Gill Stephen

Download as pdf or txt
Download as pdf or txt
You are on page 1of 135

Evaporation of Thin Lubricant Films under Laminar and Turbulent

Boundary Layers

by

Stephen James Gill

A thesis
presented to the University of Waterloo
in fulfilment of the
thesis requirements for the degree of
Master of Applied Science
in
Mechanical Engineering

Waterloo, Ontario, Canada, 2019

© Stephen James Gill 2019


Author’s Declaration
This thesis consists of material all of which I authored or co-authored: see Statement of
Contributions included in the thesis. This is a true copy of the thesis, including any required
final revisions, as accepted by my examiners.

I understand that my thesis may be made electronically available to the public.


!ii
Statement of Contributions
Section 3.2 of this thesis paraphrases content featured within a paper which has been
submitted for publication to ASME, Journal of Fluids Engineering. The paper was coauthored
by myself, my supervisor (Prof. Xianguo Li), Dr. Burak Ahmet Tuna, and Prof. Serhiy
Yarusevych. I primarily authored the text of the paper, and performed the lab experiments
presented therein. The passages in question are located on pp. 43 – 44 and concern the
experimental setup for the PIV measurements presented in this study. Likewise, Sections 3.3.4
and 4.2 feature figures initially prepared for the same publication (Figures 3-8, 3-9, and 4-4 –
4-8).

Section 3.2 and 3.3.3 also reference an accepted paper by Dr. Burak Ahmet Tuna, Prof.
Li, and Prof. Yarusevych [46] which details the duct flow development measurements
performed with the same apparatus. I had no part in authoring the paper or analyzing the post-
processed results, although I contributed to conducting the experimental measurements.


!iii
Abstract
The evaporation of engine oil hydrocarbons from the cylinder liner of internal
combustion engines is a significant contributor to engine emissions. This evaporation occurs
from the thin films deposited by the piston compression ring as it sweeps along the liner.
Establishing the magnitude of the liner evaporation with respect to different lubricant and
engine conditions is therefore important to both engine performance and oil consumption. The
contribution from liner evaporation to oil consumption has been the subject of several
experimental studies which show reasonable agreement with liner evaporation models
employing one-dimensional heat and mass transfer. However, models normally eschew
specification of the convective mass transfer conditions at the liquid interface, opting instead to
employ heat-mass transfer analogies for given convective heat transfer conditions. Direct
measurement of the evaporation of discrete oil films under turbulent boundary layer flow is
therefore a useful comparative tool for evaluating model performance and methodology,
especially for full-engine simulations which employ submodels for liner evaporation.

This study experimentally investigates the effects of laminar and turbulent boundary
layers on the evaporation of thin, liquid oil films. A wind tunnel, incorporating a flow
conditioner and test section, is designed and validated in order to produce repeatable boundary
layer flows within an aspect ratio 2 rectangular duct (20 mm × 40 mm) with a length of 1 m
(~37.5!Dh). Particle image velocimetry (PIV) is used to characterize the velocity fields and
turbulence statistics within the test section. Reynolds numbers based on hydraulic diameter of
10,650, 17,750, and 35,500 are considered, with grid-generated inlet turbulence being
manipulated by wire-meshes and validated by hot-wire measurement. The evaporation of oil
films with initial thicknesses of 50 μm and temperatures of 50°C are then measured directly by
an analytical microbalance for different exposures under the laminar and turbulent boundary
layers with varying levels of near-wall turbulence intensity and shape. Reynolds number is
found to have a significant effect on the evaporation rate. Specifically, increasing near-wall
velocity gradient is found to increase the rate of evaporation. Varying near-wall turbulence
intensity is shown to have little effect for the length scales of the study, implying that the shear
velocity and transport within the viscous sublayer are the predominant parameters governing
the convection-limited mass transfer.


!iv
Acknowledgements
First and foremost, I wish to thank my supervisor, Prof. Xianguo Li for his contributions
to my work. His advice was greatly appreciated when I struggled and invaluable throughout my
degree.

Similarly, I would like to thank the members of my examining committee, Prof. Serhiy
Yarusevych and Prof. Roydon Fraser, for their patience and insight.

I would also like to express my profound thanks to Dr. Burak Ahmet Tuna, without
whom I would not have completed this thesis. Burak was my constant mentor during my
experimental work, and introduced me to the finer points of fluid measurement techniques.

My fellow research group members have also been a great source of support, for their
advice both academic and practical.

To my friends and family, I wish to say thank you for your tireless support while I
completed my degree.

Finally, I wish to thank the machinists working for the University of Waterloo
Engineering Department and Machine Shop. When I couldn’t mill or turn something myself
you folks did so peerlessly.


!v
For my kith and kin,
and those no longer with us,
who kept me going.

!vi
Table of Contents
Author’s Declaration ................................................................................................................. ii
Statement of Contributions ..................................................................................................... iii
Abstract ....................................................................................................................................iv
Acknowledgements ...................................................................................................................v
List of Figures ..........................................................................................................................ix
List of Tables ..........................................................................................................................xii
Nomenclature ........................................................................................................................xiii
1 Introduction ......................................................................................................................... 1
1.1 Automotive Lubricant Overview 2
1.2 Lubricant Contribution to Emissions 3
1.3 Engine Oil Consumption Mechanisms 5
1.4 Boundary Layer Flow 8
1.4.1 Developing Duct Flow 11
1.4.2 Law of the Wall 12
1.4.3 Airflow Within an Engine Cylinder 14
1.5 Lubricant Films on the Cylinder Liner 15
1.6 Scope and Outline of Thesis 16
2 Literature Review ............................................................................................................... 18
2.1 Evaporation under Boundary Layer Flow 18
2.2 Multicomponent Mixture Evaporation 21
2.3 Cylinder Liner Evaporation Modelling 24
2.5 Gap In Literature 26
2.6 Approach 26
3 Boundary Layer Measurements ......................................................................................... 28
3.1 Wind Tunnel Design 28
3.2 Experimental Setup 37
3.3 Flow Characterization 45
3.3.1 Experimental Conditions 45
3.3.2 Flow Conditioner Performance 46
3.3.3 Duct Flow Measurements 48

!vii
3.3.4 Evaporation Test Section PIV Measurements 50
3.3.5 Law of the Wall 52
4 Evaporation Measurements ............................................................................................... 56
4.1 Experimental Setup 56
4.2 Evaporation Results 64
5 Conclusions ........................................................................................................................ 74
6 Future Work ....................................................................................................................... 76
References ............................................................................................................................... 77
Appendix A – Blasius Laminar Boundary Layer Solution ..................................................... 84
Appendix B – Wind Tunnel Design Drawings ....................................................................... 86
Appendix C – Detailed Test Section Flow ............................................................................. 96
Appendix D – Evaporation Measurement Data ................................................................... 115

!viii
List of Figures
Figure 1-1. ........Schematic of a typical, two-dimensional boundary layer flow over a flat plate where the
flow transitions from laminar to turbulent ........................................................................9
Figure 3-1. ........Overall layout of the flow conditioner, including major changes between inlet and outlet
cross-sections for different components ...........................................................................30
Figure 3-2. ........Final contraction geometry resulting from the Morel axisymmetric design method,
including equivalent inlet and outlet geometries .............................................................33
Figure 3-3. ........Major components of the evaporation test section and their assembly. Also shown are
the sample holder inset geometry and coordinate system ...............................................40
Figure 3-4. ........General coordinate system and duct geometry implemented for the experimental fluids
measurements ..................................................................................................................41
Figure 3-5. ........Experimental setups for experimental fluids measurements ...........................................42
Figure 3-6. ........Perturbation of the streamwise velocity and turbulence intensity for ReDh = 1.0 × 104,
1.7 × 104, and 3.5 × 104 at the contraction exit .............................................................47
Figure 3-7. ........Shape factor variation of the boundary layer within the x! -!y plane at the duct centreline
(!z = 0) for Grid 3 (red) and Baseline (black) inlet conditions ...........................................49
Figure 3-8. .......Convergence of downstream velocity profiles for each Reynolds number and inlet
condition. Grid 1 and 2 cases omitted for clarity in a) ....................................................51
Figure 3-9. .......Velocity profiles above the sample holder at each axial location (!x /a = 2 and 45), shown
over the length of inset (!z = 0) and across its half-width (!x′ /a = 0.5) .............................52
Figure 3-10. ......u! + profiles within the x! -!y plane at x′
! /a = -0.1 (!x′ /a = 45) for each Reynolds number
case under the a) Baseline and b) Grid 3 inlet conditions. Profiles are compared to either
the Musker profile (black) or !u + = y + (red) ......................................................................54
Figure 4-1. .......Main components of heated parts within the evaporation test section and the clearances
between them ..................................................................................................................57
Figure 4-2. .......IR measurements of a 50 μm oil film and the sample holder. Showing a) the change in
film thickness with time under approximately the Baseline, ReDh = 1.7 × 104, b) the
associated experimental setup, and c) temperature distribution over the sample holder
inset for !T = 150°C (!σ = 1.1°C) .....................................................................................62
Figure 4-3. .......Evaporation results and estimated evaporation rates for oil films for each airflow case,
including the natural evaporation case ...........................................................................65
Figure 4-4. .......Comparison of Baseline, Re ! Dh = 10,650, 17,750, and 35,500 evaporation cases at
x! /a = 2. Showing a) film evaporation and b) velocity profiles. Subfigures show same
properties with relevant parameters scaled by δ* ! . Solid black line in b) denotes the
Blasius laminar boundary layer .......................................................................................67
Figure 4-5. .......Comparison of Baseline and Grid 3 cases for Re ! Dh = 10,650 at x! /a = 2. Showing a)
film evaporation, b) velocity profiles, and c) turbulence intensities .................................68

!ix
Figure 4-6. .......Comparison of Baseline and Grid 1 – 3 cases for Re ! Dh = 17,750 at x! /a = 2. Showing a)
film evaporation, b) velocity profiles, and c) turbulence intensities .................................69
Figure 4-7. .......Comparison of Baseline, Re ! Dh = 10,650 and 17,750 cases at x! /a = 45. Showing a) film
evaporation, b) velocity profiles, and c) turbulence intensities ........................................71
Figure 4-8. .......Evaporation rates from linear regression of forced convection data (!m· f) relative to
natural evaporation rate (!m· n). Shown with respect to a) Re ! c and b) Re ! δ* for each
Baseline and Grid case ....................................................................................................72
Figure B-1. .......Drawing of wind tunnel settling chamber and component assemblies (assy.) .................87
Figure B-2. .......Drawing of test section mounting plate and insert geometry .........................................88
Figure B-3. .......Drawing of top, bottom, and side contraction components ...........................................89
Figure B-4. .......Drawing of settling chamber mounting plates ................................................................90
Figure B-5. .......Drawing of settling chamber screens and honeycomb holder ........................................91
Figure B-6. .......Drawing of diverging section (diffuser) side wall components ........................................92
Figure B-7. .......Drawing of diverging section (diffuser) top and bottom wall components .....................93
Figure B-8. .......Drawing of top and bottom halves of the circular-rectangular transition ......................94
Figure B-9. .......Drawing of transition mounting plates ...........................................................................95
Figure C-1. .......Summary of PIV measurements for Baseline, !ReDh = 10,650 case at x! /a = 2 ..............97

Figure C-2. .......Summary of PIV measurements for Baseline, !ReDh = 35,500 case at x! /a = 45 ............97

Figure C-3. .......Summary of PIV measurements for Baseline, !ReDh = 17,750 case at x! /a = 2 ..............98

Figure C-4. .......Summary of PIV measurements for Baseline, !ReDh = 35,500 case at x! /a = 2 ..............98

Figure C-5. .......Summary of PIV measurements for Grid 3, !ReDh = 10,650 case at !x /a = 2 ................99

Figure C-6. .......Summary of PIV measurements for Grid 1, !ReDh = 17,750 case at !x /a = 2 ..............100

Figure C-7. .......Summary of PIV measurements for Grid 2, !ReDh = 17,750 case at !x /a = 2 ..............101

Figure C-8. .......Summary of PIV measurements for Grid 3, !ReDh = 17,750 case at !x /a = 2 ..............102

Figure C-9. .......Summary of PIV measurements for Grid 1, !ReDh = 35,500 case at !x /a = 2 ..............103

Figure C-10. .....Summary of PIV measurements for Grid 2, !ReDh = 35,500 case at !x /a = 2 ..............104

Figure C-11. .....Summary of PIV measurements for Grid 3, !ReDh = 35,500 case at !x /a = 2 ..............105

Figure C-12. .....Summary of PIV measurements for Baseline, !ReDh = 17,750 case at x! /a = 45 ..........106

Figure C-13. .....Summary of PIV measurements for Baseline, !ReDh = 35,500 case at x! /a = 45 ..........107

Figure C-14. .....Summary of PIV measurements for Grid 3, !ReDh = 10,650 case at !x /a = 45 ............108

Figure C-15. .....Summary of PIV measurements for Grid 1, !ReDh = 17,750 case at !x /a = 45 ............109

!x
Figure C-16. .....Summary of PIV measurements for Grid 2, !ReDh = 17,750 case at !x /a = 45 ............110

Figure C-17. .....Summary of PIV measurements for Grid 3, !ReDh = 17,750 case at !x /a = 45 ............111

Figure C-18. .....Summary of PIV measurements for Grid 1, !ReDh = 35,500 case at !x /a = 45 ............112

Figure C-19. .....Summary of PIV measurements for Grid 2, !ReDh = 35,500 case at !x /a = 45 ............113

Figure C-20. .....Summary of PIV measurements for Grid 3, !ReDh = 35,500 case at !x /a = 45 ............114

!xi
List of Tables
Table 3-1. .........Screen design parameters for different Re
! Dh. ..................................................................36

Table 3-2. .........Wire mesh dimensions for each turbulence grid. ............................................................39
Table 3-3. .........Dry air properties at 19°C. .............................................................................................46
Table 4-1. .........Experimental conditions for film evaporation measurements. ........................................61
Table C-1. ........Colour scheme for !z /c locations. .....................................................................................96
Table D-1. ........Evaporation data for Natural Convection case (!Grc = 3838) .......................................115
Table D-2. ........Evaporation data for Baseline, !ReDh = 17,750 case (!x /a = 2) ......................................116

Table D-3. ........Evaporation data for Baseline, !ReDh = 10,650 case (!x /a = 2) ......................................117

Table D-4. ........Evaporation data for Baseline, !ReDh = 35,500 case (!x /a = 2) ......................................117

! Dh = 10,650 case (!x /a = 2) .........................................118


Table D-5. ........Evaporation data for Grid 1, Re

! Dh = 17,750 case (!x /a = 2) .........................................118


Table D-6. ........Evaporation data for Grid 1, Re

! Dh = 17,750 case (!x /a = 2) .........................................119


Table D-7. ........Evaporation data for Grid 2, Re

! Dh = 17,750 case (!x /a = 2) .........................................119


Table D-8. ........Evaporation data for Grid 3, Re

Table D-9. ........Evaporation data for Baseline, !ReDh = 10,650 case (!x /a = 45) ....................................120

Table D-10. ......Evaporation data for Baseline, !ReDh = 17,750 case (!x /a = 45) ....................................120


!xii
Nomenclature
⟨! ⟩ Time-averaged measured variable
u
! Time or spatially averaged variable, e.g. u

! Subscript designating quantity outside a boundary layer
′! Superscript designating fluctuating variable, or indicating a derivative
A Cross-sectional area (m2)
a
! Duct height (m)
B
! Law of the wall constant
B
! e Experimentally measured cylinder bore diameter (m),
b
! Duct width (m)
C
! f Friction coefficient
c! Film length, inset length, and inset width (m)
ci Concentration of species i (mol m-3)
cp Specific heat with respect to constant pressure (kJ kg-1 K-1)
D
! h Hydraulic diameter (m)
! 1, !D2
D Equivalent contraction inlet and outlet diameters (m)
D, Dm Molecular diffusivity (m2 s-1)
Dt,y Turbulent molecular diffusivity (m2 s-1)
d
! Wire diameter (m)
!Gr Grashof number
gm Mass convection coefficient (kg m-2 s-1)
H
! Shape factor
h
! Cylinder liner oil film thickness (m)
h
! f Film thickness in two-phase flow (m)
h
! min Minimum oil film thickness (m)
K
! Static pressure drop coefficient
k Thermal conductivity (W m-2 K)
L
! Total length (m)
L
! e Duct entrance length (m)
l! , !m Wavenumbers
M
! Mesh wire spacing (centre to centre) (m)
m
! A Mass of air (kg)
m evap Evaporated mass per initial film area (kg m-2)
m evap,t=0 Evaporated mass per initial film area prior to t = 0 (kg m-2)
m
! F Mass of fuel (kg)
m
! f Final film mass (kg)

!xiii
m
! i Initial film mass (kg)
m
! o Sample holder mass (kg)
m
! 1 Initial mass of sample holder and film (kg)
!m 2 Final mass sample holder and film (kg)
m· , m·
f Evaporation rate (mass flux) from film (kg m-2 s-1)
m· n Evaporation rate from film under natural convection (kg m-2 s-1)
N
! Number of samples
!
Nu Nusselt number
n
! i Number of molecules or species !i (mol), or a constant
n
! tot Total number of molecules
P
! Wetted perimeter of duct (m)
!
Pr Prandtl number
p
! Pressure (Pa)
p
! i Vapour pressure of species !i in solution (Pa)
p*
! i Vapour pressure of pure species !i (Pa)
!Re Reynolds number
!
Sc Schmidt number
!
Sh Sherwood number
!St Stanton number
s! Musker profile constant
T
! b Base temperature (°C)
T
! f Film temperature (°C)
T
! g Gas temperature (°C)
T
! s Surface temperature (°C)
t! Time, or exposure time (s)
t! 0 Preheat time (s)
U Characteristic velocity (m s-1)
u Streamwise (x) velocity component (m s-1)
u 0 Incoming perturbation (across y, z) of axial velocity to a screen (m s-1)
uN Outgoing perturbation (across y, z) of axial velocity from a screen (m s-1)
u τ Shear velocity (m s-1)
!u + Streamwise velocity, wall scaling
! m+
u Musker profile for velocity within inner region
V Velocity vector (m s-1)
·
Vf /B Film volume flux per unit breadth in two-phase flow (m2 s-1)
v Wall-normal (y) velocity component (m s-1)
v! Hot-wire voltage (V)

!xiv
w Cross-stream (z) velocity component (m s-1)
we Average cylinder gas velocity, empirical function (m2 s-1)
X
! Settling chamber screen spacing (m)
x! Streamwise/axial coordinate (m)
x! i Mole fraction of species i!
x! m Location of contraction cubic match point (m)
x! min Location of minimum velocity wall pressure coefficient (m)
x! o Virtual origin of contraction boundary layer (m)
x′
! Axial coordinate from inset leading edge (m)
Y
! Mass fraction
y! Wall normal coordinate (m)
y! vt Smooth boundary layer viscous sublayer height (m)
y! 0 Pool leading edge land roughness (m)
y! + Wall-normal direction, wall scaling
z! Cross-stream coordinate (m)
α Thermal diffusivity (m2 s-1)
β
! Mesh porosity
δ! Boundary layer thickness (m)
δ*
! Displacement thickness (m)
δ! d Mass boundary layer thickness (m)
ϵ! Emissivity
ζ! Contraction design parameter (m)
η! Similarity variable in Blasius solution
Θ
! Crank angle (degrees)
θ! Momentum thickness (m)
θi, j Excess temperature at ith and jth location (°C)
Φ
! Relative error function
ϕ
! Equivalence ratio
κ! von Kármán constant
μ
! Dynamic viscosity (kg m s)
π
! Ratio of a circles circumference to its diameter
ρ Density (kg m-3)
ρo Oil density (kg m-3)
σ! Standard deviation, or mesh solidity
τ! w Wall shear stress (Pa s)
ν Kinematic viscosity (m2 s-1)
ψ
! Streamfunction
ω Engine speed (s-1)


!xv
1 Introduction
The modern internal combustion (ICE) engine is ubiquitous in today’s society, generating
the mechanical power used by automobiles, aircraft, electric generators, and many other
machines. Among the chief design concerns of an ICE are efficiency and the emission of
combustion products. Climate change and air quality are paramount concerns, and as such
controls have been placed on the emissions of several compounds found in ICE exhaust and
are enforced by government regulation. Catalytic converters have been used for decades to limit
the emission of pollutants such as carbon monoxide (CO), unburned hydrocarbons (HC), and
nitrous oxides (NOX)., although the emission of carbon dioxide (CO2) remains a challenge as a
primary product of combustion and catalytic reactions. In addition, particulates (soot) are a
significant product of diesel engines, although catalytic converters are capable of removing a
large portion of the hydrocarbon species condensed on the solid carbon particles, known as the
soluble organic fraction (SOF). Engine load, equivalence ratio (or fuel-air ratio), engine speed,
and timing directly effect both cycle efficiency and the foregoing emissions. The equivalence
ratio, !ϕ, is defined as,

mF /mA
ϕ
! = (1-1)
(mF /mA)st

which is a ratio of the actual fuel-to-air ratio and the stoichiometric fuel-air ratio, where there is
sufficient air to react with all of the fuel. Lean combustion (φ < 1) increases engine efficiency,
and reduces the emission of HC and CO, yet operates at near maximum production of NOX
compounds and results in less power than richer combustion (φ > 1). The overall efficiency of
an engine is also heavily impacted by the action of friction between components. Mechanical
losses account for approximately 30% of the losses from power produced by the combustion
cycle [1-3], and are mitigated by lubrication between the moving and stationary components.

Lubrication within the piston-cylinder tribosystem of a modern internal combustion


engine is accomplished by liquid oil between the piston rings and cylinder liner. Thin, liquid
lubricant films remain on the liner after the passage of the piston rings and are exposed to
transient gas temperatures, liner temperatures, and turbulent airflow present in the cylinder

!1
during operation. The lubricant used is almost always a liquid hydrocarbon mixture (oil), and
due to the dynamics of the moving components within an ICE, and the heat generated, the
hydrocarbons participate in the combustion cycle due to several different oil consumption
mechanisms. Along with inertial loss mechanisms, evaporation from the cylinder liner is a
significant contributor to engine oil consumption and consequently engine emissions [4, 5].

The hydrocarbon species composing the multicomponent lubricants, which are


predominantly n-paraffins for synthetic and conventional oils, evaporate at varying rates
governed by the vapour pressures of species at the liquid-gas interface, gas phase mass transfer
conditions, and diffusion of species within the liquid phase. Especially in diesel engines,
unburned lubricant present in the gas phase is a major contributor to the soluble organic
fraction (SOF) of particulate emissions [6, 7]. In some cases, lubricant consumption can be the
predominant source of n-paraffins and polycyclic aromatic hydrocarbons found in the SOF [7].
Therefore, the evaporation of oil films from the cylinder liner is of significant interest with
respect to both engine performance and environmental considerations. The following sections
provide context for the nature of engine lubricants and the conditions under which lubricant
films are created and evaporate.

1.1 Automotive Lubricant Overview

In general, an automotive lubricant or oil is a mixture of liquid hydrocarbons


components which facilitates lubrication between mechanical parts. The hydrocarbon mixtures
which are combined to form the oil are referred to as base stocks. Automotive lubricants
commonly fall into two categories which will be familiar to the layman: conventional and
synthetic oils. The difference between the two is in how their base stocks are derived. For
conventional oils, base stocks are distilled from crude oil (petroleum). Base stocks are defined as
paraffinic, naphthenic, or intermediate based on the tendency of their composition, although each
base stock is a combination of those compounds [1]. Synthetic oils are, generally, composed of
base stocks which have been produced through chemical processes other than distillation from
crude oil, including: synthesized hydrocarbons (from carbon monoxide and hydrogen), organic
esters, and silicones, among others [1]. However, the practical definition of a synthetic oil is

!2
very broad due to the varying degrees modern base stocks will include a blend of conventional
and synthesized hydrocarbons [1]. SAE J357 [1] defines base stocks based on the following
classifications, which provides a useful summary of the above: refined petroleum base stocks, re-refined
or recycled petroleum base stocks, synthetic base stocks, and partial synthetic base stocks. Oil additives are
also added in small quantities to increase a facet of the lubricants performance, including: pour
point depressants, viscosity modifiers, anti-oxidants, detergents, detergent-inhibitors, and
dispersants [1].

Since the primary goal of an engine oil is mechanical lubrication they are first and
foremost classified based on their viscosities (rheological performance). SAE J300 [1] provides a
classification method for engine oils based on viscosities measured at high and low temperatures
via procedures outlined in several ASTM standards. This is necessary due to the decrease in
viscosity with temperature experienced by liquid oils [1]. The classifications therein are
represented by number and letter grades such as “5W-30”, and are the most common in North
America. SAE J300 [1] uses two series of viscosity grades to classify single viscosity-grade oils,
which are represented by a number with and without a ‘W’. The W designation means that the
oil grade is based on both maximum low-temperature viscosity and minimum kinematic
viscosity at 100°C, while no W indicates that the grade is based on a set of minimum and
maximum kinematic viscosities at 100°C and a minimum high-shear (106 s-1) viscosity at
150°C. Furthermore, multigrade oils are defined based on the maximum low-temperature
viscosity of a W grade oil combined with the high-shear and kinematic viscosities of a non-W
grade. An example of a multigrade oil would be the aforementioned “5W-30” oil. Many
automotive oils are multigrade, which ensures predictable performance at both high and low
temperatures. This is important because of the wide temperature conditions automotive
engines are expected to operate under, from cold-start conditions during winter to the high
engine load conditions where the cylinder temperature is highest.

