the case of Experiment #6 at 0.15m

location, the deposition-removal cycles had an average temperature change of 12°C from the

initially clean state. This corresponded to a calculated thickness of the deposit right before the

removal of 42 µm.

94
The cool down procedure was repeated in a similar manner for all experiments for direct

comparison between the experiments. Significant amount of particles were found in the effluent

of the system during the cooling/depressurization for both high pH experiments, indicating that

removal rates post-injection were much higher for this deposit. This may be attributed to the

increase in flow rate during depressurization. Turner et al. (1990) suggested that the change in

flow rates are a larger factor in removal rates compared to steady state conditions.

5.4 Low Concentration and Blank Experiment

The experiment with lower concentration of 0.5 mmol/L and proportionally longer duration of

400 minutes was conducted to determine the effects of concentration on deposition without

changing the total amount of ferrous species injected. All three filters showed ferrous-chromium

exfoliation occurring in the system indicating corrosion of the tube. Hence, the results from this

experiment were not analyzed any further.

A blank test was also run with reference conditions but without a precursor. SEM analysis of the

HTHP and LTLP filters showed almost no particles collected as expected. The LTHP filter had

few particles which likely exfoliated from the heat exchanger, but the amount was insignificant

compared to the experiments with a precursor.

5.5 Overall Summary and Mass Balance

A summary of the significant results from each of the experiments are provided in Table 5-2.

95
Table 5-2: Results summary table - Part I

Filters
Parameter Precursor Tube Samples SEM
HTHP LTHP LTLP
1 Reference FeCl2 Almost no Large amount Almost no Small particles at inlet,
case for particles of submicron particles large micron sized
ferrous and some particles at high
chloride micron sized temperatures
particles

2 Reference FeSO4 Nanometer to Very Small Almost no Small and few deposits
case for several micron particles of particles found at inlet, large deposits
ferrous unknown size at center of test
sulfate section

3 Low heat flux FeCl2 Several micron Very Small N/A consistent micron sized
size particles of particles found at inlet
unknown size and outlet of test
section

4 Low heat flux FeSO4 Consistent ≈ 1 Very Small Almost no Very few deposits
micron size particles of particles found
unknown size

5 No heat flux, FeSO4 Hundred Approximately Almost no Combination of


all nanometer to 500nm particles found submicron and micron
supercritical several micron consistent size sized particles
size

6 High pH FeSO4 Approximately Approximately Small amount Very few deposits


100-200nm 100-200nm of particles,
consistent sizes consistent sizes hundred-
nanometer
sized particles

7 High pH, all FeSO4 Approximately Approximately Almost no N/A


subcritical 100-200nm 100-200nm particles found
consistent sizes consistent sizes

8 Low FeCl2 Nanometer to Large amount Large amount N/A


Concentration several micron of particles and of very small
flakes particles

Blank - Almost no Few particles Almost no N/A


particles of sub-micron particles
size

96
Table 5-3: Results summary table - Part II

Temperature Deposit Mass


Deposit Strength XRD Comments
Measurements Thickness Balance
1 Temperature rise at 10µm More depost strength Recovery Analyzed tube Secondary line flow rate was
supercritical maximum in supercritical of 79% deposits, found much lower resulting in very
temperatures, no conditions magnetite few particles on secondary line
temperature difference filters
for subcritical

2 No difference found 12µm More deposit strength Recovery Analyzed tube


maximum in supercritical of 82% deposits, found
conditions magnetite and
some hematite

3 No difference found N/A N/A Recovery Analyzed HTHP Test section is 1.0m long, test
of 52% Filter, found section not examined using
magnetite cleaning method

4 No difference found Less than No difference Recovery N/A


1µm of 72%
thickness

5 N/A Less than No difference Recovery N/A


1µm of 77%
thickness

6 Temperature rise at Less than No difference Recovery N/A Large amount of black particles
supercritical 1µm of 53% found in the effluent after
temperatures, oxide thickness cooling the system
deposition and
exfoliation found at
subcritical

7 No difference found N/A N/A Recovery N/A Did not dissassemble test
of 39% section for this experiment

8 No difference found N/A N/A Recovery N/A Did not dissassemble test
of 83% section for this experiment

No difference found N/A N/A N/A N/A Did not dissassemble test
section for this experiment

97
A mass balance calculation was completed for all experiments, however, not all sections for each

experiment were analyzed and therefore, are presented to provide an approximation only. The

majority of the experiments had good recovery of the ferrous species injected into the system

with the exception of experiment #7 which was below 50%. To determine the yield of magnetite

for input into the simulation, the iron content found in the effluent was subtracted from the total

injected amount. Hence, it was assumed that the iron mass which was not recovered had

deposited as magnetite particles onto other surfaces within the system not analyzed, such as the

heat exchanger or the tube fittings between the mixing tee and test section. Mass balance for

future experiments may be improved by cleaning the heat exchanger after each experiment and

analyzing the iron concentration in the cleaning solution. The equation for recovery and yield are

given in equations 5-7 and 5-8 respectively.

5-7

5-8

In experiment #6 and #7, it was found that the effluent contained significant amount of oxide

particles which passed through the LTHP filter, perhaps due to overloading. The experiment

produced a higher yield of iron oxide particles indicated by the relative number of particles

collected on the HTHP filter. To gain insight into the yield in the high pH experiments, the

effluent from experiment #7 was first filtered using a Fisherbrand filter (Grade - P2) to separate

the residue from the filtrate. The residue and filtrate were independently analyzed using AAS. It

was found that the effluent was 57% particulate matter. Therefore, for experiments #6 and #7,

98
the same percentage of particulate matter in the effluent was assumed and the amount was added

to the total yield of particles in the system.

Table 5-4: Mass balance of experiments #1 - 8

2+ Flow Run Test Total Filter Total Iron Total Iron


Experiment Fe Effluent Recovery Estimated
Rate Time Section Deposit Expected Measured
# [mmol/L] [mg] [%] Yield
[L/hr] [min] [mg] [mg] [mg] [mg]
1 5.48 3.54 42 353 244 0.69 758 598 79% 52%
2 4.97 3.54 39 141 380 1.28 639 522 82% 78%
3 5.31 3.60 44 404 NM NM 783 404 52% 48%
4 5.04 3.54 40 478 11.5 NM 664 478 72% 28%
5 5.32 3.50 40 536 26.1 NM 692 536 77% 23%
6 5.11 3.50 37 310 12.0 2.53 615 324 53% 78%
7 5.17 3.47 40 260 NM NM 667 260 39% 83%
8 0.56 3.50 400 605 NM NM 729 605 83% 17%

99
6 Results and Discussion of Simulations

6.1 Comparison between Simulation and Experimental

Simulations were run with parameters from experiments and these were compared to

experimental deposition thickness obtained from the cleaning method. The magnetite

concentration used for the simulation was obtained from the yield estimated from mass balance

in Table 5-4, while the average particle size was estimated from the SEM images of the tube

deposit and filters. For the reference conditions, the average particle size was estimated to be 4

µm and 1 µm for ferrous chloride and ferrous sulfate precursors respectively. For these reference

condition experiments, the simulation results predicted magnetite deposition reasonably well.

The simulation predicted similar deposition thickness for each case, however, with larger

particles the peak deposit shifted closer towards the inlet of the test section.

(a) (b)

Figure 6-1: Comparison between simulation and experimental for reference conditions (a)
Experiment #1, ferrous chloride with average particle size = 4 µm, magnetite
concentration = 200 mg/L (b) Experiment #2, ferrous sulfate with average particle
size = 1µm, magnetite concentration = 299 mg/L.

100
For low heat flux conditions, the deposition thickness was overestimated by nearly a factor of

four. However, the simulation did predict a lower deposition rate compared to the other

experiments and a uniform deposition rate along the test section compared well to experimental

results, as shown in Figure 6-2.

(a) (b)

Figure 6-2: Comparison between simulation and experimental for ferrous sulfate experiments
with (a) Experiment #4, low heat flux with average particle size = 1 µm, magnetite
concentration = 108 mg/L and (b) Experiment #5, no heat flux with average particle
size = 2 µm, magnetite concentration = 89 mg/L.

Finally, for the experiments at high pH, the simulation greatly overestimated the deposition

thickness. This is likely due to the removal that occurred during and after the deposition process

described in section 5.3.3. Higher removal rates in the high pH conditions suggest that the

deposit had a weaker adhesion to the tube. As discussed in section 5.1.3, the strength of oxide

adhesion was enhanced with the precipitation of ferrous species, and one explanation is that the

101
reduced amount of dissolved ferrous ions in the solution resulting from the high yield of particles

formed under high pH condition may have caused the formation of a weak deposit. A second

possibility is that the formation of molten sodium hydroxide salts had formed and deposited

along with the magnetite particles in supercritical conditions. During cool down, the molten salt

would have re-dissolved in the water, possibly carrying the magnetite particles along with it.

