You are on page 1of 279

THE RATE OF OXYGEN EVOLUTION FROM AVIATION

TURBINE FUEL WITHIN AIRCRAFT FUEL TANKS

by

Adam Paul Harris

A thesis submitted in partial fulfilment of the requirements of the University of the West
of England, Bristol for the degree of Doctor of Philosophy

Faculty of Health and Life Sciences, University of the West of England, Bristol
February 2012

Abstract
Managing the effects of dissolved air evolution from aviation fuel has presented long-standing
issues for the design and operation of aircraft fuel systems. This phenomenon, known colloquially
as fuel outgassing, is responsible for a broad spectrum of fuel system issues, including; increased
fuel tank flammability, two-phase flow in pipes, fuel pump cavitation and fuel tank overpressurisation. The rate and effects of oxygen evolution from Jet A-1 aviation turbine fuel is
studied here using experimental techniques, dimensional modelling and aircraft flight testing. The
rate of fuel agitation present within a laboratory fuel tank was demonstrated to have the greatest
effect on the rate of oxygen evolution from the fuel.

Oxygen evolution rate increased

hyperbolically with increasing fuel agitation rate under pressure and temperature conditions
consistent with an aircraft fuel tank during flight. Dimensional modelling was used to estimate
the rate of oxygen evolution in an Airbus A320-200 aircraft fuel tank from measurements made
on a dimensionally similar laboratory model. The extrapolated rate of oxygen evolution from
similarity laws was found to be over 200% greater in the A320 inner wing fuel tank than that
measured in the laboratory model. Further work is required to validate the similarity laws of fuel
outgassing with flight test data if dimensional modelling is to be adopted for estimating fuel
outgassing rates in aircraft fuel tank flammability studies. Flight testing on an Airbus A340-300
aircraft revealed the effect of fuel outgassing on a nitrogen inerted Centre Wing Fuel Tank
(CWT) ullage to be minimal. CWT ullage oxygen concentration increased primarily due to
atmospheric air inspired via the vent system, resulting from a reducing fuel quantity in the CWT.
This unexpected result is believed to have been influenced by a combination of the fuels
tendency to absorb nitrogen from the ullage during CWT refuel, a large ullage to fuel ratio and
near quiescent CWT fuel conditions.
ii

In memory of the 230 passengers and flight crew who lost their lives in the TWA 800 accident on
July 17th 1996 over the Atlantic Ocean

iii

Acknowledgements
I would like to thank Prof. Norman Ratcliffe for his unending support and motivation. His advice,
guidance and enthusiasm aided the writing of this thesis in innumerable ways.

I would like to express my gratitude to Dr. Joseph Lam, Dr. Kevin Golden and Dr. Paul White for
their assistance and guidance with mathematical and statistical analysis of test data. I would also
like to thank Dr. Fritjof Korber for constructive feedback and words of wisdom during my
progression examination.

I gratefully acknowledge the support of Moreton Sandford and James Henshaw at Airbus, Filton
for their participation in stimulating and thought provoking technical discussion on the physics of
fuel outgassing. I would like to thank David Oram for his suggestions and profound advice on
tackling the mathematical challenges encountered within this work. My thanks are also extended
to Mike Wilson and Philippe Foucault at Airbus, Toulouse who made flight testing on an A340
aircraft a reality. In addition, I gratefully acknowledge the financial support provided by Airbus.

I will be forever indebted to the eminent Dr. Thomas Szirtes for introducing me to the fine art of
dimensional modelling and for sharing his skills developed over an illustrious career. I would like
to thank Penny Szirtes for her first rate secretarial and administrative support, which eased transAtlantic communications.

Finally, I would like to thank my wife, Katie, whose encouragement, support and enduring
patience has enabled me to focus the last five years of my life on the completion of this work.
iv

Table of Contents
Abstract........................................................................................................................................... ii
Acknowledgements ....................................................................................................................... iv
Chapter 1 Introduction.................................................................................................................. 1
1.1 Background ............................................................................................................................ 1
1.2 Scope of Research .................................................................................................................. 3
1.3 Thesis Layout ......................................................................................................................... 4
Chapter 2 The Behaviour of Dissolved Air in Aviation Turbine Fuel ...................................... 7
2.1 Introduction ............................................................................................................................ 7
2.2 Air Solubility in Aviation Fuel .............................................................................................. 7
2.2.1 Ostwald Coefficient ........................................................................................................ 8
2.2.2 Bunsen Coefficient.......................................................................................................... 8
2.2.3 Absorption Coefficient.................................................................................................... 8
2.3 Effect of Pressure on Air Solubility in Aviation Fuel .......................................................... 10
2.4 Effect of Temperature on Air Solubility in Aviation Fuel ................................................... 11
2.5 Composition of Air in Aviation Fuel ................................................................................... 19
2.5.1 Effect of Fuel Temperature on Dissolved Air Composition ......................................... 21
2.5.2 Composition of Air Released from Aviation Fuel ........................................................ 21
2.5.3 Effect of Fuel Properties on Air Solubility ................................................................... 23
2.6 Mechanisms of Dissolved Air Evolution ............................................................................. 25
2.6.1 Diffusive Mass Transfer to Gas Bubbles in Liquids ..................................................... 25
2.7 Conclusions .......................................................................................................................... 27
Chapter 3 The Effect of Dissolved Air Evolution on Aircraft Fuel Systems .......................... 29
3.1 Introduction .......................................................................................................................... 29
3.2 Fuel Tank Inerting Systems ................................................................................................. 30
3.2.1 Effect of Air Evolution on Inerting System Performance ............................................. 32
3.3 Engine Fuel Feed Systems ................................................................................................... 35
3.3.1 Functionality of the Engine Fuel Feed System ............................................................. 35
3.3.2 Effect of Air Evolution on Engine Fuel Feed ............................................................... 38
3.3.3 Vapour/Liquid Ratio ..................................................................................................... 38
3.3.4 Effect of Air Evolution on Engine Backing Pumps ...................................................... 41
v

3.4 Fuel Quantity Indication Systems (FQIS) ............................................................................ 43


3.5 Vent Systems ....................................................................................................................... 44
3.5.1 Effect of Air Evolution on Vent System Performance.................................................. 45
3.6 Conclusions .......................................................................................................................... 49
Chapter 4 Measurement of Air, Oxygen and Nitrogen in Aviation Fuels .............................. 51
4.1 Introduction .......................................................................................................................... 51
4.2 Physical Methods ................................................................................................................. 52
4.3 Chemical Methods ............................................................................................................... 54
4.3.1 Gas Chromatographic Methods .................................................................................... 54
4.3.2 Electrochemical Methods .............................................................................................. 56
4.4 Optical Sensing Methods ..................................................................................................... 58
4.4.1 Optical Absorption Spectroscopy ................................................................................. 58
4.4.2 Measurement Principle ................................................................................................. 59
4.4.3 Luminescence Quenching ............................................................................................. 60
4.4.4 Measurement Principle ................................................................................................. 60
4.5 Conclusions .......................................................................................................................... 62
Chapter 5 Development of an Experimental Apparatus to Determine Oxygen Evolution
Rate from Aviation Turbine Fuel ............................................................................................... 64
5.1 Preliminary Considerations .................................................................................................. 64
5.1.1 Preliminary Tests .......................................................................................................... 65
5.2 Experimental Design ............................................................................................................ 66
5.2.1 Overall Assembly.......................................................................................................... 66
5.2.2 Thermal-Altitude Test Chamber ................................................................................... 68
5.2.3 Fuel Tank ...................................................................................................................... 69
5.2.4 Fuel Agitation ............................................................................................................... 71
5.2.5 Impeller Pumping Capacity .......................................................................................... 75
5.2.6 Impeller Pumping Capacity in Aviation Fuel ............................................................... 81
5.2.7 Fuel Sparging ................................................................................................................ 85
5.2.8 Oxygen Sensing ............................................................................................................ 87
5.2.9 Tunable Diode Laser Absorption Spectroscopy (TDLAS) Oxygen Sensor ................. 88
5.2.10 Polarographic Oxygen Sensor ..................................................................................... 92
5.3 Simulation of Aircraft Fuel Tank Pressures ......................................................................... 94
vi

5.3.1 Climb Profiles ............................................................................................................... 95


5.3.2 Pressure Profiles............................................................................................................ 96
5.4 Oxygen Sensor Measurement Uncertainty .......................................................................... 99
5.4.1 Measurement Uncertainty Calculation Methods ........................................................... 99
5.5 Summary ............................................................................................................................ 105
Chapter 6 Experimental Examination on the Rate of Oxygen Evolution from Aviation
Turbine Fuel ............................................................................................................................... 106
6.1 Introduction ........................................................................................................................ 106
6.2 Experimental Design .......................................................................................................... 106
6.3 Experimental Method......................................................................................................... 107
6.3.1 Repeatability of Experimental Method ....................................................................... 109
6.3.2 Calculation of Oxygen Evolution Rates...................................................................... 110
6.3.3 ASTM D2779-92 Oxygen Solubility Estimates ......................................................... 110
6.3.4 Non-Linear Regression Analysis ................................................................................ 112
6.4 Experimental Results ......................................................................................................... 115
6.4.1 Measurement of Oxygen Partial Pressures ................................................................. 115
6.4.2 Rate of Oxygen Evolution Results Fuel Agitation Rate .......................................... 119
6.4.3 Rate of Oxygen Evolution Results Fuel Temperature ............................................. 121
6.4.4 Rate of Oxygen Evolution Results Rate of Ullage Pressure Change ....................... 124
6.4.5 Time Constant of Oxygen Evolution Results ............................................................. 126
6.5 Statistical Analysis of Test Data ........................................................................................ 128
6.6 Discussion of Results ......................................................................................................... 130
6.6.1 Effect of Fuel Agitation Rate ...................................................................................... 130
6.6.2 Limitations of the Reciprocal Model .......................................................................... 139
6.6.3 Statistical Significance of Fuel Agitation Rate ........................................................... 140
6.6.4 Effect of Fuel Temperature ......................................................................................... 140
6.6.5 Effect of Rate of Change of Ullage Pressure .............................................................. 145
6.6.6 Time Constant of Oxygen Evolution, ...................................................................... 148
6.7 Conclusions ........................................................................................................................ 151
Chapter 7 Dimensional Modelling of Oxygen Evolution Rate from Aviation Turbine Fuel in
Aircraft Fuel Tanks ................................................................................................................... 153
7.1 Introduction ........................................................................................................................ 153
vii

7.2 Dimensional Similarity and Modelling .............................................................................. 154


7.3 Identifying Relevant Variables .......................................................................................... 155
7.3.1 Consideration of Other Physical Variables ................................................................. 158
7.4 The Dimensional Method .................................................................................................. 160
7.4.1 Sequence of Physical Variables in the Dimensional Set ............................................. 161
7.4.2 Scale Factors ............................................................................................................... 161
7.4.3 The Model Law ........................................................................................................... 162
7.5 Model Design ..................................................................................................................... 163
7.5.1 Flight Test Data........................................................................................................... 163
7.5.2 A320 Engine Fuel Feed System Performance Analysis ............................................. 165
7.5.3 Model Fuel Agitation Rate (2) .................................................................................. 168
7.5.4 Model Rate of Ullage Pressure Change ( p 2 ) ............................................................. 170
7.6 Model Tests........................................................................................................................ 172
7.6.1 Surface Tension Measurements .................................................................................. 173
7.6.2 Measurement of Oxygen Partial Pressures in the Model ............................................ 174
7.7 Results ................................................................................................................................ 176
7.7.1 Rate of Oxygen Mass Release .................................................................................... 176
7.8 Discussion .......................................................................................................................... 179
7.8.1 Oxygen Evolution Rate Measurements from Dimensionally Dissimilar Models ....... 180
7.8.2 Reducing the Number of Dimensionless Variables Fusion of Physical Variables .. 183
7.8.3 Altering Variable Sequence in the Dimensional Set ................................................... 186
7.8.4 Relationship between Dimensionless Variables ......................................................... 187
7.8.5 Analytical Sensitivity Analysis ................................................................................... 190
7.9 Incorrect Determination of Impeller Rotational Speed ...................................................... 194
7.10 Conclusions ...................................................................................................................... 197
Chapter 8 Effect of Fuel Outgassing on Oxygen Concentration in an Aircraft Fuel Tank
Ullage........................................................................................................................................... 200
8.1 Introduction ........................................................................................................................ 200
8.2 Fuel Tank Flammability Assessment Method ................................................................... 201
8.2.1 FRM Performance Degradation due to Fuel Outgassing ............................................ 203
8.2.2 FTFAM Oxygen Evolution Time Constants ............................................................... 203
8.3 Flight Test Aircraft A340-300 MSN001 ......................................................................... 204
viii

8.3.1 A340-300 CWT Fuel System ...................................................................................... 205


8.4 Ullage Oxygen Concentration Measurement ..................................................................... 207
8.4.1 Flight Test F1437 ........................................................................................................ 212
8.5 Results from Flight Test F1437 ......................................................................................... 212
8.6 Discussion .......................................................................................................................... 217
8.6.1 Effect of Air Evolution on Inerted CWT Ullage ......................................................... 218
8.6.2 Validation of Fuel Outgassing Model Law ................................................................. 221
8.7 Conclusions ........................................................................................................................ 223
Chapter 9 Conclusions and Future Work................................................................................ 225
9.1 Conclusions ........................................................................................................................ 225
9.1.1 Measurement of Oxygen Evolution Rates .................................................................. 225
9.1.2 Effect of Aircraft Fuel Tank Environment on Oxygen Evolution Rate ...................... 226
9.1.3 Time Constant of Oxygen Evolution .......................................................................... 227
9.1.4 Dimensional Modelling of the Fuel Outgassing Phenomenon ................................... 228
9.1.5 Effect of Fuel Outgassing on an Inerted Aircraft Fuel Tank Ullage ........................... 228
9.2 Future Work ....................................................................................................................... 230
9.2.1 Experimental Set-up and Further Testing ................................................................... 230
9.2.2 Validation of Fuel Outgassing Model Law ................................................................. 231
9.2.3 Dissolved Oxygen Sensor Development..................................................................... 232
Appendix A Rolls Royce Engine Fuel Feed Pipe Arrangements used in V/L Measurements
..................................................................................................................................................... 244
Appendix B Sizing Calculation for Fuel Tank Sparger Surface Area .................................. 245
Appendix C Performance Calculations for In-line Fuel Vapour Condenser ....................... 248
Appendix D Conversion of RAE Oxygen Release Rates ........................................................ 254

ix

List of Figures
Figure 2-1 Ostwald solubility coefficients for air gases in aviation turbine fuels as a function of
temperature (Coordinating Research Council, 2004, Section 2 pg. 42)......................................... 12
Figure 2-2 Gas chromatography measurements of air solubility in Avtur aviation fuel as a
function of fuel temperature (Ross, 1970, p.29) ............................................................................ 13
Figure 2-3 Clausius-Clapeyron equation predictions of Ostwald solubility coefficients compared
with CRC solubility data for oxygen dissolved in Jet A-1 aviation fuel as a function of
temperature .................................................................................................................................... 17
Figure 3-1 Generic On Board Inert Gas Generation System (OBIGGS) schematic (Langton,
Clark, Hewitt and Richards, 2009, p.228)...................................................................................... 31
Figure 3-2 Effect of dissolved air evolution from aviation kerosene on the oxygen concentration
of an on demand nitrogen purged fuel tank ullage as determined by (Gatward and Wyeth, 1951,
p.13) ............................................................................................................................................... 34
Figure 3-3 Schematic of a twin engine fuel feed system of a commercial transport aircraft
(Langton, Clark, Hewitt and Richards, 2009, p.67) ....................................................................... 36
Figure 3-4 Photograph of the Eaton Aerospace FRH280002 air release valve incorporated within
the engine fuel feed gallery of the Airbus A340-500/600 aircraft (Eaton Corporation, 2004) ...... 37
Figure 3-5 Simplified engine fuel feed system schematic (Langton, Clark, Hewitt and Richards,
2009, p.10) ..................................................................................................................................... 38
Figure 3-6 Vapour/liquid ratio in the horizontal feed pipe as a function of altitude (Wright, 1976,
p.16) ............................................................................................................................................... 40
Figure 3-7 Effect of engine fuel feed pipe geometry on vapour/liquid ratio as a function of
altitude (Wright, 1976, p.16).......................................................................................................... 41
Figure 3-8 Wing vent arrangement of the Airbus A340-600 aircraft (Langton, Clark, Hewitt and
Richards, 2009, p.46) ..................................................................................................................... 45
Figure 3-9 Maximum tank pressure rise and fuel volume carry-over as a function of vent pipe ID
from tests conducted by (Ross, 1972, p.21) ................................................................................... 48
Figure 3-10 Calculated rate of air evolution and maximum tank pressure rise from Avtur aviation
turbine fuel carryover measured by (Ross, 1972, p.21) ................................................................. 48
Figure 4-1 Air solubility measurement apparatus schematic developed by Cann of the
RAE(Cansdale, 1978, p.21) ........................................................................................................... 53
x

Figure 4-2 Anatomy of an electrochemical oxygen sensor based on the polarographic design
(Hach Lange, 2009, p.15) .............................................................................................................. 56
Figure 4-3 Diagrammatic description of the TDLAS measurement principal (Frish, 2004, p.5) .. 60
Figure 5-1 Photograph of pilot apparatus set-up ............................................................................ 66
Figure 5-2 Schematic & photograph of the small-scale experimental fuel tank set-up and oxygen
evolution rate measurement apparatus ........................................................................................... 67
Figure 5-3 Weiss WK/1000/70-100/DS ATEX certified thermal-altitude test chamber with 1m3
internal volume .............................................................................................................................. 69
Figure 5-4 Range of fuel depth within the Airbus A380-800 wing fuel tanks at maximum fuel
capacity (LH wing-box shown) ..................................................................................................... 70
Figure 5-5 Experimental fuel tank dimensions (all dimensions in mm) ........................................ 71
Figure 5-6 Closed-loop PID controlled mixing impeller system architecture ............................... 74
Figure 5-7 IKA R1373 mixing impeller rotational speed in Jet A-1 fuel at 20C achieved by the
pneumatically driven impeller mixing system over the controllable airflow range ....................... 75
Figure 5-8 Dimensions of the IKA R1373 paddle type mixing impeller ....................................... 77
Figure 5-9 Radial flow pattern of a paddle type mixing impeller in a liquid tank ......................... 78
Figure 5-10 Photograph of the experimental set-up to measure impeller torque using a Brookfield
R/S Rheometer and water tank of identical internal dimensions to the experimental fuel tank .... 79
Figure 5-11 IKA R1373 impeller torque as a function of rotational speed measured using a
Brookfield R/S Rheometer in water at 17C with power-law regression fit and extrapolated
impeller torque values at 450 and 500 rpm .................................................................................... 80
Figure 5-12 Dimensionless plot of Np as a function of Re number for the IKA R1373 impeller
running in water at 17C over a 50 rpm to 400 rpm speed range .................................................. 83
Figure 5-13 RP4 paddle type impeller with Npstd = 3.4 and Nqstd = 0.62 (Post, 2011) .................. 83
Figure 5-14 IKA R1373 impeller pumping capacity as a function of impeller Re number running
in water at 17C ............................................................................................................................. 84
Figure 5-15 Calculated fuel agitation rate as a function of impeller Re number over a 0C to 40C
fuel temperature range (agitation rate of water shown for comparison) ........................................ 85
Figure 5-16 Photographs of the air metering system for in-tank sparging of aviation turbine fuel86
Figure 5-17 Re-saturation curve for dissolved air in Jet A-1 aviation fuel whilst agitating with the
mixing impeller at 300 rpm during sparging ................................................................................. 87
Figure 5-18 Schematic of Oxigraf O2G1 TDLAS oxygen sensing system architecture ............... 90
xi

Figure 5-19 Photograph of the Oxigraf O2G1 TDLAS oxygen sensor used to measure oxygen
concentration in the experimental fuel tank ullage ........................................................................ 91
Figure 5-20 Photograph of the in-line condenser designed to strip hydrocarbon fuel vapours from
the gas sample train of the Oxigraf O2G1 TDLAS oxygen sensor (exterior insulation removed) 92
Figure 5-21 Orbisphere polarographic oxygen sensor installation geometry within the
experimental fuel tank (all dimensions in mm) ............................................................................. 93
Figure 5-22 Climb rate performance envelope of the Boeing 777-200 aircraft at altitudes up to
40,000 ft (James and ODell, 2005, p.3) ........................................................................................ 95
Figure 5-23 Linear experimental climb profiles with rates ranging from 1000 ft/min to 5500
ft/min .............................................................................................................................................. 96
Figure 5-24 Pressure vs. time profiles developed to simulate the effect of aircraft altitude .......... 97
Figure 5-25 Fuel tank vent valve functionality at a climb rate of 2500 ft/min and 75% fuel load 98
Figure 5-26 Photograph of the Orbisphere polarographic oxygen sensor, compensation pressure
sensor and oxygen calibration gas supply hose inside the plastic bag ......................................... 101
Figure 5-27 Oxygen concentration measurement plot for the TDLAS oxygen sensor and oxygenin-nitrogen calibration gas certified to 5.04 % oxygen by volume .............................................. 102
Figure 5-28 Measured oxygen concentrations in certified oxygen-in-nitrogen calibration gases for
the polarographic and TDLAS oxygen sensors demonstrating a high level of linearity over a 5%
to 35% oxygen concentration range ............................................................................................. 103
Figure 6-1 Partial pressure of dissolved oxygen measured in Jet A-1 fuel as a function of time for
three separate tests conducted under identical environmental conditions to establish experimental
repeatability where =0.96707 kg/s, p =495.1 Pa/s and fuel temperature =20C ...................... 109
Figure 6-2 Mass of oxygen released from Jet A-1 fuel as a function of time calculated using
ASTM D2779-92 for three separate tests conducted under identical environmental conditions
where =0.96707 kg/s, p = 495.1 Pa/s and fuel temperature =20C ........................................... 112
Figure 6-3 Instantaneous mass release rates of oxygen from Jet A-1 fuel as a function of time for
three separate tests conducted under identical environmental conditions in the experimental fuel
tank where=0.96707 kg/s, p = 495.1 Pa/s and fuel temperature =20C ................................... 114
Figure 6-4 Oxygen evolution from Jet A-1 aviation fuel in the experimental fuel tank at a
temperature of 0C, fuel agitation rate of 0.355 kg/s and a 295.64 Pa/s rate of ullage pressure
change .......................................................................................................................................... 115
xii

Figure 6-5 Dissolved and ullage oxygen partial pressures as a function of time in the experimental
fuel tank at a temperature of 0C, fuel agitation rate of 0.355 kg/s and a 295.64 Pa/s rate of ullage
pressure change ............................................................................................................................ 117
Figure 6-6 Oxygen mass released from Jet A-1 fuel and Janoschek regression model fit as a
function of time in the experimental fuel tank at a temperature of 0C, fuel agitation rate of 0.355
kg/s and a 295.64 Pa/s rate of ullage pressure change ................................................................. 118
Figure 6-7 Instantaneous oxygen mass release rate as a function of time for Jet A-1 fuel in the
experimental fuel tank at a temperature of 0C, fuel agitation rate of 0.355 kg/s and a 295.64 Pa/s
rate of ullage pressure change ...................................................................................................... 118
Figure 6-8 Oxygen release rate at t = as a function of fuel agitation rate at a fuel temperature of
20C and rate of change of ullage pressure ranging from 73.85 to 405.75 Pa/s .......................... 120
Figure 6-9 Oxygen release rate at t = as a function of fuel agitation rate at a fuel temperature of
40C and rate of change of ullage pressure ranging from 73.85 to 405.75 Pa/s .......................... 120
Figure 6-10 Oxygen release rate at t = as a function of fuel agitation rate at a fuel temperature
of 0C and rate of change of ullage pressure ranging from 73.85 to 405.75 Pa/s ........................ 121
Figure 6-11 Oxygen release rate at t = as a function of fuel temperature and a 73.85 Pa/s rate of
change of ullage pressure ............................................................................................................. 122
Figure 6-12 Oxygen release rate at t = as a function of fuel temperature and a 184.86 Pa/s rate
of change of ullage pressure ........................................................................................................ 122
Figure 6-13 Oxygen release rate at t = as a function of fuel temperature and a 295.64 Pa/s rate
of change of ullage pressure ........................................................................................................ 123
Figure 6-14 Oxygen release rate at t = as a function of fuel temperature and a 405.75 Pa/s rate
of change of ullage pressure ........................................................................................................ 123
Figure 6-15 Oxygen release rate at t = as a function of the rate of ullage pressure change at a
fuel temperature of 20C.............................................................................................................. 124
Figure 6-16 Oxygen release rate at t = as a function of the rate of ullage pressure change at a
fuel temperature of 0C................................................................................................................ 125
Figure 6-17 Oxygen release rate at t = as a function of the rate of ullage pressure change at a
fuel temperature of 40C.............................................................................................................. 125

xiii

Figure 6-18 Semi-log transformation and linear regression fit of oxygen release rate as a function
of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage pressure of
295.64 Pa/s ................................................................................................................................... 131
Figure 6-19 Log-log transformation and linear regression fit of oxygen release rate as a function
of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage pressure of
295.64 Pa/s ................................................................................................................................... 132
Figure 6-20 Reciprocal transformation and linear regression fit of oxygen release rate as a
function of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage pressure
of 295.64 Pa/s .............................................................................................................................. 132
Figure 6-21 Reciprocal model fit to oxygen release rate multiplied by 106 as a function of fuel
agitation rate at a fuel temperature of 40C and a rate of change of ullage pressure of 295.64 Pa/s
..................................................................................................................................................... 133
Figure 6-22 Oxygen release rate 106 plotted as a function of impeller power at a fuel
temperature of 40C and a rate of change of ullage pressure of 295.64 Pa/s .............................. 135
Figure 6-23 Reciprocal of oxygen release rate 106 plotted as a function of impeller power at a
fuel temperature of 40C and a rate of change of ullage pressure of 295.64 Pa/s ....................... 136
Figure 6-24 Reciprocal model fit to oxygen release rate multiplied by 106 as a function of
impeller power at a fuel temperature of 40C and a rate of change of ullage pressure of 295.64
Pa/s ............................................................................................................................................... 137
Figure 6-25 Experimental data determined by the Shell Oil company for the volumetric release
rate of air from aviation kerosene at varying levels of fuel agitation (Ross, 1972, p.22) ............ 139
Figure 6-26 Oxygen diffusion coefficient in Jet A-1 over a 0C to 40C temperature range
estimated from the Wilke-Chang correlation for diffusion of dilute gases in liquids .................. 141
Figure 6-27 Rate constant of nitrogen gas evolution as a function of fuel temperature in RJ-1 fuel
subjected to horizontal shaking (Coordinating Research Council, 1958, Figure 39) .................. 142
Figure 6-28 Diffusivity of oxygen in Jet A-1 aviation fuel as a function of fuel viscosity over a
0C to 40C temperature range .................................................................................................... 145
Figure 6-29 % air by vol. at STP dissolved in aviation gasoline as a function of time in agitated
and unagitated conditions at 43C (110F) (Beal, Hilburger and Porter, 1945) .......................... 147
Figure 6-30 Comparison of the half-life of oxygen evolution from oxygen saturated RJ-1 fuel
with for oxygen evolution from Jet A-1. Both fuels were subjected to agitation by either shaking
or stirring at a temperature of ~20C ........................................................................................... 150
xiv

Figure 7-1 The Dimensional Set for oxygen evolution rate from aviation turbine fuel............... 160
Figure 7-2 Flight test data from flight No.0073 from the A320-200 MSN 659 aircraft (Bonjour,
2009, p.14) ................................................................................................................................... 164
Figure 7-3 Fuel tank configuration of the Airbus A320-200 aircraft (Walker, 2005, p.5) .......... 165
Figure 7-4 Schematic of the A320-200 engine fuel feed system architecture (left hand wing and
centre tank shown) (Walker, 2005, p.38) ..................................................................................... 167
Figure 7-5 Photograph of an A320-200 aircraft inner wing collector cell illustrating fuel boost
pumps and sequence valves from which fuel is ejected resulting in collector cell fuel agitation.
View is from the wing front spar looking aft towards RIB 1....................................................... 168
Figure 7-6 IKA R1373 mixing impeller pumping capacity as a function of impeller rotational
speed ............................................................................................................................................ 169
Figure 7-7 Pressure as a function of time curve generated from Equation 7-11 and exponential
regression equation fit for the A320-200 aircraft climb to 11582.4 m in flight No.0073 ............ 171
Figure 7-8 Ullage pressure and altitude as a function of time profiles for A320-200 aircraft (flight
No.0073) and model determined from flight test data and the Model Law ................................. 172
Figure 7-9 Physical model test set-up .......................................................................................... 174
Figure 7-10 Mass of oxygen released from Jet A-1 aviation fuel as a function of time in the model
tests .............................................................................................................................................. 176
Figure 7-11 Instantaneous mass release rates of oxygen from Jet A-1 aviation fuel as a function of
time in 3 separate model tests with identical test conditions ....................................................... 177
Figure 7-12 Oxygen partial pressures as a function of time due to outgassing of dissolved air from
aviation turbine fuel under reduced ullage pressure and agitated fuel conditions during Test 2. 184
Figure 7-13 Reconstruction of the Dimensional Set with variables pu and pf combined (pf pu) . 185
Figure 7-14 Re-arranged variables in the Dimensional Set ......................................................... 186
Figure 7-15 Dimensionless plot of 1 as a function of 3 using data from Test 1 in the ullage
depressurisation phase ................................................................................................................. 189
Figure 7-16 Relationship between (pf -pu) and p represented by a linear mathematical function 189

O2 in terms of p during ullage


Figure 7-17 An empirical model describing the behaviour of m
depressurisation in Test 1............................................................................................................. 190
Figure 7-18 Absolute sensitivity of oxygen evolution rate resulting from a 1 to 10% change in the
rate of change of ullage pressures simulated in the dimensional modelling tests ........................ 192
xv

Figure 7-19 Relative sensitivity of oxygen evolution rate resulting from a 1 to 10% change in the
rate of change of ullage pressure simulated in the dimensional modelling tests ......................... 193
Figure 7-20 Pumping capacity of IKA R1373 impeller running in Jet A-1 aviation fuel at 20C as
a function of impeller Re number ................................................................................................ 194
Figure 8-1 Flow chart illustrating the main computations used in the Fuel Tank Flammability
Assessment Method (FTFAM) and FRM performance (Summer, 2008, p.7) ............................. 202
Figure 8-2 Airbus A340-300 MSN001 flight test aircraft (Mervelet, 2009) ............................... 204
Figure 8-3 Schematic of A340-300 CWT fuel system architecture (Airbus Technical Description
Volume 3A System AI/ED-N 433.0023/91 Iss. 6) ...................................................................... 205
Figure 8-4 A340-300 architecture (Airbus Technical Description Volume 3A System AI/ED-N
433.0023/91 Iss. 6)....................................................................................................................... 206
Figure 8-5 Oxigraf O2N2 flight test oxygen analyser and O2G8 gas sampling system architecture
(schematic courtesy of Oxigraf Inc.) (McCaul, 2007, p.29) ........................................................ 207
Figure 8-6 Photograph of the Oxigraf O2N2 flight test ullage oxygen analyser and O2G8 oxygen
sampling system installation within the Airbus A340-300 MSN001 flight test aircraft cabin .... 208
Figure 8-7 Cross-section of the A340-300 MSN001 centre wing fuel tank (Left-hand side)
showing ullage oxygen concentration and temperature sensing locations (all dimensions in mm)
(Boulet, 2009, p.4) ....................................................................................................................... 210
Figure 8-8 Illustration of ullage oxygen gas sampling points - float valve arrangements in the
centre wing fuel tank of the A340-300 MSN001 flight test aircraft (Boulet, 2009, p.5) ............. 211
Figure 8-9 Illustration of ullage oxygen gas sample return points - float valve arrangement at R10
within the centre wing fuel tank of the A340-300 MSN001 flight test aircraft(Boulet, 2009, p.6)
..................................................................................................................................................... 211
Figure 8-10 Flight test data from flight No. F1437 showing ullage oxygen concentration
measurements (left hand side) and corresponding aircraft altitude, CWT fuel quantity and fuel
temperature (right hand side) ....................................................................................................... 215
Figure 8-11 Effect of fuel transfer from CWT to inner feed tanks on CWT ullage oxygen
concentration during flight no. F1437. Fuel temperature and altitude data removed for clarity . 216
Figure 8-12 Comparison of theoretical CWT ullage oxygen concentration increase resulting from
air inspired via the vent system due to fuel transfer with measured ullage oxygen concentration.
Fuel temperature and altitude data removed for clarity ............................................................... 219

xvi

Figure 8-13 The effect of a nitrogen enriched ullage on the dissolved oxygen content of mildly
agitated Jet A-1 fuel at ambient temperature and barometric pressure ........................................ 221

xvii

List of Tables
Table 2-1 A summary of air, oxygen and nitrogen gas solubility measurements in aviation fuels
reported by various workers at an air partial pressure of 760 mmHg. Derry, Evans, Faulkner and
Jelfs (1952), Logvinyuk, Makarenkov, Malyshev and Panchenkov (1970), Astafev and Kozinova
(1988), Schweitzer and Szebehely (1950) ..................................................................................... 10
Table 2-2 Compositions of air dissolved in aviation fuels as determined at 25C using Gas
Chromatography (Ross, 1970, p.25) .............................................................................................. 20
Table 3-1 Effect of vent pipe ID on fuel tank pressurisation and fuel carryover due to air release
from supersaturated Avtur aviation turbine fuel subjected to agitation and a simulated altitude of
60,000 ft (Ross, 1972, p.21) ........................................................................................................... 47
Table 5-1 Outline specification of Weiss WK 1000/70-100/DS thermal-altitude test chamber .... 68
Table 5-2 IKA R1373 impeller pumping capacity from measured torque values in water at 17C.
Data values shown in red are calculated from extrapolated torque data ........................................ 82
Table 5-3 Fuel agitation rates calculated over a 200 rpm to 450 rpm impeller rotational speed
range............................................................................................................................................... 84
Table 5-4 Rate of change of ullage pressure for each climb profile .............................................. 97
Table 5-5 Certified oxygen-in-nitrogen calibration gas mixtures and associated standard
uncertainties assuming a normal distribution............................................................................... 103
Table 5-6 Uncertainty of measurement analysis of the TDLAS oxygen sensor over a 5% to 35%
oxygen by vol. concentration measurement range ....................................................................... 104
Table 5-7 Uncertainty of measurement analysis of the polarographic oxygen sensor over a ...... 105
Table 6-1 Experimental test conditions ....................................................................................... 107
Table 6-2 Time constants of oxygen evolution over the experimental ranges of fuel agitation, fuel
temperature and rate of change of ullage pressure ....................................................................... 127
Table 6-3 Two-way ANOVA results for oxygen release rate test data ....................................... 129
Table 6-4 Two-way ANOVA results for data ........................................................................... 129
Table 6-5 Calculated impeller power dissipated to the Jet A-1 aviation fuel at 40C over the range
of fuel agitation rates investigated ............................................................................................... 134
Table 7-1 Variables relevant to the rate of oxygen evolution from aviation turbine fuel using the
SI dimensional system ................................................................................................................. 156

xviii

Table 7-2 Summary of flight test data taken from flight No.0073 on MSN 659 required for design
of the laboratory model (Bonjour, 2009, p.14) ............................................................................ 165
Table 7-3 Collector cell fuel agitation rate calculated from A320-200 MSN 659 flight test data
(flight No.0073) and analysis of engine fuel feed system performance parameters .................... 166
Table 7-4 Surface tension measurements of Jet A-1 aviation fuel used in the model tests
compared to published CRC Fuels Properties Handbook surface tension values for Jet A1(Intertek, 2011) .......................................................................................................................... 173
Table 7-5 Instantaneous oxygen mass release rates at t = and time constants of oxygen
evolution measured from Jet A-1 fuel in the dimensionally similar model tests ......................... 178
Table 7-6 Summary of dimensional modelling results on the rate of oxygen evolution from
aviation fuel in an A320 aircraft fuel tank ................................................................................... 179
Table 7-7 Oxygen release rates measured by the RAE under simulated flight conditions in
quiescent aviation kerosene using a dimensionally dissimilar laboratory model in comparison
with rates of oxygen release measured in the dimensionally similar model fuel tank (Bedwell,
1952, p.10) ................................................................................................................................... 181

xix

Nomenclature
Upper Case
1, m2

Constant, Condenser surface area

Bunsen solubility coefficient

Bubble diameter, Impeller diameter

DBA

Gas-liquid diffusivity

m2/s

Activation energy

J/mol

Molar volume of solvent

Solubility of Oxygen

Luminescence intensity in the presence of oxygen

cd

I0

Luminescence intensity in the absence of oxygen

cd

Transmitted intensity

W/m2

I,0

Initial laser intensity

W/m2

Rate constant of oxygen evolution from fuel

s-1

Ostwald solubility coefficient

Lo

Ostwald solubility coefficient at 0C and 850 kg/m3 density

Lc

Corrected Ostwald solubility coefficient

Mass flow rate of condensate

Mg

Molecular weight

(NA)avg

m3/mol

kg/s
kg/mol

Target gas number density, Impeller rotational speed


Molar rate of gas diffusion from liquid to gas bubble
averaged over bubble surface area

molecule/m3,
rev/min, s-1
m2/s

Np

Impeller power number

Nq

Impeller flow number

Npstd

Ref. Impeller power number

Nqstd

Ref. Impeller flow number

Number of dimensionless variables

Nd

Number of fundamental dimensions

O2,i

% volume fraction of oxygen at time step, i

xx

O2

% volume fraction of oxygen in ullage

Gas Pressure, Power in stirred tank

P(a)

Pressure at altitude

Pa

P1D

Partial pressure of dissolved air in fuel upstream of fuel pipe

Pa

Partial pressure of dissolved air in fuel downstream of fuel

Pa

P2D

1
Pa, W

pipe

P2

Pressure in pipe

Pa

Ps

Fuel vapour pressure

Pa

PTVP

True vapour pressure of fuel

Pa
3

m /s, m2kg/s3

Volumetric flow rate, Rate of heat transfer

Qimp

Volumetric impeller flow

m3/s

Qm

Mass flow rate

kg/s

Gas constant

Re

Impeller Reynolds number

%S

Percentage solubility

S(T)

Temperature dependent line strength

Sc

Schmidt number

Sp

Percentage air solubility

Temperature

Tfilm

Condensate film temperature

Tsat

Vapour saturation temperature

Twall

Condenser wall temperature

Temperature difference between condenser wall and

T0

J/kgK

m-1/moleculem-2

saturated vapour

Tq

Impeller torque

V/L
~
VA

Vapour/liquid ratio

Nm
1
m3/mol

Molar volume of solvent

xxi

Lower Case
a

Constant

Constant

Constant

cA0

Solubility of gas A in solvent B

Liquid density, Diameter

Acceleration due to gravity

m/s2

g(-v0)

Frequency dependence of the line strength

cm

hcond

Condensing heat transfer coefficient

Time step

kg/m3, m

W/m2K
s

Constant, Solubility coefficient, Boltzmann constant, Stern-Volmer


constant, Rate constant of oxygen evolution

1,1,J/K,
W/m2K

kf

Thermal conductivity of condensate film

kH

Henrys Law constant

Pa

Constant

Condensation rate

mf

Mass of fuel in tank

kg

mO2(t)

Mass of oxygen dissolved in fuel at time, t

kg

m O2

Rate of oxygen evolution from fuel

mO2(t)

Mass of oxygen released from fuel into ullage at time, t

kg

Number of measurements in set

Gas partial pressure, Significance level, Shaping parameter

pO2

Partial pressure of oxygen

Pa

pf

Partial pressure of oxygen dissolved in fuel

Pa

pt

Total pressure in ullage

pu

Partial pressure of oxygen in ullage

pv

Vapour pressure of liquid, Vapour density

kg/s

kg/s

Pa, mmHg,1

Pa, mbar

Rate of change of ullage pressure

(pf pu)

Delta oxygen partial pressure between fuel and ullage

Radius of solute molecule, Volume fraction

Standard deviation

Pa
Pa, kg/m3
Pa/s
Pa
m, 1
1

xxii

Fuel Temperature

K,C,F

Standard uncertainty of measurement

uc

Combined standard uncertainty of measurement

Altitude

Greek Letters

Bunsen solubility coefficient, Fuel agitation rate

Absorption coefficient

Condensate loading

Surface tension

1, kg/s
1
kg/sm-1
N/m

Partial eta square (Effect size)

1/2

Half-life of oxygen evolution

Heat of vapourisation

Fuel viscosity

Laser frequency

s-1

Dimensionless variable

Fuel density, Gas density

Surface tension

N/m

Fuel surface tension

N/m

Time constant of oxygen evolution

Terminal bubble rise velocity

m/s

Association factor of solvent

J/kg
kg/(ms)

kg/m3

xxiii

Chapter 1
Introduction
1.1 Background
The tendency of dissolved air to emerge from aviation fuels during aircraft flight has presented
arduous fuel system design challenges since the advent of jet-propelled aircraft. During an
aircrafts ascent to cruising altitude, reducing atmospheric pressure above the fuels surface
promotes the release of dissolved air from the fuel. This phenomenon, known as fuel outgassing,
is closely related to the environmental conditions present within the aircraft fuel tank. The rate at
which air is evolved is governed by several key variables, namely; the degree of fuel agitation,
the rate of change of atmospheric pressure, degree of air supersaturation, air solubility and fuel
temperature. Under near quiescent fuel tank conditions the rate of air evolution is reported to be
slow where the fuel is highly supersaturated with air (Schweitzer and Szebehely, 1950). However,
operating fuel pumps and other fuel system equipment in air supersaturated fuel can produce
significant levels of fuel agitation where air release rates are high, causing a foam layer on the
fuels surface (Poulston and Thomas, 1959). Fuel outgassing can seriously impact the
performance and safe operation of an aircraft fuel system. Entrained air within the fuel flow to an
engine can lead to a condition known as vapour lock, which, if sustained, can result in fuel pump
cavitation and loss of engine operation (Beal, Hilburger and Porter, 1945; SAE International,
1997). Air evolution from supersaturated fuel has been responsible for over-pressurisation of fuel
tanks where tank venting systems have been designed with insufficient margin to accommodate

Supersaturation defines the condition when the amount of gas dissolved in a liquid is greater than the
amount which would correspond to the gas pressure above the liquids surface

air released at a high rate from the fuel (Beal, Hilburger and Porter, 1945; Lee 1974). The sudden
release of air, comprising up to ~35% oxygen by volume, can lead rapidly to the formation of
flammable fuel-air mixtures in the ullage of aircraft fuel tanks (Bragg, Kimmel and Jones, 1969).
Even the protection provided by nitrogen gas inerting systems, designed to mitigate the risk of
aircraft fuel tank explosion, can be compromised (Gatward and Wyeth, 1951; McConnell, Dalan
and Anderson, 1986; Cavage, 2005). Despite ongoing design efforts to manage the effects of fuel
outgassing, many problems still exist and arise during aircraft service. Previous studies on the
fuel outgassing problem have failed to examine the synergistic effects of key environmental
variables, such as pressure, temperature and fuel agitation on air release from aviation fuel.
Consequently, failure to simulate the true aircraft fuel tank environment within such studies has
lead to a gulf in understanding between fuel system designers and equipment manufacturers on
how to tackle the problem.
The problem of fuel outgassing has become more prevalent as greater demands are placed on fuel
system performance and operational safety. Aircraft fuel systems have become more complex as
a direct result of continually evolving requirements for larger, multi-engined aircraft flying longer
distances with greater fuel loads. Apart from supplying the propulsion system, fuel on-board
modern aircraft fulfils many other duties (Morris, Miller and Limaye, 2006). Fuel is used as a
heat sink for cooling hydraulic system oil, electronic equipment and for controlling aircraft flight
attitude and centre of gravity. Increasing fuel system complexity and the number of system
functions creates a greater susceptibility to the problems of fuel outgassing. In view of this it is
imperative that the aerospace industry fosters a need for comprehensive understanding of the
behaviour of air and its evolution characteristics within aviation fuel under all aircraft operating

Ullage describes the space occupied by air above the fuels surface in an aircraft fuel tank

envelopes. Such understanding will underpin the improved utilisation and development of multiphase computational fluid dynamics (CFD) models and mathematical modelling tools used in
fuel systems design. The design of fuel management systems for the next generation aircraft will
benefit as CFD modelling becomes more capable of accurately simulating the gas-liquid
environments encountered within aircraft fuel systems and tanks. Perhaps most importantly, it
will be possible to optimise fuel tank inerting strategies to counter the effects of oxygen-rich air
evolution on fuel tank flammability to further improve aircraft safety. Fuel tank inerting systems
in use on modern commercial aircraft are rather wasteful, generating far more nitrogen enriched
air than is necessary to replenish the gas lost to atmosphere. This increases airline operating costs
through increased fuel consumption and reduced engine performance. Characterising fuel
outgassing behaviour and its effect on flammability in aircraft fuel tanks will enable airlines to
adopt more economic inerting strategies, leading to cost savings and improved performance
without compromising safety. To achieve these goals this research will focus on quantifying the
rate of oxygen evolution from aviation turbine fuel over a range of fuel tank environments
encountered within the flight envelope of commercial transport aircraft.