1.2 Lubricant Contribution to Emissions

Analysis of automotive engine exhaust shows that lubricating oil within the engine
contributes directly to emissions. Engine emissions are the gaseous, solid, and liquid byproducts

!3
of combustion that constitute engine exhaust. The common products of hydrocarbon
combustion make up the majority of the engine exhaust: namely, nitrogen gas (N2), carbon
dioxide (CO2), water vapour (H2O), and some oxygen gas (O2). However, more harmful
compounds are also produced during and after the complexities of combustion within the
cylinder, including: nitrous oxides (NOX), sulphur dioxide (SO2), and carbon monoxide (CO).
These gases are joined by unburned hydrocarbons, volatile organic compounds (VOCs), and
solid particulate matter nucleated in the exhaust or formed on the cylinder/piston surfaces. In
general these substances produce unwanted side-effects and therefore their production is
limited by design, such as through catalytic converters and careful engine operation. Some
prime examples of this would be the greenhouse effect caused by CO and CO2 and the adverse
health affects of particulate matter, VOCs, and SO2.

The role of lubricating oil in affecting emissions is primarily found in solid, particulate
emissions. Due to loss mechanisms within the engine and fuel spray interaction with the engine
surfaces, liquid oil and vaporized oil components are present among the cylinder gasses. Oil
compounds can then become involved in the combustion process, or exhausted directly from
the cylinder. Therefore, when particulate matter (soot) nucleates, organic compounds from the
oil are adsorbed into the particle surfaces. These compounds, including paraffins and polycyclic
aromatic hydrocarbons (PAH), are generally found in the soluble organic fraction (SOF) of the
particles.

Several studies have determined that a significant portion of the particulate matter SOF
found in diesel engine exhaust can be attributed to engine lubricants [6, 8], regardless of the
fuel used except where low-quality fuels cause the particles to become saturated with fuel
hydrocarbons [6]. In a study by P. T. Williams et al. [8] it was determined that lubricant
contributes to 80 – 90% of the particulate SOF, and accounts for approximately 10% of the
PAH found therein with fresh oil. Additionally, the lubricant contribution to polycyclic aromatic
compounds (PAC) levels in the exhaust was found to be exacerbated by unburned diesel fuel
“leaking” into the engine oil [8-10]. The PAH levels found in aged engine oil are also generally
higher than that of diesel fuel [7], and can therefore be a more significant source of unburned
PAH than the fuel itself. Paraffins from the engine oil are also a major part of its SOF
contribution and account for 40% – 80% of emitted paraffins [7]. Higher contributions are

!4
noticeable at higher mean “engine loads” [7], where engine load is the torque output of the
engine (and proportional to power output). Similarly, lubricant makes significant contributions
to the composition of nanoparticles (< 50 nm), where most of their mass composed of paraffins
sourced in unburned fuel and engine oil [11].

The direct effect of engine oil on particulate emissions in general can also be considered,
meaning the effect of lubricant properties such as viscosity and Noack volatility on the amount
of particulates found in a DI diesel engine exhaust. Noack volatility is the result of a
specification [10] for quantifying oil mixture volatility based on the percentage mass loss of the
mixture when held at a given temperature (250°C). An increase in viscosity has been shown to
result in decreased particulate emissions, with the effects of high and low-shear viscosities being
similar [9, 12]. This is attributed to the formation of thicker films at lower viscosities, and
encouragement of oil entrainment into reverse flow past the piston rings, which allows larger
volumes of oil to enter the combustion chamber (higher consumption) [9]. Higher volatility was
also shown to increase the amount of particulates present in the exhaust [12]. Additionally, the
contribution of hydrocarbons from the engine oils to the particulate SOF was shown to
increase with engine speed and decrease with engine load [6, 8, 12].

1.3 Engine Oil Consumption Mechanisms

The consumption of oil within an IC engine is controlled by the motion and geometry of
the piston-cylinder assembly as well as lubricant properties. Of particular importance to this
study is the number and type of piston rings located in grooves around the circumference of
the piston. These piston rings form the primary point of contact between the piston and the
cylinder liner, and are where the lubricant acts to reduce friction and wear. These rings serve
several purposes: forming a mechanical seal between the combustion chamber and crankcase;
transporting lubricant along the cylinder liner; improving heat transfer from the piston to the
cylinder liner; and scraping oil from the cylinder liner back to the piston skirt and sump. The
top ring, or compression ring, provides the primary seal for the combustion chamber. The top
ring groove is designed such that the high pressures within the combustion chamber act to press
the ring against the cylinder liner, which improves the seal. The second ring, or wiper ring, is

!5
intended to aid in removing excess oil from the cylinder liner while providing additional sealing.
The last ring, or oil ring, is closest to the crankcase and is used to control the supply of oil to the
upper regions of the piston. The most common number of rings found in four-stroke IC
engines is three [2].

Oil consumption refers to the removal of liquid oil from the engine during operation,
The mechanisms leading to oil consumption can be grouped into two categories: evaporative
and inertial. Evaporative loss mechanisms are governed by convective mass transport of the oil
components. One example is liner evaporation, where liquid oil present on the cylinder liner
evaporates under high temperatures and airflow during the engine cycle. Similar evaporation
occurs beneath the piston as well, where liquid oil can also be entrained into any blow-by gas
flow (combustion gases which breach the piston rings and enter the crankcase). Oil constituents
can therefore enter the combustion chamber if the blow-by gasses are vented into the intake
manifold, and represent a loss mechanism regardless [2, 13].

Conversely, inertial loss mechanisms are controlled by the inertia of the piston-cylinder
system. During operation liquid oil can accumulate at the piston top land [2, 13], and may be
“thrown off ” into the combustion chamber due to inertia donated by the piston during an
upward stroke. A similar effect is created when pressures in the piston second land become
higher than the combustion chamber pressure, creating a positive pressure gradient. Gas may
then flow around the top ring groove, or through the gap in the compression ring, which can
carry liquid and evaporated oil into the combustion chamber [2, 13]. In each case, the oil is
most often removed from the cylinder via exhaust gases after liquid oil or evaporated
constituents enter the bulk flow within the combustion chamber [2, 4, 13].

The contributions to oil consumption made by each mechanism have been explored
extensively through bench and field testing with both test and production engines, respectively.
In general, these tests were accomplished by subjecting oils with differing viscosities and
volatilities to the same set of conditions and measuring the amount of oil consumption directly.
While both viscosity and volatility are important characteristics to the rheological performance
and longevity of an oil, they are also the most significant controllable parameters that relate to
the consumption mechanisms. As such, relative differences in consumption can be related to
differences in oil properties.

!6
Volatility has been highlighted as one of the most important sources of oil consumption
in both field tests [14, 15] and bench tests [14, 16]. Where the role of viscosity, and the
contribution of inertial mechanisms to overall oil consumption, has been found to depend
heavily on engine design [16], the role of evaporation and volatility is directly dependent on
engine conditions and oil composition. In this case, volatility is the tendency for a liquid to form
a vapour. Specifically, it can be defined as the standardized mass fraction lost when the mixture
is held at a specific temperature (Noack volatility) [10], by mass or volume fraction lost based
on a temperature range (distillation curve), or a similar boiling point distribution generated
through a procedure such as in ASTM D2887 (gas chromatography) [17]. Test engine oil
consumption has been found to increase by at least 1% for every 1% increase in species distilled
at 371°C [14, 17] in test engines. However, consumption trends as high as 4.5% [14] and 3%
[15] have been found in field tests. The contribution of volatility to consumption is also affected
by engine temperature, where higher volatility oils (based on distillation curves) have been
found to not only exhibit increasing oil consumption with increasing surface temperatures
(piston and cylinder liner), but also a relatively larger increase for higher volatility oils [18]. This
is expected based on how volatility is defined above, and the behaviour of vapour pressure with
temperature [19], and serves to emphasize the importance of high temperature operation to
evaporation within an engine.

The relative contribution from each loss mechanism is more difficult to measure than the
role of bulk oil properties. However, a combination of modelling and experimental
measurements can allow estimation of the relative contributions of each mechanism, as in the
work of E. Yilmaz et al. [2, 13]. A sulphur-trace method was used to measure the transient oil
consumption from a production engine concurrently with liner temperature, oil film thickness,
and cylinder pressure. These measurements were conducted for various engine loads and
speeds, and blow-by gas venting conditions. Combined with a cylinder liner evaporation model
[4], the relative effects of liner evaporation, blow-by entrainment, and inertial mechanisms (the
remainder) were determined. While oil consumption was found to increase with engine speed,
the contributions from each source remained relatively the same [2, 13]. However, at low
engine load, the inertial mechanisms were found to be the highest contributors to consumption
(> 90%), while higher engine load was dominated by liner evaporation (> 40%) [2, 13]. Blow-

!7
by entrainment was found to increase with engine load as well, but was relatively insignificant
(< 15%). These results are supported by additional consumption modelling based on measured
engine operating conditions [20, 21].

One key conclusion is that liner evaporation is the predominant source of evaporation
within the cylinder, which is itself a major contributor to oil consumption (and thus emissions)
at high engine-loads. Cylinder liner evaporation occurs primarily from the film of oil deposited
onto the liner surface by the piston rings. Numerous models have been created to estimate the
liner evaporation, and are affected by a number of factors: oil film thickness, distribution of
liquid oil along the cylinder [22], oil composition, oil and gas temperatures (heat transfer to the
film), and the gas phase velocity field. The gas phase flow within the cylinder is therefore
particularly important, as it controls the convective mass transport from the film [19], and is
largely responsible for the heat transfer at the liner.

1.4 Boundary Layer Flow

The heat and mass transfer between the liquid film and the gas phase within an engine
cylinder is governed primarily by thin regions above the liquid surface known as boundary layers.
In each case a quantity such as velocity, temperature, or species concentration varies from a
bulk value in the gas phase to its value at the liquid surface or wall; the nature and degree of
these gradients controls the transport of momentum, heat, and mass within the gas phase. In
each case, there exists a quantity which governs the transport; kinematic viscosity (!ν), thermal
diffusivity (!α), and molecular diffusivity (!D), respectively. The momentum boundary layer is of
paramount importance in this case, as it jointly governs the rate at which heat and vaporized
species are convected to and from the film.

With respect to the momentum (velocity) boundary layer, viscous fluid flow over a surface
interacts via friction. Generally, fluid particles near a surface will tend to match the velocity of
the surface because the force of cohesion between fluid particles is weaker than the adhesion
force between the particles and the wall; this is known as the “no-slip” condition. This results in
a boundary layer, a region where the relative velocity varies from zero at the surface to some
relative velocity farther from the wall (!U∞). This smooth variation is governed by viscosity,

!8
Figure 1-1. Schematic of a typical, two-dimensional boundary layer flow over a
flat plate where the flow transitions from laminar to turbulent.

which causes a shear stress between fluid moving at different speeds, decelerating the faster
fluid. An increase in thickness (!δ) of the region of decelerated fluid is normally evident as the
fluid flows downstream, as fluid near the edge of the boundary layer is in turn decelerated.

The simplest form of boundary layer flow is over a planar surface, shown schematically in
Figure 1-1. In this case, the fluid velocity field is described in cartesian coordinates (!x, y! , !z) as,

!V = u i ̂ + v j ̂ + w k̂ (1-2)

with each component given in terms of time-averaged and fluctuating components:

!u (x, y, z, t) = ⟨u⟩(x, y, z) + u′(x, y, z, t) (1-3)

A time-averaged quantity such as !⟨u⟩ is given in terms of a sample of size !N as,

N
∑i=1 ui
!⟨u⟩ = (1-4)
N

Together, the time-averaged and fluctuating velocity components serve to describe the fluid
flow both spatially and in time, which is relevant for unsteady or turbulent flows where the
velocity might vary rapidly at a single point. A useful statistic for describing the level of

!9
turbulence at a specific point in space is the root mean square (rms) of the fluctuating velocity
component, given by:

∑i=1 (ui − ⟨u⟩)


N 2

!ur′ms = (1-5)
N

While Figure 1-1 demonstrates two-dimensional flow over a flat plate, boundary layer flows
exist in the vicinity of all surfaces (liquid or solid) and need not be described in cartesian
coordinates. The geometry of an engine cylinder is complex, although the airflow is generally
described in cylindrical coordinates [23]. Similarly, the boundary layer flow over curved
surfaces such as an aerofoil can be described in terms of cartesian coordinates or in terms of
tangential, normal, and parallel components relative to the local surface.

Like all fluid flows, boundary layer flow may be laminar or turbulent. For a given set of
flow conditions and geometry, the state of a boundary layer is local and depends on Reynolds
number, given for a velocity !U as:

Ux
!Rex = (1-6)
ν

While a boundary layer may begin laminar, downstream it may eventually transition to
turbulence (Figure 1-1b) due to the onset of instabilities [24]. In terms of laminar flow, the
Blasius laminar boundary layer solution is an excellent descriptor of steady, two-dimensional (as
in Figure 1-1a). It provides a simple comparative tool for evaluating whether a given boundary
layer flow is laminar, transitioning, (Figure 1-1b), or turbulent (Figure 1-1c). It is presented in
terms of a similarity solution to simplified versions of the mass and momentum conservation
equations (Navier-Stokes equations), which is given in Appendix A.

Several integral quantities are useful for describing boundary layers based on their shape.
The displacement (!δ*) and momentum (!θ) thicknesses, given by:

∫0 ( U∞ )
δ
u
!δ* = 1− dy (1-7)

∫0 U∞ ( U∞ )
δ
u u
!θ = 1− dy (1-8)

together describe a boundary layer’s shape factor:

!10
δ*
!H = (1-9)
θ

The shape factor is a useful way of defining the state of a boundary layer in terms of laminar
and turbulent flow. For laminar flow, a boundary layer will exhibit a shape factor of
approximately 2.59 (as in the Blasius solution), while turbulent boundary layers have been
shown experimentally to reach shape factors of 1.4 – 1.6 [25].

1.4.1 Developing Duct Flow

Internal boundary layer flow within a duct may be distinguished from external flow over
a flat plate by both a pressure gradient and boundary layer growth which is limited by the duct
walls. For an axisymmetric duct (circular) the boundary layer growth is also symmetric towards
the centreline of the duct. Similarly, for rectangular ducts boundary layer growth occurs
simultaneously at each wall. The boundary layer growth is accompanied by an acceleration in
the inviscid core of the flow (outside the boundary layers) with boundary layer growth due to
conservation of mass. This is further complicated by the advent of transition from laminar to
turbulent flow, when the initial flow is laminar, which occurs at some downstream point
depending on Reynolds number, similar to flat plate flow.

The boundary layer growth is limited in that, eventually, the boundary layer edges at each
wall must meet. Transition aside, once this occurs the velocity field within the duct will become
invariant with axial distance, which is known as “fully developed” flow. The centreline velocity
also reaches a maximum at this point, although it may overshoot its final amplitude [25] prior
to settling. The length from the duct inlet to the point where the flow is fully developed is
known as the entrance length (!Le) and defines the “developing region”. The boundary layer
development for laminar flow within ducts of various cross-sections has been the subject of
numerous experimental [26-28] and numerical studies [29-31]. The length of the development
region has similarly been the subject of numerous studies [32-34] and is known to increase with
Reynolds number, since with increasing Reynolds number the rate of boundary layer growth
decreases [32-34].

!11
The behaviour of flows transitioning from laminar to turbulent, either before or after the
flow becomes fully developed, is less simple to predict. Flows will remain laminar for Reynolds
numbers based on the duct hydraulic diameter less than approximately 2300 [35], where the
hydraulic diameter is defined as:

4A
!Dh = (1-10)
P

For 2300 ≤ Re
! Dh ≤ 10,000, flows are transitional, meaning they may change rapidly between

laminar and turbulent flow randomly [35]. At higher Reynolds numbers (≥ 10,000), transition
to turbulent flow can occur within the development region [35-37]. However, locating and
quantifying the precise point at which a flow begins transition is still an open question and has
been the subject of numerous experimental and analytical studies [24, 36-47]. The degree of
inlet turbulence has been found to have a significant effect on where transition begins and its
duration [38-43]. Changes in small disturbances in the earlier stages of transition (Tollmien-
Schlichting waves) have been shown to lead to more rapid transition [24, 44, 45] in flat plate
flow. A similar effect has been found in duct flow, where higher inlet turbulence intensity has
been shown to move transition towards the inlet of the duct [40, 46, 47].

Internal duct flow is convenient for experimental work as the flow conditions may be
tightly controlled, as in a wind tunnel test section where flow conditioning produces an initially
uniform, and quiet, velocity field. Further manipulation of the flow from that point produces
the desired measurement conditions. Similarly, knowledge of the boundary layer development
and turbulence generation along the length of any test section is required for producing the
desired boundary layer conditions, especially when the duct boundary layers themselves are the
subject of experimentation. Therefore, determining the sizing for an appropriate test section,
and having adequate inlet conditions, is important.

1.4.2 Law of the Wall

Characterizing turbulent boundary layers is useful for establishing repeatable


experimental conditions. Turbulent boundary layers have been found to match a universal

!12
velocity profile when scaled by the wall shear stress, called the “law of the wall” [48, 49]. The
wall scaling is normally presented in terms of a shear velocity, given by:

τw
!uτ = (1-11)
ρ

The scaling of the streamwise velocity (!u) and wall-normal (!y) distance is then given by:

u
!u + = (1-12)

yu
y! + = τ (1-13)
ν

Experimental measurements of turbulent flow over flat plates has shown that turbulence
intensity increases from a minimum outside the boundary layer edge to a maximum near the
wall, due to the high velocity gradient present across the boundary layer. However, the
turbulent fluctuations then proceed to zero at the wall, where the velocity gradient is highest. It
has also been shown that the shear stress within the flow remains constant and equivalent to the
wall shear stress within a similar region close to the wall. This region is named the “laminar
sublayer” where the influence of turbulence disappears (e.g. u′
! v′ → 0) and the velocity profile is
essentially laminar and linear in terms of wall scaling [48, 49]:

!u + = y + (1-14)

This region has been found to occur for y! + ≤ 5. The region within 30 ≤ y! + ≤ 200 also tends to
follow a universal profile, given by [48, 49],

1
!u + = ln(y +) + B (1-15)
κ

over which the viscous shear stress wanes (e.g. μ∂u


! /∂y → 0). However, experimental studies
disagree on an exact value for both the von Kármán constant (!κ ≅ 0.39 – 0.41) and B
! (5 – 5.5)
[49]. This region is known as the “logarithmic region”, while the region for 5 ≤ y! + ≤ 30 where
both viscous and turbulent stresses are significant is known as the “buffer region”. Together
they compose the “inner region”, with which experimental studies for duct flow also tend to
agree [43].

!13
1.4.3 Airflow Within an Engine Cylinder

As opposed to boundary layer flow over a flat plate, or within ducts, the gas phase flow
within an engine cylinder is highly complex. Boundary layers which form at the cylinder liner
are influenced by the three-dimensional and transient nature of the airflow, which varies cycle-
to-cycle (intake through exhaust) as the piston and valves move. Due to this, experimental
studies generally present the velocities and turbulence statistics in terms of the mean piston
velocity (!Vp) and crank angle (!Θ) [23, 50, 51]. While the exact airflow behaviour is unique to
each engine design and operating conditions, in general engine flows are characterized in terms
of tumble, swirl, and squish flows [23, 50]. The charge (volume of air), which begins as jet flow
from the intake valves, has a rotational character with an axis of rotation at a relative angle to
the cylinder axis. During intake, tumble flow occurs when this angle is approximately 90°, while
swirl flow occurs at 0°. The exact nature of this flow depends on the valve geometry and the
interaction between the valve jet flow and the cylinder liner [23, 50]. Squish flow occurs during
compression, where the cylinder volume decreases and the general direction of airflow is
rotationally inwards (depending on piston geometry) [23, 50].

In-cylinder velocities are generally several times higher than the mean piston velocity. For
example, ensemble velocities (averaged over numerous cycles) near the cylinder head (within 2
mm) exhibit velocities well within ±Vp (Vp ≈ 2.3 ms-1) [23, 50]. These velocity measurements
were shown to have good agreement with the law of the wall within the laminar sublayer.
Farther from the cylinder head, ensemble velocities may reach values within ±4!Vp (!Vp ≈ 2.8
ms-1) [51]. However, measurements near the cylinder liner of ensemble tangential velocities at
half the clearance height were found to reach between 4Vp – 6Vp (Vp ≈ 4.6 ms-1) at 90% of the
bore radius (4 mm from the liner), tapering to zero at the cylinder axis and reaching zero at the
cylinder liner [52]. The clearance height is between the piston top and cylinder head once the
piston reaches top dead centre (TDC), where TDC is the position of the piston when it is
highest in the cylinder. TDC is therefore also when the piston is farthest from the crankshaft, at
the end of am upwards stroke. The turbulence intensities near the wall were found to reach
between 0.2!Vp – 0.4!Vp in the same region (10% of the maximum) [52].

!14
1.5 Lubricant Films on the Cylinder Liner

Lubrication via engine oil reduces the friction between the piston rings and the cylinder
liner. In the process, a film of oil is deposited on the cylinder surface as the compression ring
(top ring) sweeps along it from top to bottom dead centre. A minimum oil film thickness (!hmin) is
present between the top ring and the liner, and is ultimately responsible for controlling the
thickness of the deposited film, h! [53]. The minimum oil film thickness is also directly related to
the frictional forces acting between the piston and liner, and is therefore critical to engine
performance [3, 54, 55]. The geometry of each ring, the circumferential difference in ring
pressure (between the thrust and anti-thrust sides of the piston), and varying engine conditions
each have an impact on the oil available stroke-to-stroke at the top ring, and results in an oil
film of nonuniform thickness [53, 56, 57]. The oil film thickness is relevant to its evaporation
because of the fundamental impact on mass diffusion through the liquid to the surface where
evaporation occurs [19, 35]. The height of the film also affects the relative scale of waves at its
surface under shear flow, which may significantly impact the boundary layer flows above it [58,
59].

Measurements of h! or h! min are accomplished in-situ by means of capacitance [60, 61]


and ultrasonic [3, 55] sensors which incorporate the piston rings (measuring h! min ) or optical
measurements through the cylinder liner leveraging laser fluorescence [54, 56]. Measurements
by Furuhama et al. [60] via top ring capacitance sensors give a range of h! min ≤ 8 μm for a
139.7 mm × 152.4 mm (bore × stroke) cylinder. Similarly, measurements using local
capacitance sensors by Söchting and Sherrington [61] yielded h! min ≤ 10 μm for a 95 mm × 127
mm cylinder. However, measurements by ultrasonic sensors in an experimental rig simulating
the contact between the liner and piston rings for various loads found a lower limit of h! min ≤ 3
μm [3, 55]. In addition, laser fluorescence measurements by Dearlove and Cheng [54] in a
similar reciprocating test rig found h! min ≤ 3 μm for various engine speeds and loads. Similar
measurements at an elevated liner temperature (80°C versus 30°C) by Takiguchi et al. [56]
determined h! min ≤ 2 μm at both the thrust and anti-thrust sides of the piston, with a resulting h!
< 1 μm after each stroke. These measurements serve more as typical values than deterministic
ones due to the variability in lubricants, part geometries, conditions, and measurement
techniques.

!15
Models which directly model the heat transfer and oil film dynamics between the piston
ring and cylinder liner provide values for both the oil film temperature and thickness. For
example, Harigaya et al. reported values of h! ≤ 12 μm depending on temperature and
lubricant viscosities [53, 57] for liner temperatures between 80°C < T! s < 120°C depending on
engine speed and decreasing with distance from TDC. Markedly higher values for h! (5 μm < h!
< 22.5 μm) were calculated at low film temperatures (!Ts = 30°C) [57], which emphasizes its
importance to the rheological performance of the oil aside from its importance to evaporation.
The cylinder liner temperature cover a large range, with a familiar variation with engine design
and operating conditions. Direct measurements of the cylinder liner temperature provides a
range of 90°C < T! s < 200°C [5, 18, 60] with calculated values from liner film evaporation
models falling within the same range.

1.6 Scope and Outline of Thesis

The focus of the present work is to establish the evaporation rates of a common engine
lubricant (5W-30) under engine-like conditions, and in a form similar to that of a wall-film
found on a cylinder liner. The evaporation of the lubricant is studied as a thin film under
various boundary layer conditions at an elevated temperature (150ºC). In particular, the bulk
velocity of the flow is varied while the shape of the boundary layer is controlled. The effect of
near-wall turbulence is also explored by varying the turbulence intensity within the boundary
layer. The required gas flow is generated through use of a wind tunnel designed for this study.
The evaporation measurements are conducted within the test section of the wind tunnel, where
the geometry of the test section is adjusted in order to produce the varying boundary layer
conditions. The gas flow conditions are validated through the use of particle image velocimetry
and hot-wire velocimetry measurements.