Figure 6-3: Comparison between simulation and experimental for Experiment #6, high pH,
ferrous sulfate. Simulation was run with an average particle size = 0.15µm,
magnetite concentration = 321 mg/L.

In all cases, the simulation overestimated the deposition of magnetite suggesting that surface

attachment and/or removal rates cannot be neglected in magnetite particle deposition in

supercritical water.

102
6.2 Simulation for CANDU SCWR

As one of the design criteria in the proposed SCWR, the maximum fuel cladding temperature

cannot exceed 850⁰C (Naidin et al., 2009). This is to ensure that the material’s mechanical and

chemical properties can withstand the aggressive environment of supercritical water. Li et al.

(2009) developed a 3-dimensional heat transfer simulation for a CANFLEX fuel bundle in a

SCWR which showed that the maximum temperature of the cladding can reach as high as

802.2⁰C at certain locations of the fuel bundle.

The CANDU SCWR deposition simulation developed in this study does not account for oxide

growth from the fuel cladding. This is due to the fact that the alloy for the fuel cladding has not

been decided and the rate of oxide growth from the alloy varies dramatically depending on the

material (Was et al., 2006). Additionally, only the surface temperature of the fuel cladding

exposed to supercritical water is calculated since the design criteria of maximum fuel cladding is

concerned with the surface temperature (Li et al., 2009).

Figure 6-4 shows the expected fuel cladding-surface temperature with a heat flux of 1000

kW/m2 with no deposition (see Table 3-1 for simulation parameters). The fluid enters the test

section at 350ºC and 25MPa, saturated with 11µg/L of ferrous ions. As the temperature

increases beyond the pseudocritical temperature, the solubility drops in accordance with

Figure 2-3 and particles form in the bulk fluid. Figure 6-5 shows the temperature profile of the

fuel bundle after one year of operation assuming a particle diameter of 150 nm.

103
Figure 6-4: Bulk fluid and fuel cladding-surface temperature without particle deposition

Figure 6-5: Bulk fluid and fuel cladding-surface temperature after 1 year of operation with
magnetite particles of 150 nm diameter

104
Figure 6-6: Magnetite deposition thickness on fuel cladding after 1 year of operation for 150
nm and 1 µm average particle diameters

In Figure 6-6, the effects of particle diameter is compared between 150 nm and 1 µm. The

experimental results from section 5.3 suggest that at a pH ≈ 9-10 where a typical CANDU

reactor operates (Burrill, 2000), the formation of small, ~100 nm particles are likely. For the 1

µm scenario where 0.279 ≤ tp+ ≤ 8.284, inertial coasting is the dominant mechanism for

deposition and the deposition velocity is approaching maximum rate. On the other hand, for

the particle diameter of 150 nm where 0.006 ≤ tp+ ≤ 0.215, the deposition mechanism changes

from diffusion to inertial coasting inside the reactor core and the deposition velocity remains

closer to the minimum of the deposition velocity curve.

Maximum temperature difference between fouled and un-fouled cladding-surface

temperatures was found to be 23.9⁰C occurring approximately 3 m downstream of the inlet.

105
Since the simulation was based from 1 fuel element, the cladding temperature represents an

average among all 43 actual elements. The Li et al. (2009) simulation which looked at all 43

elements, found that cladding-surface temperatures can vary by several hundred degrees

depending on the fuel element. The highest fuel cladding temperature of 802.2⁰C occurred at

the 4 m location where an increase of 19.6⁰C due to fouling was found by this study.

Therefore after several years of operation, fouling on the fuel cladding may cause

temperatures to exceed the design criteria of 850⁰C if magnetite deposition considerations are

not taken into account.

Calculation of the deposit thickness found in the simulation by Burrill (2000) discussed in

section 2.3.4 suggest a peak deposit of 297 µm assuming mass transport of ferrous ions in a

CANDU SCWR after one year of operation. This is in relatively good agreement to the results

presented in this study considering several key differences in the parameters and assumptions

in the model; a slightly higher saturation concentration of 18µg/L (Burrill, 2000) compared to

11 µg/L, temperature of core outlet of 384°C (Burrill, 2000) compared to 625°C, and the use

of an ionic diffusion model.

Similarly, the transport and deposition model of Cook and Fatoux (2009) concluded a deposit

thickness of 22.8 µm near the outlet of a CANDU SCWR reactor. The arbitrary deposition

velocity assumed by Cook and Fatoux of 10-5 m/s which is slightly lower than the 4.2×10-5 –

28.7×10-5 m/s range for the deposition velocity calculated here using mass transport for 150

nm sized particles. Despite these differences, the maximum deposition thickness of 50 µm

found from this study fits reasonably well with the model by Cook and Fatoux.

106
6.3 Limitations of the Simulation

In Chapter 5, surface attachment resistance and removal rates were found to be important

factors in the deposition of magnetite particles in supercritical water. However, these were

neglected in the simulation as further experimental work is required to properly include a

surface attachment and removal term. Therefore, for the situation where surface attachment

and removal terms are unknown, mass transport models provide a relatively good estimate on

the upper limit for deposition rate. Similar results were found by Turner and Klimas (2000)

who examined magnetite particle deposition onto Alloy 600 under single-phase forced

convection at high temperature and pressure water. Significant deviations were found between

calculated and measured deposition velocities, all of which the mass transport model

overestimated deposition rates. Surface attachment was determined to be the main limitation

in this experiment while removal was found to be insignificant. Turner measured the

isoelectric point for both magnetite and Alloy 600 and determined that they have the same

sign of charge in high temperature alkaline water resulting in a repulsive force between the

particle and surface, thereby significantly limiting deposition (Turner & Klimas, 2000).

Therefore, it is possible that the magnetite particles formed in these experiments in subcritical

water experienced repulsion due to the electric double layer repulsion.

On the other hand, surface attachment behaviour can be significantly different in supercritical

water. The electric double layer force was calculated by Ghosh et al. (2006) as a function of

temperature in subcritical and supercritical water at 25MPa. These forces were found to

decrease significantly in supercritical water due to its low dielectric constant. On the other

hand, Van der Waals attraction potential was shown to increase as temperature increased in

107
subcritical water, and further increase in supercritical water (Ghosh et al., 2006). This would

suggest that in supercritical water, double layer and Van der Waals forces should not limit

deposit growth. Instead, the lower deposition would most likely be the result of the removal

process. This may be particularly important for particulate fouling which has a weaker bond

to the surface than crystallization fouling.

Another limitation to the model is that the nucleation and growth of the particles was assumed

to be uniform and instantaneous. Agglomeration of particles was also not considered as it is

difficult to model the interactions of particles in a complex system (Bott, 1995). Although the

formation of nanometer sized particles can be very fast (Adschiri et al., 1992), SEM images

along the test section has shown that micron-sized particles require time to grow. Therefore,

better models could be produced by including crystal growth along the test section.

108
7 Conclusion

In this study, fouling and transport of magnetite particles in supercritical water were examined

experimentally and through computer simulation models. Experimental work was conducted

with the objective of utilizing several online and offline techniques for characterizing the

deposition and transport phenomena under different heat flux, temperature ranges, and pH

conditions. To form simulated corrosion products, a hydrothermal synthesis technique was

adapted for producing magnetite particles in deaerated supercritical water using ferrous chloride

and ferrous sulfate as precursors. EDX analysis of the particles collected on the filter clearly

show an iron oxide composition and XRD of the tube and filter deposits confirm the formation of

magnetite.

An online monitoring method utilizing thermal resistance properties of magnetite was

implemented in the test section to infer the deposition thickness on the tube wall. Deposits led to

an increase in outer wall temperature under supercritical conditions during the injection of

ferrous chloride with no pH adjustment and ferrous sulfate at high pH. It is speculated that the

mixed deposition of magnetite and precipitated ferrous precursor was the source of very high

fouling rates in supercritical water. For high pH ferrous sulfate condition, deposition-removal

cycles were observed at subcritical temperatures which are believed to be only magnetite

particulate fouling. These cycles continued throughout the injection with complete removal

occurring when deposit thickness approached 42 µm. It was observed that removal during cool

down was particularly important for high pH experiments where sub-micron sized particles of

100 - 200 nm diameter were observed.

109
Using ultrasound and acid wash cleaning procedures, a qualitative method for determining

deposit strength was developed. Results suggest that under supercritical conditions, the

particulate deposit had a stronger adhesion to the tube and required an aggressive cleaning

procedure to completely remove the deposit. It is possible that the precipitation of ferrous species

between the depositing particle and surface was responsible for the stronger bond. In addition,

SEM imaging of the test section showed the morphology of the deposit varied along the tube

length and corresponded to the particles found on the filters.

All four techniques individually provide unique data on the particles while the combination of

the techniques give an overall understanding of deposition/transport in the system. The online

thermal resistance monitoring provided valuable information about deposition and removal

cycles, while particle size and adhesive strength could only be determined from offline methods.