1.2 Scope of Research


The work presented in this thesis covers the measurement of oxygen concentration in small-scale
laboratory fuel tanks and a flight test aircraft fuel tank from which oxygen evolution rates are
calculated. The effects of environmental variables, such as air pressure, temperature and the level
of fuel agitation on oxygen evolution rate are investigated to establish a clear understanding of
how these factors influence fuel outgassing behaviour. Dimensional modelling is used to project
oxygen evolution rate results from a small-scale fuel tank to an aircraft fuel tank to understand if
such an approach can provide a viable, cost effective alternative to flight testing.
3

The specific objectives are as follows;

To design and develop an environmentally conditioned experimental fuel tank capable of


simulating the environment within an aircraft fuel tank during flight. Oxygen sensing
apparatus and an experimental method for measuring the rate of oxygen evolution from
aviation turbine fuel in the fuel tank is to be developed.

To establish, from statistical analysis of empirical test data, the effect size of individual
environmental variables and associated interactions of these variables on the rate of
oxygen evolution from aviation turbine fuel.

To derive a Model Law of fuel outgassing rate using dimensional modelling so that
oxygen release rates from aviation turbine fuel measured in the laboratory can be
extrapolated to full scale aircraft fuel tanks.

To establish the effect of fuel outgassing on oxygen concentration within an aircraft fuel
tank ullage through aircraft flight testing.

1.3 Thesis Layout


This thesis is divided into three main sections. Chapters 2 to 4 form a literature review of
dissolved air-in-fuel behaviour, its effects on aircraft fuel system performance and methods for
measuring gas solubility and concentration in aviation fuels. The literature review creates a
foundation on which the experimental set-up, method and testing activities are developed in
Chapters 5 to 8. Finally, Chapter 9 presents a summary of the key conclusions from experimental
testing together with recommendations for further work. In detail, the thesis is organised as
follows:
Chapter 2 is a literature review of the work previously undertaken to study the behaviour of
dissolved air within aviation fuels and the effects of pressure and temperature on air, oxygen and
4

nitrogen gas solubility in aviation fuels. The influence of thermophysical fuel properties on gas
solubility is also reviewed.

Chapter 3 reviews literature on the effects of fuel outgassing on aircraft fuel system performance
and fuel tank flammability. Specifically, each fuel sub-system in modern commercial transport
aircraft is examined in terms of its susceptibility to the fuel outgassing phenomenon.

Chapter 4 examines the different approaches previously taken for measuring both dissolved and
gaseous phase oxygen in aircraft fuel tanks and systems. The suitability of each measurement
method is explored for determining the rate of oxygen evolution from aviation fuel in the
experimental phases of this research.

Chapter 5 covers the development of an experimental fuel tank, oxygen measuring apparatus and
an experimental method for determining oxygen evolution rates in a simulated aircraft fuel tank
environment.

Chapter 6 details the laboratory testing undertaken to determine oxygen evolution rates in a
simulated aircraft fuel tank environment and presents experimental results and a discussion of key
findings. A statistical analysis of the results is conducted to examine the effect size of each
environmental variable and associated interactions of each variable on the rate of oxygen
evolution from the fuel.

Chapter 7 explores the derivation of a Model Law of fuel outgassing from dimensional analysis,
which is used to extrapolate laboratory results of oxygen evolution rate from a dimensionally
similar experimental fuel tank to a full scale aircraft fuel tank. Oxygen evolution rate results are
contrasted with those generated by other workers from dimensionally dissimilar fuel tanks and
experiments.

Chapter 8 describes the testing undertaken on a long-range flight test aircraft to establish the
effect of fuel outgassing on the concentration of oxygen within an aircraft fuel tank ullage during
flight.
Chapter 9 summarises the key conclusions made in the results Chapters, with recommendations
for future work.

Chapter 2
The Behaviour of Dissolved Air within Aviation Turbine Fuel
2.1 Introduction
As introduced briefly in Chapter 1, dissolved air is released from aviation fuel as the partial air
pressure above the fuels surface is reduced. Although problematic from a fuel handling and
management perspective, this behaviour is part of a more serious problem. The solubility
characteristics of air within aviation fuels results in an oxygen-rich dissolved air composition,
such that the released air is also oxygen-rich imposing serious flammability risks for aircraft fuel
tanks. The effect of temperature further compounds these issues as aviation fuel at higher
temperatures dissolves greater quantities of air. Within this Chapter a detailed review of the
factors which affect air solubility and the mechanisms of air evolution in aviation fuels will be
conducted. Experimental investigations carried out by previous workers to quantify air solubility
and its behaviour in hydrocarbon fuels and liquids under a variety of environmental conditions is
reviewed and the key findings relevant to this research presented to build an understanding prior
to experimental work.

2.2 Air Solubility in Aviation Fuel


The solubility of a gas in a liquid under equilibrium conditions can be expressed in a variety of
different ways (Markham and Kobe, 1941). Typically, the solubility of atmospheric gases in
aviation fuels are expressed using coefficients (Hetherington and Bellerby, 2004). These
coefficients differ depending on whether the gas pressure of interest, above the fuel, is expressed

In an aircraft fuel tank ullage air constitutes a partial pressure as the total ullage pressure is the sum of the
fuel vapour pressure and the air partial pressure

in terms of the partial pressure of that gas or the total pressure of all the gases present. The main
three coefficients used for expressing gas solubility in aviation fuels are the Ostwald, Bunsen and
absorption coefficients.

2.2.1 Ostwald Coefficient


The Ostwald coefficient, L is the ratio;

concentrat ion of gas in liquid phase


concentrat ion of gas in gaseous phase

It is equivalent to the volume of gas dissolved by a unit volume of liquid where the gas volume is
measured at the temperature and partial pressure of the solution conditions. For aviation fuel the
Ostwald coefficient is expressed in mm3 gas/mm3 fuel.

2.2.2 Bunsen Coefficient


The Bunsen coefficient, is defined as the volume of gas reduced to 273.15 K and 760 mmHg
that is dissolved by a unit volume of fuel under a gas partial pressure of 760 mmHg.

2.2.3 Absorption Coefficient


The absorption coefficient, is defined as the volume of gas reduced to 273.15 K and 760
mmHg which is dissolved by a unit volume of fuel when the total pressure is maintained at 760
mmHg. The total pressure is defined as a summation of the gas partial pressure and the fuels
vapour pressure.
Markham and Kobe (1941) established that the solubility coefficients, L, and were interrelated via the expressions detailed in Equation 2-1 and Equation 2-2 as follows;

760
760 p s

Equation 2-1
8

T
T 760
L

273
273 760 p s

Equation 2-2

where T is temperature in Kelvin and Ps is the vapour pressure of the fuel in mmHg.
It can be seen clearly from Equation 2-2 that the Ostwald coefficient L, can be calculated from

through correction of the gas volume to the specific temperature at which gas absorption was
carried out.
Further research (Derry, Evans, Faulkner and Jelfs, 1952) found it convenient to express the
solubility of air, oxygen and nitrogen gases in aviation fuels from experimental measurements as
a percentage using the following expression;

Sp k p

Equation 2-3

where Sp is the number of ml of gas dissolving in 100 ml of liquid under a gas partial pressure, p
in mmHg. k is the experimentally determined solubility coefficient for each gas. Solubility
coefficients measured by Derry, Evans, Faulkner and Jelfs (1952) correlate well with those of
Logvinyuk, Makarenkov, Malyshev and Panchenkov (1970), Astafev and Kozinova (1988) and
Schweitzer and Szebehely (1950). Table 2-1 summarises the solubility data of air, oxygen and
nitrogen in various aviation fuels measured by those workers.

Table 2-1 A summary of air, oxygen and nitrogen gas solubility measurements in aviation
fuels reported by various workers at an air partial pressure of 760 mmHg. Derry, Evans,
Faulkner and Jelfs (1952), Logvinyuk, Makarenkov, Malyshev and Panchenkov (1970),
Astafev and Kozinova (1988), Schweitzer and Szebehely (1950)
Reference

Fuel Type

Temperature
(C)

Solubility Coefficient, k

% Solubility

Air

oxygen

nitrogen

Air

oxygen

nitrogen

Derry et al

Kerosene

15.5

0.0184

0.0285

0.0157

13.98

4.54

9.43

Logvinyuk
et al

TS-1

20

0.0195

0.029

0.017

14.82

4.62

10.21

T-1
T-6

20
20

0.0183
0.0174

0.027
0.026

0.016
0.015

13.9
13.22

4.3
4.14

9.61
9.01

RT

20

0.0215

0.0267

0.020

16.35

4.25

12.1

T-6

20

0.0177

0.0188

0.017

13.5

3.0

10.5

Aircraft
Engine
Fuel

21.1

0.0226

17.2

Astafev &
Kozinova

Schweitzer
&
Szebehely

2.3 Effect of Pressure on Air Solubility in Aviation Fuel


The solubility of air in aviation fuel is described by Henrys law, where the volume of air
dissolved is directly proportional to the partial air pressure above the fuels surface at a constant
temperature. Mathematically, Henrys law can be expressed as;

p kH c

Equation 2-4

where p is the partial pressure of the solute in the gas above the solution, c the concentration of
the solute and kH is a constant (Henrys law constant), which is dependent on the solute, solvent
and the temperature. For a mixture of gases dissolved in a solvent, which is the case for air
dissolved in aviation fuel, Henrys law for each gas i.e. oxygen and nitrogen, applies
independently.
10

Derry, Evans, Faulkner and Jelfs (1952) have shown that the solubility of air gases in aviation
kerosene closely obeys Henrys law obtaining straight-line relationships between the volume of
gas evolved from the fuel and gas pressure above the fuels surface. This behaviour was
demonstrated to hold for air, oxygen and nitrogen separately. Ross (1970) also reported the
solubility behaviour of air in Avtur aviation turbine fuel at various fixed temperatures obeyed
Henrys law. Air solubility was found to decrease approximately linearly in Avtur aviation
turbine fuel with decreasing air partial pressure.

2.4 Effect of Temperature on Air Solubility in Aviation Fuel


In general the solubility of gases in non-aqueous solvents is reported to increase with increasing
temperature. Ostwald solubility data for air gases, published by the Coordinating Research
Council (2004), for a variety of military and civil aviation turbine fuels, exhibits this trend as
shown in Figure 2-1.

11

Figure 2-1 Ostwald solubility coefficients for air gases in aviation turbine fuels as a function
of temperature (Coordinating Research Council, 2004, Section 2 pg. 42)

Ross (1970) used gas chromatography to show the solubility of air in Avtur aviation turbine fuel,
at atmospheric pressure, increased with increasing temperature over a 26C to 120C range.
Solubility data was expressed as Ostwald, Bunsen and absorption coefficients as shown in Figure
2-2.

12

Figure 2-2 Gas chromatography measurements of air solubility in Avtur aviation fuel as a
function of fuel temperature (Ross, 1970, p.29)
Ostwald coefficients of air solubility increased linearly with increasing temperature over a 26C
to 120C temperature range. Bunsen coefficients increased almost linearly over this temperature
range, whilst the absorption coefficient reached a maximum at 60C and then decreased as
temperature increased beyond this value. The most likely explanation for a decreasing absorption
coefficient above a certain fuel temperature is attributable to the way in which the vapour
pressure of aviation turbine fuel increases exponentially with increasing temperature. In general,
the vapour pressure of aviation turbine fuel is low at temperatures below the fuels flash point. It
increases rapidly at temperatures above the flash point and becomes significant in the calculation
of in Equations 2-1 and 2-2. In the case of Avtur or Jet A-1, the flash point can vary between

13

38C and 66C. The fuel used by Ross (1970) may well have had a flash point of around 60C,
hence a rapid decrease in calculated absorption coefficient above this temperature.
Contrasting the results of Ross (1970) with Glendinning and Bedwell (1949) reveals some
contradiction in the behaviour of Bunsen solubility coefficients of oxygen and nitrogen in
aviation kerosenes with varying temperature. Both Ross (1970) and Glendinning and Bedwell
(1949) report a slight decrease in the Bunsen solubility of oxygen in aviation kerosene with
increasing temperature. However, for nitrogen solubility the results of these workers are not in
agreement. Ross (1970) reports a distinct increase in Bunsen coefficient with increasing
temperature and Glendinning and Bedwell (1949) a slight decrease in nitrogen solubility over a
similar temperature range. Barnett and Hibbard (1956) show Bunsen solubility coefficients for
air, oxygen and nitrogen in petroleum fractions of varying density increase with temperature over
a -30C to 60C range. Logically it would appear that nitrogen, being more abundant within the
dissolved air composition of aviation fuel, results in the Bunsen coefficient for air increasing with
fuel temperature as reported by Ross (1970) and Barnett and Hibbard (1956). Given the various
ways in which gas solubility can be expressed, it seems sensible that when considering the effect
of temperature on gas solubility, comparison between solubility results should be made, if
possible using the Ostwald solubility coefficient. The air solubility measurements of Ross (1970),
Derry, Evans, Faulkner and Jelfs (1952) and Barnett and Hibbard (1956) in kerosene based
aviation fuels when expressed as Ostwald coefficients were all in good agreement over a -40C to
120C temperature range. The gas volume is measured at the temperature and partial pressure
conditions of solution in each experimental method reducing the likelihood of possible errors
associated with conversion of solubility measurements to other temperatures and partial
pressures.
14

Markham and Kobe (1941) reported the Clapeyron equation had been used to relate the Ostwald
solubility of gases in liquids with temperature. Provided the heat of solution of the gas in the
liquid is constant over the temperature range of interest then it follows that;

ln

L1 H

L2
R

1 1

T1 T2

Equation 2-5

where L1 and L2 are the Ostwald solubility coefficients at temperatures T1 and T2, H is the heat
of solution and R is the gas constant. Graphically, Ross (1970) showed ln L is a linear function of
1/T with the slope equal to H/R for air solubility in aviation fuels. Equation 2-5 is the
integrated form of the Clausius-Clapeyron equation. For air solubility data in aviation fuels
exhibiting variation with temperature it is perhaps more prudent to use an alternative form of the
Clausius-Clapeyron equation. Letting H =A then;

dL
LA

dT RT 2

Equation 2-6

where T is equal to the fuel temperature. Integrating Equation 2-6 provides the solution for L for
constant A = a;

a
L k exp

RT

Equation 2-7

where k is a constant. Now, if the heat of solution of air gases in aviation fuel is a linear function
of temperature e.g. A a bT then the solution is;
b


a
L k1 T R . exp

RT

Equation 2-8

15

If A is considered to vary quadratically with temperature e.g. A a b T c T 2 then the


solution is;

L k2 T

b

R

RT c T

exp
R

Equation 2-9

Equations 2-8 and 2-9 were fitted to Ostwald solubility vs. temperature data for Jet A-1 in Figure
2-1using SPSS17 non-linear regression analysis software. As can be seen from Figure 2-3
whether the heat of solution of air gases in aviation fuels is considered to vary linearly or
quadratically, the Clausius-Clapeyron equation closely describes the behaviour of dissolved air
gases in aviation fuel with varying temperature. Interestingly, the result obtained in Equation 2-8
is of the same form as that obtained by Valentiner in studies of inert gas solubility in water
reported by Markham and Kobe (1941);

ln L

a
b ln T c
T

Equation 2-10

Senese (2009), described using a molecular model of gas solubility, how in organic solvents there
is usually net absorption of heat (endothermic reaction) when gases are dissolved. The energy
associated with forming a cavity between solvent molecules is greater than the energy released
during dissolution. Le Chateliers principle predicts that when heat is absorbed by the dissolution
process it will be favoured at higher temperature where gas solubility is expected to increase as
temperature rises (Holmes, 1996). By contrast, water exhibits a net release of heat (exothermic
reaction) when gases are dissolved because of much stronger intermolecular attractions between
dissolved gas molecules and surrounding solvent molecules where gas solubility is expected to
decrease with increasing temperature.
16

Figure 2-3 Clausius-Clapeyron equation predictions of Ostwald solubility coefficients


compared with CRC solubility data for oxygen dissolved in Jet A-1 aviation fuel as a
function of temperature

The solubility of air gases in aviation fuels can be estimated using the method given in American
Society for Testing Materials (ASTM), D2779-92 (2002). Utilising calculations based on the
Clausius-Clapeyron equation, Henrys law and the ideal gas law, the Ostwald solubility and mass
concentration of air gases in aviation fuels can be estimated over a -50F to 302F (-45.6C to
150C) temperature range. Universal Oil Products (UOP) Method 678-04 (2004) provides a
procedure for estimating the concentration of dissolved oxygen in liquid hydrocarbons from
measurements of dissolved oxygen partial pressure in a liquid sample, the Henrys law constant
and the Ostwald solubility coefficient estimated at the measurement temperature. As will be seen
17

in the experimental section of Chapter 6, these methods are used to determine the mass of
dissolved oxygen in the fuel, from which the rate of oxygen evolution is calculated. Considering
the numerous ways in which gas solubility in liquids can be expressed, as reviewed in this
Chapter, it can be concluded that the Ostwald solubility is perhaps the most relevant to this
research work.
Boeing and the US Air Force (2010) jointly examined the theoretical solubility of air gases
(oxygen and nitrogen) in bio-derived and Fischer-Tropsh-derived synthetic paraffinic kerosenes
using Ostwald solubility coefficients. Dissolved gas concentrations in mixtures of these bio-fuels
with petroleum aviation fuels were also examined. They contrasted the dissolved oxygen and
nitrogen mass concentrations with typical petroleum based fuels over a 0C to 120C temperature
range. Apart from the data showing a slight decreasing trend in dissolved oxygen concentration
with increasing temperature and a slight concentration increase of nitrogen with temperature, the
dissolved gas concentrations reported appear very high for all the fuels examined. Even the biofuels lower density in comparison with petroleum derived fuel does not explain the magnitude of
the concentration values presented for each gas. The mass concentrations were calculated using
the ASTMD2779-92 method, which requires the fuel density, temperature and partial pressure of
the solute gas to be known. The error which appears to have occurred, resulting in higher than
expected concentrations, most probably stems from use of the air partial pressure instead of the
individual gas partial pressure used to compute the concentration of each dissolved gas. The
ASTMD2779-92 calculation method should report a mass concentration of oxygen (ppm) at 1
atmosphere air pressure, in Jet A-1 fuel with a density of 0.8 kg/l at 20C of 83.5ppm. Boeing and
the US Air Force report a dissolved oxygen concentration of ~400 ppm for the same fuel and
temperature. In close agreement with the ASTM method, UOP Method 678-04, using the authors
18

measurements of dissolved oxygen partial pressure at 1 atmosphere in Jet A-1 fuel provides an
oxygen concentration estimate of 82 ppm at 20C. Correspondence with the Boeing Commercial
Airplane Company revealed that their theoretical solubility calculations using ASTMD2779-92
for both dissolved oxygen and nitrogen had indeed been based on the air partial pressure. This
leads to the assumption of either a pure oxygen or nitrogen atmosphere above the fuel surface,
which in an aircraft fuel tank ullage is unrealistic (Belieres, 2011).

2.5 Composition of Air in Aviation Fuel


When aviation fuel is exposed to the atmosphere oxygen and nitrogen are dissolved into the fuel
until the partial pressures of these dissolved gases are in equilibrium with the partial pressures of
the atmospheric gases. Oxygen is more readily dissolved than nitrogen where the oxygen
solubility coefficient can be 1.5 to 2 times that of nitrogen. As a result of differing solubility
coefficients, dissolved air in aviation fuel contains approximately 32% oxygen by volume with
the balance being nitrogen. Using a gas chromatographic method, Ross (1970) determined the
composition of air in terms of oxygen and nitrogen at 25C in a wide range of aviation fuels.
Table 2-2 presents this data.

19

Table 2-2 Compositions of air dissolved in aviation fuels as determined at 25C using Gas
Chromatography (Ross, 1970, p.25)
Fuel Type

Composition (% by volume)
Oxygen

Nitrogen

Avgas 115/145

29.9

70.1

Avtag 673/66

31.9

68.1

Avtur 417K

31.9

68.1

Avtur 4766K

32.2

67.8

Avtur 883/63

25.9

74.1

Avtur 1472/67

32.0

68.0

Avtur 1931/68

31.5

68.5

Avtur 845/64

30.8

69.2

Avtur 1373/68

31.1

68.9

Avtur 1004/63

32.3

67.7

Avcat 1135/66

32.9

67.1

TS-1 (Moscow)

32.3

67.7

Shellsol T

31.9

68.1

Medium-Heavy Gas Oil

33.0

67.0

Mathematical expressions have been derived by Cansdale (1978) relating the volumetric fractions
of oxygen and nitrogen present in atmospheric air to the dissolved composition of air in aviation
fuel. Using Henrys law and Bunsen solubility coefficients for oxygen and nitrogen in aviation
fuel the expressions were given as follows;

xO

O rO
O rO N rO

xN

N rN
O rO N rN

20

Equation 2-11

where xO and xNare volume fractions of oxygen and nitrogen dissolved in the fuel,O and N are
Bunsen solubility coefficients for oxygen and nitrogen in aviation fuel respectively and rO and rN
the volume fraction of oxygen and nitrogen in air above the fuel.
2.5.1 Effect of Fuel Temperature on Dissolved Air Composition
The composition of dissolved air in aviation fuels was found by Ross (1970) to change with
increasing temperature. As fuel temperature is increased Ross (1970) reported the % volume of
oxygen in the dissolved air within aviation fuel decreased from ~33% at 0C down to ~27% at
120C. A corresponding increase in nitrogen content within the air dissolved in the fuel was
observed over this temperature range. This behaviour results from oxygen and nitrogen having
different thermal expansion rates. From the ideal gas law;

P RT

Equation 2-12

where P is pressure, R is the specific gas constant and T is temperature. For a constant density,
i.e. a fixed amount of air in a sealed container, increasing temperature results in an increase in
pressure. Since the specific gas constant for nitrogen is greater than that of oxygen, the pressure
increase for nitrogen would be greater than that of oxygen for the same temperature increase.
From Henrys law a higher nitrogen partial pressure above the fuel would result in a higher
nitrogen solubility, explaining the drop in dissolved oxygen content with increasing fuel
temperature.

2.5.2 Composition of Air Released from Aviation Fuel


Ross (1972) calculated the compositions of air (oxygen and nitrogen) released from aviation fuels
using gas chromatographic measurements of the amounts and compositions of dissolved air prior
21

to and following agitation of fuel at sub atmospheric pressures. Values ranged from 24% to 30%
oxygen by volume in the released air over a 10,000 ft to 50,000 ft simulated altitude range. As
altitude was increased over this range the composition of the dissolved air became progressively
more oxygen-rich, from 32% at sea-level to 37% at a simulated altitude of 70,000 ft.
At 20C, Logvinyuk, Makarenkov, Malyshev and Panchenkov (1970) reported the average %
oxygen content by volume in air released from MK-8 engine turbine oil was 30.9%. Poulston and
Thomas (1959) reported the oxygen concentration in the ullage above kerosene fuel increased
from 23% to 34% by volume over a simulated 0 ft to 40,000 ft altitude range. Theoretically based
calculations of the oxygen concentration in the released air over this altitude range using a
nitrogen : oxygen solubility ratio of 0.625:1 at 20C was in good agreement with the
experimental data. In line with Poulston and Thomas (1959), Wyeth and Cooper (1956) made an
initial assumption that air was released from kerosene into a fuel tank ullage with fixed oxygen
content (32.5% by volume), equal to that in the dissolved air. Wyeth and Cooper (1956) later
pointed out that this could not be true if aviation fuel obeys Henrys law because the oxygen
content of the dissolved air would have to vary as the fuel tank ullage becomes progressively
more oxygen-rich due to air evolution. This assumption appears to be substantiated by the gas
chromatographic data of Ross (1970), although it is difficult to explain how good correlation was
achieved between the theoretical work and the experimental findings of Poulston and Thomas
(1959). Wyeth and Cooper (1956) demonstrated poor correlation between theoretical calculations
of ullage oxygen concentration due to air release with fixed oxygen content and experimental
measurements. It was clear that air was released initially with a higher nitrogen content in
comparison with higher altitudes where the oxygen concentration approached 31% by volume.
Derry, Evans, Faulkner and Jelfs (1952) suggested the ratio of nitrogen to oxygen evolved from a
22

fuel saturated with air as being 2.07:1, which is in reasonable agreement with experimental values
obtained by Wyeth and Cooper (1956). Derry, Evans, Faulkner and Jelfs (1952) do not provide
an explanation as to how this ratio was arrived at.

2.5.3 Effect of Fuel Properties on Air Solubility


Early theoretical assumptions suggested that the solubility of gases in liquids was related to the
surface tension of the liquid through the bulk modulus and free space within the liquid. Derry,
Evans, Faulkner and Jelfs (1952) reported the solubility of air, oxygen and nitrogen gases in
aviation kerosene were closely related to the surface tension of the fuel. A linear relationship
between gas solubility coefficient and fuel surface tension was shown to exist where gas
solubility increased with reducing surface tension. Mathematically the relationship was stated as;

k m

34

Equation 2-13

where m and c are constants and is the surface tension of the fuel. Estimates of air solubility in
aviation fuels are thus possible from Equation 2-13 provided the surface tension of the fuel of
interest is known. From a theoretical perspective, Uhilg (1937) proposed that gas solubility was
connected with liquid surface tension via the following expression;

4r 2 E
ln
kT

Equation 2-14

where;
= Ostwald solubility coefficient
r = radius of solute molecule
= liquid surface tension
E = Interaction Energy of solvent-solute system
23

k = Boltzmann Constant
T = Absolute Temperature
Inspection of Equation 2-14 indicates with constant interaction energy, E, gas solubility decreases
with increasing solute molecule radius and liquid surface tension. Plotting ln as a function of
gives a linear relationship similar to that proposed by Derry, Evans, Faulkner and Jelfs (1952) in
Equation 2-13.
Air solubility measurements made by Szebehely (1951) in various aviation fuels and oils were
similarly related by the linear function of surface tension presented in Equation 2-13, supporting
the findings of Derry, Evans, Faulkner and Jelfs (1952). Ross (1970) also established that the
solubility of air in aviation fuel was related to fuel surface tension. However, only air solubility
measurements expressed as absorption coefficient provided a linear function of surface tension.
The role of fuel surface tension in the mechanism of dissolved gas evolution from aviation fuel
will be explored more thoroughly in Chapter 7. Szebehely (1951) examined the variation in air
solubility with viscosity for a variety of oils and aircraft engine fuels. Viscosity was connected
with air solubility via a non-linear relationship, where solubility increased with decreasing
viscosity. It can be established from the air solubility-viscosity data, published by Szebehely
(1951), for the range of hydrocarbon liquids examined, including aircraft fuels, that;

% S 4.44 1.62
0.81
a b c

Equation 2-15

where a, b and c are constants and is the fuel viscosity at the temperature of interest.

24

Ross (1970) and Szebehely (1951) independently concluded variations in aviation fuel volatility
and density both had an effect on air solubility; however no precise mathematical relationship
could be established.

2.6 Mechanisms of Dissolved Air Evolution


2.6.1 Diffusive Mass Transfer to Gas Bubbles in Liquids
Burrows and Preece (1954) investigated dissolved gas evolution from vacuum pump oils
supersaturated with air at reduced pressures. They reported that in quiescent liquids air could
evolve via two distinct processes; by diffusion at the liquids surface and via gas bubble
formation, growth and rise up to the liquid surface. Schweitzer and Szebehely (1950) found, by
experiment, that the mechanism of gas evolution was strongly controlling of the rate at which gas
was released from solution. Burrows and Preece (1954), like Schweitzer and Szebehely (1950)
found the rates of gas evolution from hydrocarbon liquids were significantly higher when bubble
formation and growth accompanied gas evolution. Burrows and Preece (1954) suggested the
motion of gas bubbles in the liquid produced agitation which was associated with further bubble
formation and a corresponding rise in gas release rate. Dean (1943) developed a theory based on
free vortex motion to explain bubble formation in agitated liquids supersaturated with gas.
Borisov, Logvinyuk, Makarenkov and Malyshev (1969) investigated bubble formation and
growth in supersaturated aviation fuels. From experimental work, the conclusion reached was that
micro cracks, scratches and irregularities in the walls of aircraft fuel tanks and internal tank
structure remain unwetted by the fuel where pockets of air become trapped. Acting as air bubble
nucleation centres, the trapped gas pockets produce bubbles in the fuel, which grow upon the
reduction of pressure above the fuel surface. Burrows and Preece (1954) provided the same
theory for bubble nucleation centres stemming from trapped gas pockets in micro cavities at the
25

liquid vessel wall. Dismissing the hypothesis that gas bubble growth in liquids is a function of
reducing hydrostatic pressure acting on the rising bubble, Shafer and Zare (1991) conclude that
bubbles grow through the accumulation of dissolved gas during ascent. The wealth of study, both
theoretical and empirical, conducted on bubble growth in liquids has shown that diffusive mass
transfer of dissolved gas across the gas-liquid interface is responsible for bubble growth. Lewis
and Whitman (1924) describe, in an early paper on the theory of gas absorption in liquids, how
rising gas bubbles continually expose fresh liquid surface where the area available for diffusive
mass transfer increases as the bubble grows. Examining the bubble evolution process, Lubetkin
(1995) identifies interfacial tension at the gas-liquid interface as the dominant force controlling
growth rate when the bubble is small. Bird, Stewart and Lightfooot (2002) provide an equation
for estimating the molar absorption rate of a gas A from a bubble rising in liquid B due to mass
diffusion. As pointed out by Lewis and Whitman (1924), the process of gas evolution is simply
the reverse of absorption where the liquid is supersaturated with the escaping gas. On this basis,
the equation provided by Bird, Stewart and Lightfoot (2002) can be stated to describe the molar
rate of gas diffusion from the liquid to the rising bubble, averaged over the bubble surface as;

N A avg

4 D BA t
c A0
D

Equation 2-16

where DBA is the gas-liquid diffusivity where dissolved gas A is diffusing from liquid B to the
bubble, t the terminal rise velocity of bubble, D the bubble diameter and cA0 the solubility of gas
A in liquid B.

26

Brennen (2009) pointed out that diffusive mass transfer is enhanced by convection caused by
relative motion of the bubble in the liquid. In summary, it can be seen that the variables key to the
mass transfer process for gas bubble growth in liquids are; diffusivity of the gas in the liquid,
concentration of the dissolved gas and level of supersaturation, liquid surface tension, viscosity,
bubble surface area and the level of liquid agitation. Of these variables, diffusivity, concentration
of dissolved gas, liquid surface tension and viscosity are all functions of temperature.

2.7 Conclusions
The following conclusions capture the important aspects of air-in-fuel behaviour as they relate to
fuel outgassing from the reviewed literature;

Aviation turbine fuel obeys Henrys law where the relationship between the volume of
dissolved air evolved from the fuel was found to be a linear function of air pressure above
the fuels surface.

Oxygen dissolves more readily than nitrogen within aviation turbine fuel when exposed
to the atmosphere. Solubility coefficients of oxygen in aviation turbine fuels are 1.5 to 2
times that of nitrogen resulting in an oxygen-rich composition in the released air. Under
certain conditions air released from aviation turbine fuels was found to contain up to 35%
oxygen by volume.

Oxygen solubility coefficients in aviation turbine fuels vary with temperature where at
higher temperatures oxygen solubility increases.

Two distinct mechanisms of air evolution within aviation fuels are known to occur; under
highly quiescent conditions dissolved gas escapes from the liquid surface by diffusion at
a relatively slow rate. Higher rates of gas evolution are associated with bubble formation,

27

growth and rise up to the liquid surface. Diffusive mass transfer of dissolved gas across
the gas-liquid interface is responsible for bubble growth.

The solubility of air in aviation fuels has been shown by a number of workers to be
related by a linear function to the fuels surface tension where air solubility increases
with decreasing surface tension. The fuels surface tension has also been identified as the
dominant force controlling gas bubble growth rate.

28

Chapter 3
The Effect of Dissolved Air Evolution on Aircraft Fuel Systems
3.1 Introduction
For many decades airframe and engine manufacturers, hydro-mechanical equipment suppliers and
aircraft operators have faced fuel system performance problems resulting from the evolution of
dissolved air. Despite significant advances in system design methods through computational
fluid dynamics (CFD), fuel system problems, due to dissolved air release, still exist. Historically
each aspect of the fuel system has demonstrated vulnerability to the fuel outgassing problem. For
example the release of oxygen-rich air can counteract the effectiveness of fuel tank inerting,
leaving the aircraft exposed to flammability and explosion risks. Air bubbles developing in fuel
pipes create two-phase flow conditions that can be responsible for fuel pump cavitation,
disrupting the supply of fuel to engines and damage to the pump. Even the fuel quantity
indication system (FQIS) has demonstrated a sensitivity to the fuel outgassing phenomenon,
where system accuracy errors and in some cases the complete loss of system operation has
occurred. In cases of sudden air evolution from fuel supersaturated with air the fuel tank venting
system has been unable to cope, resulting in fuel carryover and tank over-pressurisation (Lee,
1974). Air can also be released from aviation fuel on the downstream side of a valve, which has
been closed quickly when fuel is flowing within a pipe, where the downstream fuel column
separates, forming a low pressure vapour cavity (Swaffield, 1972). A review of the literature
covering the effects of air evolution from aviation fuel on aircraft fuel systems now follows.

Carryover occurs when air evolving violently from the fuel tank causes fuel to be displaced from the tank
and into the vent system

29

3.2 Fuel Tank Inerting Systems


The necessity for fuel tank inerting was first realised in military aircraft service during the late
1950s with the introduction of gaseous nitrogen systems (Langton, Clark, Hewitt and Richards,
2009). The principal design objective was to mitigate the risk of aircraft explosion following fuel
tank penetration by enemy fire. In these early systems the gas was stored at high pressures, which
added a significant amount of weight to the aircraft and provided only a limited gas supply time.
Fuel tank inerting systems using liquid nitrogen emerged in the late 1960s, however in common
with their high pressure, gas predecessors, they added significant weight and complexity which
was unsuitable for use in fighter aircraft. In the mid to late 1980s significant developments were
made with air separation technologies, making it possible for both military and commercial
aircraft to generate Nitrogen Enriched Air (NEA) from engine bleed air for fuel tank inerting.
Today, following a spate of aircraft fires and fuel tank explosions, it has become mandatory for
new commercial transport aircraft with heated Centre Wing Tanks (CWT) to be fitted with
flammability reduction measures, such as fuel tank inerting (Federal Aviation Administration,
2008). The typical architecture of a fuel tank inerting system on board a modern commercial
transport aircraft is presented in Figure 3-1.