The following are the objectives of the work presented in this thesis:

i. To validate the flow conditions through the use of PIV and hot-wire measurements.

ii. To study the effect of gas velocity and boundary layer shape on the evaporation of a
discrete thin liquid film of engine lubricant at an elevated temperature.

!16
iii. To study the effect of near-wall turbulence on the evaporation of the same.

Chapter 1 of this thesis has established the context for the evaporation of thin liquid oil
films, including: the impact of oil consumption on engine emissions; the relative contribution
of wall-film evaporation to oil consumption; typical engine conditions; and the generation of
the film at the cylinder liner. Chapter 2 will present a review of literature relating to the
evaporation of hydrocarbon liquids, including the evaporation of bulk volatile liquids under
atmospheric conditions and the implications of direct modelling of the wall-film. Chapter 3
discusses the validation of the wind tunnel and channel flow, which includes the validation of
the flow within the evaporation test section (incident upon the film). Chapter 4 explores the
evaporation results. Finally, Chapters 5 and 6 will discuss the conclusions drawn from this study,
and recommendations for any future work, respectively.

Several appendices are also included which contain additional relevant material.
Appendix A provides an overview of the Blasius laminar boundary layer solution for flow over
a flat plate. Appendix B shows technical drawings detailing the geometry of the designed wind
tunnel flow conditioner. Lastly, Appendix C compiles the full results of the PIV measurements
within the evaporation test section described in Chapter 3


!17
2 Literature Review
The following section contains details of the literature reviewed with respect to the
evaporation of engine lubricant films under airflow. This includes a review of experimental
studies for both the evaporation of volatile, liquid species under turbulent boundary layers and
the behaviour of liquid films in two-phase flow, both of which are relevant to the evaporation
of engine liner films. Experimental studies relating to the evaporation of multi-component
hydrocarbon mixtures are also reviewed, which differ in behaviour compared to single
component substances. Finally, modelling studies of the evaporation of cylinder wall films are
reviewed, which are generally informed by both experimental and nominal engine conditions.

2.1 Evaporation under Boundary Layer Flow

The evaporation of single species liquids in air, including pure hydrocarbons, is a


relatively simple phenomena limited by diffusion above the liquid surface. In general, mass
transfer is governed by a concentration gradient present above the liquid. This process is in
turn affected by convection, which in boundary layer flow (forced convection) acts to increase
the concentration gradient by carrying vaporized species downstream; in this way, a mass or
concentration boundary layer forms [19, 35]. In 1983, J. Schröppel and F. Thiele [62] provided
an excellent example of the familiar, fundamental problem of boundary layer flow over a
wetted flat plate, calculating the momentum, mass, and temperature boundary layers
numerically. In contrast with forced convection, under natural convection – where the inertial
movement of the fluid is governed by buoyancy – the evaporation rates for pure hydrocarbon
liquids are well-defined by one-dimensional diffusion models, as in a 2009 study by P. L. Kelly-
Zion et al. [19, 63].

Among the first experimental considerations of this model problem was conducted by F.
Pasquill [64] in 1943 to extend the theory proposed earlier by O. G. Sutton [65]. In 1934,
Sutton [65] proposed an evaporation model from free liquid surfaces of discrete length under
turbulent boundary layers based on vapour transport through momentum transfer. Pasquill
[64] extended Sutton’s theory by introducing molecular diffusivity as opposed to purely the

!18
mechanical action of eddies. This provided a revised theory with much better agreement to
Pasquill’s wind tunnel evaporation measurements, and explained the disparity between previous
measurements for different pure substances where no consideration was given to the
aerodynamics [64]. For a given pure substance, it was shown that evaporation rate increases
with increasing mean air velocity. A 1952 study by Davies and Walters [66] further extended
Pasquill’s experiments further for longer aspect ratio plane areas, and showed good agreement
with Pasquill and Sutton’s two-dimensional solution [64, 65]. In 1997, C. H. Huang [67]
provided an overview of the Pasquill-Sutton theory’s applicability, showing that the evaporation
in these problems is dependent on the -2/3 power of the Schmidt number as opposed to simply
the kinematic viscosity, !ν or molecular diffusivity, !D:

ν
!Sc = (2-1)
D

Pasquill [64] alluded to this dependence on both the molecular diffusivities and viscosity,
although the molecular diffusivity was considered dominant. This is a well known result, as
correlations for mass transfer coefficients, or the Sherwood number (!Sh), for convective mass
transfer are known to follow Sc
! and Re
! , similar to how Nusselt number (!Nu) correlations with
the Prandtl number (!Pr) instead of Sc
! [19]. However, Huang [67] also suggests a model which
accounts for an “interfacial sublayer” (i.e. the viscous sublayer), where there exists a linear
velocity profile directly above the liquid surface and molecular diffusivity dominates.

While the evaporation rate dependence on the vapour’s molecular diffusivity and the
mean air velocity was shown by Pasquill [64], the notion that the evaporation rate depends
more fully on molecular diffusion within the laminar region as suggested by Huang [67] was
also explored by R. Reijnhart et al. [68]. In 1980, Reijnhart et al. [68] presented a theoretical
approach to modelling free surface evaporation under smooth boundary layers. The simplified,
two-dimensional model for the vapour concentration, !ci, was given as:

{ ∂y }
∂⟨ci⟩ ∂ ∂⟨ci⟩
!⟨u⟩
∂x
=
∂y ( Dm + Dt,y) (2-2)
νm
if y ≤ yvt
!Dm = Sc m (2-3)
0 if y > yvt

!19
{uτ κ y if y > yvt
0 if y ≤ yvt
!Dt,y = (2-4)

where D
! m is the molecular diffusivity and D
! t,y is the turbulent diffusivity normal to the surface.
Furthermore, the velocity profiles for the smooth boundary layer case were given as,

u τ2 y
ν
if y ≤ yvt
!⟨u⟩ = (2-5)
ln ( ) + 5.1uτ if y > yvt
uτ uτ y
κ ν

which is equivalent to equations (1-13) and (1-14) given in §1.4.2 where the viscous sublayer
extends to a height y! νt and the buffer region is neglected. The goal of Reijnhart et al. [68] was
to provide an improved model for where the wall prior to the liquid leading edge has significant
roughness (!y0). Future experimental measurements by R. Reijnhart and R. Rose [69] meant to
verify the applicability of the previous model [68] showed that the roughness transition cannot
be neglected, as artificially higher evaporation rates for a given shear velocity result. In
addition, where the evaporation rate was found to be lower with increasing roughness, while
neglecting the transition yields the opposite trend. However, excellent agreement to the smooth
boundary layer model was found in a smooth wall case (!y0 ≈ 22 μm for a 25 × 25 × 5 cm pool),
which implied that the evaporation was governed significantly by molecular diffusion in the
laminar sublayer, near the gas-fluid interface. In each case, evaporation rate was shown to
increase with increasing u! τ , and also to increase rapidly as the liquid temperature approached
its boiling point and the interfacial vapour pressure increased [69].

While the above studies generally considered plane (or smooth) liquid surfaces, in reality
surface waves form on a liquid pool or film under shear from turbulent boundary layer flow.
The wavy nature of liquid films along a flat plate under turbulent flow was investigated by S.
Wittig et al. in 1992 [58]. The goal of Wittig et al. [58] was to model the heat, mass and
momentum transfer of the two-phase flow within pre-filming airblast atomizers, which was
modelled and experimentally verified using water flow along a flat-plate (60 mm × 340 mm)
within a duct (wide 60 mm × 4 mm) at environmental temperatures (!Tg ≤ 40°C), high gas phase
·
velocities (U ≤ 120 ms-1), and low film volume fluxes per unit breadth, B (Vf /B ≤ 1 cm2s-1).
When laminar liquid flow was assumed the model matched experimental data well under fully

!20
developed turbulent flow, resulting in low average film surface velocities (⟨uf ⟩ < 2 ms-1) and
average film thicknesses (!⟨h f ⟩) less than 150 μm, depending on gas phase Reynolds number.
Wittig et al. [58] posited that the film remains laminar for the velocity ranges studied, and that
the waviness of the film surface does not increase the turbulence levels of the gas phase flow.

The laminar nature of the liquid phase flow was tested in 1994 by Himmelsbach et al.
[59], which extended the previous study by Wittig et al. [58] with the same apparatus for higher
gas temperatures. Himmelsbach et al. determined similar behaviour, where the film flow was
predominantly laminar. Interestingly, this was despite the average film height being larger than
the laminar sublayer for most liquid flow rates [59]. However, in both studies the laminar
·
nature of the film was questionable at higher V! f /B , and it was observed that for those
measurements the film became unstable, meaning droplets begin to tear from the film surface
[58, 59]. This occurred for Reynolds numbers based on the local film height within
! f = ⟨uf ⟩⟨h f ⟩/νf ≈ 200 – 250.
Re

2.2 Multicomponent Mixture Evaporation

The evaporation of multi-component hydrocarbon mixtures is significantly more


complex than for pure substances. For pure liquids, the equilibrium vapour mole fraction at the
liquid-air interface is governed by its vapour pressure at a given temperature [19, 35]. However,
for multi-component mixtures each species is evaporating simultaneously while their vapour
pressures (!pi) depend directly on the mole fraction (!xi = ni /ntot) of each species in solution, as in
Raoult’s Law for ideal liquid mixtures:

!pi = xi p*
i (2-6)

where p*
i
is the vapour pressure for the i th pure species. While the mixture may be initially
homogenous, each species may have different rates of evaporation, depending on their initial
proportions and p*
! i . Consequently, the concentrations of each species at the interface will vary
with the progress of evaporation, leading to concentration gradients within the liquid phase
and finally transport of species to the surface via diffusion (in the case where the liquid is static)
[19]. The resulting variation in evaporation rate (of each species) with time was demonstrated

!21
effectively by Okamoto et al. [70] for several hydrocarbon mixtures with two, three, and five
components. In 2010, Okamoto et al. [70] determined that the varying vapour pressures and
evaporation rates of each mixture were predicted well by a model based on Raoult’s law with
activity coefficients, which account molecular interactions in the liquid phase which leads to
non-ideal behaviour. The evaporated mass for the five-component mixture increased
logarithmically with time, while the overall evaporation rate was exponentially decreasing;
similar behaviour was demonstrated by each other mixture [70]. The evaporation rates in each
case increased with temperature [70], primarily due to the increase in vapour pressure with
temperature. This was demonstrated in an earlier study by Okamoto et al. [71] in 2009 for
gasoline, in a similar study where the evaporation rate of gasoline was shown to also decrease
exponentially with weight loss fraction [71].

The evaporation of multicomponent hydrocarbons has also been studied under


convective mass transfer conditions. Specifically, several studies focus on establishing
evaporation behaviour under environmental conditions, meaning moderate gas phase velocities
and temperatures. In 1973, Mackay and Matsugu [72] provided an evaporation model for
shallow pools (19 mm) based on a mass transfer coefficient calibrated by the evaporation of
water and cumene. The resulting correlation was given in terms of bulk velocity, pool size, and
Schmidt number. The evaporation of gasoline was also considered, each under the same
simplified, atmospheric velocity conditions given by:

n
!⟨u⟩ = U1 y 2 − n (2-7)

with U
! 1 being the velocity at 1 m of height and n! = 0.25 – 1.0, where n! = 0.25 yields the 1/7
power law profile for fully developed turbulent pipe flow. The evaporation rate of the mixtures
was found highly dependent on “liquid mass transfer resistance” (diffusion) and the species
vapour pressures, consistent with the work of Okamoto et al. [70]. Mackay and Matsugu [72]
also posited that the higher evaporation rates relative to flat plate correlations is due to waves at
the liquid surface [58, 59], where their correlation predicted evaporation similar to smooth flat
plate correlations under turbulent flow.

An additional study by M. Gerendas and S. Wittig [73] in 2001 further explored the
behaviour of a multi-component mixture of ethanol and water (for 1:0, 1:1, 1:4, and 1:3 mass

!22
ratios, respectively) in the same apparatus for an elevated gas temperature (!Tg = 200°C). In this
case the film thickness decreased markedly with axial distance due to evaporation (for each
·
!Vf /B case). While “partially turbulent” velocity profiles were required to match the
experimental results for higher film thicknesses, a laminar profile (!u + = y +) was sufficient at
·
lower film thicknesses (unspecified). Also, for the same gas phase velocity, cases with higher V
! f /B
maintained nearer the initial concentration of ethanol over the length of the plate. This implies
that for the cases where the film thickness decreased rapidly the ethanol evaporated in higher
proportion than water, indicating an expected lower resistance to the more volatile component’s
diffusion [19]. Additionally, a reduced concentration gradient of ethanol in the gas phase was
observed for an elevated pressure (!po = 260 kPa versus 110 kPa). In each of the above studies by
S. Wittig [58, 59, 73] the inlet turbulence intensity of the fully developed turbulent flow was
measured to be 5% (centreline location assumed).

In 1997, M. Fingas [74, 75] conducted experimental evaporation studies for numerous
hydrocarbon mixtures at low temperatures (21 – 25°C), stagnant gas or low bulk airflow
conditions (0 – 2.5 m/s), and large pool sizes. Logarithmic or hyperbolic evaporation behaviour
was observed, roughly depending on the number of major species; as the concentrations of
more volatile species depleted, the overall evaporation rate dropped. Higher rates of
evaporation were observed under airflow while differing little with velocity [74, 75]. This lead
to the conclusion that the evaporation was not boundary layer regulated [74, 75]. Boundary
regulation is where the downstream evaporation would be limited by the increasing mass
fraction of hydrocarbons at the liquid-gas interface. Fingas [74-76] posited that boundary layer
regulated evaporation would increase with air velocity and turbulence, while non-regulated
flows would reasonable for high vapour pressures (low temperatures). In 2004, Fingas [76]
presented empirical models with respect to temperature and time for the evaporation of many
hydrocarbon mixtures (oils) based on previous results [68, 69] each demonstrating logarithmic
or square root behaviour. As expected, the evaporated mass increased was shown to increase
with temperature, although the experimental temperature range was low (!T ≤ 60°C).

!23
2.3 Cylinder Liner Evaporation Modelling

The evaporation of engine oil from the cylinder liner has been the subject of several
modelling studies. Modelling is a useful tool for studying an otherwise complex problem; the
liner evaporation process incorporates not only a vaporizing multicomponent film under shear
flow but also the transient behaviours of engine temperature, pressure, airflow, and oil film
generation. Each study [5, 20, 77-80] generally includes some empirical formulae or
experimental engine data to simplify or verify the modelling process. Simplified cylinder
geometry is also used, although wall film sub-models are employed as part of numerical engine
simulation codes such as KIVA [81, 82].

The convective mass transfer from the liquid phase is predominantly solved one-
dimensionally through a heat and mass balance at the liquid-gas interface. In 1992, S.
Wahiduzzaman et al. [83] proposed a one-dimensional method whereby a mass boundary layer
thickness (!δd) is calculated based on a heat-mass transfer analogy (!Sc = (δ /δd) = 2.5). This
3

length is then used in a solution to the diffusion equation [19], given as:

Dρg
( 1 − Ys )
Y − Y∞
!m· = log 1 + s (2-8)
δd

The temperatures and pressures required to calculate the thermodynamic properties for each
species were then determined via a one-dimensional heat transfer model based on a separate
engine simulation code for the gas phase velocity, temperature, and heat flux to the liner. A
separate “ring pack” simulation code was used to provide the film thickness distribution in
conjunction with an engine simulation code, allowing computation for each part of the engine
cycle along an axial discretization of the liner. A similar procedure was employed by Audette
and Wong [79, 80] in 1999, again using an external engine simulation code and film thickness
distribution model. Distillation curves for two 15W40 oils were used to determine the virtual
distribution of n-paraffins composing the modelled liquid. A Sherwood number correlation was
used to calculate the relevant mass transfer parameters of the form:

gm L
!Sh = = 0.035Re0.8Sc0.667 (2-9)
ρD

!24
where gm is the mass convection coefficient (kg m-2 s-1).

In 1997, De Petris et al. [78] presented an approach similar to Wahiduzzaman et al. [83],
yet limited to single component liquid films (C24H50, C27H56, C30H62). In relatively simplistic
approach, an instantaneous convective heat transfer coefficient (!h) at the liquid-gas interface
was provided by a correlation given by G. Woschini [77] in 1967. Woschini’s correlation, which
empirically includes radiative heat transfer, is given by:

!h (Θ) = 3.26Be−0.2 pe(Θ)0.8Tg(Θ)−0.55we(Θ)0.8 (2-10)

where p! e is the in-cylinder gas pressure and w


! e is the average cylinder gas velocity. Both p! e and
! g are provided by De Petris et al. [78] with respect to crank angle (!Θ), while w
T ! e is an empirical
function. In 2009, Harigaya et al. [20] performed another study on the evaporation of single-
component hydrocarbons using a model which employed both Woschini’s [77] correlation and
experimentally measured engine conditions. In addition, the model’s predictions were verified
by measured oil consumption data. A “thermal-hydrodynamic lubrication model” [53, 57] was
subsequently used to calculate the oil film thickness based on the model results. In 2017,
Soejima et al. [5] extended Harigaya et al.’s [20] model to include multi-component liquid
films. Measured oil consumption and engine condition data for a supercharged diesel engine
was used to inform the model. The model was implemented for films composed of a
distribution of n-paraffins estimated by comparing vapour pressures measured for test oils (SAE
10W30, SAE 30) with vapour pressure-temperature correlations for CnHn+2 [5].

Each model agreed that the oil film evaporation rate depended strongly on the film
temperature and oil composition, and weakly on the oil film thickness due to the time scale of
each cycle [5, 20, 77-79, 83]. Soejima et al. [5] also conclude that evaporation rates decrease
with increasing cylinder pressure, due to the phenomenon’s reliance on vapour pressure.
Similarly, each model agreed that evaporative losses are highest during the intake stroke and
lowest during the expansion stroke due to the elevated pressure [5, 20, 77-79, 83].

The evaporation of wall films is also considered by sub-models which model spray-wall
interactions and film dynamics as part of numerical engine simulations, such as KIVA [81].
Modified wall functions are used to model the interaction between the vaporizing film and the

!25
boundary layer flow, as part of modelling the near-wall region during simulation. O’Rourke
and Amsden [81] employ a mass vaporization term as part of source terms in the gas phase
conservation, momentum, and energy equations. This vaporization term incorporates separate
functions for the laminar and turbulent regions within the boundary layer. Foucart et al. [82]
use a similar method in their model which only considers evaporation with respect to a
conservation source term. As opposed to O’Rourke and Amsden [81], Foucart et al. [82] do not
consider turbulent kinetic energy or a turbulent Schmidt number. Instead, two source terms are
employed depending on whether inertial or thermal effects are dominating mass transfer.

2.5 Gap In Literature

Studies which experimentally measure the evaporation of hydrocarbon mixtures under


boundary layer conditions frequently do not include a rigorous description of their airflow
conditions. Additionally, measurements do not consider boundary layer velocities similar in
order of magnitude to those in-cylinder during typical engine operating conditions (§1.5.3).
The effect of turbulence is also typically restricted to fully developed turbulent flow, or in
modelling heat-mass transfer analogies. In realistic cylinder flow, turbulence is present in thin
boundary layers near the liner [52], with turbulence generated by a combination of liner
boundary layer generation and via the intake valve jet flows [23]. Since the mass transfer is
predominantly controlled by the nature of these boundary layer flows, providing experimental
measurements of the oil species evaporation for similar length scales and at an elevated
temperature is valuable.

2.6 Approach

Evaporation of engine lubricant from a liquid film at the cylinder liner is synonymous
with, if not equivalent to, evaporation from a plane liquid surface under boundary layer flow.
In each case a free liquid surface is present beneath a boundary layer while the transport of the
evaporating species is theoretically encouraged by any turbulence present within the
concentration (mass) boundary layer, if it is of sufficient thickness [19]. Therefore, the film

!26
evaporation rates can be characterized by subjecting discrete samples of oil to boundary layer
flows for specific durations.

Based on the foregoing literature analysis the purpose of this study is therefore to
determine the relative importance of boundary layer shape, near-wall turbulence intensity, and
Reynolds number on the evaporation of discrete film samples. To fulfill this purpose the
evaporation of lubricant films of a known initial thickness and temperature are measured after
exposure to laminar and turbulent boundary layers with varying velocity and turbulence
intensities. Evaporation measurements from liquid samples using an analytical microbalance
under steady flow [68, 69] and for discrete samples [63, 70, 71, 74-76] have been successful.
Since a microbalance was available, this gravimetric measurement technique was chosen.
While cylinder film thicknesses are generally less than 10 μm (§1.5) initial film thicknesses of 50
μm were chosen to improve measurement accuracy. Finally, a constant oil temperature typical
of liner temperatures near TDC was chosen for all experiments, where evaporation and
cylinder temperatures are highest [81, 83].

It was decided that the evaporation measurements should be performed within a wind
tunnel, which lends itself to the study of discrete liquid samples under boundary layer flow
(§1.4.1). This necessitated designing, building, and validating an appropriate flow conditioner
and test section, which is detailed in the following section along with the accompanying PIV
and hot-wire analysis. Within the test section, a shallow (< 0.5 mm) inset in a copper plate was
chosen to hold the oil film samples. As Reijnhart and Rose [69] observed that change in
evaporation rate with pool height (relative to vessel edge) was near negligible, this was
considered reasonable. Despite the 50 μm film samples being subjected to relatively high
velocities, it was assumed that there would be no droplet tear-off from the film surface waves
based on the expected film Reynolds number [58, 59]; no film instability was observed during
the experiments.


!27
3 Boundary Layer Measurements
The purpose of this work is to investigate the evaporation of thin lubricant films under
various boundary layer flows, and particularly with varying levels of near-wall turbulence.
Generating appropriate boundary layer flows was therefore necessary prior to conducting the
evaporation measurements. A wind tunnel was designed and built for this purpose, including a
flow conditioner and an extended channel (the test section). Hot-wire anemometry (hot-wire)
was used to judge the performance of the wind tunnel for a range of applicable Reynolds
numbers. The turbulence generated by a number of wire meshes placed at the inlet of the test
section was also measured using hot-wire, to determine how the turbulence intensity might be
controlled within the duct. PIV was then used to ascertain the development of the velocity and
turbulence fields along the entire length of the duct at the plane of symmetry. An additional
test section for holding and heating liquid film samples was also designed and built (evaporation
test section). After post-processing the initial PIV results, two axial lengths within the duct were
chosen which provide an acceptable range of boundary layer conditions. A second set of PIV
measurements was taken at both locations to validate the flow field within the evaporation test
section for the chosen range of Reynolds numbers and inlet conditions. Multiple planes were
measured within the duct, from two directions, in order to measure each velocity component.
These procedures are detailed further in the following sections.

3.1 Wind Tunnel Design

Traditionally, measurements within a wind tunnel test section are conducted within the
inviscid region of the flow produced by the conditioner, meaning outside the boundary layer
region where velocity gradients are present. This provides a predictable and uniform flow for
many experiments. However, the boundary layers considered in this study are generated along
the walls of the test section duct, with the intent that the lower wall of the duct will
accommodate the liquid film surface coplanar with the lower wall. This is primarily due to
space constraints; a long duct length is required to produce developing boundary layers and
observe the generation and decay of inlet turbulence. Due to space constraints, and to match

!28
preexisting apparatus, the test section cross section was chosen to be 20 × 40 mm2. This
corresponds to a rectangular cross section with D
! h = 26.67 mm. For the initial flow validation, a
duct length of 37.5!Dh was chosen to accommodate development length required for “fully
developed” turbulent duct flow [19].

The goal of most wind tunnels is to deliver uniform flow to the inlet of the test section,
both spatially and in time (minimum turbulence intensity). A flow conditioner is used to
accomplish this, in that it attempts to remove any swirl and non-uniformity from the flow. The
secondary purpose of the flow conditioner is to link the geometry of the flow from the air
source to that of the test section. For this study a preinstalled centrifugal blower (APPL
RB80-4BU) was employed as the air source, connected via a round hose (L = 3 m) from a
separate room within the lab. The maximum bulk airflow to be delivered to the test section was
defined as 50 ms-1 (ReDh = 89,000) based on requirements separate from this study.

The requirements listed above lend themselves to an open-circuit tunnel, due to the
length of the test section, space considerations, and cost [84, 85]. Flow conditioners for tunnels
of this type are generally composed of some common components: a contraction; a settling
chamber; and a diffuser or transition to accommodate the air source geometry. A contraction is
used to increase the velocity of the flow to the desired velocity at the test section, which limits
losses within the contraction. It is also a primary contributor to turbulence reduction and flow
uniformity at the test section. The honeycomb and screens found within the settling chamber
serve a similar purpose, with the primary purpose of the former being swirl reduction and the
latter being reduction of non-uniformities. In this case, both a transition and diffuser were
required to connect the air supply to the settling chamber. The settling chamber geometry and
dimensions are set by the chosen contraction ratio and test section geometry, which both
differed from the blower/hose outlet. Finally, the number, placement and dimensions of the
honeycomb and screens is determined by the required amount of attenuation, and must
perform robustly across the operating Reynolds number range. Drawings of the final wind
tunnel components can be found in Appendix B, while an overall layout is shown in Figure 3-1.