In the future, high temperature high pressure filters could be implemented in full scale SCW

systems which could provide valuable information on the effects of water chemistry on corrosion

product transport. On the other hand, destructive testing techniques such as surface SEM and

cleaning methods would likely be useful for small test setup experiments such as the one

demonstrated in this study.

Simulation using heat and mass transport equations produced comparable predictions to

experiments for the deposition thickness in the test section. In all cases, the simulation

overestimated the deposit thickness, particularly for the high pH experiment where high removal

was observed experimentally. The results suggest that both surface attachment and removal rates

should be included in the simulation model which would effectively reduce the deposition

110
thickness. In supercritical conditions, the effects of surface attachment forces become less

significant and removal is likely to be the dominant limitation to deposition. However, difficulty

in modeling both removal and surface attachment stems from the lack of fundamental theory for

predicting their coefficients, can vary significantly depending on environmental conditions.

Finally, the simulation was adapted to predict fouling rates in a hypothetical CANDU SCWR.

The design parameters of a current ACR-700 fuel bundle along with proposed SCWR parameters

provided a baseline for the simulation. Thermodynamic solubility predictions using HKF models

from literature were used to predict the concentration of magnetite particles which would form in

the reactor core. These simulations (assuming mass-transfer limited deposition) suggested that

fouling may lead to an increase of fuel cladding-surface temperatures of up to 23.9⁰C after one

year of operation assuming the formation of submicron particles. At 4 m down from the inlet, the

temperature rise is expected to be 19.6ºC, raising the peak temperature of one section from

802.2ºC to 823.7ºC. This temperature increase in the fuel cladding can potentially be a

significant problem for long term operation of a CANDU SCWR, and therefore requires further

research.

111
8 Recommendations

This study showed that fouling and transport of corrosion products can be a significant issue for

the development of a Generation IV CANDU SCWR. Therefore there is a need for developing

new techniques and instruments for monitoring and characterizing deposition and transport,

particularly for magnetite particles. It remains a challenging task to produce these instruments

which can withstand the high temperature and pressure of supercritical water. The following

provide recommendations for further development for the techniques presented in this study.

 Use nickel alloys for test section tube material which will provide better resistance to

hydrochloric acid and are immune to sulfuric acid, both a byproduct of the hydrolysis of

ferrous chloride and sulfate respectively, making it more suitable for hydrothermal

synthesis of magnetite and cleaning method analysis (Kritzer, 2004).

 Experiments of longer duration would be useful for validating the online thermal

resistance measurements in more realistic deposition rates. Thermocouple drift may

become an issue when the thermocouples are exposed to high temperature for extended

period of time.

 Experiments of very long durations (>1000 hrs) could simulate deposition and transport

of corrosion products without the need for artificially introduced ferrous species. Instead,

the water could be saturated with ferrous species at 350°C which is then pumped into a

test section for deposition. This would require a fully automated system and the system

piping would need to be constructed from an iron free alloy or lined with ceramic or

glass.

112
References

Adschiri, T., Hakuta, Y., & Arai, K. (2000). Hydrothermal synthesis of metal oxide fine particles
at supercritical conditions. Industrial & Engineering Chemistry Research, 39, 4901-4907.

Adschiri, T., Kanazawa, K., & Arai, K. (1992). Rapid and continuous hydrothermal
crystallization of metal oxide particles in supercritical water. Journal of American
Ceramic Society, 4, 1019-1022.

American Society for Metals. (1985). Metals Handbook Desk Edition. Ohio: American Society
for Metals.

Anthony, J. W., Bideaux, R. A., Bladh, K. W., & Nichols, M. C. (1997). Handbook of
Mineralogy (Vol. 3). Chantilly, VA: Mineralogical Society of America.

Arden, T. V. (1950). The solubility products of ferrous and ferrosic hydroxids. Journal of
Chemical Society, 24, 882-885.

Basset, M., McInerney, J., Arbeau, N., & Derek, L. H. (2000). The fouling of alloy-800 heat
exchange surfaces by magnetite particles. The Canadian Journal of Chemical
Engineering, 78, 40-52.

Bastidas, J. M., Torres, C. L., Cano, E., & Polo, J. L. (2002). Influence of molybdenum on
passivation of polarised stainless steels in a chloride environment. Corrosion Science, 44,
625-633.

Bazargan, M., & Fraser, D. (2009). Heat transfer to supercritical water in a horizontal pipe:
modeling, new empirical correlation, and comparison against experimental data. Journal
of Heat Transfer, 131, 1-9.

Bazargan, M., Fraser, D., & Chatoorgan, V. (2005). Effect of buoyancy of heat transfer in
supercritical water flow in a horizontal round tube. Journal of Heat Transfer, 127, 897-
902.

113
Beal, S. K. (1970). Deposition of particles in turbulent flow on channel or pipe wall. Nuclear
Science and Engineering, 40, 1-11.

Bott, T. R. (1995). Fouling of Heat Exchangers. Netherlands: Elsevier Science.

Burrill, K. A. (1977). Corrosion product transport in water-cooled nuclear reactors. Part I:


Pressurized water operatoin. The Canadian Journal of Chemical Engineering, 55, 54-61.

Burrill, K. A. (2000). Water chemistries and corrosion product transport in supercritical water in
reactor heat transport systems. Water Chemistry of Nuclear Reactor Systems, 1, 357-363.

Cengel, Y. A., & Cimbala, J. M. (2006). Fluid Mechanics. New York: McGraw-Hill.

Chen, Q., & Ahmadi, G. (1997). Deposition of particles in turbulent pipe flow. Journal of
Aerosol Science, 28(5), 789-796.

Cleaver, J. W., & Yates, B. (1973). Mechanism of detachment of colloidal particles from a flat
substrate in a turbulent flow. Journal of Colloid and Interface Science, 44, 464-474.

Cook, W. G., & Fatoux, W. (2009). A CANDU-SCWR with a steam generator: Thermodynamic
assessment and estimation of fouling rates. Paper presented at the 4th International
Symposium on Supercritical Water-Cooled Reactors, Heidelberg.

Cooper, K., Gupta, A., & Beaudoin, S. (2001). Simulation of the adhesion of particles to
surfaces. Journal of Colloid and Interface Science, 234, 284-292.

Crockett, H. M., & Horowitz, J. S. (2010). Erosion in nuclear piping systems. Journal of
Pressure Vessel Technology, 132, 1-3.

Dai, L., Yu, M., & Dai, Y. (2007). Nozzle passage aerodynamic design to reduce solid particle
erosion of a supercritical steam turbine control stage. Wear, 262, 104-111.

Electric Power Research Institute Inc. (2003). Deposition in boilers - Review of Soviet and
Russian literature. Concord, CA.

Epstein, N. (1987). General Thermal Fouling Models. Fouling Science and Technology.

114
Epstein, N. (1997). Elements of particle deposition onto nonporous solid surfaces parallel to
suspension flows. Experimental Thermal and Fluid Science, 14, 323-334.

Gao, X., Wu, X., Zhang, Z., Guan, H., & Han, E. (2007). Characterization of oxide films grown
on 316L stainless steel exposed to H2O2-containing supercritical water. Journal of
Supercritical Fluids, 42, 157-163.

Generation IV International forum SCWR Committee. (2002). A Technology Roadmap for


Generation IV Nuclear Energy Systems.

Ghosh, S. K., Alargova, R. G., Deguchi, S., & Tsujii, K. (2006). Dispersion stability of colloids
in sub- and supercritical water. Journal of Physical Chemistry, 110, 25901-25907.

Guzonas, D., Tremaine, P., & Brosseau, F. (2009). Predicting activity transport in a
supercritical water cooled pressure tube reactor. Paper presented at the 4th International
Symposium on Supercritical Water-Cooled Reactors, Heidelberg.

Helling, R. K., & Tester, J. W. (1988). Oxidation of simple compounds and mixtures in
supercritical water: Carbon monoxide, ammonia, ethanol. Environmental Science &
Technology, 22(11), 1319-1324.

Holman, J. P. (2002). Heat Transfer (9 ed.). New York: McGraw-Hill.

Hong, Y. (2007). Composite Fouling on Heat Exchanger Surfaces. New York: Nova Science
Publishers Inc.

International Atomic Energy Agency. (2010). Energy, Electricity and Nuclear Power Estimates
for the Period up to 2050. Vienna, Austria: IAEA.

Jones, D. A. (1996). Principles and Prevention of Corrosion. Upper Saddle River, NJ: Prentice-
Hall.

Jones, R. L., Gilman, J. D., & Nelson, J. L. (1993). Controlling stress corrosion cracking in
boiling water reactors. Nuclear Engineering and Design, 143, 111-123.

115
Khan, M. S. (2005). Deposition of sodium carbonate and sodium sulfate in supercritical water
oxidation systems and its mitigation. PhD Thesis, University of British Columbia.