30

Figure 3-1 Generic On Board Inert Gas Generation System (OBIGGS) schematic (Langton,
Clark, Hewitt and Richards, 2009, p.228)
In an On-Board Inert Gas Generation System (OBIGGS) bleed air from the engine passes through
an Air Separation Module (ASM) where it is separated into its molecular constituents as it passes
over bundles of specially treated fibres. Oxygen molecules are encouraged to migrate towards the
ASMs vent, leaving NEA to pass through into the fuel tank ullage, displacing flammable fuel/air
mixtures and providing protection against potential fuel tank ignition. For commercial transport
aircraft it is required that the fuel tank ullage is maintained at an oxygen concentration of 12% or
lower to achieve an inert status (Federal Aviation Administration, 2005). Ullage oxygen
concentration is primarily a function of ASM performance, evolution of oxygen-rich air from the
fuel, variations in fuel tank quantity (and thus ullage quantity) and the flow of air into and out of
the vent system during flight. Airworthiness authorities (Federal Aviation Administration &
31

European Aviation Safety Authority) mandate that airframe manufacturers must take these factors
into account when modelling the performance of a fuel tank inerting system. The Fuel Tank
Flammability Assessment Method (FTFAM), engendered by the FAA requires that the inerting
system maintains the fleet average flammability exposure time to below 7% (Summer, 2008).
This value describes the amount of time an aircraft fuel tank was flammable divided by the total
flight time for a number of simulated aircraft flights. A survey of literature examining the effect
of oxygen-rich air evolution on fuel tank inerting system performance will now be reviewed.

3.2.1 Effect of Air Evolution on Inerting System Performance


Gatward and Wyeth (1951) conducted early work on the effects of air evolution from aviation
kerosene on the inert gas protection systems of aircraft fuel tanks. Their approach was to use a
laboratory scale fuel tank fitted with an on demand nitrogen gas inerting system. The inerting
system pressurised the ullage space of the fuel tank with nitrogen to values between 0.75 psi and
3 psi above that of the outside ambient pressure in a test chamber which was reduced to simulate
aircraft climb. Fuel was withdrawn from the tank to simulate engine demand allowing a quantity
of nitrogen to be admitted into the ullage. Figure 3-2 shows graphical plots of ullage oxygen
concentration at different nitrogen pressures measured by Gatward and Wyeth (1951). Air
released from the fuel was found to affect the inerting system performance in two ways. Air
evolution increased ullage space oxygen content to a maximum of 32.5% by volume at the lowest
nitrogen pressure. Air was released from the fuel at a rate greater than fuel consumption and for a
significant period the internal tank pressure remained at a higher level than the nitrogen
admission pressure. The rate of air evolution was significant because nitrogen gas could not enter
the tank ullage to reduce oxygen concentration below the critical limit until the ullage pressure
had fallen below the nitrogen admission pressure. In later work, conducted by Wyeth, Oldfield
32

and Wiltshire (1957) the ineffectiveness of this type of nitrogen inerting system designed to
combat the effects of air evolution was demonstrated in a flight test. The ullage oxygen
concentration remained well above the critical limit (13.6% oxygen in this case) for the entire
flight. To achieve the 5% ullage oxygen concentration target with the on demand system, nitrogen
gas was injected into the fuel during the refuelling process to lower the fuels dissolved oxygen
content, in conjunction with inerting the empty fuel tank prior to refuelling. Roth (1987) was also
able to demonstrate the merits of saturating the fuel with NEA to manage oxygen concentration
within the fuel tank ullage during simulated aircraft climb. A nitrogen and 5% oxygen gas
mixture was injected into the fuel until the fuel was saturated and in equilibrium with the ullage at
a 5% oxygen concentration. Following a simulated C-17 aircraft climb the ullage had reached
~8.5% oxygen concentration in comparison to 26% oxygen with untreated fuel, saturated with air.
Cavage (2005) provided similar findings to Gatward and Wyeth (1951). Even though different
inerting system strategies in both studies were examined, where Gatward and Wyeth (1951) used
on demand pressure regulated nitrogen purging, and Cavage (2005) a pre-inerted ullage with no
purging, air evolution in both cases significantly increased ullage oxygen concentration. Cavage
(2005) reported an increase in oxygen content of up to 7% for an ullage pre-inerted to 6% oxygen
by volume at sea-level pressure with NEA (95% N2/5% O2) and an 80% fuel load agitated by a
pump. Reducing ullage pressure to simulate aircraft climb increased ullage oxygen concentration
further, behaviour consistent with the findings of Gatward and Wyeth (1951). OBIGGS
performance studies conducted by McConnell, Dalan and Anderson (1986) of the Boeing
Military Airplane Company investigated the effect of dissolved air evolution from JP-4 aviation
fuel on a pre-inerted ullage. Using a laboratory fuel tank the ullage was pre-inerted to an oxygen
concentration of 5% by volume by injecting an NEA gas mixture (95% N2/5% O2) into the ullage.
33

Following NEA injection the ullage was depressurised in accordance with the climb schedule of a
KC-135 aircraft and fuel withdrawn from the tank at the aircrafts engine fuel burn rate. During
the simulated climb the fuel tank was vibrated and the fuel internally re-circulated with pumps to
simulate the aircraft fuel tank environment. The combined effects of fuel-recirculation and tank
vibration were found to increase the ullage oxygen concentration at the highest rate, where
oxygen concentration rose from ~6% to 22.5% oxygen by volume over the test duration. By
comparison, the rate at which oxygen was liberated from the fuel under quiescent conditions was
much slower.

Figure 3-2 Effect of dissolved air evolution from aviation kerosene on the oxygen
concentration of an on demand nitrogen purged fuel tank ullage as determined by
(Gatward and Wyeth, 1951, p.13)
34

Kosvic, Zung and Gerstein (1971) illustrated, through experimental studies and computer
modelling, the role that dissolved oxygen released from fuel plays in the flammability of aircraft
fuel tanks. It was established that the fuel-air mixture in an aircraft tank ullage, which was
initially fuel vapour-rich, could become combustible during aircraft ascent due to fuel outgassing.
High levels of fuel tank vibration were seen to acerbate the problem. Although the effectiveness
of fuel scrubbing to limit ullage oxygen concentration was demonstrated in an experimental
capacity by Wyeth, Oldfield and Wiltshire (1957) and Roth (1987), Bragg, Kimmel and Jones
(1969) highlights the impracticality of such an approach for commercial airline operations. The
cost and logistical problems associated with treating large quantities of fuel with nitrogen gas has
driven the development and introduction of OBIGGS in modern commercial transport aircraft.

3.3 Engine Fuel Feed Systems


3.3.1 Functionality of the Engine Fuel Feed System
The engine fuel feed system in todays modern transport aircraft is perhaps the most critical
function of the fuel system (Langton, Clark, Hewitt and Richards, 2009). Continued and safe
operation of the aircraft engines depends upon its reliable operation. The typical architecture of
an airframe engine fuel feed system for a twin engine commercial transport aircraft is presented
in Figure 3-3.

Fuel scrubbing describes the process of reducing the dissolved oxygen content of aviation fuel by forcing
nitrogen gas through the fuel

35

Figure 3-3 Schematic of a twin engine fuel feed system of a commercial transport aircraft
(Langton, Clark, Hewitt and Richards, 2009, p.67)
The feed system must ensure that fuel delivered to the engine interface is maintained at a pressure
above the fuels vapour pressure during flight to prevent adiabatic fuel boiling and to limit the
amount of air evolution. This requirement is typically achieved through use of electric motordriven fuel pumps located within the fuel tank. These pumps typically raise the fuels pressure for
supply to the engine from the static head pressure at the pumps inlet to between 140 kPa to 350
kPa. There are many design attributes associated with motor-driven fuel boost pumps that must
be taken into account, not least the pumps ability to continue operation in the presence of air or
fuel vapour at the inlet. Accommodating entrained air within the fuel is a major issue in the
design of the pump and although design strategies to remove entrained air and vapour (liquid ring
elements) have been adopted, problems still exist (Harris, 2011). Design of the engine fuel feed
gallery, connecting fuel tank pumps to the engine interface has also received significant attention
with respect to air evolution. It is desirable to ensure all fuel feed line geometries are smooth with
36

no sharp bends or sudden changes in cross section which can encourage air bubble formation.
Installation of air release valves within the feed gallery that separate air from the fuel and
discharge it back to the tank ullage have provided some improvement, although such devices are
far from a complete design solution. Locating the valves within complex feed line geometries,
where they are able to effectively remove moving air bubbles, is challenging whilst one of the
valve failure modes is to allow fuel to leak back to the tank compromising feed system integrity.
A photograph of an air release valve manufactured by Eaton Aerospace fitted to the main engine
fuel feed gallery of the Airbus A340-500/600 long range aircraft is shown in Figure 3-4.
Downstream of the engine interface contained within the engine nacelle is the engine fuel system.
This typically comprises a low pressure centrifugal pump stage (backing pump) and a high
pressure positive displacement pump stage (HP pump). Unlike the fuel tank electric pumps, both
of these pumps are driven mechanically from the engine shaft through an auxiliary gearbox. A
simplified schematic of an engine fuel system is shown in Figure 3-5.
The role of the backing pump is two-fold. Firstly, it improves the suction performance of the
engine if operation of fuel tank boost pumps is lost. Secondly, it provides the required fuel
pressure at the inlet of the high pressure pump to prevent cavitation.

Figure 3-4 Photograph of the Eaton Aerospace FRH280002 air release valve incorporated
within the engine fuel feed gallery of the Airbus A340-500/600 aircraft (Eaton Corporation,
2004)
37

Figure 3-5 Simplified engine fuel feed system schematic (Langton, Clark, Hewitt and
Richards, 2009, p.10)
3.3.2 Effect of Air Evolution on Engine Fuel Feed
During the 1970s Rolls Royce conducted an extensive programme of research on the effects of air
release from aviation fuels on the performance of engine fuel feed systems (Loxley, 1976; Loxley
and Wright, 1976; 1978; Wright, 1976; Loxley and Goddard, 1980). Their focus was to
understand how the released air affected flow conditions within different engine feed pipe
configurations and geometries operating under suction over a range of simulated altitudes.
Schematics of the feed pipe geometries and dimensions studied are presented in Appendix A.

3.3.3 Vapour/Liquid Ratio


Rolls Royce developed an equation for calculating the vapour/liquid ratio for aviation kerosene
flowing in an engine fuel feed pipe under suction conditions (Wright, 1976). The equation, based
on the calculation method developed in ARP429C, allowed dissolved oxygen partial pressure
values, measured in the fuel flow tests to be used (Society of Automotive Engineers, 1994).
Oxygen partial pressure measurements were used to represent dissolved air partial pressure values
38

used in the vapour/liquid ratio calculations by assuming a constant dissolved oxygen/nitrogen gas
ratio in the fuel.

P P2 D t 460
V

1D
P2 PTVP
492
L

Equation 3-1

where;

Bunsen solubility coefficient (set = 0.15 in the Rolls Royce tests)

P1D = partial pressure of dissolved air in fuel upstream of feed pipe


P2D = partial pressure of dissolved air in fuel downstream of feed pipe
P2 = pressure in pipe
PTVP = true vapour pressure of fuel at temperature, t
t = fuel temperature
The calculated vapour/liquid ratio results for a horizontal feed pipe and VC10 aircraft doublehump feed pipe are shown in Figure 3-6 and Figure 3-7 respectively. Climb 1 simulated a
commercial transport aircraft climbing at a rate of 2000 ft/min to 10,000 ft and then at 1000
ft/min beyond this altitude up to 45,000 ft. Climb 2 was aligned to that of a fighter aircraft
climbing at a rate of 10,000 ft/min up to 10,000 ft and then at 5000 ft/min beyond this altitude.
The ground level fuel flow rate in all tests was 1250 gallons per hour to achieve a fuel flow
velocity of 5 ft/s in the 1.5 inch diameter pipe. Figure 3-6 shows very clearly the effect of aircraft
climb rate on the vapour/liquid ratio in the same horizontal feed pipe. The vapour/liquid ratio
peak under Climb 2 conditions is nearly three times that of Climb 1 demonstrating a much larger
volume of air was contained in the pipe at approximately 30,000 ft. The slope of the vapour/liquid
ratio curve for Climb 2 is significantly steeper than for Climb 1, indicating the rate of air
evolution from the fuel was much higher at the higher climb rate. An increase in vapour/liquid
39

ratio in all tests as the maximum altitude was approached was thought to occur due to changes in
dissolved air composition. The air, remaining in solution became more oxygen rich. This
hypothesis was supported by the fact that increases in dissolved oxygen partial pressure, at these
high altitudes, were not accompanied by visible signs of gas evolution from the fuel. Chapter 4
expands on the problem of using dissolved oxygen measurements to infer the quantity of
dissolved air in aviation fuel using the assumption that air composition remains fixed at sub
atmospheric pressures. Figure 3-7 illustrates the effect of fuel feed pipe geometry on
vapour/liquid ratio. Although not as pronounced as the effect of climb rate, the inclusion of bends
in the feed pipe were responsible for almost doubling the vapour/liquid ratio at approximately the
same altitude.

Figure 3-6 Vapour/liquid ratio in the horizontal feed pipe as a function of altitude (Wright,
1976, p.16)

40

Figure 3-7 Effect of engine fuel feed pipe geometry on vapour/liquid ratio as a function of
altitude (Wright, 1976, p.16)
Various types of two-phase flow were visually observed through clear glass sections incorporated
within the feed pipes over the range of altitudes examined. In summary, low altitudes gave rise to
dispersed bubbly flow patterns whilst at medium to high altitudes, stratified, slug and channel
flow behaviour was observed.

3.3.4 Effect of Air Evolution on Engine Backing Pumps


As mentioned briefly in Section 3.3 backing pumps were introduced into the engine fuel system
to improve suction performance of the engine if boost pumps in the wing tank fail and to ensure
the high pressure positive displacement pump is fed with fuel at a pressure necessary to prevent
cavitation. Backing pump inlet pressure requirements are based on two operating conditions; the
first being normal operation with tank boost pumps operative, the second, without boost pumps
working, termed suction feed. The backing pump inlet pressure requirements (to be met by the
41

airframe manufacturer) at the engine interface are 5 psi plus True Vapour Pressure (TVP) of the
fuel above fuel tank ullage pressure with a V/L of 0. For suction feed conditions the margin is 3
psi plus TVP above tank ullage pressure with a maximum V/L of 0.45. Even with these margins
the low vapour pressure aviation fuels based on kerosene are susceptible to fuel outgassing at the
engine interface, making the backing pumps operating conditions arduous.
Allied to their research work on engine fuel feed pipes, Rolls Royce conducted extensive research
on the effect of air evolution from fuel on engine backing pumps (Loxley and Goddard, 1980).
The relationship between fuel pump performance and the fuels dissolved air content was
explored in a series of tests. The tests demonstrated that the limiting gas/liquid ratio was 0.13,
which indicated that a pumps ability to manage free air would be determined by the efficiency of
the impeller to re-dissolve free air into the fuel. Backing pumps tested without a fuel feed system
attached to their inlet provided satisfactory pressure rise at suction inlet pressures down to 1.5 psi,
compared with 2.5 to 7 psi suction inlet pressures with backing pumps fitted with different engine
feed system arrangements. The specific feed system arrangement influences the pressure loss and
type of two-phase flow at the pumps inlet, as has been discussed earlier. Careful design of the
engine fuel feed gallery is critical therefore to the performance of the engine backing pump under
suction feed duty. Airframe fuel system design strategies must ensure wherever possible fuel
pressure losses, due to bends, long pipe lengths and internal geometry changes, are minimised to
reduce susceptibility to air evolution and the resultant two-phase flow conditions that present
arduous operating environments for the engine backing pump. Achieving this is extremely
challenging as fuel pipe routing options are heavily constrained by structural design requirements
and keep out zones imposed by engine rotor burst requirements.

42

3.4 Fuel Quantity Indication Systems (FQIS)


The fuel quantity indication system (FQIS) in commercial transport aircraft provides the flight
crew with an indication of the total mass of fuel on board. Mass is an acceptable measure of the
fuels thermal energy content which enables engine thrust and thus aircraft range to be calculated.
Simplistically, the FQIS measures fuel heights within each fuel tank using vertically mounted
probes. Fuel height measurements are converted to fuel volume using on-board computer
algorithms with height-volume data specific to each tanks geometry. A device measuring the
fuels density (densitometer) is used to convert fuel volume in each tank to a total mass of fuel on
board. A number of methods have been used to gauge fuel level in aircraft fuel tanks, including,
capacitance level sensing, ultrasonic level sensing, and hydrostatic pressure measurement. Of
these different technologies the ultrasonic approach was noted as exhibiting a particularly high
level of sensitivity to fuel outgassing and free air-in-fuel behaviour. Fuel aeration occurring
during both aircraft refuel, and in flight, as emerging air bubbles escaped from the fuel, caused
significant problems for ultrasonic system operation (Kumar, 1987). In aerated fuel the acoustic
wave emitted from the ultrasonic sensor was attenuated due to scattering and reflection from the
gas bubble surface leading to erroneous interpretation of the fuel level. Computations of fuel
mass using algorithms which relied on fuel level measurements generated significant fuel
quantity errors. Investigations into these malfunctions concentrated on the level of fuel aeration at
which the system would operate reliably, the effects of surface air bubbles on acoustic echo and
fuel filter techniques to reduce sensitivity to fuel aeration. Results of these investigations
prompted the development of design features which significantly improved the ultrasonic level
sensors performance in aerated fuel. The use of a still well tube improved the confinement of
acoustic energy, increasing the strength of reflected echoes from the fuel surface to the sensing
43

transducer. A bubble baffle arrangement dramatically increased sensor performance in aerated


fuel during refuelling. The baffle device reduced fuel flow in the region local to the ultrasonic
sensor allowing gases to emerge out of the fuel before they entered the still well tube.

3.5 Vent Systems


Commercial transport aircraft feature open vent systems which connect the ullage of each fuel
tank to the outside atmosphere. A key requirement of this type of vent system is to enable free
breathing of the fuel tanks throughout the aircrafts operational flight envelope. Without this
provision large pressure differences between ullage air and the outside atmosphere would
develop, resulting in the build up of large forces capable of damaging the aircraft structure. A
schematic of the Airbus A340-600 wing vent system is shown in Figure 3-8. Typically vent
system sizing requirements are principally driven by the aircraft emergency descent case, where
an aircraft is required to descend from maximum cruise altitude to below 10,000 ft as quickly as
possible following loss of cabin pressurisation. There has however been reported instances of
rapid air evolution from fuel in aircraft that has resulted in tank over pressurisation and vent
system flooding with fuel due to under-sizing of the vent system.

44

Figure 3-8 Wing vent arrangement of the Airbus A340-600 aircraft (Langton, Clark, Hewitt
and Richards, 2009, p.46)
3.5.1 Effect of Air Evolution on Vent System Performance
The British Aircraft Corporation (BAC) experienced fuel tank venting problems due to air release
from fuel during the development of Concorde in the 1970s (Lee, 1974). As it was recognised
that the rate of air release from supersaturated fuel could be very high during the aircrafts rapid
rate of climb to high altitude, specific fuel tanks were fitted with de-aeration systems. These
systems, comprising centrifugal pumps and high flow velocity nozzles, continually agitated the
fuel during climb to maintain a steady rate of gas evolution. In general this approach proved
successful by lowering the volume of dissolved air content in the fuel from 15% at sea-level to
3% at 65,000 ft. There was, however, one exception associated with fuel tank No. 11. Tank No.
11 experienced over-pressurisation and fuel carry-over into the vent system prompting a
45

modification by BAC to increase the diameter of the vent in tank No. 11 from 2.5 to 3 inches. In a
much earlier study, Beal, Hilburger and Porter (1945) highlighted that instantaneous air evolution
from aviation gasoline in a nearly full tank with a constricted vent might raise the ullage pressure
by several psi, which could be more than the amount allowable. The research work conducted by
Beal, Hilburger and Porter (1945) on air evolution from aviation gasoline made a
recommendation that for vent pipe design purposes the volume of gasoline vapour escaping
through an aircrafts vent pipe should be increased from 3% to 5% to allow for the released air. In
small-scale laboratory experiments Ross (1972) examined the effect of fuel tank vent size on tank
pressurisation and fuel carryover due to air evolution from air supersaturated Avtur aviation
turbine fuel. The results obtained by Ross (1972) are presented in Table 3-1. Pressure build-up in
the small test tank increased with decreasing vent diameter which supports the comments of Beal,
Hilburger and Porter (1945). The volume and rate of fuel carry-over decreased with decreasing
vent diameter. The relationships between tank pressure and fuel carry-over, as a function of vent
diameter under air evolution conditions, are plotted in Figure 3-9. It is interesting to note how the
rate of fuel carryover indicative of air evolution rate from the fuel has been affected by the vent
pipe diameter. An increased amount of resistance to fuel flow with smaller diameter vent pipes
has retarded the rate of gas evolution and resulted in tank pressurisation. If it is assumed that the
amount of displaced fuel over a given time period is proportional to the amount of air evolved
over that same time period and provided no air is re-absorbed by the fuel, then the rate of air
evolution as a function of vent pipe diameter and tank pressurisation can be found from the data
in Table 3-1. Figure 3-10 shows that the rate of air evolution, inferred from the rate of fuel
carryover, increases significantly with vent pipe diameter and reduces the maximum pressure rise
experienced in the tank exponentially.
46

Table 3-1 Effect of vent pipe ID on fuel tank pressurisation and fuel carryover due to air release from supersaturated Avtur aviation
turbine fuel subjected to agitation and a simulated altitude of 60,000 ft (Ross, 1972, p.21)

Vent Pipe I D
(mm)

Fuel Carryover

Maximum Tank
Pressure Rise

% of
tank
3

Depth
(mm)
4.5

Duration
(sec)
600

(mbar)

Volume
(ml)
265

Time to reach
maximum tank
pressure
(sec)

724

12.5

4.8

685

7.9

33

407

5.5

9.5

1040

12

13.5

12

200

131

4.5

12

1165

13.4

15

10

159

117

3.5

16

1273

14.6

16.5

117

103

22

1353

15.6

17

103

103

5.5

47

Maximum Vent
Pressure Rise
(mbar)

Time to reach
maximum vent
pressure
(sec)

Figure 3-9 Maximum tank pressure rise and fuel volume carry-over as a function of vent
pipe ID from tests conducted by (Ross, 1972, p.21)

Figure 3-10 Calculated rate of air evolution and maximum tank pressure rise from Avtur
aviation turbine fuel carryover measured by (Ross, 1972, p.21)
48

3.6 Conclusions
The following conclusions capture from the reviewed literature the effects of air evolution on
aircraft fuel systems;

The effectivity of early fuel tank inerting concepts based on purging the ullage with
nitrogen gas were significantly reduced due to the release of oxygen-rich air from the
fuel. Where oxygen rich air evolution was greater than the fuel consumption rate, the
inerting system failed to reduce ullage oxygen concentrations to below critical limits.
Scrubbing the fuel with nitrogen gas was shown to be a more effective approach for
reducing ullage flammability. Scrubbing the fuel with nitrogen gas was time consuming
and deemed financially unviable for large commercial aircraft.

The effect of aircraft climb rate and engine fuel feed pipe geometry has a significant
effect on fuel outgassing rate in engine fuel feed systems operating under suction
conditions. Increasing climb rate and introducing bends into engine feed lines increased
the vapour/liquid ratio by a factor of approximately 3.

The ability of an engine backing pump to manage free-air evolved from the fuel is
largely governed by the efficiency of the impeller to re-dissolve free gas into the liquid
and the pressure-loss characteristics of the engine feed gallery upstream of the pump.

Fuel Quantity Indication Systems based on ultrasonic level gauging have shown
susceptibility to fuel outgassing and entrained air within the fuel. Erroneous interpretation
of the fuel level resulted as air bubbles escaping from the fuel causing scattering of the
systems acoustic measuring waves.

The rate of fuel outgassing is an extremely important factor when sizing atmospheric
venting systems for aircraft fuel tanks. Undersized vents with insufficient flow area have
49

been responsible for aircraft fuel tank over-pressurisation and fuel carryover, particularly
where the rate of fuel outgassing was high.

50

Chapter 4
Measurement of Air, Oxygen and Nitrogen in Aviation Fuels
4.1 Introduction
A number of methods have been developed and adapted to quantify the amounts of atmospheric
air, oxygen and nitrogen gases dissolved in aviation fuel. The recent introduction of fuel tank
inerting systems into commercial aircraft has extended the need for the measurement of oxygen
concentration to the ullage of aircraft fuel tanks. These measurement methods can be generically
split into three separate categories; physical, chemical and optical. Physical methods usually
exploit the physical behaviour of the gas in the liquid in accordance with Henrys law, whereas
chemical methods are dependent on specific chemical properties of the gas being analysed.
Perhaps most promising for fuel tank based measurements are the optical oxygen sensing
methods with their inherent intrinsic safety and broad environmental operating envelope.
Whichever methodology is selected, consideration must be given to the environment under which
the measurement is to be made. The environment associated with an aircraft fuel tank is harsh,
where extremes of temperature, pressure and vibration are present throughout the aircrafts flight
envelope. Such conditions can limit the practicality of some measurement methods due to
temperature, pressure and other environmental factors which lie outside of the instruments
operating capability. The method should also be capable of measuring dissolved and gaseous
phase oxygen concentrations both on-line and in real-time, as opposed to delayed sampling and
post test measurement which can introduce significant error. As the requirement in this research

51

work is to quantify dissolved oxygen release rate from aviation turbine fuel, gas phase
measurement approaches will also be reviewed alongside liquid phase methods.

4.2 Physical Methods


Physical methods using either dissolved gas extraction or gas-liquid saturation have been
developed for determining the total dissolved quantity of gases in liquids. In the former approach
the gas saturated liquid is subjected to a medium vacuum in a sealed chamber which, following
liberation of the gas from the liquid, results in a pressure increase above the liquids surface.
From the measured pressure rise the original gas content of the liquid is calculated. The limitation
however with this method is an inability to distinguish between the partial pressure of the
released gas and the liquids vapour pressure. As such, only the measurement of dissolved gas
content in low vapour pressure liquids (<0.1 mmHg) is possible if large errors are to be avoided.
Cann of the Royal Aircraft Establishment (RAE) developed a physical method which was capable
of measuring the partial pressure of released air from aviation fuel to estimate air solubility
(Cansdale, 1978). Figure 4-1 shows a schematic of the apparatus developed by Cann.

A medium vacuum as defined by the National Physical Laboratory spans a pressure range of 3000 to 0.1
Pa

52

Figure 4-1 Air solubility measurement apparatus schematic developed by Cann of the
RAE(Cansdale, 1978, p.21)
As with previous methods, the fuel sample is introduced into an evacuated chamber of known
volume. After deaeration is complete the volume of the chamber is reduced by a known quantity
via a set of manually adjusted bellows. The pressure in the headspace of the chamber above the
liquid is measured with a manometer. The rise in pressure caused by the volume reduction is
solely proportional to the partial pressure of air released from the fuel as the pressure of the
saturated vapour is not altered by compression. From the measured pressure rise and using
Boyles, Charles and Henrys laws the fuels original dissolved air content is obtained.
Assumptions are made in that all of the dissolved gas is released from solution and during
compression none of it is redissolved and the pressure in the deaeration chamber prior to sample
admission is approximately zero. Using the method of Cann, Cansdale (1978) demonstrated that
measurements of dissolved air content in aviation turbine fuel were in good agreement with those
from a gas-liquid chromatography method suggesting that the methods assumptions were valid.
Hayward and Dallas (1965) developed an almost identical apparatus to Cann for measuring the
53

gas contents of liquids, except the method of reducing volume in the sample chamber was
achieved by a moving piston rather than bellows. They reported an accuracy of +/- 5% with their
method and apparatus. Derry, Evans, Faulkner and Jelfs (1952) measured the solubility of air in
aviation fuels using the Van Slyke-Neill manometric apparatus.
Although the physical based methods reviewed here have been demonstrated with relatively good
accuracy, they are ultimately not designed or suitable for continuous, in-situ dissolved gas
measurements in a fuel tank. The environmental conditions under which these methods are
designed to operate is fixed and dissociated from the range of environments present in aircraft
fuel tanks, making an in-situ approach impossible. Even though fuel samples can be extracted
from the fuel tank and sampled offline, the complexity associated with such sampling activities
increases the probability of contamination and associated error.

4.3 Chemical Methods


4.3.1 Gas Chromatographic Methods
Rubey, Striebich, Tissandier and Tirey (1995) developed a gas chromatographic method to
measure dissolved oxygen in thermally- stressed hydrocarbon liquids designed to emulate
aviation fuels flowing in an aircraft fuel pipe. A tandem column assembly was used, comprising
two fore columns and an analytical column. This was essential to ensure dissolved oxygen gas
was separated from the hydrocarbon liquids prior to introduction into the measurement
instrument. Dissolved oxygen concentrations in flowing aviation fuel simulants were made with
values ranging from 70 ppm to 0.1 ppm by weight where the average relative standard deviation
at 70ppm was 0.98%. Petrocelli and Lichtenfels (1959) developed a gas chromatographic method
for determining the concentration of dissolved oxygen and nitrogen in petroleum fractions. Like
Rubey, Striebich, Tissandier and Tirey (1995) the use of a separation column in series with a
54

molecular sieve column was found to be essential, as was a robust sampling method to avoid air
contamination. Dissolved oxygen and nitrogen were separated from the hydrocarbons in a precolumn and swept into the molecular sieve column by helium carrier gas where they were
resolved. The pre-column retained the hydrocarbons. Results of dissolved oxygen were in good
agreement with polarographic oxygen sensor measurements. The advantage offered by the gas
chromatographic method over oxygen specific sensing methods was highlighted where
simultaneous determinations of oxygen and nitrogen concentration can be made. Kitanin, et al.
(2007) used a series of gas chromatograms to measure dissolved air concentrations within
aviation turbine fuel flowing within a pipe between two fuel reservoirs subjected to reduced
atmospheric pressures. Gas chromatographic measurements made at various points along the pipe
enabled the quantity of evolved air in the pipe to be estimated. Although this method was
successful in measuring dissolved air quantities, there were a number of issues. Taking fuel
samples from the pipe whilst the fuel was flowing and injecting it into the gas chromatograms
was complex and sometimes unreliable. Further, the level of regular maintenance required to
clean the column within the chromatograms to remove traces of fuel between tests, required to
minimise error, was high. Error was reported to be as high as 25% in measured samples. Ross
(1970) used gas chromatography to determine the quantities of air, oxygen and nitrogen in
aviation fuels. The method and gas chromatographic configuration used was based on that of
Petrocelli and Lichtenfels (1959), described previously.
Despite the key advantage offered by gas chromatographic techniques in that the total air content
within the fuel can be determined from individual measurements of oxygen and nitrogen, the
complexity of sampling, ultimate destruction of the sample and difficulties of in-situ

55

measurement at low pressure and temperature with the method make it unviable for this research
work.

4.3.2 Electrochemical Methods


Electrochemical oxygen sensing methods have been successfully used in previous investigations
to determine dissolved oxygen concentration in aviation fuels and in studies to measure the
oxygen content of air in a fuel tank ullage. The electrochemical sensors utilised were of the
galvanic micro-fuel cell and polarographic type. An electrochemical oxygen sensor in its simplest
form consists of an anode and a cathode immersed in an aqueous electrolyte solution. The anode,
cathode and electrolyte solution are separated from the gaseous or liquid sample being analysed
by an oxygen permeable membrane. The anatomy of an electrochemical oxygen sensor based on
the polarographic design is shown in Figure 4-2.

Figure 4-2 Anatomy of an electrochemical oxygen sensor based on the polarographic design
(Hach Lange, 2009, p.15)

56

As oxygen molecules diffuse through the gas permeable membrane, the molecules are reduced at
the cathode to form negatively charged hydroxyl ions. Hydroxyl ions migrate to the sensor anode
where an oxidation reaction takes place. The resultant oxidation/reduction reaction generates an
electrical current proportional to the partial pressure of oxygen in the measured sample.
Laboratory standard UOP 678-04 details a method for determining dissolved molecular oxygen
concentration in liquid hydrocarbons using electrochemical oxygen sensors. Measured oxygen
partial pressures are converted to dissolved oxygen concentration (ppm by weight) via a
calculation procedure. The use of electrochemical oxygen sensing methods is restricted at subzero temperatures due to freezing of the electrolyte solution, which in most electrolytes occurs at
approximately -10C. Air pressures reduced significantly below the barometric ambient can
affect sensor performance where some dissolved gas in the electrolyte can escape. Gas escaping
from the electrolyte inflates the membrane leading to an apparent reduction in measured oxygen
partial pressure. Employing an electrochemical method for measuring oxygen in aviation fuel
tanks at reduced pressures and temperatures can thus require sensor modification or
environmental conditioning of the gas sample to ensure sample temperatures and pressures are
within the environmental measurement range of the sensor. Cansdale (1978) used a polarographic
oxygen sensor to determine the variation in dissolved air content of aviation turbine fuel as air
was liberated at reduced pressures. The oxygen sensor (Beckman model 777) was calibrated to
give a reading of 14% dissolved air by volume in the fuel saturated with air at atmospheric
pressure. The fuels volumetric air content was inferred from dissolved oxygen measurements,
assuming the ratio of oxygen to nitrogen in the dissolved air remained constant with altitude.
Readings from the polarographic sensor for dissolved air were higher at altitude than those from
an alternative air measuring method. Poor correlation between measurement methods was
57

attributed to changes in dissolved air composition, implying air becomes more oxygen rich as
altitude increases. This hypothesis of oxygen enrichment was subsequently proven by gas-liquid
chromatography measurements, highlighting the limitations of the polarographic sensor for
estimating dissolved air content in aviation fuel. Wright (1974) and Loxley (1976) used
polarographic oxygen sensors to measure the partial pressures of dissolved oxygen in aviation
turbine fuel flowing within pipes. Oxygen partial pressure readings were used to estimate air
partial pressures in the fuel, assuming the ratio of oxygen and nitrogen in the dissolved air
remained constant. Oxygen sensor readings indicated that the vapour/liquid ratio in the fuel feed
pipes increased dramatically as the maximum simulated test altitude was approached.
Observations of air evolution were not consistent with sensor readings at these higher altitudes
indicating the dissolved air composition was becoming more oxygen rich, rather than air being
released, the same conclusion reached by Cansdale (1978). Gas chromatographic measurement of
air compositions in aviation fuels made by Ross (1970) as presented in Chapter 2 further
substantiates this conclusion.
In this research work the electrochemical dissolved oxygen sensor, specifically the polarographic
type, is well suited to the measurement task, providing in-situ, continuous measurements of
dissolved oxygen partial pressure in the fuel, provided fuel temperatures below the electrolyte
freeze point are avoided.

4.4 Optical Sensing Methods


4.4.1 Optical Absorption Spectroscopy
Tunable Diode Laser Absorption Spectroscopy (TDLAS) is an optical absorption spectroscopy
technique that has been developed for measuring oxygen concentration in an aircraft fuel tank
ullage (Frish, 2004; McCaul, 2007; Cavage, 2009). The technique, with pressure and temperature
58

compensation, provides accurate oxygen concentration measurements in air down to -40C and
pressures over a 110 kPa to 20 kPa range. The method demonstrates an insensitivity to fuel
vapour due to the extremely narrow spectral bandwidth used for the measurement. TDLAS based
oxygen sensing was used for validating the performance of fuel tank inerting systems in the
Airbus A320 and A340 aircraft (Blanchard, 2008). Gas samples continuously drawn from the
ullage of the aircraft fuel tanks, with a dedicated pump, were fed into the TDLAS sensor located
within the aircraft cabin.

4.4.2 Measurement Principle


Gas molecules absorb light at specific wavelengths, referred to as absorption lines. Oxygen
absorbs light in the near infra-red spectrum at approximately 760 nanometers. In the TDLAS
technique for oxygen detection a laser diode is used to produce light in the near IR spectrum at
~760 nanometers. The laser diode is tuned to repeatedly cross an absorption line at this
wavelength specifically for oxygen. The laser beam is focused through the gas sample containing
oxygen and onto a photodetector where the amount of light reaching the photodetector is
inversely proportional to the concentration of oxygen in the gas sample. The intensity of light
passing through the gas sample is given by the well known Beer-Lambert equation;

I I ,0 expS T g v0 N

Equation 4-1

where:
I = transmitted intensity
= laser frequency
= optical path length
N = target gas number density
I,0 = initial laser intensity
S(T) = temperature dependent line strength
g(-0) = frequency dependence of the line strength
59

Figure 4-3 Diagrammatic description of the TDLAS measurement principal (Frish, 2004,
p.5)
4.4.3 Luminescence Quenching
Optical sensing methods based on luminescence quenching have been exploited for measuring
gaseous and dissolved oxygen concentrations in aircraft fuel tank ullages and fuel de-oxygenation
systems (Morris, Miller and Limaye, 2006; Cavage, 2006). The method has also been used to aid
studies of thermally induced fuel oxidation. The method offers many attractions for use in fuel
tank applications as it uses fibre optic technologies which are intrinsically safe, can function
under harsh operating environments (temperature, pressure and hydrocarbon exposure) and can
be packaged into small, lightweight sensor housings.

4.4.4 Measurement Principle


Luminescence quenching, more commonly referred to as collisional quenching, results from
molecular collisions between an oxygen sensitive luminophore in its excited state and oxygen
molecules. Collision results in a transfer of energy from the excited luminophore to the oxygen
molecules, raising them from a ground state to an excited state. This leads to a detectable
decrease in luminescence intensity where the luminophore is said to be quenched. Oxygen
60

sensors using luminescence quenching typically feature the luminophore coated onto the end of a
probe. An optical fibre carrying light of a certain wavelength excites the luminophore coating on
the probe tip encouraging it to fluoresce. When the probe encounters oxygen molecules the
luminescence is quenched. The degree of quenching correlates to the partial pressure of oxygen in
the luminophore which is in equilibrium with the measured sample.
The partial pressure of oxygen is inversely proportional to the intensity of the luminescence
exhibited by the luminophore coating and is described by the Stern-Volmer Equation;

I0
1 k pO2
I

Equation 4-2

where;
I = luminescence intensity in the presence of oxygen
I0 = luminescence intensity in the absence of oxygen
k = Stern-Volmer constant (quantifies the quenching efficiency and sensitivity of the sensor
luminophore)
pO2 = partial pressure of oxygen
Gord, Buckner and Weaver (1995) doped aviation turbine fuels with trace amounts of pyrene and
irradiated the fuel/pyrene solutions with UV light to produce a fluorescence emission which was
quenched by dissolved oxygen. The motive was to develop an in-situ dissolved oxygen sensing
method to aid understanding of thermally induced fuel oxidation. Using a fluorimeter the
fluorescence of the pyrene/fuel solution was measured and converted to dissolved oxygen
concentrations in aviation turbine fuel using the Stern-Volmer relation. The system was capable
of measuring dissolved oxygen from saturation, (~70 ppm) down to 10 ppm. The method proved
an attractive alternative to chromatographic and electrochemical approaches offering lower cost
and non-destructive in situ monitoring of dissolved oxygen. Morris, Miller and Limaye (2006) in
studies on aviation fuel de-oxygenation methods, used a dissolved oxygen probe coated with an
61

oxygen quenched luminophore to measure the partial pressure of dissolved oxygen. The probe
enabled the dissolved oxygen content of the fuel to be related to the coking propensity of the fuel
in hot aero engine components. Cannoletta, et al. (2007) used oxygen sensing probes based on
fluorescence quenching in a fuel tank ullage to characterise the fuel tank inerting system
performance of an Alenia Aeronautica C27-J aircraft during flight test. The probes gave good
results on the ground, however the reduced temperature and pressure associated with the flight
test programme resulted in oxygen measurement issues. Further work associated with pressure
and temperature compensation of the oxygen probe signal was required to achieve accurate
oxygen measurements during flight.
With more development effort having been expended in recent years upon the TDLAS technique
for gaseous phase oxygen sensing in aircraft fuel tanks in low pressure and temperature
environments, it is this technology that presents a more attractive solution to measuring oxygen
concentration in the ullage of a fuel tank for this research work.