The contraction ratio, C


! , was chosen to be 9 based on recommendations by Mehta [86]
and Mehta and Bradshaw [84]. Large values of c are advantageous, as lower speeds within the
settling chamber improves the pressure drop across the screens/honeycomb and lowers the

!29
Figure 3-1. Overall layout of the flow conditioner, including major changes
between inlet and outlet cross-sections for different components.

Reynolds number based on the wire diameter, Red, which is important to the design of the
settling chamber. Additionally, with high C! the contraction will be more tolerant of
“irregularities” at its entrance due to the high acceleration [87]. However, a larger ratio also
increases the length of the contraction required to avoid separation at the exit, increasing size
and cost, although due to the scale of the equipment this was considered to be a negligible
concern as most of the cost was associated with fabrication.

Due to the relatively high aspect ratio of the tunnel, three-dimensional geometry was
chosen for the contraction. A two-dimensional contraction, with plane walls along one
dimension, produces a very high aspect ratio settling chamber for large values of C
! . A high

!30
aspect ratio settling chamber is problematic when designing the settling chamber, due to both
boundary layer growth affecting uniformity and the required mesh wire diameters. The length
of a two-dimensional contraction will also need to be excessively longer for the same
performance, since it is only expanding in one plane. Morel [87, 88] suggests similar design
techniques for axisymmetric and two-dimensional contractions which aim to produce wall
geometries with minimum contraction length and exit boundary layer thickness while avoiding
separation and ensuring flow uniformity in the test section. The guidelines in Morel’s work
produces wall geometries based on simple matched cubic curves. In both studies [87, 88] Morel
concludes that while the design procedures cannot account for the complexities of the flow in
three-dimensional, rectangular nozzles (with all four walls curved), the design of an
axisymmetric contraction provides a good first estimate of the average conditions for an
equivalent diameter given by D
! eq = 2 A1 /π if the aspect ratio is not too high (unspecified) [87].
This is supported by a brief study of a rectangular nozzle of aspect ratio 1 which was designed
using Morel’s axisymmetric method, where the design wall pressure coefficient was reduced to
account for the expected high pressure gradients in the corners [88]. Deviation from the design
pressure coefficient was measured at the wall centreline along with variation from the centreline
to the corners, as expected [88]. A region of constant pressure along the corners indicated
some separation was occurring, but Morel comments that the effect is likely restricted to small
bubbles at the corners; no low frequency fluctuations or non-uniformities were evident in the
test section. Consequently, the design of the contraction was conducted using Morel’s
axisymmetric guidelines as a first estimate [87].

Yao-xi Su’s [89] numerical, potential flow simulations of three-dimensional contractions


further informed the design of the contraction. The flow analysis was conducted for a number
of geometries, varying the inlet and exit aspect ratios, match-point, length, contraction ratio,
and contour power factor. Each geometry was compared on the basis of inlet and exit flow
uniformity (maximum variation), pressure coefficients, and crossflow amplitude (relative
difference between the corner and wall velocities). As in Morel’s guidelines [87, 88], increasing
contraction ratio was found to improve exit uniformity. Similarly, increasing length was found
to improve performance. Also, the contour match-point was found to be the parameter most

!31
important in optimization due to an opposite effect on inlet and outlet performance (wall
pressure coefficients), which is the case in Morel’s work [87].

Importantly, the affect of test section aspect ratio has limited affect on exit uniformity and
pressure coefficients, where the inlet (settling chamber) and exit (test section) aspect ratios are
the same [89]. The major affect of increasing aspect ratio was shown to be an increase in Su’s
[89] crossflow parameter. An increase in the exit pressure coefficient was also evident, which is
detrimental to performance as it increases the risk of separation. Concurrently, a considerable
difference in crossflow was shown for contractions with different inlet and exit aspect ratios,
with the best performance found for contractions where both ratios are 1. A small improvement
to the exit pressure coefficient and flow uniformity at the inlet and exit was demonstrated for an
inlet ratio of 1 and a narrow exit (~2.5). However, results were not presented exhaustively for
this case (or the exit ratio 2 case), so the aspect ratio of the contraction was chosen to be
constant between the inlet and exit.

The final length of the contraction was increased by 25% from the results of these
axisymmetric calculations. This is based on a recommendation from Su [89], where it is noted
that an increase of 20 – 25% is necessary to ensure similar pressure coefficients and uniformity
compared to axisymmetric contractions. The increase of 25% was chosen due to the overall
positive effect of increasing the contraction length. The final geometry is shown in Figure 3-2
along with the relevant parameters from Morel’s procedure. The final length of the contraction
was L = 107 mm, with A2 = 20 × 40 = 800 mm2 and A1 = 60 × 120 = 7200 mm2. The final
cubic match point was calculated to be !xm = 52.89 mm.

The chosen contraction ratio sets the cross-section dimensions for the settling chamber at
60 mm × 120 mm. The goal of the settling chamber is to make the incoming flow more
uniform. The turbulence intensity of the flow should also be reduced [84], although the
contraction accomplishes both goals [84]. The composition of the settling chamber is generally
a segment of honeycomb followed by a series of mesh screens. The screens within the settling
chamber serve to make the flow more uniform, as higher velocities are locally damped more
than slower velocities as they flow over the mesh. A “honeycomb” in this case is composed of
many cells aligned in the same direction and joined together into a sheet. The cells normally
have the same cross-sectional shape, often hexagonal, and with a length several times their

!32
Figure 3-2. Final contraction geometry resulting from the Morel axisymmetric
design method, relevant design parameters, and equivalent inlet and outlet
geometries.

hydraulic diameter. The purpose of the honeycomb is to remove swirl and transverse lateral
velocity from the incoming flow, which is accomplished by the extended length of each cell.
Mehta [86] suggests that the honeycomb cell length be 7 – 10 times their diameter, with a more
modern simulation study by V. Kulkarni et al. [90] suggesting 8 – 10 for screened honeycombs.
Longer lengths are not recommended due to boundary layer growth and subsequent
turbulence generation within the cells. Additionally, Mehta [86] recommends there be at least
150 cells per settling chamber diameter (smaller than the smallest lateral wavelength of the
velocity variation) if the goal of the honeycomb is turbulence suppression. Since this equates to
cell diameters of approximately 0.5 mm, which would be prohibitive to manufacture or
purchase, it was initially decided to forgo including a honeycomb.

The screen parameters are a more challenging part of the settling chamber design. In
general, each screen is composed of a mesh of wires and can be defined by its porosity, β! ,
which is a ratio of its open area to the total area of the mesh. It follows that porosity is directly
related to the wire diameter (for circular wires), d! , and the centre-to-centre wire mesh spacing,
M
! . For square meshes, porosity is defined as:

( M)
2
d
!β = 1 − (3-1)

!33
The critical dimensions for each screen is wire diameter and porosity. The Reynolds number
based on the wire diameter, Re
! d , should not exceed a critical number or excessive turbulence
! d = 50
generation will occur due to shedding [84, 86]. Mehta and Hoffman [91] recommend Re
! d = 40. Similarly, the porosity must be at least
while Groth and Johansson [92] recommend Re
0.57 to avoid instability [84, 86]. Based on continuity and contraction ratio the maximum
velocity within the settling chamber is 5.56 ms-1. For Red,max = 40, this corresponds to a
maximum wire diameter of !d max = 0.112 mm.

A series of screens of varying porosity and wire diameter are generally used to produce
the required amount of attenuation of the incoming flow non-uniformities [93]. More
simplistically, a series of equidistantly spaced, identical screens can be used. A method for
determining the screen parameters and their spacing in this case is provided by a numerical
and experimentally verified study by P. E. Hancock [94]. Hancock [94] considers a series of N
!
screens with an upstream, streamwise velocity field defined as U
! = U + u, where U
! is the mean
flow velocity and !u is the spatial perturbation defined by:

!u = {u 0}(l,m)sin(l y)sin(m z) (3-2)

Attenuation is defined for a given mode of this incoming flow (based on wavenumbers l and m)
by {u
! N}(l,m) /{u 0}(l,m) , where u! 0 and u! N are the amplitudes of the incoming and outgoing flow
perturbation. Hancock’s study [94] requires specification of the static pressure drop coefficient,
! , shared by the screens and provides the spacing, X! , required for a given degree of
K
attenuation. Estimating the static pressure drop coefficient of a screen may be accomplished by
a number of correlations; in this case a relation given by Groth and Johansson [92] is used
which is based on the solidity of each screen, !σ = 1 − β, and an empirical function f! (Red ):

1 − (1 − σ)2
!K ≡ f (Red ) (3-3)
(1 − σ)2

This correlation is used because Groth and Johansson [92] provide empirical measurements of
f! (Red ) for a number of screens with wire diameters similar to the maximum determined above.
An attenuation to at least 3% (±0.03) was deemed adequate, as any spacing of screens with
! ΣK > 2.5 will provide an attenuation of at least 11% and 3% is the best attenuation supported
by the studies experimental verification. For ΣK
! = NK > 4, Hancock [94] gives the attenuation

!34
for ΣK
! and spacing parameter N
! X /H , where H
! is the duct height, and shows that for a given
level of attenuation there is a minimum N
! X /H required to ensure at least that level regardless
of ΣK
! . Choosing N
! X /H ≈ 0.64 therefore ensures that the attenuation for any given ΣK
! is
theoretically better than 3%, regardless of changes to the screen pressure drop coefficient (due
! d or K
to variations in Re ! , for example). It is important to note that these results are given for the
largest scale perturbation present in the flow, and an assumption is made that the highest
amplitude perturbations belong to the lower wave numbers.

Based on the foregoing specifications of porosity and wire diameter, a wire mesh with
! = 0.043 inch (0.11 mm) and β! = 62.7% was chosen, considering availability. At Re
d ! d,max = 40,
f! (Red ) ≈ 0.9 and K
! = 1.4. N
! = 4 screens was chosen to ensure that ΣK
! > 4 for the operating
range of the wind tunnel. The final spacing for the screens was therefore determined to be X
! =
19.2 mm, based on a tunnel height of 120 mm (the longest tunnel dimension). However, in the
final design a honeycomb was included in order to aid in maintaining the flow direction. An
aluminum honeycomb sheet with cell diameters of 3 mm and a depth of 14 mm was selected
from available materials based on these requirements, and one of the four screen mounting
plates was modified to accommodate the honeycomb. Table 3-1 summarizes the resulting
parameters for different values of Re
! d , from the minimum required flow to the design
maximum, and considering the reduction to three screens. Although with three screens N
! X /H
< 0.64 the increased K
! of the screens with decreasing Re
! d allows for similar attenuation to the
maximum case with four screens. For higher velocities nearer the design maximum it would be
beneficial to machine a replacement screen mounting plate and operate the wind tunnel with
four screens.

A circular-rectangular transition was required to connect the circular hose outlet (1⅛
inch, 28.575 mm inner diameter) of the flow source to the rectangular inlet of the diffuser.
Good performance of both the diffuser and transition required avoiding transitory stall, where
separation periodically occurs at any of the diverging walls. The key parameter for achieving
this for a plane-wall diffuser was the larger half-angle between the diverging walls, and the area
ratio [86]. For the transition, the area ratio and length are of similar importance [95, 96]. The
design proceeded as follows: the outlet size of the diffuser was set by the settling chamber
dimensions (60 mm × 120 mm) while the inlet dimensions needed to match those of the

!35
transition outlet. The transition, changing from a circle to an aspect ratio 2 rectangle, provided
some diffusion with the final height matching the inlet diameter in order to adhere to the design
guidelines by Miller [95]. Based on design guidelines by Miller [95], circular-rectangular
transitioning diffusers operate similarly to conical diffusers of equivalent half-angle, where the
half-angle should be less than 5°. For steady, unseparated flow the inlet radius to length ratio
was therefore set to 8.5 [95] yielding a length of 149 mm and outlet dimensions of 35.052 mm
× 70.104 mm (equivalent half-angle of 4°). The final design parameter was therefore the
diffuser half-angle, chosen to be 6° again based on guidelines by Miller [95] as well as Mehta
[86], where a wire mesh screen was assumed present at the inlet. Wire mesh screens have been
found to have a favourable impact on the stall performance of wide-angle diffusers [97, 98];
based on design guidelines by Farell and Xia [97] the foregoing diffuser benefits from a screen
of arbitrary pressure drop at the outlet. Therefore, screens equivalent to those chosen for the
settling chamber were added to the inlet and outlet (Figure B-7), with no appreciable stall
regime expected [98]. The final length of the diffuser was therefore set to 238.018 mm.

Table 3-1. Screen design parameters for different Red.


! Ū NX ! X uN
!Red !f (Red ) !K !N !Σ K ! !
(m/s) H (mm) u0

5 4 3.1 4.79 3 14.36 0.48 19.2 < 1%

10 8 2.0 3.09 3 9.26 0.48 19.2 < 3%

20 16 1.5 2.32 3 6.95 0.48 19.2 < 5%

50 40 0.9 1.39 4 5.56 0.64 19.2 < 3%

Each component of the flow conditioner was CNC milled from 6061 aluminum alloy to
the specifications outlined in Appendix A. Also, each interior surface of the flow conditioner
was stoned smooth in the direction of flow to remove tool marks. The screen holders (Figure
B-5), honeycomb holder (Figure B-5), and diffuser mounting plates (Figure B-7) were each
designed to accommodate 0.060 in rubber gaskets in order to seal the screen against the facing
surface. The depth of each inset (1.52 mm) is set to the thickness of these gaskets, where the
thickness of the wire mesh (0.14 mm) is accommodated by 10% compression of the gasket.

!36
Accounting for the wire mesh and a deformation tolerance of 0.5 mm, the screen mount
thickness was machined to 19.84 mm in order to match the desired inter-screen spacing of 19.2
mm. Also, as opposed to maintaining the inlet aspect ratio, the width of the transition outlet
was changed slightly to 69.967 mm in order to hold the smaller half-angle to 3°, which was
more convenient to machine (Figures B-8 and B-9).

Finally, pressure taps were installed into the final screen mount (Figure B-5) and the
turbulence grid holding plate (Figure B-2) in order to measure the pressure drop across the
contraction. For the same outlet condition, the pressure drop across the contraction provides a
measure of the flow rate to the test section, and is therefore useful in metering the flow without
the use of other apparatus. Four pressure taps were included in both components and were
positioned at the wall centrelines. At the outlet, the pressure taps are placed prior to any
turbulence grid. Both the high and low pressure tap sets are measured in parallel through
conjoined tubes of equal length by a differential pressure transducer (Omega PX653, 10
inH20).

3.2 Experimental Setup

In order to validate the performance of the wind tunnel, and to characterize the flow
fields for the desired experimental conditions, two flow measurement techniques were
employed. The first of these is hot-wire anemometry, where the voltage required to heat a thin
filament to a constant temperature is measured. When exposed to airflow the heat required
changes with the velocity of the gas; after calibration the velocity may then be interpreted from
the voltage measurements directly. The second technique is particle image velocimetry (PIV).
This technique involves illuminating particles seeded into the gas flow, and rapidly imaging the
particles with a known time delay. After image processing, the movement of the particles in the
field of view between frames allows non-invasive measurement of the velocity field within the
focus plane.

The flow conditions in each case were manipulated through a combination of the air
supply, settling chamber, and turbulence grids placed at the contraction exit (or test section
inlet). Air was supplied to the wind tunnel by a centrifugal blower equipped with a variable

!37
frequency controller. The blower frequency was held constant at 20 Hz while the volume of
airflow to the wind tunnel was controlled by two ball valves which opened to the ambient, with
the remainder flowing to the flow conditioner. The  conditioner then reduced the turbulence
intensity and promoted uniformity in the airflow before it reached the test section (or duct)
inlet. This provided a baseline flow at each Reynolds number (Baseline cases) with a minimum
inlet turbulence intensity. However, the inlet turbulence intensity was finally varied by placing
wire meshes (Grid cases) at the inlet of the duct, which increased the inlet turbulence intensity
compared to the Baseline case where no grid was present. Together, this allowed variation of
the flow Reynolds number and inlet turbulence intensity.

Three Grids were employed in addition to the open inlet case. Each Grid was mounted in
an interchangeable plate (Figure B-2), with an empty plate serving as the Baseline case. These
plates were then held in a machined inset in a separate plate (Figure B-2) attached between the
exit of the contraction and the test section (Figure B-1). The wire mesh dimensions for each
Grid, numbered 1 through 3, are detailed in Table 3-2. As with the settling chamber screens,
each grid was composed of a square mesh of circular wires with diameter d! and centre-to-
centre wire spacing M
! . The wire spacing was chosen such that each grid was a unique
combination of two diameters and two different porosities, β! . The wire diameters were chosen
based on criteria given by Roach [99] for generating near-isotropic turbulence, where the wire
diameter should not exceed 10% of the duct cross-section dimensions. The largest wire
diameter in this case was !d = 0.4 mm, yielding d! /a = 2%.

The test sections employed were also interchangeable, with a number of different ducts
employed depending on the situation, as described below. Regardless of the experiment, the
test section remained rectangular with a 20 × 40 mm cross-section. Several aluminum ducts
were employed, where each duct was CNC machined from 6061 aluminum and stoned in the
axial direction to remove tool marks. A single-piece glass duct was also employed where
ubiquitous optical access was required. A separate test section was also used for the evaporation
measurements, which was designed to hold the evaporating lubricant samples. Mounting plates
fixed at the ends of each duct allowed attachment to the contraction exit. Similarly, mounting
plates at the end of the aluminum ducts allowed placement of the evaporation test section at a
specific axial position.

!38
Table 3-2. Wire mesh dimensions for each turbulence grid.
u′
! rms /U u′
! rms /U
d
! M
! β
! ([25], x! /d = 30, ([25], x! /d = 30,
Case
(mm) (mm) (%) ! Dh = 1.7 ×
Re ! Dh = 3.5 ×
Re
104) 104)

Baseline – – – 0.87 0.68

Grid 1 0.4 1.2 44 5.68 4.97

Grid 2 0.25 1.35 66 7.16 6.11

Grid 3 0.4 2.1 66 8.32 7.21

The major components of the evaporation test section are shown in Figure 3-3,
including: borosilicate glass windows for optical access a copper heater block with two 150 W
cartridge heaters installed, a copper film holder composed of a shallow inset and a reservoir
which are 250 μm and 2.5 mm deep, respectively, and a ceramic insulator plate which allowed
contact between the film holder and heater block while insulating the remainder of the
assembly. The film holder and windows were mounted flush with the walls of the duct. The
inset of the film holder was square with a width of c! = 25 mm. During PIV measurements, the
ceramic insulator plate and copper film holder were replaced with a black, non-reflective
polycarbonate piece which replicates their geometry.

Three experimental setups were created, employing both hot-wire and PIV, in order to
characterize the flow. The shared coordinate system used in each experiment is denoted in
Figure 3-4; regardless of length or construction the test section was always a duct with a 20 ×
40 mm cross-section. The layout of each experiment is shown in Figure 3-5. First, hot-wire
measurements were taken at the exit of the contraction in order to establish the flow uniformity
and inlet turbulence (Figure 3-5a). A duct 200 mm in length was attached to allow access by the
hot-wire probe from the duct exit without allowing the contraction to eject into ambient air.
Measurements were also completed along the duct centreline to establish the variation in
turbulence intensity with axial distance (!x = 0 – 200 mm) for each inlet condition. Second, a
transparent glass duct 1 m in length was attached to allow optical access for PIV measurements
along a significant duct length (Figure 3-5b). The purpose of these measurements was to

!39
Figure 3-3. Major components of the evaporation test section and their assembly.
Also shown are the sample holder inset geometry and coordinate system based
on the inset leading edge.

!40
Figure 3-4. General coordinate system and duct geometry implemented for the
experimental fluids measurements.

establish the development of the flow for a range of potential operating conditions. The goal of
this work is to establish the effect of boundary layer conditions on the evaporation of engine
lubricant films, and so a variety of boundary layer conditions (near wall turbulence intensity,
shape factor, boundary layer thickness, etc.) was desired. Third, PIV measurements were
completed within the evaporation test section shown in Figure 3-3 at axial locations along the
duct where appropriate boundary layer properties were identified (Figure 3-5c). The purpose of
these PIV measurements was to characterize the boundary layer flows incident on the
evaporating liquid films.

The hot-wire measurements were accomplished using a single-component, normal hot-


wire probe (Dantec 55P11) and a constant temperature hot-wire anemometry system (Dantec
Streamline Constant Temperature Anemometry System). The hot-wire voltage was sampled at
16 kHz and low pass filtered at 8 kHz. The corresponding velocities were inferred through

!41
Figure 3-5. Experimental setups for experimental fluids measurements.
!42
calibration. During calibration, static pressure measurements via a pitot tube and voltage
measurements via the hot-wire probe were taken at symmetric locations within the test section
at x! = 0 (!y = a /2, z! = ± b /4 ). A calibration for the hot-wire measurement was then produced
by a fourth-order polynomial fit of the time-averaged velocity data with respect to the time-
averaged voltage:

!u (x) = 20.45v 4 − 129.94v3 + 329.70 v 2 − 381.85v + 166.45 (3-4)

The system used to perform the PIV measurements for the experiments shown in Figure
3-5b and Figure 3-5c included a dual pulsed Nd:YAG laser with a maximum output of 70 mJ, a
1600 × 1200 pixel 14-bit Imager Pro X CCD camera fitted with a Nikon 200 mm fixed focal
length macro lens, and a programmable timing unit. The resulting combination of spherical
lenses, a right-angle mirror, and a cylindrical lens produced a laser sheet approximately 1 mm
thick at its waist. The laser sheet was then introduced into the test section through either the
glass walls of the duct or the glass windows mounted in the evaporation test section. The
camera and laser were mounted and aligned using a set of servo driven traverses which allowed
simultaneous movement in the x! , y! , and z! directions. This ensured that the focus plane of the
camera always remained aligned with the laser sheet, and together remained aligned to the
tunnel geometry during repositioning. Flow seeding was accomplished using a fog machine
located at the blower inlet. This resulted in a uniform seeding of particles 3 μm in diameter.

Within the glass duct shown in Figure 3-5b, the velocity fields in the x! -!y plane at z! = 0
were interrogated at 17 different axial duct locations through PIV: x! /a = 1.5, 2, 2.5, 3.75, 5,
6.25, 7.5, 10, 12.5, 17.5, 22.5, 25, 27.5, 32.5, 35, 40, and 45. The particle images, and resulting
velocity fields, at each location are referred to as lying within a field of view (FOV). For each
FOV, 2000 image pairs were acquired at a rate of 14.7 Hz for statistical analysis of the velocity
fields, which were computed through the cross-correlation of successive image pairs. The
images in all cases were processed in DaVis 8 using an advanced multi-pass method where the
initial and final correlation passes were 64 × 64 pixels with 50% overlap and 24 × 24 pixels
with 75% overlap, respectively. Each FOV was 13.1 × 9.8 mm2, with an image magnification of
0.9, yielding 30,000 velocity vectors with a 0.076 mm vector pitch. The random errors in the
PIV measurements were evaluated using the particle image disparity method [100], with the

!43
associated average uncertainties in the velocity fields estimated to be less than 4.5% and 2%
within the 95% confidence for the near-wall and inviscid core (outside the boundary layers)
regions, respectively.

Detailed PIV measurements were taken in the volume above the film holder within the
evaporation test section, as shown in Figure 3-5c. Two axial locations were chosen for the
evaporation measurements: x! /a = 2 (!x = 40 mm) and x! /a = 45 (!x = 900 mm). These locations
were chosen based on the flow development observed within the glass duct (Figure Bc), which
are detailed in the following sections. With the evaporation test section attached the leading
edge of the film holder was located at each position; for x! /a = 2 the test section was attached
directly to the outlet of the contraction, whereas an aluminium extension duct was machined to
locate the test section at x! /a = 45. In both cases, FOVs in the x! -!y plane were spaced across the
film holder in 1 mm intervals within 0 mm ≤ z! ≤ 9 mm (0 ≤ z! /c ≤ 0.36). For x! /a = 2, two
FOVs were employed due to the presence of a thin boundary layer in the vicinity of duct inlet,
with a scale factor of 120 pixels/mm for each FOV (13.3 × 10 mm2) and a 3.6 mm overlap
between them in order to cover the length of the film, c! = 25 mm. For x! /a = 45, a single FOV
with a scale factor of 62 pixels/mm (25.8 × 19.4 mm2) was used. In each case, 1500 images
were captured at 14.7 Hz in order to produce time-averaged vector fields and statistics. The
resulting vector spacing was 76 μm and 150 μm for the x! /a = 2 and x! /a = 45 cases,
respectively.

Similarly, PIV measurements were also taken for FOVs in the x! -!z plane For both x! /a = 2
and x! /a = 45 FOVs were spaced above the inset surface within 0 ≤ y! /a ≤ 0.5, each with a scale
factor of 62 pixels/mm (25.8 × 19.4 mm2). Specifically, FOVs were located at y /a = 0.025,
0.05, 0.075, 0.1, 0.125, 0.175, 0.225, 0.3, 0.375, and 0.45. As in the x! -!y plane measurements,
1500 images were captured at 14.7 Hz and the resulting vector spacing was 150 μm for both
axial locations.