Khartabil, H. (2009). SCWR: Overview. Paper presented at the Gen-IV International Forum,
Paris, France.

Kritzer, P. (2004). Corrosion in high-temperature and supercritical water and aqueous solutions:
a review. The Journal of Supercritical Fluids, 29, 1-29.

Li, C., Shan, J., & Leung, L. K. H. (2009). Subchannel analysis of CANDU-SCWR fuel.
Progress in Nuclear Energy, 799-804.

Lister, D. H. (1980). Corrosion products in power generating systems. New York: Hemisphere
Publishing Corporation.

Marrone, P. A., Cantwell, S. D., & Dalton, D. W. (2005). SCWO system designs for waste
treatment: Application to chemical weapons destruction. Industrial & Engineering
Chemistry Research, 44, 9030-9039.

Masuyama, F. (2001). History of power plants and progress in heat resistance steels. ISIJ
International, 612-625.

McCabe, L. P., Sargent, G. A., & Conrad, H. (1985). Effect of microstructure on the erosion of
steel by solid particles. Wear, 105, 257-277.

Miller, D. G. (1982). Estimation of tracer diffusion coefficients of ions in aqueous solution.


Livermore, California: Lawrence Livermore Laboratory.

Müller-Steinhagen, H., Reif, F., Epstein, N., & Watkinson, A. P. (1988). Influence of operating
conditions on particulate fouling. The Canadian Journal of Chemical Engineering, 66,
42-50.

Naidin, M., Mokry, S., Baig, F., Gospodinov, Y., Zirn, U., Pioro, I., & Naterer, G. (2009).
Thermal-design options for pressure-channel SCWRs with cogeneration of hydrogen.
Journal of Engineering for Gas Turbines and Power, 131, 1-8.

116
Newson, I. H., Bott, T. R., & Hussain, C. I. (1983). Studies of magnetite deposition from a
flowing suspension. Chemical Engineering Communications, 20, 335-353.

Papavergos, P. G., & Hedley, A. B. (1984). Particle deposition behaviour from turbulent flows.
Chemical Engineering Research & Design, 62, 275-295.

Propp, A. W., Carleson, T. E., Wai, C. M., Taylor, P. R., Daehling, K. W., Huang, S., & Abdel-
Latif, M. (1996). Corrosion in Supercritical Fluids. Idaho Falls: Idaho National
Engineering Laboratory.

Schwarz, T. (2001). Heat transfer and fouling behaviour of Siemens PWR steam generators -
long-term operating experience. Experimental Thermal and Fluid Science, 25, 319-327.

Shaw, R. W., Brill, T. B., Clifford, A. A., Eckert, C. A., & Franck, E. U. (1991). Supercritical
water: A medium for chemistry. Chemical Engineering News, 69(51), 26-39.

Shreir, L. L., Jarman, R. A., & Burstein, G. T. (1994). Corrosion (3rd Edition) Volumes 1-2.
Oxford: Butterworth-Heinemann.

Sinha, A. K. (1989). Ferrous Physical Metallurgy. Stoneham: Butterworth Publishers.

Srisukvatananan, P., Lister, D. H., Svoboda, R., & Daucik, K. (2007). Assessment of the state of
the art of sampling of corrosion products from water/steam cycles. PowerPlant
Chemistry, 9(10), 613-626.

Sue, K., Adschiri, T., & Arai, K. (2002). Predictive model for equilibrium constants of aqueous
inorganic species at subcritical and supercritical conditions. Industrial & Engineering
Chemistry Research, 41, 3298-3306.

Sugimoto, T., & Matijevic, E. (1980). Formation of uniform spherical magnetite particles by
crystallization from ferrous hydroxide gels. 74(1), 227-243.

Sweeton, F. H., & Baes Jr., C. F. (1970). Th solubility of magnetite and hydrolysis of ferrous ion
in aqueous solutions at elevated temperatures. Journal of Chemical Thermodynamics, 2,
479-500.

117
Swenson, H. S., Carver, J. R., & Kakarala, C. R. (1965). Heat transfer to supercritical water in
smooth-bore tubes. Journal of Heat Transfer, 477-484.

Teshima, P. (1997). Fouling rates from a sodium sulphate - water solution in supercritical water
oxidation reactors. MASc Thesis, University of British Columbia.

Thomas, D., & Grigull, U. (1974). Experimental investigation of the deposition of suspended
magnetite from the fluid flow in steam generating boiler tubes. Brennst-Warme-Kraft, 26,
109-117.

Torgerson, D. F., Shalaby, B. A., & Pang, S. (2006). CANDU technology for Generation III+
and IV reactors. Nuclear Engineering and Design, 236, 1565-1572.

Tremaine, P. R., & LeBlanc, J. C. (1980). The solubility of magnetite and the hydrolysis and
oxidation of Fe2+ in water to 300C. Journal of Solution Chemistry, 9(6), 415-442.

Turner, C. W. (1993). Rates of particle deposition from aqueous suspensions in turbulent flow: A
comparison of theory with experiment. Chemical Engineering Science, 48, 2189-2195.

Turner, C. W., & Klimas, S. J. (2000). Deposition of magnetite particles from flowing
suspensions under flow-boiling and single-phase forced-convective heat transfer. The
Canadian Journal of Chemical Engineering, 78, 1065-1075.

Turner, C. W., Lister, D. H., & Smith, D. W. (1990). The deposition and removal of sub-micron
particles of magnetite at the surface of alloy 800. Paper presented at the Steam Generator
and Heat Exchanger Conference, Toronto, Canada.

Viswanathan, R., Henry, J. F., Tanzosh, J., Stanko, G., Shingledecker, J., Vitalis, B., & Purgert,
R. (2005). U.S. program on materials technology for ultra-supercritical coal power plants.
Journal of Materials Engineering and Performance, 14(3), 281-292.

Was, G. S., Ampornrat, P., Gupta, G., Teysseyre, S., West, E. A., Allen, T. R., . . . Pister, C.
(2007). Corrosion and stress corrosion cracking in supercritical water. Journal of Nuclear
Materials, 176-201.

118
Was, G. S., Teysseyre, S., & Jiao, Z. (2006). Corrosion of austenitic alloys in supercritical water.
Corrosion, 62(11).

Weingartner, H., & Franck, E. U. (2005). Supercritical water as a solvent. Angewandte Chemie
International Edition, 44(18), 2672-2692.

Willard, H. H., Merritt, L. L. J., & Dean, J. A. (1969). Instrumental Methods of Analysis (4 ed.).
London: D. Van Nostrand Company Inc.

Yamagata, K., Nishikawa, K., Hasegawa, S., Fujii, T., & Yoshida, S. (1971). Forced convective
heat transfer to supercritical water flowing in tubes. International Journal of Heat and
Mass Transfer, 15, 2575-2593.

Yeon, J., Jung, Y., & Pyun, S. (2006). Deposition behaviour of corrosion products on the
zircaloy heat transfer surface. Journal of Nuclear Materials, 354, 163 - 170.

Zhang, L., Zhu, F., & Tang, R. (2009). Corrosion mechanisms of candidate structural materials
for supercritical water-cooled reactor. Front. Energy Power Engineering China, 3(2),
233-240.

Zhao, L., Zhang, H., Tang, J., Song, S., & Cao, F. (2009). Fabrication and characterization of
uniform Fe3O4 octahedral microcrystals. Materials Letters, 63, 307-309.

119
Appendices

Appendix A: MATLAB Code


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% main.m
%
% This is the main program to run to calculate heat transfer and deposition
% Written by Keigo Karakama
% Date created: September 25, 2010
% Last modified: October , 2011

% Clear everything from previous programs


clc
clear all

%-------------------------------------------------------------------------
% Section 1: User defined inputs

flowrate = 3.50; % Total flow rate (Pump 1 and 2)


Ti = 350; % Temp. of bulk fluid going in [deg C]
To = 400; % Temp. of bulk fluid going out [deg C]
P = 237; % Pressure [bar]
di = 0.069*0.0254; % Inside tube dia. of test section [m]
do = 0.125*0.0254; % Outside tube dia. of test section [m]
tlength = 1.75; % Length of tube [m]
runtime_min = 40; % Run time of injection [min]
Cb(1) = 500*10^-6; % Initial concentration of suspended...
% magnetite in solution [kg/L]
dp = 0.1 *10^-6; % Diameter of particle [m]
corr = 2; % Correlation to use for heat transfer
flowrate_pump2 = 0.5; % Injection pump flow rate [L/hr]

%---------------------------------------------------------------------------
% Section 2: Runtime Variables

m = flowrate/3600; % Flow rate in [kg/s]