4.5 Conclusions
The following conclusions capture from the reviewed literature the key aspects of measuring air,
oxygen and nitrogen in aviation fuels and fuel tanks;

Methods to determine the concentration of air and air gases within aviation fuels and fuel
tank ullages fall into three main categories; physical, chemical and optical. Of these
methods the chemical and optical approaches are best suited for this research as they can
be applied in-situ within a fuel tank or provide real-time, continuous remote sampling.

Gas chromatographic methods provide the advantage of being able to measure the total
air content in both the dissolved and gaseous phases from individual determinations of
oxygen and nitrogen. Sampling complexity, difficulties of in-situ measurement at low
62

pressures and temperatures and ultimate destruction of the sample make it unviable from
this research.

Electrochemical oxygen sensors have been used successfully in measuring dissolved


oxygen concentration in aviation fuels and liquid hydrocarbons. The principal drawback
associated with this type of oxygen sensor however is the tendency for electrolyte
freezing and subsequent sensor malfunction to occur at sub-zero temperatures.
Electrolyte freezing constrains the fuel temperature range that can be investigated within
this study.

Flight test aircraft have utilised both Tunable Diode Laser Absorption Spectroscopy
(TDLAS) and luminescence quenching techniques for measuring ullage oxygen
concentrations during flight. Of these techniques TDLAS was demonstrated to be far
superior providing gaseous phase oxygen concentration measurements at temperatures
and pressures down to -40C and 20 kPa.

63

Chapter 5
Development of an Experimental Apparatus to Determine Oxygen
Evolution Rate from Aviation Turbine Fuel

5.1 Preliminary Considerations


The experimental objective is straightforward to determine the rate of oxygen evolution from
aviation turbine fuel under simulated aircraft fuel tank conditions as outlined in Chapter 1. In this
Chapter the development of an experimental apparatus for that purpose will be described.
Preliminary tests using a small-scale, thermally conditioned fuel tank were conducted to establish
the feasibility of measuring oxygen evolution rate from aviation fuel at reduced pressures and
temperatures. The pilot study enabled verification of oxygen sensing equipment performance and
development of the main experimental design prior to conducting tests at a larger scale.
Foremost in the experimental design is the fuel tank and the quantity of fuel to be used. Ideally, a
minimum volume of fuel is sought, such that its safe management within the test laboratory can
be readily achieved. Consideration must also be given to the control of fuel temperature, tank
ullage pressure and the means by which a repeatable level of fuel agitation can be achieved.
Careful selection of oxygen sensing methods in the dissolved and gaseous phases are essential to
ensure accurate, repeatable and problem free measurement over the range of temperatures and
pressures that will be encountered within the tests. An experimental apparatus will be designed to
control and measure these parameters to a high degree of accuracy. Provision will also be made
within the apparatus design to achieve dimensional similarity between the laboratory set-up and a
64

full size aircraft, such that at a later stage within the research work dimensional modelling can be
used to project test results from the laboratory fuel tank to an aircraft fuel tank.

5.1.1 Preliminary Tests


A series of preliminary tests were conducted using a pilot apparatus to develop both the
experimental method and oxygen measurement instrumentation subsequently used during the
main empirical study (Chapter 6). This apparatus consisted of a 1.6 litre thermally conditioned
fuel tank, a variable speed, magnetic stirrer mechanism for fuel agitation and a closed-loop
controlled vacuum pump for altitude simulation. Figure 5-1 shows a photograph of the pilot
apparatus set-up. These early tests proved important in establishing the feasibility of the
experimental method for measuring oxygen evolution rate from aviation turbine fuel in a
simulated aircraft fuel tank. The key findings from the pilot study are summarised below;

Oxygen evolving from the fuel increased ullage oxygen concentration in the fuel tank
from 21% up to 30 % oxygen by volume.

Oxygen concentration increase in a fuel tank ullage due to fuel outgassing appears to be
an exponential function of time.

Using a stirring mechanism within the fuel whose rotational speed could be varied proved
that the level of fuel agitation could be repeatedly controlled allowing the basic effect of
fuel agitation on oxygen release rate to be established.

Ullage oxygen sensing equipment based on Tunable Diode Laser Absorption


Spectroscopy (TDLAS) was proven to operate successfully over pressure and
temperature ranges encountered within commercial aircraft flight.

The study revealed issues with using ullage oxygen concentration measurement data to
determine oxygen release rate from the fuel. Variations in ullage gas temperature and
65

density as a function of ullage depressurisation made calculation of a mass release rate of


oxygen from the fuel difficult and prone to error.

The tests highlighted the need to measure the dissolved oxygen partial pressure using a
sensor placed directly within the fuel.

Altitude simulation
system

Thermally
conditioned fuel
tank and magnetic
stirrer

Figure 5-1 Photograph of pilot apparatus set-up

5.2 Experimental Design


5.2.1 Overall Assembly
The basic requirement for an experimental examination of oxygen evolution rate from aviation
fuel in an aircraft fuel tank environment calls for precise control of the governing parameters such
as the degree of fuel agitation, fuel temperature and rate of change of ullage pressure. The
conceptual design displayed in Figure 5-2 was assembled to meet this requirement. Aviation
66

turbine fuel within the fuel tank is initially saturated with air through an in-tank gas sparging
system. The fuel is mechanically stimulated, via a radial flow mixing impeller, which is driven
via a pneumatic motor and 4:1 reduction ratio gearbox. The fuel tank, mixing impeller with
associated drive motor and gearbox are all located within a 1 m3 thermal-altitude test chamber.
Fuel temperature and tank ullage pressure are conditioned by time dependent profiles, preprogrammed into the test chambers control system. Vent valves incorporated into the lid of the
fuel tank allow ullage air to be expelled from the tank, or air from the thermal-altitude test
chamber to be admitted, as chamber pressure varies.

Figure 5-2 Schematic & photograph of the small-scale experimental fuel tank set-up and
oxygen evolution rate measurement apparatus

Sparging describes the process of introducing gas into a liquid typically using small, finely dispersed
bubbles

67

5.2.2 Thermal-Altitude Test Chamber


A bespoke thermal-altitude test chamber (Weiss WK 1000/70-100/DS) was designed and
manufactured to simulate the pressure and temperature environments within an aircraft fuel tank
during flight. The chambers 1 m3 test space is ATEX type certified (TUV 06 ATEX 553169 X)
as explosion proof, incorporating special safety features to ensure the risk of igniting an explosive
atmosphere caused by fuel vapours is mitigated. An outline specification of the test chamber is
summarised in Table 5-1 and a photograph of the test chamber is shown in Figure 5-3.

Table 5-1 Outline specification of Weiss WK 1000/70-100/DS thermal-altitude test chamber


Temperature range

-70C to +100C

Pressure range

1-101.325 kPa

Humidity range

15-95% relative humidity

Internal volume

1 m3

68

Figure 5-3 Weiss WK/1000/70-100/DS ATEX certified thermal-altitude test chamber with
1m3 internal volume
5.2.3 Fuel Tank
The fuel tank geometry of commercial transport aircraft is complex. Typically, the wing tank
structure itself bounds the fuel between the forward and aft wing spars. The depth of fuel
contained within the wing can vary considerably between the root and tip. In the case of the
A380-800 aircraft carrying a maximum fuel load, fuel depth varies from 2.27 m in the inner tank
at the wing root to 0.34 m in the surge tank near the wing tip as shown in Figure 5-4.

69

Figure 5-4 Range of fuel depth within the Airbus A380-800 wing fuel tanks at maximum
fuel capacity (LH wing-box shown)
The principal constraint governing fuel tank size in this experimental research is the available
working volume within the test chamber in which the tank will be installed. A steel fuel tank of
square-section with internal dimensions of 0.75 m x 0.75 m x 0.3 m was designed and fabricated,
providing a maximum internal fuel volume of 0.168 m3. The fuel tank features connection flanges
for refuelling, air sparging, defuelling and oxygen sensing. The fuel tank lid is manufactured from
10mm polycarbonate sheet to enable observation of fuel outgassing behaviour during testing. The
tank lid incorporates two vent valves (Swagelok SS-CHS8 & Swagelok SS-12-C-1/3) of the
check valve type. The valves are installed to prevent air within the test chamber from entering
and mixing with the tank ullage during the simulated aircraft climb and cruise phases of a test.
Chamber air entering the fuel tank ullage during a test would tend to suppress any increase in
ullage oxygen concentration resulting from fuel outgassing, making it difficult to characterise
outgassing behaviour. Each valve has a cracking pressure of 1/3 psi that enables ullage air to be
expelled from the tank during depressurisation, or air from the thermal-altitude test chamber to be
admitted, as chamber pressure is recovered at the end of a test. This action mimics the freebreathing characteristics of an aircraft fuel tank, ensuring fuel tank ullage and altitude test
chamber pressures are always equal. Dimensions of the fuel tank are shown in Figure 5-5.
70

Figure 5-5 Experimental fuel tank dimensions (all dimensions in mm)


5.2.4 Fuel Agitation
As discussed in Chapters 1 and 4 there are a number of environmental factors responsible for fuel
agitation within an aircrafts fuel tanks. These include airframe vibration, wing bending and the
operation of pumps and jet pumps within the fuel system. Controlled simulation of the agitated
fuel tank environment, in order to understand its effect on fuel outgassing within the laboratory, is
challenging. Perhaps the most onerous task is reproducing airframe vibration and the complex
71

fuel motions and surface wave oscillations that result from this type of mechanical stimulus.
Under most circumstances the fuel system will transmit far higher levels of agitation to the fuel,
than the airframe, having the greatest effect on oxygen gas release rate. Observations of fuel
motion in studies using small transparent tanks containing aircraft fuel pumps, jet pumps and
discharging flows has demonstrated that the fuel tends to boil and swirl with a broadly radial
motion within the fuel tanks collector cells. Video footage taken during these fuel system
development studies has served as a useful means for recording this fuel flow behaviour
(Henshaw, 2007). In this experiment a method is sought by which fuel within the tank can be
displaced, in a reproducible and controlled way without the complexity associated with fuel tank
vibration. Such a method would attempt to mimic the critical action of a fuel pumps impeller on
the fuel or the internal transfer/movement of fuel both within and between tanks via jet pumps
and sequence valves. The method must be measurable with a high degree of accuracy, variable to
achieve a range of fuel agitation rates and operate reliably over a range of sub-atmospheric
pressures and hot or cold test chamber temperatures. Furthermore all equipment selected must be
safe for use in potentially flammable atmospheres and be of a non-spark producing design. To
meet with these stringent criteria a pneumatic, motor driven, mixing impeller system was
developed.
Impeller mixing plays an important role in a number of industries which include, but are not
limited to; chemicals, pharmaceuticals, petrochemicals, biotechnology, pulp and paper, food and
mineral processing (Paul, Atiemo-Obeng and Kresta, 2004). Typically the common mixing
process objectives within these industries are to promote mass transfer, chemical reaction or to
control a product within a certain process. Although some of these goals are not necessarily

72

required in this study, impeller mixing provides a convenient means of quantifying the effect of
fuel agitation on fuel outgassing rate in a controlled and repeatable way.
A pneumatic powered motor (Atlas Copco LZL 03M) and 4:1 reduction ratio gearbox (Tramec
EP90/1C 4 T/T) were selected to drive the mixing impeller. This motor and gearbox combination
provides the desired range of mixing impeller rotational speeds over the airflow range available
from the laboratory compressed air supply. A closed-loop Proportional-Integral-Derivative (PID)
control system was developed to achieve mixing impeller rotational speeds over a 200 rpm to 450
rpm range. The mixing impellers rotational shaft speed is measured with an optical speed sensor
(Compact Instruments MINI VLS 111). The optical speed sensors output signal is fed into a
digital tachometer (Contrec 405 LA) and PID controller (Data Track 331) located outside of the
thermal-altitude test chamber. The PID controller compares the actual shaft speed, measured by
the optical sensor, with a pre-programmed speed and adjusts airflow to the pneumatic motor via a
flow control valve (Burkert 2712) until the demanded and measured speeds are matched. The
targeted system accuracy is +/- 1 rpm. Figure 5-6 illustrates the closed-loop mixing impeller
system architecture.

73

Pneumatic
Motor
Atlas Copco

Exhaust Air
Compressed Air Inlet

LZL 03M
Tramec EP90/1C
4T/T 4:1 Gearbox

Data
Logger

5VDC
Regulator

15

4 20 mA I/P

IKA R1373 Mixing


Impeller

16

RS232 Config

Laser

PID Controller
Data Track 331

4 20 mA O/P

Optical Speed
Sensor MINI VLS
111

0-700 l/min
Rotameter

PC

10

Airflow
Control
Valve
24 21
--

24VDC

11.5 28.5VDC In

DC Gnd

Pullse In

8 - 24VDC Out

8 9 11 12 13

Signal Gnd

Prog Sw

1 2

4 20 Ma O/P
25

Burkert 2712
& Positioner

PSU

26

12VDC

Digital Speed Sensor Contrec 405 LA

Thandar
TS3022S

24VDC

Air Supply
(6.5 barg)

Figure 5-6 Closed-loop PID controlled mixing impeller system architecture


The impeller mixing system was set up and the fuel tank loaded with 0.126 m3 of Jet A-1 fuel at a
temperature of approximately 20C. This corresponded to the tank being 75% full, the quantity
selected to conduct all oxygen evolution rate tests. A separate, calibrated, laser tachometer
(Compact Instruments) was used to evaluate system accuracy. Reported impeller speed readings
from the Contrec sensor agreed to within +/-1 rpm of the laser tachometer over a 175 rpm to 500
rpm speed range. Figure 5-7 shows the achieved impeller rotational speed as a function of airflow
for the closed-loop impeller mixing system in Jet A-1 fuel at 20C.

74

Figure 5-7 IKA R1373 mixing impeller rotational speed in Jet A-1 fuel at 20C achieved by
the pneumatically driven impeller mixing system over the controllable airflow range
5.2.5 Impeller Pumping Capacity
As outlined in section 5.2.4 characterising the level of fuel agitation imparted by the mixing
impeller on a mass per unit time basis is necessary. To achieve this goal an experimental
approach for stirred liquid tanks will be used to determine the impellers pumping capacity from
dimensionless impeller power and flow numbers.
In a stirred liquid tank the power draw, P of an impeller is characterised by its power number, Np.
The power draw of an impeller can be calculated from the following expressions;

P Np N 3 D 5

Equation 5-1

P 2NTq

Equation 5-2

75

where N is the impeller rotational speed, D the impeller diameter, the liquid density and Tq
impeller torque. Combining Equations 5-1 and 5-2 and rearranging gives the expression for
impeller power number, Np;

Np

2Tq

Equation 5-3

N 2 D 5

Equation 5-3 gives the impeller power number in terms of the torque, liquid density, impeller
speed and diameter. Measuring impeller torque at known speeds in a liquid of constant density
thus enables determination of the impeller power number. Impeller power numbers are typically
constant in the turbulent regime where Re > 10,000. The impeller Reynolds number can be
written as;

Re

ND 2

Equation 5-4

where is liquid viscosity. The dimensionless flow number is a measure of the pumping capacity
of an impeller and is given by the following relation;

Nq

Qimp

Equation 5-5

ND 3

where Qimp is the volumetric pumping capacity. Since there is relatively little data published on
impeller flow numbers it is necessary to estimate the flow number for an impeller through the
1/3rd power-law rule (Post, 2007);

Np
Nq Nq Std 3
Np Std

Equation 5-6

The 1/3rd power-law rule enables the flow number for a particular impeller to be estimated from
known power and flow numbers for impellers of similar design.
76

A paddle type mixing impeller (IKA R1373) was selected for agitating the fuel. Dimensions of
the R1373 impeller are shown in Figure 5-8. This type of impeller produces a radial flow pattern
in the fuel tank as shown in Figure 5-9. Two circulating loops are produced, one above and one
below the impeller. This type of radial flow impeller induces high turbulence around the impeller
region, promoting good gas dispersion necessary for efficient and uniform gas sparging prior to
the start of each test. Furthermore the action of this type of impeller, in very broad terms mimics
the highly turbulent conditions arising from internal fuel system flows in the rib bays and
collector cells of aircraft fuel tanks.

8 mm

12 mm

70 mm

70 mm
Figure 5-8 Dimensions of the IKA R1373 paddle type mixing impeller

77

Figure 5-9 Radial flow pattern of a paddle type mixing impeller in a liquid tank
The first step was to measure the torque imparted by the fuel on the rotating impeller over the
mixing systems speed range (~200 rpm to 500 rpm). A transparent polycarbonate tank of
identical internal dimensions to the experimental fuel tank was manufactured and filled to a total
depth of 0.225 m with ambient temperature water. This liquid height corresponded to the tank
being 75% full, the chosen quantity of fuel that would be used throughout the oxygen evolution
rate tests.
Water was selected as a fuel simulant because it exhibits the same Newtonian behaviour as
aviation fuel in the turbulent mixing regime, (Re > 10,000). By using water in the open top,
polycarbonate tank, impeller flow patterns could be easily visualized without the problem of
hazardous fuel vapour generation in the laboratory. The mixing impeller was fitted to a
Rheometer, (R/S Brookfield), configured to measure torque and the assembly positioned on a
stable platform above the tank. Figure 5-10 shows a photograph of this experimental set-up.
Impeller torque was actually measured at set rotational speeds over a 50 rpm to 400 rpm range.
Torque measurements were repeated three times and the average value found at each speed within
78

this range. The Rheometer was unable to measure impeller torque beyond 400 rpm as the
instruments measurement range (0.00125 to 0.05 Nm) was exceeded. Impeller torque values
beyond 400 rpm were found by extrapolation with a power law regression equation fitted, using
Microsoft Excel to the test data.
Figure 5-11 shows the measured average impeller torque as a function of rotational speed with
the fitted regression equation and extrapolated torque values at 450 rpm and 500 rpm.

Figure 5-10 Photograph of the experimental set-up to measure impeller torque using a
Brookfield R/S Rheometer and water tank of identical internal dimensions to the
experimental fuel tank

79

Figure 5-11 IKA R1373 impeller torque as a function of rotational speed measured using a
Brookfield R/S Rheometer in water at 17C with power-law regression fit and extrapolated
impeller torque values at 450 and 500 rpm

Using the measured impeller torque data the power number was calculated and plotted against Re
number. Figure 5-12 shows that beyond a Re number of approximately 20,000 the value of Np is
almost constant. At Re numbers above 20,000 the flow regime is believed to be fully turbulent
and the drag force on the impeller constant (Paul, Atiemo-Obeng and Kresta, 2004). Between Re
numbers of 5,000 and 20,000 the flow is in transition and below 5,000 the flow is laminar. Using
the 1/3rd power-law rule as given in Equation 5-6 the pumping capacity of the IKA R1373
impeller, (kg/s) was calculated for the water system. Published values of Npstd and Nqstd for the
RP4 paddle impeller of 3.4 and 0.62 respectively were used as the reference impeller data (Post,
80

2011). Table 5-2 shows these calculated values from the measured torque data and Figure 5-13
illustrates the design of the RP4 paddle type impeller used as the reference.

5.2.6 Impeller Pumping Capacity in Aviation Fuel


To determine pumping capacity for the impeller running in aviation fuel at different temperatures
and impeller speeds, the differences in density and viscosity between water and fuel must be
accounted for. The impeller pumping capacity in fuel is scaled from the water pumping capacity
data using the impeller Re number. Impeller pumping capacity as a function of Re number in
water was plotted from the data shown in Table 5-2 and fitted with a linear regression equation as
shown in Figure 5-14. The impeller Re number at each rotational speed was then calculated using
values of fuel viscosity and density at each fuel temperature over the 0C to 40C range. Inputting
these Re numbers into the regression equation provides the impeller pumping capacity in aviation
fuel over the 200 rpm to 450 rpm speed range. The rate of fuel agitation was then found by
multiplying the impeller pumping capacity for fuel by the fuel density at the test temperature of
interest. Table 5-3 shows calculated fuel agitation rates over a 200 rpm to 450 rpm impeller
rotational speed range. Figure 5-15 shows the calculated fuel agitation rates plotted as a function
of impeller Re number at each fuel temperature over the 0C to 40C range in contrast with water
at 17C.

81

Table 5-2 IKA R1373 impeller pumping capacity from measured torque values in water at 17C. Data values shown in red are calculated
from extrapolated torque data
Impeller Speed (rpm)

Torque (Nm)

Power (W)

Np

Nq

Q (m3/s)

Qm (kg/s)

Re

50
(rpm)
75
100
125
150
175
200
225
250
300
350
400
450
500

0.0009
0.0013
0.00225
0.00365
0.006
0.0095
0.0115
0.015
0.175
0.250
0.03450
0.04693
0.05709
0.07063

0.0047
0.0102
0.0236
0.0478
0.0942
0.1741
0.2409
0.3534
0.4581
0.7854
1.2645
1.9656
2.6901
3.6980

4.85
3.11
3.03
3.14
3.59
4.17
3.87
3.99
3.77
3.74
3.79
3.95
3.79
3.80

0.698
0.602
0.597
0.604
0.631
0.664
0.647
0.654
0.642
0.640
0.643
0.652
0.643
0.644

0.00020
0.00026
0.00034
0.00043
0.00054
0.00066
0.00074
0.00084
0.00092
0.00110
0.00129
0.00149
0.00165
0.00184

0.2
0.26
0.34
0.43
0.54
0.66
0.74
0.84
0.92
1.1
1.29
1.49
1.65
1.84

4075
6113
8150
10188
12226
14263
16301
18338
20376
24451
28526
32601
36677
40752

82

Figure 5-12 Dimensionless plot of Np as a function of Re number for the IKA R1373
impeller running in water at 17C over a 50 rpm to 400 rpm speed range

Figure 5-13 RP4 paddle type impeller with Npstd = 3.4 and Nqstd = 0.62 (Post, 2011)

83

Figure 5-14 IKA R1373 impeller pumping capacity as a function of impeller Re number
running in water at 17C

Table 5-3 Fuel agitation rates calculated over a 200 rpm to 450 rpm impeller rotational
speed range

Impeller
Rotational
Speed
(rpm)

Re

200
300
400
450

6533
9800
13067
14700

0C Fuel
Qimp
(m3/s)

(kg/s)

Re

20C Fuel
Qimp
(m3/s)

(kg/s)

Re

40C Fuel
Qimp
(m3/s)

(kg/s)

0.00029
0.00044
0.00059
0.00066

0.234
0.355
0.476
0.536

10867
16301
21734
24451

0.00049
0.00073
0.00098
0.0010

0.386
0.583
0.78
0.879

13611
20417
27222
30625

0.00061
0.00092
0.00123
0.00139

0.479
0.722
0.966
1.087

84

Figure 5-15 Calculated fuel agitation rate as a function of impeller Re number over a 0C to
40C fuel temperature range (agitation rate of water shown for comparison)
5.2.7 Fuel Sparging
To ensure the Jet A-1 aviation fuel was saturated with air at the beginning of each test, an
efficient and repeatable method for re-introducing air into the fuel was required. Sparging, the socalled process of introducing gas into a liquid, employs fine bubble propagation, maximising the
interfacial surface area between gas and liquid for effective mass transfer. An in tank gas sparger
(Mott Corporation 2312-A04-06-A00-2-AB) was selected using the design guide procedure set
out in the sparger manufactures literature. Sparger sizing is based on the superficial gas exit
velocity from the porous sparger surface, expressed in feet per minute (fpm) and calculated from
actual cubic feet per minute per square foot of sparger surface area. The required sparger surface
area was calculated to be 0.008324 ft2 (773.32 mm2). The nearest sparger surface area provided
by the manufacturer was 0.1 ft2 (9290.3 mm2) with a 2 micron porosity, giving a more than
85

sufficient capability. A calculation of the required sparger surface area is shown in Appendix B.
A mass flow controller (Bronkhorst In-Flow), located outside of the thermal-altitude test
chamber, was used to meter dry air (dew point -75C) into the sparger located within the fuel
tank. Figure 5-16 shows a photograph of the air mass flow controller and associated air supply
equipment. In line with the sparger sizing calculation, the mass flow controller was set-up to
meter air into the fuel via the sparger at 0.4 kg/hr. Figure 5-17 shows the effectivity of the sparger
for re-saturating the fuel with air, following a test in which the fuel was partially de-gassed. As
can be seen a high rate of saturation is initially achieved levelling off after approximately 1000
seconds. The fuel is considered to be fully saturated with air at 1500 seconds since further
sparging does not change the dissolved oxygen partial pressure in the fuel.

Key
a) Air inlet to thermal-altitude test chamber and in-tank sparger
b) Non-return valve
c) 10 mm diameter manual ball valve
d) Air mass flow controller
e) Pressure regulator/ filter

Figure 5-16 Photographs of the air metering system for in-tank sparging of aviation turbine
fuel

86

Figure 5-17 Re-saturation curve for dissolved air in Jet A-1 aviation fuel whilst agitating
with the mixing impeller at 300 rpm during sparging
5.2.8 Oxygen Sensing
Oxygen sensors were installed both in the fuel (dissolved phase) and in the fuel tank ullage
(gaseous phase). This approach provides a complete picture of the air outgassing process. Oxygen
evolution rates were determined from the dissolved phase gas measurements as proven methods
for measuring dissolved oxygen concentration in hydrocarbon liquids are established. These
methods avoid the complication of changing gas temperature and density in the ullage as pressure
is reduced. Oxygen concentration measurements made in the ullage were used to determine ullage
oxygen partial pressures required for the dimensional modelling study conducted in Chapter 7. As
presented in Chapter 4, a number of oxygen sensing methods have previously been applied to the
measuring of oxygen concentration in aircraft fuel tanks. There are a number of key requirements

87

which must be met to ensure the measurements are made safely and with a high level of accuracy.
These requirements include;

Fast response time to changes in oxygen concentration

High accuracy (maximum error 0.5% of reading)

Maintain accuracy over a 20 kPa to 120 kPa ullage pressure range

Maintain accuracy over a -50C to +60C gas sample temperature range

Unaffected by liquid hydrocarbon fuels and vapours

High level of intrinsic safety for operation in potentially flammable atmospheres

5.2.9 Tunable Diode Laser Absorption Spectroscopy (TDLAS) Oxygen Sensor


The principle of operation of the TDLAS based oxygen sensor has been covered in Chapter 4.
The specificity of laser diode spectroscopy for oxygen, at 760 nm in the near infra-red spectrum,
where there is no interference from other gases, coupled with cross-sensitivity correction for
foreign gas molecules, makes TDLAS an ideal choice for oxygen concentration measurement in a
fuel tank ullage. Pressure and temperature correction within the sensor, using proprietary
algorithms, enables gas phase measurements to be made accurately between 10 kPa to 120 kPa
and -50C to +50C. A bespoke TDLAS oxygen analyser (Oxigraf O2G1) was designed and
supplied by Oxigraf Inc for measuring the concentration of oxygen in the fuel tank ullage. The
analysers on-board diaphragm pump draws a continuous ullage gas sample at ~1.2 Litres/min
from the fuel tank. A check valve at the gas sample inlet manifold creates a pressure drop, which
drives a small portion of the fast loop gas sample to and from the analysers sample cell.
Restrictor tubing and a differential pressure sensor downstream of the sample cell regulate the
flow of the measured gas sample to ~0.2 Litres/min. A schematic showing the architecture of the

88

Oxigraf O2G1 oxygen analyser can be seen in Figure 5-18. A photograph of the oxygen analyser
is shown in Figure 5-19.
Whilst performing commissioning tests with the Oxigraf oxygen sensing system and fuel in the
experimental fuel tank above 20C, it was noticed that fuel vapour was condensing out in the gas
sample lines connected to the fuel tank ullage. Even though the Oxigraf sensor configuration is
capable of accurate oxygen measurement with hydrocarbon fuel vapours entrained within the gas
sample stream, the effect of fuel condensate in the sample train system was undesirable. If a thin
film of fuel vapour condensate was to collect on the optical window of the sensors gas sampling
cell, problems with measurement accuracy would undoubtedly occur. An in-line condenser was
designed and manufactured to strip hydrocarbon fuel vapours from the gas sample, upstream of
the inlet to the Oxigraf analyser. The condenser was machined to form an aluminium cylinder and
featured a copper cooling coil embedded into the condensers exterior surface. A chilled water
and glycol solution was circulated through the cooling coil to cool the internal surface of the
condenser through which the gas sample stream is routed. A drain located at the bottom of the
vertically mounted condenser was periodically opened allowing the condensate to drain away
under gravity. Figure 5-20 shows a photograph of the condenser with the exterior insulation
removed. A detailed calculation of the condensers performance, based upon condensing heat
transfer theory, is given in Appendix C.

89

O
X
Y
G
E
N

TEE
FLOW
RESTRICTOR
TUBING

DIFFERENTIAL
PRESSURE
SENSOR

A
N
A
L
Y
S
E
R

TEE

CLIPPARD EVO-3-12
VALVE
(CAL ONLY)

LOW CAL
PORT
5% O2 by
Vol.

IN

OUT

IN
FILTER

CLIPPARD EVO-3-12
VALVE
HI CAL
PORT
35% O2 by
Vol.

IN

OUT

IN
AIRTROL PRESSURE
REGULATOR
(ORS-810-3.5)

MANIFOLD

RED SILICON
TUBING

KEY
BLANK
FAST LOOP FLOW
SAMPLE LOOP FLOW

DAMPING
VOLUME

OUT
SAMPLE PUMP

CLEAR PU TUBING

IN

NEEDLE VALVE
(Set to regulate pump
flow to 1.2L/Min)

TO TANK

CHECK
VALVE 1

F
L
O
W

TEE
INLET
MANIFOLD

CHECK
VALVE 2

SAMPLE IN
FROM
TANK

OUTLET
MANIFOLD

1.2 L/Min
M
E
T
E
R

SAMPLE OUT

Figure 5-18 Schematic of Oxigraf O2G1 TDLAS oxygen sensing system architecture
90

Figure 5-19 Photograph of the Oxigraf O2G1 TDLAS oxygen sensor used to measure
oxygen concentration in the experimental fuel tank ullage

91

Figure 5-20 Photograph of the in-line condenser designed to strip hydrocarbon fuel vapours
from the gas sample train of the Oxigraf O2G1 TDLAS oxygen sensor (exterior insulation
removed)
5.2.10 Polarographic Oxygen Sensor
An Orbisphere 3660Ex analyser and 31120E polarographic oxygen sensing system was selected
to measure dissolved oxygen partial pressure in the fuel. The sensor was mounted into the base of
the fuel tank with a purpose made, leak proof connector. The oxygen sensor was screwed into the
leak-proof connector such that the sensing head was exposed to the fuel. For correct measurement
the oxygen sensor requires a continuous flow or movement of fuel over the sensing head. This
was accomplished by ensuring the head of the sensor was placed in local proximity to the mixing
impeller. Polarographic oxygen sensor installation geometry within the fuel tank is shown in
Figure 5-21.

92

370

STIRRER

239.5

TANK
332

70

65

ORBISPHERE SENSOR HEAD

70

20

300

SPARGER

152.4

TANK BASE

405
SPACER

600
ORBISPHERE
ORBISPHERE SENSOR

LOCKING NUT

Figure 5-21 Orbisphere polarographic oxygen sensor installation geometry within the experimental fuel tank (all dimensions in mm)

93

The polarographic oxygen sensor was set-up with a 2958A type Tefzel gas permeable membrane,
which gives the sensor a 9.5 second response time to a 90% change in the sensors sample
measurement range. According to the manufacturers data, this type of oxygen sensor is limited to
dissolved oxygen partial pressure measurements between 0C and +50C. The potassium
hydroxide (KOH) electrolyte within the sensor has a freeze point of -5C. It was decided
therefore, to prevent electrolyte freezing and subsequent deterioration or loss of signal, to limit
the minimum fuel temperature to 0C in the tests. Typically in commercial transport aircraft
service, the fuel temperatures during refuel, aircraft climb and the initial phase of cruise, when the
majority of fuel outgassing occurs, are above 0C. There are, of course, exceptions to this in
colder climates, where aircraft are refuelled and the bulk fuel temperature remains sub-zero
throughout the flight mission. However, in general, this is relatively uncommon.

5.3 Simulation of Aircraft Fuel Tank Pressures


A series of pressure vs. time profiles that would represent the pressure conditions experienced
within the ullage of in-service commercial transport aircraft were required. Of particular interest
in this investigation is how the rate of change of ullage pressure affects dissolved oxygen
evolution rate. The starting point, therefore, was examination of the climb rate performances of a
wide range of commercial transport aircraft. Complicating this however, is the fact that aircraft
climb rate is not constant with increasing altitude. The climb rate is higher at lower altitudes
because engine thrust is high compared with aircraft drag forces. At higher altitudes the thrust
produced by the engines is greatly reduced and the rate of climb diminishes. Other factors such as
the weight of the aircraft, which varies with fuel and passenger load, and weather, all affect climb
performance. Figure 5-22 shows, by way of example, the climb rate performance envelope of the

94

Boeing 777-200 aircraft at different weights for altitudes up to 40,000 ft (James and ODell,
2005).

Figure 5-22 Climb rate performance envelope of the Boeing 777-200 aircraft at altitudes up
to 40,000 ft (James and ODell, 2005, p.3)
5.3.1 Climb Profiles
A total of four climb profiles were derived to cover all the possible climb performance envelopes
of commercial transport aircraft. A simplification was made, in that unlike the aircraft the rates of
climb were made linear between minimum and maximum altitude, such that the effect of this
variable on oxygen evolution rate could be clearly understood. The four linear climb profiles
range from 1000 ft/min to 5500 ft/min, up to a common altitude of 38,000 ft and are shown in
Figure 5-23.

95

Figure 5-23 Linear experimental climb profiles with rates ranging from 1000 ft/min to 5500
ft/min
5.3.2 Pressure Profiles
Pressure varies non-linearly with altitude in the atmosphere and can be represented by the
following equation, taken from the 1996 CRC Handbook relating pressure to altitude;
1

44331.514 z 0.1902632
P 100

11880.516

Equation 5-7

where P is pressure in Pascals and z is altitude in metres. The linear climb profiles were converted
into pressure vs. time profiles using Equation 5-7. Pressure vs. time profiles for each climb rate
were programmed into the thermal-altitude test chambers control system. A 3-hour isobaric
dwell at the pressure condition, equivalent to 38,000 ft, was added to each pressure vs. time
profile. The dwell is designed to simulate the high altitude cruise phase of an aircraft flight
96

mission providing the fuel with sufficient opportunity to outgas and reach an equilibrium state
with the ullage.

Figure 5-24 Pressure vs. time profiles developed to simulate the effect of aircraft altitude
The rate of change of ullage pressure was calculated for each pressure vs. time profile by
differentiating with respect to time an exponential regression equation fitted to each of the
profiles shown in Figure 5-24. Table 5-4 shows the rate of change of ullage pressure for each
climb profile.
Table 5-4 Rate of change of ullage pressure for each climb profile
Climb Rate (ft/min) Regression Equation (Pa)
1000
1.05548 E 05 e 6.99645 E 04t
2500
1.05598 E 05 e 1.75063 E 03t
4000
1.05580 E 05 e 2.80015 E 03t
5500

1.05514 E 05 e 3.84550 E 03t

97

Differential (Pa/s)
73.846 e

6.99645 E 04t

184.863 e

1.75063 E 03t

295.639 e 2.80015 E 03t


405.754 e 3.84550 E 03t

dP/dt at t = 0

73.846
184.863
295.639
405.754

The effectiveness (pressure equalisation) of the vent valve arrangement in the fuel tank was
examined by running the thermal-altitude test chamber at a constant climb rate of 2500 ft/min
from a barometric ambient pressure of 97135 Pa down to 16000 Pa (43,000 ft). A pressure
transmitter (0 kPa to 160 kPa, Druck PTX 1400) located within the lid of the fuel tank was used
to measure fuel tank ullage pressure. The fuel tank contained a 75% Jet A-1 fuel load. Figure
5-25 shows very good agreement between the fuel tank ullage and thermal-altitude test chamber
pressure readings, indicating the effectiveness of the fuel tank vent valves. Slight oscillation of
both pressure signals during the depressurisation phase is a characteristic of the thermal-altitude
test chambers vacuum control system.

Figure 5-25 Fuel tank vent valve functionality at a climb rate of 2500 ft/min and 75% fuel
load

98

5.4 Oxygen Sensor Measurement Uncertainty


Prior to commencing tests to examine the rate of oxygen gas evolution from aviation turbine fuel
using the developed apparatus, both oxygen sensor types were examined for linearity and
measurement uncertainty over a range of known oxygen gas concentrations. Examination of these
parameters in the gas phase for both sensor types was preferred, as obtaining a series of known
dissolved oxygen-in-fuel concentration samples would have been extremely difficult.

An

uncertainty of measurement analysis was conducted and the measurements for each sensor type,
was summarised in uncertainty budgets. It is important to distinguish measurement uncertainty,
which is a measure of the bounds within which a value may be reasonably presumed to lie, and
measurement error, which is the difference between an indicated value and the corresponding
presumed true value.

5.4.1 Measurement Uncertainty Calculation Methods


Two types of approach are used to estimate uncertainty in a measurement system, they are termed
Type A and Type B evaluations. Type A evaluations are statistically based, typically on a number
of repeated measurements made on the system. Type B evaluations are based on uncertainty from
other information sources, such as calibration certificates. Typically individual uncertainties,
calculated from Type A and B evaluations, are combined to give the combined uncertainty for the
system, as uncertainty exists within the system in a number of different forms.
In Type A evaluations a number of repeated measurements are made from which the mean, x and
standard deviation, s are calculated from the data set. From these the estimated standard
uncertainty, u of the mean is calculated by;

Equation 5-8

n
99

where n is the number of measurements in the set.


For Type B evaluations, where uncertainties are imported from a calibration certificate, the
standard uncertainty for a normal distribution is found from;
u

a
2

Equation 5-9

where a is the semi-range or half width between upper and lower limits on the calibration
certificate. From these evaluations the combined standard uncertainty is found by summation in
quadrature as follows;

uc u12 u 22 u32 .....u n2

Equation 5-10

Type A measurement of uncertainty evaluations were made on both the TDLAS oxygen sensor
and the polarographic oxygen sensor, using a range of oxygen-in-nitrogen calibration gas
mixtures (BOC Ltd) spanning a nominal oxygen concentration range of 5 % to 35 % oxygen by
volume. The gas mixtures were of a standard, having a 2% uncertainty of the actual certified
oxygen concentration. The compressed oxygen-in-nitrogen gas mixtures, stored in pressure
regulated cylinders, were individually fed into a partially sealed plastic bag via a supply hose.
The oxygen sensor located inside of the plastic bag was thus exposed to the gas environments.
The continuous calibration gas flow conditioned the gas volume contained within the bag to a
steady state oxygen concentration. For the TDLAS oxygen sensor the sample pumps suction
inlet hose was placed inside the bag with the pump running. Calibration gas was drawn into the
TDLASs sensing cell from the plastic bag. For each oxygen gas concentration, five repeat tests
were made and the sensors oxygen concentration recorded over time using the data acquisition
system. The sample mean for each repeated test was based upon 20 oxygen concentration data
points chosen from within the stabilised oxygen concentration section of the test data. After each
100

measurement the bag was purged with clean, dry air. Figure 5-26 shows the calibration set-up for
the polarographic oxygen sensor.