Additional measurements were taken in the x! -!y plane to verify that boundary layer flow
incoming to the film adhered to the law of the wall. An FOV at z! = 0 covering -0.1 ≤ x′
! /c ≤ 0
was employed, where the bottom wall was unbroken by the sample holder inset. A scale factor
of 120 pixels/mm was present for these measurements, resulting in an FOV 13.3 mm high in
order to fully capture 0 ≤ y! /a ≤ 0.5. The camera was oriented 90° relative to the similar x! /a =

!44
2 measurements in order to orient the longer FOV dimension in the y! direction. The
measurement and processing parameters were otherwise the same as in the other
measurements.

3.3 Flow Characterization

The following section details the results of the PIV and hot-wire measurements described
in foregoing chapters. A summary of the experimental conditions is followed by a validation of
the flow conditioner performance through hot-wire measurements at the contraction outlet
(Figure 3-5a). The PIV measurements within the full-length glass duct (Figure 3-5b) are then
considered with respect to boundary layer development. Finally, the PIV measurements within
the evaporation test section for the boundary layer flow incident on the film samples are
explored (Figure 3-5c).

3.3.1 Experimental Conditions

The foregoing experiments were conducted for three Reynolds numbers based on the
duct hydraulic diameter, Dh = 26.67 mm: ReDh = 10,650 (1.0 × 104), 17,750 (1.7 × 104), and
35,500 (3.5 × 104). Measurements were taken for each combination of ReDh and inlet condition
! Dh = 10,650 cases under Grid 1 and 2, which were omitted. The flow rate
excepting the Re
through the duct was constant in each case, and corresponded to mean velocities of
approximately 6, 10, and 20 m/s. An additional ReDh = 71,000 (7.1 × 104) case is included to
inspect the operation of the wind tunnel with three screens, outside of its original four screen
design.

These velocities were considered suitable for the evaporation experiments based on their
similarity to the range of velocities experienced near the cylinder wall during the compression
and expansion strokes. Under typical engine operation (1000 – 3000 RPM, 80 mm stroke) the
mean piston velocity, V! p , is approximately 2.7 – 8 m/s, with the mean tangential velocity
component near the liner falling in the range of V
! p – 6!Vp [52]. This conservatively places the

!45
near-wall velocities within 3 – 50 m/s. The velocities in real engines obviously considerably, but
for the above conditions these values are approximately representative.

Prior to running each experiment both the blower and wind tunnel were allowed to reach
a steady temperature. In all cases, the test section wall and air temperatures were held at
19±1°C. Table 3-3 below tabulates the various air properties used during post-processing based
on that temperature.

Table 3-3. Dry air properties at 19°C.


T
! ρ
! ν! μ
! k! c! p
!Pr
(K) (kg/m3) (10-5 m2/s) (10-5 kg/ms) (10-5 W/mK) (kJ/kgK)
292 1.211 1.496 1.807 2.561 1.0045 0.709

3.3.2 Flow Conditioner Performance

The hot-wire experiments described above and in Figure 3-5a measured the
instantaneous velocity across the cross-section of the duct at the outlet of the contraction. The
purpose of those measurements was to ascertain whether the flow conditioner produces airflow
of sufficient uniformity and with low turbulence intensity at each Reynolds number under
consideration. Figure 3-6 below shows the resulting velocity measurements at x! /a = 0.5, z! = 0,
from y! /a = 0.05 – 0.65.

Figure 3-6 presents the difference between the measured velocity and the centreline
velocity, with respect to the centreline velocity, U
! c. This gives a quantitative measure of the non-
uniformity in the streamwise velocity across the height of the channel. For each Reynolds
number, the velocity varies between 0% and 0.5% of the centreline velocity from the centreline
to y! /a = 0.15. Nearer to the wall, between y! /a = 0.05 and 0.15, the existence of a boundary
layer is evident. This is expected, since the boundary layer within the contraction will reach
minimum thickness downstream of the vena contracta and will continue to grow from that point.
The vena contracta is the point at which the flow cross-section is minimum, due to abrupt change
in flow direction which exists at the outlet of the contraction. The location of this minimum

!46
Figure 3-6. Perturbation of the streamwise velocity and turbulence intensity for
ReDh = 1.0 × 104, 1.7 × 104, and 3.5 × 104 at the contraction exit.

will change with Re


! Dh as the vena contracta changes size [87, 95]; it is for this reason that the wind
tunnel adaptor plate which holds the turbulence grids is 5/8 inches (15.875 mm) thick, and the
origin of the duct is considered at the downstream edge of the same plate (Figure 3-2).

The baseline turbulence intensity for each Re


! Dh case falls within 1.2 – 1.9% in the inviscid
core flow, which is acceptable [84]. The magnitude of the turbulence intensity is shown to
decrease with Re
! Dh , which is expected [25]. In addition, the turbulence intensity shown in
Figure 3-6 is uniform across the height of the tunnel for each Re
! Dh, with a maximum deviation
of ±0.02%. No irregularities are present in the ReDh = 1.7 × 104 case, which implies either that
operating the settling chamber with three screens instead of four provides adequate attenuation
or any adverse effects are removed by the contraction.

The centreline turbulence intensity produced by each wire mesh Grid was established by
Tuna et al. [25] through hot-wire measurement. The resulting values at x! /d = 30 are listed in
Table 3-2 for ReDh = 1.7 × 104 and 3.5 × 104. This determined the relative magnitudes of inlet
turbulence produced by each mesh and their numbering (1 – 3) in order of ascending
! Dh =
centreline turbulence intensity. Tuna et al. [25] provide a more in-depth analysis of the Re

!47
1.7 × 104 and 3.5 × 104 data-set referenced in this study, with a focus on the development of
the duct flow and the influence of grid generated turbulence.

3.3.3 Duct Flow Measurements

The PIV measurements conducted using the glass tunnel (Figure 3-5b) describe the flow
field within the x! -!y plane at the duct centreline for the entire domain considered in this study (0
< x! /a < 50). The primary purpose of these measurements was to ascertain the development of
the boundary layers which would interact with the oil films during the evaporation
measurements. The goal of this study is to evaluate both the affect of near-wall turbulence, the
affect of boundary layer shape (shear stress and velocity gradient), and Reynolds number on the
lubricant film evaporation. Consequently, boundary layer conditions which allow comparison
of different near-wall turbulence intensities for similar boundary layer shapes (thickness) as well
as comparison of different boundary layer shapes for the same Reynolds number were
required. Simultaneously, it was considered advantageous if no more than two measurement
locations were chosen in order to limit cost and time.

The latter requirement was met by choosing locations within the domain in order to take
advantage of the duct boundary layer growth. Figure 3-7 below demonstrates the evolution of
the boundary layer shape factor with axial distance for the Baseline, ReDh = 1.7 × 104 and 3.5
× 104 cases. For the Reynolds numbers chosen (ReDh > 10,000) it was expected that fully
developed turbulent flow would occur within the length of the duct (!L = 37.5!Dh) for each of
the Grid inlet cases [35]. This requires the flow to transition (or remain) turbulent and finally
reach steady state with axial distance, where the boundary layer thickness reaches the centre of
the duct. For the Baseline inlet condition cases, as in Figure 3-7, it was unknown whether
transition would necessarily occur within the length of the duct. Figure 3-7 demonstrates that
transition occurs for ReDh = 1.7 × 104 and 3.5 × 104 under the Baseline condition, with the
originally laminar flow condition (!H ≈ 2.59) transitioning to turbulent flow (!H ≈ 1.4 – 1.6) by
x /a = 32.5 and x /a = 25 respectively. The Baseline, ReDh = 1.0 × 104 case did not transition

within the length of the duct. However, each Grid case was found to be turbulent for the
majority of the duct length, including the ReDh = 1.0 × 104 cases; while only the Grid 3 case is

!48
Figure 3-7. Shape factor variation of the boundary layer within the x" -"y plane at the
duct centreline ("z = 0) for Grid 3 (red) and Baseline (black) inlet conditions.

shown in Figure 3-7 (red) the Grid 1 and 2 cases can be seen in the work of Tuna et al. [25] for
ReDh = 1.7 × 104 and 3.5 × 104. The shape factor uncertainties are the result of random error
estimates from the post-processed PIV results. The error is proportionally higher when
calculating the integral boundary layer properties over smaller distances (nearer to the error
magnitude).

Based on these results, x! /a = 45 was chosen to contrast the fully developed turbulent
flows, and a single developed laminar case, with the thinner boundary layers present at an
upstream location. In each turbulent case the boundary layer thicknesses (within the x! -!y plane)
reached the centre of the duct prior to x! /a = 45. However, after that point, the centreline
velocity in duct flow have been shown to overshoot their fully developed values prior to settling,
[25]. This also occurs for each of the foregoing turbulent conditions [25], and so the choice of
x! /a = 45 provides the greatest leeway for each case to reach a true fully developed condition
where the centreline velocity is invariant with axial distance.

!49
Finally, to provide comparison between different levels of turbulence intensities near the
film surface, a second measurement location at x! /a = 2 was chosen. This location provided
different levels of turbulence intensity in each Grid case, as well as a minimum intensity case in
the Baseline cases, for each Reynolds number. The boundary layer thickness was also at
minimum closest to the inlet, which provided the greatest contrast in terms of boundary layer
shape to the x! /a = 45 location. However, while the inlet flow was perturbed by the wire meshes
in each Grid case, the flow became more homogeneous after x! /a = 1 [25]. Therefore, x! /a = 2
was chosen, and together with !x /a = 45 satisfied each of the requirements stated above.

3.3.4 Evaporation Test Section PIV Measurements

The PIV measurements above the sample holder at both x! /a = 2 and x! /a = 45 were
conducted primarily to characterize the boundary layer conditions for future evaporation
measurements. In all FOVs the velocity vector fields and statistics were determined through
post-processing of the PIV images, which included the streamwise (!u) and wall-normal (!v, w
! )
velocities. This in turn allowed calculation of the boundary layer parameters and near-wall
turbulence intensities, which are the differentiating features of each convective mass transfer
case considered during the evaporation measurements. A representation of the full PIV results
for both locations is found in Appendix C, while results relevant to each evaporation
measurement are shown in §4.2.

The main differentiating factor between the upstream (!x /a = 2) and downstream (!x /a =
45) boundary layers was the degree of development over the intervening length of duct; each
case save the Baseline, ReDh = 1.0 × 104 case was found to attain a similar shape factor by
x! /a = 45, which indicated that the boundary layers had transitioned or remained turbulent
(!H = 1.4 – 1.6). As shown in Figure 3-8, this also translated to a common boundary layer profile
for the turbulent cases when scaled by the displacement thickness, δ*
! . The laminar boundary
layer exhibited in the Baseline, ReDh = 1.0 × 104 case is shown in Figure 3-8a, where each other
case (Figures 3-8a – 3-8c) demonstrated velocity profiles with shape factors and boundary layer
thicknesses of fully developed, turbulent boundary layers. Similarly, the turbulence intensities
reached similar values for those cases [25], as shown in Figure C-12d – C-20d.

!50
Figure 3-8. Convergence of downstream velocity profiles for each Reynolds
number and inlet condition. Grid 1 and 2 cases omitted for clarity in a).

In Figure 3-8, and in each velocity profile presented in §4.2, the full results from the
sideview measurements are abbreviated to only the z! = 0 plane. However, the velocity fields
were found to vary little over the volume above the sample holder (within 0 ≤ x′
! /c ≤ 1, 0 ≤ z! /c
≤ 0.36) in the region near the wall (!y /a ≤ 0.15), although under the Grid inlet conditions
significant differences near the boundary layer edges were observed due to the presence of the
wire meshes. Figure 3-9 shows the ⟨u⟩
! velocity profiles above the sample holder inset for the
Baseline, ReDh = 1.7 × 104 case, which demonstrates the small variation for the same condition
at each axial location. In the upstream cases, the small variation in the x! -direction (Figure 3-9a)
were due to the relatively small length of the inset versus the entry length at each Re
! Dh, whereas
for x! /a = 45 the duct flow was fully developed and mostly invariant with axial distance (Figures
C-12 – C-20). In the z! -direction (Figure 3-9b), small variation was expected due to the similar
growth rate of the side-wall boundary layers and the aspect ratio of the duct. However, the
effect of the growing side wall boundary layers is evident in the downstream cases as a small

!51
Figure 3-9. Velocity profiles above the sample holder at each axial location ("x /a = 2
and 45), shown over the length of inset ("z = 0) and across its half-width ("x′ /a = 0.5).

decrease in velocity, relative to the z! = 0 case, with increasing distance from the centreline
(Figures C12b – C20b).

3.3.5 Law of the Wall

The additional PIV measurements conducted within the x! -!y plane at z! = 0, for x! /a = 45
just prior to the sample holder were completed in order to determine if the turbulent boundary
layers adhered to the “law of the wall”. Turbulent boundary layers typically follow a linear
relationship under wall scaling within the viscous sublayer (!y + ≤ 5) and a logarithmic
relationship within the logarithmic region (30 ≤ y! + ≤ 200), as stated in §1.4.2. As the behaviour
of turbulent boundary layers within these regimes is supported by numerous experimental

!52
measurements, showing that the “law of the wall” applies to the measured velocity profiles is an
indicator that the boundary layer flow is typical and repeatable.

However, accurately scaling velocity measurements is difficult since this requires an


accurate measurement or estimation of the wall shear, τ! w , in order to calculate the friction
velocity, u! τ . Measurement of the wall shear directly is difficult, generally requiring invasive
methods such as oil film interferometry. Measuring the wall shear from pressure gradient
measurements was also inconvenient, requiring minute pressure taps, small tap spacing, and an
accurate pressure transducer within the expected range of differential pressures. Alternatively,
the wall shear could be calculated by establishing the velocity gradient at the wall from
measurements within the viscous sublayer. This in turn is also difficult, as for high Re
! Dh the
thickness of this laminar region is prohibitively small (y + = 5 → y = 75 μm for uτ = 1 ms-1) so
obtaining measurements (if any) within this region is not guaranteed.

A common indirect method for determining the shear velocity is to find a best fit for the
velocity measurements within the logarithmic region, known as the Clauser method [49]. This
method is problematic because of the uncertainty of which measurements lay within the
logarithmic region and the choice of von Kármán constant, κ! [49]. A similar method is to fit
the data to a model mean velocity profile which incorporates the logarithmic region, buffer
region, and viscous sublayer such as the method suggested by Kendall and Koochesfahani [49].
Kendall and Koochesfahani [49] recommend fitting both zero pressure gradient and favourable
pressure gradient boundary layer measurements to the Musker profile, based on the eddy
viscosity model, given by the solution of:

( y +)2 1
∂u+
κ
+ s
! + = (3-5)
∂y ( y+)2 1
(y +)3 + κ
+ s

In implementing this method, Kendall and Koochesfahani [49] recommend fitting to the
Musker profile through an iterative method which minimizes an aggregate error function, Φ
! ,
! data points (!ui+) and the model (!um,i
based on a fractional difference between the N +
) at the same
!y +:

!53
Figure 3-10. u" + profiles within the x" -"y plane at x′
" /a = -0.1 ("x′ /a = 45) for each
Reynolds number case under the a) Baseline and b) Grid 3 inlet conditions.
Profiles are compared to either the Musker profile (black) or u" + = y + (red).

+ +
1 N ui − um,i
!Φ = ∑ (3-6)
N i=0 um,i
+

Another, less accurate method is to approximate the wall shear using a correlation for the
friction coefficient [49], such as the first Petukhov equation:

τw 1
!Cf = = (0.790 ln ReDh − 1.64)−2 (3-7)
1
ρU 2 4
2

given by Petukhov [35, 101] for smooth, fully developed turbulent flow within ducts (104 < ReDh
< 106).

Figure 3-10 shows the results of fitting the Baseline (Figure 3-10a) and Grid 3 (Figure
! Dh to the Musker profile for κ! = 0.41 and B
3-10b) velocity measurements for each Re ! = 5, and
in the laminar Baseline, ReDh = 1.0 × 104 case to u + = y +. The values of κ and B were chosen
based on common values in each case [38, 43, 49]. In each case, the iterative approach was
limited to data falling under y! + ≤ 200 in order to limit the influence of measurements which
are close to the outer region nearer to the duct centreline, which is expected to deviate from

!54
logarithmic behaviour [38, 43]. In each turbulent case, the measurements closely adhere to the
Musker profile (or linear curve in the laminar case) for their respective shear velocity
estimations, up to the edge of the viscous sublayer (!y + = 5) at the limit of the PIV FOV. In
addition, the shear velocities given in Figure 3-10 are within 5% of the values given by
equation (3-7) for fully developed turbulent flow.


!55
4 Evaporation Measurements
The evaporation behaviour of thin lubricant films under airflow was investigated by
directly measuring the evaporated mass under differing boundary layer conditions with the
liquid oil at an elevated temperature. The evaporation rates in each case were determined by
exposing oil film samples to the measurement conditions for several intervals, which
demonstrated the evaporation behaviour of the oil with time. The evaporation results for the
chosen conditions are explored below, following an outline of the experimental setup and
measurement procedure.

4.1 Experimental Setup

Each film evaporation measurement was conducted within the evaporation test section
described in §3.2. The relevant components are reiterated here along with a more complete
explanation of their functions. These components are also shown below in Figure 4-1, along
with notes on the clearances between parts for thermal insulation. Each oil film sample was
generated within the inset of the sample holder in order to facilitate weighing the oil before and
after exposure to airflow. The sample holder itself was constructed in two parts so that a fin
overhanging the downstream reservoir could be machined. Both parts together are referred to,
here, as the “sample holder”. Any weight measurements and cleaning operations always
included both parts. The removable roof and front wall allowed access to the test section so
that each prepared sample could be quickly and repeatably positioned.

While positioned within the test section the ceramic insulator plate provided both thermal
insulation and mechanical support for the sample holder, preventing it from moving under
airflow and ensuring contact with the copper heating block (heater). The top surface of the
heating block was raised partially above the lower surface of the insulator plate such that the
sample holder contacted the heater while its upper surface was flush with the duct bottom. The
vertical position of the heater was adjusted to account for thermal expansion; once at
temperature, the desired position was verified by a depth micrometer. In addition, the heater
block was wrapped in ceramic-fibre insulation in order to limit heat loss. Two 150 W cylindrical

!56
Figure 4-1. Main components of heated parts within the evaporation test section
and the clearances between them.

cartridge heaters (Omega CIR-1014/120V) inserted into the heater block were used as heat
sources. A stainless-steel sheathed K-type thermocouple (Omega KTSS-116) was also inserted
equidistantly between the two cartridge heaters to measure the heater base temperature, T
! b (the
maximum temperature). Finally, an enclosure was used to safely contain the heater and the
associated electrical wiring; the enclosure is not shown in Figure 4-3.

In order to perform constant surface temperature measurements the heat flux from the
heating block to the sample holder surface had to be equivalent to the heat lost to convection.
To approximate these conditions the heater base temperature was controlled through a tuned
PID controller which was informed by the thermocouple installed at its base. With the base
held at a constant temperature, and under a single set of boundary layer conditions (convective
heat transfer conditions), the surface temperature, T! s , was expected to assume a single, steady-
state temperature. Calibration of T! s relative to T! b was accomplished using a K-type surface

!57
mounted thermocouple (Omega SA3-K-SRTC) attached to the inset surface of a sample
holder clone, which allowed direct measurement of T
! s. The surface thermocouple was mounted
at x′! /c = 0.5, y! /c = 0. These calibrations were performed prior to the evaporation
measurements for each set of airflow conditions, the choice of which is explored below. Each
thermocouple was self-calibrated through an NI 9214 module with provides an estimated
uncertainty of ±0.5°C. The temperature control system itself was composed of an FPGA (NI
cRIO-9074) a digital IO module (NI 9375), a 20 mA variable current supply module (NI 9265),
the multi-channel NI 9214 module, and separate relays for each cartridge heater. Each
calibration was composed of two procedures: first, the required base temperature at steady-
state was identified; second, the time required to heat the cold sample holder to the desired
surface temperature at that base temperature was measured. For the latter, the cold sample
holder was placed on the heater and allowed to heat to the desired surface temperature without
airflow. Once the surface temperature was reached the airflow was applied and any deviation
from the surface temperature was observed. The resulting heating time and base temperature
were then used for each measurement in order to approximate a constant film temperature
during the measurement duration (under airflow).

The oil film mass loss due to evaporation was measured to investigate the evaporation
rate of the oil under specific boundary layer conditions. For each case, oil film samples were
prepared, heated to the measurement temperature, and then exposed to the airflow conditions.
Each sample was exposed to airflow for a specific duration, with a fresh oil sample being
prepared for each measurement from a single, reserved volume of 5W-30 synthetic engine
lubricant (!ρo = 8.3 g/mL). In order to determine the evaporated mass loss each sample was
weighed before and after exposure. The entire measurement procedure was as follows: First,
the sample holder was flushed and cleaned with isopropyl alcohol. Next, the sample holder was
weighed to establish the non-oil mass of future measurements, m
! o . Handling and cleaning the
sample holder, even with sterile gloves, removes oxides at the metal surface and therefore causes
sample-to-sample variation of the copper mass. Weight measurements were accomplished via a
Sartorius analytical balance (MSE125P-100-DU) with a resolution of 10 μg and repeatability
of 15 μg. A draft shield prevented ambient air movement from interfering with the
measurements while care was taken to ensure no other vibrations affected each weight

!58
measurement. Each weighing was repeated 3 times, with the measurement taken only once the
readout was stable for at least 1 minute, with the final weight taken as the average. The
MSE125P self-calibrates upon startup or if the ambient temperature deviates by ±1°C. If auto-
calibration occurred prior to evaporation the measurement process was repeated; if it occurred
post-evaporation the current sample was rejected. Next, a volume of oil was deposited onto the
sample holder using a syringe. The volume of oil deposited corresponded roughly to the film
volume at the desired initial thickness over the 25 mm × 25 mm inset. A ground, stainless steel
pin was then used to spread the oil over the inset. After spreading, the film sample was weighed
to establish the initial weight of the sample, m
! 1. The initial weight of the oil film was then given
by:

!mi = m1 − mo (4-1)

Next, the film was placed into the pre-heated evaporation test section (at the correct base
temperature) and allowed to heat for the time required for the surface to reach the target
measurement temperature. The film was then exposed to airflow by actuating a valve at the
entrance of the flow conditioning unit. After the measurement duration the airflow was
stopped. The sample was then removed and immediately cooled via contact with a large
aluminum block to await re-measurement. Finally, to eliminate bias in the weight due to
buoyancy the final weights of the films, m
! 2, were measured once the samples reached the initial
measurement temperature. The final mass of the oil film was then given by,

!m f = m 2 − mo (4-2)

with the evaporated mass of the oil film being,

!mevap = m f − mi = m 2 − m1 (4-3)

During the measurement process, gravity caused each sample to settle into an approximately
even film, excepting a meniscus around the inset edge; pre-heating within the evaporation test
section also contributed to this due to the reduction in viscosity with temperature [53].
Additionally, the MSE125P remained in the same location for the entirety of the study and in a
separate room from the blower and other apparatus to minimize vibrations.

!59
Prior to any calibrations or measurements the blower was allowed to run for 1 hour at its
operating frequency (20 Hz) with the flow conditioning valve open so that the system could
reach a stable operating temperature (and blower output). Similarly, post-calibration the
heating block and evaporation test section were allowed to heat to the required base
temperature with flow conditioner open for at least 1 hour. In addition, the ambient
temperature of the scale room where each weight measurement was conducted remained at 19
±2°C, which is similar to the air temperature at the test section inlet since both the blower and
room were supplied by the same dedicated air conditioning unit.

Each evaporation measurement was conducted for an initial film temperature, T! f , of


150±2.5°C, which is typical of regions nearer to TDC on the cylinder liner at high engine
loads [4, 5, 22]. Measurements for each forced convection case were conducted for t = 0, 25,
50, and 150 s, where t! = 0 s is the point at which airflow begins after heating. Although the oil
species are stable at low temperatures (< 90°C) [5], evaporation occurred during preheating
from the ambient temperature to T! f . This initial mass loss during heating, m
! evap,t=0 , was
established for each case naturally by performing measurements for t! = 0. However, the
evaporation results in the following section are presented corrected such that m′
! evap,t=0 = 0:

!mevap
′ = mevap − mevap,t=0 (4-4)

The prime is omitted from the notation in the next section. The maximum time exceeds
common exposure times within IC engines (e.g. 30 ms at 2000 RPM) but was necessary to
vaporize a significant amount of liquid and produce reliable estimations of evaporation rates.
In addition, a relatively high film thickness (50 μm versus 5 μm) mitigates the difference in
diffusion times compared to engine timings. An average film thickness of 50 μm was chosen for
all experiments based on the repeatability of the balance, corresponding to a volume of 31.25
mm3 (26.14 mg) within the film holder inset. Consequently, the film mass was constrained to
26.64±0.5 mg during preparation. Table 4-3 below summarizes the boundary layer conditions
chosen from among the characterizations described in §3.6. For each set of conditions, each
evaporation experiment was repeated five times to prevent possible outliers and reduce random
errors in the reported data. Five measurements was deemed an appropriate compromise

!60
between accuracy and time based on an extended study of 10 measurements producing similar
repeatability for the Baseline, ReDh = 1.0 × 104 case.