Nlength = 100; % Number of length steps
dt = runtime_min*60 % Time step [s]
dx = tlength/Nlength; % Increment [m]
x = 0:dx:tlength; % Array
h_upstream = XSteam('h_pT',P,Ti);
h_downstream = XSteam('h_pT',P,To);
delta_h = (h_downstream - h_upstream)/Nlength;
Q = (h_downstream - h_upstream)*m; % [kW]
q = Q*1000/(pi * di); % [W/m]
hflux = Q*1000/(pi * di * tlength); % [W/m^2]

%--------------------------------------------------------------------------
% Section 3: Constants

rho_oxide = 5200; % Density of magnetite [kg/m^3]


K316L = 20.2; % Thermal conductivity of 316L [W/mK]

120
% Other Variables
% KFe3O4 Thermal conductivity of Fe3O4 [W/mK]
% Tw1 Temperature at the bulk/surface interface
% Tw2 Temperature at the point between the oxide and alloy
% Tw3 Outside wall temperature and is measured experimentally
%--------------------------------------------------------------------------
% Section 4: Set Initial Values

count = 0; % Set initial count of loop to zero


Tb(1) = Ti; % Temperature of bulk fluid
dmdt(1) = 0; % Deposition mass flux
dm(1) = 0; % Deposition mass
thickness = zeros(1,Nlength+1);
[buo(1),Re(1),rhoB(1),Tw1(1),hconv(1)] ...
= fluidproperties(m,Tb(1),di,P,hflux,corr);
Tw2(1) = Tw1(1); % Temperature inside wall
Tw3(1) = twecalc(Tw2(1),hflux,K316L,di,do);
h(1) = h_upstream;

%--------------------------------------------------------------------------
% Section 5: Main part of the code

for j = 2:Nlength+1,
% Track and print to display
count = 1 + count

% Calculate fluid properties of next step


h(j)=h(j-1) + delta_h; % Enthalpy at position x
Tb(j) = XSteam('T_ph',P,h(j)); % Bulk temperature at position x
[buo(j),Re(j),rhoB(j),Tw1(j),hconv(j)]...
= fluidproperties(m,Tb(j),di,P,hflux,corr);

% For preformed particle transport, saturation is negligible


Csat(j) = 0;

% Calculate bulk concentration calculated from previous deposition


Cb(j) = Cb(j-1) - dmdt(j-1)/m;

% If bulk concentration is not zero, then particles are in the fluid


if (Cb(j) > Csat(j))
[Kt(j),TP(j), LN(j),viscouslayer(j)] = depositionvelocity(m,...
rho_oxide,P,Tb(j),dp,di,thickness(j));
dmdt(j) = depositionfun(Kt(j),dx,di,rho_oxide,Cb(j),...
Csat(j));
dm(j) = dmdt(j)*dt;
rfoul(j) = thicknessfun(dmdt(j),rho_oxide,di/2,dx,dt);

% If bulk concentration is zero, set values to default value of 0


else
Kt(j) = 0;
TP(j) = 0;
LN(j) = 0;
viscouslayer(j) = 0;
dmdt(j) = 0;

121
dm(j) = 0;
rfoul(j) = 0;
end

% Calculate the cumulative thickness of the deposit


thickness(j) = rfoul(j) + thickness(j);

% Calculate the thermal conductivity of magnetite at the temperature


KFe3O4(j) = 4.133 - 0.852*10^-2*Tw1(j) + 0.757*10^-5*Tw1(j);

% Calculate the temperature at the surface interfaces


Tw2(j) = twocalc(Tw1(j),hflux,KFe3O4(j),...
(di-2*thickness(j)),di);
Tw3(j) = twecalc(Tw2(j),hflux,K316L,di,do);
end

%--------------------------------------------------------------------------
% Section 6: Plotting
xnew = dx:dx:tlength; %For graphs which does not need value at x=0
for k=1:Nlength,
yfoul(k) = thickness(k+1)*1000000;
end

% Plot of result summary


figure(1)
subplot(2,2,1)
plot(xnew,yfoul)
xlabel('Length [m]')
ylabel('Thickness [\mum]')
title('Thickness vs Length')

subplot(2,2,2)
semilogy(x,TP,x,0.2,x,20)
xlabel('Length [m]')
ylabel('tp*')
title('tp* vs Length')

subplot(2,2,3)
plot(x,Tb)
xlabel('Length [m]')
ylabel('Temperature [(Khan)C]')
title('Bulk Temperature vs Length')

subplot(2,2,4)
plot(x,Cb*1000000)
xlabel('Length [m]')
ylabel('Concentration [mg/kg]')
title('Concentration vs Length')

% Plot of temperature profile


figure(2)
plot(x,Tb,'k.-',x,Tw1,'k:',x,Tw2,'k--',x,Tw3,'k')
legend('Bulk fluid temperature','Oxide temperature', ...
'Tube inside wall temperature','Tube outside wall temperature');
xlabel('Length (m)')

122
ylabel('Temperature (C)')

% Plot of fouling thickness vs. distance


figure(3)
plot(xnew,yfoul)
xlabel('Distance [m]')
ylabel('Magnetite thickness [\mum]')

% Plot of Reynolds vs. distance


figure(4)
plot(x,Re)
xlabel('Distance [m]')
ylabel('Re')

% Plot of heat transfer coefficient vs. distance


figure(5)
plot(x,hconv)
xlabel('Distance [m]')
ylabel('Heat transfer coefficient')

% Plot of Gr/Grth vs. distance for buoyancy


figure(6)
plot(x,buo,'k')
xlabel('Distance [m]')
ylabel('Gr/Grth')

%--------------------------------------------------------------------------
% Section 6: Reporting to MATLAB display

thickp = thickness'*1000000;
xp = x';
Bulkfluid = Cb(Nlength+1);

section_mass = zeros(10,1);
count1 = 1;
count2 = 1;
for p=1:Nlength,
section_mass(count1) = section_mass(count1)+dm(p);
if count2 == (Nlength/10)
count1 = count1+1;
count2 = 1;
else
count2 = count2+1;
end
end

section_mass_mg = section_mass*1000000
total_mass_mg = Cb(1)*flowrate*runtime_min*1000000/60
deposited_mass_mg = sum(section_mass_mg)
mass_to_filter_mg = total_mass_mg - deposited_mass_mg
deposited_percentage = deposited_mass_mg*100/total_mass_mg
concentration_tank2_mgL = Cb(1)*(flowrate/flowrate_pump2)*1000000
total_mass_g = Cb(1)*flowrate*runtime_min*1000/60
transpose(thickness);
hflux

123
Functions
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Buoyancy Effects Calculation
% Bouyancy Calculations based from Petukhov, evaluated by Bazargan
% Written by Keigo Karakama
% Date created: April 18, 2011
% Last modified: July 19, 2011

function [A,Grth,Grq] = bouyancy(Tw,Tb,Reb,d,P,m,hflux)

% Grth = Grashof number threshold


% Grq = Grashof number heat-flux-related
% Pr = Prandtl number
% Hw,Hb = Enthalpy at wall and bulk respectively
% Tw,Tb = Temperature at wall and bulk respectively
% mu = bulk dynamic viscosity
% k = bulk thermal conductivity
% hflux = heat flux

g = 9.81;
Tf = (Tw + Tb)/2;
rhoF = XSteam('rho_pT' ,P,Tf);
rhoB = XSteam('rho_pT' ,P,Tb);
rhoW = XSteam('rho_pT' ,P,Tw);
Hw = XSteam('h_pT' ,P,Tw);
Hb = XSteam('h_pT' ,P,Tb);
k = XSteam('tc_pT' ,P,Tb);
mu = XSteam('my_pT' ,P,Tb);

Pr = (Hw-Hb)*1000*(mu/k)/(Tw-Tb);
Refun_pipe(m,d,mu);
B = (1/rhoF)*(rhoB - rhoW)/(Tw-Tb);

Grth = (3*10^-5)*(Reb^2.75)*(Pr^0.5)*(1+2.4*(Reb^-0.125)*(Pr^(2/3)-1));
Grq = g*B*hflux*d^4/((mu/rhoB)^2*k);
A = Grq/Grth;

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Deposition Velocity
% Written by Keigo Karakama
% Date created: January 9, 2011
% Last modified: January 20, 2011

function dmdt = depositionfun(Kt,dx,di,rho_oxide,Cb,Csat);

% Kt = mass transport deposition velocity [m/s]


% di = inner diameter of tube [m]
% dx = step size [m]
% rhoB = bulk fluid density [kg/m^3]
% Cb = bulk concentration [kg Fe3O4/kg H2O]
% Csat = Saturation concentration [kg Fe3O4/kg H2O]

124
dmdt = rho_oxide*Kt*(Cb-Csat)*pi*di*dx;

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Deposition Velocity
% Written by Keigo Karakama
% Date created: January 9, 2011
% Last modified: January 10, 2011

function [Kd TP LN delta] = depositionvelocity(m,rhop,P,Tb,dp,di,thickness)

% m = mass flow rate [kg/s]