Oxygen calibration
gas supply hose

Pressure Sensor
Polarographic
Oxygen Sensor

Figure 5-26 Photograph of the Orbisphere polarographic oxygen sensor, compensation


pressure sensor and oxygen calibration gas supply hose inside the plastic bag
Figure 5-27 shows an oxygen concentration measurement plot for the TDLAS oxygen sensor,
following injection of an oxygen-in-nitrogen sample gas certified to 5.04% oxygen by volume
into the bag. A stabilised oxygen concentration can be observed to occur between ~50 and 90
seconds.

101

Figure 5-27 Oxygen concentration measurement plot for the TDLAS oxygen sensor and
oxygen-in-nitrogen calibration gas certified to 5.04 % oxygen by volume
Figure 5-28 demonstrates both the polarographic and TDLAS oxygen sensors to be highly linear
over a 5% to 35% oxygen by vol. calibration gas range. Table 5-5 shows the certified oxygen
values of the -standard calibration gases and their associated standard uncertainties assuming a
normal distribution. The calculated, combined standard uncertainty and the expanded uncertainty
for both TDLAS and polarographic oxygen sensors are shown in uncertainty analyses in Table
5-6 and

Table 5-7.

102

Figure 5-28 Measured oxygen concentrations in certified oxygen-in-nitrogen calibration


gases for the polarographic and TDLAS oxygen sensors demonstrating a high level of
linearity over a 5% to 35% oxygen concentration range

Table 5-5 Certified oxygen-in-nitrogen calibration gas mixtures and associated standard
uncertainties assuming a normal distribution
Certified Oxygen
Concentration (% vol.)
5.04

Gas Product Codes/Cylinder


Nos.
226686L-B/S1139842

Standard Uncertainty
( % vol. oxygen)
0.0252

11.97

226826L-B/S1245815

0.0599

24.98

226823L-B/S1262081

0.1249

29.08

226822L-B/S1075213

0.1454

34.78

226776L-B/S1358868

0.1739

103

Table 5-6 Uncertainty of measurement analysis of the TDLAS oxygen sensor over a 5% to
35% oxygen by vol. concentration measurement range
Source of Uncertainty

Value

5 repeated measurements with


5.04% oxygen by vol. gas

0.0154

5.04% calibration gas uncertainty

0.0504

5 repeated measurements with


11.97% oxygen by vol. gas

0.0238

11.97% calibration gas uncertainty

0.1197

5 repeated measurements with


24.98% oxygen by vol. gas

0.0186

24.98% calibration gas uncertainty

0.2498

5 repeated measurements with


29.08% oxygen by vol. gas

0.0860

29.08% calibration gas uncertainty

0.2908

5 repeated measurements with


34.78% oxygen by vol. gas

0.0518

34.78% calibration gas uncertainty

0.3478

Unit

% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.

Probability
Distribution

Divisor

Standard
Uncertainty,
u

Normal

0.0154

Normal

0.0252

Normal

0.0238

Normal

0.0599

Normal

0.0186

Normal

0.1249

Normal

0.0860

Normal

0.1454

Normal

0.0518

Normal

0.1739

Combined Standard Uncertainty

0.2871

Expanded Uncertainty

0.5742

104

Table 5-7 Uncertainty of measurement analysis of the polarographic oxygen sensor over a
5% to 30% oxygen by vol. concentration measurement range

Source of Uncertainty

Value

5 repeated measurements with


5.04% oxygen by vol. gas

0.0115

5.04% calibration gas uncertainty

0.0504

5 repeated measurements with


11.97% oxygen by vol. gas

0.0026

11.97% calibration gas uncertainty

0.1197

5 repeated measurements with


24.98% oxygen by vol. gas

0.0181

24.98% calibration gas uncertainty

0.2498

5 repeated measurements with


29.08% oxygen by vol. gas

0.0383

29.08% calibration gas uncertainty

0.2908

Unit
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.
% oxygen
by vol.

Probability
Distribution

Divisor

Standard
Uncertainty, u

Normal

0.0115

Normal

0.0252

Normal

0.0026

Normal

0.0599

Normal

0.0181

Normal

0.1249

Normal

0.0383

Normal

0.1454

Combined Standard Uncertainty

0.2071

Expanded Uncertainty

0.4142

5.5 Summary
A small-scale experimental fuel tank and associated apparatus capable of simulating the
environmental conditions present within an aircraft fuel tank, critical to the evolution of oxygen
from aviation fuel during flight has been developed. A closed-loop controlled, impeller mixing
system, required for repeatable fuel agitation within the tank has been developed and proven.
Oxygen sensors for determining both dissolved and gaseous phase oxygen concentrations in both
the fuel and tank ullage have been developed and their uncertainties of measurement established.

105

Chapter 6
Experimental Examination on the Rate of Oxygen Evolution from
Aviation Turbine Fuel
6.1 Introduction
A small-scale experimental fuel tank and test apparatus, designed to simulate the environment of
an aircraft fuel tank during flight, has been developed for the purpose of measuring fuel
outgassing rates from aviation turbine fuel. The work conducted in this Chapter has used this
experimental set-up to investigate the effects of the aircraft fuel tank environment on fuel
outgassing rate. A series of environmental variables have been investigated to establish their
relative effect on oxygen release rate and its associated time constant, () in an attempt to gain a
clear understanding of how these factors control and influence fuel outgassing behaviour. A clear
understanding of the fuel outgassing phenomenon and the factors which affect it will enable novel
design strategies to be developed to alleviate fuel tank flammability and fuel system performance
problems in both current and future commercial transport aircraft.

6.2 Experimental Design


A 3 by 4 by 4 independent experimental design was developed to investigate the effect of fuel
agitation rate, fuel temperature and the rate of change of ullage pressure on the rate of oxygen
evolution from aviation turbine fuel. The effects of two-way interactions between fuel agitation
rate, fuel temperature and rate of change of ullage pressure on oxygen evolution rate were also
examined, via statistical analysis of the test data. The experimental design features sufficient
factor levels to enable mathematical relationships to be found between the selected independent
variables (factors) and dependent variables (oxygen release rate and time constant, ) over the
106

chosen experimental ranges. The range of each factor was selected to encompass that typically
encountered within the flight envelope of commercial transport aircraft. As discussed in Chapter
5, fuel temperature has been limited to a minimum of 0C due to sub-zero operating constraints
imposed by the polarographic oxygen sensor. Although it is recognised in commercial transport
aircraft that fuel temperatures can, on occasion, reach -50C, the significant increases in fuel
viscosity and surface tension at this temperature tends to retard the rate of gas evolution. From an
environmental standpoint the worst case fuel outgassing conditions are presented by much
warmer fuel, so this experimental limitation is not considered to have a significant impact on this
study. The basic experimental approach adopted in the 48 tests conducted was to systematically
increase fuel agitation rate from test to test for combinations of fuel temperature and rate of
change of ullage pressure. Table 6-1 presents the experimental test conditions examined. Note
that the rate of change of ullage pressure values shown in Table 6-1 are initial values at the
beginning of each simulated aircraft climb profile.
Table 6-1 Experimental test conditions
Fuel Temperature
(C)
0
20
40

Fuel Agitation Rate, (kg/s)


0.234
0.386
0.479

0.355
0.583
0.722

0.476
0.780
0.966

0.537
0.879
1.087

Rate of Change of Ullage Pressure, p


(Pa/s)
73.85
184.86
295.64
405.75
73.85
184.86
295.64
405.75
73.85
184.86
295.64
405.75

6.3 Experimental Method


The fuel tank was filled with 0.126 m3 (101 kg) of Jet A-1 aviation turbine fuel. This quantity of
fuel corresponded to the tank being 75% full, giving a 3:1 fuel: ullage ratio. This ratio enabled
both ullage and dissolved oxygen concentration measurements to be made simultaneously using
the oxygen sensing systems described in Chapter 5. Following fuelling of the tank, the thermal107

altitude test chamber was programmed to run at a specific test temperature within the 0C to 40C
temperature range. The thermal-altitude test chamber conditioned the fuel and tank assembly to
the desired temperature set-point. Platinum Resistance Temperature (PRT) pt100 sensors, located
within the fuel tank, were used to monitor fuel and ullage temperature. When the desired fuel
temperature had been reached a 10 mm diameter manual ball valve, located within the air
sparging line, was opened. Dry air was then metered into the fuel through the sparger and mass
flow controller at a rate of 0.4 kg/hr. During the sparging period the fuel was stirred using the
mixing impeller set at a constant rotational speed of 300 rpm. The mixing impeller ensured that
air bubbles leaving the sparger were uniformly distributed throughout the fuel. This provided a
large interfacial gas-liquid surface area necessary for efficient air absorption. The sparging
process was complete and the fuel saturated with air when the readings of the oxygen sensors,
located within the ullage and fuel, were in equilibrium. The sparging process conducted prior to
the beginning of each test ensured the dissolved air content within the fuel was consistent at each
fuel temperature examined.
With the mixing impeller rotational speed set to provide the required rate of fuel agitation, the
thermal-altitude test chamber was started and the fuel tank ullage depressurised in accordance
with the pre-programmed pressure vs. time profile. Following ullage depressurisation the pressure
vs. time profile featured a 3 hour dwell at the minimum absolute pressure of 20.6 kPa. This dwell
simulated the ullage pressure within a fuel tank during aircraft cruise at 38000 ft. Throughout
each test, ullage and fuel oxygen concentration/partial pressure, temperature, ullage pressure, and
impeller rotational speed were measured and recorded at a sample frequency of 1 Hz with a data
acquisition system and laptop computer. A total of 48 tests combining fuel temperature, agitation
rate and rate of change of ullage pressure were conducted on the experimental set-up.
108

6.3.1 Repeatability of Experimental Method


The experimental method was found to produce highly repeatable results between tests with
identical fuel tank environmental conditions. A demonstration of this fact can be observed in
Figure 6-1 where 3 separate tests were conducted under the same experimental conditions of fuel
agitation rate, fuel temperature and rate of ullage pressure change. In each test the decay of
dissolved oxygen partial pressure measured in the fuel was highly repeatable confirming the
robustness of the experimental method and test apparatus.

Figure 6-1 Partial pressure of dissolved oxygen measured in Jet A-1 fuel as a function of
time for three separate tests conducted under identical environmental conditions to
establish experimental repeatability where =0.96707 kg/s,
temperature =20C

109

p =495.1 Pa/s and fuel

6.3.2 Calculation of Oxygen Evolution Rates


A two-step calculation procedure was used to determine oxygen evolution rates from each of the
48 test data sets. Firstly, dissolved oxygen partial pressure measurements (from the polarographic
oxygen sensor) were converted to oxygen solubility values using the ASTM D2779-92 method.
The oxygen solubility data allowed the mass of oxygen released from the fuel as a function of
time to be found. The second step involved fitting a four-parameter exponential growth model
(Janoschek) to the mass of oxygen released data (Gille and Salomon, 1995). Regression equations
describing the mass of oxygen released from the fuel as a function time were then differentiated
to find instantaneous rates of oxygen release.

6.3.3 ASTM D2779-92 Oxygen Solubility Estimates


ASTM D2779-92 is a calculation method for estimating the equilibrium solubility of air gases in
petroleum fractions with densities falling within a 630 to 980 kg/m3 range. Using this method, gas
solubility can be estimated over a 0C to 150C liquid temperature range. The method is based
upon the Clausius-Clapeyron equation, Henrys law and the perfect gas law. Empirically assigned
constants are used to account for variations in liquid density and gas type. The first step is to find
the Ostwald solubility coefficient (L) for oxygen in the fuel at the test temperature using the
following equation;
L 0.300 exp 0.639 700 T / T ln 3.333 Lo

Equation 6-1

where T is the fuel temperature and Lo is the Ostwald solubility coefficient of oxygen in
petroleum liquids at 0C with density = 850 kg/m3.

110

As the calculated value of L applies to hydrocarbon liquids with density = 850 kg/m3, correction
of L to Jet A-1s fuel density at the test temperature is required using;
Lc 7.70 L 0.980 d

Equation 6-2

where d is the density of Jet A-1 fuel. Using this value of Lc the Bunsen solubility coefficient is
calculated as follows;
B 2697 p pv Lc / T

Equation 6-3

where p is the partial pressure of dissolved oxygen measured using the polarographic oxygen
sensor in MPa, and pv the vapour pressure of the fuel at the test temperature. The vapour pressure
of Jet A-1 over the 0C to 40C test temperature range is so low it can be neglected in the
solubility calculations. Oxygen solubility is finally calculated in mg/kg from the following
expression;

BMg
T 288.6
d 1 0.000595
G

1.21

d

0.0224

Equation 6-4

where Mg is the molecular weight of oxygen in g/mol. The mass of oxygen dissolved in the fuel
at time, t in kg is calculated from the solubility values as follows;

mO2 (t )

Gmf

Equation 6-5

10 6

where mf is the mass of fuel contained within the tank. The mass of oxygen released from the fuel
into the tank ullage at time, t in kg is thus calculated from;
mO2 (t ) mO2 (t 0) mO2 (t )

Equation 6-6

The mass of oxygen dissolved in the fuel is calculated from the solubility values when the fuel is at
saturation. The calculation approach under-estimates the solubility if the fuel is supersaturated with air, and
over-estimates the solubility if the fuel is under-saturated with air.

111

Figure 6-2 shows, by way of example, the calculated mass of oxygen released from Jet A-1 fuel
as a function of time using the ASTM D2779-92 method for 3 separate tests conducted under
identical environmental conditions in the experimental fuel tank.

Figure 6-2 Mass of oxygen released from Jet A-1 fuel as a function of time calculated using
ASTM D2779-92 for three separate tests conducted under identical environmental
conditions where =0.96707 kg/s, p = 495.1 Pa/s and fuel temperature =20C
6.3.4 Non-Linear Regression Analysis
Instantaneous mass release rates of oxygen from Jet A-1 aviation fuel for each of the 48 tests
were found by fitting an exponential growth function to the oxygen mass released vs. time data
using non-linear least squares regression, and then, differentiating the regression equations.

112

A re-parameterised form of the Janoschek growth function was chosen to model the exponential
oxygen mass release behaviour;

t p
mO2 (t ) a a exp

Equation 6-7

where a represents the asymptote (total mass of oxygen released), , the time constant of oxygen
evolution and p, a shape parameter adjusting the mass of oxygen released when the rate of
oxygen release peaks. The functions first derivative was found and used to determine the rate of
oxygen mass release from the fuel at time t = ;

t p p p 1
d (mO2 )
a exp p t
dt

Equation 6-8

The Janoschek growth function was chosen as its flexibility allows application to most
exponential and sigmoid growth curves, in particular those that appear to result when dissolved
oxygen evolves from aviation fuel. At time t = the exponential term in Equation 6-7 is e 1 .The
time constant represents the time taken for 1 1 e , or 63.2% of the total mass of dissolved oxygen
that will evolve from the fuel to be transferred into the ullage. If p 1 a sigmoid curve is modelled
where d mO2 dt becomes 0 and if p<1 (exponential curve) d mO2 dt at t = 0 tends to
infinity. It must be stressed that the motive here isnt to seek the underlying physical formula that
best describes oxygen evolution, rather, the formula that best matches the measured data which
can then be used for interpolation between data points within the experimental range of
measurements. In this case one is at liberty to choose a goodness criterion, commonly the
minimum least squares deviation.
113

Figure 6-3 shows instantaneous oxygen mass release rates from Jet A-1 aviation fuel as a function
of time for three separate tests conducted under identical environmental conditions in the
experimental fuel tank.

Figure 6-3 Instantaneous mass release rates of oxygen from Jet A-1 fuel as a function of
time for three separate tests conducted under identical environmental conditions in the
experimental fuel tank where=0.96707 kg/s, p = 495.1 Pa/s and fuel temperature =20C

114

6.4 Experimental Results


6.4.1 Measurement of Oxygen Partial Pressures
As outlined earlier, the experimental method involved depressurising the ullage of a thermally
conditioned, small-scale fuel tank containing a quantity of stirred aviation turbine fuel to promote
fuel outgassing. Whilst under this simulated aircraft fuel tank environment, dissolved oxygen
partial pressure was measured within the fuel and volumetric oxygen concentration in the ullage.
An example of a typical test data set is given in Figure 6-4.

Figure 6-4 Oxygen evolution from Jet A-1 aviation fuel in the experimental fuel tank at a
temperature of 0C, fuel agitation rate of 0.355 kg/s and a 295.64 Pa/s rate of ullage
pressure change

A number of experimental observations can be made from the graphical data set presented in
Figure 6-4. Firstly, ullage oxygen concentration increases monotonically until equilibrium is
115

reached within the fuel tank ullage. As ullage oxygen concentration increases the dissolved
oxygen partial pressure decreases, illustrating the oxygen mass transfer process due to fuel
outgassing between fuel and ullage. Beyond ~4500 seconds into this test and after ullage oxygen
concentration has reached an asymptotic condition, ullage oxygen concentration begins to decay
very slightly through to the end of the test. The observed decrease in ullage oxygen concentration
beyond equilibrium is most probably due to an increasing build-up of fuel vapour within the
tanks ullage. The partial pressure of oxygen in the fuel, at and beyond this point is not decaying
significantly, indicating that the majority of dissolved oxygen has already evolved from the fuel
into the ullage. With continuous fuel vapour generation due to low pressure conditions and
mixing impeller agitation, hydrocarbon vapours will displace oxygen and nitrogen gases in the
ullage reducing their concentrations. Similar behaviour supporting this hypothesis was observed
by the US Air force in their studies on oxygen evolution from aviation fuels with varying vapour
pressures and fuel temperatures in small-scale laboratory test tanks (Roth, 1987). The fuel
temperature change during the test is within 0.5C, however a temporary sub-zero drop in ullage
temperature during the depressurisation phase is observed. Once the minimum ullage pressure is
reached and held to simulate aircraft cruise, the ullage temperature recovers and tracks well with
the fuel temperature. Whilst it exerts no influence on the fuel temperature throughout a test, this
effect, due to adiabatic cooling, is recovered by the altitude chambers thermal control system
after a short time. Ullage oxygen concentration data illustrated in Figure 6-4 was converted to
partial pressure values using the total ullage pressure measurements. Figure 6-5 shows the
relationship between dissolved and ullage oxygen partial pressures. The fuels outgassing
behaviour is clearly illustrated in the partial pressure plot where a difference in the fuel and ullage

116

oxygen partial pressures (initiated via ullage pressure reduction) drives gas evolution from the
fuel until equilibrium is finally achieved.

Figure 6-5 Dissolved and ullage oxygen partial pressures as a function of time in the
experimental fuel tank at a temperature of 0C, fuel agitation rate of 0.355 kg/s and a
295.64 Pa/s rate of ullage pressure change
Using the calculation methodology outlined in Sections 6.3.3 and 6.3.4 the total mass of oxygen
released and the instantaneous mass release rates of oxygen from the fuel as a function of time are
shown in Figure 6-6 and Figure 6-7.

117

Figure 6-6 Oxygen mass released from Jet A-1 fuel and Janoschek regression model fit as
a function of time in the experimental fuel tank at a temperature of 0C, fuel agitation
rate of 0.355 kg/s and a 295.64 Pa/s rate of ullage pressure change

Figure 6-7 Instantaneous oxygen mass release rate as a function of time for Jet A-1 fuel in
the experimental fuel tank at a temperature of 0C, fuel agitation rate of 0.355 kg/s and a
295.64 Pa/s rate of ullage pressure change
118

The instantaneous rate of oxygen evolution at a time t = was plotted as a function of each of
the three independent variables examined; fuel agitation rate, fuel temperature and rate of ullage
pressure change for all 48 test data sets.
6.4.2 Rate of Oxygen Evolution Results Fuel Agitation Rate
Figure 6-8 to Figure 6-10 illustrates the effect of fuel agitation rate at each fuel temperature over
the 0C to 40C temperature range examined. The observable trend is consistent, in that
increasing fuel agitation rate increases the rate of oxygen evolution from the fuel. Over the range
of fuel temperatures examined the effect of fuel agitation rate on oxygen evolution rate becomes
more pronounced as the rate of ullage pressure change increases. The graphical plots show clearly
that the relationship between oxygen release rate and fuel agitation rate is non-linear, being
possibly of a power-law or exponential form. The mathematical function which best describes
this non-linear relationship will be examined later in the discussion section.

119

Figure 6-8 Oxygen release rate at t = as a function of fuel agitation rate at a fuel
temperature of 20C and rate of change of ullage pressure ranging from 73.85 to 405.75
Pa/s

Figure 6-9 Oxygen release rate at t = as a function of fuel agitation rate at a fuel
temperature of 40C and rate of change of ullage pressure ranging from 73.85 to 405.75
Pa/s
120

Figure 6-10 Oxygen release rate at t = as a function of fuel agitation rate at a fuel
temperature of 0C and rate of change of ullage pressure ranging from 73.85 to 405.75 Pa/s
6.4.3 Rate of Oxygen Evolution Results Fuel Temperature
Figure 6-11 to Figure 6-14 presents oxygen evolution rate at t = as a function of fuel
temperature over a 73.85 to 405.75 Pa/s rate of change of ullage pressure range. Without
exception, the relationship between oxygen evolution rate and fuel temperature appears to be
linear over the range of fuel temperatures investigated where increasing fuel temperature
increases oxygen evolution rate. The effect of increasing fuel temperature on oxygen evolution
rate becomes more pronounced at the higher fuel agitation rates. Increasing the rate of change of
ullage pressure appears to increase the rate of oxygen evolution from the fuel at all fuel
temperatures examined.

121

Figure 6-11 Oxygen release rate at t = as a function of fuel temperature and a 73.85 Pa/s
rate of change of ullage pressure

Figure 6-12 Oxygen release rate at t = as a function of fuel temperature and a 184.86 Pa/s
rate of change of ullage pressure
122

Figure 6-13 Oxygen release rate at t = as a function of fuel temperature and a 295.64 Pa/s
rate of change of ullage pressure

Figure 6-14 Oxygen release rate at t = as a function of fuel temperature and a 405.75 Pa/s
rate of change of ullage pressure
123

6.4.4 Rate of Oxygen Evolution Results Rate of Ullage Pressure Change


Figure 6-15 to Figure 6-17 show graphical plots of oxygen release rate as a function of the rate of
change of ullage pressure. In general, the rate of change of ullage pressure appears to have had
the least impact upon the rate of oxygen release in comparison with the other independent
variables (fuel temperature and agitation rate). There is, however, one exception to this
observation. At all three fuel temperatures (0C, 20C and 40C) the rate of change of ullage
pressure has a significant effect on oxygen release rate at the highest rates of fuel agitation (0.537,
0.879 and 1.087 kg/s). These plots highlight the existence of a broadly linear trend; increasing the
rate of change of ullage pressure increases the rate of oxygen evolution from the fuel.

Figure 6-15 Oxygen release rate at t = as a function of the rate of ullage pressure change at
a fuel temperature of 20C

124

Figure 6-16 Oxygen release rate at t = as a function of the rate of ullage pressure change at
a fuel temperature of 0C

Figure 6-17 Oxygen release rate at t = as a function of the rate of ullage pressure change at
a fuel temperature of 40C
125

6.4.5 Time Constant of Oxygen Evolution Results


The test data was analysed and tabulated to establish the effect of the three independent variables;
fuel agitation rate, fuel temperature and rate of change of ullage pressure on the time constant of
oxygen evolution . The time constant represents the time taken for 1 1 e , or 63.2% of the total
mass of oxygen that will evolve from the fuel, to be transferred into the tank ullage. Values of
were calculated from parameter estimates for the Janoschek growth model, fitted to oxygen mass
released vs. time data using non-linear least squares. The Janoschek growth model is shown in
Equation 6-9;

t p
mO2 t a a exp

Equation 6-9

from which the expression for is as follows;

1
p

Equation 6-10

where k is the rate constant of oxygen evolution and p is a shaping parameter adjusting the point
of inflexion (POI) of the fitted oxygen mass released curve.
Table 6-2 presents the calculated time constants in seconds for oxygen evolution at each level of
fuel agitation rate, fuel temperature and rate of change of ullage pressure. Table 6-2 reveals the
effect of each variable on the time constant. Increasing fuel agitation rate, fuel temperature and
the rate of change of ullage pressure all reduce the time constant of oxygen evolution. This trend
demonstrates that as the time constant decreases the rate of oxygen release at a time t =
increases where a given amount (i.e. 63.2%) of dissolved oxygen gas has evolved from the fuel in
a successively shorter time period. Across all of the 48 test cases examined the time constant of
oxygen evolution spanned a 3647 to 554 second range.
126

Table 6-2 Time constants of oxygen evolution over the experimental ranges of fuel agitation, fuel temperature and rate of change of ullage
pressure

Fuel
Agitation
Rate
kg/s
0.234
0.355
0.476
0.537
0.234
0.355
0.476
0.537
0.234
0.355
0.476
0.537
0.234
0.355
0.476
0.537

Fuel Temperature
& Rate of Change
of Ullage Pressure
C
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0

Pa/s
73.85
73.85
73.85
73.85
184.86
184.86
184.86
184.86
295.64
295.64
295.64
295.64
405.75
405.75
405.75
405.75

Fuel
Agitation
Rate

seconds
3647
3025
2276
1932
3089
2314
1575
1214
3007
2139
1329
989
3061
2208
1263
812

kg/s
0.386
0.583
0.780
0.879
0.386
0.583
0.780
0.879
0.386
0.583
0.780
0.879
0.386
0.583
0.780
0.879

Fuel Temperature
& Rate of Change
of Ullage Pressure
C
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20
20

Pa/s
73.85
73.85
73.85
73.85
184.86
184.86
184.86
184.86
295.64
295.64
295.64
295.64
405.75
405.75
405.75
405.75

127

Fuel
Agitation
Rate

seconds
3013
2503
2038
1684
2310
1756
1191
1023
2111
1401
990
746
2062
1441
913
636

kg/s
0.479
0.722
0.966
1.087
0.479
0.722
0.966
1.087
0.479
0.722
0.966
1.087
0.479
0.722
0.966
1.087

Fuel Temperature
& Rate of Change
of Ullage Pressure
C
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40
40

Pa/s
73.85
73.85
73.85
73.85
184.86
184.86
184.86
184.86
295.64
295.64
295.64
295.64
405.75
405.75
405.75
405.75

seconds
2780
2430
1885
1692
1952
1477
1056
848
1909
1217
831
649
1753
1091
745
554

6.5 Statistical Analysis of Test Data


Two-way Analysis of Variance (ANOVA) was used to investigate the effect size of each
independent variable (main effects) and all possible two-way interactions between fuel
temperature, agitation rate and rate of change of ullage pressure on both the rate of oxygen
evolution and . As the independent experimental design was single replicate (each test conducted
only once), all three-way interactions were suppressed in the two-way ANOVA. Oxygen release
rate values in the ANOVA have been multiplied by 106 for numeric convenience within the
statistical analysis.
Table 6-3 presents two-way ANOVA results for oxygen release rate test data using SPSS17
statistical analysis software. All main effects (Fuel Temperature, Fuel Agitation Rate and Rate of
Change of Ullage Pressure) and all possible two-way interactions between main effects are all
statistically significant (p < 0.05) where p represents significance level. Fuel agitation rate is seen
to have the largest effect upon oxygen release rate with a partial 2 value = 0.967. In the case of ,
values presented in Table 6-4, similarly indicate that all main effects and all possible two-way
interactions are statistically significant. The partial 2 statistic provides a useful measure of effect
size and enables each individual variable and variable interactions to be ranked according to the
size of their effect upon the dependent variables (oxygen release rate and ). Higher values of
partial 2 indicate a stronger effect, where a partial 2 value of 0.14 is classed as a large
effect(Kinnear and Gray, 2010).

128

Table 6-3 Two-way ANOVA results for oxygen release rate test data
Tests of Between-Subjects Effects
Dependent Variable: Oxygen Release Rate * 10
Source

Type III Sum

Partial Eta

of Squares

Df

Mean Square

Sig.

Squared

Fuel Temperature

15.673

7.836

82.976

.000

.902

Fuel Agitation Rate

50.508

16.836

178.273

.000

.967

Rate of Change of Ullage Pressure

4.635

1.545

16.360

.000

.732

Fuel Agitation Rate * Rate of Change of

7.402

.822

8.708

.000

.813

Fuel Temperature * Fuel Agitation Rate

3.604

.601

6.360

.001

.679

Fuel Temperature *Rate of Change of

1.780

.297

3.142

.027

.512

Error

1.700

18

.094

Corrected Total

85.302

47

Ullage Pressure

Ullage Pressure

Table 6-4 Two-way ANOVA results for data


Tests of Between-Subjects Effects
Dependent Variable: Tau
Source

Type III Sum of

Mean

Partial Eta

Squares

Df

Square

Sig.

Squared

Fuel Temperature

4061144.390

2030572.195

543.706

.000

.984

Fuel Agitation Rate

8072377.298

2690792.433

720.487

.000

.996

1.576E7

5253697.699

1406.731

.000

.992

98523.688

10947.076

2.931

.025

.594

Fuel Temperature * Fuel Agitation Rate

875468.325

145911.387

39.069

.000

.929

Fuel Temperature * Rate of Change of

90647.409

15107.902

4.045

.010

.574

67224.326

18

3734.685

2.903E7

47

Rate of Change of Ullage Pressure


Fuel Agitation Rate * Rate of Change of
Ullage Pressure

Ullage Pressure
Error
Corrected Total

129

6.6 Discussion of Results


The purpose of this experimental investigation was to examine the effects of fuel agitation rate,
fuel temperature and the rate of change of ullage pressure on oxygen release rate from aviation
turbine fuel. A small-scale experimental fuel tank and associated test apparatus was developed,
enabling each variable to be systematically evaluated through a series of tests, which simulated a
range of aircraft fuel tank environments. The results of this investigation have clearly shown each
independent variable to have a statistically significant effect on oxygen release rate and its
associated time constant, .

6.6.1 Effect of Fuel Agitation Rate


The relationship between oxygen evolution rate and fuel agitation rate was shown to be nonlinear. To better understand this relationship, in terms of the effect that increasing fuel agitation
rate has on the rate of oxygen evolution, a mathematical function was sought. Samples of the test
data were transformed using semi-log, log-log and reciprocal transformations as shown in Figure
6-18 to Figure 6-20. All transformations linearised the test data to which linear regression
equations were fitted and the corresponding values of R2 compared. Prior to transformation,
oxygen release rate values were re-scaled by multiplying by a factor of 106 to achieve values in a
more manageable numeric range. Figure 6-20 demonstrates that the reciprocal model provides the
best fit to the experimental data with the highest R2 = 0.997. Coefficient values from the linear
equation in Figure 6-20 were substituted into the reciprocal model as follows;

1
1

a bx 1.298 1.039 x

Equation 6-11

Substituting values of fuel agitation rate used in the experiment for x into Equation 6-11 gives the
reciprocal model fit for the sample test data as shown in Figure 6-21.
130

Figure 6-18 Semi-log transformation and linear regression fit of oxygen release rate as a
function of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage
pressure of 295.64 Pa/s

131

Figure 6-19 Log-log transformation and linear regression fit of oxygen release rate as a
function of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage
pressure of 295.64 Pa/s

Figure 6-20 Reciprocal transformation and linear regression fit of oxygen release rate as a
function of fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage
pressure of 295.64 Pa/s
132

Figure 6-21 Reciprocal model fit to oxygen release rate multiplied by 106 as a function of
fuel agitation rate at a fuel temperature of 40C and a rate of change of ullage pressure of
295.64 Pa/s
As the value of fuel agitation rate in the reciprocal model increases, a successively larger number
is subtracted from the constant (a), thus the model predicts, as expected, an increase in oxygen
evolution rate. Conversely, if fuel agitation rate decreases then so does oxygen release rate as the
value of (a-bx) increases. The limitation of the reciprocal model occurs as (bx) approaches (a)
where oxygen release rate is predicted to increase without bound. This limitation of the reciprocal
model will be discussed later. It can be concluded therefore, that the reciprocal model is effective
at modelling the hyperbolic growth behaviour of oxygen release rate from aviation turbine fuel
with increasing agitation rate over the range of experimental conditions examined. A hypothesis
in terms of the fuel outgassing process supporting this finding will now be set out.
133

At higher fuel agitation rates more energy is imparted to the fuel through the mixing impeller, or
in the aircraft, through a fuel pumps impeller or by the fuel itself, discharging into a segregated
part of the fuel tank. The rate of mass transfer between dissolved gas in the fuel and evolving gas
bubbles increases as the mass transfer coefficient increases, due to higher levels of liquid
turbulence reaching a maximum, local to the tips of the mixing impeller. Mechanical agitation of
the fuel by impeller mixing also breaks up the evolving gas bubbles into smaller bubbles and
distributes them throughout the fuel. This action results in a greater oxygen release rate, because
the interfacial surface area available for mass transport between the fuel and gas bubbles, is
increased. Oxygen release rate should, therefore, be similarly related, via a hyperbola to the
power dissipated to the fuel by the mixing impeller. To investigate this, power numbers for the
mixing impeller, determined in water at different rotational speeds, were used to plot the rate of
oxygen release from the fuel as a function of impeller power. Power in a stirred tank is given by;

P NpN 3 D 5

Equation 6-12

From Equation 6-12 the power imparted to the fuel by the impeller over the range of fuel
agitation rates is presented in Table 6-5.
Table 6-5 Calculated impeller power dissipated to the Jet A-1 aviation fuel at 40C over the
range of fuel agitation rates investigated
Impeller Rotational
Speed, N
3.33
5
6.66
7.5

Power Number, Np

Power, P

3.87
3.74
3.95
3.79

0.19
0.62
1.54
2.11

Fuel Agitation
Rate (kg/s)
0.479
0.722
0.966
1.087

Re No.
13611
20417
27222
30625

The plot in Figure 6-22 was linearised by taking the reciprocal of the oxygen release rate and replotting as a function of the impeller power as shown in Figure 6-23. The linear regression fit to
134

this data then provided the coefficients needed to fit the reciprocal model to the data, as shown in
Figure 6-24.

Figure 6-22 Oxygen release rate 106 plotted as a function of impeller power at a fuel
temperature of 40C and a rate of change of ullage pressure of 295.64 Pa/s

135

Figure 6-23 Reciprocal of oxygen release rate 106 plotted as a function of impeller power at
a fuel temperature of 40C and a rate of change of ullage pressure of 295.64 Pa/s

136

Figure 6-24 Reciprocal model fit to oxygen release rate multiplied by 106 as a function of
impeller power at a fuel temperature of 40C and a rate of change of ullage pressure of
295.64 Pa/s
Even though at the highest impeller power value the reciprocal model tends to over-predict the
oxygen release rate, the model still fits the data very well, supporting the hypothesis. A possible
factor contributing to the over-prediction could be associated with the torque value used to
calculate the impellers power number at a 7.5 rev/s impeller speed. When impeller torque was
measured, using the Rheometer, the range of the instrument was just exceeded (> 0.050 Nm) at
this rotational speed. For this rotational speed an extrapolated torque value was determined from
the measured torque vs. rotational speed curve and used to calculate impeller power. The
extrapolated torque value may have been higher than an actual measured value.
Using coefficients generated from the linear regression equation in Figure 6-23 rather than a
reciprocal regression equation has also contributed to some error in the reciprocal model fit
137

shown in Figure 6-24. As this data is well behaved and the R2 value associated with the linear
regression fit in Figure 6-23 is close to 1, the discrepancy in R2 values between the two regression
equations is considered to be small. In fact, fitting the reciprocal model of the form shown in
Equation 6-11 to the oxygen release rate data in Figure 6-24, using non-linear regression software
has generated coefficient values of a=3.041and b=2.364, yielding an R2 value of 0.958.
The empirical relationship between oxygen release rate and fuel agitation rate determined in this
experimental study can be loosely correlated with results obtained by the Shell oil company
(Ross, 1972). Ross (1972) examined air evolution rate from kerosene-based turbine fuel under
conditions of varying fuel agitation. Reciprocal and exponential regression models have been
fitted to the Shell data as shown in Figure 6-25. It can be seen, with one exception, that at the
highest level of fuel agitation investigated the reciprocal model describes the Shell data
reasonably well. The reason for the over-prediction at the highest level of fuel agitation is most
probably associated with key differences between the two experimental methods. Firstly, Ross
(1972) determined air release rate by measuring the time taken for a volume of liquid fuel to be
expelled from the test vessel (due to pressure rise from evolving air). To achieve this, the test
vessel was completely full of fuel, containing no ullage space. The rate of pressure reduction was
very high where the time taken to reach a simulated altitude of 60000 ft above the full fuel vessel
was almost instantaneous. Agitation of the fuel did not commence until after air pressure
reduction had taken place producing a highly air supersaturated fuel. This experimental approach
did not simulate closely the conditions in an aircraft fuel tank, which has probably lead to a
different type of mathematical function providing a better description of the test data. This
finding highlights the following important point; if the behaviour of dissolved air evolution from
aviation turbine fuel in an aircraft fuel tank is to be correctly characterised, then the
138

environmental conditions within the experiment must closely emulate those in an aircraft fuel
tank during flight.

Figure 6-25 Experimental data determined by the Shell Oil company for the volumetric
release rate of air from aviation kerosene at varying levels of fuel agitation (Ross, 1972,
p.22)
6.6.2 Limitations of the Reciprocal Model
Some caution has to be exercised when applying the reciprocal model to oxygen release rate data
outside of the experimental conditions investigated in this study. According to Equation 6-11, as
the value of fuel agitation rate approaches 1.249 kg/s the oxygen release rate tends to infinity.
This is impossible in reality as other factors, outside of fuel agitation rate, such as gas-liquid
diffusivity, viscosity and surface tension (which all affect the bubble growth rate), ultimately
limit the rate at which dissolved air can be released from the fuel. Unfortunately, it was not
possible to conduct further tests at fuel agitation rates approaching this value to establish the
139

limiting rate of fuel agitation beyond which an increase in oxygen release rate would not be
observed. The maximum impeller stirring speed achievable, from the pneumatic motor and
gearbox set-up, was ~500 rpm due to flow rate limitations of the compressed air supply. Further,
at higher impeller mixing speeds a large vortex would have appeared in the fuel around the
impeller, which would be undesirable experimentally and unlikely to be representative of an
aircraft fuel tank condition.

6.6.3 Statistical Significance of Fuel Agitation Rate


A two-way ANOVA analysis showed the rate of fuel agitation, under the experimental conditions
examined, to have the strongest effect on oxygen release rate. This indicates that the rate of fuel
agitation plays the most significant role in influencing the mass transport properties of the gasliquid system, through the mechanisms previously discussed. Statistical analysis of the test data
underpins the experimental observations which lead to a clear conclusion; fuel agitation rate is the
most strongly controlling factor on oxygen release rate in an aircraft fuel tank. This finding is
consistent with that of a significant number of other workers.

6.6.4 Effect of Fuel Temperature


Increasing fuel temperature increased the rate of oxygen evolution from the fuel in all 48 test
cases examined. Over a 0C to 40C temperature range the increase in oxygen release rate with
temperature was found to be linear. As fuel temperature increased over this range, the interfacial
tension between the fuel and gas bubbles would be lowered, increasing the rate of diffusion from
dissolved gas in the fuel to the growing bubbles. As gas bubbles increase in size, the area
available for diffusion also increases, supporting further bubble growth. Gas bubble growth and
rise was observed to be the predominant mechanism by which dissolved air was released from the

140

fuel, so it is of value here to examine the effect of temperature on the diffusivity of oxygen in
aviation fuel to help explain this empirical relationship.
The Wilke-Chang correlation (Equation 6-13) as applicable to the diffusion of dilute gases in
liquids, was used to estimate the diffusion coefficient of oxygen in Jet A-1 aviation fuel over a
0C to 40C temperature range (Perry, Green and Maloney, 1997).

D AB 7.4 10 8

T B M B
~
V A0.6

Equation 6-13

Figure 6-26 shows a graphical plot of estimated diffusion coefficients for oxygen in Jet A-1 fuel
over the experimental temperature range.