The results of the calibrations for each set of conditions are also summarized in Table
4-1, including the heater base temperature and preheat time. Infrared (IR) imaging of the
sample holder was used to both measure the temperature variation at the sample holder inset
surface and visualize the oil film motion as it shears under airflow. An optris® PI 640 camera
(640 × 120 pixels, 32 Hz) was used to capture the required radiometric data. Due to the
reflective properties of copper a clone of the sample holder was coated in an emissive, high-
temperature paint to provide surfaces with consistent emissivity and low reflectance. The
emissivity of the surface (!ϵ = 0.85) was determined by mounting a surface thermocouple and
matching the measured temperature to the PI 640 measurements. Figure 4-2 demonstrates the
experimental setup (Figure 4-2b) used to image the sample holder, with the camera oriented
above the test section with the top and borosilicate glass window removed due to its absorption
in the IR spectrum.

Table 4-1. Experimental conditions for film evaporation measurements.


T
! b t! o m
! evap,t=0
!x /a Inlet Case !ReDh
(°C) (s) (10-4 kg/m2)
2 Baseline 10,650 (1.0 × 104) 194 60 2.72

17,750 (1.7 × 104) 208 50 2.72

35,500 (3.5 × 104) 216 40 2.72

Grid 1 17,750 204 45 2.40

Grid 2 17,750 204 45 2.40

Grid 3 10,650 194 55 2.88

17,750 204 45 2.40

45 Baseline 10,650 188 56 1.76

17,750 209 42 2.24

!61
Figure 4-2. IR measurements demonstrating a) the change in film thickness with
time, b) the associated experimental setup, and c) temperature distribution over
the sample holder inset for "T = 150°C ("σ = 1.1°C).

!62
Figure 4-2a shows a visualization of the oil film under the Baseline, ReDh = 1.0 × 104
condition. Due to variation in the film surface, the reflectance/transmissivity of the liquid oil,
and small differences in surface temperature, the IR data reveals the changes in surface profile
as the oil shears over the maximum measurement duration. This is a qualitative, while typical,
measure of the film thickness with time, with the thicker regions coloured in blue. Initially, a
thicker region of oil is found within a meniscus at the film edge while a thinner region is
present at the centre of the holder. As the film shears under airflow it begins to pool at the
downstream edge of the inset, with significant surface deformations reflected in the images
(Figure 4-2a). It can be inferred from the results that the scale and magnitude of surface waves
subside progressively as the film thickness reduces over the majority of the surface, with the
main film volume collected in the aft part of the holder.

Finally, Figure 4-2c demonstrates the even temperature distribution of the clean inset at
steady-state without airflow. The inset surface was imaged with the PI 640 at its minimum
measurement distance (200 mm) with its standard 33° × 25° lens (!f = 18.7 mm), resulting in a
125 × 125 pixel image where each pixel represents a temperature measurement. Once an
average, steady-state temperature of T
! = 150°C was reached 150 measurements were gathered
at 32 Hz. The resulting time-averaged temperature field is shown in Figure 4-2c, shown in the
form of an excess temperature,

!θi, j = ⟨T ⟩i, j − T (4-5)

where i! and j! represent pixel locations in the x! and y! locations respectively and ⟨T
! ⟩i, j represents
the time-averaged temperature at each location. The mean temperature is in turn given by,

n m
∑i ∑j ⟨T ⟩i, j
!T = (4-6)
nm

with a standard deviation of σ! = 1.1°C. The distribution shows that the majority of the sample
holder lies within ±2!σ = 2.2°C, while the accuracy of the PI 640 is reported to be ±2°C.

!63
4.2 Evaporation Results

In this section, the results of the oil film evaporation measurements are presented.
Specifically, the evaporation is compared and contrasted with respect to the flow
characterization for each measurement case detailed in the foregoing section, including
evaporation at ambient conditions under natural convection. The effects of Reynolds number,
near-wall turbulence intensity, flow regime, and boundary layer shape are discussed.

The evaporated masses with time for each case listed in Table 4-1 are shown in Figure
4-3. Furthermore, a full tabulated list of the evaporation results may be found in Appendix D.
Firstly, vaporization under free convection (natural convection) at 150°C (!Grc = 3,838) was
completed to provide a baseline for the oil film evaporation measurements. Figure 4-3a shows
the resulting evaporation within t! = 0 – 600 s. An approximately linear evaporation rate with
time is observed, with m· n = 3.06 × 10-6 kg m-2 s-1. This is expected given the low evaporation
rate, where species concentrations within the film will not decrease substantially such that the
mass transport is limited by liquid-phase diffusion, or dominated by heavier hydrocarbons with
lower vapour pressures [71]. A similar linear regression is performed to approximate the
evaporation rate for each other evaporation case in Figure 4-3. An increase with Reynolds
number is evident for the Baseline cases (Figure 4-3b – 4-3d). This trend is mimicked by the
Grid cases (Figure 4-3e – 4-3h) and downstream baseline cases (Figure 4-3i and 4-3j). Little
! Dh = 10,650 and
variation is evident in the evaporation between Grid and Baseline cases for Re
17,750. Error bars are present in Figure 4-3 and subsequent figures which denotes ±!σ, which is
the standard deviation of the mass measurements. The error bars tend to be smaller than the
symbology used.

Comparisons of evaporation results with respect to their boundary layer conditions are
presented in Figures 4-4 – 4-7. In each case, the evaporation and airflow results are presented
non-dimensionally to assess the relative effects of different parameters. The exposure times (t)
are scaled by the characteristic velocity (!U), and film length (!c), representing exposure in terms
of mean transit across the inset length. For duct flows, U
! is equivalent to the centreline velocity.
The evaporated mass (!mevap) is given with respect to the initial film mass (!mi) for contextual
purposes, as the initial film masses are within 26.64±0.5 mg. The boundary layer profiles

!64
Figure 4-3. Evaporation results and estimated evaporation rates for oil films
initially at 50 μm and held at 150°C for each airflow case, including the natural
evaporation case.

!65
velocities are scaled with respect to (!U), while the wall-normal distance (!y) is scaled by both the
tunnel height (!a) and the displacement thickness (!δ*). The exposure time is also scaled by δ*
! in
addition to c! , with both the mass and velocity δ*
! scaling shown inset in their respective
subfigures. Theroetically, laminar boundary layers in two-dimensional flow have similar shape
when scaled by the δ*
! , described by the Blasius laminar boundary layer solution (Appendix A).
Comparison to the Blasius profile allows evaluation of the boundary layer shape, and so it is
included as a solid black line in the inset of each velocity subfigure. The exposure time scaled
by δ*
! (!U/δ*) is relatable to the shearing rate within the boundary layer and is therefore
proportionate to the shearing rate within the film (! ∼ ⟨uf,o⟩/⟨h f ⟩) for laminar boundary layers.
The same velocity scalings are used in a similar fashion for the presented turbulence statistics
(!u rms) in Figures 4-5 – 4-7 and the PIV results in Appendix C.

A comparison of the evaporated masses under laminar boundary layers with similar
thicknesses serves to demonstrate the effect of shear velocity on the film evaporation. Figure 4-4
provides this comparison, showing the evaporation data (Figure 4-4a) and boundary layer
profiles (Figure 4-4b) for each case under the Baseline inlet condition at the upstream location,
x! /a = 2. In this case the boundary layers are obviously laminar, demonstrated by each profile
collapsing to the Blasius profile (Figure 4-4b). The results in Figure 4-4a show a collapse of the
scaled evaporated masses, implying a dependence on U
! . This also reflects the increase in
evaporation rate with increasing U
! and Re
! Dh demonstrated in Figure 4-3b – 4-3d. Here, an
increase in U
! is associated with a proportional increase in the local wall shear stress (and shear
velocity) due to the laminar nature of the boundary layers. This trend matches with the results
of Reijnhart and Rose [69], where the evaporation rate of a single component liquid phase
increased with increasing shear velocity under turbulent flow.

A different comparison between boundary layer cases with differing levels of near-wall
turbulence intensity demonstrates its effect on the film evaporation. This is shown in Figures
4-5 and 4-6, which show thin boundary layer cases at the upstream location (!x /a = 2) with
varying turbulence intensity maxima due to the wire-meshes (Grids) present at the inlet [25].
Firstly, Figure 4-5 shows a comparison between the evaporation data (Figure 4-5a) and
! Dh = 10,650
boundary layer profiles (Figure 4-5b and 4-5c) for the Baseline and Grid 3, Re
cases. Some perturbations in the core velocity profile are evident for the single Grid case (Figure

!66
Figure 4-4. Comparison of Baseline, Re " Dh = 10,650, 17,750, and 35,500 evaporation
cases at x" /a = 2. Showing a) film evaporation and b) velocity profiles. Subfigures
show same properties with relevant parameters scaled by δ* " . Solid black line in b)
denotes the Blasius laminar boundary layer.

4-5a) due to the proximity of the wire mesh, and so the core velocity U
! is estimated as an
average in the core flow region for each Grid case. This estimation is also used in the scaling of
Figure 4-6 and elsewhere. Despite the difference in u! rms profiles (maximum of 13% vs. 1.5% for
the Baseline case), Figure 4-5a shows no significant differences in the evaporation between the
Baseline and Grid cases (relative to the measurement uncertainty). This is possibly due to the
similarity in δ*
! and U
! between the two cases, with the mass transfer being dependent primarily

!67
Figure 4-5. Comparison of Baseline and Grid 3 cases for Re " Dh = 10,650 at x" /a = 2.
Showing a) film evaporation, b) velocity profiles, and c) turbulence intensities.

!68
Figure 4-6. Comparison of Baseline and Grid 1 – 3 cases for Re " Dh = 17,750 at x" /a =
2. Showing a) film evaporation, b) velocity profiles, and c) turbulence intensities.

!69
on the mean wall shear as in Figure 4-4. Therefore, a significant increase in flow fluctuations
produces a minimal impact on the gas-phase mass transport in the entrance region as long as
the mean boundary layer characteristics are not remarkably different.

This trend is also evident in Figure 4-6, which displays similar conditions of varying near-
! Dh = 17,750. Here, each of the three Grid cases are considered. Similar
wall turbulence for Re
! Dh = 10,650 case, no significant change in evaporation is evident despite notably
to the Re
higher near-wall u! rms (9%, 12%, and 14% maxima for Grids 1 – 3 respectively). Again, this
implies that the level of velocity fluctuations in the boundary layer plays a less significant role
compared to the mean boundary layer characteristics. Based on the results in Figures 4-5 and
4-6 it is likely that the influence of near-wall turbulence can be neglected when modelling
convective mass transfer from the evaporating oil film surface for similar length scales.

The dominant effect of mean boundary layer characteristics are demonstrated further
through comparison between the thin upstream (!x /a = 2) boundary layers and more
developed, downstream (!x /a = 45) boundary layers. Figure 4-7 presents a comparison is
! Dh = 10,650 and 17,750 cases at x! /a = 45, where the Re
between the Baseline, Re ! Dh = 17,750
case has transitioned (see Figure 4-7b inset) and generated significant near wall turbulence
! Dh = 10,650 case has remained laminar yet increased in
(Figure 4-7c). Concurrently, the Re
! Dh =
thickness. As in Figure 4-3 and 4-4, notably higher evaporation rate is seen for the Re
17,750 case. A similar evaporation rate under δ*
! time scaling (Figure 4-7a inset) is
demonstrated in both cases. However, the nondimensional evaporation rate based on δ*
! is
higher than in the upstream laminar cases shown in Figure 4-4a or the turbulent cases shown in
Figures 4-5a and 4-6a by approximately 100% (0.06 versus 0.03). This implies that similarity in
this scaling is dependent on displacement thickness (which is greater in Figure 4-7), and
therefore that the evaporation over the discrete samples is dependent primarily on Reynolds
number and mean wall shear (which is ill-described by !δ* in thick, turbulent boundary layers).

The dependence on bulk flow properties over the length of the film is demonstrated
further by Figure 4-8, which summarizes the effects of flow parameters on film evaporation. In
both Figure 4-8a and 4-8b the forced convection evaporation rates (!m· f) from Figures 4-3b – 4-3i
are scaled by the natural convection evaporation rate (!m· n) derived from the data in Figure 4-3a.

!70
Figure 4-7. Comparison of Baseline, Re " Dh = 10,650 and 17,750 cases at x" /a = 45.
Showing a) film evaporation, b) velocity profiles, and c) turbulence intensities.

!71
Figure 4-8. Evaporation rates from linear regression of forced convection data ("m· f)
relative to natural evaporation rate ("m· n). Shown with respect to a) Re
" c and b) Re
" δ*
for each Baseline and Grid case.

In Figure 4-8a the evaporation rates are plotted with respect to Reynolds number based on the
film length (!Rec) which is the characteristic length based on the film area. As expected, the
evaporation rate increases with Re ! · n as Re
! c , approximately decreasing towards m ! c approaches
zero. Conversely, in Figure 4-8b, the relative evaporation rates are plotted with respect to the
Reynolds number based on the local boundary layer displacement thickness (!Reδ*). In this case,
the downstream cases have significantly different behaviour compared to the upstream cases, as

!72
in Figure 4-7. The results in Figure 4-8a highlight the prominent dependence of the
evaporation rate on bulk flow parameters (!ReDh) or mean wall shear for the appropriate length
scales.

These results agree with the similarity of evaporation rates observed earlier, where those
rates are well defined with respect to convective time scales based on length scales and
characteristic velocities (inset in Figure 4-4a) which better describe the wall shear stress. For
turbulent flows, this is likely the velocity gradient within the laminar sublayer (!y + = y uτ /ν < 5),
as opposed to the displacement thickness. This is further supported by the apparently negligible
effect of turbulence above the film. The thinness of the concentration boundary layer over the
short, discrete samples is likely the culprit for these cases: once diffusion through the sublayer
has proceeded it is expected the near-wall turbulence will encourage the convective mass
transfer of hydrocarbons from the film. Therefore, for the length scales and conditions studied,
it is likely the effect of gas phase transport on the evaporation may be restricted to the viscous
sublayer.


!73
5 Conclusions
In this study the evaporation of thin, liquid oil films was investigated experimentally.
The oil films 50 μm in thickness were raised to an elevated temperature of 150°C and
subjected to constant temperature evaporation under laminar and turbulent boundary layers.
Oil films which are deposited onto a cylinder liner by the compression ring as a piston
reciprocates are subjected to similar conditions; the goal of this study was to evaluate the effects
different boundary layer parameters have on the evaporation under representative
circumstances.

Characterization of boundary layer flows within a duct (!2a × a! ) was completed


successfully through a combination of PIV and hot-wire measurements. Several sets of
! Dh = 10,650, 17,750, and 35,500. Three
boundary layer conditions were chosen therein, for Re
different inlet turbulence conditions were also employed which varied the initial levels of
turbulence within the boundary layers. Evaporation measuyrements were conducted for two
separate locations within the duct (!x /a = 2 and 45) to take advantage of the boundary layer
development and high inlet turbulence. This allowed comparison between laminar, turbulent,
and transitional boundary layers of similar and different thicknesses. PIV measurements at
both the upstream and downstream locations showed that the boundary layers remain
predominantly two-dimensional (negligible variation in z! ) across the film inset with little
variation in the axial direction due to the inset’s short length. Additionally, the fully-developed,
turbulent boundary layers at the downstream location were found to abide by the law of the
wall.

The evaporation of the oil hydrocarbons was found to be highly dependent on bulk
Reynolds number, and demonstrated a linear increase in evaporation with velocity. The
evaporation rate was also found to not depend directly on the boundary layer shape, but rather
on the mean wall shear based on comparison between thin and developed boundary layers.
Near-wall turbulence intensity (maxima of 1.5%, 9%, 12%, 14%, and 16%) were found to
have negligible impact on the evaporation rate for the length scales studied. Consequently, it
was concluded that when modelling cylinder liner evaporation under similar conditions the
effect of convective mass transfer on the evaporation may be restricted to the laminar sublayer.

!74
A wind tunnel was designed to provide a quiet, uniform airflow source at the inlet of the
test section. Hot-wire measurements at the outlet of the contraction showed that the settling
chamber and contraction were operating adequately, with mean streamwise velocity (!⟨u⟩/U)
remaining within 1% of the centreline velocity (!U) for 90% of the test section height (!x-!y plane,
z! = 0) for the lowest Re
! Dh case. Likewise, no instabilities or noise indicative of separation within
the flow conditioner components was detected for the Re
! Dh range considered. The baseline
inlet turbulence intensity remained less than 2% for each !ReDh case.


!75
6 Future Work
The foregoing study presented conclusions with respect to the evaporation of thin,
lubricant films that is by no means comprehensive. Specifically, the actual engine conditions
which cylinder wall films are subjected to are highly transient, variable, and three-dimensional.
Therefore, the following are some suggestions for future work which could follow this work.

Firstly, while this study considered flow tangent to a wall film it would be beneficial to
experimentally study the effect of impinging jet flow on the film evaporation. This is
synonymous to the flow from the intake valve which both generates a considerable amount of
the in-cylinder turbulence and is incident on or near the hottest portion of the cylinder liner
film.

Secondly, the foregoing experiments could be completed for both a longer film lengths (!c).
This would allow for mass boundary layer growth. It is expected that the effect of turbulence
would be evident for the mean evaporation for longer c! where the concentration boundary
layer significantly exceeds the laminar sublayer.

Thirdly, thinner initial film thicknesses (!h f,i ≪ 10 μm) could be considered for the same or
higher Reynolds numbers (!ReDh). Where the rate of evaporation would be similar, this would
allow for the depletion of species within the film and relatively significant changes in thickness
with evaporation [73]. When combined with longer film lengths, a true study of boundary layer
regulation could be conducted relative to the chosen hydrocarbon mixture. Where thinner films
are considered, higher Reynolds numbers could also be considered more conveniently for
discrete liquid samples (to avoid overflow under shear). Where the increase in c! is not
proportionate to the decrease in film thickness a more sensitive analytical microbalance would
be required. Another method of measurement entirely could also be explored, if necessary.

Lastly, lubricants of vastly different viscosity and volatility could be considered under
similar conditions. Where surrogate oils could be tailored with species of different, known
molecular diffusivities, the above mass transfer conditions could be evaluated with respect to
the individual components.

!76
References
[1] Caines, A. J. and Haycock, R. F., 1996, Automotive Lubricants Reference Book, Society of Automotive Engineers
Inc.

[2] Yilmaz, E., 1997, “Sources and Characteristics of Oil Consumption in a Spark-Ignition Engine,” Ph. D
dissertation, Massachusetts Institute of Technology, Cambridge, MA.

[3] Spencer, A., Avan, Almqvist, A., E. Y., Dwyer-Joyce, R. S., and Larsson, R., 2013, “An experimental and
numerical investigation of frictional losses and film thickness for four cylinder liner variants for a heavy
duty diesel engine,” Proc. IMechE Part J: J. Engineering Tribology, 227(12), pp. 1319 – 1333

[4] Yilmaz, E., Tian, T., Wong, V. W., and Heywood, J. B., 2002, “An Experimental and Theoretical Study of
the Contribution of Oil Evaporation to Oil Consumption,” SAE Technical Paper No. 2002-01-2684

[5] Soejima, M., Harigaya, Y., Hamatake, T., and Wakuri, Y., 2017, “Study on Lubricating Oil Consumption
from Evaporation of Oil-Film on Cylinder Wall for Diesel Engine,” SAE Int. J. Fuels Lubr., 10(2), pp. 487 –
501

[6] Lapuerta, M., Hernández, J. J., Ballesteros, R., and Durán, A, 2003, “Composition and size of diesel
particulate emissions from a commercial European engine tested with present and future fuels,” Proc. Instn
Mech. Engrs, 217(D), pp. 907 – 919

[7] Brandenberger, S., Mohr, M., Grob, K., and Neukom, H. P., 2005, “Contribution of unburned lubricating
oil and diesel fuel to particulate emission from passenger cars,” Atmospheric Environment, 39, pp. 6985 –
6994

[8] Williams, P. T., Abbass, M. K., Andrews, G. E., and Bartle, K. D., 1989, “Diesel Particulate Emissions: The
Role of Unburned Fuel,” Combustion and Flame, 75, pp. 1 – 24

[9] Laurence, R. B., Wong, V. W., and Brown, A. J., 1996, “Effects of Lubrication System Parameters on Diesel
Particulate Emission Characteristics,” SAE Technical Paper No. 960318

[10] ASTM D5800-18, 2018, “Standard Test Method for Evaporation Loss of Lubricating Oils by the Noack
Method”, ASTM International, West Conshohocken, PA

[11] Tobias, H. J., Beving, D. E., Ziemann, Sakurai, H., Zuk, M., McMurry, P. H., Zarling, D., Waytulonis, R.,
and Kittelson, D. B., 2001, “Chemical Analysis of Diesel Engine Nanoparticles Using a Nano-DMA/
Thermal Desorption Particle Beam Mass Spectrometer,” Environ. Sci. Technol., 35, pp. 2233 – 2243

[12] Dowling, M., 1992, “The Impact of Oil Formulation on Emissions from Diesel Engines,” SAE Technical
Paper No. 922198

[13] Yilmaz, E., Tian, T., Wong, V. W., and Heywood, J. B., 2004, “The Contribution of Different Oil
Consumption Sources to Total Oil Consumption in a Spark Ignition Engine,” SAE Technical Paper No.
2004-01-2909

[14] Didot, F. E., Green, E., and Johnson, R. H., 1987, “Volatility and Oil Consumption of SAE 5W-30 Engine
Oil,” SAE Technical Paper No. 872126

!77
[15] Carey, L. R., Roberts, D. C., and Shaub, H., 1989, “Factors Influencing Engine Oil Consumption in
Today’s Automotive Engines,” SAE Technical Paper No. 892159

[16] Manni, M., and Ciocci, G., 1992, “An Experimental Study of Oil Consumption in Gasoline Engines,” SAE
Technical Paper No. 922374

[17] ASTM D2887-16a, 2016, “Standard Test Method for Boiling Range Distribution of Petroleum Fractions
by Gas Chromatography”, ASTM International, West Conshohocken, PA

[18] Furuhama, S., Hiruma, M., and Yoshida, H., 1981, “An Increase of Engine Oil Consumption at High
Temperature of Piston and Cylinder” SAE Technical Paper No. 810976

[19] Carey, V. P., 1992, Liquid – Vapor Phase-Change Phenomena: An Introduction to the Thermophysics of Vaporization and
Condensation Processes In Heat Transfer Equipment, 1st ed., CRC Press

[20] Harigaya, Y., Yamasuga, K., Suzuki, M., Iijima, N., Takiguchi, M., and Shimada, A., 2009, “The Effect of
Oil Evaporation From the Cylinder Wall on Oil Consumption of a Gasoline Engine,” Proceedings of the
ASME 2009 Internal Combustion Engine Division Fall Technical Conference, Lucerne, Switzerland,
September 20 – 24, 2009, ASME Paper No. ICEF2009-14051, pp. 523 – 531

[21] Zhang, J., Zhang, G., He, Z., Lin, J., and Liu, H., 2013, “Analysis of Oil Consumption in Cylinder of
Diesel Engine for Optimization of Piston Rings,” Chinese Journal of Mechanical Engineering, 26(1), pp.
207 – 216

[22] Liu, L., Tian, T., and Yilmaz, E., 2005, “Modelling Oil Evaporation From the Engine Cylinder Liner With
Consideration of the Transport of Oil Species Along Liner,” Proceedings of World Tribology Conference
III, Washington, D.C., September 12 – 16, 2005, ASME Paper No. WTC2005-63984, 2, pp. 579 – 580

[23] Park, D., 2005, “The Influence of Different In-Cylinder Flows on Combustion in an SI Engine,” Ph.D.
dissertation, University of Toronto, Toronto, ON

[24] Amitay, M., Tuna, B. A., and Dell’Orso, H., 2016, “Identification and mitigation of TS waves using
localized dynamic surface modification,” Physics of Fluids, 28(6), 064103

[25] Tuna, B. A., Li, X., and Yarusevych, S., 2017, “Investigation of the Effect of Inlet Turbulence on
Transitional Wall-Bounded Flows,” Proceedings of the ASME 2017 Fluids Engineering Division Summer
Meeting, Waikoloa, Hawai’i, July 31 – August 3, 2017, ASME Paper No. FEDSM2017-69321

[26] Patel, V. C., and Head, M. R., 1969, “Some observations on skin friction and velocity profiles in fully
developed pipe and channel flows,” Journal of Fluid Mechanics, 38(1), pp. 181 – 201

[27] Westerweel, J., Draad, A. A., Van der Hoeven, J. T., and Van Oord, J., 1996, “Measurement of fully
developed turbulent pipe flow with digital particle image velocimetry,” Experiments in Fluids, 20(3), pp.
165 – 177