% rhop = density of particle [kg/m3]
% P = pressure [bar]
% Tb = temperature of bulk fluid [C]
% dp = diameter of particle [m]
% di = diameter of inner wall of tube [m]

rhof= XSteam('rho_pT',P,Tb);
mu = XSteam('my_pT',P,Tb);
Re = Refun_pipe(m,di,mu);
eps = 0.000002 + thickness;
f = frictionfun(eps,di,Re);
u = 4*m/(rhof*pi*di^2);
U = u*sqrt(f/8);
D = diffusioncoefficient(Tb,mu,dp);
TP = rhop*(rhof*U*dp/mu)^2/(18*rhof);
Sc = Scfun(mu,rhof,D);

% Wood, Fan and Ahmadi Equation


if TP < 0.2
KD = (4.5*10^-4)*TP^2+0.057*Sc^(-2/3);
LN = 1;
elseif (((TP > 0.2)||(TP == 0.2))&&((TP < 20)||(TP == 20)))
KD = (3.5*10^-4)*TP^2;
LN = 2;
elseif TP > 20
KD = 0.18;
LN = 3;
end

% Calculate the deposition velocity [m/s]


Kd = KD*U;

% Calculate the viscous sublayer distance [m]


delta = 5*mu/(rhof*U);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Diffusion Coefficient
% Written by Keigo Karakama
% Date created: January 9, 2011

125
% Last modified: January 9, 2011

function D = diffussioncoefficient(T,mu,dp)

% kB = Boltzmann constant [J/K]


% T = temperature [K]
% mu = dynamic viscosity [Pa.s]
% dp = diameter of particle [m]

kB = 1.38065*10^-23;
Tk = T + 273;
D = kB*Tk/(3*pi*mu*dp);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Fluid/thermal Properties Function
% Written by Keigo Karakama
% Date created: August 23, 2010
% Last modified: October 9, 2011

function [A,Reb,rhoB,Tw,hconv] = ...


fluidproperties(m,Tb,di,P,hflux,select)

% m = mass flow rate [kg/s]


% Tw = wall temperature [C]
% Tb = bulk fluid temperature [C]
% di = inside diameter [m]
% P = pressure [bar]
% hflux = heat flux [W/m^2]
% k = thermal conductivity of fluid [W/m.K]

% Calculate bulk fluid properties


rhoB = XSteam('rho_pT' ,P,Tb);

% First estimate to Tw
Tw = Tb + 1;
Tw_old = Tb + 2;

% Determine pseudocritical temperature


Tpc = pseudocritical(P);

% Correlation by Swenson et al.


if select == 1
while (abs(Tw-Tw_old) > 0.2)
% Calculate fluid properties
Hw = XSteam('h_pT' ,P,Tw)*1000;% [J/kg]
Hb = XSteam('h_pT' ,P,Tb)*1000;% [J/kg]
k = XSteam('tc_pT' ,P,Tw); % [W/m.K]
rhoW = XSteam('rho_pT' ,P,Tw); % [kg/m^3]
viscosity = XSteam('my_pT' ,P,Tw); % [Pa.s]
cp = (Hw - Hb)/(Tw-Tb); % [J/kgK]

% Calculate non-dimensional numbers


Re = Refun_pipe(m,di,viscosity);

126
Pr = Prfun(viscosity,cp,k);
Nu = Nufun(Re,Pr,rhoB,rhoW,Tb,P);
hconv = Nu*k/di;

% Store the old wall temperature and calculate a new one


Tw_old = Tw;
Tw = Twfun(hflux,Nu,k,Tb,di);
Tw = (Tw+Tw_old)/2;
end

% Correlation by Bazargan and Fraser


elseif select == 2;
% Calculate enthalpy of pseudocritical and bulk temperatures
Hpc = XSteam('h_pT' ,P,Tpc);
Hb = XSteam('h_pT' ,P,Tb);

% Initial guess of Tw
Tw = Tb + 1;
Tw_old = Tb + 2;

while (abs(Tw-Tw_old) > 0.2)


Hw = XSteam('h_pT' ,P,Tw);
Izone = (Hpc - Hb) / (Hpc - Hw);

if ((Izone > -0.9)&&(Izone < 1))


Hfac = (1-Izone)*(Hpc)/(1.9*Hw);
LM = 1;
else
Hfac = 0;
LM = 2;
end

Href = Hb + Hfac*(Hw - Hb);


Tref = XSteam('T_ph' ,P,Href);
cp = XSteam('Cp_pT' ,P,Tref) * 1000; %J/kg.K
k = XSteam('tc_pT' ,P,Tref); % W/mK
rhoW = XSteam('rho_pT' ,P,Tref); % kg/m3
viscosity = XSteam('my_pT' ,P,Tref); % Pa.s
Re = Refun_pipe(m,di,viscosity);
Pr = Prfun(viscosity,cp,k);

% Dittus Boelter Correlation


Nu = 0.023*Re^0.8*Pr^0.4;

hconv = Nu*k/di;
Tw_old = Tw;
Tw = Twfun(hflux,Nu,k,Tb,di);
Tw = (Tw+Tw_old)/2;
end
end
end

% Calculate bouyancy effects for this temperature and pressure


viscB = XSteam('my_pT' ,P,Tb); % Pa.s
Reb = Refun_pipe(m,di,viscB);

127
[A,Grth,Grq] = bouyancy(Tw,Tb,Reb,di,P,m,hflux);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Friction Factor Calculation
% Written by Keigo Karakama
% Date created: January 12, 2011
% Last modified: January 12, 2011
%
% Note: Uses explicit relation developed by S.E.Haaland

function f = frictionfun(eps,di,Re)

% Re = Reynolds number
% di = diameter of inside of tube [m]
% eps = roughness value of tube [m]

f = (-1.8*log10((6.9/Re)+(eps/(3.7*di))^1.11))^-2;

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Nusselt number function
% Written by Keigo Karakama
% Created: November 28, 2010
% Modified: December 9, 2010

function Nu = Nufun(Re,Pr,rhoW,rhoB,Tb,P)

% Re = Reynolds number
% Pr = Prandtl number
% rhoW, rhoB = fluid density at wall and bulk respectively [kg/m^3]
% Tb = Bulk fluid temperature [C]
% P = Pressure [bar]

Tcritical = 374;
Pcritical = 221;

if ((Tb > Tcritical)&&(P > Pcritical))


% Swenson et al. equation
Nu = 0.00459*Re^0.923*Pr^0.613*(rhoW/rhoB)^.231;
else
% Dittus and Boelter equation
Nu = 0.023*Re^0.8*Pr^0.4;
end

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Prandtl Function
% Written by Keigo Karakama
% Date created: August 23, 2010
% Last modified: August 23, 2010

128
function prandtl = Prfun(viscosity,cp,k)

% viscosity = viscosity [Pa.s]


% cp = specific isobaric heat capacity [J/kg.K]
% k = thermal conductivity [W/mK]

prandtl = cp*viscosity/k;

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Reynolds Pipe Function
% Written by Keigo Karakama
% Date created: August 23, 2010
% Last modified: December 9, 2010

function reynolds = Refun_pipe(m,D,viscosity)

% D = inner tube diameter [m]


% m = mass [kg/s]
% viscosity = viscosity [Pa.s]
% Laminar Re ~< 2300
% Transitional 2300 < Re < 4000
% Turbulent Re > 4000

reynolds = 4*m/(pi*D*viscosity);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Schmidt Number = viscous diffusion rate / mass diffusion rate
% Written by Keigo Karakama
% Date created: January 9, 2011
% Last modified: January 9, 2011

function Sc = Scfun(mu,rho,D)

% mu = dynamic viscosity [Pa.s]


% rho = density [kg/m^3]
% D = diffusion coefficient [m^2/s]

Sc = mu/(rho*D);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Thickness of deposit calculation
% Written by Keigo Karakama
% Date created: December 1, 2010
% Last modified: January 24, 2011
% Equation from Holman (see reference)

function t = thicknessfun(m_oxide,rho_oxide,r2,dx,dt)

129
% m_oxide = mass of deposit [kg]
% rho_oxide = density of oxide [kg]
% r2 = radius [m]

t = r2 - sqrt(r2^2 - m_oxide*dt/(rho_oxide*pi*dx));

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% External Wall Temperature
% Program to calculate the outside wall temperature with internal heat
% Written by Keigo Karakama
% Date created: December 1, 2010
% Last modified: December 1, 2010

function To = twecalc(Ti,hflux,k,di,do)

% Ti = Inside wall temperature [deg C]


% To = Outside wall temperature [deg C]
% hflux = Heat flux [W/m2]
% k = Thermal conductivity of solid [W/mK]
% di = Inside diameter [m]
% do = Outside diamter [m]

A = di/do;
To = Ti + hflux*(di/2)/(2*k)*((A^2-log(A^2)-1)/(1-A^2));