Figure 6-26 Oxygen diffusion coefficient in Jet A-1 over a 0C to 40C temperature range
estimated from the Wilke-Chang correlation for diffusion of dilute gases in liquids

141

Although no other data in the literature could be found for oxygen diffusivity in aviation fuel,
Burrows and Preece (1954) reported empirically determined values for air diffusivity in vacuum
pump oils at 20C of between 2-310-5cm2/s. As can be seen in Figure 6-26, this range of values
correlates reasonably well with the values estimated for oxygen diffusivity in Jet A-1 through the
Wilke-Chang relation.
In laboratory studies of gas evolution from aviation fuels, the Atlantic Research Corporation
found that the rate constant of gas evolution from nitrogen saturated RJ-1 aviation fuel to be
approximately linear with temperature over a -1C to ~50C range (Co-ordinating Research
Council, 1958). In agreement with the trends exhibited by graphical plots in Section 6.4.3, the
rate constant of gas evolution increased approximately linearly with increasing fuel temperature.
Figure 6-27 shows a graphical plot of the rate constant as a function of fuel temperature for the
nitrogen-RJ-1 system.

Figure 6-27 Rate constant of nitrogen gas evolution as a function of fuel temperature in RJ1 fuel subjected to horizontal shaking (Coordinating Research Council, 1958, Figure 39)
142

To help quantify the effects of changes in temperature-dependent fuel properties on gas evolution
rate, the Atlantic Research Corporation derived an equation based on mass transfer theory which
related the rate constant of dissolved gas evolution to the viscosity, density and diffusivity of the
fuel. The equation states that the rate constant of gas evolution is proportional to the square root
of the Schmidt number, Sc D AB ;

K 2 D AB 1

K 1 D AB 2

1/ 2

Equation 6-14

where subscripts 1 and 2 represent fuel type 1 and 2.


Using the method of Wilke the diffusivity of the liquid phase is given as follows;

D AB

T
F

Equation 6-15

where F is equal to the molar volume of the liquid.


Combining Equations 6-14 and 6-15 allowed estimation of the effect of temperature on gas
evolution rate constant for a particular fuel type as follows;

K 2 1

K1 2

T
2 2
T1 1

Equation 6-16

where subscripts 1 and 2 refer to temperatures 1 and 2. Equation 6-16 supports the experimental
findings made in this study; a lower diffusivity yields a higher Schmidt number, which in turn
leads to a lower rate of gas evolution. Workers at the Atlantic Research Corporation found,
however, that Equation 6-16 yielded a lower bound to the ratio of gas evolution rate constants,
measured empirically at two different temperatures, for nitrogen evolving from shaken aviation
fuel. This suggests that fuel temperature not only plays a central role in changing the mass
143

transport properties which influence gas evolution rate, but in changing the thermophysical
properties of the fuel (viscosity and density), which affect its hydrodynamic behaviour under
mechanical stimulus. A viscous liquid, when agitated with a mixing impeller, would flow less
easily, exhibiting less turbulence and therefore less agitation. However, in this research work the
Reynolds numbers associated with impeller mixing at the fuel agitation rates investigated were
mainly in the turbulent regime (>10,000), where inertial forces would have overshadowed
viscous forces. So from this perspective the increase in gas evolution rate, due to increasing
temperature, is largely governed by the influence of fuel viscosity on diffusivity, rather than
changes in hydrodynamic behaviour due to a reduction in viscosity. However, in studies where
laminar or transitional impeller mixing regimes are investigated, where viscous forces are the
more dominant, the rate of gas evolution would possibly exhibit a higher sensitivity to changes in
the fuels thermophysical properties, influencing hydrodynamic behaviour. Figure 6-28 shows the
effect of fuel viscosity on diffusivity of oxygen in Jet A-1 over a 0C to 40C temperature range.

144

Figure 6-28 Diffusivity of oxygen in Jet A-1 aviation fuel as a function of fuel viscosity over
a 0C to 40C temperature range
6.6.5 Effect of Rate of Change of Ullage Pressure
At all three fuel temperatures examined the rate of change of ullage pressure had the greatest
effect on oxygen release rate at the highest rates of fuel agitation (0.537, 0.879 and 1.087 kg/s).
Below these values the rate of change of ullage pressure was seen to have little effect on oxygen
release rate. This finding appears to be supported by the statistical analysis where partial 2 values
for main effects were the lowest by a significant margin. However, it is the interaction between
the rate of change of ullage pressure and fuel agitation rate that is perhaps more important here. It
is likely that the highest levels of fuel agitation significantly lowered the level of air
supersaturation in the fuel during the most rapid ullage depressurisation. Dissolved oxygen came
out of solution at a higher rate than in comparison with tests conducted at lower agitation rates.
This experimental finding is consistent with results obtained from the sensitivity analysis
145

conducted in Chapter 7 where small changes in the rate of ullage pressure change, in the
dimensional modelling tests have little effect on oxygen release rate soon after ullage
depressurisation has begun. As will be seen in these tests, the level of fuel agitation was much
lower than the highest agitation rates examined in this empirical study. Results obtained by Beal,
Hilburger and Porter (1945) for air evolution in aviation gasoline under simulated aircraft climb,
showed that an increasing rate of climb increased the rate of oxygen evolution. Graphical data
presented by Beal, Hilburger and Porter (1945) has been re-plotted to highlight this effect and is
shown in Figure 6-29.

146

Figure 6-29 % air by vol. at STP dissolved in aviation gasoline as a function of time in
agitated and unagitated conditions at 43C (110F) (Beal, Hilburger and Porter, 1945)
Although rate of climb tends to have a greater effect on air evolution rate in aviation gasoline
than in comparison to the results obtained in this study with Jet A-1, the underpinning trend
exhibited by the two fuel types is the same. An increasing rate of ullage depressurisation has the
greatest effect on gas evolution rate when the fuel is more vigorously agitated. Beal, Hilburger
and Porter (1945) reported that in the gasoline tests, intense fuel boiling occurred when the
pressure condition reached in the ullage was equal to the liquids vapour pressure. Vapour bubble
evolution associated with fuel boiling disturbed and agitated the gasoline promoting an increase
in the rate of air evolution. Fuel agitation due to adiabatic boiling is most probably responsible for
climb rate having a more pronounced effect on air evolution rate in the aviation gasoline than
compared to Jet A-1. The boiling point altitude for aviation gasoline at 43C (110F) was 13,000
ft whereas for Jet A-1 the theoretical boiling point altitude at this temperature would be ~80,000
147

ft. Therefore in this study, fuel boiling would not have contributed towards fuel agitation which
helps to explain the lower rates of oxygen evolution observed at all but the highest rates of fuel
agitation.
6.6.6 Time Constant of Oxygen Evolution,
The time constant of oxygen evolution () was found to be a reliable indicator of how each
independent variable influenced the rate of oxygen evolution from the fuel. As will be covered in
Chapter 8, the time constant of oxygen evolution is used to model the effects of gas evolution rate
on fuel tank inerting system performance to establish the flammability risk in aircraft fuel tanks.
It is of interest therefore, to contrast the values of determined in this and other empirical
laboratory studies with those measured in aircraft fuel tanks in an attempt to understand how
closely laboratory results correlate with the aircraft.
It is apparent that even between laboratory based studies with small-scale fuel tanks there is a
large variation in measured time constants of oxygen evolution from aviation fuels. Figure 6-30
contrasts the half-life of oxygen evolution from RJ-1 fuel with that of oxygen rich air evolution
from Jet A-1 fuel. Not only are significant differences seen in the magnitudes of the time
constants but the relationship between time constant and fuel agitation is very different. In the
two laboratory test cases the method of fuel agitation differed where the Jet A-1 fuel was stirred
with a radial flow mixing impeller and the RJ-1 fuel shaken on the horizontal axis. In the case of
agitating the fuel by shaking, a wave motion was set-up in the fuel tank. Increasing shaking
frequency brought about further resonance of waves on the fuels surface, dramatically increasing
gas evolution rate and lowering half-life. Hydrodynamic behaviour of the fuel due to impeller
mixing used in this study was, in comparison, more uniform where the fuel was set in circulatory
motion, with a higher flow rate as impeller speed was increased. Whilst these differences in fuel
148

agitation method help to explain the gulf between results obtained in the laboratory, how do they
relate to the aircraft? As will be covered in Chapter 8, Boeing and the FAA derived time
constants of oxygen evolution from Jet A fuel of 100 and 3500 minutes. The 100 minute time
constant represents the gas evolution transfer rate from the fuel to ullage during aircraft climb
above 15000 ft. The 3500 minute time constant represents gas transfer rate in cruise where the
fuel is not subject to further altitude changes. In contrast with the laboratory results produced by a
number of other workers, these values are much, much higher, suggesting that the level of fuel
agitation in the aircrafts CWT was significantly lower than that examined in laboratory based
studies. The highest value of (slowest rate of oxygen evolution) measured in this study is 65%
greater than that measured in the 737-700 aircraft CWT climbing above 15,000 ft. This situation
must be considered as a key driver for establishing the time constant of oxygen evolution from
fuel in other areas of aircraft fuel tanks, such as collector cells and in the vicinity of discharging
jet pumps, where the level of fuel agitation is much higher. As will be demonstrated in Chapters 7
& 8 through Dimensional Modelling and flight testing, the rate of oxygen evolution varies
considerably in aircraft fuel tanks, reminding us that the use of generic time constants to model
the effect of gas evolution may be inadequate for fuel tank flammability studies. In a situation
where time constants of oxygen evolution are measured in a specific location, i.e. within the
CWT where the fuel is more quiescent than compared to other areas and then those time constants
are used to model the effect of oxygen evolution on fuel tank flammability for the entire aircraft
ullage, it is clear to see that a significant problem will occur. Consequently, in this situation the
period of time that the aircraft fuel tank is flammable will be drastically underestimated by the
analysis.

149

Figure 6-30 Comparison of the half-life of oxygen evolution from oxygen saturated RJ-1
fuel with for oxygen evolution from Jet A-1. Both fuels were subjected to agitation by
either shaking or stirring at a temperature of ~20C

150

6.7 Conclusions
The basic theme explored within this experimental work has been to investigate the effect of
environmental variables present within an aircraft fuel tank on the rate of oxygen evolution from
aviation fuel. Concluding remarks will now be made.

Effect of Environmental Variables on Oxygen Evolution Rate


Fuel agitation rate was shown to have the greatest effect on the rate of oxygen evolution from Jet
A-1 aviation fuel. The statistical significance of the test data underpins this finding. Oxygen
evolution rate exhibited a hyperbolic increase with increasing fuel agitation rate over the range of
experimental conditions examined. This behaviour was mathematically modelled with a
rectangular hyperbola.

Temperature influenced the rate of oxygen evolution from the fuel by altering fuel viscosity,
which plays a central role in the diffusivity of oxygen in the fuel. Increasing fuel temperature
lowers fuel viscosity which increases diffusivity, leading to increased rates of gas evolution. The
overall effect of fuel temperature on oxygen evolution rate was not as pronounced as that for fuel
agitation rate.

The rate of change of ullage pressure was observed to have only a marginal influence on the rate
of oxygen evolution from the fuel at lower rates of fuel agitation. At the highest fuel agitation
rates, the rate of change of ullage pressure significantly increased the rate of oxygen evolution
from the fuel. This experimental finding was consistent with the behaviour of dissolved air
evolution from aviation gasoline under simulated aircraft climb conditions.
151

Effect of Experimental Conditions on Gas Evolution Rate


The environmental conditions employed in experimental examination of oxygen evolution rate
from aviation turbine fuel are of critical importance to the correct characterisation of fuel
outgassing behaviour. If fuel outgassing behaviour in aircraft fuel tanks is to be correctly
understood, the environmental conditions simulated in the laboratory must closely emulate the
aircraft fuel tank environment.

Time Constant of Oxygen Evolution


The time constant of oxygen evolution provided a consistent measure of the influence of each
environmental variable on the rate of oxygen evolution from Jet A-1 aviation fuel. The
experimental study showed that the time constant of oxygen evolution can vary dramatically,
depending on the environmental conditions in an aircraft fuel tank. It is, therefore, recommended
that the use of universal time constants to model fuel outgassing rate in different aircraft fuel
tanks, for flammability analyses, is avoided in favour of an approach that adopts the use of
specific time constants for a particular aircraft type.

152

Chapter 7
Dimensional Modelling of Oxygen Evolution Rate from Aviation
Turbine Fuel in Aircraft Fuel Tanks
7.1 Introduction
The work conducted within this chapter aims to understand how measurements of oxygen
evolution rate from aviation turbine fuel made in the laboratory relate to oxygen evolution rates in
aircraft fuel tanks. What effect does experimental scale have on the magnitude of oxygen
evolution rate from aviation turbine fuel? As previously discussed, measuring the rate of oxygen
evolution from aviation fuel within the laboratory presents its challenges, however, measurement
of this phenomenon in a flying aircraft is far more onerous. The cost and complexity of making
fuel tank oxygen measurements in flight test aircraft is extremely high. A method is sought,
therefore, in which measurements of oxygen evolution rate from small-scale laboratory fuel tanks
can be used to determine aircraft fuel tank oxygen evolution rates. Within this Chapter oxygen
evolution rates measured in the laboratory will be used to estimate the oxygen evolution rate in an
aircraft fuel tank by dimensional modelling. The importance of using a similarity approach in
extrapolating data from laboratory scale experiments to the aircraft is also discussed and the
results contrasted with those of other workers who employed an alternative experimental
approach.
The dimensional modelling method offers a key advantage in that the basic equations governing
the phenomenon need not be known. Instead it is only necessary to understand the underlying
physical phenomenon to the extent that only the relevant variables, parameters and constants are
identified. This aspect of the dimensional method is a key advantage over other modelling tools
153

such as CFD where the analysis must begin with a mathematical model of the physical problem
(Bakker, 2006). It is recognised however that for certain fluid mechanical studies CFD offers
other advantages such as reduced experimental costs, expediency of analysis and the ability to
cover a much greater range within the parameter space than is practical or sometimes possible
with dimensional modelling. The lack of a mathematical model describing fuel outgassing rate in
conjunction with access to experimental facilities in which to conduct fuel tank based
measurements has made dimensional modelling the primary choice in this study.

7.2 Dimensional Similarity and Modelling


Studies on a physical model can furnish useful qualitative information on the characteristics of a
prototype (full size article) (Langhaar, 1980; Szirtes, 2007). Usually though, quantitative
information is also sought and in many cases the primary result of a model study is a single
numerical value. The numerical value obtained from a test on a model depends on the values of
the independent variables in the problem.
Using dimensional analysis, the behaviour of a physical system can be described by
Buckinghams theorem (Buckingham, 1914);

1 f 2 , 3 ,... n

Equation 7-1

In Buckinghams theorem the behaviour of the physical system is defined by a set of


dimensionless variables, formed from the physical variables relevant to that system. If the two
systems have the same numerical values for all the defining dimensionless variables then the two
systems are dimensionally similar. Their behaviour can be closely correlated and the results of
measurements on either one can be projected to the other. Dimensional similarity is the
cornerstone of dimensional modelling and achieving it, between model and prototype, is

154

imperative if measurements on the model are to be successfully projected to the prototype


(Szirtes, 2007).
To be effective, the modelling exercise must be carried out in distinct phases in a well-established
chronological order;
i.

Establish the theoretical background, define relevant variables and most importantly the
Model Law

ii.

Design the model

iii.

Build the model

iv.

Execute the experiments and tests

v.

Evaluate the data

vi.

Draw conclusions

7.3 Identifying Relevant Variables


The first step in the dimensional modelling process is to identify all the variables relevant to the
rate of oxygen evolution in aviation turbine fuel. Studies conducted by previous workers have
examined the effect of fuel agitation, fuel temperature, rate of change of ullage pressure, gas
solubility, tank size/shape and the level of air supersaturation on the gas evolution rate.
The literature survey of this previous work, detailed in Chapter 2, now provides a valuable
information source from which key variables relevant to the rate of oxygen evolution can be
selected. Further, the statistical significance of the results of the experimental investigation
conducted in Chapter 6 confirms the validity of some of these aforementioned variables. The
selected variables relevant to the rate of oxygen evolution from aviation fuel are presented in
Table 7-1. The dimension for each variable is given using the SI (kilogramme, metre, second)
dimensional system.
155

Table 7-1 Variables relevant to the rate of oxygen evolution from aviation turbine fuel using
the SI dimensional system
Variable

Symbol
m O2

oxygen evolution rate


partial pressure of oxygen in ullage
partial pressure of oxygen dissolved in fuel
fuel agitation rate
fuel surface tension
rate of change of ullage pressure
mass of fuel in tank

pu
pf

f
p

mf

Dimension
kg/s
kg/(m.s2)
kg/(m.s2)
kg/s
kg/s2
(kg/m.s2)/s
kg

Rationale supporting selection of each of the variables listed in Table 7-1 will now be discussed.
Partial Pressures of oxygen in Ullage (pu) and Fuel (pf)
As an aircraft climbs, the ambient pressure reduces above the fuels surface in the tank as
pressure is equalised via the atmospheric vent system. Previous workers have shown that
dissolved air evolution is not directly proportional to this pressure reduction. Initially, if the fuel
is quiescent no air will be released. The level of air supersaturation in the fuel increases until a
critical difference between the fuels dissolved gas partial pressures and the tank ullage partial
pressures initiates air evolution. Dissolved oxygen partial pressure, also known as fugacity
(meaning escaping tendency), is related to dissolved oxygen concentration via Henrys law.
Oxygen fugacity measurements made in the fuel, together with an estimation of the Henrys law
constant under the temperature and pressure conditions of the experiment, allow the fuels
dissolved oxygen content to be determined and from this the rate of oxygen release estimated.
Fuel Surface Tension (f)
Numerous theoretical and experimental investigations have examined the effects of surface
tension on gas bubble growth rate for bubbles of certain sizes in liquids. The liquids surface
tension acts as a barrier to interfacial mass transfer where increasing surface tension retards the
156

rate of gas bubble growth. In agitated aviation fuel at sub-atmospheric pressures, bubble
formation and growth has been observed and is recognised as the governing mechanism by which
the gas evolution process occurs. Thus fuel surface tension, a factor influencing bubble formation
and growth rate, is considered central to this gas evolution rate modelling study. The surface
tension of aviation turbine fuels varies linearly with temperature where in-service fuel tank
temperature changes can bring about significant shifts in fuel surface tension.
Fuel Agitation Rate (
The degree of fuel agitation has been shown to have a significant affect on the rate of gas
evolution from aviation fuel. Higher levels of agitation intensity increase the rate of mass transfer
between the gas and liquid phases. Mechanical agitation of the fuel plays a vital role, breaking up
larger gas bubbles into smaller ones, increasing the interfacial surface area available between gas
and liquid for mass transfer to take place. Higher numbers of smaller, well dispersed bubbles in
the fuel provides a greater number of gas bubble nucleation sites, thereby increasing gas release
rate. As the degree of fuel agitation can vary considerably within aircraft fuel tanks and for the
aforementioned reasons fuel agitation is a key variable within this modelling study.
Rate of Change of Ullage Pressure ( p )
Commercial transport aircraft climb at different rates. Engine performance, weather, fuel and
passenger loads all contribute towards in-service aircraft climb performance. The rate of pressure
reduction in the ullage of an aircraft fuel tank can, therefore, vary considerably within the flight
envelope. As reducing ullage pressure creates a difference between dissolved and gaseous partial
pressures of oxygen in the fuel and ullage, which is the mechanism responsible for driving gas
evolution, it is prudent to consider how the rate at which pressure reduction occurs affects the rate
of gas evolution.
157

Mass of Fuel in Tank (mf)


This variable is closely linked with the fuel agitation rate. In a tank containing a small quantity of
fuel, mechanically agitated by an impeller of specific pumping capacity, the total mass of fuel in
the tank will be displaced by the impeller in a much shorter period of time than in comparison to
the tank containing a much larger mass of fuel operating at the same specific pumping capacity.
The power density (W/m3) imparted by the mechanical agitation is much higher for the smaller
fuel quantity, thus overall the level of fuel agitation is higher. During the course of a flight the
aircraft will have, overall, a continually varying fuel mass (reducing) due to fuel burn. Further,
the mass of fuel determines the quantity of gas available for outgassing. Small quantities of fuel
will de-gas very quickly.

7.3.1 Consideration of Other Physical Variables


The following variables were, after consideration discounted as being relevant to the gas
evolution rate modelling study. Supporting rationale is given.
Fuel Tank Size, Shape and Orientation
Studies conducted by the Atlantic Research Corporation revealed that there was a trend toward
reduced gas evolution rates from aviation fuel with increasing fuel tank size, where the fuel was
agitated by shaking the tank (Co-ordinating Research Council, 1958). Fuel tank shape and
orientation were found to affect gas evolution rate through the effect on wave motion during
shaking. It would imply that under this regime the fuels surface area is less relevant with respect
to the rate of gas evolution than the combined effects of fuel tank size, shape and orientation on
the fuels hydrodynamic behaviour. Wave motion within the tank will ultimately govern the level
of agitation imparted to the fuel. As the chosen method of fuel agitation in this study is
significantly different to that achieved via shaking, any benefits of investigating tank size, shape
158

and orientation on gas evolution rate are thus lost. Although tank size, shape and orientation may
not be physically irrelevant, it is believed that their effects on gas evolution rate are second-order
and overshadowed by fuel agitation. In an attempt therefore to reduce the number of
dimensionless variables and the complexity of the modelling study these variables are dispensed
with.
Fuel Temperature
As discussed in previous Chapters fuel temperature influences the thermophysical fuel properties
(surface tension, viscosity and density) central to the mass transport processes responsible for
bubble growth in aviation fuel. Having selecting fuel surface tension as physically relevant for the
reasons previously described and noting it to be a linear function of temperature, it is not
necessary to account for fuel temperature separately.
Viscosity and Density
Fuel viscosity and density have been shown to influence the rate of gas evolution from aviation
fuels by either altering the diffusivity of the gas in the liquid or by changing the hydrodynamic
behaviour of the fuel under mechanical stimulus. In the latter the effect of changes in the fuels
hydrodynamic behaviour due to changes in these properties has been discounted with the method
of agitation used as impeller Re numbers are into the turbulent regime. Although the influence of
viscosity on diffusivity is important, the net effect of viscosity on gas evolution rate is likely to be
similar to that of fuel surface tension as these thermophysical properties behave similarly in
aviation fuel as functions of temperature. On this basis and in the interests of keeping the
dimensional modelling approach as simple as possible viscosity and density are not required.

159

7.4 The Dimensional Method


By the Szirtes algorithm the variables in Table 7-1 were used to construct the Dimensional Set
(Szirtes, 2007). The Dimensional Set consists of 4 matrices. The elements of the A and B
matrices are simply the exponents of the fundamental dimensions involved in the particular
variable. The C matrix is determined from the Fundamental Formula;

C D A 1 B

Equation 7-2

and the D matrix is an Identity Matrix.

Accordingly the Dimensional Set for oxygen evolution rate follows as;

B matrix
m O2

pu

pf

mf

-1

-1

-1

kg

-1

-2

-2

-3

-1

-2

-1

-1

-2

-1

-1

A matrix

C matrix

D matrix
Figure 7-1 The Dimensional Set for oxygen evolution rate from aviation turbine fuel

Thus we have 7 physical variables and 3 fundamental dimensions. Therefore, by Buckinghams


theorem, we have 4 dimensionless variables. They are, by the Dimensional Set presented above;

160

m O2

; 2

f mf
p m f
pu
; 3
; 4
2
pf
pf

Equation 7-3

where as required all of the above variables have the dimension of 1.

7.4.1 Sequence of Physical Variables in the Dimensional Set


The sequence of physical variables appearing in the Dimensional Set presented in Figure 7-1was
arranged such that the dependent variable, m O2 features in the leftmost position. In accordance
with the dimensional method, the placement of the dependent variable in the leftmost position of
the Dimensional Set is usually advantageous as it can be expressed easily and explicitly (Szirtes,
2007). This fact then greatly facilitates the manipulation of the variables in the Dimensional Set.
It follows that this arrangement is not a necessity but a convenience. Variables, pf and pu were
arbitrarily split between the A and B matrices respectively as no two or more variables, with
identical dimensions and hence identical columns, may exist within the A matrix as the value of
A must be non-singular.

7.4.2 Scale Factors


Scale factors facilitate and enhance the dimensional modelling procedure; scale factors make the
handling of data easier, the technique more comprehensible and application of the process to a
given problem simpler. Scale factors are associated with the particular physical variables and
therefore in any dimensional modelling study there are exactly as many scale factors as there are
physical variables. The scale factor, with respect to a physical variable, is the quotient of the
magnitudes of that variable for the prototype and its model.
For example, if the mass of the fuel in an aircraft fuel tank was m f1 = 3000 kg, and of its model
m f 2 = 100 kg, then the Fuel Mass Scale Factor is;

161

Smf

m f2
m f1

100
0.033
3000

Equation 7-4

Note that in the quotient the models value is always in the numerator and bears the subscript 2,
whilst the prototypes value, in this case, the aircraft, is always in the denominator, designated 1.
From the variables given in Equation 7-3 the scale factors are written as follows;

S m O2

m O2 2
m O21

; S pu

pu 2
pu1

; S pf

p f2
p f1

; S f

f2
f1

; S

mf
2
p
; S m f 2 ; S p 2
1
m f1
p 1

Equation 7-5

7.4.3 The Model Law


The Model Law is a set of relations amongst scale factors relevant to a particular modelling
instance (Szirtes, 2007). Even though more than one relation amongst scale factors may exist,
there is only one Model Law. Thus a Model Law may comprise several relations, each of which
is part of the same Model Law. As the Model Law is the principal design tool in a modelling
experiment, it must be determined before the experiments are conducted or apparatus designed. In
this dimensional modelling activity constraints such as the test chamber volume, laboratory air
supply and safety regulations, discussed in Chapter 5, has limited the number of aircraft design
cases to which results can be extrapolated with the derived Model Law. However, as will be seen
in Section 7.5, sufficient capacity within the apparatus design has been incorporated to achieve
dimensional similarity with an aircraft case. The Model Law is determined from the scale factors
in Equation 7-5 and the variables in Equation 7-3 as;

S m O2 S ; S pu S p f ; S f

Smf
S2

; Smf

S S p f
S2
; S p
S f
Smf

Equation 7-6

The scale factor for fuel surface tension becomes equal to unity if the aviation fuel (Jet A-1) in
the small-scale experimental fuel tank (model) is maintained at the same temperature as the
162

prototype (aircraft fuel tank). By reducing ullage pressure in the model tank to the same condition
experienced by an aircraft during flight, ullage and fuel oxygen partial pressures in the model
tank and prototype will reach similar values (due to fuel outgassing) as equilibrium is eventually
established. Taking advantage of the fact that aviation fuel will obey Henrys law for both
systems, the Model Law is further simplified to;
S m O2 S ; S m f S2 ; S p

S
Smf

Equation 7-7

The scale factor relations in Equation 7-7 constitute the Model Law of the oxygen evolution
system, by which the experimental model will be designed.

7.5 Model Design


As discussed in Section 7.4.3 the Model Law is used to design the physical fuel tank model i.e.
the quantity of fuel, rate of ullage pressure change and rate of fuel agitation needed to ensure
dimensional similarity with the chosen aircraft case. The fuel type used in the model (Jet A-1)
will be the same as the aircraft. Fuel temperature between model and aircraft is to be common,
such that the simplified Model Law in Equation 7-7 can be used.

7.5.1 Flight Test Data


To design the laboratory model, historic flight test data (flight No. 0073, 18/08/2009) from an
A320-200 aircraft (MSN 659) was examined to determine typical values for a commercial
transport aircrafts rate of climb, fuel tank temperature, fuel quantity and rate of engine fuel burn
(Bonjour, 2009). Figure 7-2 shows a plot of these parameters during the flight. Analysis of engine
fuel feed system parameters under these flight conditions provide the fuel agitation rate within the
inner wing tank collector cell. Using these values for the aircraft in the Model Law (Equation 77), enables values for the rate of ullage pressure change, fuel mass and fuel agitation rate to be
163

found for the model. The laboratory model can then be designed appropriately using the correct
scale factors.

Figure 7-2 Flight test data from flight No.0073 from the A320-200 MSN 659 aircraft
(Bonjour, 2009, p.14)
Table 7-2 summarises the values selected from the flight test data history of flight No. 0073 for
use in design of the laboratory model whilst Figure 7-3 illustrates the basic configuration of the
A320-200 fuel tanks.

164

Table 7-2 Summary of flight test data taken from flight No.0073 on MSN 659 required for
design of the laboratory model (Bonjour, 2009, p.14)
Parameter
Aircraft altitude at Top-of-Climb
Time taken for aircraft to reach 11582.4 m
Aircraft avg. rate of climb
LH inner tank fuel quantity
LH inner tank collector cell fuel quantity
LH inner tank collector cell fuel temperature
Engine No. 1 fuel burn rate

Value
Dimension
11582.4
m
1080
s
10.72
m/s
2460
kg
1038.4
kg
20
C
0.333
kg/s

Figure 7-3 Fuel tank configuration of the Airbus A320-200 aircraft (Walker, 2005, p.5)

7.5.2 A320 Engine Fuel Feed System Performance Analysis


In the A320 aircraft a total of three centrifugal boost pumps, located within the wing and centre
tanks, deliver fuel to each engine fuel system. The wing pumps, located in collector cells, feature
sequence valves which reduce their fuel delivery pressure to a value below that of the centre
tank pump, biasing fuel delivery from the centre tank. When the centre tank fuel quantity falls to
a minimum level the pump is switched off and engine fuel supply is maintained by the collector
165

cell pumps. During aircraft operation the mass of fuel held within the collector cells is
continuously agitated by fuel re-circulating through the pumps and sequence valves. This level of
agitation varies as the engine fuel flow rate changes during flight. Fuel flow from the pumps is
also provided to drive water scavenge jet pumps (WSJP) located in the surge and outer wing
tanks and to cool engine oil. Thus the collector cell fuel agitation rate is simply the mass
discharge rate of fuel from the sequence valves, estimated from the engine fuel feed rate (engine
burn rate), jet pump motive flow, engine oil fuel-cooling flow and fuel pump performance curves.
Table 7-3 shows the calculated collector cell fuel agitation rate for the A320-200 aircraft
determined by fuel system performance analysis under the flight test conditions of flight No.0073
when the aircraft had reached the top of climb (TOC) at 11582.4 m (38000 ft).
Table 7-3 Collector cell fuel agitation rate calculated from A320-200 MSN 659 flight test
data (flight No.0073) and analysis of engine fuel feed system performance parameters
Parameter
WSJP motive flow
Engine No. 1 fuel burn rate
Engine oil fuel cooling flow
Single boost pump performance characteristic at
11582.4 m (38000 ft) and 20C fuel temperature
Collector cell fuel agitation rate

Value (kg/s)
0.125
0.333
0.333
1.944
3.097

Figure 7-4 shows a schematic of the A320-200 engine fuel feed system architecture and the fueloil cooling arrangement of the aircrafts International Aero Engine V2500 engine variant. The
photograph in Figure 7-5 shows the collector cell fuel boost pumps and sequence valve
arrangement.

166

Figure 7-4 Schematic of the A320-200 engine fuel feed system architecture (left hand wing and centre tank shown) (Walker, 2005, p.38)
167

Sequence Valves

Boost Pumps

Figure 7-5 Photograph of an A320-200 aircraft inner wing collector cell illustrating fuel
boost pumps and sequence valves from which fuel is ejected resulting in collector cell fuel
agitation. View is from the wing front spar looking aft towards RIB 1
7.5.3 Model Fuel Agitation Rate (2)
The simplified Model Law in Equation 7-7 states that the Fuel Mass Scale Factor is proportional
to the square of the Fuel Agitation Rate Scale Factor;

S m f S2

Equation 7-8

In Chapter 5 a small-scale experimental fuel tank was designed, based upon the available volume
within the thermal-altitude test chamber. Maximising this available volume, the fuel tank was
designed with an internal volume of 0.16875 m3. Filling the tank 75% full gives a fuel volume of
0.1265625 m3 and a fuel mass in the model tank of 101.25 kg based on a fuel density for Jet A-1
of 800 kg/m3 at 20C.

168

Using the mass of fuel identified in the aircraft inner wing tank collector cell at the top of aircraft
climb in Table 7-2, the Fuel Agitation Rate Scale Factor is given by;

101.25
0.31226
1038.4

Equation 7-9

The collector cell fuel agitation rate in the aircraft was calculated as 3.097 kg/s, therefore the fuel
agitation rate required in the model is;

2 0.31226 3.097 0.96707 kg / s

Equation 7-10

Linear interpolation, using a linear regression equation fitted to the pumping capacity data of the
IKA R1373 impeller for Jet A-1 fuel at 20C (Figure 7-6), gives a required impeller rotational
speed of 324.82 rpm for the calculated model fuel agitation rate. The impeller will thus be driven
at 325 rpm in the model tests.

Figure 7-6 IKA R1373 mixing impeller pumping capacity as a function of impeller
rotational speed
169

7.5.4 Model Rate of Ullage Pressure Change ( p 2 )


Examining the test data history of flight No. 0073 revealed that the A320 aircraft climbed linearly
from just above sea-level to 11582.4 m (38000 ft) in 18 minutes. The average rate of climb,
therefore, was 643.46 m/min (2111 ft/min). This aircraft climb rate was used to generate a linear
altitude vs. time profile from which a corresponding pressure vs. time curve was generated, using
the following equation for pressure in terms of altitude in the standard atmosphere;
1

44331.514 z 0.1902632
Pa 100

11880.516

Equation 7-11

To determine the rate of change of ullage pressure in the A320 inner wing tank an exponential
regression equation was fitted to the pressure vs. time curve, using Microsoft Excel and then
differentiated with respect to time (Figure 7-7). This gave instantaneous rates of ullage pressure
change in the aircrafts inner wing tank at 1 second time intervals. At t = 0 the rate of change of
ullage pressure was equal to 154.58 Pa/s. From the Model Law the Rate of Change of Ullage
Pressure Scale Factor, S p is given by;

S p

S
Smf

Equation 7-12

where;
S p

0.31226
3.20266
0.09750

Equation 7-13

from which the corresponding models rate of change of ullage pressure at t = 0 is;
p 2 3.20266 154.58 495.0671828 Pa / s

Equation 7-14

A pressure vs. time curve for the model was generated by multiplying the rate constant in the
exponential regression equation fitted to the aircraft pressure vs. time data by the Rate of Change
170

of Ullage Pressure Scale Factor, S p given in Equation 7-13. Pressure and time data points from
this profile were programmed into the Weiss WK1000 thermal-altitude test chambers control
system. Figure 7-8 shows the pressure vs. time curve for the model. The slight discrepancy
between the intercepts of the aircraft and model pressure curves on the pressure axis at t = 0 is
due to the fit of the Excel generated regression equation.

Figure 7-7 Pressure as a function of time curve generated from Equation 7-11 and
exponential regression equation fit for the A320-200 aircraft climb to 11582.4 m in flight
No.0073

171

Figure 7-8 Ullage pressure and altitude as a function of time profiles for A320-200 aircraft
(flight No.0073) and model determined from flight test data and the Model Law

7.6 Model Tests


The experimental fuel tank (physical model) was installed in the thermal-altitude test chamber
and filled with Jet A-1 aviation turbine fuel to 75% of its internal capacity. With the fuel at 20C
this corresponded to a fuel mass of 101.25 kg in the tank. Ullage pressure within the model fuel
tank was reduced at a rate equivalent to an aircraft climbing linearly at 2061.18 m/min. Vent
valves located within the lid of the model tank allowed ullage air to be expelled from the tank, or
air from the thermal-altitude test chamber to be admitted. The fuel was mechanically stimulated
using a mixing impeller (IKA Model No. R1373) rotating at 325 rpm. Impeller rotational speed
was regulated to within +/-1 rpm using a closed loop PID control system. Fuel and ullage
temperatures were also conditioned within the test chamber and model fuel tank to 20C +/-1 C.

172

7.6.1 Surface Tension Measurements


The surface tension of a sample of Jet A-1 fuel taken from the same batch as that used in the
model tests was measured using the Du Nouy ring method over a -20C to +40C temperature
range. Measurements were repeated twice at each temperature and the average value compared
with surface tension values published in the CRC Aviation Fuels Properties Handbook.
Table 7-4 Surface tension measurements of Jet A-1 aviation fuel used in the model tests
compared to published CRC Fuels Properties Handbook surface tension values for Jet A1(Intertek, 2011)
Fuel Temperature C
-20
-10
0
10
20
30
40

Surface Tension (N/m) Du


Nouy ring method
0.03
0.03
0.0295
0.03
0.029
0.0295
0.0294
0.0291
0.028
0.0282
0.0275
0.0275
0.0255
0.026

Avg. (N/m)

CRC Data (N/m)

0.03
0.02975
0.02925
0.02925
0.0281
0.0275
0.02575

0.0264
0.0257
0.025
0.0242
0.0236
0.0226
0.0218

Over the -20C to +40C fuel temperature range, the Du Nouy ring measured fuel surface
tensions were higher than those published by the CRC whose values were based on the theoretical
correlation of Ramsey and Shields. For this dimensional modelling study, the average value of
surface tension measured using the Du Nouy ring method at 20C will be used.

173

Figure 7-9 Physical model test set-up


7.6.2 Measurement of Oxygen Partial Pressures in the Model
The % volume fraction of oxygen within the model fuel tank ullage was measured in real-time
using an Oxigraf O2G1 analyser. This analyser utilises Tunable Diode Laser Absorption
Spectroscopy (TDLAS) to measure oxygen concentration. A gas sample is drawn continuously
from the model fuel tank ullage into the analysers sample cell via an on-board diaphragm pump.
The measurement is pressure and temperature compensated by the analysers hard-coded
algorithms to minimise error over the pressure and temperature ranges encountered within the
tanks ullage during a test.
Measurement of dissolved oxygen partial pressure in the fuel was made using an Orbisphere 3660
polarographic oxygen sensor and analyser. This sensor was located inside the base of the fuel
tank adjacent to the tip of the mixing impeller. The sensors location relative to the mixing
174

impeller ensures a continuous flow of fuel is achieved over the face of the sensor, sufficient to
renew the oxygen diffusion layer at the sensors gas permeable membrane for stable and accurate
readings.The sensor is constructed from two metal electrodes immersed in an electrolytic solution
and separated from the fuel with a Tefzel gas permeable membrane. An electrical potential is
applied between the two electrodes to reduce oxygen that is driven through the membrane by a
partial pressure gradient. An electrical current is generated, proportional in magnitude to the
partial pressure of dissolved oxygen in the fuel.
The partial pressure of oxygen within the ullage was calculated in units of kPa from the Oxigraf
analysers concentration readings (O2) and the total pressure in the ullage as follows;
p u p t O2

Equation 7-15

Prior to the beginning of a test, the fuel was saturated with dry air (dew point -75C) using a gas
sparger (Mott Corporation) positioned within the tank. Air was introduced into the fuel via the
sparger at a constant rate of 0.4 kg/hr until the % volume fraction of oxygen in the ullage and the
partial pressure of dissolved oxygen were in equilibrium. Equilibrium was reached when the
readings from the two sensors were approximately equal in magnitude and stable to within 0.1 %
oxygen by volume in the ullage and 0.1 kPa in the fuel. Partial pressures of oxygen in the ullage
and fuel, ullage pressure, temperature and mixing impeller rotational speed were logged at a
frequency of 1 Hz using a National Instruments SCXI 1100 data-logger, PC and Labview 7.1
software. The test was repeated 3 times under identical conditions to assess experimental
repeatability.