[28] Hellström, L. H., Ganapathisubramani, B., and Smits, A. J., 2015 “The evolution of large-scale motions in
turbulent pipe flow,” Journal of Fluid Mechanics, 779, pp. 701 – 715

[29] Kim, J., Moin, P., and Moser, R., 1987, “Turbulence statistics in fully developed channel flow at low
Reynolds number,” Journal of fluid mechanics, 177, pp. 133 – 166

!78
[30] Jiménez, J., Hoyas, S., Simens, M. P., and Mizuno, Y., 2010, “Turbulent boundary layers and channels at
moderate Reynolds numbers,” Journal of Fluid Mechanics, 657, p. 335

[31] El Khoury, G. K., Schlatter, P., Noorani, A., Fischer, P. F., Brethouwer, G., and Johansson, A. V., 2013,
“Direct numerical simulation of turbulent pipe flow at moderately high Reynolds numbers,” Flow,
turbulence and combustion, 91(3), pp. 475 – 495

[32] McComas, S. T., 1967, “Hydrodynamic entrance lengths for ducts of arbitrary cross section,” Journal of
Basic Engineering, 89(4), pp. 847 – 850

[33] Shah, R. K., 1978, “A correlation for laminar hydrodynamic entry length solutions for circular and
noncircular ducts,” Journal of Fluids Engineering, 100(2), pp. 177 – 179

[34] Durst, F., Ray, S., Ünsal, B., and Bayoumi, O. A. 2005, “The development lengths of laminar pipe and
channel flows,” Journal of fluids engineering, 127(6), pp. 1154 – 1160

[35] Bergman, T. L., Lavine, A. S., Incropera, F. P., and DeWitt, D. P., 2011, Fundamentals of Heat and Mass
Transfer, 8th ed., John Wiley and Sons

[36] Darbyshire, A. G., and Mullin, T., 1995, “Transition to turbulence in constant-mass-flux pipe flow,” Journal
of Fluid Mechanics, 289, pp. 83 – 114

[37] Mullin, T., 2011, “Experimental studies of transition to turbulence in a pipe,” Annual Review of Fluid
Mechanics, 43, pp. 1 – 24

[38] Patel, R. P., 1974, “A note on fully developed turbulent flow down a circular pipe,” The Aeronautical
Journal, Vol. 78, pp. 93 – 97

[39] Barr, D. I. H., 1980, “Technical note. The transition from laminar to turbulent flow,” Proceedings of the
Institution of Civil Engineers, 69(2), pp. 555 – 562

[40] Klein, A., 1981, “Review: Turbulent Developing Pipe Flow,” Journal of Fluids Engineering, 103, pp. 243 –
249

[41] Da Silva, D. F., and Moss, E. A., 1994, “The stability of pipe entrance flows subjected to axisymmetric
disturbances,” Journal of Fluids Engineering, 116, pp. 61 – 65

[42] Sharp, K. V., and Adrian, R. J., 2004, “Transition from laminar to turbulent flow in liquid filled micro
tubes,” Experiments in fluids, 36(5), pp. 741 – 747

[43] Zanoun, E. S., Durst, F., Bayoumy, O., and Al-Salaymeh, A., 2007, “Wall skin friction and mean velocity
profiles of fully developed turbulent pipe flows,” Experimental Thermal and Fluid Science, 32(1), pp. 249 –
261

[44] Thomas, A. S., 1983, “The control of boundary-layer transition using a wave-superposition principle,”
Journal of Fluid Mechanics, 137, pp. 233 – 250

[45] Widmann, A., Kurz, A., Simon, B., Grundmann, S., and Tropea, C. (2013). “Characterization of the
interaction between Tollmien-Schlichting waves and a DBD plasma actuator using phase-locked PIV,” In
PIV13; 10th International Symposium on Particle Image Velocimetry, Delft, The Netherlands, July 1-3,
2013

!79
[46] Eckhardt, B., Schneider, T. M., Hof, B., and Westerweel, J., 2007, “Turbulence transition in pipe flow,”
Annu. Rev. Fluid Mech., 39, pp. 447 – 468

[47] Lee, Y., and Goel, K. C., 1972, “Free stream turbulence and transition in a circular duct,” Fluid machinery
and fluidics, pp. 11 – 20

[48] Musker, A. J., 1979, “Explicit expression for the smooth wall velocity distribution in a turbulent boundary
layer,” AIAA J, 17, pp. 655 – 657

[49] Kendall, A., and Koochesfahani, M., 2008, “A method for estimating wall friction in turbulent wall-
bounded flows,” Exp Fluids, 40, pp. 773 – 780

[50] Alharbi, A., 2010, “High Speed High Resolution Vector Field Measurements and Analysis of Boundary
Layer Flows In An Internal Combustion Engine,” Ph.D. dissertation, University of Michigan, Ann Arbor,
MI.

[51] Alharbi, A. and Sick, V., 2010, “Investigation of boundary layers in internal combustion engines using a
hybrid algorithm of high speed micro-PIV and PTV,” Exp. Fluids, 49, pp. 949 – 959

[52] Hall, M. J. and Bracco, F. V., 1986, “Cycle-Resolved Velocity and Turbulence Measurements Near the
Cylinder Wall of a Firing S.I. Engine,” SAE Transactions, 95(6), pp. 526 – 539

[53] Harigaya, Y., Suzuki, M., and Takiguchi, M., 2003, “Analysis of Oil Film Thickness and Heat Transfer on
a Piston Ring of a Diesel Engine: Effect of Oil Film Temperature,” Journal of Engineering for Gas
Turbines and Power, 125, pp. 596 – 603

[54] Dearlove, J. and Cheng, W. K., 1995, “Simultaneous Piston Ring Friction and Oil Film Thickness
Measurements in a Reciprocating Test Rig,” SAE Technical Paper No. 952470

[55] Avan, E. Y., Spencer, A., Dwyer-Joyce, R. S., Almqvist, A., and Larsson, R., 2012, “Experimental and
numerical investigations of oil film formation and friction in a piston ring – liner contact,” Proc. IMechE
Part J: J. Engineering Tribology, 227(2), pp. 126 – 140

[56] Takiguchi, M., Nakayama, K., Furuhama, S., and Yoshida, H., 1998, “Variation of Piston Ring Oil Film
Thickness in an Internal Combustion Engine – Comparison Between Thrust and Anti-Thrust Sides,” SAE
Technical Paper No. 980563

[57] Harigaya, Y., Suzuki, M., Toda, F., and Takiguchi, M., 2006, “Analysis of Oil Film Thickness and Heat
Transfer on a Piston Ring of a Diesel Engine: Effect of Lubricant Viscosity,” Journal of Engineering for
Gas Turbines and Power, 128, pp. 685 – 693

[58] Wittig, S., Himmelsbach, J., Noll, B., Feld, H. J., and Samenfink, W., 1992, “Motion and Evaporation of
Shear-Driven Liquid Films in Turbulent Gases,” Journal of Engineering for Gas Turbines and Power, 114,
pp. 395 – 400

[59] Himmelsbach, J, Noll, B., and Wittig, S., 1994, “Experimental and numerical studies of evaporating wavy
fuel films in turbulent air flow,” Int. J. Heat Mass Transfer, 37(8), pp. 1217 – 1226

[60] Furuhama, S., Asahi, C., and Hiruma, M., 1983, “Measurement of Piston Ring Oil Film Thickness in an
Operating Engine,” ASLE Transactions, 26(3), pp 325 – 332

!80
[61] Söchting, S. J., and Sherrington, I., 2009, “The effect of load and viscosity on the minimum operating oil
film thickness of piston-rings in internal combustion engines,” Proc. IMechE Part J: J. Engineering
Tribology, 223, pp. 383 – 391

[62] Schröppel, J. and Thiele, F., 1983, “On the calculation of momentum, heat, and mass transfer in laminar
and turbulent boundary layer flows along a vaporizing liquid film,” Numerical Heat Transfer, 6(4), 475 –
496

[63] Kelly-Zion, P. L., Pursell, C. J., Booth, R. S., and VanTilburg, A. N., 2009, “Evaporation rates of pure
hydrocarbon liquids under the influences of natural convection and diffusion,” International Journal of
Heat and Mass Transfer, 52, pp. 3305 – 3313

[64] Pasquill, F., 1943, “Evaporation from a plane, free-liquid surface into a turbulent air stream,” Proc. R. Soc.
Lond. A, 182, pp. 75 – 95

[65] Sutton, O. G., 1934, “Wind structure and evaporation in a turbulent atmosphere,” Proc. R. Soc. Lond. A,
146, pp. 701 – 722

[66] Davies, D. R. and Walters, T. S., 1952, “Further Experiments on Evaporation from Small, Saturated, Plane
Areas into a Turbulent Boundary Layer,” Proc. Phys. Soc. B, 65(8), pp. 640 – 645

[67] Huang, C. H., 1997, “Pasquill’s Influence: On the Evaporation from Various Liquids into the
Atmosphere,” Journal of Applied Meteorology, 36, pp. 1021 – 1026

[68] Reijnhart, R., Pieper, J., and Toneman, L. H., 1980, “Vapour Cloud Dispersion and the Evaporation of
Volatile Liquids in Atmospheric Wind Fields – I. Theoretical Model,” Atmospheric Environment, 14, pp.
751 – 758

[69] Reijnhart, R., and Rose, R., 1980, “Vapour Cloud Dispersion and the Evaporation of Volatile Liquids in
Atmospheric Wind Fields – II. Wind Tunnel Experiments,” Atmospheric Environment, 14, pp. 759 – 762

[70] Okamoto, K., Watanabe, N., Hagimoto, Y., Miwa, K., and Ohtani, H., 2010, “Evaporation characteristics
of multi-component liquid,” Journal of Loss Prevention in the Process Industries, 23, pp. 89 – 97

[71] Okamoto, K., Watanabe, N., Hagimoto, Y., Miwa, K., and Ohtani, H., 2009, “Changes in evaporation rate
and vapor pressure of gasoline with progress of evaporation,” Fire Safety Journal, 44, pp. 756 – 763

[72] Mackay, D., and Matsugu, R. S., 1973, “Evaporation Rates of Liquid Hydrocarbon Spills on Land and
Water,” The Canadian Journal of Chemical Engineering, 51, pp. 434 – 439

[73] Gerendas, M., and Wittig, S., 2001, “Experimental and Numerical Investigation on the Evaporation of
Shear-Driven Multicomponent Liquid Wall Films,” Journal of Engineering for Gas Turbines and Power,
123, pp. 580 – 588

[74] Fingas, M. F., 1997, “Studies on the evaporation of crude oil and petroleum products: I. the relationship
between evaporation rate and time,” Journal of Hazardous Materials, 56, pp. 227 – 236

[75] Fingas, M. F., 1997, “Studies on the evaporation of crude oil and petroleum products II. Boundary layer
regulation,” Journal of Hazardous Materials, 57, pp. 41 – 58

[76] Fingas, M. F., 2004, “Modeling evaporation using models that are not boundary-layer regulated,” Journal
of Hazardous Materials, 107, pp. 27 – 36

!81
[77] Woschni, G., 1967, “A Universally Applicable Equation for the Instantaneous Heat Transfer Coefficient in
the Internal Combustion Engine,” SAE Technical Paper No. 670931

[78] De Petris, C., Giglio, V., and Police, G., 1977 “A Mathematical Model of the Evaporation of the Oil Film
Deposed on the Cylinder Surface of IC Engines,” SAE Technical Paper No. 972920

[79] Audette, W. E., Wong, V. W., 1999, “A Model for Estimating Oil Vaporization From The Cylinder Liner As
A Contributing Mechanism to Engine Oil Consumption,” SAE Technical Paper No. 1999-01-1520

[80] Audette, W. E., 1999, “Estimation of Oil Consumption Due to In-Cylinder Vaporization in Internal
Combustion Engines,” MASc thesis, Ph. D dissertation, Massachusetts Institute of Technology, Cambridge,
MA

[81] O’Rourke, P. J. and Amsden, A. A., 1996, “A Particle Numerical Model for Wall Film Dynamics in Port-
Injected Engines,” SAE Technical Paper No. 961961

[82] Foucart, H., Habchi, C., Le Coz, J. F., and Baritaud, T., 1998, “Development of a Three Dimensional
Model of Wall Fuel Liquid Film for Internal Combustion Engines,” SAE Technical Paper No. 980133

[83] Wahiduzzaman, S., Keribar, R., Dursunkaya, Z., and Kelley, F., 1992, “A Model for Evaporative
Consumption of Lubricating Oil in Reciprocating Engines,” SAE Technical Paper No. 922202

[84] Mehta, R. D. and Bradshaw, P., 1979, “Design rules for small low speed wind tunnels,” The Aeronautical
Journal of the Royal Aeronautical Society, 83(827), pp. 443 – 453

[85] Cattafesta, L., Bahr, C., and Mathew, J., 2010, “Fundamentals of Wind‐Tunnel Design,” Encyclopedia of
Aerospace Engineering, John Wiley and Sons

[86] Mehta, R. D., 1977, “The Aerodynamic Design of Blower Tunnels With Wide Angle Diffusers,” Prog.
Aerospace Sci., 18, pp. 59 – 120

[87] Morel, T., 1975, “Comprehensive Design of Axisymmetric Wind Tunnel Contractions,” ASME Journal of
Fluids Engineering, 97(2), pp. 225 – 233

[88] Morel, T., 1977, “Design of Two-Dimensional Wind Tunnel Contractions,” ASME Journal of Fluids
Engineering, 99(2), pp. 371 – 377

[89] Su, Y., 1991, “Flow analysis and design of three-dimensional wind tunnel contractions,” AIAA Journal,
29(11), pp. 1912 – 1920

[90] Kulkarni, V., Sahoo, N., and Chavan, S. D., 2011, “Simulation of honeycomb – screen combinations for
turbulence management in a subsonic wind tunnel,” Journal of Wind Engineering and Industrial
Aerodynamics, 99(1), pp. 37 – 45

[91] Mehta, R. D. and Hoffmann, P. H., 1987, “Boundary layer two-dimensionality in wind tunnels,”
Experiments in Fluids, 5(5), pp. 358 – 360

[92] Groth, J., and Johansson, A. V., 1988, “Turbulence reduction by screens,” Journal of Fluid Mechanics, 197,
pp. 139 – 155

[93] Mehta, R. D., 1985, “Turbulent Boundary Layer Perturbed by a Screen,” AIAA Journal, 23(9), pp. 1335 –
1342

!82
[94] Hancock, P. E., 1998, “Plane multiple screens in non-uniform flow, with particular application to wind
tunnel settling chamber screens,” European Journal of Mechanics – B/Fluids, 17(3), pp. 357 – 369

[95] Miller, D. S., 1978, Internal Flow Systems, 1st ed., British Hydromechanics Research Association

[96] Atligan, M., and Calvert, J. R., 1980, “Geometry of transition sections between ducts of equal area,”
Journal of Wind Engineering and Industrial Aerodynamics, 6(1 – 2), pp. 25 – 37

[97] Farell, C., and Xia, L., 1990, “A note on the design of screen-filled wide-angle diffusers,” Journal of Wind
Engineering and Industrial Aerodynamics, 33(3), pp. 479 – 486

[98] Ward-Smith, A. J., Lane, D. L., Reynolds, A. J., Sahin, B., and Shawe, D. J., 1991, “Flow regimes in wide-
angle screened diffusers,” International Journal of Mechanical Sciences, 33(1), pp. 41 – 54

[99] Roach, P. E., 1987, “The generation of nearly isotropic turbulence by means of grids,” International
Journal of Heat and Fluid Flow, 8(2), pp. 82 – 92.

[100] Sciacchitano, A., Neal, D. R., Smith, B. L., Warner, S. O., Vlachos, P. P., Wieneke, B., and Scarano, F.,
2015, “Collaborative framework for PIV uncertainty quantification: comparative assessment of methods,”
Measurement Science and Technology, 26(7), 074004

[101] Petukhov, B. S., 1970 “Heat Transfer and Friction in Turbulent Pipe Flow with Variable Physical
Properties.” In Advances in Heat Transfer, ed. T. F. Irvine and J. P. Hartnett, Vol. 6. New York: Academic
Press


!83
Appendix A – Blasius Laminar Boundary Layer
Solution
The following is a solution to the Navier-Stokes equations for two-dimensional, steady
flow with constant fluid properties, given by:

∂u ∂v
! + =0 (A-1)
∂x ∂y

( ∂x 2
∂ 2u ∂ 2u
∂y )
∂u ∂u 1 ∂p
!u +v =− +ν + 2 (A-2)
∂x ∂y ρ ∂x

( ∂x 2 ∂y 2 )
∂v ∂v 1 ∂p ∂ 2v ∂ 2v
!u +v =− +ν + (A-3)
∂x ∂y ρ ∂y

Observing that δ! /L ≪ 1, where δ! is the boundary layer thickness and L


! is the dimension of the
surface in the !x-direction, equations A1 – A3 may be simplified to the following, respectively:

∂u ∂v
! + =0 (A-3)
∂x ∂y

∂u ∂u ∂u ∂ 2u
!u +v = u∞ ∞ + ν 2 (A-4)
∂x ∂y ∂x ∂y

∂p
! ≅0 (A-5)
∂y

where u! = u∞ for y! ≥ δ . For the case of flow over a flat plate, where u! ∞ = constant and p! ∞ =
constant, a solution proceeds by assuming similarity of the following form,

y y
!η = = Rex
νx x (A-6)
u∞

and also by defining a “streamfunction” !ψ, which satisfies:

∂ψ
!u = (A-7)
∂y
∂ψ
!v = (A-8)
∂x

!84
The solution then proceeds by finding solutions for the functions G
! (x) and F(η)
! by
separation of variables, which is a problem defined by:

!ψ (x, η) = G (x)F(η) (A-9)

When the first solution is assumed to be !G (x) = u∞νx , the following solution results:

∂F u
!F′ = = (A-10)
∂η u∞

x ( )
∂ψ u∞ν F − ηF′
!−v = = (A-11)
∂x 2

( 2x )
∂u u∞
! =− ηF′′ (A-12)
∂x
3
∂u u∞
! = F′′ (A-13)
∂y νx
2
∂ 2u
( νx )
u∞
! 2 = F′′′ (A-14)
∂y

Finally, this yields the Blasius equation for flow over a flat plate from equation A4 and equations
A10 – A14, given by,

1
!F′′′ + FF′′ = 0 (A-15)
2

for the boundary conditions:

!y = 0 : u = 0, v = 0 ⟶ η = 0 : F = 0, F′ = 0 (A-15)
!y = ∞ : u = u∞ ⟶ η = ∞ : F′ = 1 (A-16)

Equation A15 may then be solved numerically, notably resulting in the following:

!η = 0 : F′′ = 0.33206 (A-15)


νx
!δ* = 1.72 (A-16)
u∞
νx
!θ = 0.665 (A-17)
u∞
δ*
!H = = 2.59 (A-18)
θ

!85
Appendix B – Wind Tunnel Design Drawings
Below is a set of drawings describing the design of the wind tunnel used to produce the
foregoing measurements. Critical dimensions are shown for each separate part along with the
design diameters and depths for each hole. In general, the final geometry may differ due to
changes made during machining. It is recommended that any modifications be made only after
inspection of any related parts. The original CAD files are available upon request.

The drawings are subject to the following:

i. Counterbored hole dimensions are shown for ideal dimensions. Clearances for standard
socket-head cap screws were ensured during machining.

ii. Any holes of the same diameter between parts are intended for stainless-steel alignment
dowels with a ground finish. Tight, non-interference fits were ensured during machining.

iii. Parts which are numbered the same may be rotated/flipped for assembly.

iv. Additional alignment dowel holes were included at the machinist’s discretion, and are
not shown.

Some unfamiliar symbols may include:

⌀ Diameter

⌴ Counterbore. Following diameter and depth is for counterbore geometry.

↧ Depth. Indicates a blind depth into a surface. When not shown for a hole callout, a
through-hole can be assumed.

TYP Typical dimension. Indicates where a dimension is shared by similar geometry


within the same drawing.

xx.xx Theoretically Exact. Design intent is for these dimensions to be exact to the
precision shown.

Third-Angle Projection. Designates how projected views are shown with respect
to a base view.


!86

(assy.).
A

4
3 1
2

!87
651.67
ASSY. DESCRIPTION

1 DUCT MOUNTING PLATE

2 CONTRACTION

3 SETTLING CHAMBER

4 DIVERGING SECTION
200 A A-A
5 CIRCULAR — RECTANGULAR TRANSITION

UNLESS OTHERWISE TITLE


100 SPECIFIED, DIMENSIONS
ARE IN MM. Wind Tunnel Settling
Chamber Assembly

Figure B-1. Drawing of wind tunnel settling chamber and component assemblies
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION —


DRAWN SHEET
S. J. GILL 1 OF 1

1 Ø6.35 x2 Ø0.51 x4
Ø6 Ø0.81 4
.35
x4

20
20
55
3.5 40 A
75
40 x4
1
B C 19.4 x4
2
2
A Ø6.35 x2 B-B 14.6 x4

!88
B C
Ø9.525 x4
Ø6.35 x2 50.65
Ø15.875 9.525
30.65

C-C
3.5

40.65

20

55
100
PART DESCRIPTION

1 DUCT MOUNTING PLATE

Figure B-2. Drawing of test section mounting plate and insert geometry.
2 BASELINE INSERT
75
UNLESS OTHERWISE TITLE
120 SPECIFIED, DIMENSIONS
ARE IN MM. MOUNTING PLATE &
INSERT
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION DUCT MOUNTING PLATE


DRAWN SHEET
S. J. GILL 1 OF 1

1/4-20-UNC 12.7 x2
Ø6.35 12.7

2
Ø9.525 x4
Ø15.875 9.525 60 40
20

3/8-16-UNC 12.7
52.89

52.89
107

107
3/8-16-UNC

200 40 20 80

!89
1
1 3/8-16-UNC x2 Ø6.35 12.7
25.4

PART DESCRIPTION
2
1 TOP/BOTTOM CONTOUR

Figure B-3. Drawing of top, bottom, and side contraction components.


2 SIDE CONTOUR
1
UNLESS OTHERWISE TITLE
SPECIFIED, DIMENSIONS
2 ARE IN MM. CONTRACTION
CONTOUR SECTIONS
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION CONTRACTION


DRAWN SHEET
S. J. GILL 1 OF 1

Ø6.35  x4
1/4-20-UNC x4 120 TYP Ø9.525   6.35 4
Ø6.35  x4 3
Ø4.76  x2  TYP Ø9.525   6.35
5 2

60 TYP

100 TYP
1 5

10

42.5 TYP

22.5 TYP

!90
75 TYP 200 PART DESCRIPTION
 Ø6.35  x2  TYP

Figure B-4. Drawing of settling chamber mounting plates.


1 CONTRACTION MOUNTING PLATE
90 TYP Ø6.35  x4
Ø9.525   6.35 2 PRESSURE TAPPED SCREEN HOLDER

3 SCREEN HOLDER

4 HONEYCOMB HOLDER

5 DIVERGING SECTION MOUNTING PLATE

UNLESS OTHERWISE TITLE


SPECIFIED, DIMENSIONS
ARE IN MM.
MOUNTING PLATES
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION SETTLING CHAMBER


DRAWN SHEET
S. J. GILL 1 OF 2

2 Ø6.53  x4  TYP 3 4
F Ø4.76  x2  TYP

C C D D E E

100 TYP
F-F
165 TYP
F Ø.51 x4
Ø.81 6
19.84 TYP

!91
B
G

2.76 TYP
C-C
1.52

D-D
14.29

E-E

1.52

Figure B-5. Drawing of settling chamber screens and honeycomb holder.


12.5 TYP UNLESS OTHERWISE TITLE
SPECIFIED, DIMENSIONS
ARE IN MM. SCREEN & HONEYCOMB

Ø9.53  x4
G HOLDERS
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION SETTLING CHAMBER


DRAWN SHEET
S. J. GILL 2 OF 2

Ø4.76   10 3/8-16-UNC 25.4
3

12.7
45

35.052
31

A' A 3

25
2
2
2 1

!92
238.018
3/8-16-UNC 14 x3

30
20.5
40
Ø4.76   10
6° A
PART DESCRIPTION

30
1 MOUNTING PLATE

60
2 SIDE WALL

3 TOP/BOTTOM PLATE

Figure B-6. Drawing of diverging section (diffuser) side wall components.