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% Oxide/Wall temperature
% Written by Keigo Karakama
% Date created: December 1, 2010
% Last modified: January 24, 2011
% Equation from Holman (see reference)

function To = twocalc(Ti,hflux,k,di,do)

% Ti = Inside wall temperature [C]


% To = Outside wall temperature [C]
% hflux = Heat flux [W/m2]
% k = Thermal Conductivity of solid [W/mK]
% di = Inside diameter [m]
% do = Outside diamter [m]

To = Ti + hflux*do*log(do/di)/(2*k);

end

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

130
Appendix B: Filter Parts Drawings

131
132
Appendix C: Results of Experiments

C. 1. Reference Condition, Ferrous Chloride

Table C-1: Summary table of experiment #1

Summary Units Values Comments


Main Date of Experiment 19-Apr-11
Parameters Precursor FeCl2∙4H2O
Average Pressure [MPa] 23.7
Concentration [mmol/L] 5.48
Pump 1 Flow Rate [L/hr] 3.00
Pump 2 Flow Rate [L/hr] 0.54
Total Flow Rate [L/hr] 3.54
Secondary Line Flow Rate [L/hr] 0.03
pH 3
pH Base -
Injection Time [min] 42
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 350
Bulk Fluid Temperature Outlet [°C] 400
Heat Flux In (based on enthalpy) [kW/m2] 102
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes (1)
Deposit with SEM Yes
Deposit with Cleaning Methods Yes
XRD Yes
Effluent Yes

Comments:

(1) Flow rates of the secondary line were found to be significantly lower than originally set and
therefore there was very little filter deposits.

133
Injection

Figure C-1: Temperature measurements of test section vs. time for reference condition with
ferrous chloride precursor. (Experiment #1)

Figure C-2: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
reference condition with ferrous chloride precursor. Low flow rate in secondary line
resulting in very few particles collected. (Experiment #1)

134
Figure C-3: SEM photograph of particles collected on LTHP filter, 1.8k magnification, for
reference condition with ferrous chloride precursor. (Experiment #1)

Figure C-4: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
reference condition with ferrous chloride precursor. Low flow rate in secondary line
resulting in very few particles collected. (Experiment #1)

135
(a)

(b)

(c)

(d)

Figure C-5: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m, (b)
0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #1)

136
Figure C-6: Deposit thickness on test section for reference condition with ferrous chloride
precursor. (Experiment #1)

Figure C-7: Comparison between simulation and experimental for reference condition ferrous
chloride. Simulation was run with an average particle size = 4 µm, magnetite
concentration = 200 mg/L.

137
C. 2. Reference Condition, Ferrous Sulfate

Table C-2: Summary table of experiment #2

Summary Units Values Comments


Main Date of Experiment 24-Jun-11
Parameters Precursor FeSO4·7H2O
Average Pressure [MPa] 23.9
Concentration [mmol/L] 4.97
Pump 1 Flow Rate [L/hr] 3.05
Pump 2 Flow Rate [L/hr] 0.49
Total Flow Rate [L/hr] 3.54
Secondary Line Flow Rate [L/hr] 0.21
pH 3.5
pH Base -
Injection Time [min] 39
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 350
Bulk Fluid Temperature Outlet [°C] 397
Heat Flux In (based on enthalpy) [kW/m2] 97
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM Yes
Deposit with Cleaning Methods Yes
XRD Yes
Effluent Yes

138
Injection

Figure C-8: Temperature measurements of test section vs. time, temperature fluctuations were
found in both the inlet and outlet bulk temperatures which may be attributed to the
inconsistent flow. (Experiment #2)

Figure C-9: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2)

139
Figure C-10: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2)

Figure C-11: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for
reference condition with ferrous sulfate. (Experiment #2)

140
Figure C-12: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m, (b)
0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #2)

141
Figure C-13: Deposit thickness on test section for reference condition with ferrous sulfate
precursor. (Experiment #2)

Figure C-14: Comparison between simulation and experimental for reference condition ferrous
sulfate. Simulation was run with an average particle size = 1 µm, magnetite
concentration = 190 mg/L.

142
C. 3. Low Heat Flux, Ferrous Chloride

Table C-3: Summary table of experiment #3

Summary Units Values Comments


Main Date of Experiment 30-Mar-11
Parameters Precursor FeCl2∙4H2O
Average Pressure [MPa] 22.4
Concentration [mmol/L] 5.31
Pump 1 Flow Rate [L/hr] 3.06
Pump 2 Flow Rate [L/hr] 0.54
Total Flow Rate [L/hr] 3.60
Secondary Line Flow Rate [L/hr] 0.21
pH 3
pH Base -
Injection Time [min] 44
Test Section Length of Test Section [m] 1 (1)
Thermocouple Spacing [cm] 10
Bulk Fluid Temperature Inlet [°C] 372
Bulk Fluid Temperature Outlet [°C] 376
Heat Flux In (based on enthalpy) [kW/m2] 24
# of Thermocouples (test
section) 10
Analysis Thermal Resistance Yes
Filters Yes (2)
Deposit with SEM Yes
Deposit with Cleaning Methods No
XRD Yes
Effluent Yes

Comments:

(1) Experiment used a 1 m long test section and lower line voltage.
(2) LTLP filter had not been added yet to the system for this experiment.

143
Injection

Figure C-15: Temperature measurements of test section vs. time for low heat flux ferrous
chloride precursor. (Experiment #3)

Figure C-16: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for low
heat flux with ferrous chloride precursor. (Experiment #3)

144
Figure C-17: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for low
heat flux with ferrous chloride precursor. (Experiment #3)

Image Not Available

(See comments in Summary)

Figure C-18: SEM photograph of particles collected on LTLP filter, - magnification, for low heat
flux with ferrous chloride precursor. (Experiment #3)

145
(a)

(b)

(c)

Figure C-19: SEM photograph of magnetite deposit on test section tube surface, (a) 0.10 m, (b)
0.30 m, and (c) 0.90 m location. (Experiment #3)

146
Data Not Available

(See comments in Summary)

Figure C-20: Deposit thickness on test section for low heat flux ferrous chloride precursor.
(Experiment #3)

147
C. 4. Low Heat Flux, Ferrous Sulfate

Table C-4: Summary table of experiment #4

Summary Units Values Comments


Main Date of Experiment 31-Aug-11
Parameters Precursor FeSO4·7H2O
Average Pressure [MPa] 23.0
Concentration [mmol/L] 5.04
Pump 1 Flow Rate [L/hr] 3.03
Pump 2 Flow Rate [L/hr] 0.51
Total Flow Rate [L/hr] 3.54
Secondary Line Flow Rate [L/hr] 0.20
pH 3.5
pH Base -
Injection Time [min] 40
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 372
Bulk Fluid Temperature Outlet [°C] 376
Heat Flux In (based on enthalpy) [kW/m2] 13
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM Yes
Deposit with Cleaning Methods Yes
XRD No
Effluent Yes

148
Injection

Figure C-21: Temperature measurements of test section vs. time for low heat flux with ferrous
sulfate precursor. (Experiment #4)

Figure C-22: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for low
heat flux ferrous sulfate precursor (Experiment #4)

149
Figure C-23: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for low
heat flux ferrous sulfate precursor. (Experiment #4)

Figure C-24: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for low
heat flux ferrous sulfate precursor. (Experiment #4)

150
(a)

(b)

(c)

(d)

Figure C-25: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m, (b)
0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #4)

151
Figure C-26: Deposit thickness on test section for low heat flux ferrous sulfate precursor.
(Experiment #4)

Figure C-27: Comparison between simulation and experimental for low heat flux, ferrous sulfate.
Simulation was run with an average particle size = 1 µm, magnetite concentration =
108 mg/L.

152
C. 5. No Heat Flux, Ferrous Sulfate

Table C-5: Summary table of experiment #5

Summary Units Values Comments


Main Date of Experiment 20-Sept-11
Parameters Precursor FeSO4·7H2O
Average Pressure [MPa] 23.3
Concentration [mmol/L] 5.32
Pump 1 Flow Rate [L/hr] 2.97
Pump 2 Flow Rate [L/hr] 0.53
Total Flow Rate [L/hr] 3.50
Secondary Line Flow Rate [L/hr] 0.21
pH 3.5
pH Base -
Injection Time [min] 40
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 384
Bulk Fluid Temperature Outlet [°C] 383
Heat Flux In (based on enthalpy) [kW/m2] -
# of Thermocouples (test
section) 11
Analysis Thermal Resistance No (1)
Filters Yes
Deposit with SEM Yes
Deposit with Cleaning Methods Yes
XRD No
Effluent Yes

Comments:

(1) Because there was no net heat flux in this experiment, thermal resistance measurements on
the test section were not taken.