175

7.7 Results
A total of 3 tests were performed on the model fuel tank. Oxygen concentrations measured in the
tank ullage were converted to oxygen partial pressures (kPa) using the total ullage pressure
readings. Dissolved oxygen partial pressures (kPa), proportional to the concentration of oxygen
dissolved in the fuel, were used to calculate the rates of oxygen outgassing using the ASTM
D2779-92 method and regression analysis approach set out in Chapter 6.
Figure 7-10 shows the calculated mass of oxygen released from the fuel as a function of time for
the 3 repeated tests.

Figure 7-10 Mass of oxygen released from Jet A-1 aviation fuel as a function of time in the
model tests
7.7.1 Rate of Oxygen Mass Release
Inspection of Figure 7-10 revealed the mass of oxygen released from the fuel increased
exponentially with time until some asymptotic mass of oxygen had been released. Instantaneous
176

mass release rates of oxygen for each test were found by fitting the Janoschek exponential growth
function to the oxygen mass release data using non-linear least squares and then, differentiating
the function at a particular time, t.
Figure 7-11 presents the instantaneous mass release rates of oxygen from Jet A-1 aviation fuel in
the dimensionally similar laboratory fuel tank.

Figure 7-11 Instantaneous mass release rates of oxygen from Jet A-1 aviation fuel as a
function of time in 3 separate model tests with identical test conditions
Table 7-5 provides the instantaneous values of oxygen mass release rate for each test at time t =

177

Table 7-5 Instantaneous oxygen mass release rates at t = and time constants of oxygen
evolution measured from Jet A-1 fuel in the dimensionally similar model tests

Test No.
1
2
3
Avg.

m O2 t (kg/s)
1.5082910-6
1.5111310-6
1.4895910-6
1.50310-6

(Seconds)
1287.99
1314.13
1281.72
1294.61

Using an average oxygen mass release rate value from the 3 model tests in Table 7-5, the sought
after rate of oxygen evolution in the A320 aircraft collector cell could now be found from the
Model Law;
S m O2 S

where from Equation 7-9;


S m O2 0.31226

from which, for the A320 aircraft case, the rate of oxygen mass release is given as;

m O21

1.503 10 6
4.8132 10 6 kg / s
0.31226

A summary of the dimensional modelling results shown in Table 7-6 indicates that for model and
prototype dimensional similarity has been successfully achieved where the values of each
dimensionless variable for model and prototype is almost identical.

178

Table 7-6 Summary of dimensional modelling results on the rate of oxygen evolution from
aviation fuel in an A320 aircraft fuel tank
Variable
name
release rate of
oxygen
partial pressure
of oxygen in
ullage at 38 kft
fuel surface
tension at 20C
rate of change
of ullage
pressure
fuel agitation
rate
partial pressure
of oxygen
dissolved in
fuel at 38 kft
mass of fuel
dimensionless
dimensionless
dimensionless
dimensionless
categories of
variables

Scale factor S
symbol

dimension

prototype

model

m O2

kg/s

4.813210-6

1.50310-6

pu

kg/s2.m =
(Pa)

5718.46

kg/s2 =
(N/m)

model/prototype

Category
prototype

model

0.31226

5718.46

0.0281

0.0281

(kg/s2.m)/s =
(Pa/s)

154.58

495.0671828

3.20266

kg/s

3.097

0.96707

0.31226

pf

kg/s2.m =
(Pa)

5060.11

5060.11

mf
1
2
3
4
1
2
3

kg
1038.4
101.25
0.09750
1
1.554110-6 1.554110-6
1
1.13
1.13
1
3.0422
3.0421
1
10.2427
10.2433
freely chosen, a priori given, or determined independently
determined by application of the model law
determined by measurement on the model

1
-

2
-

7.8 Discussion
Dimensional modelling has been used to estimate the mass release rate of oxygen within an A320
aircraft fuel tank from measurements made on a physical laboratory model. By using the Model
Law, established for oxygen release rate, it is clear to see the values measured on the physical
model differ appreciably from the aircraft projected value. This result re-iterates the importance
of ensuring dimensional similarity between the physical model and the aircraft case if oxygen
evolution data, gathered from laboratory testing, is to be used for aircraft fuel tank analysis.
179

Incorrect values for the rate of oxygen mass release from the fuel will lead to estimation errors of
ullage oxygen concentration used in flammability analyses. The impact of this may be farreaching, where in the worst case the percentage of flight time the fuel tank was flammable is
underestimated within the analysis. Oxygen evolution rate results obtained through dimensional
modelling will now be discussed and contrasted with those of other workers.

7.8.1 Oxygen Evolution Rate Measurements from Dimensionally Dissimilar Models


In the 1950s the Royal Aircraft Establishment (RAE) measured the rates of air, oxygen and
nitrogen gas evolution from aviation kerosene under simulated flight conditions (Bedwell, 1952).
Their experimental approach didnt consider dimensional similarity with a full size aircraft,
although from a geometric standpoint the depth of fuel contained in the experimental glass vessel
replicated that of a shallow aircraft wing tank, approximately 10 inches. The volumes of air gases
released from the kerosene were determined at specific sub-atmospheric pressures by measuring
pressure changes with a differential manometer in the head-space of the fuel vessel, relative to an
evacuated vessel, held at the sub-atmospheric pressure of interest. The fuel volume, quiescent
throughout the test period, was approximately 1 litre and was thermostatically conditioned to
23C 1C. The rate of pressure reduction was equal to an aircraft climbing at 3000 ft/min.
Results for oxygen release rate from the RAE tests are given in Table 7-7 and contrasted with
those obtained in Test 1 (Figure 7-11) of this dimensional modelling study.

180

Table 7-7 Oxygen release rates measured by the RAE under simulated flight conditions in
quiescent aviation kerosene using a dimensionally dissimilar laboratory model in
comparison with rates of oxygen release measured in the dimensionally similar model fuel
tank (Bedwell, 1952, p.10)
Altitude kft.
Mean

Time (mins)

Total amount of
oxygen released
VS/100 VL

Rate of oxygen
release

VS
100V L
t

*Rate of oxygen
release (RAE)
(kg/s)

Rate of oxygen
release (Test 1)
(kg/s)

5.99210-6
4.51410-6
3.73810-6

7.5
22.5
37.5
45.0
45.0

5
10
16
37
162

6.85
10.50
13.26
13.47
15.49

(ltr/min/100 ltr)
1.37
0.73
0.46
0.010
0.016

3.00710-7
1.60210-7
1.00910-7
2.19410-9
3.51110-9

45.0

234

16.35

0.012

2.63310-9

Rate of oxygen release at normal pressure (101.3 kPa)


* Rate of oxygen release at altitude pressure condition

Rates of oxygen release in litres/min/100 litres fuel in Table 7-7 at the altitude cases examined
were converted to oxygen mass release rates in kg/s such that direct comparison with
measurements in the dimensional modelling tests could be made. Appendix D details with an
example calculation the approach taken for converting the RAE oxygen release rates from
litres/min/100 litres fuel into kg/s. Clearly the rates of oxygen evolution measured by the RAE are
much lower than those measured in the A320 similarity model. The principal reason for this is
due to the effect of fuel agitation on oxygen release rate in the dimensionally similar model. In
the RAE tests the fuel would have become highly supersaturated with air during the simulated
aircraft climb until a point in time was reached when the fuel was agitated and equilibrium
eventually established. For most commercial transport aircraft highly supersaturated fuel
conditions are considered unlikely to occur in wing tanks and collector cells due to airframe
vibration and fuel recirculation agitating the fuel. The volumetric release rate unit used by the
181

RAE suggests that the release rate of oxygen can be scaled linearly with fuel volume as the values
quoted in Table 7-7 are per 100 litres of fuel, even though only just less than 1 litre of fuel was
used in their tests. Although it is not mentioned as to how their laboratory results relate to an
aircraft case, it is possible to compare both linear and similarity law scaling methods for the A320
aircraft case. The RAEs oxygen release rate value of 1.00910-7 kg/s measured in their model at
37500 ft is thus scaled to find the oxygen release rate from the volume of fuel held in the A320
collector cell as follows;

1038.4
4
m O2 1.009 10 7
1000 1.309 10 kg/s
800

Equation 7-16

where fuel density at 20C = 800 kg/m3 and mass of fuel in the A320 collector cell =1038.4 kg.
By comparison the rate of oxygen release measured in the A320 physical model at 37500 ft from
the Model Law in Equation 7-7;

m O2

3.738 10 6
1.197 10 5 kg/s
0.31226

Equation 7-17

It is apparent that despite the RAEs quiescent fuel conditions yielding lower oxygen release rate
values than those measured in the agitated physical model, the RAEs linear scaling method
actually over-estimates the rate of oxygen release in the A320 aircraft collector cell in comparison
with the Model Law. This highlights an extremely important point made earlier; environmental
conditions and scaling laws must be correctly defined in the laboratory model if oxygen release
rate data is to be projected accurately to full scale aircraft fuel tanks.
The application of dimensional modelling to the problem of fuel outgassing has shown that
dimensional similarity can be achieved between model and prototype with a geometrically
dissimilar model. In the case of oxygen evolution rate from agitated aviation fuel, fuel tank
182

geometry is not thought to play a significant role as the regime by which outgassing occurs is
dominated by bubble growth and rise rather than surface diffusion. This is highly convenient and
allows the experimenter a free hand to shape the model without the need to reproduce the
complex geometry of an aircraft fuel tank at a scale suitable for use in the laboratory.
7.8.2 Reducing the Number of Dimensionless Variables Fusion of Physical Variables
In Section 7.4.3 the Model Law was simplified by using Jet A-1 fuel at the same temperature in
the physical model as in the chosen A320 aircraft case where the Fuel Surface Tension Scale
Factor equals unity. Observing the fact that aviation turbine fuels obey Henrys law, a further
simplification of the Model Law was made by reducing ullage pressure in the model to the same
pressure condition as the A320 aircraft fuel tank. With absolute ullage pressures between model
and aircraft at 38,000 ft being common, eventual equilibration of partial gas pressures across the
fuel surface would lead to both Fuel and Ullage Oxygen Partial Pressure Scale Factors equalling
unity. A demonstration of this behaviour is shown below from the test data in Figure 7-12 which
appears to support this Model Law simplification.

183

Figure 7-12 Oxygen partial pressures as a function of time due to outgassing of dissolved air
from aviation turbine fuel under reduced ullage pressure and agitated fuel conditions
during Test 2.
As discussed previously, it is the difference in partial pressures between the fuel and ullage that
drives gas evolution which is clearly shown in Figure 7-12. This provides an opportunity to
reduce the number of dimensionless variables describing the behaviour of the physical system
from 4 to 3. Reducing the number of dimensionless variables makes the modelling process easier
to handle, both analytically and graphically, and will be used to great advantage when analysing
the effect of physical variables on gas evolution rate behaviour.
The first step is to combine the physical variables pu and pf given in Table 7-1 into delta oxygen
partial pressure between fuel and ullage (pf pu).

184

Using the Szirtes algorithm the Dimensional Set given in Figure 7-1 is reconstructed as follows;

m O2

mf

(pf -pu)

-1

-1

kg

-1

-2

-1

-3

-2

-1

-2

-1

-1

Figure 7-13 Reconstruction of the Dimensional Set with variables pu and pf combined (pf pu)
With 6 physical variables and 3 fundamental dimensions by Buckinghams theorem we have
reduced the number of dimensionless variables from 4 to 3 by combining pu and pf as follows;

m O2

; 2

m f f

; 3

pu f

Equation 7-18

from which the Model Law follows as;

S m O2 S ; S m f S2 ; S p

1
S

Equation 7-19

Inspection of the Model Law shows it to give the same numerical values for the Scale Factors as
those achieved in Equation 7-7. Combining physical variables pu and pf better describes the
process of gas evolution with a reduced number of dimensionless variables, simplifying the
dimensional modelling process. The further advantage is that the Model Law remains unchanged,

185

so the design of the physical model is still valid upon which to conduct further tests allowing
projection of laboratory data to the aircraft.

7.8.3 Altering Variable Sequence in the Dimensional Set


As mentioned in Section 7.4.1 the sequence of variables appearing in the dimensional set can be
arbitrary provided some basic rules are followed; the dependent variable is placed in the leftmost
position of the B matrix, and no two or more variables with identical dimensions exist in the A
matrix. To prove that this feature of the method does not affect the end result the mass of fuel in
the tank, (mf) and the rate of change of ullage pressure,( p ) variables are swapped between the A
and B matrices forming a new dimensional set;

B matrix

m O2

pu

mf

pf

-1

-1

-1

kg

-1

-2

-2

-1

-2

-3

-1

-1

-1

-1

-1

-1

A matrix

C matrix

D matrix

Figure 7-14 Re-arranged variables in the Dimensional Set


from which the following dimensionless variables result;

m O2

; 2

f pf
p m f
pu
; 3
; 4
pf
p
pf

186

Equation 7-20

The action of swapping the sequence of variables in the original dimensional set, shown in Figure
7-1, has changed the form of the dimensionless variable 3 appearing in Equation 7-20. Although
this change in 3 alters the subsequently derived Model Law from that shown in Equation 7-7, it
can be demonstrated that both Model Laws provide identical scale factor values. From Equation
7-20 the simplified Model Law is given as;

S m O2 S ; S S p 1 ; S S m f S p

Equation 7-21

Using the scale factor value for S p = 3.20266 from Table 7-6 to determine S from Equation 721 gives;

1
0.31224
3.20266

where the values of S from the two types of Model Law are almost identical. Similarly the value
for S m f 0.09750 in Table 7-6 is confirmed;

S 0.31224

0.0975
S p 3.20266
7.8.4 Relationship between Dimensionless Variables
A mathematical relationship between dimensionless variables 1 and 3 in Equation 7-22 during
ullage depressuriation was revealed by the dimensionless plot shown in Figure 7-15. This single
dimensionless plot provides useful insight into how the independent physical variables, contained
in 3, influence the rate of oxygen evolution. In all 3 tests both surface tension and rate of fuel
agitation were held constant. Thus, the combined effects of (pf -pu) and p are jointly responsible
for the logarithmic decay in oxygen evolution rate from the fuel. There are two key questions
arising at this point. Firstly, which of these independent variables has the greatest effect on
187

O2 to changes in these variables? To


oxygen evolution rate and secondly, how sensitive is m
answer these questions the first step was to inspect the relationship between the independent
variables, (pf -pu) and p . A graphical plot presented in Figure 7-16 shows (pf -pu) p for the
majority of the depressurisation phase, suggesting each variable has an equal effect upon oxygen

O2 an
evolution rate decay. Working on the basis that each variable has an equal effect on m
expression for oxygen release rate is developed in terms of just one variable. Selecting p as the
independent variable the logarithmic regression equation shown in Figure 7-15 can be re-written
in terms of 1 and 3 giving;

m O2

1.0279 10 6 ln

pu f

2.0217 10 6

Equation 7-22

multiplying through by gives;

m O2 1.0279 10 6 ln

pu f

2.0217 10 6

Equation 7-23

letting A = (1.0279 10 6 ) and b = (2.0217 10 6 ) the expression for (pf -pu) p from
Figure 7-16 is substituted into Equation 7-23;

16.01 0.031 p f
m O2 A ln
p

Equation 7-24

O2 in terms of experimental values of p and the results


Equation 7-24 was used to predict m
plotted and compared with the experimental data.

188

Figure 7-15 Dimensionless plot of 1 as a function of 3 using data from Test 1 in the ullage
depressurisation phase

Figure 7-16 Relationship between (pf -pu) and p represented by a linear mathematical
function
189

O2 in terms of p during
Figure 7-17 An empirical model describing the behaviour of m
ullage depressurisation in Test 1
It can be seen from Figure 7-17 that a very good correlation between the empirical model and test
data from Test 1 during ullage depressurisation has been achieved.

7.8.5 Analytical Sensitivity Analysis


The sensitivity of m O2 to small changes in p in the empirical model (Equation 7-24) will now be
examined using an analytical approach. Both absolute and relative sensitivity will be determined.
Absolute Sensitivity
An empirical model was developed from plotting dimensionless variables;

16.01 0.031 p f
m O2 A ln
p

Equation 7-25

O2 is the rate of oxygen evolution from


where p is the rate of change of ullage pressure and m
the fuel. All other parameters in the empirical model were held constant in the dimensional
190

O2 with respect
modelling tests. From Equation 7-25 the rate of change of oxygen evolution m
to the rate of change of ullage pressure p may be determined directly by differentiation, thus,
re-arranging Equation 7-25,

16.01 0.031 p
f
m O2 Aln
ln

Equation 7-26

therefore

16.01
dm O2
p
A

dp
16.01 0.031 p p 2

Equation 7-27

dm O2
16.01 A

dp
p 16.01 0.031 p

Equation 7-28

Now for small changes in p ( p ), to a first approximation is given as;

m O2 dm O2

p
dp

Equation 7-29

therefore;

m O2 p

dm O2
dp

Equation 7-30

O2 (i.e. m O2 ) for a (defined) proportional change


Thus the absolute change in m

p
in p may
p

be determined as;

m O2

p
dm O2
p
p
dp

Equation 7-31

191

Changes in the value of p from 1 to 10% over the range encountered within the tests, i.e.~500 to
100 Pa/s, were inserted into Equation 7-28 and the effect on the absolute change in oxygen
evolution rate calculated using Equation 7-31.

O2 to % changes in p . The
The plot presented in Figure 7-18 shows the absolute sensitivity of m
O2 to small changes in p increases as the value of p increases noting that %
sensitivity of m
increases in the rate of change of ullage pressure leads to a decrease in the rate of oxygen
evolution. Figure 7-18 highlights a particularly interesting feature of oxygen evolution rate
behaviour. At the beginning of a test the rate at which the ullage is depressurised has a significant
effect on the reduction in gas evolution rate. This effect diminishes rapidly as a greater quantity
of oxygen is liberated from the fuel and the rate of change of ullage pressure decreases.

Figure 7-18 Absolute sensitivity of oxygen evolution rate resulting from a 1 to 10% change
in the rate of change of ullage pressures simulated in the dimensional modelling tests
192

Relative Sensitivity

O2 i.e.
To find the relative change in m

m O2
for a proportional change in p , then from
p

Equation 7-28;

m O2
1 p dm O2

m O2
m O2 p
dp

Equation 7-32

O2 can be determined from Equation 7-25 and


where on the right hand side of Equation 7-32, m
dm O2
can be determined from Equation 7-28 for any prescribed value of p .
dp
Figure 7-19 shows the relative change in oxygen evolution rate to % changes over a 1 to 10%
range in rate of change of ullage pressure.

Figure 7-19 Relative sensitivity of oxygen evolution rate resulting from a 1 to 10% change in
the rate of change of ullage pressure simulated in the dimensional modelling tests
193

7.9 Incorrect Determination of Impeller Rotational Speed


In Section 7.5.3 an error was made when calculating the required impeller rotational speed to
achieve a fuel agitation rate (2) of 0.96707 kg/s in the dimensional modelling tests. The error
resulted from incorrect scaling between impeller pumping capacity in water and impeller
pumping capacity in fuel. The correct method of scaling using the impeller Reynolds number, as
described and used in Section 5.2.6 of Chapter 5 accounts for the differences in viscosity and
density between the two liquid types, from which, the correct impeller speed can be determined.
The impeller pumping capacity in fuel at 20C was calculated from the water pumping data and
plotted as a function of the impeller Re number over a 50 rpm to 500 rpm impeller speed range,
as shown in Figure 7-20.

Figure 7-20 Pumping capacity of IKA R1373 impeller running in Jet A-1 aviation fuel at
20C as a function of impeller Re number

194

From the linear regression equation fitted to the data in Figure 7-20, the impeller Re number at
the required model fuel agitation rate (2) of 0.96707 kg/s is found;

0.96707 0.0000363 Re 0.007559

Equation 7-33

Re-arranging Equation 7-31 to give impeller Re number;

Re

0.96707 0.007559
26674
0.0000363

Equation 7-34

From Equation 5-4 the impeller Re number is;

Re

ND 2

The impellers rotational speed, required to generate a fuel agitation rate at 20C of 0.96707 kg/s
in the model fuel tank is given by;

0.001196689 26674
8.18 rps
796.2 (0.07) 2
N 8.18 60 491 rpm
N

Equation 7-35

Ideally, 3 dimensional modelling tests, repeated under identical conditions would have been rerun at an impeller speed of 491 rpm to generate new oxygen release rate values. From these
values a projected rate of oxygen evolution in the A320 aircraft could have been calculated using
the derived Model Law. Unfortunately, the experimental apparatus had been decommissioned
prior to the detection of this error and the opportunity to conduct further experimental testing was
lost.
As shown in Chapter 6 the rate of oxygen evolution was seen to increase with increasing fuel
agitation rate and the behaviour modelled with a rectangular hyperbola. The effect therefore of
using a lower than required fuel agitation rate (0.633 kg/s) at an impeller Re number of 17657,
195

and rotational speed of 325 rpm, has led to the oxygen release rate being under-estimated in this
modelling study.

196

7.10 Conclusions
The principal aim set out within this phase of the research was to understand how the rates of
oxygen evolution from aviation turbine fuel, measured within a model fuel tank, relate to oxygen
evolution rates in an aircraft fuel tank. The following conclusions summarise the key findings;

It has been made possible through dimensional modelling, to estimate the rate of oxygen
evolution from aviation turbine fuel in an aircraft fuel tank, from measurements made on a
dimensionally similar physical model. A dimensional modelling study has revealed the rate of
oxygen evolution from aviation fuel in an A320 aircraft fuel tank to be ~200% greater than the
rate actually measured in a laboratory model.

Careful design of the laboratory model, through derivation of appropriate scaling factors, is
essential to ensure accurate projection of oxygen release rate data from the physical model to the
aircraft. Previous laboratory studies conducted into gas evolution rate from aviation fuels have
failed to achieve dimensional similarity between model and prototype, restricting the use of gas
evolution data from such studies for aircraft analysis.

Oxygen evolution rate values obtained by the RAE from a dimensionally dissimilar physical
model when extrapolated to an A320 aircraft fuel tank differ significantly from those determined
in the dimensional modelling study. The RAEs linear scaling method, in comparison to the
dimensional modelling approach, overestimates the rate of oxygen release from aviation fuel in
an aircraft fuel tank.

197

The number of dimensionless variables applicable to the rate of oxygen evolution from aviation
turbine fuel was reduced from 4 to 3 by combining the physical variables pf and pu to form a new
physical variable, (pf pu). Reducing the number of dimensionless variables simplified the
analysis of how each physical variable affects the rate of oxygen evolution from aviation fuel.

A single graphical relationship between dimensionless variables, applicable to the rate of oxygen
evolution from aviation fuel, has revealed interesting behavioural characteristics of the oxygen
evolution process;
i)

O2 decreases logarithmically with decreasing p


At a constant rate of fuel agitation m
and (pf pu). Both p and (pf pu) were found to have an approximately equal effect

O2 .
upon m
ii)

O2 was significant at the beginning of


The absolute magnitude of their effect upon m
ullage depressurisation when the rate of change of ullage pressure was high.
However, this was seen to rapidly diminish as more oxygen was released from the
fuel and the rate of ullage pressure change fell.

Dimensional modelling of oxygen evolution rate from aviation turbine fuel, requires only that
dimensional similarity between the physical model and aircraft fuel tank be achieved without the
need for geometrical similarity. This is highly convenient as the complex geometry associated
with an aircraft fuel tank, at a scale suitable for use in the laboratory, does not need to be
reproduced, allowing time and cost savings to be made in the modelling process.

198

Use of an incorrect impeller rotational speed (325 rpm) in the dimensional modelling tests has
resulted in a lower than required fuel agitation rate in the model fuel tank. The extrapolated
oxygen evolution rate value in the A320 aircraft fuel tank, determined from measurements made
on the dimensionally similar model is therefore lower than the value that could be expected if the
correct fuel agitation rate (0.96707 kg/s) had been used.

The dimensional modelling method may provide an alternative approach to flight testing for
determining the rate of fuel outgassing in aircraft fuel tanks. To gain greater confidence in the
method, and the derived Model Law, it is imperative that further dimensional modelling studies
are conducted using fuel tanks containing larger quantities of fuel at higher agitation rates. If the
Model Law is valid, oxygen release rates measured during these tests should correlate with the
aircraft values projected from the smaller scale tests conducted in this study. Reducing the
amount of flight testing previously conducted to determine oxygen release rate through an
increased use of dimensional modelling will provide the aerospace industry with significant cost
and timescale saving benefits. Dimensional modelling will enable the effects of a broader range
of environmental conditions affecting the rate of oxygen evolution to be examined, thus
improving the accuracy of flammability analyses for future aircraft designs.

199

Chapter 8
Effect of Fuel Outgassing on Oxygen Concentration in an Aircraft Fuel
Tank Ullage
8.1 Introduction
Experimental work conducted in Chapter 6, using a small-scale fuel tank, revealed how
environmental variables within an aircraft fuel tank influence the rate of oxygen evolution from
aviation fuel. In Chapter 7, oxygen evolution rates from the experimental fuel tank (physical
model) were extrapolated to a full scale aircraft fuel tank using dimensional modelling to
understand the effects of experimental scale on the fuel outgassing phenomenon. Finally, the key
objective within this Chapter is to understand how fuel outgassing affects the flammability of an
aircraft fuel tank ullage during flight.
The evolution of oxygen from aviation fuel can significantly affect the flammability of an aircraft
fuel tank and limit the effectiveness of Flammability Reduction Measures (FRMs) designed to
mitigate flammability issues. The fuel outgassing phenomenon has driven airworthiness
authorities to mandate that for aircraft fitted with an FRM, such as Fuel Tank Inerting (FTI) the
outgassing of fuel must be taken into account within fuel tank flammability analyses. The effect
of fuel outgassing on the performance of a FRM must be quantified such that analysis tools can
provide accurate assessment of fuel tank flammability. The Fuel Tank Flammability Assessment
Method (FTFAM), engendered by the Federal Aviation Administration (FAA), uses exponential
time constants of oxygen evolution to model fuel outgassing rate and ullage oxygen concentration
using Equation 8-1. The exponential time constant () represents the time taken for (1-1/e), or

200

63.2% of the total mass of oxygen that will evolve from the fuel to be transferred into the tank
ullage.

O2,i O2,i 1
20.9 O2,i 1

1 e

Equation 8-1

A key issue for the flammability analyst, when quantifying the effects of fuel outgassing on FRM
performance, is that the time constants of oxygen evolution used in the FTFAM are derived from
a single aircraft type and set of fuel tank conditions. Consequently, using the FTFAMs universal
time constants to model fuel outgassing effects on FRM performance in other aircraft fuel tanks,
under different environmental conditions, may lead to flammability estimation errors. In the worst
case, the period of time the aircraft fleet is exposed to a flammable ullage condition is
underestimated within the analysis.

8.2 Fuel Tank Flammability Assessment Method


The FTFAM, developed by the FAA, is a computer model designed as a comparative analysis
tool to determine aircraft fuel tank flammability. The FTFAM calculates the fleet flammability
exposure time within the ullage of a particular aircraft type using Monte Carlo statistical methods.
The model simulates a large number of flights, comparing the bulk average fuel temperature at
each time increment of flight with lower and upper flammability limits (LFL & UFL). When the
bulk fuel temperature falls within these flammability limits the fuel vapour is considered
flammable. Figure 8-1 illustrates a flow chart depicting the main computation of the FTFAM,
including FRM performance computations.

201

Figure 8-1 Flow chart illustrating the main computations used in the Fuel Tank
Flammability Assessment Method (FTFAM) and FRM performance (Summer, 2008, p.7)

202

8.2.1 FRM Performance Degradation due to Fuel Outgassing


Within the FTFAM the FRM performance module refers to the times throughout a flight that the
FRM is operating as expected, but is unable to protect the fuel tank due to limitations in its
performance under certain operating conditions. The effect of oxygen evolution, due to fuel
outgassing degrading FRM performance, is approached by assuming the fuel is saturated with air
at the beginning of the flight. It is then assumed that oxygen evolution is driven by differences in
the oxygen partial pressure in the ullage and dissolved in the fuel. The process has an exponential
mass transfer time constant to achieve equilibrium. The effect of supersaturation where the
concentration of oxygen in the fuel has not reached equilibrium with the ullage can, if necessary,
be simulated by keeping gas dissolved in the fuel (>100% saturation, i.e. over saturation) until a
specific partial pressure differential is reached. The FRM performance module features two time
constants of oxygen evolution. Under quiescent fuel tank conditions, without altitude changes, the
time constant used for oxygen transfer from fuel to ullage is 3500 minutes. Under aircraft climb
conditions the time constant of oxygen evolution used is 100 minutes, where it is assumed no
oxygen transfer takes place until the aircraft is above 15000 ft.

8.2.2 FTFAM Oxygen Evolution Time Constants


The oxygen evolution time constants of 100 and 3500 minutes for aviation fuel used in the
FTFAM were derived by the FAA and Boeing from studies of oxygen concentration data
measured in two different types of aircraft fuel tank. Their approach was to use a spreadsheet
which computed ullage oxygen concentration levels in a fuel tank using Equation 8-1, knowing
the starting fuel quantity and assuming the fuel was saturated with air from the ullage originally
comprising 21% oxygen by volume. The spreadsheet incorporated aircraft flight profiles and
computed the ullage oxygen concentration for a variety of cases, with and without the inerting
203

system running. Time constants of oxygen and nitrogen gas evolution were adjusted in the
spreadsheet to try and match the aircraft flight test data. Both parties felt that the derived
constants were reasonably representative of a fuel tank condition during an aircraft climb with
fuel boost pumps operating. One major concern at the time was a major fuel outgassing event
during climb if an air-supersaturated fuel developed.

8.3 Flight Test Aircraft A340-300 MSN001


An Airbus A340-300 flight test aircraft, MSN001, was made available in which ullage oxygen
concentration in the centre wing fuel tank (CWT) could be measured. The A340-300 is a long
range, four-engined wide body commercial passenger transport aircraft. The aircraft type entered
service in 1993 and has the capability of carrying 295 passengers over a range of 7300 nautical
miles. Figure 8-2 shows a photograph of the A340-300 MSN001 flight test aircraft. The aircraft is
stationed at the Airbus Flight and Integration Test Centre in Toulouse, France.

Figure 8-2 Airbus A340-300 MSN001 flight test aircraft (Mervelet, 2009)

204

The aircrafts CWT has a capacity of 33,300 kg of fuel (42420 litres). It is contained within the
centre wing section of the aircraft, located within the body. The aircraft is fitted with a CWT
inerting system and ullage oxygen measuring system specifically developed to characterise FTI
system performance necessary for design certification. The aircrafts CWT configuration
presented the ideal opportunity in which to measure dissolved oxygen evolution in flight,
provided the FTI system was temporarily disabled, such that oxygen concentration increase could
be observed and recorded.

8.3.1 A340-300 CWT Fuel System


Figure 8-3 illustrates the CWT fuel system architecture of the A340-300 aircraft. The main duty
of the fuel system located in the CWT is to provide fuel to both the left and right inner feed tanks
and the auxiliary power unit (APU). Fuel transfer, using two dedicated centrifugal pumps from
the CWT to the inner feed tanks, is achieved by opening the left hand and right hand inner feed
tank inlet valves F and H. Figure 8-4 shows a schematic of the CWT location within the A340300 aircraft.

Figure 8-3 Schematic of A340-300 CWT fuel system architecture (Airbus Technical
Description Volume 3A System AI/ED-N 433.0023/91 Iss. 6)
205

Figure 8-4 A340-300 architecture (Airbus Technical Description Volume 3A System AI/ED-N 433.0023/91 Iss. 6)

206

8.4 Ullage Oxygen Concentration Measurement


Ullage oxygen concentration measurements in the CWT of the A340-300 MSN001 aircraft were
made using an Oxigraf O2N2 flight test oxygen analyser and O2G8 ullage gas sampling system.
The flight test oxygen analyser utilises the same Tunable Diode Laser Absorption Spectroscopy
(TDLAS) oxygen measurement technology as the Oxigraf O2G1 laboratory oxygen analyser
described in Chapter 5. Figure 8-5 shows a schematic of the Oxigraf oxygen sensing system and
Figure 8-6 a photograph of the oxygen analyser and gas sampling system installed within the
cabin of MSN001.
A sampling pump system continuously draws ullage gas samples from four discrete locations in
the CWT to the Oxigraf sensor located in a remote enclosure within the aircraft cabin prior to
returning the gas samples back to the tank. Figure 8-7 to Figure 8-9 illustrates the oxygen sensing
positions and sense line installations within the CWT of the flight test aircraft.

Figure 8-5 Oxigraf O2N2 flight test oxygen analyser and O2G8 gas sampling system
architecture (schematic courtesy of Oxigraf Inc.) (McCaul, 2007, p.29)

207

Figure 8-6 Photograph of the Oxigraf O2N2 flight test ullage oxygen analyser and O2G8
oxygen sampling system installation within the Airbus A340-300 MSN001 flight test aircraft
cabin
The Oxigraf system was calibrated to measure oxygen concentration in the ullage of the aircrafts
CWT over a 2 to 21% oxygen by volume range. Whilst this satisfied the requirements for
assessing FTIS performance, this oxygen measurement range was not suitable for gas evolution
tests within a un-inerted CWT. The ullage oxygen concentration, prior to any fuel outgassing,
would normally be at the limit of the instruments measuring range i.e. ~21%. Two options
existed to overcome this problem. First was the option to re-scale the instruments oxygen
measurement range to a value above 21%. Alternatively the CWT could be pre-inerted prior to
flight using the FTIS to a value below 21%. It was decided that pre-inerting the CWT to ~12%
oxygen by volume, prior to disabling the FTIS for the flight, would be easier to implement than

208

re-scaling the Oxigraf system and would provide a closer emulation of the FAA/Boeing flight test
activities.

209

ULLAGE GAS SAMPLES DRAWN TO OXIGRAF O2N2 ANALYSER


IN AIRCRAFT CABIN

TEMPERATURE SENSING POINTS


OXYGEN SENSING PIPES

TEMPERATURE DATA
TO AIRCRAFT CABIN

OXYGEN SENSING POINTS

Figure 8-7 Cross-section of the A340-300 MSN001 centre wing fuel tank (Left-hand side) showing ullage oxygen concentration and
temperature sensing locations (all dimensions in mm) (Boulet, 2009, p.4)
210

Figure 8-8 Illustration of ullage oxygen gas sampling points - float valve arrangements in
the centre wing fuel tank of the A340-300 MSN001 flight test aircraft (Boulet, 2009, p.5)

Figure 8-9 Illustration of ullage oxygen gas sample return points - float valve arrangement
at R10 within the centre wing fuel tank of the A340-300 MSN001 flight test aircraft(Boulet,
2009, p.6)
211

8.4.1 Flight Test F1437


A flight test request, FTA No. 201/47 F 00 D 12, was raised and submitted to the Airbus EV
Flight and Integration Test Centre in Toulouse, France. The request called for the aircraft to be
refuelled with Jet A-1 aviation turbine fuel such that the CWT was at least 50% full at take-off.
This quantity of fuel in the CWT would ensure that the effects of fuel outgassing and oxygen
concentration increase in the CWT ullage would be detectable with the Oxigraf ullage oxygen
measuring system. Unfortunately, because the aircraft was required for other systems related test
activities (acoustic noise measurements) during the planned flight, it was not possible to refuel
the aircraft with more than 10,000 kg (12726 litres) in the CWT. Prior to refuelling, the CWT
ullage was inerting with nitrogen enriched air (NEA) using the on-board inert gas generation
system (OBIGGS), such that the bulk oxygen concentration level within the CWT ullage was
12% oxygen by volume. The FTI system was then disabled and with the Oxigraf system running
the aircraft proceeded to climb from just above sea-level to FL 350 (35,000 ft) at a requested
average climb rate of 2500 ft/min. The aircraft achieved this climb to altitude in approximately 15
minutes. From take off to 47 minutes after FL350 was established the fuel quantity in the CWT
remained approximately constant at 10,000 kg. Fuel from the CWT was then transferred to both
inner wing tanks via transfer valves F and H. The fuel transfer was initiated to agitate the fuel in
the CWT through transfer pump impeller motion to promote fuel outgassing.

8.5 Results from Flight Test F1437


To determine the effect of fuel outgassing on oxygen concentration in an aircraft fuel tank ullage,
a flight test was performed (flight no. F1437) on an A340-300 aircraft (MSN001) in which
oxygen concentration was measured at various locations in the aircrafts CWT ullage. The CWT
ullage was pre-inerted prior to flight using the aircrafts FTI system to an oxygen concentration of

212

~ 12% by volume, such that oxygen concentration increases due to fuel outgassing could be
detected over the limited oxygen sensing systems measurement range (2-21% oxygen by vol.).
A plot featuring the measured ullage oxygen concentrations in the CWT during the flight is
shown in

Figure 8-10. The plot shows that the aircraft climbed to the designated cruising altitude of 35000
ft in ~15 minutes achieving a climb rate of ~2333 ft/min, which was reasonably close to the 2500
ft/min target. The plot shows that just over an hour after take-off the 10,000 kg fuel quantity
contained in the CWT was transferred to the inner wing tanks at an average rate of approximately
333 kg/min. It can be seen from the ullage oxygen concentration measurements that only a small
quantity of oxygen rich air was released from the fuel into the ullage during the climb and cruise
phases of the flight up to the initiation of fuel transfer. Oxygen concentration in the CWT ullage
had only increased by approximately 0.5% from 12.2% to 12.7%.
Just after fuel transfer had begun significant increases in oxygen concentration at the measured
points within the CWT ullage were observed. Oxygen measurement traces at locations 28, 29 and
30 show more erratic behaviour than those at locations 31 and 32. Beyond 8000 seconds the
aircraft performed a series of flight manoeuvres associated with acoustic noise measurement tests
that were not related to the fuel outgassing investigation. At just prior to 10000 seconds before
aircraft descent the CWT inerting system was turned on, resulting in the eventual drop of the
ullage oxygen concentration to 12% on landing. Oxygen concentration at location 32 read
slightly higher than the other traces at other locations in the CWT up until the point at which fuel
transfer was initiated. Figure 8-11 provides a more detailed view of ullage oxygen concentration
213

behaviour from just prior to fuel transfer and up to the end of the flight. Fuel temperature data has
been removed in Figure 8-11for clarity. Temporal breaks in the oxygen concentration traces are
associated with in-flight calibration of the Oxigraf measuring system, where a small quantity of
oxygen-in-nitrogen gas of known concentration was introduced directly into the analyser from
compressed gas bottles located in the aircraft cabin.

214

Figure 8-10 Flight test data from flight No. F1437 showing ullage oxygen concentration measurements (left hand side) and corresponding
aircraft altitude, CWT fuel quantity and fuel temperature (right hand side)
215

Figure 8-11 Effect of fuel transfer from CWT to inner feed tanks on CWT ullage oxygen concentration during flight no. F1437. Fuel
temperature and altitude data removed for clarity
216

8.6 Discussion
A flight test was successfully performed on an A340-300 aircraft. Oxygen concentration at
various locations within the ullage of the CWT were measured and recorded. Results from the
flight test have provided an opportunity to examine the effects of fuel outgassing on ullage
oxygen concentration in an aircraft fuel tank.