1/4-20-UNC 17.46 x2
UNLESS OTHERWISE TITLE
SPECIFIED, DIMENSIONS
ARE IN MM. DIVERGING SECTION
22.5 TYP

SIDE WALLS
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION DIVERGING SECTION


DRAWN SHEET
S. J. GILL 1 OF 2

140
120
87 1 1.52
° A

components.
B Ø9.525  x6
60
80

Ø15.875 7.5

15
A A-A
Ø6.35  x4
 Ø9.525   6.35

1/4-20-UNC x4
Ø6.35   10 x2

!93
100

75 TYP
C 90 TYP
22.5 TYP

200
3
87
°

UNLESS OTHERWISE TITLE


SPECIFIED, DIMENSIONS
C ARE IN MM. DIVERGING SECTION
PLATES

Figure B-7. Drawing of diverging section (diffuser) top and bottom wall
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY

THIRD ANGLE PROJECTION DIVERGING SECTION


DRAWN SHEET
S. J. GILL 2 OF 2
Ø4.76   12.5 TYP 55 TYP

4
3

25 TYP
1/4-20-UNC

transition.
12.5 x2 TYP

R1
Ø6.35   12.5 x6

7.5

4.76
1/4-20-UNC 12.5 x4
2
1

150 TYP

130 TYP
Ø6.35  x6

112.5 TYP
77.5 TYP
Ø9.525 12.7

!94
130 TYP Ø6.35  x6
3.5°  TYP
49 TYP

25 TYP
42.5 TYP
PART DESCRIPTION
69.967
3 1 DIVERGING SECTION MOUNTING PLATE
17.526

A
2 2 TRANSITION BOTTOM

3 TRANSITION TOP

YP
T
4 BLOWER ATTACHMENT PLATE

1.5
R
A UNLESS OTHERWISE TITLE
SPECIFIED, DIMENSIONS
ARE IN MM. TRANSITION TOP &
BOTTOM

Figure B-8. Drawing of top and bottom halves of the circular-rectangular


.xx ± 0.01
.xxx ± 0.01 ASSEMBLY
CIRCULAR — RECTANGULAR
THIRD ANGLE PROJECTION TRANSITION

DRAWN SHEET
S. J. GILL 1 OF 2

Ø6.35  x4
Ø4.76   10 x2 1-1/4 NPT Ø4.76  x2  Ø11.11   6.35
DEPTH FOR FLUSH FIT

70

Ø9.525  x2
 Ø15.875   7.5 130 4

15
B

Ø6.35  x4

!95
B  Ø9.525   6.35
11.43

Figure B-9. Drawing of transition mounting plates.


70
35.052

B-B
1.52

69.967
1
10 TYP

200

UNLESS OTHERWISE TITLE


SPECIFIED, DIMENSIONS
ARE IN MM. TRANSITION MOUNTING
PLATES
.xx ± 0.01
.xxx ± 0.01 ASSEMBLY
CIRCULAR — RECTANGULAR
THIRD ANGLE PROJECTION TRANSITION

DRAWN SHEET
S. J. GILL 2 OF 2
Appendix C – Detailed Test Section Flow
A summary of the PIV measurements within the evaporation test section, detailed in §3.2
and §3.3.4, are presented in the Figures below. The entire dataset is available upon request.
The PIV measurements in the x! -!y plane for z! /c = 0, 0.04, 0.08, 0.12. 0.16, 0.20, 0.24, 0.28,
0.32, and 0.36 were conducted for two overlapping FOVs. The region of overlap extends from
0.38 ≤ x′
! /c ≤ 0.52. Within this region, the ⟨u⟩
! and u′
! rms data is presented as an interpolation
from one FOV to the other. In some cases, namely Figures C-5, C-6, C-10, and C-11, this
produces a visible distortion in the flow near the boundary layer edge (!⟨u⟩/U ≥ 0.9). The flow
outside the boundary layer in the Grid inlet cases (!x /a = 2) displays some non-uniformity due
to the presence of the wire meshes, which is likely the cause of the differences between the
FOVs.

In each b) and d) subfigure profiles for ⟨u⟩


! and u′
! rms are presented at x′
! /c = 0.5 for each
z! /c measurement plane. The following colour scheme is used to differentiate each !z /c location:

Table C-1. Colour scheme for "z /c locations.


!z /c 0 0.04 0.08 0.12 0.16 0.20 0.24 0.28 0.32 0.36

Legend ● ● ● ● ● ● ● ● ● ●

!96
"
" Dh = 10,650 case at
Figure C-1. Summary of PIV measurements for Baseline, Re
x" /a = 2.

"
" Dh = 10,650 case at
Figure C-2. Summary of PIV measurements for Baseline, Re
x" /a = 45.


!97
"
" Dh = 17,750 case at
Figure C-3. Summary of PIV measurements for Baseline, Re
x" /a = 2.

"
" Dh = 35,500 case at
Figure C-4. Summary of PIV measurements for Baseline, Re
x" /a = 2.


!98
"
" Dh = 10,650 case at x" /a =
Figure C-5. Summary of PIV measurements for Grid 3, Re
2.


!99
"
" Dh = 17,750 case at x" /a =
Figure C-6. Summary of PIV measurements for Grid 1, Re
2.


!100
"
" Dh = 17,750 case at x" /a =
Figure C-7. Summary of PIV measurements for Grid 2, Re
2.

!101
"
" Dh = 17,750 case at x" /a =
Figure C-8. Summary of PIV measurements for Grid 3, Re
2.

!102
"
" Dh = 35,500 case at x" /a =
Figure C-9. Summary of PIV measurements for Grid 1, Re
2.


!103
"
" Dh = 35,500 case at
Figure C-10. Summary of PIV measurements for Grid 2, Re
x" /a = 2.


!104
"
" Dh = 35,500 case at
Figure C-11. Summary of PIV measurements for Grid 3, Re
x" /a = 2.


!105
"
" Dh = 17,750 case at
Figure C-12. Summary of PIV measurements for Baseline, Re
x" /a = 45.


!106
"
" Dh = 35,500 case at
Figure C-13. Summary of PIV measurements for Baseline, Re
x" /a = 45.


!107
"
" Dh = 10,650 case at
Figure C-14. Summary of PIV measurements for Grid 3, Re
x" /a = 45.


!108
"
" Dh = 17,750 case at
Figure C-15. Summary of PIV measurements for Grid 1, Re
x" /a = 45.


!109
"
" Dh = 17,750 case at
Figure C-16. Summary of PIV measurements for Grid 2, Re
x" /a = 45.


!110
"
" Dh = 17,750 case at
Figure C-17. Summary of PIV measurements for Grid 3, Re
x" /a = 45.


!111
"
" Dh = 35,500 case at
Figure C-18. Summary of PIV measurements for Grid 1, Re
x" /a = 45.


!112
"
" Dh = 35,500 case at
Figure C-19. Summary of PIV measurements for Grid 2, Re
x" /a = 45.


!113
"
" Dh = 35,500 case at
Figure C-20. Summary of PIV measurements for Grid 3, Re
x" /a = 45.


!114
Appendix D – Evaporation Measurement Data
Each evaporation measurement taken in the foregoing work is tabulated below in Tables D-1
through D-10. The mean evaporated mass over each of the 5 (or 10) measurements for each
exposure time (!t) is given as !mevap. Similarly, the mean initial film mass is given by !mo.
Consequently, the standard deviation for the !mevap,i samples is given by:

1 N
( evap,i evap)
2

N∑
!σ = m − m (D-1)
i=1

Table D-1. Evaporation data for Natural Convection case ("Grc = 3838)
t! (s) m
! o (g) m
! 1 (g) m
! i (g) m
! 2 (g) m
! f (g) m
! evap (g) m
! o (g) m
! evap (g) σ! (g)
0 31.50051 31.52660 0.02609 31.52643 0.02592 0.00017 0.02629 0.00020 0.00003
31.50201 31.52877 0.02676 31.52861 0.02660 0.00016
31.50056 31.52659 0.02603 31.52636 0.02580 0.00022
31.50211 31.52820 0.02609 31.52801 0.02590 0.00019
31.50054 31.52702 0.02648 31.52678 0.02624 0.00024
150 31.50051 31.52827 0.02776 31.52780 0.02729 0.00047 0.02660 0.00048 0.00003
31.50201 31.52721 0.02520 31.52672 0.02471 0.00049
31.50056 31.52861 0.02805 31.52817 0.02761 0.00044
31.50211 31.52655 0.02444 31.52606 0.02395 0.00050
31.50054 31.52808 0.02754 31.52756 0.02702 0.00052
300 31.50197 31.52867 0.02670 31.52789 0.02592 0.00078 0.02641 0.00079 0.00004
31.50048 31.52665 0.02617 31.52589 0.02541 0.00076
31.50192 31.52822 0.02630 31.52747 0.02555 0.00074
31.50045 31.52694 0.02649 31.52610 0.02565 0.00084
31.50191 31.52831 0.02640 31.52749 0.02558 0.00082
450 31.50213 31.52820 0.02607 31.52722 0.02509 0.00098 0.02615 0.00104 0.00006
31.50077 31.52678 0.02601 31.52575 0.02498 0.00103
31.50191 31.52820 0.02629 31.52719 0.02528 0.00101
31.50057 31.52659 0.02602 31.52555 0.02498 0.00103
31.50207 31.52815 0.02608 31.52713 0.02506 0.00102
31.50045 31.52665 0.02620 31.52568 0.02523 0.00097
31.50207 31.52823 0.02616 31.52715 0.02508 0.00108
31.50056 31.52686 0.02630 31.52568 0.02512 0.00117
31.50210 31.52815 0.02605 31.52711 0.02501 0.00104
31.50056 31.52687 0.02631 31.52576 0.02520 0.00111
600 31.50052 31.52686 0.02634 31.52549 0.02497 0.00137 0.02628 0.00135 0.00006
31.50196 31.52810 0.02614 31.52683 0.02487 0.00127
31.50050 31.52670 0.02620 31.52534 0.02484 0.00136
31.50191 31.52832 0.02641 31.52690 0.02499 0.00142
31.50046 31.52675 0.02629 31.52545 0.02499 0.00130

!115
" Dh = 17,750 case ("x /a = 2)
Table D-2. Evaporation data for Baseline, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 31.26419 31.29062 0.02643 31.29038 0.02619 0.00024 0.02640 0.00017 0.00005
31.26409 31.29039 0.02630 31.29021 0.02612 0.00018
31.26405 31.29063 0.02658 31.29052 0.02647 0.00011
31.26409 31.29018 0.02609 31.28998 0.02589 0.00020
31.26414 31.29076 0.02662 31.29063 0.02649 0.00013
25 56.51333 56.53975 0.02642 56.53913 0.02580 0.00063 0.02647 0.00051 0.00008
56.51318 56.53968 0.02650 56.53925 0.02607 0.00043
56.51314 56.53937 0.02623 56.53889 0.02575 0.00048
56.51306 56.54000 0.02694 56.53955 0.02649 0.00044
56.51313 56.53939 0.02626 56.53883 0.02570 0.00057
50 56.51332 56.54013 0.02681 56.53933 0.02601 0.00081 0.02657 0.00075 0.00007
56.51325 56.53984 0.02659 56.53905 0.02580 0.00079
56.51319 56.53964 0.02645 56.53893 0.02574 0.00072
56.51312 56.53993 0.02681 56.53930 0.02618 0.00063
56.51311 56.53931 0.02620 56.53851 0.02540 0.00080
150 56.51382 56.54053 0.02671 56.53872 0.02490 0.00180 0.02637 0.00172 0.00013
56.54640 56.57263 0.02623 56.57116 0.02476 0.00147
56.51371 56.54033 0.02662 56.53875 0.02504 0.00158
56.51366 56.53980 0.02614 56.53806 0.02440 0.00174
56.51361 56.53962 0.02601 56.53793 0.02432 0.00168
56.51336 56.53946 0.02610 56.53787 0.02451 0.00159
56.51323 56.53932 0.02609 56.53746 0.02423 0.00186
56.51316 56.53975 0.02659 56.53798 0.02482 0.00177
56.51315 56.53999 0.02684 56.53822 0.02507 0.00177
56.51329 56.53966 0.02637 56.53777 0.02448 0.00190

!116
" Dh = 10,650 case ("x /a = 2)
Table D-3. Evaporation data for Baseline, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51300 56.53944 0.02644 56.53918 0.02618 0.00025 0.02663 0.00017 0.00006
56.51286 56.53973 0.02687 56.53953 0.02667 0.00020
56.51282 56.53910 0.02628 56.53894 0.02612 0.00016
56.51271 56.53943 0.02672 56.53930 0.02659 0.00013
56.51270 56.53955 0.02685 56.53943 0.02673 0.00012
25 56.51295 56.53978 0.02683 56.53937 0.02642 0.00041 0.02660 0.00037 0.00003
56.51290 56.53972 0.02682 56.53939 0.02649 0.00033
56.51293 56.53934 0.02641 56.53896 0.02603 0.00038
56.51291 56.53936 0.02645 56.53898 0.02607 0.00038
56.51275 56.53925 0.02650 56.53891 0.02616 0.00034
50 56.51286 56.53973 0.02687 56.53918 0.02632 0.00056 0.02662 0.00055 0.00007
56.51283 56.53926 0.02643 56.53870 0.02587 0.00056
56.51263 56.53913 0.02650 56.53868 0.02605 0.00045
56.51278 56.53935 0.02657 56.53872 0.02594 0.00063
56.51264 56.53937 0.02673 56.53881 0.02617 0.00056
150 56.51262 56.53940 0.02678 56.53822 0.02560 0.00118 0.02656 0.00104 0.00008
56.51243 56.53895 0.02652 56.53799 0.02556 0.00096
56.51242 56.53896 0.02654 56.53792 0.02550 0.00104
56.51230 56.53908 0.02678 56.53804 0.02574 0.00103
56.51220 56.53837 0.02617 56.53735 0.02515 0.00101

" Dh = 35,500 case ("x /a = 2)


Table D-4. Evaporation data for Baseline, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51360 56.54008 0.02648 56.53988 0.02628 0.00020 0.02649 0.00017 0.00006
56.51348 56.53990 0.02642 56.53967 0.02619 0.00023
56.51375 56.54007 0.02632 56.53986 0.02611 0.00020
56.51352 56.54041 0.02689 56.54031 0.02679 0.00010
56.51336 56.53971 0.02635 56.53960 0.02624 0.00011
25 56.51313 56.53982 0.02669 56.53902 0.02589 0.00079 0.02661 0.00073 0.00009
56.51305 56.53984 0.02679 56.53926 0.02621 0.00059
56.51316 56.53925 0.02609 56.53853 0.02537 0.00072
56.51324 56.53979 0.02655 56.53897 0.02573 0.00082
56.51322 56.54014 0.02692 56.53940 0.02618 0.00074
50 56.51349 56.53977 0.02628 56.53862 0.02513 0.00115 0.02657 0.00123 0.00005
56.51348 56.54028 0.02680 56.53904 0.02556 0.00124
56.51353 56.54030 0.02677 56.53902 0.02549 0.00128
56.51338 56.54006 0.02668 56.53880 0.02542 0.00126
56.51346 56.53980 0.02634 56.53856 0.02510 0.00124
150 56.51333 56.53959 0.02626 56.53689 0.02356 0.00270 0.02636 0.00272 0.00005
56.51353 56.54004 0.02651 56.53731 0.02378 0.00272
56.51323 56.53976 0.02653 56.53703 0.02380 0.00273
56.51323 56.53952 0.02629 56.53674 0.02351 0.00278
56.51310 56.53933 0.02623 56.53667 0.02357 0.00265

!117
" Dh = 10,650 case ("x /a = 2)
Table D-5. Evaporation data for Grid 1, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51162 56.53786 0.02624 56.53774 0.02612 0.00012 0.02649 0.00018 0.00005
56.51157 56.53801 0.02644 56.53788 0.02631 0.00013
56.51146 56.53811 0.02665 56.53791 0.02645 0.00020
56.51155 56.53800 0.02645 56.53775 0.02620 0.00025
56.51147 56.53813 0.02666 56.53794 0.02647 0.00018
25 56.51153 56.53817 0.02664 56.53784 0.02631 0.00033 0.02648 0.00031 0.00004
56.51141 56.53787 0.02646 56.53758 0.02617 0.00029
56.51156 56.53782 0.02626 56.53750 0.02594 0.00032
56.51141 56.53781 0.02640 56.53757 0.02616 0.00024
56.51144 56.53805 0.02661 56.53771 0.02627 0.00035
50 56.51143 56.53800 0.02657 56.53751 0.02608 0.00049 0.02645 0.00050 0.00005
56.51135 56.53766 0.02631 56.53722 0.02587 0.00044
56.51133 56.53775 0.02642 56.53724 0.02591 0.00051
56.51140 56.53807 0.02667 56.53751 0.02611 0.00057
56.51136 56.53763 0.02627 56.53712 0.02576 0.00051
150 56.51165 56.53813 0.02648 56.53683 0.02518 0.00130 0.02640 0.00119 0.00013
56.51152 56.53814 0.02662 56.53693 0.02541 0.00121
56.51162 56.53773 0.02611 56.53640 0.02478 0.00133
56.51145 56.53806 0.02661 56.53701 0.02556 0.00105
56.51134 56.53752 0.02618 56.53645 0.02511 0.00107

" Dh = 17,750 case ("x /a = 2)


Table D-6. Evaporation data for Grid 1, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51308 56.53969 0.02661 56.53959 0.02651 0.00010 0.02654 0.00015 0.00003
56.51300 56.53940 0.02640 56.53921 0.02621 0.00019
56.51294 56.53955 0.02661 56.53939 0.02645 0.00015
56.51290 56.53938 0.02648 56.53923 0.02633 0.00015
56.51295 56.53954 0.02659 56.53941 0.02646 0.00013
25 56.51298 56.53959 0.02661 56.53907 0.02609 0.00052 0.02646 0.00053 0.00003
56.51296 56.53920 0.02624 56.53866 0.02570 0.00055
56.51283 56.53947 0.02664 56.53899 0.02616 0.00048
56.51299 56.53947 0.02648 56.53891 0.02592 0.00056
56.51308 56.53939 0.02631 56.53884 0.02576 0.00055
50 56.51306 56.53964 0.02658 56.53878 0.02572 0.00086 0.02668 0.00081 0.00003
56.51294 56.53971 0.02677 56.53891 0.02597 0.00080
56.51299 56.53933 0.02634 56.53853 0.02554 0.00080
56.51295 56.53975 0.02680 56.53895 0.02600 0.00080
56.51284 56.53973 0.02689 56.53896 0.02612 0.00077
150 56.51299 56.53972 0.02673 56.53815 0.02516 0.00157 0.02673 0.00155 0.00003
56.51292 56.53950 0.02658 56.53794 0.02502 0.00156
56.51289 56.53951 0.02662 56.53795 0.02506 0.00156
56.51283 56.53963 0.02680 56.53808 0.02525 0.00155
56.51281 56.53971 0.02690 56.53821 0.02540 0.00150

!118
" Dh = 17,750 case ("x /a = 2)
Table D-7. Evaporation data for Grid 2, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51303 56.53950 0.02647 56.53935 0.02632 0.00015 0.02652 0.00015 0.00003
56.51295 56.53951 0.02656 56.53937 0.02642 0.00014
56.51298 56.53956 0.02658 56.53938 0.02640 0.00018
56.51300 56.53946 0.02646 56.53930 0.02630 0.00016
56.51270 56.53924 0.02654 56.53913 0.02643 0.00011
25 56.51304 56.53926 0.02622 56.53868 0.02564 0.00058 0.02643 0.00050 0.00006
56.51301 56.53927 0.02626 56.53879 0.02578 0.00048
56.51292 56.53963 0.02671 56.53922 0.02630 0.00041
56.51296 56.53942 0.02646 56.53888 0.02592 0.00054
56.51290 56.53940 0.02650 56.53890 0.02600 0.00050
50 56.51290 56.53935 0.02645 56.53856 0.02566 0.00079 0.02665 0.00074 0.00005
56.51288 56.53966 0.02678 56.53894 0.02606 0.00072
56.51293 56.53950 0.02657 56.53878 0.02585 0.00073
56.51302 56.53994 0.02692 56.53927 0.02625 0.00067
56.51306 56.53958 0.02652 56.53881 0.02575 0.00077
150 56.51325 56.53942 0.02617 56.53773 0.02448 0.00169 0.02653 0.00170 0.00008
56.51315 56.53953 0.02638 56.53795 0.02480 0.00158
56.51315 56.53990 0.02675 56.53810 0.02495 0.00180
56.51303 56.53993 0.02690 56.53825 0.02522 0.00168
56.51295 56.53941 0.02646 56.53767 0.02472 0.00174

" Dh = 17,750 case ("x /a = 2)


Table D-8. Evaporation data for Grid 3, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51305 56.53949 0.02644 56.53937 0.02632 0.00012 0.02650 0.00015 0.00003
56.51293 56.53935 0.02642 56.53918 0.02625 0.00017
56.51302 56.53957 0.02655 56.53942 0.02640 0.00015
56.51291 56.53938 0.02647 56.53927 0.02636 0.00011
56.51296 56.53958 0.02662 56.53939 0.02643 0.00019
25 56.51307 56.53956 0.02649 56.53901 0.02594 0.00055 0.02655 0.00055 0.00006
56.51299 56.53947 0.02648 56.53890 0.02591 0.00057
56.51283 56.53949 0.02666 56.53902 0.02619 0.00047
56.51300 56.53940 0.02640 56.53877 0.02577 0.00063
56.51295 56.53966 0.02671 56.53910 0.02615 0.00055
50 56.51305 56.53959 0.02654 56.53869 0.02564 0.00090 0.02662 0.00083 0.00009
56.51303 56.53954 0.02651 56.53875 0.02572 0.00079
56.51306 56.53967 0.02661 56.53893 0.02587 0.00075
56.51301 56.53977 0.02676 56.53901 0.02600 0.00076
56.51306 56.53971 0.02665 56.53878 0.02572 0.00094
150 56.51301 56.53951 0.02650 56.53779 0.02478 0.00171 0.02652 0.00168 0.00007
56.51295 56.53977 0.02682 56.53812 0.02517 0.00165
56.51302 56.53935 0.02633 56.53758 0.02456 0.00177
56.51289 56.53929 0.02640 56.53772 0.02483 0.00157
56.51275 56.53933 0.02658 56.53765 0.02490 0.00168

!119
" Dh = 10,650 case ("x /a = 45)
Table D-9. Evaporation data for Baseline, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51157 56.53774 0.02617 56.53757 0.02600 0.00017 0.02631 0.00011 0.00004
56.51138 56.53760 0.02622 56.53753 0.02615 0.00007
56.51133 56.53792 0.02659 56.53784 0.02651 0.00008
56.51126 56.53755 0.02629 56.53743 0.02617 0.00012
56.51124 56.53754 0.02630 56.53745 0.02621 0.00009
25 56.51148 56.53764 0.02616 56.53737 0.02589 0.00028 0.02641 0.00025 0.00004
56.51145 56.53772 0.02627 56.53747 0.02602 0.00025
56.51149 56.53819 0.02670 56.53790 0.02641 0.00029
56.51138 56.53799 0.02661 56.53781 0.02643 0.00018
56.51146 56.53774 0.02628 56.53748 0.02602 0.00026
50 56.51149 56.53789 0.02640 56.53752 0.02603 0.00037 0.02640 0.00036 0.00003
56.51148 56.53775 0.02627 56.53733 0.02585 0.00041
56.51145 56.53783 0.02638 56.53750 0.02605 0.00034
56.51144 56.53793 0.02649 56.53758 0.02614 0.00035
56.51138 56.53782 0.02644 56.53750 0.02612 0.00033
150 56.51140 56.53773 0.02633 56.53678 0.02538 0.00095 0.02639 0.00095 0.00003
56.51138 56.53766 0.02628 56.53669 0.02531 0.00097
56.51125 56.53793 0.02668 56.53701 0.02576 0.00092
56.51129 56.53774 0.02645 56.53683 0.02554 0.00091
56.51125 56.53746 0.02621 56.53647 0.02522 0.00099

" Dh = 17,750 case ("x /a = 45)


Table D-10. Evaporation data for Baseline, Re
!t !mo !m1 !mi !m 2 !m f !mevap !mo !mevap !σ
0 56.51143 56.53756 0.02613 56.53745 0.02602 0.00011 0.02631 0.00014 0.00004
56.51142 56.53765 0.02623 56.53755 0.02613 0.00010
56.51146 56.53783 0.02637 56.53763 0.02617 0.00020
56.51153 56.53806 0.02653 56.53794 0.02641 0.00012
56.51150 56.53777 0.02627 56.53761 0.02611 0.00016
25 56.51150 56.53776 0.02626 56.53731 0.02581 0.00044 0.02646 0.00056 0.00007
56.51142 56.53771 0.02629 56.53713 0.02571 0.00057
56.51150 56.53812 0.02662 56.53750 0.02600 0.00062
56.51142 56.53786 0.02644 56.53728 0.02586 0.00058
56.51137 56.53804 0.02667 56.53745 0.02608 0.00059
50 56.51155 56.53792 0.02637 56.53710 0.02555 0.00082 0.02632 0.00090 0.00009
56.51155 56.53797 0.02642 56.53708 0.02553 0.00088
56.51168 56.53783 0.02615 56.53679 0.02511 0.00105
56.51147 56.53774 0.02627 56.53684 0.02537 0.00090
56.51150 56.53790 0.02640 56.53704 0.02554 0.00086
150 56.51159 56.53817 0.02658 56.53631 0.02472 0.00186 0.02643 0.00171 0.00013
56.51158 56.53781 0.02623 56.53605 0.02447 0.00176
56.51150 56.53829 0.02679 56.53669 0.02519 0.00161
56.51159 56.53789 0.02630 56.53613 0.02454 0.00176
56.51160 56.53785 0.02625 56.53630 0.02470 0.00155

!120

You might also like