153
Image Not Available

(See comments in Summary)

Figure C-28: Temperature measurements of test section vs. time for no heat flux with ferrous
sulfate precursor. (Experiment #5)

Figure C-29: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for no
heat flux with ferrous sulfate precursor. (Experiment #5)

154
Figure C-30: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for no
heat flux with ferrous sulfate precursor. (Experiment #5)

Figure C-31: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for no
heat flux with ferrous sulfate precursor. (Experiment #5)

155
(a)

(b)

(c)

(d)

Figure C-32: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m, (b)
0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #5)

156
Figure C-33: Deposit thickness on test section for no heat flux with ferrous sulfate precursor.
(Experiment #5)

Figure C-34: Comparison between simulation and experimental for no heat flux with ferrous
sulfate precursor. Simulation was run with an average particle size = 2 µm,
magnetite concentration = 89 mg/L.

157
C. 6. High pH, Ferrous Sulfate

Table C-6: Summary table of experiment #6

Summary Units Values Comments


Main Date of Experiment 03-Aug-11
Parameters Precursor FeSO4·7H2O
Average Pressure [MPa] 23.8
Concentration [mmol/L] 5.11
Pump 1 Flow Rate [L/hr] 2.99
Pump 2 Flow Rate [L/hr] 0.51
Total Flow Rate [L/hr] 3.50
Secondary Line Flow Rate [L/hr] 0.23
pH 9
pH Base NaOH
Injection Time [min] 37
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 350
Bulk Fluid Temperature Outlet [°C] 396
Heat Flux In (based on enthalpy) [kW/m2] 97
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM Yes
Deposit with Cleaning Methods Yes
XRD No
Effluent Yes

158
Injection

Figure C-35: Temperature measurements of test section vs. time for high pH with ferrous sulfate
precursor. (Experiment #6)

Figure C-36: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for high
pH with ferrous sulfate precursor. (Experiment #6)

159
Figure C-37: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for high
pH with ferrous sulfate precursor. (Experiment #6)

Figure C-38: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for high
pH with ferrous sulfate precursor. (Experiment #6)

160
(a)

(b)

(c)

(d)

Figure C-39: SEM photograph of magnetite deposit on test section tube surface, (a) 0.15 m, (b)
0.60 m, (c) 1.20 m, and (d) 1.65 m location. (Experiment #6)

161
Figure C-40: Deposit thickness on test section for high pH with ferrous sulfate precursor.
(Experiment #6)

Figure C-41: Comparison between simulation and experimental for high pH, ferrous sulfate.
Simulation was run with an average particle size 0.15 µm, magnetite concentration
= 321 mg/L.

162
C. 7. High pH & Subcritical, Ferrous Sulfate

Table C-7: Summary table of experiment #7

Summary Units Values Comments


Main Date of Experiment 22-Aug-11
Parameters Precursor FeSO4·7H2O
Average Pressure [MPa] 23.7
Concentration [mmol/L] 5.17
Pump 1 Flow Rate [L/hr] 2.95
Pump 2 Flow Rate [L/hr] 0.51
Total Flow Rate [L/hr] 3.47
Secondary Line Flow Rate [L/hr] 0.27
pH 9
pH Base NaOH
Injection Time [min] 40
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 200
Bulk Fluid Temperature Outlet [°C] 370
Heat Flux In (based on enthalpy) [kW/m2] 92
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM No (1)
Deposit with Cleaning Methods No (1)
XRD No
Effluent Yes

Comments:

(1) Test section was not disassembled after experiment.

163
Injection

Figure C-42: Temperature measurements of test section vs. time for high pH & subcritical with
ferrous sulfate precursor. (Experiment #7)

Figure C-43: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for high
pH & subcritical with ferrous sulfate precursor. (Experiment #7)

164
Figure C-44: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for
high pH & subcritical with ferrous sulfate precursor. (Experiment #7)

Figure C-45: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for high
pH & subcritical with ferrous sulfate precursor. (Experiment #7)

165
C. 8. Low Concentration, Ferrous Chloride

Table C-8: Summary table of experiment #8

Summary Units Values Comments


Main Date of Experiment 26-May-11
Parameters Precursor FeCl2∙4H2O
Average Pressure [MPa] 23.7
Concentration [mmol/L] 0.56
Pump 1 Flow Rate [L/hr] 2.95
Pump 2 Flow Rate [L/hr] 0.54
Total Flow Rate [L/hr] 3.50
Secondary Line Flow Rate [L/hr] 0.24
pH 3
pH Base -
Injection Time [min] 400
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 350
Bulk Fluid Temperature Outlet [°C] 400
Heat Flux In (based on enthalpy) [kW/m2] 96
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM No (1)
Deposit with Cleaning Methods No (1)
XRD No
Effluent No

Comments

(1) Test section was not disassembled after experiment.

166
Injection

Figure C-46: Temperature measurements of test section vs. time for low concentration with
ferrous chloride precursor. (Experiment #8)

Figure C-47: SEM photograph of particles collected on HTHP filter, 2.0k magnification, for low
concentration with ferrous chloride precursor. (Experiment #8)

167
Figure C-48: SEM photograph of particles collected on LTHP filter, 2.0k magnification, for low
concentration with ferrous chloride precursor. (Experiment #8)

Figure C-49: SEM photograph of particles collected on LTLP filter, 2.0k magnification, for low
concentration with ferrous chloride precursor. (Experiment #8)

168
C. 9. Blank Run

Table C-9: Summary table of blank run

Summary Units Values Comments


Main Date of Experiment 23-Jun-11
Parameters Precursor None
Average Pressure [MPa] 23.7
Concentration [mmol/L] 0.00
Pump 1 Flow Rate [L/hr] 2.98
Pump 2 Flow Rate [L/hr] 0.50
Total Flow Rate [L/hr] 3.48
Secondary Line Flow Rate [L/hr] 0.21
pH Neutral
pH Base -
Injection Time [min] 40
Test Section Length of Test Section [m] 1.8
Thermocouple Spacing [cm] 15
Bulk Fluid Temperature Inlet [°C] 351
Bulk Fluid Temperature Outlet [°C] 399
Heat Flux In (based on enthalpy) [kW/m2] 99
# of Thermocouples (test
section) 11
Analysis Thermal Resistance Yes
Filters Yes
Deposit with SEM No (1)
Deposit with Cleaning Methods No (1)
XRD No
Effluent No

Comments

(1) Test section was not disassembled after experiment.

169
Figure C-50: Temperature measurements of test section vs. time for high pH & subcritical with
ferrous sulfate precursor. (Experiment #9)

Figure C-51: SEM photograph of particles collected on HTHP filter, 2.0k magnification, blank
run. (Experiment #9)

170
Figure C-52: SEM photograph of particles collected on LTHP filter, 5.0k magnification, blank
run. (Experiment #9)

Figure C-53: SEM photograph of particles collected on LTLP filter, 2.0k magnification, blank
run. (Experiment #9)

171
Appendix D: Filter Flow Rates

Table D-1: Flow rate through filter

Filter 1 2 3 4 5 6 7 8
Flow Rate [L/hr] HTHP 0.034 0.21 0.21 0.20 0.21 0.23 0.27 0.24
LTHP 3.506 3.33 3.39 3.339 3.285 3.265 3.195 3.255
LTLP 0.034 0.21 - 0.201 0.21 0.23 0.27 0.24
2
Area of Filter Sample [cm ] HTHP 0.71 0.71 0.71 0.71 0.71 0.71 0.71 0.71
LTHP 2.85 2.85 2.85 2.85 2.85 2.85 2.85 2.85
LTLP 4.91 4.91 4.91 4.91 4.91 4.91 4.91 4.91
Time of Injection [min] 42 39 44 40 40 37 40 400
2
Volume per Area [L/cm ] HTHP 0.033 0.191 0.216 0.188 0.196 0.199 0.252 2.244
LTHP 0.861 0.759 0.872 0.781 0.768 0.706 0.747 7.614
LTLP 0.005 0.028 - 0.027 0.029 0.029 0.037 0.326
Ratio LTHP/HTHP 25.8 4.0 4.0 4.2 3.9 3.6 3.0 3.4

172
Appendix E: Atomic Absorption Spectroscopy

The concentration of iron in each sample using AAS was determined from the calibration curve

found in Figure E-1and Figure E-2 which produced equations E-1 and E-2 respectively.

Figure E-1: Calibration curve for iron concentration using AAS, nitric acid matrix

E-1

where C is the concentration [mg/L] and A is the absorbance.

173
Figure E-2: Calibration curve for iron concentration using AAS, nitric acid and copper sulfate
matrix

E-2

The uncertainty in the thickness of the deposit was estimated from the uncertainty of several

sources in the calculation including tube length, calibration curve, background concentration, and

volume error.

Uncertainty from calibration curve: ±0.3 mg/L

Uncertainty from background iron concentration = ±0.1 mg/L

Uncertainty in volume = 5%

Uncertainty in timing, heating, loss of iron in transfer = 10%

174

You might also like