Figure 8-10 shows that very little oxygen concentration increase occurred in the CWT ullage
between take-off and the start of fuel transfer to the left and right inner wing tanks. This suggests
that the fuel within the CWT was relatively quiescent during the climb and initial cruise phase of
the flight. Under these quiescent conditions it is highly likely that the fuel would have become
supersaturated with air. Another significant factor, supporting a small oxygen concentration
increase up to the point of fuel transfer, was the relatively large ullage volume present within the
CWT. The total fuel load in the CWT was 20% less than the requested minimum. This larger
than expected ullage volume would have tended to dilute the effect that oxygen-rich air, liberated
from the fuel, would have on CWT ullage oxygen concentration. At the point of initiating fuel
transfer and beyond, a significant increase in oxygen concentration was observed at all the
available oxygen measurement locations. Figure 8-11 shows that oxygen concentration traces at
locations 28, 29 and 30 behaved more erratically than those at positions 31 and 32. Oxygen
measurements at locations 28, 29 and 30 were made in very close proximity to the CWT vent
pipe. This erratic behaviour is possibly attributable to vent air, containing 21% oxygen by volume
entering and mixing with the inerted CWT ullage. As the fuel level dropped in the CWT, due to
its transfer into the inner wing tanks, the vent system would freely admit air from the atmosphere
into the CWT in order to maintain a low pressure differential. Of perhaps more interest in this
217

investigation are the oxygen concentration traces at positions 31 and 32. These measurement
locations were well away from the CWT vent and would have been in much closer proximity to
the fuel surface at the beginning of the fuel transfer operation. These traces show a much more
stable, increasing oxygen concentration trend. Although these traces would include the effects of
air, with a higher relative oxygen content entering the CWT, they are perhaps more representative
of the effects of oxygen outgassing from the fuel. As the transfer pumps were turned on and their
impellers began to rotate, dissolved air would evolve in the form of tiny gas bubbles from the fuel
around the pump impellers inlet annulus. However, even though the pumps were running, the
fuel in the CWT does not appear to have been highly agitated other than in the area local to the
pump impeller inlets.

8.6.1 Effect of Air Evolution on Inerted CWT Ullage


To establish the effect of oxygen rich air evolution from the fuel on the inerted CWT ullage, the
% oxygen increase in the CWT ullage attributable to atmospheric air, inspired via the vent system
as a result of fuel transfer, must be determined. Using gas laws and assuming perfect mixing
within the CWT ullage, the theoretical CWT ullage oxygen concentration increase, due to
inspired air, was calculated and contrasted with ullage data from the flight. The theoretical
oxygen concentration curve, shown in Figure 8-12 suggests the small amount of air evolved from
the fuel had only a minor effect on the CWT ullage if the oxygen concentration measurements
local to the vent are ignored. Key factors associated with the flight test most likely to have
contributed to this result will now be discussed.
Inerted CWT Ullage
Whilst on the ground the aircrafts CWT was inerted with NEA to an oxygen concentration of
~12% by vol. prior to refuelling with the OBIGGS. During CWT refuelling the fuel would come
into contact with the nitrogen enriched ullage and a quantity of ullage gas would have been
218

absorbed by the fuel. The fuel would have had a more nitrogen rich dissolved air composition
than if it had been exposed to atmospheric air containing ~21% oxygen.

219

Figure 8-12 Comparison of theoretical CWT ullage oxygen concentration increase resulting from air inspired via the vent system due to
fuel transfer with measured ullage oxygen concentration. Fuel temperature and altitude data removed for clarity
220

When the dissolved air began to evolve the effect of oxygen would have been far less apparent
as observed in Figure 8-12. In fact, this situation emulates the early fuel tank inerting strategy
examined by Wyeth, Oldfield and Wiltshire (1957) for reducing ullage oxygen concentration in
flight. A simulation of this situation was conducted in the small-scale laboratory fuel tank to
demonstrate how a nitrogen enriched ullage lowers the dissolved oxygen content of the fuel.
After introducing a quantity of fuel in the laboratory tank the ullage was then inerted with a 5%
by volume oxygen-in-nitrogen gas mixture. During inerting the fuel was mildly agitated with a
mixing impeller at 200 rpm to simulate fuel agitation in the CWT due to refuelling. When the
ullage oxygen concentration was stable, a 25% oxygen-in-nitrogen gas mixture was fed into the
tank ullage. Figure 8-13 demonstrates how inerting the fuel tank with nitrogen enriched air lowers
the dissolved oxygen content in the fuel. The partial pressure gradient which develops across the
ullage-fuel interface as the tank is inerted drives gas diffusion where oxygen is driven out of the
fuel and replaced with nitrogen. Eventually, the partial pressures of oxygen and nitrogen in the
fuel and ullage approach equilibrium.
CWT Fuel Load
The CWT fuel load was 20% less than the minimum requested, therefore the larger than expected
ullage would have tended to dilute the effects of oxygen evolution on ullage oxygen
concentration. Although some mixing would have been taking place between evolved oxygen
from the fuel and the ullage, it is possible that in such a large ullage stratification of the gases
might occur.
Fuel Agitation
The level of fuel agitation in the CWT imparted by vibration and fuel transfer is clearly very low
where the fuel would have become highly supersaturated with air. Dissolved oxygen would have
been released at a much slower rate under these near quiescent conditions. In comparison, the
221

environment in a fuel collector cell would be significantly different and the level of fuel agitation
very much higher.

Figure 8-13 The effect of a nitrogen enriched ullage on the dissolved oxygen content of
mildly agitated Jet A-1 fuel at ambient temperature and barometric pressure
8.6.2 Validation of Fuel Outgassing Model Law
The flight test provided oxygen concentration measurements within the ullage of an aircraft fuel
tank during flight. Unfortunately, this data does not present an opportunity for validating
dimensional modelling predictions of oxygen evolution rate, made using the model fuel tank in
Chapter 7. Dimensional similarity between the laboratory model and A340 aircraft CWT is not
achievable. The quantity of fuel required in the model fuel tank would be far greater than could
be accommodated owing to the larger quantity of fuel carried in the A340 CWT. As discussed in
Chapter 5, calculating the rate of oxygen evolution from ullage oxygen concentration data is
complicated due to the behaviour of ullage gases at reduced pressures and temperatures.
222

Validation of fuel outgassing rates, projected from model fuel tanks, with aircraft data, requires
dissolved oxygen concentration measurements to be made in the aircraft fuel tank. At present, a
dissolved oxygen sensor that can withstand an aircraft fuel tank environment in flight is not
available or in commercial use.

223

8.7 Conclusions
Oxygen concentration in the CWT ullage of an A340-300 flight test aircraft has been successfully
measured using a TDLAS oxygen sensing system up to a flight altitude of 35,000 ft.

With a 30% Jet A-1 fuel load only a marginal increase in oxygen concentration was recorded in
the CWT ullage from take-off to a cruise altitude of 35,000 ft. The fuel in the CWT was likely to
have been relatively quiescent during the flight and highly supersaturated with air.

Transferring fuel during aircraft cruise from the CWT to the inner wing tanks resulted in an
almost instantaneous rise in ullage oxygen concentration at all measurement locations within the
CWT ullage. A theoretical calculation demonstrated that increasing oxygen concentration in the
CWT ullage immediately following the initiation of fuel transfer, was principally due to inspired
air from the atmosphere, via the vent system rather than from dissolved air evolved from the fuel.

A low fuel load (large ullage), absorption of nitrogen into the fuel from the inerted CWT ullage
during aircraft refuel and relatively quiescent CWT fuel conditions during the flight, is thought to
have been responsible for oxygen evolution having little effect on the oxygen concentration in the
CWT ullage.

Ullage oxygen concentration data from the flight test cannot be used to validate projected rates of
fuel outgassing from dimensional modelling studies, as dimensional similarity, between the A340
CWT and the existing model fuel tank cannot be achieved. Further, ullage oxygen concentration
measurements do not allow accurate rates of fuel outgassing to be calculated. A dissolved oxygen
224

sensor that can withstand the harsh environment of an aircraft fuel tank during flight must be
developed if accurate rates of oxygen evolution are to be determined in flight.

225

Chapter 9
Conclusions and Future Work
9.1 Conclusions
Detailed conclusions are given at the end of each associated results chapter: this chapter presents
the overall conclusions resulting from this research work, together with recommendations for
future work that will further the understanding of fuel outgassing phenomenon in aircraft fuel
tanks.
The aim of this work was to quantify the rate of oxygen evolution from aviation turbine fuel over
a range of fuel tank environments encountered within commercial transport aircraft. Fuel
outgassing is detrimental to the performance and safe operation of an aircraft fuel system and
increases fuel tank flammability. Rates of oxygen evolution from aviation turbine fuel were
measured in a small-scale, environmentally conditioned fuel tank. Dimensional modelling was
used to project results from the small-scale fuel tank to an aircraft fuel tank to establish if such an
approach could be used to predict oxygen release rates in aircraft fuel tanks. A flight test was also
performed on a long-range commercial aircraft to establish the effect of fuel outgassing on ullage
oxygen concentration.

9.1.1 Measurement of Oxygen Evolution Rates


The feasibility of determining oxygen evolution rates from aviation turbine fuel in small-scale
fuel tanks using dissolved and gaseous phase oxygen sensors was demonstrated. Precise control
over fuel agitation rate, ullage pressure and fuel temperature was achieved with the experimental
set-up to simulate aircraft fuel tank environments in the laboratory. The experimental set-up

226

provided an extremely useful tool for understanding how fuel outgassing rate is affected by the
environmental conditions present within an aircraft fuel tank during flight.
Variations in ullage gas temperature and density due to ullage depressurisation made it difficult to
accurately calculate oxygen mass release rates from oxygen concentration measurements made in
the tanks ullage. An electrochemical dissolved oxygen sensor, submerged in the fuel, where
oxygen gas density and temperature was relatively stable made calculation of oxygen release
rates more straightforward.
Limitations were encountered with measuring oxygen release rates at sub-zero fuel temperatures
due to electrolyte freezing within the electrochemical oxygen sensor. To avoid this problem
oxygen release rate measurements were restricted to fuel temperatures above 0C.
Although dissolved oxygen concentrations measured in the agitated fuel were thought to be
uniform throughout the tank this could not be substantiated with measurements from a single
dissolved oxygen sensor. Further work needs to be done to establish if dissolved oxygen
concentration gradients exist in agitated fuel tanks.

9.1.2 Effect of Aircraft Fuel Tank Environment on Oxygen Evolution Rate


In an environment representative of an aircraft fuel tank, fuel agitation rate was shown
statistically to have the most significant effect on the rate of oxygen evolution from aviation
turbine fuel. The effect of fuel agitation rate was markedly greater than the effects of fuel
temperature, rate of change of ullage pressure and associated interactions of these variables on
fuel outgassing rate. Oxygen evolution rate increased non-linearly with fuel agitation rate where
the data was best fitted by a rectangular hyperbola. Oxygen release rate was similarly related to
the power dissipated to the fuel by the mixing impeller. The results of this research support the
idea that fuel agitation plays the most significant role in influencing the mass transport properties
227

between dissolved gas and evolving bubbles in the fuel. The tendency for gas bubbles to break-up
at higher impeller speeds providing a larger overall surface area for mass transport to take place is
also believed to be responsible for increasing oxygen evolution rate.

9.1.3 Time Constant of Oxygen Evolution


One of the most significant findings to emerge from this study is that the time constant of oxygen
evolution, is highly sensitive to the specific environmental conditions present within an aircraft
fuel tank. The results of this investigation have shown that at higher fuel agitation rates,
temperatures and aircraft climb rates the time constant decreases significantly. The relevance of

on fuel tank inerting system performance is clearly supported by these current findings. These
results suggest that the use of universal time constants to mathematically model the effect of
oxygen evolution on fuel tank inerting system performance in different aircraft should be
avoided. Time constants of oxygen evolution, specific to an aircraft fuel tank for a particular
aircraft type, should be used where possible to avoid flammability analysis errors.
The current findings add substantially to our understanding of oxygen evolution behaviour from
aviation fuel and the factors that affect it. The results of this study make clear the need for
developing, cost effective and robust experimental methods that will enable the time constant of
oxygen evolution to be accurately determined for specific aircraft fuel tanks and their
environments. The current study was unable to investigate oxygen evolution time constant over
the complete environmental envelope due to limitations in the experimental set-up and dissolved
oxygen measuring apparatus.

228

9.1.4 Dimensional Modelling of the Fuel Outgassing Phenomenon


The evidence from this study suggests that experimental scale has a significant bearing on the
magnitude of oxygen evolution rate from aviation fuel. The rate of oxygen evolution from
aviation fuel in an A320 aircraft fuel tank was projected to be over 200% greater than oxygen
evolution rate measurements made on a dimensionally similar laboratory model. An incorrect
mixing impeller speed, used in the dimensional modelling tests resulted in lower than required
oxygen evolution rates. This suggests that the projected values of oxygen evolution rate from the
physical model for the A320 aircraft are likely to be even greater. Results of the dimensional
modelling study were based on the assumption that fuel outgassing rate measurements can be
extrapolated to full scale aircraft fuel tanks without the need for a complex, geometrically similar
model. Further work to support this assumption is required by conducting dimensional modelling
studies in fuel tanks of differing size and geometry. For dimensional modelling to be accepted as
a viable cost effective alternative to flight testing, validation of extrapolated fuel outgassing rates
with flight test data, from aircraft fuel tanks, is required. The present challenge of developing a
robust dissolved oxygen sensor, from which, accurate rates of fuel outgassing in aircraft fuel
tanks can be calculated must therefore be overcome.
9.1.5 Effect of Fuel Outgassing on an Inerted Aircraft Fuel Tank Ullage
The findings in this study have shown that the evolution of oxygen from aviation fuel in the
inerted CWT of an A340-300 aircraft has only a minimal effect on ullage oxygen concentration
during flight. The following factors provide a possible explanation as to why fuel outgassing had
only a minor effect on ullage oxygen concentration in the flight test;

Pre-inerting the aircrafts CWT with NEA to a concentration of 12% oxygen by volume
altered the composition of dissolved air within the fuel during refuel, i.e. dissolved air
229

within the fuel contained less oxygen than normal. Air released from the fuel during
flight contained less oxygen. Experimentation in the laboratory using the small-scale fuel
tank supported this finding and demonstrated how the composition of dissolved air in the
fuel can be altered by inerted the ullage with nitrogen.

The ullage volume was unexpectedly large due to a lower than requested CWT fuel load,
which diluted the effects of evolving oxygen-rich air from the fuel within the ullage
during flight.

The CWT of the A340-300 aircraft was most likely to have been quiescent during the
flight test where a lack of mechanical stimulus would have lead to the fuel being highly
supersaturated with air.

230

9.2 Future Work


9.2.1 Experimental Set-up and Further Testing
The sensitivity of oxygen evolution rate to fuel agitation has been demonstrated by stirring fuel in
a small-scale tank with a radial flow mixing impeller at different rotational speeds. This method
was chosen to emulate the basic motion of fuel within an aircraft collector cell and provide a
means by which the level of fuel agitation, in terms of a rate, could be repeatedly controlled and
varied between tests. The mixing impeller approach enabled the effect of fuel agitation on oxygen
release rate to be clearly quantified. To validate this method of fuel agitation and prove it
emulates the aircraft fuel tank environment, a series of tests should be conducted in small-scale
fuel tanks containing aircraft fuel pumps, jet pumps and other fuel system equipment known to
agitate fuel. Oxygen evolution rates from fuel tanks containing mixing impellers and fuel system
equipment should be compared and contrasted. If it is possible to correlate the level of agitation
imparted to the fuel by a particular fuel system configuration with mixing impellers running at
specific speeds then the cost and complexity of small-scale set-ups for future fuel outgassing rate
studies could be vastly reduced.
Different designs of mixing impeller should be investigated to understand how accurately they
mimic the turbulent flow behaviour and thus agitation of fuel in mock-up aircraft fuel tanks to
determine the effect on oxygen evolution rate.
Whole field flow visualisation techniques such as Laser Doppler Anemometry (LDA) and
Particle Image Velocimetry (PIV) could be used to further characterise 2D and 3D fuel flow
behaviour and turbulence within optically transparent fuel tanks containing aircraft fuel system
equipment. Flow patterns and velocities for different fuel system set-ups and tank architectures

231

could be measured with LDA and PIV and then possibly recreated with a particular mixing
impeller configuration.
Limitations in the experimental set-up have constrained the present study to a narrower range of
fuel temperatures and agitation rates than would typically exist within an aircraft fuel tank
environment during flight. As such, certain tests following on from those conducted in this study
need to be performed with a modified experimental set-up. The set-up must permit dissolved
oxygen concentration measurements in the fuel at temperatures down to -50C and a mixing
impeller system that can generate fuel agitation rates equal in magnitude to those in the aircraft.
An array of dissolved and gaseous phase oxygen sensors should be positioned within the test tank
such that the uniformity of dissolved oxygen concentration within the fuel and ullage can be
assessed.
Fuel test tanks featuring different geometry containing varying quantities of fuel are to be
examined in future tests. In the present study the rate of gas evolution was measured from an
experimental set-up that emulated a constant fuel quantity being held in a collector cell type tank.
The effect of fuel usage during flight to represent the conditions in aircraft wing tanks as they are
gradually depleted should also be examined to explore and understand if this has any effect on
fuel outgassing rate and ullage oxygen concentration. The rate of tank depletion should be altered
to examine its effect on outgassing rate. A fuel quantity indication system is required to be
introduced to the test tanks to enable the rate of depletion to be monitored throughout these tests.

9.2.2 Validation of Fuel Outgassing Model Law


Further use of dimensional modelling for estimating fuel outgassing rates from aviation fuel
requires the Model Law, developed in this study, to be validated with oxygen evolution rates
from aircraft flight tests. As an interim step a series of tests using fuel tanks containing larger
232

quantities of fuel and fuel agitation rates approaching those present in the A320 aircraft collector
cell are to be conducted to establish the validity of the Model Law developed in this study. If the
Model Law holds then oxygen evolution rates determined from these studies should correlate
well with those projected from the dimensional modelling study in this thesis. Dimensional
modelling studies in fuel tanks of differing shape and geometry are to be conducted to justify the
rationale and assumptions made in selection of the relevant variables used in the present study.
Provided the dimensional modelling approach can be validated a database of fuel outgassing rate
information for different aircraft operating under a variety of flight environments is to be
developed from dimensional modelling studies to support flammability analysis of aircraft fuel
tanks in the future.

9.2.3 Dissolved Oxygen Sensor Development


The electrochemical dissolved oxygen sensor used in this study provided satisfactory operation
over a 0C to 40C fuel temperature range. Electrolyte freezing however at sub-zero fuel
temperatures constrained the investigation, preventing determination of oxygen evolution rates at
very cold fuel temperatures. Development of a robust oxygen sensor enabling dissolved phase
oxygen measurements to be made in aircraft fuel tanks and small-scale laboratory test tanks with
fuel temperatures down to -50C is required. Oxygen sensors based on fluorescence quenching
have shown the most promise for the low temperature application but require improvements to be
made that provide long term stability of the oxygen quenched sensing media in the harsh fuel tank
environment. Qualification of this oxygen sensor technology for use in an aircraft fuel tank also
requires a significant number of sensor design features, not least intrinsic safety, to be fully
demonstrated via laboratory tests and flight trials.

233

References
American Society for Testing Materials, 2002. Standard test method for estimation of solubility
of gases in petroleum products, D2779-92. Pennsylvania: American Society for Test Materials.
Astafev, V.A., Kozinova, L.N., 1998, Determinations of oxygen and nitrogen solubilities in fuels
at elevated temperatures, Chemistry and Technology of Fuels and Oils, 24 (2), pp. 84-87.
Bakker, A., 2006. Lecture 1 Introduction to CFD, applied computational fluid dynamics.
Available at http://www.bakker.org/
Barnett, H.C. and Hibbard, A.A., 1956. Properties of aircraft fuels. Washington: National
Advisory Committee for Aeronautics.
Beal, J.L., Hilburger, E.A., Porter, P.K., 1945. Evolution of air from aviation fuel during climb.
New York: Society of Automotive Engineers.
Bedwell, M.E., 1952. Rate of release of dissolved air, oxygen and nitrogen from kerosene during
flight .Farnborough: Royal Aircraft Establishment.
Belieres, J. 2011. Email to A. Harris 19 July 2011.
Blanchard, N., 2008. ZZ156 Evaluation test report. Toulouse: Airbus.
Boeing and US Air Force, 2010. Evaluation of bio-derived synthetic paraffinic kerosenes (BioSPKs), version 5. Boeing and US Air Force.
Bonjour, J.C., 2009. A320/MSN 659/IAE V2500-A5 & Select One Heat Management Systems
Test/Fuel Lacquering Flight Test Report. Toulouse: Airbus.
234

Borisov, V.D., Logvinyuk, V.P., Makarenkov, V.V., Malyshev, V.V., 1969, Formations of gas
bubbles in supersaturated fuel, Journal of Engineering Physics and Thermophysics, 17, pp. 920923.
Boulet, E., 2009. A340 MSN001 Inerting caisson central, Report No. H00 Z00440 IND A PL 1
Toulouse: Airbus.

Bragg, K.R., Kimmel, C.C., Jones, P.H., 1969. Nitrogen inerting of aircraft fuel tanks. New
York: Society of Automotive Engineers.
Brennen, C.E. 2009. Email to A. Harris February 2009.

Buckingham, E., 1914, On physically similar systems, Physical Reviews, 4 (4), p.385.

Burrows, G., Preece, F.H., 1954, The process of gas evolution from low vapour pressure liquids
upon reduction of pressure, Transactions of the Institution of Chemical Engineers,32, pp. 99-106.

Cannoletta, D et al., 2007. Fuel Tank Explosion Protection System Based on OBIGGS, Fifth
Triennial International Aircraft Fire and Cabin Safety Research Conference, 29th October 1st
November 2007, New Jersey.

Cansdale, J.T., 1978. The effect of using jet transfer pumps on the dissolved oxygen content of
fuel in aircraft. Farnborough: Royal Aircraft Establishment.
235

Cavage, W., 2006. State of the Art of Fuel Tank Ullage Oxygen Concentration Measurements,
International Systems Fire Protection Working Group, 16-17th April, London.

Cavage, W., 2005. The Effect of Fuel on an Inert Ullage in a Commercial Transport Airplane
Fuel Tank, Report No. DOT/FAA/AR-05/25.New Jersey: Federal Aviation Administration.

Cavage, W., 2009. Measuring Oxygen Concentration in a Fuel Tank Ullage, U.S. Air Force T&E
Days, 10 -12th February 2009, New Mexico.

Co-ordinating Research Council, 1958. Solution and evolution of gases and vapors in aircraft
fuels, report of the advisory group on air-gas solubility work conducted by the Atlantic Research
Corporation.

Co-ordinating Research Council, 2004. Handbook of Aviation Fuel Properties, Third Edition.
Pennsylvania: Co-ordinating Research Council.

Dean, R.B., 1943, The formation of bubbles, Journal of Applied Physical, 15, pp. 447-451.

Derry, L.D., Evans, E.B., Faulkner, B.A., Jelfs, E.C.G., 1952, Vapour and air release from
aviation fuels, Journal of the Institute of Petroleum, 38 (343), pp.475-525.

236

Eaton Corporation, 2007, Air Release Valve DS600-26.

Federal Aviation Administration, 2005. Advisory circular, fuel tank flammability AC No. 25.9812A. US Department of Transport.

Fernandez-Prini, R., 1982, Le Chatelier's principle and the prediction of the effect of temperature
on solubilities, Journal of Chemical Education, 59 (7), pp. 550-553.

Frish, M.B., et al., 2004. Progress in Reducing Size and Cost of Trace Gas Analyzers Based on
Tunable Diode Laser Absorption Spectroscopy, Optics East, 26th October 2004, Philadelphia.
Gatward, B.F.A, Wyeth, H.W.G., 1951. The effect of air evolution from fuel on the inert gas
protection of aircraft fuel tanks, Report M.E. 96. Farnborough: Royal Aircraft Establishment.

Gille, U., Salomon, F.V., 1995, Bone Growth in Ducks through Mathematical Models with
Special Reference to the Janoschek Growth Curve, Growth, Development and Aging,59, pp. 207214.

Glendinning, W.G., Bedwell, M.E., 1949. Solubility of gases in liquids used in aircraft Part 1.
Measurement of gas solubility, Report No. 464.Farnborough: Royal Aircraft Establishment.

Gord, J.R., Buckner, S.W., Weaver, W.L., 1995. Dissolved oxygen quantitation in fuel through
measurements of dynamically quenched fluorescence lifetimes, 16th International Congress on
Instrumentation in Aerospace Simulation Facilities, 18-21st July 1996, Ohio.
237

Hach Lange, 2009. Orbisphere model 31xxx electrochemical sensors, installation and
maintenance manual, revision H, Geneva: Hach Lange.

Harris, A.P., 2011. Determination of dissolved oxygen in Jet A-1 subject to mechanical agitation
and reduced ullage pressure, Report No. V28RP1101048. Filton: Airbus.

Hayward, A.T.J., Dallas, J.T., 1965. The NEL universal apparatus for measuring the gas content
of liquids and semi-liquids. East Kilbride: National Engineering Laboratory.

Henshaw, J., 2008. Jet pumps and suction pipes Airbus A320 Phase 3 testing, Report No.
SCF/SYS/N/93/0057. Filton: Airbus.

Hetherington, J.I., Bellerby, J.M., 2004. Measurement of oxygen content of Jet A-1. Shrivenham:
Cranfield University.

Intertek, 2011. Characterisation of jet fuel ITS (20053), Report No. RT/ELE/8741.Bristol: ITS
Testing Services (UK).

James, W., ODell, P., 2005. Derated climb performance in large civil aircraft, Boeing
Performance and Flight Operations Engineering Conference, 2005.

Kinnear, P.R., Gray, C.D., 2010. PASW 17 Statistics made simple. Hove: Phsycology Press.
238

Kitanin, E.L., 2007. Air Evolution Research in Fuel Systems, Report No. D2.1-D2.4. St
Petersburg: St Petersburg State Technical University.

Kosvic, T.C., Zung, L.B., Gerstein, M., 1971. Analysis of Aircraft Fuel Tank Fire and Explosion
Hazards, Report No. AFAPL-TR-71-7. Ohio: US Air Force.

Langhaar, H.L., 1980. Dimensional Analysis and Theory of Models.2nd ed. Malabar: Krieger
Publishing.

Langton, R., Clark, C., Hewitt, M., Richards, L., 2009. Aircraft Fuel Systems. Chichester: John
Wiley & Sons.

Lee, I.E., 1974. De-aeration systems; performance tests, Report No. SST/B26-FUEL/2260. Filton:
British Aircraft Corporation.

Lewis, W.K., Whitman, W.G., 1924, Principals of gas absorption, Industrial and Engineering
Chemistry,16 (12), pp. 1215-1220.

Logvinyuk, V.P., Makarenkov, V.V., Malyshev, V.V., Panchenkov, G.M., 1970, Solubility of
gases in petroleum products, Chemistry and Technology of Fuels and Oils,6 (5), pp. 353-355.

239

Loxley, R.A., 1976. Development and calibration of a fuel primary/gas secondary jet pump,
Report No. FHR90113, Hucknall: Rolls Royce.

Loxley, R.A., Wright, G.H., 1976. Air release from wide-cut gasoline at normal and high
temperature by an ARP429A type feed arrangement, Report No. FHR90095, Hucknall: Rolls
Royce.

Loxley, R.A., Wright, G.H., 1978. MOD pumping research, further work on gas release in
engine fuel feed systems, Report No. FHR90150, Hucknall: Rolls Royce.

Loxley, R.A., Goddard, S., 1980. A review of the design and operation of LP fuel systems and
pumps mainly under two-phase flow conditions, Report No. FHR90173, Hucknall: Rolls Royce.

Lubetkin, S.D., 1995, The fundamentals of bubble evolution, Chemical Society Reviews, pp. 243250.

Markham, A.E., Kobe, K.A., 1941, The solubility of gases in liquids, Chemical Reviews, 28, pp.
519-588.

McCaul, B., 2007. Verification of Airliner Fuel Tank Inerting, Fifth Triennial International
Aircraft Fire and Cabin Safety Research Conference, 29th October 1st November 2007, New
Jersey.

240

McConnell, P.M., Dalan, G.A., Anderson, C.L., 1986. Volume III On-board inert gas generator
system (OBBIGS) studies, Part 2 Fuel scrubbing and oxygen evolution tests, Report No. AFWALTR-85-2060, Ohio: US Air Force.

Mervelet, J., 2009. Photograph of Airbus A340-300 MSN001 registration F-WWAI. Available at:
http://www.airliners.net/photo/Airbus-Industrie/Airbus-A340-311/1542765

Morris, R.J.W., Miller, J., Limaye, S.Y., 2006. Fuel deoxygenation and aircraft thermal
management, 4th International Energy Conversion Engineering Conference and Exhibit, 26-29th
June 2006, San Diego.

Paul, E.L., Atiemo-Obeng, V.A., Kresta, S.M., 2004. Handbook of Industrial Mixing. Hoboken:
John Wiley & Sons.

Perry, R.H., Green, D.W., Maloney, J.O., 1997. Chemical Engineers' Handbook, 7th ed.

Petrocelli, J.A., Lichtenfels, D.H., 1959, Determination of dissolved gases in petroleum fractions
by gas chromatography, Analytical Chemistry 31 (12), pp. 2017-2019.
Post, T., 2007. Pharmaceutical mixing Mixing basics Part 3, Hoboken: Post Mixing
Solutions.
Post, T. 2010. Email to A. Harris April 2010.

241

Post, T., 2011. Photographs and electronic data for RP4 impeller. Available at:
http://www.postmixing.com/mixing%20forum/impellers/impellers.htm#rp4

Poulston, B.V., Thomas, A., 1959, Inflammability and electrical studies of foams which may
occur at altitude by de-aeration of aviation turbine fuels, Journal of the Royal Aeronautical
Society 63, pp.581-588.

Ross, K., 1970. The solubility of air in aviation turbine fuels, Report No. K.186, Thornton: Shell
Oil Company.

Ross, K., 1972. Air release from supersaturated aircraft fuel, Report No. K.195, Thornton: Shell
Oil Company.

Roth, A.J., 1987. Development and evaluation of an airplane fuel tank ullage composition model,
Volume II experimental determination of airplane fuel tank ullage compositions, Report No.
AFWAL-TR-87-2060, Ohio: Boeing Military Airplane Company.

Rubey, W.A., Striebich, R.C., Tissandier, M.D., Tirey, D.A., 1995, Gas chromatographic
measurement of trace oxygen and other dissolved gases in thermally stressed jet fuel, Journal of
Chromatographic Science 33, pp. 433-437.

SAE International, 1994. Aircraft engine fuel pump cavitation endurance test, Report No.
ARP492 REV C, Pennsylvania: Society of Automotive Engineers.
242

Schweitzer, P.H., Szebehely, V.G., 1950, Gas evolution in liquids and cavitation, Journal of
Applied Physics 21, pp.1218-1224.

Senese, F., 2009. Why does the solubility of gases usually increase as temperature goes down?
Available

at:

http://antoine.frostburg.edu/chem/senese/101/solutions/faq/temperature-gas-

solubility [Accessed 22 June 2009].

Shafer, N.E., Zare, R.E., 1991, Through a beer glass darkly, Physics Today, pp. 48-52.

Summer, S.M., 2008. Fuel tank flammability assessment method user's manual, Report No.
DOT/FAA/AR-05/8, New Jersey: Federal Aviation Administration.

Swaffield, J.A., 1972, A study of the influence of air release on column separation in an aviation
kerosine pipeline, Proceedings of the Institute of Mechanical Engineers 186 (56/72), pp.693-703.

Szebehely, V.G., 1951, Relation between gas evolution and physical properties of liquids,
Journal of Applied Physics 22 (5), pp.627-628.

Szirtes, T., 2007. Applied Dimensional Analysis and Modeling. Oxford: Elsevier.

Uhlig, H.H., 1937, The solubilities of gases and surface tension, The Journal of Physical
Chemistry 41 (9), pp.1215-1225.
243

Universal Oil Products 678-04, 2004. Dissolved molecular oxygen in liquid hydrocarbons by
electrochemical detection, Honeywell Company.

Walker, R.J., 2005. Single aisle aircraft fuel system ATA 28 A318,319, A320 & A321,Report No.
Ref. SA28TM0500268, Filton: Airbus.

Wright, G.H., 1974. MOD fuel pumping research - Interim Report - The air measuring
instrument - It's development and calibration with hydrocarbon fuels, Report No. FHR90074,
Hucknall: Rolls Royce.

Wright, G.H., 1976. MOD fuel pumping research, evaluation of gas/liquid ratios in aircraft feed
systems, Report No. FHR90111, Hucknall: Rolls Royce.

Wyeth, H.W.G., Cooper, R.F., 1956. Investigation into the variation of oxygen content in the gas
space above kerosene fuel when subjected to reduced pressures, Report No. 635, Farnborough:
Royal Aircraft Establishment.

Wyeth, H.W.G., Oldfield, G.H., 1957. Treatment of aircraft fuel with nitrogen to reduce tank
explosion hazards (Part II), Report No. 210, Farnborough: Royal Aircraft Establishment.

Wyeth, H.W.G., Oldfield, G.H., Wiltshire, P.H., 1957. Treatment of aircraft fuel with nitrogen to
reduce tank explosion hazards (Part III), Report No. 210, Farnborough: Royal Aircraft
Establishment.
244

Appendix A
Rolls Royce Engine Fuel Feed Pipe Arrangements used in V/L Measurements

Interface with the


Common Fuel
System

2.5 inch diameter


glass sight tube

30 feet (1.5 inch


ID Pipe)

Beckman Oxygen
analyser and
amplifier

2.5 inch diameter


glass sight tube
22 inches

24 feet (1.5 inch


ID Pipe)

Beckman Oxygen
analyser and
amplifier

245

Appendix B
Sizing Calculation for Fuel Tank Sparger Surface Area
Extract from Mott Corporation Sparging/Gas-Liquid Contacting Design Guide and Part Selection;

246

1. Required Gas Flow in SCFM


Volume of fuel in tank = 126.6 L = 6.26 ft3
Volume of air dissolved in fuel at saturation = 0.14 126.6 L = 17.72 L
Allowing for 10 times the saturated amount of air to be introduced into the fuel via sparger
over 30 minutes;

SCFM

10 17.72 0.0353
0.2085 ft3/min
30

2. Liquid Pressure at Sparger


Height of fuel in tank = 225 mm = 0.73823 ft
SG of Jet A-1 fuel at 20C = 0.8

P 225 0.003281 0.8 0.433 0.2557 PSIG

3. Determine ACFM
Fuel temperature = 20C = 68F

ACFM SCFM

460 68 0.2081 ft3/min


14.7

14.7 0.2557
520

4. Agitator Tip Speed


Rotational speed of impeller during sparging = 300 rpm
Impeller diameter = 70 mm

70

300
25.4

Tip Speed
3.61 ft/sec
229
247

5. Required Sparger Surface Area


The gas exit velocity design limit from the sparger for an agitated tank with an impeller tip
speed = 3.61 ft/s = 25 ft/min

0.2081
0.008324 ft2 = (773.32 mm2)
25

248

Appendix C
Performance Calculations for In-line Fuel Vapour Condenser
Calculations to determine the performance of the in-line oxygen sampling system condenser are
based on condensing heat transfer theory given in McCabe, W.L., J.C. Smith, and P. Harriott,
Unit Operations of Chemical Engineering (5th Edition), McGraw-Hill, 1993, pp. 374-85.

Condenser Parameters
Inner diameter of condenser = 0.028 m
Length of condenser = 0.132 m
Flow rate of air saturated with fuel vapour through condenser = 210-5 m3/s
Density of Jet A-1 fuel vapour = 4.5 kg/m3 at 101.325 kPa
Density of Jet A-1 liquid at 40 C = 785 kg/m3
Density of Air at 40C = 1.127 kg/m3

The condensate loading on the inside of a vertically mounted tube is the mass flow rate of
condensate per unit length of periphery;

d
The Reynolds number is given by;

Re

film

249

Assuming the air entering the condenser is completely saturated with fuel vapour, the mass flow
rate is;

M Q

M 4.5 2 10 5 9 10 5 kg / s

The condensate loading in the condenser is thus;

9 10 5

0.028

9 10 5
1.023 10 3 kg / s / m
0.088

And the Reynolds No;

Re

4 1.023 10 3
2.54
0.00161

The driving force for condensation is the temperature difference between the cold condenser wall
surface and the bulk temperature of the saturated vapour;

T0 Tsat Twall
Tvapour Tsurface

Bulk temperature of saturated vapour = 313.15 K


Temperature of condenser inner wall = 273.15 K

T0 313.15 273.15 40 K

250

The viscosity and most other properties used in the condensing correlations are evaluated at the
film temperature, a weighted mean of the cold surface (wall) temperature and the (hot) vapour
saturation temperature;

3
Tsat Twall
4
3T0

T film Tsat
Tsat

3
313.15 273.15
4
3 40
313.15

4
313.15 30
283.15 K

T film 313.15

The condensing heat transfer coefficient for the inside of a vertical tube under laminar flow
conditions is given by;

k 3f f f g

0.943

0
f

1/ 4

hcond

The following parameter values are applicable to the condenser operation;


Thermal Conductivity of Jet A-1 condensate film, kf = 0.1275 W/m K
Density of Jet A-1 condensate film at film temperature, f = 805 kg/m3
Density of Jet A-1 Vapour, v = 4.5 kg/m3
Acceleration due to gravity, g = 9.81 m/s2
Heat of Vaporization of Jet A-1, = 311580 J/kg

T0 = 40 K
251

Length of condenser, L = 0.132 m


Viscosity of Jet A-1 condensate film, f at condensate film temperature of 283.15 K = 0.00161
kg/m s

The condensing heat transfer coefficient for the condenser operation is;

0.12753 805 805 4.5 9.81 311580

0.943
40 0.132 0.00161

1/ 4

hcond

0.00207 805 800.5 9.81 311580


0.943

0.0085

1/ 4

1/ 4

4077238744
0.943

0.0085
0.943 832.2
784.8W / m 2 K

The rate of heat transfer is then given by;

Q hcond A T
784.8 0.0116 40
364 W
The condensation rate within the condenser is now given by;

364
311580
1.17 10 3 kg / s

252

The condensing heat transfer coefficient of 784.84 W/m2.K can be compared with that estimated
from the Duckler plot, as shown below (dashed red line). The line intersects the hm (vertical scale)
at 140 B.t.u./hr.ft2.F. Conversion of this value to SI units is achieved by multiplying by 5.6780,
giving a value of 794.92 W/m2 K. The condensation rate for the condenser, using this condensing
heat transfer coefficient estimated from the Duckler plot, is 1.18 10-3 kg/s, which is in very good
agreement with that obtained through condensing heat transfer theory.

253

254

Appendix D
Conversion of RAE Oxygen Release Rates
The test conditions under which oxygen gas release rates from aviation kerosene were measured
by the RAE are as follows;
Altitude = 37,500 ft
Vol. of fuel used = ~ 1 ltr
Temperature of fuel = 23 1C
Rate of Climb = 3000 ft/min
Quiescent fuel contained in cylindrical glass vessel

The measured oxygen release rate was given in; litres of O2 released/min/100 litres of fuel.
At 37,500 ft the measured rate was 0.46 litres of O2 released/min/100 litres of fuel. The gas
volumes used to calculate this value were reduced to normal pressure, i.e. 1 atm.

As only 1litre of fuel was used it is necessary to convert the value 0.46 litres of O2
released/min/100 litres of fuel into O2 released per min in the actual fuel quantity used;

VS 0.46

0.0046
t
100

Correcting a volumetric release rate value of 0.0046 litres/min at 1 atm to the pressure condition
at 37500 ft such that direct comparison can be made gives;

V2

P1 V1 1 0.0046

0.0221
P2
0.2083
255

The volumetric oxygen release rate value of 0.0221 litres/min at 37500 ft is converted into a
mass release rate (kg/s) using the density of oxygen gas at the pressure condition equivalent to
37500 ft and 23C;

p
21100
21100

0.274
RT 259.8 296.15 76939.77

Where R = 259.8 J/kg/K (Individual Gas Constant for oxygen)


It follows that with an oxygen density of 0.274 kg/m3 at 37500 ft, the mass release rate of oxygen
in kg/s at 37500 ft from the RAE results table is;

0.0221

0.274
1000

m O2
1.009 10 7 kg/s
60

The % difference between oxygen release rate measurements under very similar fuel temperature
and altitude conditions for the RAE result and the dimensional modelling experiment is;

1.58402 10

1.009 10 7
100 1470
1.009 10 7

256

You might also like